Strip rolling detection sensor online calibration method

Abstract

The application discloses a strip rolling detection sensor online calibration method, which comprises a static calibration method and / or a dynamic calibration method. The method mainly sets corresponding static calibration parts and / or dynamic calibration parts, uses a detection sensor to online detect the calibration size, analyzes the detection data, obtains the strip rolling detection sensor online calibration result, and determines the sensor working state and performance according to the mean value and variance of the deviation between the actual measured value and the nominal value. The calibration size comprises the width and / or thickness of the strip, the static calibration part comprises not less than five strip-shaped bands with different calibration sizes, the dynamic calibration part comprises at least one strip with an accurately known mean value of the calibration size, and the total length of the dynamic calibration part is not less than 5L0, wherein L0 is the minimum length of the cut strip in the next cutting process of the strip rolling process. The application solves the problems of inconvenient offline calibration and low efficiency of the high-precision detection sensor for the width and thickness of the metal rolling strip.

Description

Technical Field

[0001] This invention belongs to the field of metal rolling technology, specifically relating to an online calibration method for strip rolling detection sensors. Background Technology

[0002] With the development of precision machining technology, some metal strips, in addition to the traditional requirements for shape and structural strength, also have high requirements for dimensional accuracy during the rolling process. For example, the rolling of metals such as titanium alloys, aluminum alloys, and stainless steel used in the processing and production of mobile phone frames, card trays, precision medical devices, precision automotive parts, and precision smart home parts, as well as the rolling of composite metals such as steel-aluminum, titanium-aluminum, and magnesium-aluminum, all require high dimensional accuracy.

[0003] For example, a well-known mobile phone brand requires that the rolling thickness error of the titanium alloy mobile phone frame strip not exceed 0.03mm. This is a very high requirement for traditional metal rolling processes, especially considering that the random errors in dynamic detection can exceed the specified requirements due to the movement and vibration of the rolling equipment's drive system. Therefore, this places high demands on the detection sensors and their installation and operating environment. Generally, the detection accuracy requirement for sensors is better than half of the error specification, and in actual equipment purchases, the sensor's detection accuracy requirement is better than one-third of the error specification.

[0004] For such high-precision online detection, the accuracy is limited not only by the inherent precision of the technology itself but also by factors such as the detection environment, sensor condition, and installation environment. Different installation locations can all affect the actual accuracy. When an anomaly occurs, the first step is to rule out sensor malfunction. Removing the sensor for testing at this point is not only cumbersome and inefficient, but also significantly reduces the usability and reliability of offline test results due to differences in the online environment. Therefore, a method for online sensor calibration is needed.

[0005] In particular, after equipment maintenance, repair, and production line adjustment and installation are completed, it is also necessary to determine the accuracy of the sensors online. Summary of the Invention

[0006] To address the above problems, this invention designs an online calibration method for strip rolling detection sensors. By designing calibration devices and operating procedures, it achieves online calibration of high-precision online detection sensors for the width or thickness of metal strips.

[0007] This invention discloses an online calibration method for a strip rolling detection sensor. The calibration method includes a static calibration method and / or a dynamic calibration method. By setting corresponding static calibration components and / or dynamic calibration components, the detection sensor is used to perform online detection of the calibration dimensions of the calibration components, and the detection data is analyzed to obtain the online calibration result of the strip rolling detection sensor, that is, to obtain the accuracy performance of the detection sensor (including systematic error and / or random error, i.e., accuracy and / or precision). The calibration dimensions include width and / or thickness.

[0008] The static calibration component includes no less than 5 strips with different calibration sizes. Each static calibration component is used to calibrate the same calibration size within the detection section (i.e., the width and / or thickness are consistent within the detection section), and the deviation fluctuation range is less than 1 / 3 of the product error index or even less than 1 / 3 of the sensor's nominal measurement accuracy.

[0009] The dynamic calibration component includes at least one strip with a precisely known average calibration dimension. The total length of the dynamic calibration component is not less than 5L0, where L0 is the minimum segment length to be cut in the next cutting process of the strip rolling process. The dynamic calibration component includes at least one calibration segment, the length of which is not less than L0 and is generally an integer multiple of L0. The calibration dimensions of the calibration components within the calibration segment are consistent, and the deviation fluctuation range is less than 1 / 3 of the product error index or even less than 1 / 3 of the sensor's nominal measurement accuracy. Different calibration segments are set with different average calibration dimensions, and the variance can also be set with different values.

[0010] Furthermore, the static calibration method includes placing static calibration parts one by one onto the production line testing device, obtaining the calibration dimensions (including width and / or thickness) of each calibration part, and recording the measured value of the i-th static calibration part as h. i The corresponding nominal value is denoted as h. i0 For all h i -h i0 Find the statistical mean and variance σ h The static statistical mean This refers to the static calibration systematic error of the sensor (or its zero value; the direct measurement value in subsequent engineering testing should be subtracted from this zero value to obtain the size of the target product), and its variance σ. h This refers to the total random error of the sensor's static measurement. The random error of the sensor's static measurement to be calibrated is...

[0011]

[0012] Where σ0 is the variance of the natural dimensions (true value of width or thickness) of the static calibration part;

[0013] After the mean and variance statistics are completed, the sensor performance is determined: if σ J >σ max If the sensor's performance is deemed to be out of specification, its installation structure and operating environment need to be checked. If no abnormalities are found, the sensor should be removed for individual inspection and maintenance or handled by a professional. σ max To set a threshold.

[0014] Furthermore, the static calibration method also includes a method for determining the consistency of calibration results of the same static calibration component in multiple consecutive calibrations. If the fluctuations (fluctuations, if the deviation curve is plotted as the peak-to-peak value of the curve, outliers can be eliminated first) of the deviations from the nominal value in multiple measurements exceed a certain set threshold, then the sensor or the detection equipment is considered to be abnormal.

[0015] Furthermore, the dynamic calibration method includes the following steps:

[0016] S1. Place the dynamic calibration part on the production line detection device. The equipment is started and running normally, but there is no rolling operation. The dynamic calibration part passes through the detection sensor at the linear speed during normal production.

[0017] S2. The sensor performs sampling and testing on the dynamic calibration component according to the normal detection cycle;

[0018] S3. Perform segmented detection data analysis and processing, and record the i-th measured value of the j-th segment as h. j,i The nominal value of the calibration dimension corresponding to the j-th segment is denoted as h. j0 For all h j,i -h j0 Find the statistical mean and variance σ j For all Calculate the average again The zero value of the sensor's dynamic measurement can be obtained, and the dynamic statistical mean can be obtained. This refers to the dynamic calibration system error of the detection sensor; the variance σ j That is, the total random error of the dynamic measurement of the j-th segment of the sensor, then the random error of the dynamic measurement of the sensor to be calibrated Where N is the total number of segments in the dynamic calibration component, σ j0 This is the variance of the natural size of the j-th segment of the dynamic calibration part. Generally, when the size consistency within the segment is very good, it can be ignored, i.e., the default value is 0. However, when the calibration part is deliberately made to have certain fluctuations in size, i.e. the variance is artificially set, it needs to be considered.

[0019] S4. Sensor performance determination: If σ D >σ max If σ is found to be above the limit, then the sensor performance is considered to be out of specification. max The settings are the same as the sensor performance determination method in the static calibration method.

[0020] Furthermore, the sampling detection includes setting a sampling interval ΔL or sampling period Δt for calibration detection, satisfying Δt=ΔL / V, where V is the linear velocity output by strip rolling, and there are no less than 10 sampling points in each segment.

[0021] Furthermore, the dynamic calibration method also includes a method for judging the consistency of multiple calibration results for the same segment after multiple consecutive calibrations of the same dynamic calibration component. If the fluctuation (fluctuation, if the deviation curve is plotted as the peak-to-peak value of the curve, outliers can be eliminated first) of the multiple measurement results deviating from the nominal value exceeds a certain set threshold, then the sensor or the detection equipment is considered to be abnormal.

[0022] Furthermore, after the unidirectional motion sampling and detection of the dynamic calibration component is completed in step S2, the reverse motion sampling and detection is performed (the starting end of the motion is changed from one end to the other end). Subsequent data processing still processes the data of each segment, except that the amount of data for the same segment increases. Of course, the mean and variance statistics can also be performed on the forward and reverse sides of the same segment, and then the consistency of the forward and reverse motion results can be compared.

[0023] Furthermore, the dynamic calibration component has the same calibration mean and variance set for different segments.

[0024] Furthermore, outlier values ​​of the actual detected values ​​are removed before step S3.

[0025] The advantages and beneficial effects of this invention are as follows: The online calibration method for strip rolling detection sensors designed in this invention, by setting specific calibration components and conducting online detection of the calibration components, determines whether the sensor's status and / or measurement accuracy meet the requirements for online strip rolling detection. For situations where high-precision sensors for width and / or thickness detection in metal strip rolling experience abnormal actual detection results on continuous production lines (such as large changes in measured values, making it difficult to determine in a short time whether the cause is the strip being measured or the monitoring device), production line adjustments or modifications, or sensor replacements or maintenance, this method can assist in analysis and judgment, thereby reducing the workload of offline sensor calibration and improving the efficiency of equipment and production line status confirmation. Especially for high-precision online detection, when sensors malfunction or their accuracy decreases, offline laboratory testing or calibration using more advanced measurement methods is time-consuming, labor-intensive, inconvenient, and inefficient. Furthermore, the online results still have a certain degree of uncertainty because changes in the detection environment significantly affect the accuracy of high-precision sensors; static accuracy or laboratory accuracy is difficult to represent the accuracy under actual working conditions. The online calibration method of this invention perfectly solves this problem. Attached Figure Description

[0026] Figure 1 This is a flowchart of the dynamic calibration method steps. Detailed Implementation

[0027] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings and examples. The following examples are only used to more clearly illustrate the technical solutions of the present invention and should not be construed as limiting the scope of protection of the present invention.

[0028] Example 1

[0029] This invention discloses an online calibration method for a strip rolling detection sensor. The method employs a static calibration approach, which involves setting up corresponding static calibration components and using a detection sensor to perform online detection of the calibration dimensions of these components. The detection data is then analyzed to obtain the online static calibration results of the strip rolling detection sensor. This method eliminates the need for disassembling the sensor for offline calibration, thus obtaining the sensor's static accuracy performance (including systematic and / or random errors, i.e., accuracy and / or precision). The calibration dimensions include width and / or thickness.

[0030] The static calibration components include no fewer than five strips with different calibration dimensions. Each static calibration component is used to calibrate a test section with consistent calibration dimensions (i.e., consistent width and / or consistent thickness within the test section). The deviation fluctuation range is less than 1 / 3 of the product error index or even less than 1 / 3 of the sensor's nominal measurement accuracy (this requirement applies to the manufacturing of the calibration components; for a sensor with an accuracy of 1µm, a processing dimension fluctuation of less than 1 / 3µm for the static calibration components is quite high and not very necessary in actual engineering). To ensure the reliability of the static calibration results and meet statistical requirements, the number of static calibration components is generally no less than five. In this embodiment, ten static calibration components with different widths and thicknesses but precisely known lengths are customized. Generally, the length of each calibration component is not less than the minimum segment length L0 cut in the next cutting process of strip rolling. Each static calibration component is used for calibration at least once.

[0031] Preferably, the static calibration method includes placing static calibration parts one by one onto the production line testing device, obtaining the calibration dimensions (including width and / or thickness) of each calibration part, and recording the measured value of the i-th static calibration part as h. i The corresponding nominal value is denoted as h. i0 For all h i -h i0 Find the statistical mean and variance σ h The static statistical mean This refers to the static calibration systematic error of the sensor (or its zero value; the direct measurement value in subsequent engineering testing should be subtracted from this zero value to obtain the size of the target product), and its variance σ. hThis refers to the total random error of the sensor's static measurement. The random error of the sensor's static measurement to be calibrated is...

[0032]

[0033] Where σ0 is the variance of the natural dimensions (true value of width or thickness) of the static calibration part. In engineering, the natural dimension error is very small, that is, the consistency is very good, and it can generally be ignored, that is, σ0 is assumed to be 0. Alternatively, the measured values ​​can be directly statistically analyzed, and the final analysis conclusions are consistent. The dynamic calibration analysis method is the same as above.

[0034] After the mean and variance statistics are completed, the sensor performance is determined: if σ J >σ max If the sensor's performance is deemed to be out of specification, its installation structure and operating environment need to be checked. If no abnormalities are found, the sensor should be removed for individual inspection and maintenance or handled by a professional. σ max To set a threshold, a value of 3 times the sensor's nominal accuracy is typically used. In this embodiment, the sensor's nominal accuracy is 1µm, while the target control accuracy for the strip product is 0.03mm. Therefore, σ... max It can also be taken as 0.01 mm. The basic principle of this determination is that when σ J >σ max At that time, the deviation between the measured value and the nominal value of different calibration parts is relatively large, that is, the accuracy of the measured value is not high.

[0035] Example 2

[0036] The difference from Embodiment 1 is that the static calibration method also includes a method for determining the consistency of calibration results of the same static calibration component in multiple consecutive calibrations. If the fluctuation (fluctuation, if the deviation curve is plotted as the peak-to-peak value of the curve, outliers can be eliminated first) of the deviation between multiple measurement results and the nominal value exceeds a certain set threshold, the sensor or the detection equipment is considered to be abnormal. In this embodiment, the threshold is set to 10 times the nominal accuracy index of the sensor. The main purpose is to determine the sensor performance by checking the repeatability and random error of the measurement results. If there is a significant inconsistency, the specific cause needs to be investigated.

[0037] Example 3

[0038] The difference from Example 1 is that a dynamic calibration method is adopted. By setting up corresponding dynamic calibration components, the calibration dimensions of the calibration components are detected online using a detection sensor, and the detection data is analyzed to obtain the online dynamic calibration results of the strip rolling detection sensor. The static accuracy performance (including systematic error and / or random error, i.e., accuracy and / or precision) of the detection sensor can be obtained without disassembling the sensor for offline calibration detection. The calibration dimensions include width and / or thickness.

[0039] The dynamic calibration component includes at least one strip with a precisely known average calibration dimension. The total length of the dynamic calibration component is not less than 5L0, where L0 is the minimum segment length to be cut in the next cutting process of the strip rolling process. The dynamic calibration component includes at least one calibration segment, the length of which is not less than L0 and is generally an integer multiple of L0. The calibration dimensions of the calibration components within the calibration segment are consistent, and the deviation fluctuation range is less than 1 / 3 of the product error index or even less than 1 / 3 of the nominal measurement accuracy of the sensor. Different calibration segments are set with different average calibration dimensions, and the variance can also be set with different values. If the calibration dimensions are consistent, they are considered to be the same segment. In this embodiment, the dynamic calibration component is set to 5 segments, each segment with a length of 2L0, and nominal dimensions of 2.95mm, 2.95±0.005mm, and 2.95±0.01mm, respectively.

[0040] Preferably, the dynamic calibration method includes the following steps:

[0041] S1. Place the dynamic calibration part on the production line detection device. The equipment is started and running normally, but there is no rolling operation. The dynamic calibration part passes through the detection sensor at the linear speed during normal production.

[0042] S2. The sensor performs sampling and testing on the dynamic calibration component according to the normal detection cycle;

[0043] S3. Perform segmented detection data analysis and processing, and record the i-th measured value of the j-th segment as h. j,i The nominal value of the calibration dimension corresponding to the j-th segment is denoted as h. j0 For all h j,i -h j0 Find the statistical mean and variance σ j For all Calculate the average again The zero value of the sensor's dynamic measurement can be obtained, and the dynamic statistical mean can be obtained. This refers to the dynamic calibration system error of the detection sensor; the variance σ j That is, the total random error of the dynamic measurement of the j-th segment of the sensor, then the random error of the dynamic measurement of the sensor to be calibrated Where N is the total number of segments in the dynamic calibration component, σ j0 This is the variance of the natural size of the j-th segment of the dynamic calibration part. Generally, when the size consistency within the segment is very good, it can be ignored, i.e., the default value is 0. However, when the calibration part is deliberately made to have certain fluctuations in size, i.e. the variance is artificially set, it needs to be considered.

[0044] S4. Sensor performance determination: If σ D >σ max If σ is found to be above the limit, then the sensor performance is considered to be out of specification. maxThe settings are the same as the sensor performance determination method in the static calibration method.

[0045] Preferably, the sampling detection includes setting a sampling interval ΔL or sampling period Δt for calibration detection, satisfying Δt=ΔL / V, where V is the linear velocity output by strip rolling, and there are no less than 10 sampling points in each segment; that is, the sampling interval ΔL≦1 / 10L1, and setting no less than 10 sampling detection points in each segment is mainly to ensure the accuracy of statistical calculation, and the interval between each sampling point is generally uniform sampling; this setting is essentially carried out during real-time detection in the strip rolling process, and the setting during calibration is the same to ensure that the calibration and real-time detection equipment status are consistent.

[0046] Preferably, the dynamic calibration component has the same calibration mean and variance for different segments. That is, the actual width and / or thickness of the calibration component has the same average value for different segments, but the actual size of each point has random characteristics based on the average value, and the variance is fixed. That is, the width and / or thickness of each different part has a certain fluctuation value near the same mean value. In other words, the nominal value itself fluctuates, and the statistical variance of this fluctuation is the design variance of the corresponding mean value.

[0047] Generally, the nominal value of a calibration component is a constant value, without considering random variations. Statistical mean calibration is relatively intuitive. This embodiment incorporates a certain error factor in the design of the calibration component. To reflect the statistical effect or confidence level, according to statistical theory, the actual calibration sampling points should be no less than 100 points per calibration. This embodiment designs 400 points, so the length of the dynamic calibration component should be no less than 400Δt*V. Wherein, Δt is the sampling period, and V is the linear velocity output by strip rolling.

[0048] This invention employs categorized or segmented statistical analysis of the mean and variance of calibration components, summarizing the overall mean and variance. The key lies in designing suitable calibration components and determining sensor performance through statistical measures. Generally, calibrating an instrument to its zero value (or mean deviation, or accuracy) using calibration components is relatively easy. However, calibrating the random error (precision or accuracy) of instrument measurements is not easy, especially in online calibration. This is because the variance of calibration measurements can generally be equivalent to the total random error, which includes both sensor accuracy error and the random error of the calibration components (similar to state noise in cybernetics). Strictly separating the sensor accuracy from this is difficult. This invention achieves the goal of giving the calibration component dimensions random characteristics through the design of the calibration component dimensions (the calibration dimension for each sampling period is designed as "nominal value + random value"). The length of each dimension segment matches the sampling period. This calibration component requires high customization and generally necessitates precision machining tools.

[0049] Taking dynamic calibration as an example, the total random error after calibration is σ. DLet σ be the sensor accuracy error. X Let be the quantity to be determined, and let σ0 be the random error of the calibration component, which is a known quantity. Then we have:

[0050] In step S1 of the dynamic calibration method, the dynamic calibration piece passes through the detection sensor at the linear velocity during normal production. Generally, an auxiliary device can be used to make the dynamic calibration piece pass through the detection sensor at the same speed as the strip rolling, dynamically measuring its real-time width and / or thickness. The auxiliary device mainly avoids changes in the dimensions and other characteristics of the calibration piece caused by the output of the rolls, which would lead to inaccurate calibration data. A relatively convenient method is to use an existing output traction motor or to set a traction motor at the output end of the strip to pull the dynamic calibration piece through the detection sensor at a set speed. During the dynamic calibration process, in order to more realistically reflect the dynamic environment, the rolling equipment such as the rolls is turned on, but the strip is not actually rolled. If the length of the calibration piece exceeds the distance between the detection sensor and the rolls, the calibration piece can pass through the rolling mill, provided that the rolls do not apply rolling force to the calibration piece to avoid changing the dimensional characteristics of the calibration piece.

[0051] Preferably, outlier values ​​of the actual detection values ​​are removed before step S3.

[0052] Example 4

[0053] The difference from Embodiment 3 is that the dynamic calibration method also includes a method for judging the consistency of multiple calibration results for the same segment after multiple consecutive calibrations of the same dynamic calibration component. If the fluctuation (fluctuation, if the deviation curve is plotted as the peak-to-peak value of the curve, outliers can be eliminated first) of the deviation between the multiple measurement results and the nominal value exceeds a certain set threshold, the sensor or detection equipment is considered abnormal. The multiple measurement results include the measurement values ​​of multiple sampling points in the same segment during the same calibration. In this embodiment, the threshold is set to 15 times the nominal accuracy index of the sensor. The repeatability of the calibration measurement is detected. The specific method and steps are the same as S1 to S4. At the same time, the consistency judgment of multiple dynamic detection results for the same segment is added. If there is obvious inconsistency, the specific cause needs to be investigated.

[0054] Example 5

[0055] The difference from Example 3 is that after the unidirectional motion sampling and detection of the dynamic calibration component is completed in step S2, the reverse motion sampling and detection is performed (the starting end of the motion is changed from one end to the other end). The subsequent data processing is still to process the data of each segment, but the amount of data for the same segment is larger. Of course, the mean and variance statistics can also be performed on the forward and reverse sides of the same segment, and then the consistency of the forward and reverse motion results can be compared.

[0056] This invention determines whether the sensor's status and / or measurement accuracy meet the requirements for online strip rolling inspection by setting specific calibration components and performing online testing on these components. In situations such as abnormal actual test results (e.g., large changes in measured values, making it difficult to determine in a short time whether the cause is the strip being tested or the monitoring device), production line modifications and maintenance, or sensor replacement or maintenance, this invention can assist in analysis and judgment, thereby reducing the workload of sensor calibration and improving the efficiency of equipment status confirmation.

[0057] Example 6

[0058] The difference from Example 3 is that in this example, the dynamic calibration component is set as one segment with a total length of 10L0 and a nominal size of 2.95mm.

[0059] The basic principle of this invention is as follows: By designing static and dynamic calibration components and utilizing the known nominal values ​​of these components, the mean and variance of the deviation between the online detection value and the nominal value are statistically analyzed. The working state and accuracy characteristics of the sensor are then determined based on the mean and variance; alternatively, the consistency of multiple measurements can be used to determine if the sensor is malfunctioning. This allows for online calibration of high-precision width and / or thickness detection sensors in metal strip rolling on continuous production lines, accurately determining the sensor's working state. This avoids the inconvenience of sensor disassembly and testing, as well as the poor usability and reference value of offline testing results.

[0060] The above description is only a partial, relatively comprehensive embodiment of the online calibration method for strip rolling detection sensors of the present invention. In fact, there are many design forms for strip calibration components, including the number of calibration components, the setting of nominal values, and the end settings of dynamic calibration components, etc. On the other hand, there are also many calibration methods and data processing methods. For example, the present invention performs statistical analysis on the deviation between the measured value and the nominal value, or it can directly perform statistical analysis on the measured value, etc. These preferred solutions and combinations should also be considered within the scope of protection of the present invention, and will not be listed one by one here.

Claims

1. An online calibration method for a strip rolling detection sensor, the calibration method comprising a static calibration method and / or a dynamic calibration method, characterized in that, By setting up corresponding static calibration parts and / or dynamic calibration parts, the calibration dimensions of the calibration parts are detected online using a detection sensor, and the detection data is analyzed to obtain the online calibration results of the strip rolling detection sensor. The calibration dimensions include the width and / or thickness of the strip. The static calibration component includes no less than 5 strips with different calibration sizes. The calibration size is consistent within the detection section calibrated by each static calibration component, and the deviation fluctuation range is less than 1 / 3 of the product error index or less than 1 / 3 of the sensor's nominal measurement accuracy. The dynamic calibration component includes at least one strip with a precisely known average calibration dimension. The total length of the dynamic calibration component is not less than 5L0, where L0 is the minimum segment length to be cut in the next cutting process of the strip rolling process. The dynamic calibration component includes at least one calibration segment, and the length of the calibration segment is not less than L0. The calibration dimensions of the calibration components within the calibration segment are consistent, and the deviation fluctuation range is less than 1 / 3 of the product error index or less than 1 / 3 of the sensor's nominal measurement accuracy. Different calibration dimension averages are set for different calibration segments. The dynamic calibration method includes the following steps: S1. Place the dynamic calibration part on the production line detection device. The equipment is started and running normally, but there is no rolling operation. The dynamic calibration part passes through the detection sensor at the linear speed during normal production. S2. The sensor performs sampling and testing on the dynamic calibration component according to the normal detection cycle; S3. Perform segmented detection data analysis and processing, and record the i-th measured value of the j-th segment as h. j,i The nominal value of the calibration dimension corresponding to the j-th segment is denoted as h. j0 For all h j,i -h j0 Find the statistical mean and variance σ j For all Calculate the average again The zero value of the sensor's dynamic measurement can be obtained, and the dynamic statistical mean can be obtained. This refers to the dynamic calibration system error of the detection sensor; the variance σ j That is, the total random error of the dynamic measurement of the j-th segment of the sensor, then the random error of the dynamic measurement of the sensor to be calibrated Where N is the total number of segments in the dynamic calibration component, σ j0 Let be the variance of the natural dimension of the j-th segment of the dynamically calibrated part; S4. Sensor performance determination: If σ D >σ max If σ is found to be out of specification, then the sensor performance is considered to be exceeding the limit. max To set a threshold.

2. The online calibration method for a strip rolling detection sensor according to claim 1, characterized in that, The static calibration method includes placing static calibration parts one by one onto the production line testing device, obtaining the calibration dimensions of each calibration part, and recording the measured value of the i-th static calibration part as h. i The corresponding nominal value is denoted as h. i0 For all h i -h i0 Find the statistical mean and variance σ h The static statistical mean This refers to the static calibration system error of the detection sensor, and its variance σ h This refers to the total random error of the sensor's static measurement. The random error of the sensor's static measurement to be calibrated is... ； Where σ0 is the variance of the natural dimensions of the static calibration part; After the mean and variance statistics are completed, the sensor performance is determined: if σ J >σ max If σ is found to be out of specification, then the sensor performance is considered to be exceeding the limit. max To set a threshold.

3. The online calibration method for a strip rolling detection sensor according to claim 2, characterized in that, The static calibration method also includes a method for determining the consistency of calibration results of the same static calibration component in multiple consecutive calibrations. If the fluctuation of the deviation between the multiple measurement results and the nominal value exceeds a certain set threshold, the sensor or the detection equipment is considered to be abnormal.

4. The online calibration method for a strip rolling detection sensor according to claim 1, characterized in that, The sampling and detection includes setting a sampling interval ΔL or sampling period Δt for calibration detection, satisfying Δt=ΔL / V, where V is the linear velocity output by strip rolling, and there are no less than 10 sampling points in each segment.

5. The online calibration method for a strip rolling detection sensor according to claim 1, characterized in that, The dynamic calibration method also includes a method for judging the consistency of multiple calibration results for the same segment by calibrating the same dynamic calibration component multiple times. If the fluctuation of the deviation between the multiple measurement results and the nominal value exceeds a certain set threshold, the sensor or the detection equipment is considered to be abnormal.

6. The online calibration method for a strip rolling detection sensor according to claim 1, characterized in that, After the unidirectional motion sampling and detection of the dynamic calibration component is completed in step S2, the reverse motion sampling and detection is then performed.

7. The online calibration method for a strip rolling detection sensor according to claim 1, characterized in that, The dynamic calibration component has the same calibration mean and variance set for different segments.

8. The online calibration method for a strip rolling detection sensor according to claim 1, characterized in that, Before step S3, outlier values ​​of the actual detection values ​​are removed.

Images