Method for detecting electromagnetic interference and method for suppressing it

Abstract

The present invention provides an ultrasonic probe and its signal processing that take into account the effects of temperature and temperature fluctuations at different locations within a substrate. [Solution] Disclosed is a method and apparatus for providing inspection of a weld, the method comprising: preparing a welding inspection apparatus in close proximity to a weld on a substrate to be inspected; performing an inspection, wherein the substrate is subjected to a high temperature exceeding the ambient temperature by heating during the inspection, the performance of the inspection comprising: emitting ultrasonic waves into the volume of the substrate and the weld; receiving at least a portion of the ultrasonic waves from the substrate and the weld, thereby obtaining a plurality of signal sets; and processing one or more of the plurality of signal sets to provide welding inspection data, the processing comprising correction of the temperature distribution in the volume of the substrate and / or the weld at high temperatures.

Description

[Technical Field]

[0001] This disclosure relates to improvements in ultrasonic probes for nondestructive ultrasonic testing and related matters. Improvements to the processing and compensation of those signals, and their use, in particular, their high temperature This relates to, but is not limited to, placement. [Background technology]

[0002] Ultrasonic testing is used for non-destructive testing of various objects. The transmitting transducer is... It emits ultrasound, which enters an object and interacts with the object and its secondary features, and then receives Return to the transducer. The temperature of an object affects the speed of sound within the object, therefore the test and This can negatively affect the quality of the image.

[0003] When testing a weld, after the welding is complete or after each individual weld pass, the weld or weld It is desirable that ultrasound testing can be performed immediately after the path is formed, thereby enabling practical Minimizing the time required for line, test, and corrections needed for welding or welding passes. It is possible. However, this is because the object or substrate is hot, and between each part of the object or substrate... This means that the ultrasound examination is performed while there are temperature fluctuations. [Overview of the project]

[0004] One of the potential purposes of this disclosure is to consider the effects of temperature and temperature fluctuations depending on location within the substrate. The objective is to provide an ultrasonic probe and its signal processing that does the following. One of the objectives is to provide accurate images over a wide range of temperatures, especially in high-temperature conditions. This involves considering the various temperatures in the location where the product is being grown.

[0005] According to a first aspect of the present disclosure, a method for inspecting a welded joint, comprising: (a) preparing a welding inspection device in proximity to the welded joint on a substrate to be inspected; (b) performing an inspection, wherein the substrate is subjected to a high temperature exceeding the ambient temperature by heating during the inspection, and the performance of the inspection comprises: a. emitting ultrasonic waves within the volume of the substrate and the welded joint; b. receiving at least a portion of the ultrasonic waves from the substrate and the welded joint, thereby obtaining a plurality of signal sets; and c. processing one or more of the plurality of signal sets to provide welding inspection data; wherein the processing includes correcting for the temperature within the volume of the substrate and / or the welded joint at the high temperature, providing a method for inspecting a welded joint. The method may include a high temperature that is consistent across the entire substrate and / or a temperature distribution such as a temperature gradient in the substrate and / or the welded joint and / or the welding inspection device. The processing includes correcting for the temperature distribution, such as a temperature gradient, within the volume of the substrate and / or the welded joint at the high temperature.

[0006] The method may be defined such that the correction includes correcting at least a portion of the path of the ultrasonic waves passing through the volume of the welding inspection device and / or the substrate and / or the welded joint to provide a corrected path.

[0007]

[0008]

[0009] This can be defined as follows: In this method, a portion of the ultrasound in the welding inspection device is transmitted through one or more media. It can be defined that the substrate has path characteristics. This method defines that a portion of the ultrasound within the substrate is 1 It may be specified that the substrate elements have path characteristics within a larger or larger substrate element. This method involves ultrasonic waves within the weld. It may be specified that a portion of the weld has path characteristics within one or more weld elements. This method is a welding method. A portion of the ultrasound at the interface between the inspection device and the substrate passes through one or more interface elements. It may be defined that it possesses certain characteristics.

[0010] Ultrasound is transmitted to each of the multiple elements, for example, to each element adjacent to the preceding and succeeding elements. Ultrasound may have path characteristics in each of one or more medium elements. Ultrasound may have path characteristics at each of one or more interface elements. Each of the above base material elements may have path characteristics. Ultrasound is applied to each of one or more welding elements. Each may have path characteristics.

[0011] Ultrasound has path characteristics in each of one or more medium elements, and then one or more base Each material element may have path characteristics. Ultrasound may, for example, be applied after the medium element and / or It may have path characteristics at each of one or more interface elements, such as in front of the substrate element. Ultrasound is Each of the welding elements, for example, after the base material element, may have path characteristics.

[0012] Ultrasound, for example, has path characteristics in each of one or more substrate elements in the return path. And thereafter, each of the one or more media elements may have path characteristics. Ultrasound, for example, Each of the welding elements, such as in front of the base material element, may have path characteristics. Ultrasound is For example, odors in each of one or more interface elements, such as after the substrate element and / or before the medium element. It may have path characteristics.

[0013] This method involves the ultrasonic waves entering a first element of the substrate, for example, the first substrate element, before a portion of the ultrasonic waves enter the substrate. A portion of the wave has path characteristics within the elements of the welding inspection device, such as the medium element, and / or boundary It is possible to define that a path characteristic exists within a plane, for example, within an interface element. This method involves the first element The temperature has a temperature within a temperature distribution such as a temperature gradient relative to the volume of the base material and / or weld, With temperature-compensated path characteristics determined for the ultrasonic portion within one element, temperature compensation The path characteristics are those of the elements of the welding inspection device, for example, the medium element and / or interface element, and It is determined that it is based on the temperature change between the first element of the substrate, for example, the first substrate element. It can be determined.

[0014] The change in path characteristics can be calculated according to Snell's law of refraction.

[0015] This method involves the ultrasonic waves entering a second element of the substrate, for example, a second substrate element, before the ultrasonic waves enter the second element. It is defined that a portion of the wave has path characteristics within a first element of the substrate, for example, within the first substrate element. This method involves the first element being such that a temperature gradient is applied to the volume of the base material and / or weld. The temperature within the temperature distribution, and the second element has a temperature gradient relative to the volume of the substrate and / or weld. The temperature within the temperature distribution such as the distribution and the temperature-corrected path characteristics are a second element, for example, a second The path characteristics, determined for the ultrasonic portion within the substrate element and temperature-compensated, are for the first element and the second element. Based on the temperature change between two elements, for example, between the first base material element and the second base material element. It can be further defined that it is based on.

[0016] This method involves using ultrasound to direct a portion of the ultrasound to a second further element of the substrate, for example, a second further substrate element. Before entering, a portion of the ultrasound is directed to a first further element of the substrate, for example, a first further substrate element. It may be defined that it has path characteristics within. This method is characterized in that the first further element is a substrate and / Alternatively, the temperature has a temperature within a temperature distribution such as a temperature gradient relative to the volume of the weld, and a second further requirement The element has a temperature within a temperature distribution such as a temperature gradient relative to the volume of the base material and / or welded area. Temperature-compensated path characteristics in a second further element, for example, a second further substrate element, ultrasonic The path characteristics, determined for the wave portion and temperature-corrected, are the first further element and the second further Between elements, for example, between a first further base material element and a second further base material element It can be further defined that it is based on temperature changes.

[0017] This method involves a portion of the ultrasonic waves being used in the welding inspection device and / or the substrate and / or the weld. Each element that passes through, for example, the welding inspection device and / or the base material and / or the welded area. For each element, it may be stipulated that temperature-compensated path characteristics be determined.

[0018] This method assumes that a change in temperature corresponds to a change in the speed of sound between the speed of sound in one element and the speed of sound in the next element. It can be defined that it will be expressed in this way.

[0019] This method involves ultrasonic testing within the welding inspection device and / or within the volume of the base material and / or within the weld. Multiple different parts of the wave, for example, a welding inspection device and / or a substrate and / or a weld. It may be stipulated that the entire process be corrected and a corrected path is provided.

[0020] This method involves, for example, passing through a welding inspection device and / or a base material and / or a welded area, and Multiple correction paths may be provided, such as going back and / or returning. This method provides at least five correction paths. The positive path, and in some cases at least 15 corrective paths, and in some cases at least 25 Correction paths, and at least 40 correction paths as needed, for example, 64 correction paths It can provide.

[0021] For each beam of ultrasound emitted by the transducer, for example, 64 A corrected path, such as a phased array of elements, may be provided.

[0022] In this method, a region of interest is selected, and the region of interest is the body of the substrate and weld through which the ultrasonic waves have passed. Within the product, the region of interest can be defined as being subdivided into locations such as pixels. This allows for the application of signal correction to locations such as pixels. This method applies to each pixel within the region of interest. Beam paths to locations such as xels, locations such as pixels, and / or return beam paths. Signal correction can be applied to each case.

[0023] This method involves signal correction at a location such as a pixel and at least one of the corrected paths. It can be stipulated that the number is determined according to at least one pair of relationships. This relationship is such that It can be any kind of geometric relationship. This relationship is the relative geometric position of one or more correction paths. This could be a weighted correction based on placement.

[0024] This method involves signal correction at locations such as pixels and at least one, preferably at least one. The relationship between at least one, preferably at least one pair of, positions on a pair of corrected paths. It can be stipulated that it is determined according to this. This relationship can be a geometric relationship. This includes a location and at least one, preferably at least one pair of, corrected paths. Another, preferably, a relative geometric position such as the distance between at least one pair of positions. This could be a weighted correction based on the above.

[0025] This method targets an emitted ultrasonic beam passing through one or more locations, such as pixels. This may include signal correction for the corrected path. Signal correction is applied to locations such as pixels. The process is performed primarily or exclusively on a corrected path for the emitted ultrasonic beam. It is possible. Signal correction is performed at location and at least one, preferably at least one pair of corrected This is a weighting correction based on relative geometric positions, such as the distance between positions on the path. Based on the relationship, signal correction may potentially affect the emission passing through locations such as pixels. This may be based primarily or exclusively on a corrected path for the emitted ultrasonic beam.

[0026] This method addresses pixels that do not pass through the corrected path for the emitted ultrasonic beam. This may include signal correction for one or more locations. This method provides correction for the emitted beam. Based on calculations or observed corrections for the location through which the path is taken, the emitted beam is directed to This may include signal correction for locations such as pixels that are not traversed by the corrected path. The correction is calculated or observed at multiple locations, for example, four locations, through which the emitted beam passed. It can be done based on the position.

[0027] Signal correction is performed at multiple locations on the first emission beam and multiple locations on the second emission beam. This can be done based on the calculated or observed position. The first and second emission beams are These may be adjacent beams within a set of beams. The first beam is in a position where signal correction is required. One beam may be on one side of the location, and the second beam may be on the other side of the location where signal correction is needed.

[0028] Signal correction for a particular location is, for example, for one or more other locations that the emitted beam has passed through. This could be a weighted combination of signal corrections for one or more other locations. The found combinations can be based on four locations. The weighted combinations are 1 This can be obtained based on two locations on one beam and two locations on another beam.

[0029] Signal correction for a given location involves the location on the first beam from which that location originates and the second beam. It can be weighted according to the ratio of the distance between the location and the location on the map. Signal correction for a certain location is , how far that location is from the location on the first beam and the location of that location on the second beam It can be weighted according to the ratio of how far it is from the location on the beam, for example, The corrections from these locations are weighted. The distance is the first beam, the location being corrected, and the second. This can be considered along an equidistant circular arc from the transducer passing through beam 2.

[0030] Signal correction for a certain location is performed on a first beam from which the signal originates and a first beam from which the signal originates. It can be weighted according to the ratio of the distance between a second location on the beam and a certain location. The signal correction depends on how far that location is from a first location on the first beam, The weighting is determined by the ratio of how far that location is from the second location of the first beam. It can be erased, for example, corrections from those locations can be weighted. The distance is to the first beam This can be considered accordingly.

[0031] The distance values ​​for each location, such as pixels within the region of interest or within the substrate, are pre-calculated and stored. It is possible.

[0032] It is possible to calculate the speed of sound at various locations, such as pixels within the region of interest and even within the substrate.

[0033] The beam path passing through each boundary between locations such as pixels within the region of interest or within the substrate is, for example, For example, it can be calculated according to Snell's law of refraction.

[0034] Temperature can be mapped to each location, such as a pixel within a region of interest or substrate. A region or substrate can be assigned, for example, temperature contour lines perpendicular to the surface of the substrate.

[0035] This method includes, for example, passing through each location determined by the distance and speed within the zone. This may include determining the interval. In this method, the travel time passing through all locations along the route is calculated. The sum of these factors can reveal the net transit time.

[0036] This method determines that the result set is one of the shapes of the substrate and / or the weld groove and / or weld. The method includes one or more measured representations, and the method includes a representation of the measured shape and a modeled shape. Further comparisons include those between the displayed and the displayed measured shape. This ensures that the measured shape representation closely matches the modeled shape representation. If this is revealed, it may be stipulated that imaging of the region of interest should be accepted.

[0037] This method determines that the result set is one of the shapes of the substrate and / or the weld groove and / or weld. The method includes one or more measured representations, and the method includes a representation of the measured shape and a modeled shape. Further comparisons include those between the displayed and the displayed measured shape. This indicates that the measured shape representation does not adequately match the modeled shape representation. If this becomes clear, the temperature distribution within the volume of the substrate and / or weld at high temperatures will be compensated for. It may be stipulated that the temperature distribution to be used is predetermined.

[0038] This method provides a thermal model and uses the thermal model for at least a portion of the substrate. This includes generating a modeled temperature distribution location in a high-temperature state, and the substrate In part, it may be defined that volume is included. This method uses a substrate at high temperature to measure multiple locations This includes measuring the temperature and obtaining the measured temperature distribution location. It may be further specified that the location be compared to a modeled temperature distribution location.

[0039] This method compares the measured temperature distribution location with the modeled temperature distribution location. It is clear that the modeled temperature distribution does not adequately fit the measured temperature distribution. If this occurs, modify the thermal model and / or the modeled temperature distribution location, then It may be stipulated that further comparisons may be made.

[0040] This method compares the measured temperature distribution location with the modeled temperature distribution location. It was revealed that the modeled temperature distribution fits the measured temperature distribution well. If so, the properties of the ultrasonic waves emitted during passage through the substrate and / or at least a portion of the weld. It may be further specified that this includes calculating the following:

[0041] The first aspect of this disclosure, including other aspects of this disclosure, is described elsewhere in the document. This may include ability, possibility, or choice.

[0042] According to a second aspect of this disclosure, an apparatus for inspecting a welded joint, (a) When in use, it is provided in close proximity to the weld on the substrate to be inspected, and the substrate is subjected to pressure during inspection. A welding inspection device is provided that is subjected to high temperatures exceeding the ambient temperature by being heated, (b) Welding inspection equipment, (1) Welding by emitting ultrasonic waves into the volume of the base material and the weld area, (2) Receiving at least a portion of ultrasonic waves from the substrate and the welded part, thereby multiple It is compatible with obtaining the signal set, (3) The welding inspection device includes one or more processors, which include one or more signals The processor includes an input for a set of numbers, and the processor receives one or more inputs. It provides correction for multiple signal sets to provide corrected welding inspection data, and processes S applies a correction to the temperature within the volume of the substrate and / or weld at high temperatures. The device is provided.

[0043] The second aspect of this disclosure, including other aspects of this disclosure, is described elsewhere in the document. This may include features, possibilities, or options.

[0044] According to a third aspect of this disclosure, determining one or more dimensions of one or more substrates to be welded. It is a method, (i) A step of preparing a dimensional determination device that emits ultrasonic waves, (ii) The step of introducing one or more substrates to be welded into a dimensional determination device, (iii) Using ultrasound to perform one or more dimensional determinations on one or more substrates Step and, (iv) One or more further operations performed on one or more substrates The step of using the dimension determination, This provides a method.

[0045] This method may be part of a welding method and may include the step of preparing a welding apparatus. This could be an arc welding method. In this method, one or more substrates to be welded are introduced into the welding apparatus. This may include: One or more further actions may be part of the welding process, and / or it may be part of the post-welding process.

[0046] Ultrasound may be provided by one or more beams emitted from the device onto one or more substrates. Ultrasound can be provided by multiple beams. Ultrasound is provided by multiple beams spreading in an arc shape. It may be provided by the frame. The arc is at least 5°, and possibly at least 10°. In some cases, it may extend over at least 15°. The arc is less than 40°, for example. It may extend over a range of less than 30°, or in some cases less than 25°.

[0047] One or more dimensions of one or more substrates are the distance between a first surface and a second surface, such as thickness. It is possible. The thickness can be established in the direction of one or more beams of ultrasound. It can be established perpendicular to one or more surfaces of the substrate above.

[0048] A dimensional determination device can also be a welding inspection device. Ultrasound can be used for placement.

[0049] This method involves the step of using heating to raise the temperature of one or more substrates above the ambient temperature. It may include.

[0050] Before performing one or more dimensional determinations on one or more substrates, use heating to determine one or more dimensions. A step may be provided to raise the temperature of the substrate above the ambient temperature. After performing one or more dimensional determinations, heating is used to raise the temperature of one or more substrates to the ambient temperature. A step to make it even higher may be provided.

[0051] This method involves determining the dimensions of one or more substrates at multiple locations on the substrate, one or more times for each substrate. This may include performing the following: The method involves moving the dimensional determination device between multiple locations on the substrate. This may include, for example, rolling. Multiple locations on the substrate are the intended trajectory of the welding. and / or the trajectory formed by the weld can be traced.

[0052] One or more dimensions can be measured in terms of angles relative to the substrate. The angle relative to the substrate is the angle of the substrate. This can be considered as an angle with respect to one surface and / or a second surface. The angle is 90°+ -25°, for example, 90°+ / -20°, and in some cases, it could be 90°+ / -15°. .

[0053] One or more dimensions may be measured within an angular range relative to the substrate. This can be considered as a range of angles with respect to the first and / or second surface of the substrate. The angle range is 90° + / - a maximum of 25°, for example 90° + / - a maximum of 20°, depending on the case. It can be 90° ± 15°, and especially 90° ± 10°.

[0054] One or more dimensions perpendicular to one or more of the first and / or second surfaces of the substrate. It can be measured directly.

[0055] This method, for example, measures the temperature and / or temperature gradient within the substrate in each of the ultrasonic beam paths. This may include correcting the thickness measurement to account for the effect of the wavelength on ultrasound.

[0056] This method measures thickness while considering, for example, the effect of ultrasonic refraction in each ultrasonic beam path. This may include correcting the values.

[0057] This method may include providing a welding inspection device.

[0058] This method may include inspecting the welded joint using a welding inspection device. The law stipulates that inspection of welds must be conducted at one or more inspection locations on the substrate, and the inspection location must not exceed the ambient temperature. This may include being provided at a temperature by an inspection device.

[0059] One or more dimensions are inspected and / or inspected by a welding inspection device. This may include the dimensions of the base material along one or more paths passing through the material.

[0060] A device combining a dimensional determination device and a welding inspection device can perform dimensional determination of a base material, and After that, one or more welding inspections may be performed. A combination of a dimensioning device and a welding inspection device. The apparatus can perform one or more dimensional determinations, then one or more welding inspections, and further dimensional determinations. Further welding inspections may be performed in some cases. Dimension determination, welding inspection, dimensional determination, and A cyclical approach to welding decisions may be used.

[0061] This method uses one or more dimensional determinations in adjusting and / or correcting weld inspections. By doing so, one of the one or more further operations performed on one or more substrates This method may include using the above dimensional determination. This method may include adjusting the depth of field. The method may include adjustment of the image analysis area, for example, the image analysis area considered to be the subject of examination. The law involves one or more adjustments and / or corrections to the position determined by the welding inspection. This may include using dimensional determination. This method involves the depth of the weld into the substrate determined by welding inspection. Adjusting and / or correcting the dimensions may include using one or more dimensional determinations.

[0062] A third aspect of this disclosure, including other aspects of this disclosure, is described elsewhere in the document. This may include ability, possibility, or choice.

[0063] According to a fourth aspect of this disclosure, an apparatus for determining dimensions in welding, if necessary, Provided, this device, (i) Welding equipment as needed, (ii) A dimensional determination device, Equipped with, The elements of the dimension determination device include an ultrasonic transmitter and receiver, and the elements of the dimension determination device include a base A material contact surface is provided, and an element of the dimension determination device outputs a data signal to another element of the dimension determination device. The device provides, and another element of the dimensioning apparatus includes a dimensionality determination device that determines one or more dimensions of a substrate. The decision-maker provides one or more dimensions to another processor, and the other processor provides one or more bases Use one or more dimensions in one or more further actions performed in relation to the material.

[0064] This device may further include a welding inspection device. This device includes a dimensional determination device and a welding inspection device. It may be equipped with a combination with the device.

[0065] The decision-maker and another processor combine a dimensional determination device and a welding inspection device into a common system. It may be part of the data processing components.

[0066] The fourth aspect of this disclosure, including other aspects of this disclosure, is described elsewhere in the document. This may include ability, possibility, or choice.

[0067] According to a fifth aspect of the present invention, a method for inspecting a weld is provided, which is: (a) The step of preparing a welding inspection device in close proximity to the weld on the substrate to be inspected, (b) Emitting ultrasonic waves onto the substrate and welded area, and returning ultrasonic waves from the substrate and welded area Receive and thereby obtain multiple signal sets, including a first signal set and a second signal. The steps to take, (c) Processing at least the first signal set and the second signal set to obtain welding inspection data A step to provide, The first signal set includes a first sequence of first data elements, and the first of the first data elements The sequence extends to the first period, and each first data element consists of the first variable value of the variable and its first The first time period includes the first time value of the variable value, and the first period includes sub-periods of the first time, The second signal set includes a second sequence of the second data element, and the second of the second data element The sequence extends to a second period, and each second data element is the second variable value of the variable and its second The second time period includes the second time value of the variable value, and the second period includes a sub-period of the second time, The sub-periods of the first time and the sub-periods of the second time have the same time value range, step and, Processing of the first and second signal sets contributes to the welding inspection data. A set of corrected signals is determined, and the set of corrected signals is used for the first time sub-period and the second time sub-period. Includes a sub-period of the corrected time with the same time value range as the main period, and the first variable value and the second If the variable values ​​are in a predetermined relationship with respect to the time value, the first variable with respect to that time value Includes a numerical representation and a second variable value representation for that time value, and / or the first If the first variable value and the second variable value are not in a predetermined relationship with each other, then, for the time value, One representation of the first variable value for the interval value or the representation of the second variable value for the time value include.

[0068] The first variable value is, for example, the amplitude at a specific time value, and in particular, the representation of the amplitude of a set of rectified signals. It is possible.

[0069] The second variable value is, for example, the amplitude at a given time value, particularly the representation of the amplitude of a rectified signal set. It is possible.

[0070] The first period, the second period, and / or one or more further periods may be the same period. The first period, the second period, and / or one or more additional periods are the same period + / - 25. It may be %. The first period, the second period, and one or more further periods are ultrasonic beams. This could be the duration of the received signal for a wave or other wave.

[0071] The first time subperiod and / or the second time subperiod and / or one or more The sub-periods may be individual time values. The first time sub-period and / or the second time Intermediate subperiods and / or one or more further subperiods are small sets of individual time values. It is possible.

[0072] The first period can be considered as multiple different or partially overlapping first sub-periods. The second period can be considered as multiple different or partially overlapping second sub-periods. To obtain. Further periods are considered as multiple different or partially overlapping further time subperiods. This is possible. Multiple time subperiods can be paired with each other when they have the same time value.

[0073] The modified signal set is adapted according to this method, taking into account several time subperiods. It is possible.

[0074] This method represents the first variable value for its time value in the modified signal set. And the representation of the second variable value for that time value is the average of the first variable value and the second variable value. It is possible to define what is possible. This method, in the modified signal set, applies to its time value The representation of the first variable value and / or the representation of the second variable value for its time value are further It can be defined that, along with a given variable value, it can be the average of all variable values ​​within a given relationship. .

[0075] This method determines if the first variable value and the second variable value are within a predetermined threshold, and It can be defined that the second variable value and the second variable are in a predetermined relationship with each other. The properties of the predetermined threshold and The definition function and / or value may be variable. The threshold is set as the system gain increases. In cases where and / or noise in the signal set increases, it may change, especially if it increases. The threshold can change, and in particular can increase, as the interference level in the signal set increases. The properties and / or defining functions and / or values ​​may be determined by the calibration method. ru.

[0076] This method further involves the first variable value and the second variable value being further determined by a predetermined threshold, and in some cases the same threshold. If the value is outside the range, then the first variable value and the second variable value are not in a predetermined relationship with each other. The properties and / or defining functions and / or values ​​of a given threshold can be defined. Furthermore, a predetermined threshold is obtained when the system gain increases and / or the signal set It can change, and especially increase, when the internal noise increases. Furthermore, a predetermined threshold is a signal. As the interference level within the set increases, it can change, and in particular, it can increase.

[0077] This method further considers the case where the first variable value and the second variable value are not in a predetermined relationship with each other. One of the included expressions, selected from 1 and 2, has a lower variable value relative to its time value. It may be specified that the expression will have the following characteristics: This method further includes cases where the variable value being compared and the variable value being compared are not in a predetermined relationship with each other. One expression may be defined as an expression with a lower variable value for that time value. The limit variable value can be the lower limit variable value, which is considered as an absolute value.

[0078] A predetermined threshold can be determined relative to the amplitudes of the first and second variable values.

[0079] The amplitude can be a predetermined amplitude value. The amplitude is, for example, the predicted noise amplitude. This can be the predicted amplitude value. The amplitude is a defined percentage of the predicted noise amplitude. Or it could be a multiple.

[0080] The amplitude may be the observed amplitude.

[0081] The amplitude may relate to the maximum amplitude observed for a set of variable values ​​where no interference is assumed.

[0082] The amplitude is the entire set of variable values ​​under consideration, e.g., the first variable value, the second variable value, and one Across all of the above further variable values, the minimum value observed over a certain time or period is related to the following: The amplitude can be the minimum value plus a coefficient. This amplitude is considered an interference-free threshold. It is possible.

[0083] The threshold can be the difference between the observed value and the analyzed value. The observed and analyzed values ​​can be amplitudes. The analysis values ​​include the entire set of variable values ​​under consideration, for example, the first variable value, the second variable value, and one The values ​​obtained from the minimum observed over a certain time or period across all of the above further variable values. It is possible. The analytical value may be the minimum value plus a coefficient.

[0084] This method involves acquiring one or more additional signal sets as part of multiple signal sets. It may also include the following.

[0085] This method processes one or more additional signal sets to provide welding inspection data. This could further include:

[0086] This method involves one or more further signal sets containing a further series of further data elements. This further includes all of the above, and each of the further series of further data elements extends over a further period. And each of the further data elements is the further variable value of the variable and the further variable value It includes a range of time values, and each of these further periods includes further time sub-periods.

[0087] This method further includes at least the first time subperiod and / or the second time subperiod. It may also include one of several further time subperiods that share the same time range.

[0088] This method further determines the modified signal set that contributes to the welding inspection data. It may further include one or more or all of the processes of the set.

[0089] This method further modifies the time range to include the same time values ​​as further time subperiods. Changes that extend to the main period and to one or both of the first and second time sub-periods. It may include a set of signals.

[0090] This method further involves the first variable value, the second variable value, and the further variable value relating to each other in a predetermined manner. If it is within the section, for the time value, the representation of the first variable value for that time value and that time A representation of a second variable value for the value, and a representation of one further variable value for that time value. It may include a modified signal set.

[0091] This method alternatively or additionally involves the first variable value, the second variable value, and the third variable value interacting with each other. If the relationship is not as specified, the expression of the first variable value for that time value or a representation of a second variable value for that time value, or a representation of a further variable value, It may include a set of correction signals, one of which is one of the following.

[0092] This method alternatively or additionally considers the time value as a first variable for that time value. Representation of the value, representation of the second variable value for that time value, further variable value for that time value It further includes a modified signal set containing two of one representations of the two variable values, and the two variable values ​​are relative to each other. They are in a predetermined relationship.

[0093] This method alternatively or additionally considers the time value as a first variable for that time value. Representation of the value, representation of the second variable value for that time value, further variable value for that time value It may further include a set of modified signals containing only one of the one representations of, and the variable values ​​are relative to each other. They are not within the specified relationship.

[0094] This method further involves the modified signal set, each being a first time subperiod and / or The second time subperiod and / or one or more further time subperiods have the same time values ​​as a range. It may be specified that compilation will be performed from multiple modified time subperiods.

[0095] A fifth aspect of the present invention, including other aspects of the present disclosure, is described elsewhere in this document. This may include any of the following features, additions, or possibilities.

[0096] According to a sixth aspect of the present invention, an apparatus for inspecting a weld is provided, which apparatus (a) A welding inspection device, a. Ultrasonic transmitters for substrates and welds in use, and for substrates and welds in use A receiver for ultrasonic waves returning from the unit, the receiver is connected to a processor and the first signal The system provides the processor with a set of signals and a second signal, and multiple acquired signal sets, including the first signal set and the second signal. A transmitter and receiver, b. Multiple acquired signal sets, including the first signal set and the second signal set. A processor adapted to receive, Equipped with welding inspection equipment, The first signal set includes a first sequence of first data elements, and the first of the first data elements The continuity extends to a first time period, and each first data element is the first variable value of the variable and its first The first time period includes the first time value of the variable value, and the first time period includes the first time sub-period, and the second The signal set includes a second sequence of the second data element, and a second sequence of the second data element. This extends to a second time period, and each second data element is the second variable value of the variable and its second change The second time value of the numerical value is included, and the second time period includes the second time sub-period, and the first time The inter-subperiod and the second time subperiod have the same time value range. The processor uses the first and second signal sets to process the welding inspection data. Adapted to determine the contributing modified signal set, The processor uses the same time values ​​as the first and second time sub-periods. For the modified time subperiod, the modified signal set is the first variable value and the second If the variable values ​​are in a predetermined relationship with respect to the time value, the first variable with respect to that time value Includes a numerical representation and a second variable value representation for that time value, and / or the first If the first variable value and the second variable value are not in a predetermined relationship with each other, then, for the time value, One representation of the first variable value for the interval value or the representation of the second variable value for the time value It will be adapted to specify that it must include.

[0097] A sixth aspect of the present invention is a feature, addition, or This may include any of the possibilities, and other forms of disclosure.

[0098] According to a seventh aspect of this disclosure, a welding method, (a) Prepare welding equipment, (b) Prepare multiple sensor types, (c) Define a first set of welding conditions for the welding method, (d) Introducing one or more substrates to be welded into the welding apparatus, (e) Perform welding of one or more substrates, (f) Acquiring data from multiple sensor types during welding, (g) data obtained from at least two selected sensor types out of multiple sensor types To provide correlations between the data, (h) One of the data obtained from at least one of two selected sensor types The above data points are used with at least two selected sensor types and one or more other data points. Synchronizing with the data point, (i) Data acquired from multiple sensor types, reference data of one or more sensor types Comparing it to Ta, (j) Based on one or more comparisons, the weld is of acceptable quality or unacceptable quality. This includes determining whether it is of quality, (k) If the quality of the weld is unacceptable, the method shall perform one or more actions. Including, A welding method is provided.

[0099] A seventh aspect of this disclosure relates to the inspection of welds, as described in the first and / or second aspects of this disclosure. This may include any of the features, possibilities, or additions described in that manner.

[0100] This method could be arc welding.

[0101] This method specifies that the correlation is both a temporal correlation and a spatial correlation. obtain.

[0102] This method can define that the correlation is a temporal correlation.

[0103] This method combines data from at least two selected sensor types with one of the data types The occurrence time of the above data points may be specified. This method further includes correlation The person in charge determines the generation of data points in the data from at least two selected sensor types. It may be possible to stipulate that the times coincide.

[0104] This method involves temporally correlating data from at least two selected sensor types. It may be stipulated that the time provided by the staff includes multiple hours.

[0105] This method uses data from at least two selected sensor types to access all data within the data. This method may specify that the occurrence time of each data point is included. The data from the selected sensor type is subdivided into each range (such as bandwidth) within It may be specified that the occurrence time of all data points be included.

[0106] This method provides a correlation between two or more selected sensor types. The time of occurrence is determined based on the timestamp introduced into the data from the selected sensor type. It is possible to define what is advantageous. This method obtains the occurrence time directly from the timestamp at that moment. For example, it may be stipulated that a timestamp is applied simultaneously with the time of occurrence. This method, for example, generates a timestamp when the time of occurrence has reached the timestamp. When the time of occurrence is calculated from the elapsed time since the timestamp, It may be defined that the data is acquired by the time of the data point and the duration of the overtime period.

[0107] The timestamp is used by the processor and / or clock generator, for example. The buck signal can be applied by the processor to which it is provided.

[0108] This method can define that the correlation is a spatial correlation.

[0109] This method uses data from at least two selected sensor types to determine the spatial correlation. It may be specified that this includes multiple locations where it is provided.

[0110] This method uses data from at least two selected sensor types, and one of the data is This method may specify that the occurrence locations of the above data points are included. Furthermore, the correlation is , the location of data points in the data from at least two selected sensor types It can be stipulated that they must be identical.

[0111] This method ensures that data from at least two selected sensor types is available in the data. This method may specify that the origin of each data point is included. Data from the selected sensor type is divided into each range (such as bandwidth) within which the data is subdivided. It may be specified that the origin location of all data points be included.

[0112] This method provides a correlation between two or more selected sensor types. The location of the event is obtained based on the position stamp introduced into the data from the selected sensor type. This method may stipulate that the location of occurrence is obtained directly from the location stamp at that time. For example, it may be stipulated that a location stamp is applied simultaneously with the location of occurrence. The method determines the location of occurrence based on the elapsed time from the location stamp and the time of the data point. It can be defined that a stamp is obtained. For example, when a location stamp is generated and the location of generation is reached From the elapsed time since the location stamp, and / or the location stamp when the originating location was reached. The location of the event is calculated from the elapsed distance from the point.

[0113] Positional correlation refers to the position relative to one or more substrates being welded, for example, on one or more substrates. It may be relative to the position. Multiple positional correlations with respect to one or more substrates are the same It may be provided at times. The position is relative to the location where welding is performed, for example, relative to the starting point of the welding. A known relationship may exist with respect to the end point of the weld and / or the end point of the weld.

[0114] Positional correlation is relative to the position related to the welding equipment, for example, the position on the welding equipment. It may be such. Multiple positional correlations may be provided simultaneously to the welding apparatus. One or more positions on the welding device may be close to the welding electrode. These positions may, in some cases, be in a position where welding is actually performed. A known relationship to the location where the welding is performed, for example, the start point and / or end point of the welding. They may have a known relationship, and that known relationship may be separation between the electrode and the substrate.

[0115] A position relating to one or more substrates to be welded, for example, a position on one or more substrates The positional correlation with the position of the welding equipment, for example, the position relative to the position on the welding equipment. Both a direct correlation and a direct correlation can be provided.

[0116] Location stamps can be applied by a processor. The processor can apply one or more location signals The position signal can be received by one or more position sensors and / or position indicators. Therefore, it can be emitted. The position sensor can be a vision-based sensor such as a camera. The indicators include, for example, the occurrence of movement and / or the direction of movement and / or encoding. A motion detection-based position indicator, such as an incremental encoder, provides the position of the object. It's possible. Tracking signal-based location indicators are radio frequency identification-based systems. In some cases, one or more radio frequency identification tags may be placed on the substrate and / or welding equipment. To possess.

[0117] This method uses at least three, and possibly at least four, of multiple sensor types. In some cases, the small difference between data acquired from at least five selected sensor types. It may include providing at least one type of correlation for the selected sensor type. At least two of them may provide the same type of correlation. The same type of correlation can be provided between all of the pairs.

[0118] This method uses at least three, and possibly at least four, of multiple sensor types. In some cases, the small difference between data acquired from at least five selected sensor types. It may include providing at least two different types of correlations. A correlation between the same different types of ipu can be provided between at least two types of ipu. The same different types of correlations may be provided among all selected sensor types. This type can include both temporal and spatial correlations.

[0119] This method involves selecting one or more data points in the data acquired from each selected sensor type. This may include synchronizing the data. This method involves at least two, and possibly at least two. Data obtained from each of three, and possibly at least four, selected sensor types. This may include synchronizing one or more data points within a database using the same correlation. .

[0120] This method involves data within data acquired from at least one of the selected sensor types. Obtain at least 10% of the points from at least one other of the selected sensor type. This may include synchronizing at least 10% of the data points within the data.

[0121] This method uses at least one of several sensor types, and possibly a selected sensor. At least one of the sensor types is a voltage sensor, current sensor, welding arc noise emission sensor, The system is to be selected from welding topology sensors, welding image sensors, and ultrasonic image sensors. It can be determined.

[0122] A voltage sensor can be one of several sensor types. A voltage sensor is selected. It could be one of the sensor types.

[0123] A current sensor can be one of several sensor types. The current sensor is selected. It could be one of the sensor types.

[0124] A welding arc noise emission sensor can be one of several sensor types. The sound emission sensor may be one of the selected sensor types. (e.g., welding arc sound emission sensor) The sensor can be an acoustic sensor such as a microphone. A welding arc sound emission sensor is a welding sound sensor. Emissions generated from the arc forming the part, and / or from the interaction between the arc and the substrate. Sensitive to emissions generated and / or emissions arising from the interaction between the arc and the shielding gas. It is possible. The welding arc noise emission sensor detects the welding speed and / or welding location sidewall arc. To notify of characteristics related to welding site sidewall melting and / or shielding gas flow rate. We can provide data.

[0125] A welding topology sensor can be one of several sensor types. The sensor can be one of the selected sensor types. The welding topology sensor is a radiation sensor. It can be a line-based sensor. A welding topology sensor can be an image-based sensor. A welding topology sensor may be a laser sensor. The welding topology sensor is a substrate and / or sensitive to radiation irradiated onto the weld and reflected by the substrate and / or the weld. It is possible. The applied radiation and / or reflected radiation are the welding topology sensor. It is focused by. The welding topology sensor is related to the shape of the substrate and / or the weld. We can provide information regarding its characteristics.

[0126] A welding image sensor can be one of several sensor types. It could be one of the selected sensor types. Welding image sensors are camera-based. It can be a sensor. The welding imaging sensor can determine the size of the welding pool, the shape of the welding pool, Temperature of the weld pool, presence of deposits on the weld and / or substrate, weld and / or Pattern of deposits on the substrate, welds and / or shape of deposits on the substrate, welds and / or may provide data indicating characteristics related to the presence of anomalies on the substrate.

[0127] An ultrasonic imaging sensor can be one of several sensor types. This could be one of the selected sensor types.

[0128] One or more, or all, sensor types may provide raw datasets. This method allows you to do so. This includes processing raw datasets to generate processed datasets.

[0129] When determining the type of sensor, raw and / or processed data may be considered. In determining the individual sensor type, the data points are a known set of data points (for example) Consider how far it is from the known data point set that indicates acceptable welding. It is conceivable. A known set of data points is, for example, the average of a known set of data points. It can be represented by a single element.

[0130] A known set of data points includes the sensor type and its signal value and / or data value. This can be represented by a principal component analysis (PCA) model.

[0131] The distance could be the Mahalanobis distance.

[0132] Distance can be represented as an outlier score.

[0133] The decision may be made based on the distance of the data point relative to a threshold, for example, a threshold distance. The decision is made based on whether the data point is within the boundaries of a known set of data points, or the boundary of a known set of data points. This can be done based on whether it is outside the bounds.

[0134] This decision is based on the data point or signal value, the sensor type and its signal value and / or This is done by comparing the data values ​​with previously developed principal component analysis (PCA) models. This can be done using a Mahalambis distance novelty detection model.

[0135] Processing raw datasets may include applying denoising algorithms. This can be processed by dividing it into a series of bandwidths.

[0136] The decision is made using a model such as a PCA model with acceptable performance for subsequent data points. Decisions may be made based on and / or signals.

[0137] Data points and / or data sets determined to be acceptable welds, and / or and / or data points and / or data sets determined to be unacceptable welds are added, the model and / or the average of the model can be recalculated. One or more or all sensor types may provide raw data sets, and the method includes processing the raw data set to generate a processed data set, which processing uses a model. The method may include using different deformations of the model for different parts of the welding process. In particular, the deformation of the model can be used for one or more welding paths, and different models can be used for one or more additional welding paths within the overall weld generation process. Different deformations of the model can be used for each welding path in the overall weld generation process. For repair welds within the weld generation process, for example, when a part of the weld is removed and then rewelded, one or more further different deformations can be used. For each welding path or weld bead in the entire welding process that is the target of repair welding, another further different deformation of the model can be used.

[0138] One or more or all sensor types may provide raw data sets, and the method includes processing the raw data set to generate a processed data set, which processing uses a model. The method may include using different deformations of the model for different parts of the welding process. In particular, the deformation of the model can be used for one or more welding paths, and different models can be used for one or more additional welding paths within the overall weld generation process. Different deformations of the model can be used for each welding path in the overall weld generation process. For repair welds within the weld generation process, for example, when a part of the weld is removed and then rewelded, one or more further different deformations can be used. For each welding path or weld bead in the entire welding process that is the target of repair welding, another further different deformation of the model can be used. One or more or all sensor types may provide raw data sets, and the method includes processing the raw data set to generate a processed data set, which processing uses a model. The method may include using different deformations of the model for different parts of the welding process. In particular, the deformation of the model can be used for one or more welding paths, and different models can be used for one or more additional welding paths within the overall weld generation process. Different deformations of the model can be used for each welding path in the overall weld generation process. For repair welds within the weld generation process, for example, when a part of the weld is removed and then rewelded, one or more further different deformations can be used. For each welding path or weld bead in the entire welding process that is the target of repair welding, another further different deformation of the model can be used. One or more or all sensor types may provide raw data sets, and the method includes processing the raw data set to generate a processed data set, which processing uses a model. The method may include using different deformations of the model for different parts of the welding process. In particular, the deformation of the model can be used for one or more welding paths, and different models can be used for one or more additional welding paths within the overall weld generation process. Different deformations of the model can be used for each welding path in the overall weld generation process. For repair welds within the weld generation process, for example, when a part of the weld is removed and then rewelded, one or more further different deformations can be used. For each welding path or weld bead in the entire welding process that is the target of repair welding, another further different deformation of the model can be used. One or more or all sensor types may provide raw data sets, and the method includes processing the raw data set to generate a processed data set, which processing uses a model. The method may include using different deformations of the model for different parts of the welding process. In particular, the deformation of the model can be used for one or more welding paths, and different models can be used for one or more additional welding paths within the overall weld generation process. Different deformations of the model can be used for each welding path in the overall weld generation process. For repair welds within the weld generation process, for example, when a part of the weld is removed and then rewelded, one or more further different deformations can be used. For each welding path or weld bead in the entire welding process that is the target of repair welding, another further different deformation of the model can be used. One or more or all sensor types may provide raw data sets, and the method includes processing the raw data set to generate a processed data set, which processing uses a model. The method may include using different deformations of the model for different parts of the welding process. In particular, the deformation of the model can be used for one or more welding paths, and different models can be used for one or more additional welding paths within the overall weld generation process. Different deformations of the model can be used for each welding path in the overall weld generation process. For repair welds within the weld generation process, for example, when a part of the weld is removed and then rewelded, one or more further different deformations can be used. For each welding path or weld bead in the entire welding process that is the target of repair welding, another further different deformation of the model can be used. One or more or all sensor types may provide raw data sets, and the method includes processing the raw data set to generate a processed data set, which processing uses a model. The method may include using different deformations of the model for different parts of the welding process. In particular, the deformation of the model can be used for one or more welding paths, and different models can be used for one or more additional welding paths within the overall weld generation process. Different deformations of the model can be used for each welding path in the overall weld generation process. For repair welds within the weld generation process, for example, when a part of the weld is removed and then rewelded, one or more further different deformations can be used. For each welding path or weld bead in the entire welding process that is the target of repair welding, another further different deformation of the model can be used. One or more or all sensor types may provide raw data sets, and the method includes processing the raw data set to generate a processed data set, which processing uses a model. The method may include using different deformations of the model for different parts of the welding process. In particular, the deformation of the model can be used for one or more welding paths, and different models can be used for one or more additional welding paths within the overall weld generation process. Different deformations of the model can be used for each welding path in the overall weld generation process. For repair welds within the weld generation process, for example, when a part of the weld is removed and then rewelded, one or more further different deformations can be used. For each welding path or weld bead in the entire welding process that is the target of repair welding, another further different deformation of the model can be used. One or more or all sensor types may provide raw data sets, and the method includes processing the raw data set to generate a processed data set, which processing uses a model. The method may include using different deformations of the model for different parts of the welding process. In particular, the deformation of the model can be used for one or more welding paths, and different models can be used for one or more additional welding paths within the overall weld generation process. Different deformations of the model can be used for each welding path in the overall weld generation process. For repair welds within the weld generation process, for example, when a part of the weld is removed and then rewelded, one or more further different deformations can be used. For each welding path or weld bead in the entire welding process that is the target of repair welding, another further different deformation of the model can be used. One or more or all sensor types may provide raw data sets, and the method includes processing the raw data set to generate a processed data set, which processing uses a model. The method may include using different deformations of the model for different parts of the welding process. In particular, the deformation of the model can be used for one or more welding paths, and different models can be used for one or more additional welding paths within the overall weld generation process. Different deformations of the model can be used for each welding path in the overall weld generation process. For repair welds within the weld generation process, for example, when a part of the weld is removed and then rewelded, one or more further different deformations can be used. For each welding path or weld bead in the entire welding process that is the target of repair welding, another further different deformation of the model can be used. One or more or all sensor types may provide raw data sets, and the method includes processing the raw data set to generate a processed data set, which processing uses a model. The method may include using different deformations of the model for different parts of the welding process. In particular, the deformation of the model can be used for one or more welding paths, and different models can be used for one or more additional welding paths within the overall weld generation process. Different deformations of the model can be used for each welding path in the overall weld generation process. For repair welds within the weld generation process, for example, when a part of the weld is removed and then rewelded, one or more further different deformations can be used. For each welding path or weld bead in the entire welding process that is the target of repair welding, another further different deformation of the model can be used. <​​​​​​​​​​​​​​​​

[0142] This method obtains data points from at least two selected sensor types when an event occurs. It stipulates that the information be displayed to the user in an aligned manner with respect to the interval and / or location of occurrence. obtain.

[0143] This method uses data from two or more of the selected sensor types and at least two Correlation data from sensor types is displayed to the user and / or saved. It can be defined.

[0144] This method determines whether a weld is of acceptable or unacceptable quality. It may be specified that one or more actions performed are displayed and / or saved. .

[0145] This method takes at least four selected sensor types from among multiple sensor types. It may be specified that this includes providing correlations between the obtained data.

[0146] This method may include inspection of the entire weld. This method may include, for example, inspection after each pass and the next pass. This may include inspection of multi-pass welds before all passes are completed, such as before the final pass.

[0147] The welding equipment can be mounted on any autonomous automatic placement system. The welding equipment guide It can be mounted on a rail system or a column and boom system. Welding equipment is, for example, For example, it can be attached to a robot arm such as a multi-axis arm. The welding equipment is an arc welding machine. could be.

[0148] The welding inspection device can be one of several sensor types. It can be one of the specified sensor types. The welding inspection device can use ultrasonic waves, for example, phase door array ultrasonic waves.

[0149] The welding inspection device can be attached to any autonomous automatic placement system. The welding device can be attached to a guide rail system or a column and boom system. The welding device can be attached to a robotic arm, such as a multi-axis arm for example. The welding inspection device can be a phased door array ultrasonic transducer. The welding inspection device can include an ultrasonic transmitter and a receiver and can have a base material contact surface. The welding inspection device can physically contact the inspection location and the base material contact surface can physically contact the inspection location.

[0150] The method can include rolling the welding inspection device, for example its base material contact surface, on the surface of the base material, for example on one side but in some cases parallel to the weld, if possible.

[0151] The welding inspection device can be provided with an internal cooling device. The method can include supplying a coolant to the welding inspection device and / or removing the coolant from the welding inspection device.

[0152] The method can further stipulate that at least one sensor type out of a plurality of sensor types, in some cases at least one out of the selected sensor types, is part of the welding inspection device and the method includes the step of inspecting the weld using the welding inspection device The method can stipulate that it includes the step of inspecting the weld using the welding inspection device and determining the presence or absence of welding defects at the welding location, especially at a continuous series of locations forming the weld. This method may include a step of determining the characteristics of the weld, including the shape and / or the weld. One of the following relative to the length: size, location, type of defect, shape of defect, or location of defect It may be stipulated that the above may be included.

[0153] This method may include real-time determination of the presence or absence of welding defects at the welding site. This method ensures that the welding is completed within 100 milliseconds, and in some cases 50 milliseconds, after the welding equipment leaves the welding area. Determine the presence or absence of welding defects at the welding site within milliseconds, and in some cases within 20 milliseconds. This may include comparing one or more characteristics with one or more criteria. Furthermore, whether a weld with a defect in a certain location meets the welding standards, or whether it violates the welding standards. It may be stipulated that this may include a determination of whether the condition is not met.

[0154] This method determines whether a weld with a defect in a certain location meets the welding standards or if the welding standards are met This may include determining in real time whether the condition is not met. This method involves the welding equipment welding Within 100 milliseconds of leaving the location, in some cases within 50 milliseconds, in some cases Within 20 milliseconds, determine whether a defective weld in a certain location meets the welding standards or if it is melted. This may include determining whether the criteria are not met.

[0155] In this method, if the weld meets the welding standards, a welding record is created and saved. This includes the location of the defect relative to the shape of the weld, and / or the location of the defect relative to the length of the weld. This method can define what can be seen. It may be further specified that the data must be included. In this method, the welded joint must meet the welding standards. If not, it may be stipulated that one or more repair steps be applied to the weld.

[0156] This method uses at least one sensor type from among multiple sensor types, and in some cases The method is such that at least one of the selected sensor types is a welding condition sensor, The procedure may include a step of inspecting the welding condition using a welding condition sensor.

[0157] This method determines one or more parameters of the weld when the weld is formed. The procedure may be defined to include a step of inspecting the contact condition.

[0158] This method involves comparing one or more parameters with one or more control parameters, and further This may include provisions for determining whether the risk level of welding defects has been exceeded.

[0159] This method involves one or more operations that change the welding conditions from a first set of welding conditions of the welding method. It may be specified that this includes. This method allows a change in welding conditions from a first set of welding conditions to weld This method may include stopping and / or warning the operator. Changing the welding conditions from the first welding condition set will return the welding conditions to the first welding condition set. This includes and / or changing the welding conditions to a second set of welding conditions. obtain.

[0160] The welding method may further include inspection of the weld formed by the welding method. The welding method is, (i) A step of providing a welding inspection device, (ii) The step of using heating to raise the temperature of one or more substrates above the ambient temperature. , (iii) Using a welding apparatus, weld one or more substrates at a temperature exceeding the ambient temperature. The steps to take, (iv) A step of inspecting the welded part using a welding inspection device, (v) Inspection of welds is performed at one or more inspection locations on the substrate, and the inspection location is at an ambient temperature Steps performed by the inspection device at temperatures exceeding the limit, This may further include:

[0161] The seventh aspect of this disclosure, including other aspects of this disclosure, is described elsewhere in the document. This may include ability, possibility, or choice.

[0162] According to the eighth aspect of this disclosure, an apparatus for monitoring welding, (a) Multiple inputs for data from multiple sensor types, (b) One or more processors, one of the one or more processors teeth, a. Receive input, b. Select at least two sensor types from among multiple sensor types. The acquired data is processed, and at least two selected sensors are selected from among multiple sensor types. Apply the correlation between data obtained from the sample, c. Using correlation, at least two selected sensors from multiple sensor types The data acquired from the sensor types is processed, and at least two selected sensor types are processed. One or more data points within the data obtained from one source are selected from at least two selected data points. Synchronized with one or more data points in the data acquired from another sensor type. can One or more processors, (c) One or more outputs for the processed data, To provide an apparatus that includes the following features.

[0163] The processor further processes multiple sensors, such as one or more of the selected sensor types. Data from one or more types and reference data from one or more sensor types It can provide a comparator for receiving and comparing data.

[0164] The comparator determines, based on the compared data, whether the weld is of acceptable quality or not. It can output a determination of whether the quality is acceptable. If the welding quality is determined to be unacceptable, the device This may further provide control signals to trigger one or more actions by the control unit. ru.

[0165] The apparatus, for example, is for receiving a first set of welding conditions for a welding method and / or Or one or more actions by the control unit, such as modifying the first welding condition set or ending the welding. It may further include a control unit for receiving control signals to trigger an operation.

[0166] The apparatus may be defined to include a welding inspection device for determining one or more characteristics of a defect. .

[0167] The device uses at least one of several sensor types, and possibly a selected sensor. It may provide input from at least one of the following types, sensor type and / or select The selected sensor types are voltage sensors, current sensors, welding arc noise emission sensors, and welding topography sensors. Selected from log sensors, welding imaging sensors, and ultrasonic imaging sensors. .

[0168] A voltage sensor can be one of several sensor types. A voltage sensor is selected. It could be one of the sensor types.

[0169] A current sensor can be one of several sensor types. The current sensor is selected. It could be one of the sensor types.

[0170] A welding arc noise emission sensor can be one of several sensor types. A sound emission sensor is one of the selected sensor types. A welding arc sound emission sensor is a It could be an acoustic sensor such as an 'Iku'. A welding arc sound emission sensor is a sensor that detects the arc forming the weld. Emissions generated from and / or from the interaction between the arc and the substrate and / or It can be sensitive to emissions generated from the interaction between the arc and the shielding gas. Welding arc noise The emission sensor detects the welding speed and / or the sidewall arc at the welding location and / or the sidewall at the welding location. It may provide data that indicates characteristics related to melting and / or shielding gas flow rates.

[0171] A welding topology sensor can be one of several sensor types. The sensor can be one of the selected sensor types. The welding topology sensor is a radiation sensor. It can be a line-based sensor. A welding topology sensor can be an image-based sensor. A welding topology sensor may be a laser sensor. The welding topology sensor is a substrate and / or sensitive to radiation irradiated onto the weld and reflected by the substrate and / or the weld. It is possible. Irradiated radiation and / or reflected radiation are used in welding topology sensors. The welding topology sensor can be focused on the shape of the substrate and / or weld. This may provide information regarding related characteristics.

[0172] A welding image sensor can be one of several sensor types. It could be one of the selected sensor types. Welding image sensors are camera-based. It can be a sensor. The welding imaging sensor can determine the size of the welding pool, the shape of the welding pool, Temperature of the weld pool, presence of deposits on the weld and / or substrate, weld and / or Pattern of deposits on the substrate, welds and / or shape of deposits on the substrate, welds and / or may provide data indicating characteristics related to the presence of anomalies on the substrate.

[0173] An ultrasonic imaging sensor can be one of several sensor types. This could be one of the selected sensor types.

[0174] The device further includes a first comparator that receives one or more characteristics and compares them with one or more criteria. Further comparators are provided, and the first comparator determines whether a defective weld meets the welding standard. Further outputting a first decision on whether or not the welding standards are met. The device can provide, and can provide an apparatus. The apparatus can further provide one of the welds when a weld is formed. A second comparator for receiving the above parameters and comparing them with one or more control parameters. A comparator is provided that includes a second comparator that checks whether the risk level of welding defects is exceeded. The apparatus may be defined to output a second decision regarding the risk level of welding defects. If it exceeds this limit, the control signals provided by the device are sent to the controller, and one or more The operation can be triggered, and one or more operations are the first welding of the welding method. Changing welding conditions from a set of welding conditions, for example, stopping welding, and / or The purpose is to warn the operator.

[0175] The eighth aspect of this disclosure, including other aspects of this disclosure, is described elsewhere in the document. This may include ability, possibility, or choice.

[0176] According to the ninth aspect of this disclosure, a welding method, (a) Prepare welding equipment, (b) Prepare multiple sensor types, (c) Define a first set of welding conditions for the welding method, (d) Introducing one or more substrates to be welded into the welding apparatus, (e) Perform welding of one or more substrates, (f) Acquiring data from multiple sensor types during welding, (g) Data acquired from multiple sensor types, reference data from one or more sensor types Comparing it to Ta, (h) Based on one or more comparisons, the weld is of acceptable quality or unacceptable quality. Determining whether it is of quality, Includes, (i) The acquired data is processed by processing data from multiple sensor types. The acquired data, obtained for at least one of multiple sensor types, Processed using a computer model, this method applies to different parts of the welding process. Including the use of different model variations, A welding method is provided.

[0177] In particular, the deformation of the model can be used for one or more welding passes, and another model can be used for the overall welding. It can be used for one or more additional welding passes within the manufacturing process. Different deformations of the model are used overall. It can be used in each welding pass of a typical welding process. One or more further different variations are, for example, For example, it can be used for repair welding within the welding process, such as removing a portion of the weld and then re-welding it. It may be used for each weld pass or weld bead in the overall welding process that is subject to repair welding. For D, yet another, yet another, different variation of the model may be used.

[0178] The model could be an acoustic signal processing model. The model could be a voltage and / or current and / or it could be a power processing model such as output power.

[0179] A model can consist of multiple different data sources or submodels.

[0180] The ninth aspect of this disclosure, including other aspects of this disclosure, is described elsewhere in the document. This may include ability, possibility, or choice.

[0181] According to a tenth aspect of this disclosure, an apparatus for monitoring welding, (a) Multiple inputs for data from multiple sensor types, (b) One or more processors, one of the one or more processors teeth, a. Receive input from at least one of multiple sensor types, b. Process the data to obtain information for at least one of multiple sensor types. The data is provided, the processor has access to the computer model, and this method is dissolved This includes using different deformations of the model for different parts of the tangent process. Processor and (c) One or more outputs for the processed data, A device equipped with the following features.

[0182] The tenth aspect of this disclosure, including other aspects of this disclosure, is described elsewhere in the document. This may include features, possibilities, or options.

[0183] According to the eleventh aspect of this disclosure, a welding method, (a) Prepare welding equipment, (b) Prepare multiple sensor types, (c) Introducing one or more substrates to be welded into the welding apparatus, (d) Perform welding of one or more substrates, (e) Acquiring data from multiple sensor types during welding, (f) Data acquired from multiple sensor types, reference data of one or more sensor types Comparing it to Ta, (g) Determining whether the weld is of acceptable quality or unacceptable quality. and, (h) The decision considers combined data from at least two of the multiple sensor types. Including consideration, A welding method is provided.

[0184] This method involves forming combined data, and then making a decision considering the combined data. It can be stipulated that it will be carried out.

[0185] This method involves data obtained from one or more of several sensor types and the sensor type We compare the weld to standard data from one or more of our samples to determine if the weld is of acceptable quality or not. It can offer the ability to first determine whether something is of quality.

[0186] This method uses combined data to determine whether the weld is of acceptable quality or unacceptable quality. It may be stipulated that determining whether something is of quality is a second decision, separate from the first decision. When determining whether a weld is of acceptable or unacceptable quality, the second decision The outcome may take precedence over the first decision.

[0187] Decisions based on combined data include steps based on neural networks. To obtain, integrated data-based decisions include a decision engine that supports machine learning. To obtain. Decisions based on integrated data may include the use of models.

[0188] The combined data is processed using data from only one sensor type for each decision. For one or more decisions, separate decisions on acceptable and / or unacceptable welds. It can be used for fixed purposes.

[0189] Combined data-based decisions use data from one sensor type for each decision. Decisions made using only this method cannot be made, and / or have an acceptable level of interaction. Multiple errors prevent a decision from being made, and / or allow welding and / or fail to allow welding. When different decisions are made regarding welding, what constitutes an acceptable weld and / or an unacceptable weld? We can provide decisions regarding welding.

[0190] Decisions based on combined data are close to the threshold, close to the threshold distance, and / or acceptable. Data points close to the distance of indicators for possible and / or unacceptable welds. It can be used for data points and / or a series of data. If a point is far from the threshold, threshold distance, or distance, a decision is made based on combined data. It cannot be used.

[0191] The combined data is related to one or more sensor types that are applied during welding. It could be from. Combined data is less than 1 second after welding at a certain location, for example, 1 / 1 second after welding. This can be obtained from one or more sensor types related to welding observation performed in less than 0 seconds. The combined data includes data from one or more sensor types related to the NDT of welding, and / Or, data related to welding observations performed more than one minute after welding at a certain location may be included. It is possible that this may not happen.

[0192] The neural network provides the labeled data to the neural network. It can be trained by doing so. Labeled data is data points This could be a series of data points, which may indicate acceptable welding and / or unacceptable welding. This may indicate whether it matches the welding. Labeled data This may be a tap point and / or a series of data points, and may be an acceptable weld or an unacceptable non- With respect to certain properties that may give rise to possible welds, acceptable welds and / or unacceptable welds. This could indicate whether or not it matches the welding.

[0193] Neural networks can be trained by providing supervised learning. A neural network uses labels, for example, data points and / or A series of data points indicates either an acceptable weld or / or an unacceptable weld. Supervised learning can be provided through operator-based input, such as whether or not something is being done.

[0194] Training using labeled data and / or supervised learning is calibrated. It may be provided during the ergonomics stage and / or during production welding.

[0195] Neural networks utilize existing data, especially libraries of labeled data. It can be trained from. The library can be supplied to the neural network classifier. ru.

[0196] Labeled data and / or supervised learning and / or libraries are multiple It can provide data from a number of sensor types.

[0197] The second method, a library, can be used as a starting labeled dataset. Method 1, that is, calibration in the actual welding system by operator call The implementation of a trial run or a test run may be used as an alternative from the outset. The first method is the second This method can be used to add to a dataset.

[0198] Neural networks, over time and / or through learning, can weld together acceptable welds. The definition used to determine and / or unacceptable welding may be modified.

[0199] Neural networks can be trained through unsupervised learning. Unsupervised learning can be performed during the calibration phase.

[0200] Neural networks can provide processing based on clustering or grouping. Neural networks analyze data for similarities and / or anomalies within the data. We can look for patterns.

[0201] Neural networks can be trained using only the first method, or using only the first method. Training using the second method after the first method is also possible using both the first and second methods. You can also train in parallel.

[0202] The eleventh aspect of this disclosure, including other aspects of this disclosure, is described elsewhere in the document. This may include features, possibilities, or options.

[0203] According to a twelfth aspect of this disclosure, an apparatus for monitoring welding, (a) Multiple inputs for data from multiple sensor types, (b) One or more processors, one of the one or more processors teeth, a. Receive input from at least one of multiple sensor types, b. Data acquired from multiple sensor types, reference data from one or more sensor types. Compared to Ta, c. In determining the output, determine whether the weld is of acceptable quality or unacceptable quality. Steps and Equipped with, The device further, (c) one of one or more processors, (a) Combined data from two or more sensors of multiple sensor types Form and combine data to provide, (b) A second decision is made using the joint data, and the second decision is that the weld is of acceptable quality. It is either too good or too bad in terms of quality. Processor and (d) One or more outputs for determining the output and the output of the second determination, A device equipped with the following features.

[0204] The twelfth aspect of this disclosure, including other aspects of this disclosure, is described elsewhere in the document. This may include features, possibilities, or options.

[0205] Various embodiments of this disclosure are described below, with reference to the accompanying drawings, merely as examples. . [Brief explanation of the drawing]

[0206] [Figure 1a] This is a cross-sectional side view of a probe according to an embodiment useful for carrying out the present disclosure. [Figure 1b] This figure shows the same field of view as Figure 1a, but illustrates further features. [Figure 2] Figure 2a is a schematic diagram showing the effect of temperature difference on the ultrasonic beam within the substrate. Figure 2b shows the propagation of shear wave ultrasonic waves through a substrate with defects having two different homogeneous temperature distributions and one heterogeneous temperature distribution. [Figure 3] Figure 3a is a plot of a typical heterogeneous temperature distribution along the weld cross-section, derived from a thermal model and verified by thermocouple measurements. Figure 3b is a plot of an ultrasonic beam propagating through individual thermal boundaries, showing the planned focus and actual geometric location points for a substrate with a heterogeneous temperature distribution. [Figure 4] This is a schematic diagram illustrating the scan transformation process and how pixel values ​​are derived through weighted contributions from the nearest surrounding time samples on the nearest beam. [Figure 5] This is a schematic diagram of the process flow for the modeling, ray tracing, and imaging stages. [Figure 6] This shows a series of cross-sections through the weld groove, the position and sequence of additional weld passes, and representative ultrasonic images of the weld groove in that state. [Figure 7] This diagram illustrates an approach to verifying the thickness of a substrate by reconstructing a beam generated by the same ultrasonic device used for imaging welds. [Figure 8] This figure shows ultrasonic echoes observed within a signal trace and interference bursts occurring within the same signal trace at different times. [Figure 9a] This diagram shows the signal traces from three consecutive data acquisitions and the signs of interference bursts within them. [Figure 9b]This diagram shows the signal traces from three consecutive data acquisitions and the signs of interference bursts within them. [Figure 9c] This diagram shows the signal traces from three consecutive data acquisitions and the signs of interference bursts within them. [Figure 9d] Figures 9a, 9b, and 9c show the signal traces processed based on the average of the signal traces. [Figure 10] Figure 10a is a graph in which four separate signal traces are superimposed, three of which are rectified signal traces from Figures 9a, 9b, and 9c, having interference bursts but occurring at different time positions, and the fourth is the rectified signal after processing to remove interference while retaining the actual echoes. Figure 10b is a diagram of the process rectified signal trace with the four traces, showing sample-by-sample selected contributions from each of the three original rectified traces and the fourth trace, which is their average. [Figure 11] This diagram processes the unrectified signal trace from the other three traces and also displays the selected contribution that each of those three signal traces makes to the unrectified trace. [Figure 12] This is a diagram of an unrectified waveform according to one embodiment. [Figure 13] This is a schematic flowchart related to the selection process of the interference processing technology to be used. [Figure 14] This diagram shows interference across different channels, indicated by scope display, image display, and amplitude + time-of-flight display. [Figure 15] This figure shows the effect of processing on multiple channels, as shown in Figure 14. [Figure 16] This figure shows the identification of interferences in the trace section, the identification of subsections not used in the processed trace, and further approaches to the resulting processed trace. [Figure 17] This is a schematic diagram of the adaptive control function of the welding process implemented in accordance with this disclosure. [Figure 18a] This is an image of a defect detected by an ultrasonic probe. [Figure 18b] This is an image of the ultrasonic probe used for detection in Figure 18a. [Figure 19] This is a plot of outlier scores obtained from acoustic signal v data points under various welding conditions. [Figure 20a] This is a perspective view of the profile detection device for the substrate and welded area. [Figure 20b] This is a schematic diagram showing the sequence and shape of overall weld formation through a series of welding passes. [Figure 21] These are a series of camera images of the welding site during welding. [Figure 22] This is a plot of the arc voltage during the welding process and a plot of Gaussian amplitude x Gaussian center v time. [Figure 23] This is a diagram showing the combination data types displayed to the user. [Figure 24] This diagram illustrates the second level of processing applied to data from multiple sensor types. [Modes for carrying out the invention]

[0207] Ultrasound testing is used for non-destructive testing of various objects. The transmitting transducer is ultra Sound waves are emitted, enter an object, interact with the object and its sub-characteristics, and then the received transducer Return to the user. The temperature of an object affects the speed of sound within the object, so using an inappropriate speed will result in an inappropriate speed. This can negatively impact the quality of the examination and images.

[0208] When testing a weld, after the welding is complete or after each welding pass, perform, test, and weld. Alternatively, minimize the time required for necessary modifications to the welding path or optimization of welding parameters. To achieve this, ultrasonic testing must be possible immediately after welding or welding path formation. Desirable. However, this is when the object or substrate is hot, and between parts and / or This means performing ultrasonic testing while the temperature is fluctuating within a component of an object or substrate. ru.

[0209] To accurately detect defects and, more generally, to enable imaging of the welding location, this disclosure provides: We will clarify the temperature profile and consider the effect of that temperature profile on ultrasound. A method for correcting position is provided. As a result, accurate imaging is possible under a wide range of temperatures, and especially This method provides consideration of different temperatures at different locations under high-temperature conditions, and the welding location It enables early inspection of the welding area without the need to cool it to reduce the temperature's effect on ultrasound. It is useful for doing so.

[0210] This disclosure relates in part to the creation of accurate ultrasonic images of welds in high-temperature environments. Details of probe devices suitable for use in such environments can be found in Figures 1a and 1 at the end of this document. Provided by referring to b. However, the probe does not have to be a roller probe. Wedge-type probes that are placed on the substrate and move from place to place can also be used, Please note that the advantages shown can be obtained.

[0211] The basic structure of the probe device includes a rotating axis RR extending through the probe 13. On this axis of rotation is a second axis connected to the axis element 64 by a series of releasable fasteners 72. Element 70 is provided. The second axis element 70 is a transducer 52, an echo prevention block The mounting parts for the 74 and the ultrasonic transmission block 76 are provided.

[0212] The coupling element 28 between the probe 13 and the substrate [not shown] is an integrated part of the compatible material. This will be explained later. The connecting element 28 has a nearly straight cylindrical main body portion 54 and both Inwardly curved rims 56a and 56b are provided at the ends.

[0213] The transducer 52, anti-echo block 74, and transmission block 76 are the second axis Together with element 70, axis element 64, and first mounting location 20, the probe 13 is positioned on the surface of the object. It does not rotate when rolling on a surface. Therefore, transducer 52 and related components It is always maintained in the same sensing direction opposite to the object.

[0214] The internal volume 42 may contain the maximum vertical dimension of the transducer 52 surrounding the cooling components. To maintain the probe within the acceptable temperature range up to levels exceeding 50 degrees, the probe is charged. A coolant is provided that is introduced and removed at extreme intervals.

[0215] The second shaft element 70 has a coolant at a location 86 close to the inner surface 78 adjacent to the welding site. A mounting element 82 equipped with a thermistor 84 for sensing temperature can also be attached.

[0216] Regarding the transmission of ultrasonic waves, the coupling element 28 is made of high-temperature resistant silicone rubber. The material can withstand temperatures exceeding 350°C for extended periods. Such materials are 5 It has an attenuation of 0.87 dB / mm and an acoustic impedance of 1.12 MRayls at MHz. Furthermore, it has good compatibility with other materials used.

[0217] Regarding the thickness of the coupling element 28, increasing the thickness will allow it to interact with the contents of the probe. The combination of improved thermal insulation and an unfavorable increase in damping due to increased thickness Balance is maintained in between. Under the operating conditions considered, the thickness of such material is 2mm~ 10mm, for example, 4mm to 8mm, is suitable.

[0218] The material selected for the connecting element 28 conforms to the surface of the object under a moderate level of force application. It also provides sufficient compatibility. The high force level generates it and the device test P It is undesirable from the perspective of the equipment required to move it on the surface. The surface is not highly finished and is not smooth, so it transmits ultrasound without excessive loss. To achieve good contact, suitable materials are required.

[0219] Regarding the passage of ultrasound, the echo-blocking block 74 prevents ultrasound from bouncing back within the probe. Hydrogen plays a crucial role in preventing noise or other adverse effects on the probe. Nitrile rubber (HNBR) is a material particularly suitable for its N-filler foam. I understand. This is because an attenuation of 6.4 dB / mm can be obtained at 5 MHz.

[0220] All these functions involve acoustic coupling between the probe and the object via the actual surface. It helps to make it a success.

[0221] Regarding coolants, air provides insufficient heat capacity and thermal conductivity for active cooling. Water also, The acoustic impedance is 1.5 MRayls, which is incompatible with the other components, so it is not optimal. Water is not suitable for the temperature at which it occurs, and it is more likely to bond to the sliding surface than to act as a lubricant. It tends to do so. For example, a water-soluble oyl has an acoustic impedance of 1.1MRayls. When the coolant is provided in the form of yl, the acoustic impedance of the coupling element is approximately [1.1M yls]. It is more suitable for [degree].

[0222] Transducer 52 provides a 5MHz phased array of 64 elements, and object It is mounted to generate ultrasonic waves at 55°. 0.5mm pitch and 1 An elevation angle of 0 mm can be used. The angled beam completely welds the weld from a distance in the lateral direction. It has the advantage of being able to be inspected. In many cases, welding is performed in locations that are far apart laterally. The contact between the probe and the object becomes better than at the location where it is positioned. For example, in multipass welding, Large indentations remain until welding is complete, hindering good contact with the object and the propagation of ultrasonic waves. Furthermore, the oblique inspection angle better matches the direction of typical defects. This is also true for 0°. This becomes a problem with low-angle-based approaches.

[0223] This type of transducer and transmission block configuration emits an upper beam [ [Angle away from perpendicular to the transducer surface] and lower end beam [relative to the transducer surface] It can be used to provide a fan-shaped scan beam defined by [and nearly vertical]. Regarding the performance required for probes with respect to high temperature performance, this disclosure describes the performance required for probes at approximately 300°C over a long period of time. We provide probes that can inspect objects.

[0224] The bond is dry, but the necessary wavelengths are required to propagate ultrasound into and out of the object through the interface. To make the bell a reality.

[0225] High-temperature polymers used in bonding components can withstand prolonged contact with objects at such temperatures. This makes it possible to normally propagate ultrasound between the interface and the surface.

[0226] The coolant and the gap filled with the coolant effectively transmit ultrasound to the transmission block. It is possible to propagate it.

[0227] Because the transmission block is only exposed to temperatures close to the ambient temperature, it has low ultrasonic propagation characteristics and high resistance. There is no need to select a heating material, and optimal propagation characteristics are provided for the conveying block.

[0228] All components along the acoustic path, i.e., ultrasonic transmission block 76 [transducer [Wedge between SA52 and Compatible Material 28], Compatible Material 28 [Tire], and base to be welded The speed of sound in a material changes with temperature. This is because the angle of refraction is different for each interface of the material. This means that it changes with temperature. Furthermore, the beam within each material changes with temperature within that material. It bends when there is a slope. For example, referring to Figure 2a, the beam wavefront 100 is on the substrate 10 In the hotter part 102 of 4, the heat transfer is slower than in the colder part 106 of the substrate 104. Therefore, when the beam 100 enters the substrate 104, the lower temperature portion 106 of the beam is affected. Because the minutes move ahead of the beam portion of the hotter portion 102, the straight beam 100 'It deforms from. Therefore, due to the non-uniform temperature distribution, beam 100, beam 10 0 def It bends like this.

[0229] This effect can also be seen in Figure 2b. (a) Using a plate that is homogeneous at 25°C, the plate is used as the base material. (b) A shear wave propagates to the defect location, and (b) a plate and a homogeneous plate at 150°C Using an incident angle adjusted so that the beam is pointed at the same target, the plate is used as the substrate to pinpoint defects. (c) Another shear wave propagates to the defect location under heterogeneous temperature conditions, with the plate as the substrate. This shows shear waves propagating to a certain depth.

[0230] To allow for the consideration of possible beam bending, the entire ultrasonic path, especially high It is possible to clarify the temperature of the temperature and high potential fluctuation sections as much as possible. It is necessary. If the temperature distribution is known, beam bending can be predicted, and the device configuration and Appropriate corrections can be made to the operation [for example, to direct the focus beam to the desired location], Applicable to signal processing of the incoming and / or returned beam [for example, to provide a more accurate image] Therefore.

[0231] In the first stage of the decision, the heat input from preheating and the additional heat input from the welding itself are considered in relation to time. We perform thermal modeling to determine how the number of elements is distributed across the entire metal volume of the substrate. .

[0232] One way to verify thermal modeling is to use a base in the form of a steel plate with many thermocouples distributed around it. The materials were prepared. Thermocouples were placed on both the top and bottom surfaces of the base material. Next, the base material was placed in the optimal location for welding. The temperature was preheated, and various welding head passes were performed to obtain the thermocouple response. It is possible to broaden the range of modeling by changing the bell, but in many cases, This applies to samples of the same substrate, and even to samples of different substrates [thickness, profile, type of steel, etc.]. This will result in a consistent and strict temperature range. Therefore, it is expected that there will be almost no change due to preheating. It is thought that...

[0233] By performing this thermal modeling, the safe operating temperature of the above probe can be determined [ [Based on knowing the time-dependent response of the probe when exposed to different temperatures] to the welded area It can compare temperatures at various locations far away, and the modeling tracks the area behind the welding head. This will allow us to determine the nearest safe distance for the probe.

[0234] For thermal modeling based on substrate depth, these investigations reveal the post-weld state. It was revealed that the temperature distribution rapidly stabilizes towards thermal contour lines perpendicular to the metal surface. This significantly simplifies the shape of the temperature distribution, especially the complex temperature distribution that passes through the substrate. When it is necessary to compensate for the impact of skipping numbers, the compensation calculation is greatly simplified. A kip is when ultrasonic waves entering the substrate through the interface are internally reflected at the interface on the other side of the substrate, and then enter the substrate. It means returning to the interface.

[0235] Regarding thermal modeling of the spread from the hottest welding site to the substrate, the temperature distribution is as follows: While symmetry is often observed around the center of tangent, this is not always the case. As the beam passes through the welding area, it produces a complex S-shaped distortion. This is shown in Figure 3a. This can be confirmed by the degree distribution and the ray tracing analysis in Figure 3b. The location is mm on both sides of the welding site. It is expressed in terms of position, and the welding location reaches the highest temperature [1000°C in this case]. In this example, the beam An exaggerated temperature gradient is used to emphasize the magnitude of the curvature.

[0236] Referring to Figure 3b, the deviations caused by the temperature distribution in welding inspection are shown in more detail. It can be done.

[0237] Series of plans Focus 110 1 ~110 i The planned focus is 110. 1 It has a near-plane end on which a feature is provided. It extends along the plane 112. The near-plane end points towards the probe 13. Planned focus 11 0 i It is located at the far end of the plane and is further away from the probe 13. Plane 112 is the In some cases, it is parallel to the top surface of the substrate, and therefore, in such cases, a certain depth into the substrate. represents.

[0238] Planned focus 110 1 ~110 i For, the transducer 52, and the generated beam is angled with respect to the upper and lower interfaces of the substrate. In this regard, an angle of 55° can be adopted, and the center line 114 of the scan arc is defined. By the control system of the transducer 5 2, the beam can be steered to different angles to achieve sector scanning is possible. All elements of the array [64 in this example] are used for each transmission of the beam However, the excitation time of each element is different, guiding the resulting beam to the required location to focus to align. In reality, this defines the boundaries 116 and 118 of the scan arc, the minimum and maximum angles at these boundaries are defined, and thus the planned foci 110 at both ends 1 and 1 10 i are defined.

[0239] As revealed by thermal modeling, the temperature contours 120 are often perpendicular to the surface of the substrate However, the method can also function with other contour or boundary directions and / or profiles and can fully and extensively consider various temperature gradients in different directions within the substrate can be considered.

[0240] The more thermal boundaries used to divide the substrate, the more accurate the calculation of the temperature distribution becomes, and the determination of the actual geometric location points 122 1 ~122 i can be more accurate . Of course, the temperature change is continuous, but the same effect can be accurately represented using a large number of stepwise changes. can be accurately represented.

[0241] Each time the beam crosses a thermal boundary, the beam path is corrected based on the temperature change exhibited by the boundary. This can be applied. This provides a new path for the beam, before it crosses the next thermal boundary. A new route length section is created. Further corrections will create even more new routes. This process is repeated. The result is determined as the geometric location where the signal sample actually occurs. The actual geometric location is point 122. 1 ~122 i This can be shown in Figure 3b. Geometric location point 122 1 ~122 i The profile shows that the beam gradually becomes shallower. It bends each time, and as the wavefront encounters high temperatures toward the welding center, plane 124, it slows down. This is caused by the following: The angle becomes steep again, the wavefront accelerates again, and passes through the welding center, plane 124. After passing that point, turn in the opposite direction.

[0242] The actual geometric location where the signal originates: point 122 1 ~122 i and planned focus 110 points 1 ~110 i Because of these differences between them, the scan conversion algorithm The zoom displays the observed signal to the corresponding planned focal point, e.g., 122 1 , then *, actual geometric location points, for example 110 1 It needs to be plotted on the map.

[0243] A similar process occurs during reception, and signals from each channel undergo different time delays. Therefore, the sum is consistently added from all channels, and from the specified direction and range. Net response increases. A potential drawback here is that the steering angle and focus are fixed when acquired. Because it is fixed, it is not possible to correct any arbitrary errors in the heat distribution later. Figure 4 shows the position of a pixel located between the first beam i and the next adjacent beam i+1. This further explains how this is taken into consideration. In this diagram, A ij is beam i This is the amplitude of sample j. ·W ij This is the weighting coefficient for the contribution from sample j in beam i. α is the ratio of the distance of the virtual beams where pixels exist, from beam i to beam i+1. That is the case. β is the range from sample j to sample j+1 along the virtual beam in which the pixels exist. It is a ratio of distance. thus, A pixel =(1-α).(1-β).A ij +(1-α).(β).A ij+1 + (α).(1-β).A i+1j +(α).(β).A i+1j+1 Apixel = Wij.Aij + Wij +1 .Aij+1+Wi+1j.Ai+1j+ Wi+1j+1.Ai+1j+1 That is the case.

[0244] Since the configuration does not change between acquisitions, i+α, j+β are used to increase the speed of image generation. , W ij , W ij+1 , W i+1j Wi +1j+1 The value represents all pixels in the display area. This is calculated in advance and stored in the lookup table. During the imaging operation, each The addresses of the beam and sample data contributing to pixels (i, i+1, j, j+1) are: This can be derived by simply truncating the i+α and j+β values ​​in the lookup table. Generating overlays such as beam markers for selected A-scan data is possible. If the beam is straight, the operation is simple, but if it is a curved beam with a thermal gradient, it is more difficult. It becomes complicated. This requires a reverse lookup process from the pixel to the beam and then to the sample. This is necessary, and it is convenient to store the i+α and j+β values ​​in a lookup table.

[0245] Figure 5 shows the process including the modeling, ray tracing, and imaging stages. This shows Seth's complete flow path.

[0246] During the modeling phase, a thermal model is generated that shows the temperature of each data point. The model is based on previously saved information from previous actual welding operations and / or previous thermal models. It may be based on reports.

[0247] Next, actual thermal data of the welding environment is collected. This data is input [for example, provided] The preheating temperature and / or welding conditions may be obtained, and / or the data A temperature sensor in the welding environment that provides information on temperature and position [e.g., associated with a probe] This can be obtained from a thermistor, other temperature sensors, infrared sensors, etc. The collected data is used to form the actual thermal dataset.

[0248] Next, we compare the actual thermal dataset with the thermal model to assess the degree of agreement between the two regarding temperature at a given location. The degree of agreement is evaluated. If the difference in agreement is too large, the thermal model is then updated. The new ratio A comparison is performed, and this cycle continues until it is determined that the actual thermal dataset and the thermal model match. The cruise is complete. Now we can begin the ray tracing stage.

[0249] In the ray tracing stage, the thermal model approved in the modeling stage is used to analyze acoustic velocity It is converted into a map. The substrate is known, and the acoustic velocity with respect to the temperature of that substrate is calibrated. This will be revealed through simulation testing. The calibration data will be saved and sound Accessible for use in generating resonance velocity maps. The temperature of the data points is fast It is converted to degrees. Next, set the velocity contour interval and add the same velocity data points to the acoustic velocity map. Apply contour lines passing through the lines. In this way, the substrate is divided into different velocity zones. .

[0250] The starting point of each ray [beam component] is known, along with its initial path. The projection of that path. Each time the object crosses a velocity boundary, a refraction representation is applied. The refraction representation applied is contour lines. The range of the speed contour line varies depending on the speed difference it represents compared to the previous contour line, and the speed contour line is compared to the previous contour line. The direction differs depending on whether it is increasing or decreasing compared to the previous value.

[0251] The time it takes to pass through a speed zone is determined by the speed and the length of the path between contour lines.

[0252] This process is repeated for each contour intersection within the probe as it enters the substrate. This process can be repeated for each contour line intersection echo as it exits the substrate. Once the ray tracing step for each ray is complete, the process proceeds to the imaging stage. It is visible.

[0253] In the imaging phase, the first step is to generate the region of interest for which the results need to be displayed. This depends on the shape of the base material, the shape or configuration of the weld, the method of welding provided, etc. It could be affected.

[0254] The method for generating pixel values ​​is basically linear interpolation between the four nearest real samples. To avoid the time penalty of calculating this on the fly, four steps are used. The sample addresses and their respective weight values ​​are pre-calculated and stored in a lookup table. It is stored. Next, each newly acquired dataset is scanned and transformed, and each pixel is sequentially The data is processed, and four samples are accessed and multiplied by the weighting function associated with each. The values ​​are then summed up to form the final pixel values, resulting in a linear display image.

[0255] Because existing shape information exists regarding the position, shape, and direction of the weld groove, the results are as follows: The observed shape can be matched. If these are correlated, acquisition can continue. If there is no correlation, the correction is not applied correctly, and the temperature input to that correction is incorrect. The data needs to be checked.

[0256] If the detected temperature does not match the thermal model, correct it, and use acoustics for that correction. The velocity map needs to be readjusted. The corrected acoustic velocity map is the process shown above. It can be used in the app.

[0257] Figure 6 shows an example of the signal result (in this case, gradual filling of a weld groove by multiple welding passes). As shown, in the first sub-figure, there is no welding in the groove, and the edge of the groove is detected by ultrasonic imaging. As the welding passes gradually fill the groove from the bottom up, the area of ​​the exposed groove edge shrinks, and each image The echo area becomes smaller than in the previous image and eventually disappears.

[0258] If a defect is found, it will be on the right side of the image compared to the rest of the weld groove wall. It will be displayed.

[0259] The implementation described above uses thermal contours to derive transit times and corrects the temperature distribution. This is based on the following principles. Beam reflection is fully considered, and the beam path is determined accordingly.

[0260] In another implementation, Full Matrix Capture (FMC) is a total focusing method. It can be used in combination with SODD (TFM) for the same purpose, but for larger and more complete data. Consider the set.

[0261] FMC transmits the raw data signal, including the amplitude and phase time sequence, to all transmissions (Tx). This technique is similar to holography in that it collects data from a combination of elements and receiving (Rx) elements. This is a technique. The image creates a virtual path from all elements to all pixels, and then performs post-processing operations. This FMC dataset is reconstructed by performing differential delay as a workaround. This means that all pixels are in ideal focus in both transmission and reception, Therefore, this approach can be called the Total Focusing Method (TFM). The same as above, but with the gradient approximated by multiple separate thermal zones, as required by the reconstruction algorithm. The method is used to determine the thermal gradient correction pass time from each element to each pixel.

[0262] (Verification of substrate thickness) As stated above, the temperature profile within the substrate is a critical issue that needs to be addressed. Applicant Even if the manufacturer states that the material has a specified thickness, the thickness of the material may differ from the thickness of the base material such as a board. We also discovered that there was some stuttering. This point must also be taken into consideration when optimizing imaging. There is a need.

[0263] The beam path length is affected by the thickness, and the observed variation is relative to the thickness of the material. This gives a variation of 0-5 mm in path length, compared to the value associated with the specified thickness. Note that the path length is multiplied by the number of half-skips performed. To correctly locate the insect's position, the path length should be longer than the assumed path length based on the specified substrate value. Furthermore, it is necessary to clarify the actual distance of the path to the defect and the path length from there.

[0264] As a pre-welding inspection step, the probe is moved across the substrate along the same path as during the welding inspection. This is possible. An ultrasonic beam can be used to determine the thickness of the substrate. This approach Figure 7 shows the position of Chi.

[0265] The roller probe 13 described above is shown in Figure 7 as a coupling element 28 and a transmission block 76 This is only partially shown. Transducer 52 is not shown, but the transmission It is located next to the angled surface of lock 76. The angled surface of the weld for inspection purposes as described above. Unlike the NDT beam 800, which is a wave-segment scan, the thickness measurement beam 802 It operates at 0° (i.e., perpendicular to the metal surface) relative to the substrate 804 or object. Use your electronic equipment to move the beam away from the position of the NDT beam 800, and the thickness measurement beam 8 Specify the position of 02. To include the 0° beam, select multiple options centered on 0°. Sector scanning with a vertical beam is used.

[0266] The 0° beam is used as a method to monitor the quality of the coupling between the probe 13 and the substrate 804. It is used to check the amplitude of the back wall echo from the substrate 804, but in this case, The system not only confirms the presence of a predetermined echo intensity, but also measures the thickness. It is also used.

[0267] The strongest echo received by the detector of probe 13 is when probe 13 is in contact with substrate 804. The first interface 806 corresponds to the second strongest echo at the interface 808 on the opposite side of the substrate. It corresponds to internal reflection. This method uses echoes and their repetitions to localize Measure the thickness and correct the position of the region of interest.

[0268] By using a narrow sector beam instead of a single beam for the thickness measurement beam 802, This method makes it possible to determine which angle generates the strongest repetitions. Yes, and this provides information about the tapered compression of the connecting element 28, which is a tire in this example. It can provide. Tapering can cause beam refraction, which means that the weld will be in the correct shape. This is another issue that may require correction when converting the image to a digital format.

[0269] The detected position is corrected according to the temperature as described above, and the actual position is along the length of the path. It can be used to determine the distance. 0° measurement direction, i.e., perpendicular to the substrate. Therefore, in order to achieve the minimum distance through the substrate or thickness, temperature compensation within the transfer block 76 is required. Therefore, it is necessary. This should be used at the same location along the path on the base material during inspection of the welded joint. By doing so, the image can be corrected, and the location of any defects found can be corrected.

[0270] Alternatively, the probe can advance along the weld path during the actual weld inspection to determine the thickness. In this process, the welding inspection mode and thickness determination mode can be quickly switched and the cycle can be repeated. .

[0271] As a result of either approach, any position along the weld path can be compared to the theoretical value. Since the actual thickness is known, that thickness is used to define the shape features and / or defects. The image can be corrected regarding the position.

[0272] (Handling of RF interference) Significant interference from robot motors for welding arms, probe arms, and other tasks. The bursts were observed in the ultrasonic detection signal trace. These were from the planar interface. Even if present, it was found to be several times larger than an ultrasound echo. Other bursts were not as strong. Therefore, it is difficult to distinguish it as a burst. Planar interface echoes themselves are defect echoes. It is often far stronger than that. Other sources of interference also exist, and the techniques provided here This can be addressed by technology. For example, it is used in cooling system pumps that occur in welding applications. Electrical interference occurs from various motors, including motors used in motor testing and measurement. Possible sources of interference include the elevator's drive motor and power supply. Therefore, To enable clear image capture, solutions to address such interference are desired.

[0273] The broadband nature of bursts means that bandpass filtering techniques have limitations in reducing interference amplitude. It means that it has the effect of being achieved.

[0274] The smaller the return signal, the greater the problem, and in high-temperature probe environments, the transducer Distance from the material to the substrate, dry interface with the substrate, and similar factors in the return signal The signal is completely attenuated, and noise becomes an even bigger problem.

[0275] Furthermore, the timing of the interference burst is determined by the motion task that each arm is required to perform. Accordingly, the occurrence and duration become irregular.

[0276] Figure 8 shows that an ultrasonic echo of a defect is visible towards the midpoint of the signal trace, and after the trace... This is a diagram illustrating a signal trace showing an interference burst towards the halfway point.

[0277] However, the intermittent nature of the interference may contribute to the solution.

[0278] In the first embodiment, which is intended to take interference bursts into consideration, a transducer is used. A series of beams are transmitted through the substrate using this method, and the return signal is detected as before. For example, Figures 9a, 9b, and 9c show three such detected signal traces. ru.

[0279] Interference from the robot motor occurs at a gap significantly longer than the repetition period of the ultrasonic pulse. This occurs in short high-frequency bursts separated by a "p". This time difference is due to the robot motor If a burst appears in one beam, it means it won't appear in the next beam. Therefore, using the interference-free data from this second signal, the contaminated first beam is processed. The sample can be replaced. If there is a second independent robot, it will be a second source of interference. It functions and the burst is asynchronous with the burst from the first robot. If interference from the net occurs in the sample region of the second signal that is not affected by interference, the third signal or These samples become clear and can be used in place of the contaminated samples.

[0280] In each case, the defect is applied to the same location in all three traces, therefore the defect Therefore, echoes occur simultaneously in the traces. The three traces follow essentially the same path. The beam is generated from the same beam entering and leaving the substrate, and the probe follows the path within that time frame. [Because it does not travel a significant distance], it encounters the same substrate, at the start of signal tracing. The timing does not affect the echo.

[0281] In the latter half of the first signal trace in Figure 9a, interference bursts are also present. Similar interference bursts are also present in the early part of the trace of the second signal trace in Figure 9b. Furthermore, interference bursts are present in Figure 9b, but not in Figure 9a. In Figure 9c, Furthermore, there are two potential interference bursts.

[0282] Figure 9d shows Figures 9a, 9b, and 9 used to improve the signal-to-noise ratio [SNR]. The signal trace is shown as the sample-by-sample average of the three signal traces of c. The echo portion of the signal trace is still noticeable, but in any part of the signal trace Since only one signal trace contributes to the average of the interference burst, they decrease. This averaging approach is a basic method, and depending on the situation, which echoes This method can indicate which is an interference burst. It does not attempt to identify the timing. Such bursts are in the average signal trace of Figure 9d. It contributes to this. However, averaging multiple beams significantly reduces the burst amplitude. The more beams that are averaged, the greater the reduction.

[0283] However, if the interference burst is equivalent to or greater than the echo in scale and / Alternatively, if the signal is noisy or complex, simply reducing the amplitude may be sufficient. This may be insufficient. If the rupture gives the result an echo-like impression, welding This can lead to false negative determinations in the department, requiring unnecessary further review and / or corrective action. .

[0284] The above technology correctly identifies interference bursts and the echoes that occur in the same part of the signal trace. This could be improved by ensuring that it is detected in some way.

[0285] Figure 10a shows a more sophisticated approach to removing interference from the signal trace. It is being used. Figure 10a shows three separate arrays representing the same beam in three different transmissions. The following shows a plot of the current signal traces 0, 1, and 2 superimposed on each other.

[0286] The mechanism for determining whether a sample is likely to be from an interference burst is, Compare the scale of the rectified (envelope-detected) samples from each of numbers 0, 1, and 2. This is done by comparison. The signal with the largest sample size is likely to be due to interference. Because it is complementary, selecting the minimum value across the entire signal will ultimately result in the waveform being used. If none of the signal samples are from an interference burst, the sensor or measurement Due to random noise in the instrument, the amplitude difference between signals becomes very small. Therefore, this vibration Comparing the width range to a threshold is a mechanism to indicate the absence of interference, and all signals The SNR is improved by using the average of samples over a certain period. The threshold is set to the expected noise amplitude. It may be selected as a relevant value, or as a defined percentage of the signal amplitude.

[0287] Signal trace 0 includes a high-amplitude section towards the end of the signal trace, Since this does not exist in 1 or 2, it is considered a burst from RF in stream 0. This is done. Therefore, the portion of stream 0 that contributes to the average signal trace is formed. Therefore, this section is omitted due to its contribution. This is the RF0 at the bottom of Figure 10b. As can be confirmed in the trim, trace 300 is omitted in this section. Other sections 3 02 is also omitted from RF0's contribution, and these spikes are not observed in other streams. It hasn't been done.

[0288] The averaging of trace contributions that are permitted to contribute to the average signal trace is contaminated. This means there is an improvement in the signal-to-noise ratio from averaging out traces that were not included.

[0289] In a similar manner, returning to Figure 10a, 1 has two highs towards the beginning of the signal trace. It includes an amplitude section, but this is not present in 0 or 2, therefore it is RF for stream 1. These are considered bursts. Therefore, the streams contribute to the average signal trace. If part 1 is formed, this section is omitted from its contribution. This is shown in Figure 10. This can be seen in the RF1 stream at the bottom of b, and these parts of trace 304 are omitted. It can be done.

[0290] A similar approach involves two sets of two signal trace spikes in the latter half of that trace. And the omission of those from the contribution of Stream 2 (Section 306) was also adopted. ru.

[0291] Excluding these sections, the section that contributes to the average trace is one of the sections in Figure 10b. This is displayed at the bottom, and this yields the signal trace plotted in Figure 10b. The signal trace only shows the expected echo spike 310, which is clearly observed in Figure 10a. In addition, the other two echo spikes 312 in Figure 10a are also shown. The remaining signal section, The lowest trace in Figure 10b is beneficial and indicates a significant amount of time for the defect signal to be detected. To provide.

[0292] Up to this stage of processing, there is a section for the average trace, and these are 0, 1, and 2. This represents samples that were judged to have no interference and can be replaced with the mean of 0, 1, and 2. Refer to the lowest trace among the four in Figure 10b. This averaging process ensures reliability. This offers the advantage of reducing the number-to-noise ratio. If these sections are not present, then 0. Since one or more of 1 or 2 are identified as coming from an interference burst, clear Only the average of the signal is used. This means that the signal-to-noise ratio is somewhat affected by the averaging. While this improves the result, the number of elements contributing to the average decreases due to the excluded signals.

[0293] Use this minimum or threshold approach across all potential traces This is a fast and very efficient method that combines the elimination of possible interference with the replacement of alternative trace sections. This is an efficient method. However, this method does not identify bursts and does not suggest interference. The high-amplitude section, which is significantly larger than the matching values ​​in other traces, is due to a burst. It is assumed that there is interference. If interference is identified in only one beam, average the remaining beams. Compared to the minimum approach for all streams, the SNR is improved and the signal fidelity is To improve performance, techniques for more clearly identifying bursts would be helpful.

[0294] In yet another approach, interference detection and trace section replacement are performed in separate steps. This is processed. This further approach is explained below in conjunction with Figure 16, and it is This is a sample-by-sample approach.

[0295] By providing a general approach that can identify burst regions in each trace, All trace sections without contamination can be properly identified. The burst can be identified. The criteria used and the replacement method for replacing sections containing bursts are as follows: Each can be controlled individually, and the criteria can be changed as needed.

[0296] In this general approach, the first step is to stream or tray each beam. The goal is to identify the suspicious interference section on the surface. This is achieved as follows: . 1.1. If the input data trace / stream is a complete RF waveform, each trace Envelope detection is performed on the stream waveform, and the rectified trace is applied to each beam. Generates a stream. 1.2. For each beam, each rectified sample across all traces / streams Find the minimum value. This will be a good approximation of the uninterrupted value for that time sample. 1.3. Calculate the difference between each trace / stream sample and this minimum value, and if it exceeds the threshold... If this occurs, it is marked as interference. This threshold is a value obtained by multiplying the minimum value by a predefined coefficient, etc. It can be a constant or an adaptive type.

[0297] In the second step, this method performs interference removal using the following method. 2.1. Use the interference marker (derived in 1.3) to trace the input / stream of the beam. I refuse the ream sample. 2.2. Calculate the average of the remaining samples in the entire trace / stream. 2.3. Use this average value as the time sample of the processed beam.

[0298] Referring to Figure 16, the three traces above are rectified data, and at the top of Figure 16 The upper, middle, and lower waveforms are shown as being for single-beam streams 0, 1, and 2. Highlighted section of the interference detection plot (one A towards the center of stream 0) (One B towards the end of Stream 1, two Cs and D towards the beginning of Stream 2) These are all sections with interference. Peak e is an echo.

[0299] Applying the above algorithm detects this interference, and the interference processing plot is shown in Figure The three traces at the bottom of the 16, 0'', 1'', and 2'', are marked and processed. Sections that require further consideration to reach the goal are excluded. Interference burst Each subsection within sections A, B, C, and D is represented by a separate bar-style peak. The 'ru' subsection is highlighted as a target for discarding. These identified and highlighted subsections The n is excluded from the formation of the processed trace (trace P in the center of Figure 16). The processed trace P shows the processed sample of the beam, with interference bursts removed and waves removed. This demonstrates both noise reduction in the rest of the shape.

[0300] Figures 11 and 12 show further modifications to the previous approach, which is not rectified. Applicable to any signal. This is also a sample-by-sample selection, and therefore the standard used With improved flexibility and signal-to-noise ratio, the general approach of the previous embodiment is also used here. It is possible.

[0301] Here, we identify samples without interference bursts by selecting signals with the smallest amplitude. This approach does not work. Instead, the signal is first rectified, and then the original approach... These modified waveforms are applied, and the samples of the final waveform are from signals 0, 1, and 2. It is determined whether the signal is from the average. This selector mechanism is applied to the original unrectified signal. This generates an unrectified, processed signal.

[0302] In Figure 11, a situation similar to that in Figure 10a exists for streams 0, 1, and 2. Stream 0 has an RF burst in the latter half of the signal trace, and Stream 1 has a signal trace There was a burst early in the stream, and in stream 2 there were two bursts in the latter half of the signal trace. There is a contribution to this. If we remove these contributions and create the average trace at the bottom of Figure 11, the same A missing section of the same type occurs.

[0303] However, by further processing, it is possible to obtain from one or both of the other signal traces. It is possible to reconstruct the missing section using the section. Regarding section 330, which was excluded by RF burst 332 of stream 1: Using section 334 of stream 0 or section 336 of stream 2, It is possible to fill in section 338 of the average trace. Sections 334 and 336 If you are filling in the blanks using both, you can use the average of the two.

[0304] The results of this embedding are shown in Figure 12, and there are three important sections in the entire trace. It is obtained from the average of the traces, and some of it is obtained from the average of two traces, but Other sections are filled in using one of the other traces. Stream 0 Missing sections are filled using Stream 1 and / or Stream 2. (See diagram) In 12, a continuous trace is shown with no echo loss and interference fully accounted for. The flaws and echoes are not lost.

[0305] The interference from the robot motor is significant, but the interference frequency of the welding arc is the same as that of the target ultrasonic signal. Since it is unrelated to the frequency of the signal, no special processing is required.

[0306] The approach to unrectified signals is also directly applicable to full matrix capture signals. It is applicable. Here, the SNR of each received signal is generated from a smaller excitation energy. Because it is inherently low, S is due to interference burst removal and averaging in the absence of bursts. Combinations of NR improvement are particularly effective.

[0307] Because the FMC dataset contains many signals to be processed, the computation time for processing FMC signals Minimizing the gaps is especially important. All received signals of a particular element's transmission are sent in parallel. When acquired, due to the nature of electromagnetic interference, all received signals have bursts in the same sample. Therefore, multiple pulse sets from only one receiving channel will be sampled. The selector needs to be analyzed to derive the parallel receive channel. Because it can be used throughout the entire system, computation time is reduced.

[0308] Since the above methods require different amounts of computation, computationally intensive methods should only be used when necessary. It is useful to select one method and, if not, to use an alternative method. In this regard, Figure 13 shows the latent This represents a selection approach that considers possible choices. Here, "the mean of the entire sample" Approach [see Figures 9a-9d and explained above], or "minimum sample value" approach Chi [as described above with reference to Figures 10a, 10b, 11, and 12], or full matrix In the case of cu capture, the "single-channel rectification" approach is considered. Single channel is For more general rectification, it is rectified as described above.

[0309] Therefore, referring to Figure 13, the system retrieves multiple signal traces from multiple transmissions. It is set to cancel and acquire at 1300. The repetition rate used is alias Adjusted to avoid single patterns 1302, and then the data [signal trace] Acquisition 1304 may proceed.

[0310] Regarding the acquired data 1304, in the case of the full matrix capture approach and It could be similar, and in the processing, there is interference across all parallel receive (Rx) channels. The question will be taken into consideration.

[0311] If there is no interference on all channels 1308, each signal tray of multiple transmits (Tx) The process is then examined for each beam in turn. Next, the process is carried out using the approach outlined above. Rectification 1312 is performed using the signal trace. The minimum and maximum values ​​of the signal trace are obtained 1314. It is compared to a predetermined threshold.

[0312] If the value is below the threshold of 1318, the average of the samples from each of the multiple transmit Tx streams The approach continues by using the average to obtain the result 1320. This is shown in Figure 9a. This is the approach described above, referring to ~9c.

[0313] If the value exceeds the threshold of 1322, a more discriminative analytical approach is required, and this The approach involves selecting the minimum value of each sample and constructing a rectified stream. Using the same selector, the rectified signal trace 1326 of the raw signal trace is obtained. Build the trace and get the results 1320. This is shown in Figures 10a, 10b, 11, and This is the approach described above, referring to section 12.

[0314] Returning to the initial stages of the process, we examine whether interference exists in all channels. 130 6. If 1328 is used, a different process approach is used. In that case, one chat Only the Nell needs to be analyzed 1330, and this is then performed again by rectification 1332. Then, the minimum and maximum values ​​were found at 1334, and using the sample selector at 1336, It is possible to select the appropriate Tx stream for all parallel Rx channels. Again, result 1320. To reach.

[0315] Interference is observed in all channels. This approach requires a scope display of 140 before processing. Image 1402, and other scan amplitude traces 1404 and time-of-flight map 1402. Figure 14 illustrates that there is a material interference at point 6.

[0316] The processed position is shown in Figure 15, and each scan, i.e., the processed scope display 1 500, processed image 1502, and amplitude traces of other scans 1504 and during flight Clear improvements can also be seen in the intermediate map 1506.

[0317] By using multiple transmissions and selecting the minimum rectification value for each sample, any Interference can be properly considered.

[0318] Comparing the processed results with the acquired data stream and signal trace, which... A selector waveform is generated that indicates which stream the sample came from. The selector is the original It can be used to create a processed RF stream from an RF stream, thereby This could enable the acquisition and processing of FMC data.

[0319] If no interference is detected (i.e., the range of values ​​across the entire stream is narrow), the selected process is Since only the average sample is used, the computational load is low. At the same time, in the case where interference is detected... In addition, the processing may select an alternative processing method that provides a minimum sample value approach, potentially reducing interference. It can effectively deal with such impacts.

[0320] Increasing the number of transmissions considered allows for the consideration and management of more asynchronous interference sources. Yes.

[0321] The various approaches provided by these embodiments are for preventing or suppressing interference. This is useful when the other method is insufficient.

[0322] (Further disclosure) This further disclosure concerns the use of multiple sensor types and combinations of data from them. Based on improvements to welding, quality control in welding, and their use. .

[0323] During welding, numerous variables exist that affect weld formation and weld quality. By monitoring individual variables, we can identify their deviations and the resulting decrease in weld quality and / or Alternatively, attempts have been made to detect the possibility of defects occurring. The welded area is where the weld is formed. At that time, and immediately after formation, it has various properties. The individual properties are determined during welding, and These deviations are detected, which may result in a decrease in welding quality and / or the occurrence of defects. Attempts have been made to detect the possibility of life.

[0324] One of the potential purposes of this disclosure is to mitigate the formation of welding defects or the occurrence of other quality problems. To do so, we make greater use of variable monitoring and / or characteristic determination. The objective is to provide a method for verifying welding and quality that yields profits. One of the goals is to synchronize and merge data from multiple sensor types and / or sources. A welding and welding condition verification method that aims to obtain more information than the sum of its parts. The goal is to provide [this]. Furthermore, it aims to analyze each data type in a more beneficial way. By doing so, potential advantages can be gained.

[0325] The seventh to twelfth aspects of the present invention are referenced above.

[0326] The following is a description of these aspects.

[0327] After the weld is formed and cooled, non-destructive welding is performed to check for defects. It is known that destructive testing (or inspection), NDT, is performed for any defects found. Repair work can be performed afterward. However, with conventional techniques, such NDT is not welded. Because this is done a considerable time after the part has been formed, information about defects is available. Provided a considerable time after the event, preventing the formation of defects or further defects. It's too late to mitigate the formation.

[0328] During welding, there are many variables that affect the formation and quality of the weld. Ear feed speed, voltage, current, welding speed, distance between welding device and welding location, shielding gas, Shielding gas flow rate, substrate shape / profile / composition / dimensions, substrate preheating temperature, welding groove. Shape / profile / configuration / dimensions, shape / profile / configuration / of weld or weld path May include dimensions.

[0329] During welding, individual variables are monitored to track their deviations and the resulting weld. Attempts have been made to detect the potential for quality degradation and / or the occurrence of defects.

[0330] When a weld is formed, and immediately afterward, the weld has various properties. , acoustic radiation from the welding site and / or arc, the topology of the weld formed, formed This may include the visual characteristics of the weld when it is formed and / or after it has been formed.

[0331] During welding, individual characteristics are determined, deviations from them are detected, and as a result, a decrease in weld quality is detected. Attempts have been made to detect the possibility of defects occurring.

[0332] (Overall process) This disclosure provides variable monitoring to mitigate the formation of welding defects or other quality problems. By making greater use of nomination / or characterization, we can gain increased advantages in welding. This disclosure aims to synchronize and merge data from multiple sources to partially The aim is to obtain more information than the sum of the facts. Furthermore, this disclosure will be presented in a more useful way for each individual. The goal is to analyze data types.

[0333] This disclosure improves initial weld quality and reduces the risk of defect formation. The range of types of welding data collected, analyzed, and used during the welding process will expand. This is achieved by [meaning].

[0334] This disclosure provides real-time data obtained from multiple sensors and / or sensor types. It processes and uses machine learning / artificial intelligence algorithms to control the welding process to the fullest extent. The goal is to arrive at real-time decisions. Therefore, the decision is based on multiple different factors. The sensor type involved will benefit from this.

[0335] The illustrated disclosure uses a combination of four different sensor types to collect data. Furthermore, it can provide adaptive control functions for welding and reduce the possibility of defect formation. However, this method, Suitable for handling a wider range of sensor or data types, as exemplified. It is also suitable for use with sensor types different from those of other devices.

[0336] Figure 17 shows a schematic diagram of the adaptive control function.

[0337] In the welding process, a set of variables that affect and control the welding are defined and controlled. This includes wire feed rate, voltage, current, welding speed, and the distance between the welding device and the welding location. Distance, shielding gas, shielding gas flow rate, substrate shape / profile / composition / dimensions, substrate Preheating temperature, weld groove shape / profile / configuration / dimensions, weld or weld path shape / pitch This may include file / configuration / dimensions.

[0338] In the welding process, a high-temperature resistant roller-type ultrasonic probe is used on the left side of Figure 17 [details to follow]. The NDT step [post-weld high-temperature NDT] is performed using [described], and in this step, ultrasonic Data [NDT display] is provided, which indicates the completion of the process and an overall assessment of weld qualification. It will be sent to the price.

[0339] In the welding process, it is not enough to simply monitor the generated weld and prompt repair actions for it. Furthermore, a series of parallel steps, starting with process monitoring, are executed to prevent the occurrence of defects. The goal is to minimize it.

[0340] The process monitoring step involves monitoring the time-dependent changes in the voltage and current applied to the welding device. The data is measured and collected. In this embodiment, voltage and current detection is used to determine the positive We will examine problems in forming a good or correct weld.

[0341] At this point, data from the other three types of sensors is incorporated into the data stream, and the sensor Sender data is formed. These are laser, vision, and other appropriate sensors of that type. and the audio data stream. Details are described below. In this embodiment, Detection of weld profiles using Zerscan, visual evaluation of welds, and welds Acoustic detection of the sound generated is also used to determine the correct weld formation, or any issues in the correct weld formation. The topic will be considered.

[0342] The process sequence uses laser, vision, and acoustic data streams. This includes the collection of data. Furthermore, the process monitoring step applies to the welding device. Collect data on the voltage and current applied. Post-thermal welding NDT is the fifth data type. It also provides an NDT display. These form the complete data stream that is considered. Each of these sensor types and the data types they provide will be taken into consideration. The various sensor types are as follows: • Phased array ultrasonic sensor type - for measuring welding quality. • Acoustic sensor type - Derives welding process performance from the sound of the welding arc. • Laser sensor type - Monitors the topology of the weld. • Visual monitoring type - replaces operator vision in the visualization of the weld pool. • Voltage and current monitoring type - Measures power to maintain the welding process in optimal condition. ru.

[0343] (Ultrasonic sensor type) This sensor type is related to the result of the generated welding, while other sensor types are related to the generation It focuses on the welded area. Therefore, this sensor type is relevant to the need for repair work. While it provides information to do so, the sensor types described below, in the first place, reduce the occurrence of defects. It can be used for that purpose.

[0344] The ultrasonic sensor type is a 5MHz device mounted on a substrate to generate 55° ultrasonic waves. Includes a transducer that provides a 64-element phased array of z. 0.5mm pitch An elevation angle of 10 mm can be used. Angled beams can be welded from a distance laterally. It is convenient in that the part can be inspected completely. In many cases, the areas spaced apart laterally are welded. This is more suitable for good contact between the probe and the object than where it is being done. For example, multi-pass In welding, large indentations can hinder good contact with the object and the propagation of ultrasonic waves until the welding is complete. This can occur. This becomes a problem with 0° or low-angle-based approaches.

[0345] This type of transducer and associated ultrasonic transmission block configuration emits Upper beam [angle away from perpendicular to the transducer surface] and lower beam [transducer To provide a sector-shaped scan beam defined by [approximately perpendicular to the deucer plane] It can be used.

[0346] Regarding the performance requirements for probes related to high-temperature performance, probes are suitable for temperatures of approximately 300°C. This makes it possible to inspect the substrate over a long period of time.

[0347] The bond is dry, but the necessary wavelengths are required to propagate ultrasound into and out of the object through the interface. To make the bell a reality.

[0348] High-temperature polymers used in connecting components are susceptible to prolonged contact with objects at such temperatures. It is able to withstand the forces and allows ultrasound to propagate normally between the interface.

[0349] The gap between the coolant and the sensor-type elements filled with coolant is the transmission block and It is also possible to effectively propagate ultrasound between them.

[0350] Because the transmission block is only exposed to temperatures close to the ambient temperature, it has low ultrasonic propagation characteristics and high resistance. There is no need to select a heating material, and optimal propagation characteristics are provided for the conveying block.

[0351] Sensor types include integrated surface temperature measurement sensor types and coolant temperature measurement sensor types. It can be done.

[0352] (Conditions and effects of welding location) In arc welding, a power source is used to ensure sufficient space between the electrode of the welding device and the substrate being welded. A voltage difference is generated, causing an arc to form. As a result, an electric current is generated. The base The material is heated to a molten state [electrodes may be consumed]. When cooled, the molten metal The two substrates solidify and bond together.

[0353] The speed at which the welding device moves relative to the substrate affects the degree of melting, the shape of the molten pool, etc. This affects the quality of the weld.

[0354] The welding area is usually protected by a shielding gas, which contains oxygen, water, or water vapor from the atmosphere. To prevent it from reaching the location. Shielding gases are typically inert gases such as argon or helium. Alternatively, a semi-inert gas is used. The flow rate of the shielding gas affects its ability to perform its function. ru.

[0355] Maintaining high quality in welds requires careful control of many operating variables within the welding process. It is necessary. These variables are considered indirectly, as illustrated in the section below. It is possible. In the exemplary embodiment, the movement of the welding device is not directly sensed, but other types of sensors can detect it. We measure the dynamic speed, shielding gas flow, and shielding gas flow rate, and the type of sensors used and The dataset can be added to the processing.

[0356] (Acoustic sensor type) Acoustic sensor types collect high-frequency audio signals generated during welding. These signals are The arc that forms the weld, the interaction between the arc and the substrate, and the interaction between the arc and the shielding gas. It occurs from there. The detected audio signal is sensitive to several important variables in the welding process. This was revealed. Figure 19 shows examples of outlier score values ​​for various welding characteristics. .

[0357] Outlier scores are used as tolerances for the characteristics and / or sensor type data being evaluated. A number that considers how far a particular data value is from a categorized set of known data values. It is obtained through an academic approach.

[0358] One such approach used in this disclosure involves measuring the value of the input audio signal. Previously developed principal component analysis (P) on the types of signals and their signal and / or data values. This involves using a novelty detection model for the Mahrambis distance by comparing it with the CA model.

[0359] Signal processing involves acquiring an audio signal and applying a noise reduction algorithm to the raw data. Includes. The audio signal is further processed using the Short-Time Fourier Transform, and the raw time-series data Convert to the frequency domain. Then, statistically specific values ​​are obtained from each of a series of bandwidths across the frequency range of interest. The characteristics are extracted. In this example, the bandwidth used is 39.1 kHz, and the specific signal is 312 features are generated that describe the acoustic signal of the intensity. Next, redundant features are removed. By doing so, the overall feature set [312 features per signal instance] is optimized. Standardization is performed to improve the robustness of features with small standard deviations.

[0360] In the initial settings of the PCA model, the signal resulting from the above processing, and therefore the remaining feature set The setting is configured so that the welding performance is acceptable with respect to that variable (in this case, the acoustic signal). Therefore, the remaining feature set can ultimately be reduced using PCA, weld. The principal components and model that define the model in terms of its acceptable performance are obtained.

[0361] A PCA model with acceptable performance can be used to consider subsequent signals. The subsequent signals undergo the same noise cancellation and other steps as defined above. This applies to the distribution of the characteristic values ​​of the acceptable welding performance that have been identified for each feature. The position of the feature value, i.e., the feature value, is determined. The mean of the distribution is calculated, and that mean is used as the reference. The distance between feature values ​​is then measured. The distance provides a quantification of outliers, as shown in Figure 19. , the distance variation that matches acceptable welding conditions and the distance that is more exceptional and shows a decrease in welding performance To clarify the fluctuations.

[0362] This approach provides unitless, scale-invariant quantification that considers the correlation of feature values ​​within a distribution. It is convenient for performing the transformation. The mean is the distribution of acceptable feature values ​​and / or PCA model. Recalculating each time an addition is made, or existing acceptable features used in the PCA model It is also possible to perform calculations based on a fixed set of values. This process is used to obtain the results shown in Figure 19. Referring to the sample results shown, the first set of data points indicates good solution. This indicates that contact operation parameters are occurring. These parameters are applied. These are individually examined. As can be seen from these, these are log-scale outliers. Provides a set of appropriately clustered data points A for the score axis.

[0363] The second set of data points indicates that the welding speed is too fast. In other words, the welding device The welding material and the base material are moving too fast relative to each other. Two different welding speed deviations are shown. These are two well-organized data point sets B with very few outliers. 1 oh Call B 2 Give.

[0364] The third set of data points shows the sidewall arc during welding. In other words, the arc is in the weld groove. The short circuit is occurring on the sidewall, not the intended welding location inside. Again, these are mostly outliers. No, it provides a well-organized set of data points C.

[0365] The fourth set of data points indicates sidewall melting. In other words, the arc melts the sidewall. This causes melting in the sidewall rather than the weld groove. The data points have few outliers and data Point D is properly clustered.

[0366] The fifth set of data points indicates that the shielding gas flow rate is too high. This could potentially lead to undesirable porosity issues. Other data point sets Similar to set T, this set E also has few outliers and is clearly defined. Shielding gas flow The same location is detected even when the quantity is too low.

[0367] When the aforementioned undesirable welding conditions are present, the welding performance is better than when good welding conditions are present. It exhibits a much higher outlier or feature value. As a result, the threshold Th acoustic [Mahalano The screw distance can be set, and data obtained from acoustic sensor types indicating good welding conditions can be used. To distinguish between data obtained from acoustic sensor types that indicate faulty welding conditions and data obtained from acoustic sensor types that indicate faulty welding conditions. It can be used for this purpose. Therefore, the acoustic data type is the threshold Th acoustic It was broken If this occurs or if the data remains corrupted at a specified number of points, the operation will be warned. It has the ability to trigger the cessation of welding, thereby realizing defect generation. Time-based acoustic signal identification is provided. This analysis can be provided continuously, multipath This can be provided for each path of the welding approach.

[0368] Importantly, the algorithm used depends on the type of sensor and its signal value and / or use a single principal component analysis (PCA) model previously developed for the data values. It is more complex than simply welding. As shown in Figure 20b, in multipass welding, welding Each pass gradually fills the weld groove. This is because the depth of the weld groove, its shape, and the amount of filling each pass have a corresponding effect. This means that it changes to... All these changes, and possibly other changes between paths. Changes in this affect the emitted and detected acoustic signals. Therefore, this approach So, we use a separate model for each path and apply the observational data to the predicted data for that path. To evaluate in the most accurate way.

[0369] Individual models are neural network applications that are trained based on existing paths. It can also be obtained by roaching, and as the number of paths increases during the operation, It is also possible to train the model to improve for specific paths within a sequence.

[0370] Using a separate model for each pass means, for example, as part of a welding repair job, This also applies to using separate models for repassing. The weld groove in such a repassing case... Its shape, and therefore its acoustics, are significantly different from those of a normal path.

[0371] (Laser sensor type) The next sensor type used is visual, which evaluates the shape of the created weld. The purpose is to do so.

[0372] Figure 20a shows a cross-section of the base material 2 on which the welded joint 20 is formed. In this example, the welded joint 20 is straight Although linear, other welding paths can be considered similarly. The visual sensor device 22 provides illumination. A casing is provided with a light source 26 inside that can illuminate the base material 2 and the welded area 20 over a width of 28. Includes g24. Device 22 has an operating range 30 that enables accurate imaging. Light is used. Returning to vice 22, there the receiver 32 focuses the light onto the sensor matrix 34, and A number is generated. In this example, a 2D laser profile scanner is used. It can also be replaced with other types.

[0373] The laser, acting as the light source 26, illuminates the weld and various details around it, such as the weld plate. Lofile, remaining groove profile, weld bead width, and material deposited outside the weld groove. It is used to reveal the weld groove. The inspection is performed perpendicular to the substrate surface next to the weld groove. A plane perpendicular to the vertical axis can be used. Furthermore, the weld bead profile along the weld groove can also be considered. It is possible.

[0374] Figure 20b shows a typical weld groove in the process of a series of welding passes. Each welding pass is Add welds in a predetermined order [numbered] to the welds already present in the groove, thereby completing the overall weld. To construct. As can be seen from these, the welding path is such that if the welding is proceeding correctly, the welding This contributes to the predictable shape of the path itself and the predictable changes in the shape of the weld groove.

[0375] The signal from device 22 is used to profile the entire welding track at each position along the welding track. It can be used to form a file image. The actual profile is the expected profile. It is compared to the original, and the deviation can be recorded. The deviation is determined by the threshold Th profile It can be compared to Therefore, the profile data is based on the threshold Th profile If it is broken or finger If the data remains corrupted across a specified number of data points, a warning will be issued to the operator. It has the function to trigger the cessation of welding. This allows for real-time professional Identification of defect generation based on file signals can be provided. This analysis can be continuously provided, This can be provided for each pass of the lutipass welding approach. Effective geometric validation is provided. It can be done.

[0376] (Visual sensor type) In the next sensing area, a high dynamic range camera is used to capture the weld while it is still forming. Areas where welding has not been performed, areas where welding has been formed, and areas where the weld has solidified and cooled further. Capture images of the welding location, including the exact spot.

[0377] Figure 21 shows a series of images of this type collected from various welding locations. By combining this with conventional machine vision tools to process these, sys Tem is a visual anomaly caused by image changes or defects related to abnormal welding conditions. It can be detected. This involves processing a single image, or recent images and past parts and parts. By combining and processing multiple images from previous data from both sides, the result is achieved. It can be revealed.

[0378] In this area, one or more variables may be considered. For example, the size of the weld pool [width, rear [Length of the connection, length of the lead section], shape of the weld pool [elliptical, teardrop, etc.], temperature of the weld pool This can be considered along with the analysis of patterns, the shape of the deposited material, and visible anomalies. Using a profile of known decision parameters for a particular type of defect, the algorithm The live output from each part can be compared to known good values. These values ​​are specific to visual This could be a combination of features (presence or absence), numerical bands or thresholds, or classification. If it is determined that the parameters are exceeded, the situation that occurs in one or more or all of them will be Compare the target position to one or more or all of them, and use the deviation again to trigger a warning. You can either continue or stop welding.

[0379] (Voltage and current sensor types) The applied voltage affects the formation of the arc and the current within the arc. This, in turn, affects the power. This gives and therefore affects the dissolution rate of the substrate [and electrodes if consumed]. These are important variables for the quality of the weld. They have an impact, for example, on the size of the weld pool. It is a variable that gives an effect.

[0380] The welding voltage is also automatically adjusted to reflect the distance between the welding equipment and the substrate being welded. This needs to be continuously adjusted. This involves the known and fixed position of the base material and the welding device. The mounted robotic arm has variable but known XYZ positions, that is, the division of these two positions Based on separation.

[0381] A sensing system for monitoring voltage and current is an existing automatic voltage control approach. It is orders of magnitude faster than what is built into the system. The power system handles 500A. It is possible and expandable to over 1000A, and voltage between measurements at the nanosecond level. It also provides monitoring and current monitoring. Therefore, very detailed information on voltage and current can be obtained. Short-term fluctuations should also be considered.

[0382] To avoid the sensing / detection itself interfering with the output power performance, the output power is Instead, input voltage and current, and therefore input power, tend to be detected.

[0383] The data and the approach used for processing it are the same as in the case of the acoustic sensor type described above. It seems so.

[0384] Referring to Figure 22, the arc voltage is plotted as plot V against time. In this example, the welding process is shut down after 30 seconds, and the voltage returns to zero.

[0385] Figure 22 also plots the relationship between Gaussian amplitude x Gaussian center value and time. As you can see, in the initial period of the first 2 or 3 seconds, the values ​​in this plot are above the acceptable threshold. The quality is also high, raising concerns about the quality of the weld. After the first 2-3 seconds, and definitely after 8 seconds From this point onward, the plot falls well below the threshold, indicating high-quality welding for this set of variables. It is.

[0386] In addition to these variables, Figure 22 shows the case where argon is used as the shielding gas in small quantities. This is also included. The display shows the occurrence of argon deficiency by plotted point, and the vertical position of the plotted point The position also indicates the degree of argon deficiency.

[0387] (Overall process - continued) Returning to Figure 17, the details of the sensor type's operation and the consideration of its data were revealed. Therefore, the process and the data within it undergo a data adjustment step. Data adjustment is performed by the user. History retrieved from inputs and / or storage provided through the interface Provided based on data. Storage includes early data and / or data from the execution of this welding. or data from numerous previous welds performed by the system, and / or other systems It may contain data from the system [for example, data from the calibration process, etc.] All of these contribute to historical data and therefore can contribute to data adjustment.

[0388] The live analysis step can implement two different levels of processing.

[0389] The first level of processing provides synchronization of data types from different sensor types. This synchronization involves various data types from the various sensor types exemplified above. Furthermore, it directly and / or indirectly relates to the welding system and the resulting welds. It also supports any number of other sensor types that are introduced and used to measure key characteristics. It is possible.

[0390] A programmable logic controller (PLC) controls the sensor type within the system. By providing a master timestamp at the start of data collection, various data Used to temporarily synchronize types. The PLC controls each sensor type in the system. Each sensor continuously provides data, and the provided data is received at a constant rate. Continuously check that data is being collected. During the data collection process, periodically An additional master timestamp may be applied.

[0391] The master timestamp is a microphone that functions as a sensor type for acoustic analysis. Area scan camera providing input power analysis and vision system for drawing measurement during welding. Laser profile scanners that offer 3D profile sensor types All the collected data represents data from the same time, i.e., data from the same location within the weld. This could mean that it will be adjusted accordingly.

[0392] Importantly, the PLC also receives data from an incremental encoder that provides position data. The encoder can be attached to the substrate to be welded and / or the welding apparatus at that time. This indicates the physical location of the encoder. The encoder triggers data acquisition and the sensor. It provides a consistent correlation between data and the welding position of the components. In this case, the encoder The same master timestamp is applied to the data from. This is because the synchronized data The actual locations of the same revealed positions that can be simultaneously applied in the arrangement are known. This means that the data is synchronized in time and space.

[0393] This process combines sensor data collected from various locations on the component into a single unit. It can be post-processed into a data structure with a fixed structure, from raw sensor data to post-processed information. Furthermore, it is possible to correlate the output of the analysis with the physical location of the weld. This data The user display shows the operator the welding process based on the system output. It may become possible to adjust the settings.

[0394] The second level of processing is implemented by applying data reduction and / or machine learning. This second level of processing involves obtaining correlated data and then processing and analyzing it in the second level. (Using machine learning or other appropriate analytical methods) Send the raw data and sensor type level Correlate any number of features from both outputs of the analysis to welding defects. In practice, mechanical engineering A decision-making engine that supports learning is employed.

[0395] In this method, if a defect is suspected, the defect is determined to have acceptable characteristics such as size. Defects that are either internal or exceed acceptable characteristics and require recording or repair action. A general evaluation is conducted to determine whether or not this is the case. NDT-type sensing is performed. Then, the defects are directly measured by the imaging performed. The size and, if applicable, other characteristics are reported directly. However, the defect analysis is dissolved. If based on welding conditions occurring during the weld, potential defects are considered indirectly. The question being considered is whether these conditions are likely to lead to defects. In level processing, acceptable and / or unacceptable welds in this context This could improve the decision.

[0396] Referring to Figure 24, there are two different sensor types: Type A on the left and Type B on the right. The two sensor types are either one of the sensor types described here and / or other forms of sensors that provide data on the execution or results of welding. It could be.

[0397] Referring to the Type A sensor type, within area 800, it shall indicate an acceptable weld. There is a set of data points that have been reliably established. These are other sensing and / or Or if it can be revealed or modeled from test runs validated by NDT. It is possible.

[0398] Additionally, there are data points from welding, such as data point 802, and these data The tap point is clearly outside the acceptable range, while the data point indicates an acceptable weld. The threshold 804 that may be applied when first determining whether the weld is unacceptable It is also largely on the unacceptable side. Other data points such as data point 806 This exceeds the threshold and is considered unacceptable according to the definition of threshold 804.

[0399] Instances that are difficult to interpret include data points 808 and 810, and others in region 800. These are data points located outside the threshold but below 804. Based on a single sensor, Since these welds are below the threshold, they are deemed acceptable.

[0400] To make a complete decision, data points across multiple sensor types [802, 80 Gain is obtained by a second level of processing that takes into account [6, 808, 810]. Referring to the Type B sensor type, data point 802 revealed the tolerance. It is again present within the welding area 800. Similarly, data point 806 is greater than threshold 804. It exceeds the limit, and again, this sensor type alone indicates unacceptable welding.

[0401] Looking at data points 808 and 810, both are outside region 800. However, the threshold is less than 804. In a single-sensor type approach, for these edges, It is again considered an acceptable weld. However, in the second level of processing, multiple sensor ties Additional information is obtained by considering the position between the points.

[0402] Using a neural network, the position can be taken into account across multiple sensor types. Two approaches can be adopted. These two approaches can be used as alternatives to each other. They may be used in parallel with the other, or they may be used in series. ru.

[0403] In the first approach, labeled data is provided and supervised learning is performed. Data with a special tag can be obtained from one or both of two sources. First, especially processing... In the initial stages, for example during calibration or initial production welding, labeling The data can be obtained from experimental results. Therefore, continuing with the example in Figure 24, the point is The positions of both 808 and point 810 are flagged by the operator, and by the operator It may be possible to assess whether it is acceptable or unacceptable. The operator should consider and make a decision. Not only data from a single sensor type, but also that data [data points or This is a sequence of data points from multiple different sensor types with flagged data. Data is provided, allowing for decisions to be made based on more nuanced criteria. The results are labeled with data. Because it is used to add values, it is useful within the data pool in which the neural network learns. It is used. Human knowledge and interpretations are input into the neural network by the user's decision. And monitoring is provided.

[0404] A second method for providing labeled data is to utilize existing data libraries. This data is also labeled according to the decision, and the necessary operator knowledge is also provided. Knowledge is transmitted. The library is sent to a neural network classifier, and library data is generated. The processing location of the taset is revealed. The library uses data from multiple sensor types as a reference. The area is defined as the location of data points across inspections and results from multiple sensor types, and the processing time is defined as the processing time. The second level of the classifier can then be determined. Next, any data related to the classifier library data can be determined. Determine the classifier score for each data point, and then determine the data points to be examined and decided upon. The classifier can quantify the possible errors of the `int` [or a series of errors]. An approach based on Bayes' theorem can be used to quantify it.

[0405] The second method, using libraries, is used as the starting point for labeled datasets. Obtain. The first method is calibration in the actual welding system by operator call. Conducting a ration or test run may be used as an alternative from the outset. However, Using method 1, we added to the dataset of method 2, and studied the neural network. The practice is adapted from a more general welding system position to that specific welding system. It can be moved to that position.

[0406] Returning to Figure 24, the operator determined that data point 808 was sufficiently close to the tolerance area 800. While this may be deemed acceptable, data point 810 has thresholds across multiple sensor types. It may be judged as unacceptable because it is too close. As a result, the neural network may determine that it is within an acceptable range. Small adjustments can be made to the range 800 and / or threshold 804 [value or format] over time. As this type of decision is repeated over time, the tolerance range 800 and / or threshold 8 This could lead to more pronounced and optimized modifications to the boundary of 04. For example, the tolerance region 800 could be expanded. The threshold may be increased and / or become stricter. The same results as for datapoints 808 and 810 are corrected for datapoint 808. It is within the acceptable range, but in the case of data point 810, it exceeds the corrected threshold, These can be called acceptable and unacceptable.

[0407] As an example of such analysis in a real-world scenario, the left side shows the relationship between the types of visual image sensors. If one or more consecutive images suggest a problem with the proximity of the weld sidewalls, It's possible. If we consider the acoustic sensor type to be on the right side, it could suggest a problem with the proximity of the side wall, and This can confirm the overall judgment that the welding is unacceptable.

[0408] In the example above, the positions of one type of sensor and another type of sensor are referenced to determine the acceptable range. It determines whether it is possible or unacceptable, but the determination is more detailed than that, and depends on the welding conditions. It may also be necessary to determine the nature of the problem. Therefore, data from only one type of sensor is insufficient. Even if it is only suggested, the nature of the problem can be understood from the data of two types of sensors. .

[0409] One issue to note is that importing historical data from other welding conditions and environments is acceptable. This means it's not a strong starting point for a library of data for judging welding performance. This may affect the data of other welding operations or other welding environments. This is because there are various variables that affect the outcome. For example, in the case of a visual image sensor, the welding environment The lighting and lighting angle, the properties of the substrate, the angle and spacing of the welding torch, etc., all affect the data. For example, that data may not match data searched in a different environment with different lighting. It is possible.

[0410] As an alternative to this first approach which uses labeled data and supervised learning, Additionally, neural network clustering or grouping-based processing By using the function to perform [find similarities and / or anomalies in the data] This may reduce or avoid the need for library-type data and / or supervised learning. There are various methods such as K-means clustering, which determine the centroids of clusters. This allows us to establish a cluster and clearly determine the distance around it. Several other clustering methods can also be applied. Yes, it is possible. These can be used to determine the location of the acceptable region 800 and / or threshold 804. This can be clarified and corrected by increasing the amount of data and learning. It can be started without needing a blacklist.

[0411] As mentioned above, the first and second approaches are not only alternatives, They can be used in combination. Therefore, the first approach can be used with a neural network. After starting the workbook and progressing with the first approach, you will move on to the second approach. It can be joined together. Furthermore, it can be performed using welding systems other than those specifically considered. If the library is extended from the internal welding, in order to maximize the supplied data, The neural network learns simultaneously from both the first and second approaches. It is also possible to pool learning from welding systems that are similarly configured and operated. It is also possible.

[0412] Over time, especially when this method is used in production versions that involve large-scale welding, The amount of data, the accuracy of the evaluation, and the complexity of the various cases of data that can be evaluated successfully are all factors. It increases.

[0413] Results obtained from live data analysis and / or data reduction and / or machine learning All results, once obtained, are for future use or for the knowledge of the system and / or similar systems. To contribute to awareness, it may have copies that are sent to storage.

[0414] A key step in the analysis is to assign the data value or location to one of the data types. This involves comparing with the above threshold (threshold setting). An example of a threshold setting approach is exemplified below. Although it is provided in various sections regarding the specific sensor type, each sensor It is broadly applicable to types and the data types they generate. The position relative to the threshold is a defect ( It can be considered to indicate a defect (defect indicator) and can be an important part of the welding quality result. It is displayed to the user via the user's display, and this also receives the received sensor data. Display it.

[0415] Figure 23 shows an example of a user display. The user display uses all sensors. It can provide real-time output of type analysis, so you can check it while the welding process is running. Furthermore, if poor quality welds are found in the analysis output, the process can be adjusted. After each pass, review the overall welding data and make any necessary repairs or adjustments before or during the next weld. It is also possible to examine the data to identify the necessary areas.

[0416] In the welding quality results step, if an unacceptable weld is encountered, the welding process may be stopped. , and / or the operator may be alerted via a user display or the like.

[0417] Based on the position of the user display, the operator or the system itself will determine the welding professional One or more variable parameters used to control and implement the process in real time It can be adjusted.

[0418] The welding quality results are reflected in the welding qualification assessment provided for the entire weld. Therefore, if a defect is identified, its size and location will be compared against appropriate standards, and its level The defect is then checked to ensure it is within acceptable limits. If it is not within acceptable limits, repair action is taken. If within the acceptable range, 3D lifetime record data is stored within the weld qualification evaluation, and the weld section It will be available for use throughout its lifecycle and for necessary decommissioning support.

[0419] (Data processing and storage) In addition to considering each data type individually, the data types are combined into a single data structure. Further advantages can be gained by considering it as a ream. To make this possible, each day A timestamp is added to the taset, and by synchronizing the timestamp, all data Synchronize the time of the taset.

[0420] In other words, all datasets can be combined and saved as a single data file. This also makes subsequent reuse possible.

Claims

1. A method for providing inspection of welded joints, (a) A step of preparing a welding inspection device in close proximity to the weld on the substrate to be inspected, (b) A step of performing an inspection, wherein the substrate is heated during the inspection by the surrounding The subject is subjected to a high temperature exceeding the temperature limit, and the aforementioned inspection is carried out by, a. Emitting ultrasonic waves into the volume of the substrate and the welded area, b. Receiving at least a portion of the ultrasonic waves from the substrate and the welded portion, thereby Acquiring multiple signal sets, The steps for conducting the test include, (c) Steps that process one or more sets of signals to provide welding inspection data P and, Includes, The process includes correcting the temperature distribution within the volume of the substrate and / or weld at high temperatures. nothing, A method for providing inspection of welded joints.

2. The correction corrects the path of at least a portion of the ultrasonic waves into the volume of the substrate and the welded area. The method according to claim 1, further comprising providing a correction path.

3. A portion of the ultrasound enters the first element of the substrate within the medium element before the portion of the ultrasound enters the first element of the substrate. The path characteristics are such that the first element has the temperature of the volume of the substrate and / or welded part. A path having a temperature distribution and a temperature-compensated path for a portion of the ultrasound within the first element. The characteristics are determined, and the temperature-corrected path characteristics are between the medium element and the first element. The method according to claim 2, based on temperature changes.

4. A portion of the ultrasonic waves has path characteristics in the first element of the substrate, and the first element The element has a temperature within the temperature distribution relative to the volume of the substrate and / or welded part, and the path The characteristic is that a portion of the ultrasonic path is guided to a second element, and the second element is the substrate and The temperature within the temperature distribution relative to the volume of the weld and / or the welded area is supersonic in the second element. A temperature-corrected path characteristic is determined for a portion of the wave, and the temperature-corrected path characteristic is Based on the temperature change between the first element and the second element, according to claim 2 or 3 method.

5. The temperature-compensated path characteristics are such that the ultrasonic waves in the substrate and / or welded area The method according to claim 4, which is determined for each element through which a portion passes.

6. The aforementioned temperature change is due to the change in the speed of sound between the speed of sound in one element and the speed of sound in the next element. The method according to any one of claims 3 to 5, expressed as a chemical.

7. Multiple different portions of the ultrasonic waves into the volume of the substrate are corrected to provide a corrected path. The method according to any one of claims 1 to 6.

8. The above method involves selecting a region of interest, and the region of interest is the substrate through which the ultrasonic waves have passed. and within the volume of the weld, the region of interest is subdivided into locations such as pixels, The method according to any one of claims 1 to 7, wherein a correction is applied to a location such as a lu.

9. The method described above is based on a calculation or observation correction regarding the location through which the emitted beam passes. Therefore, for one or more locations, such as pixels, through which the emitted ultrasonic beam does not pass. The method according to any one of claims 1 to 8, including the correction.

10. The correction applies to multiple locations on the first emission beam and multiple locations on the second emission beam. The method according to claim 9, based on a calculated or observed position.

11. The correction for the aforementioned location is applied to one or more other locations, for example, one or more locations through which the emitted beam has passed. Claims 8, 9, or 10 are weighted combinations of corrections for other locations. Methods used.

12. The correction for the aforementioned location is made between the location on the first beam and the location on the second beam. any one of claims 8 to 11, which is weighted according to the proportion of the distance at which the location exists. The method described in item 1.

13. The correction for the aforementioned locations is a first location on the first beam and a second location on the first beam. Between and the, weighted according to the proportion of the distance at which the location exists, from claim 8 The method described in any one of the twelve items.

14. The result set is one or more shapes of the substrate and / or the weld groove and / or weld. The method includes a measured representation of the shape and a representation of the modeled shape. Further including a comparison with the display, the display of the measured shape and the display of the modeled shape By comparing the measured shape, the representation of the measured shape is sufficiently suitable for representing the modeled shape. If it becomes clear that they match, the imaging of the region of interest is accepted, claim 1 The method described in any one of the items 13.

15. The result set is one or more shapes of the substrate and / or the weld groove and / or weld. The method includes a measured representation of the shape and a representation of the modeled shape. Further including a comparison with the display, the display of the measured shape and the display of the modeled shape By comparison, if the measured shape representation does not match the modeled shape representation If it is found to be sufficient, the body of the substrate and / or weld at high temperature Any of claims 1 to 14, the temperature distribution used to correct the temperature distribution within the product is redetermined. The method described in item 1.

16. The method provides a thermal model and, for at least a portion of the substrate, the thermal Using the model, generate the modeled temperature distribution location in the high-temperature state. The following is a description of any one of claims 1 to 15, which includes, and a portion of the base material includes the volume method.

17. The above method involves measuring the temperature at multiple locations using the substrate at a high temperature. This includes obtaining the temperature distribution location, and the measured temperature distribution location is modeled The method according to claim 16, further comprising comparing with the temperature distribution location.

18. By comparing the measured temperature distribution location with the modeled temperature distribution location, The modeled temperature distribution does not adequately fit the measured temperature distribution. If this is revealed, the thermal model and / or the modeled temperature distribution location The method according to claim 16 or 17, further comprising modifying and recomparing.

19. By comparing the measured temperature distribution location with the modeled temperature distribution location, It is clear that the modeled temperature distribution fits well with the measured temperature distribution. If this occurs, the supercharged material released during passage of at least a portion of the substrate and / or welded area The method according to claim 16, 17, or 18, further comprising calculating the characteristics of sound waves.

20. A device for inspecting welded joints, (a) Provided in close proximity to the welded portion on the substrate to be inspected during use, and during inspection the substrate A welding inspection device is provided, which is subjected to a high temperature exceeding the ambient temperature by being heated, (b) The welding inspection device is a. Welding by emitting ultrasonic waves into the volume of the base material and the weld area, b. Receiving at least a portion of the ultrasonic waves from the substrate and the welded portion, thereby multiple To obtain the signal set, It conforms to, The welding apparatus includes one or more processors, which are for one or more sets of signals. The processor includes an input, and the processor receives one or more input signals The processor provides corrections to the set and provides corrected welding inspection data, Apply a correction to the temperature within the volume of the substrate and / or weld at high temperatures, weld A device that provides testing.

21. A welding method, (a) Prepare welding equipment, (b) Prepare multiple sensor types, (c) Defining a first set of welding conditions for the welding method, (d) Introducing one or more substrates to be welded into the welding apparatus, (e) Perform welding of one or more base materials, (f) Acquiring data from the multiple sensor types during welding, (g) Take from at least two selected sensor types among the plurality of sensor types. To provide correlations between the obtained data, (h) data obtained from one of the at least two selected sensor types One or more data points to another one of the at least two selected sensor types To synchronize with the above data points, (i) Data obtained from the plurality of sensor types, one or more of the sensor types Comparing with reference data, (j) Based on one or more comparisons, the weld is of acceptable quality or unacceptable quality. Determining whether it is of quality, Includes, (k) If the quality of the weld is unacceptable, the method performs one or more operations. Including doing, Welding method.

22. The correlation is a temporal correlation and a spatial correlation, as described in claim 21. Law.

23. The method according to claim 21, wherein the correlation is a temporal correlation.

24. Data from at least two selected sensor types, one or more data in the data The correlation includes the time of occurrence of the tappoint, and the correlation is based on at least two selected sensor times. The matching of the occurrence times of data points in the data from the program, according to claims 20 to 23. The method described in any one of the items.

25. The occurrence time is determined by each of the two or more selected sensor types that provide the correlation. For each of these, the timestamp introduced into the data from the selected sensor type is used as the basis. The method according to claim 24, obtained as follows.

26. The occurrence time is obtained directly from the timestamp at that time, as described in claim 25. method.

27. The occurrence time is determined by the elapsed time from the timestamp and the time of the data point. The method according to claim 25 or 26, which is obtained as follows.

28. The method according to claim 21, wherein the correlation is a positional correlation.

29. The data from the at least two selected sensor types is one of the data. The correlation includes the locations where more than one data point originates, and the correlation is at least two selected Claim 2 is the matching of the generation locations of data points in the data from the sensor type. The method described in any one of items 1 to 28.

30. For each of the two or more selected sensor types that provide the correlation, Based on the position stamp introduced into the data from the selected sensor type, the generated position The method according to claim 24, wherein the following is obtained.

31. The aforementioned location is obtained directly from the location stamp of that location, as described in claim 25. Law.

32. The occurrence location is determined by the elapsed time from the location stamp and the time of the data point. The method according to claim 25 or 26, thereby obtained.

33. The method synchronizes data from the at least two selected sensor types. A comprehensive data structure is proposed in which the points are aligned with each other in terms of time and / or location. The method according to any one of claims 21 to 32.

34. The data points from the at least two selected sensor types are determined by the time of occurrence and Claims 21 to 3, which are displayed to the user in an aligned state with respect to the location of occurrence and / or generation. The method described in any one of item 3.

35. Data from the two or more selected sensor types and the at least two selections Correlation data from the selected sensor type is displayed to the user and / or stored. The method according to any one of claims 21 to 34.

36. Determination of whether the welding is of acceptable or unacceptable quality, / or one or more actions performed are displayed and / or saved, as described in claim 35. Method of loading.

37. The above method involves selecting at least four of the plurality of selected sensor types. Claims 21 to 3 include providing correlations between data obtained from survey types. The method described in any one of item 6.

38. Of the multiple selected sensor types, at least one selected sensor type is melting It is part of a welding inspection device, and the method involves inspecting the welded part using the welding inspection device. The method according to any one of claims 21 to 37, comprising the step of [doing something].

39. The welding inspection device is used to inspect the welded portion and determine one or more characteristics of the defect. The method according to claim 38, including a step.

40. The aforementioned characteristics are determined by referring to the shape of the weld and / or the length of the weld. Size, location, type of defect, and defect relative to the shape of the part and / or the length of the welded part. The method according to claim 39, comprising one or more of the shapes or locations of defects.

41. The method further includes comparing one or more characteristics with one or more criteria, and having the defect. The determination of whether the aforementioned welded joint meets the welding standards or does not The method according to claim 39 or 40, including the method described above.

42. If the welded joint meets the welding standards, a record of the welded joint is created and stored. The method according to claim 41.

43. The record further includes data from one or more of the plurality of sensor types. The method according to claim 42, which includes the method described in claim 42.

44. If the welded portion does not meet the welding standards, one or more repair steps are performed on the welded portion The method according to claim 41, applicable to the present invention.

45. Of the plurality of selected sensor types, at least two selected sensor types The type is a welding condition sensor, and the method uses the welding condition sensor to determine the welding condition The method according to any one of claims 21 to 44, comprising the step of inspecting the condition.

46. In order to determine one or more parameters of the weld when the weld is formed, The method according to claim 45, further comprising the step of inspecting the welding condition.

47. The process further includes comparing one or more parameters with one or more control parameters, and includes welding defects. The method according to claim 46, including determining whether the risk level of failure is exceeded.

48. The above method involves changing the welding conditions from the first set of welding conditions of the welding method. The method according to any one of claims 21 to 47, further comprising one or more operations.

49. Changing the welding conditions from the first set of welding conditions means stopping the welding and / The method according to claim 48, or the method according to claim 48, which involves warning the operator.

50. Changing the welding conditions from the first set of welding conditions is equivalent to changing the welding conditions from the first set of welding conditions. To return to the welding condition set, and / or to change the welding conditions to a second welding condition set. The method according to claim 48 or 49, wherein the method is to do the above.

51. At least one of the two selected sensor types is a voltage sensor, Current sensors, welding arc noise emission sensors, welding topology sensors, welding image sensors, and ultra The method according to any one of claims 21 to 50, selected from an acoustic image sensor.

52. A device for monitoring welding, wherein the device is (a) Multiple inputs for data from multiple sensor types, (b) One or more processors, and one of the one or more processors Ssa is, a. Receive input, b. At least two selected sensor types from the plurality of sensor types The acquired data is processed, and at least two selected from the plurality of sensor types are selected. Apply the correlation between the data obtained from the selected sensor type, c. Using the correlation described above, select at least two of the multiple sensor types. The data obtained from the selected sensor type is processed, and the at least two selected One or more data points in the data acquired from one of the sensor types, One of the data obtained from at least two other selected sensor types Synchronize with the above data points. One or more processors, (c) One or more outputs for the processed data, A device for monitoring welding, equipped with the necessary components.

53. The processor is adapted to provide temporal correlation, and prior to the processor The output data is from one of the at least two selected sensor types. The time information in the aforementioned data and the at least two selected sensors in the aforementioned data Claim 52 or the same, which includes time data in the data from another one of the IPs. The apparatus according to any one of the claims dependent on the above.

54. The processor is adapted to provide a positional correlation, and is located prior to the processor. The output data is from one of the at least two selected sensor types. The location information in the data and the at least two selected sensor types in the data Claim 52 or thereof includes position data in the data from another one of the P The apparatus according to any one of the dependent claims.

55. The processor processes the data from the at least two selected sensor types. In this, it is adapted to provide the occurrence time of one or more data points in the said data. The correlation is based on the data from the at least two selected sensor types. The matching of the occurrence times of the data points, as per claim 52 or any dependent claim. The apparatus described in any one of the items.

56. The processor processes the data from the at least two selected sensor types. In this, it is adapted to provide the location of one or more data points within the data. The correlation is based on the data from the at least two selected sensor types. The occurrence location of the data points is a match, as per claim 52 or any dependent claim. The apparatus described in any one of the items.

57. The input is at least one of the plurality of sensor types, and optionally the selected It is from at least one of the selected sensor types, and the sensor type and / or The selected sensor types are voltage sensors, current sensors, welding arc noise emission sensors, and welding to A choice selected from a porosi sensor, a welding image sensor, and an ultrasonic image sensor, as in claim 52. The apparatus described in any one of items 56.

Images