Intelligent welding robot control method and system based on deep learning model

Abstract

The invention relates to an intelligent welding robot control method and system based on a deep learning model, and relates to the technical field of welding robos.The intelligent welding robot control method comprises the steps that image detection information generated when a workpiece is welded and welding parameter information of a welding gun are obtained; analyzing and determining a welding position point according to the welding parameter information; based on the welding position points, welding spot image information corresponding to the welding position points and original image information when welding is not conducted are determined from the image detection information; analyzing and determining quality demand information according to the original image information; welding spot quality information is analyzed and determined according to the welding spot image information; and quality adjustment control information is determined according to the deviation condition between the welding spot quality information and the quality demand information, and the quality adjustment control information is output to the welding robot for welding adjustment. The welding device has the effect of improving the welding efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of welding robots, and in particular, to an intelligent welding robot control method and system based on a deep learning model. Background Art

[0002] Welding robots are indispensable automated equipment in modern manufacturing. They can improve welding quality, production efficiency, and reduce labor intensity. Welding robots are widely used in fields such as automobile manufacturing, aerospace, electronic equipment, construction, and shipbuilding.

[0003] When using a welding robot for welding, the welding robot usually consists of a robot body, a welding power source, a controller, a sensor, and a tool head (welding torch), etc. The robot body is responsible for performing precise motion control, the welding power source provides the energy required for welding, the controller commands the actions of the robot according to programming instructions, the sensor is used to detect various parameters during the welding process, and the tool head is the part that directly contacts the workpiece for welding.

[0004] The welding robot reduces the alternating current to a low voltage and high current suitable for welding through its internal transformer. When the welding electrode contacts the workpiece to be welded, the circuit is temporarily closed, resulting in a sharp increase in current. Due to the small contact area and large contact resistance, a large amount of heat is generated to melt the solder on the welding electrode and the metal on the surface of the workpiece to be welded, so that the two are combined together after cooling. At this time, the welding robot moves the welding torch according to the preset program and path, so as to weld the workpiece to be welded. After welding, the quality of the solder joints in the workpiece is detected and then subsequent processing is carried out.

[0005] When using a welding robot to weld a workpiece, the welding robot needs to first weld the workpiece to be welded according to the preset program and path, and then detect and adjust the quality of the solder joints, resulting in a reduction in welding efficiency. Summary of the Invention

[0006] In order to improve welding efficiency, the present invention provides an intelligent welding robot control method and system based on a deep learning model.

[0007] In a first aspect, the present invention provides an intelligent welding robot control method based on a deep learning model, adopting the following technical solution:

[0008] An intelligent welding robot control method based on a deep learning model includes:

[0009] Obtain image detection information and welding parameter information of the welding torch when welding the workpiece;

[0010] Analyze and determine the welding position point according to the welding parameter information;

[0011] Determine the solder joint image information corresponding to the welding position point and the original image information when not welded from the image detection information based on the welding position point;

[0012] Analyze and determine the quality requirement information according to the original image information;

[0013] Analyze and determine the solder joint quality information according to the solder joint image information;

[0014] According to the deviation between the solder joint quality information and the quality requirement information, determine the quality adjustment control information, and output the quality adjustment control information to the welding robot for welding adjustment.

[0015] Optionally, the method for determining the welding position point includes:

[0016] Retrieve the moving position point of the welding torch and the corresponding output current value based on the welding parameter information;

[0017] Perform curve analysis based on the output current value to form a current change curve corresponding to the time change process;

[0018] Select the segment where the current change curve is consistent with the preset no-current reference curve as the no-current segment;

[0019] Remove the no-current segment based on the current change curve and connect the remaining parts to form a current curve;

[0020] Analyze the change situation according to the current curve to determine the working position point, and use the working position point as the welding position point.

[0021] Optionally, the method for determining the working position point includes:

[0022] Select the moving position point corresponding to the wave peak according to the current curve and use it as the suspected welding position point;

[0023] Retrieve the dwell time value corresponding to the suspected welding position point based on the welding parameter information;

[0024] Based on the dwell time value and the preset dwell reference time value, use the suspected welding position point corresponding to the dwell time value greater than the dwell reference time value as the dwell position point, and use the suspected welding position point corresponding to the dwell time value not greater than the dwell reference time value as the remaining position point;

[0025] Calculate the distance value between two adjacent remaining position points and use it as the adjacent position distance value;

[0026] According to the falling situation of the adjacent position distance value within the preset distance reference interval, determine the continuous working position points;

[0027] Both the stop position points and the continuous working position points are regarded as working position points.

[0028] Optionally, the method for determining the quality requirement information includes:

[0029] Identifying the original image information based on the preset workpiece surface reference features and taking the different part as the gap image information;

[0030] Analyzing and determining the gap depth value, the gap width value and the gap inclination direction information according to the gap image information;

[0031] Calculating the product value between the gap depth value and the gap width value and taking it as the gap volume value;

[0032] Querying and obtaining the specification of the welding rod currently in use and taking it as the welding rod specification information;

[0033] Analyzing and determining the initial required quality information according to the welding rod specification information and the gap volume value;

[0034] According to the corresponding relationship between the gap inclination direction information and the preset inclination direction influence information, determining the inclination direction influence information corresponding to the gap inclination direction information;

[0035] Adjusting the initial required quality information based on the inclination direction influence information to form the adjusted required quality information, and taking the adjusted required quality information as the quality requirement information.

[0036] Optionally, it further includes steps before calculating the gap volume value, specifically as follows:

[0037] Obtaining the environmental light information corresponding to the current time;

[0038] Retrieving the environmental brightness value and the environmental light direction information based on the environmental light information;

[0039] Analyzing and determining the depth brightness influence value corresponding to the environmental brightness value according to the corresponding relationship between the environmental brightness value and the preset depth brightness influence value;

[0040] Calculating the deviation angle between the environmental light direction information and the preset light reference direction information and taking it as the light direction deviation angle value;

[0041] Analyzing and determining the depth light direction influence value corresponding to the light direction deviation angle value according to the corresponding relationship between the light direction deviation angle value and the preset depth light direction influence value;

[0042] Calculating the sum value between the depth brightness influence value and the depth light direction influence value and taking it as the depth comprehensive influence value, and adjusting the gap depth value based on the depth comprehensive influence value to form a new gap depth value.

[0043] Optionally, the method for determining quality adjustment control information includes:

[0044] Analyze the deviation between the solder joint quality information and the quality requirement information and use it as the quality deviation information;

[0045] Retrieve the quality deviation type information and the type deviation value based on the quality deviation information;

[0046] According to the correspondence between the quality deviation type information and the preset deviation type reference evaluation value, determine the deviation type reference evaluation value corresponding to the quality deviation type information;

[0047] Calculate the product value between the deviation type reference evaluation value and the type deviation value and use it as the single type evaluation value;

[0048] Calculate the sum value of all single type evaluation values and use it as the comprehensive evaluation value;

[0049] Input the comprehensive evaluation value into the preset quality deviation adjustment model for analysis to obtain the comprehensive evaluation adjustment information, and use the comprehensive evaluation adjustment information as the quality adjustment control information.

[0050] Optionally, it further includes the steps after calculating the sum value of all single type evaluation values and using it as the comprehensive evaluation value, which are as follows:

[0051] Judge whether the comprehensive evaluation value is greater than the preset evaluation reference value;

[0052] If it is, continue to output the comprehensive evaluation value;

[0053] If it is not, analyze and determine the workpiece type information according to the image detection information;

[0054] Analyze and determine the welding requirement angle range according to the workpiece type information and the welding position points;

[0055] Retrieve the clamping angle value between the welding torch and the welding electrode and the rotation angle value of the welding torch based on the welding parameter information;

[0056] Analyze and determine the welding angle value according to the clamping angle value and the rotation angle value;

[0057] According to the falling situation between the welding angle value and the welding requirement angle range, determine the angle influence value, and add the angle influence value to the comprehensive evaluation value to form a new comprehensive evaluation value.

[0058] Optionally, it further includes the steps after adding the angle influence value to the comprehensive evaluation value to form a new comprehensive evaluation value, which are as follows:

[0059] Retrieve the type reference shape information based on the workpiece type information;

[0060] Determine the shape centroid position point by analyzing the image detection information based on the centroid of the type-based reference shape information;

[0061] Calculate the distance vector between the shape centroid position point and the welding position point and use it as the welding centroid vector value;

[0062] Retrieve the important influence position area based on the type-based reference shape information;

[0063] Determine whether the welding position point is within the important influence position area;

[0064] If it is, analyze and determine the centroid influence value based on the welding centroid vector value and the type-based reference shape information, and add the centroid influence value to the comprehensive evaluation value to form a new comprehensive evaluation value;

[0065] If it is not, continue to output the comprehensive evaluation value.

[0066] Optionally, the method for determining the centroid influence value includes:

[0067] Retrieve the type-based reference area value and the shape vector reference influence value based on the type-based reference shape information;

[0068] Retrieve the type-based reference weight value based on the workpiece type information;

[0069] Analyze and calculate the type-based reference area value, the shape vector reference influence value, the type-based reference weight value, the welding centroid vector value, and the preset welding point reference quality value according to the preset centroid influence value calculation formula to determine the centroid influence value, where the centroid influence value calculation formula is Q is the centroid influence value, a is the welding centroid vector value, σ is the shape vector reference influence value, g is the preset welding point reference quality value, s is the type-based reference area value, and m is the type-based reference weight value.

[0070] In a second aspect, the present invention provides an intelligent welding robot control system based on a deep learning model, adopting the following technical solutions:

[0071] An intelligent welding robot control system based on a deep learning model, comprising:

[0072] An acquisition module, configured to acquire image detection information, welding parameter information, electrode specification information, and ambient light information;

[0073] A memory, configured to store the program of the intelligent welding robot control method as described in the first aspect;

[0074] A processor, configured to load and execute the program in the memory and implement the intelligent welding robot control method as described in the first aspect.

[0075] In summary, the present invention includes at least one of the following beneficial technical effects:

[0076] 1. After determining the welding position points by obtaining welding parameter information, the obtained image detection information is analyzed to determine the solder joint image information and the original image information. After separately analyzing and determining the solder joint quality information and the quality requirement information, the quality adjustment control information is determined based on the deviation between the two and output to the welding robot for welding adjustment, so as to timely understand and process the quality of the solder joints, thereby improving the welding efficiency;

[0077] 2. The moving position points and the output current values are retrieved through the welding parameter information to form a current change curve. After selecting the currentless line segments and forming a current curve, the working position is determined by analyzing the change of the current curve and used as the welding position point, thereby improving the accuracy of the obtained welding position points;

[0078] 3. The welding suspected position points are selected through the current curve and the residence time values are retrieved. The residence position points and the remaining position points are selected and defined through the residence time values, and the adjacent position distance values are calculated. The continuous working position points are determined based on the falling situation of the adjacent position distance values within the preset distance reference interval, and both the residence position points and the continuous working position points are used as the working position points, thereby improving the accuracy of the obtained working position points. Description of the Drawings

[0079] Figure 1 is a flowchart of the method for controlling an intelligent welding robot based on a deep learning model according to an embodiment of the present application.

[0080] Figure 2 is a flowchart of the method for determining the welding position points according to an embodiment of the present application.

[0081] Figure 3 is a flowchart of the method for determining the working position points according to an embodiment of the present application.

[0082] Figure 4 is a flowchart of the method for determining the quality requirement information according to an embodiment of the present application.

[0083] Figure 5 is a flowchart of the method for the steps before calculating the gap volume value according to an embodiment of the present application.

[0084] Figure 6 is a flowchart of the method for determining the quality adjustment control information according to an embodiment of the present application.

[0085] Figure 7 is a flowchart of the method for the steps after calculating the sum of the evaluation values of all single types as the comprehensive evaluation value according to an embodiment of the present application. Detailed Embodiments

[0086] The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments.

[0087] An intelligent welding robot control method based on a deep learning model obtains image detection information, welding parameter information, electrode specification information, and ambient light information during welding, so as to detect the quality of the welding point in a timely manner and make timely adjustments. Moreover, the quality requirement information is adjusted according to the electrode specification and ambient light, thereby improving the accuracy of detecting the deviation of the welding point quality.

[0088] An embodiment of the present invention discloses an intelligent welding robot control method based on a deep learning model. Referring to Figure 1 , an intelligent welding robot control method based on a deep learning model includes:

[0089] Step S100: Obtain the image detection information during welding of the workpiece and the welding parameter information of the welding torch.

[0090] Among them, the image detection information refers to the image information for detecting the welding situation of the workpiece at the current time, and the image detection information is obtained after being detected by a camera preset on the intelligent welding robot. The welding torch refers to a device on the intelligent welding robot for clamping the electrode and outputting current and voltage for welding, and the welding parameter information refers to the information corresponding to each parameter on the welding torch of the intelligent welding robot during welding of the workpiece, and the welding parameter information is obtained by uploading through the intelligent welding robot.

[0091] Step S200: Analyze and determine the welding position point according to the welding parameter information.

[0092] Among them, the welding position point refers to the position point corresponding to the welding treatment of the workpiece. By analyzing the welding parameter information, the welding position point is determined according to the parameter change situation of the electric welding machine, which is convenient for subsequent use.

[0093] Step S300: Determine the solder joint image information corresponding to the welding position point and the original image information before welding from the image detection information based on the welding position point.

[0094] Among them, the solder joint image information refers to the image information of the area where the welding position point is located after the workpiece is welded, and the original image information refers to the image information of the area where the welding position point is located before the workpiece is welded. By selecting the pre-welding image and the post-welding image through the welding position point, and taking the images of the areas where the welding position point is located in the two time detection images as the solder joint image information and the original image information respectively, it is convenient for subsequent use.

[0095] Step S400: Analyze and determine the quality requirement information according to the original image information.

[0096] Among them, the quality requirement information refers to the minimum welding quality requirement information that the un-welded workpiece needs to achieve at the solder joints after welding. By analyzing the original image information, the quality requirement information is determined to facilitate subsequent use.

[0097] Step S500: Analyze and determine the solder joint quality information based on the solder joint image information.

[0098] Among them, the solder joint quality information refers to the actual welding quality requirement information that the solder joints of the workpiece achieve after welding. By analyzing the solder joint image information, the standard thresholds for the solder joint width, smoothness, and whether there are defects such as cracks and pores are identified based on the image, and the solder joint quality information is comprehensively formed to facilitate subsequent use.

[0099] Step S600: Determine the quality adjustment control information according to the deviation between the solder joint quality information and the quality requirement information, and output the quality adjustment control information to the welding robot for welding adjustment.

[0100] Among them, the quality adjustment control information refers to the control information corresponding to controlling the welding torch to adjust the quality of the solder joints of the workpiece just after welding. By analyzing the deviation between the solder joint quality information and the quality requirement information, and forming the quality adjustment control information based on the deviation between the actual quality of the solder joints and the required quality, and outputting the quality adjustment control information to the welding robot for welding adjustment, so as to automatically understand and process the quality of the solder joints at the welding position in a timely manner by using the images before and after welding, thereby improving the welding efficiency.

[0101] In Figure 1 the shown step S200, in order to further ensure the rationality of the welding position points, it is necessary to perform a further separate analysis and calculation on the welding position points. Specifically, it is described in detail through Figure 2 the shown steps.

[0102] Referring to Figure 2 , the method for determining the welding position points includes the following steps:

[0103] Step S210: Retrieve the moving position points of the welding torch and the corresponding output current values based on the welding parameter information.

[0104] Among them, the moving position points refer to the position points where the welding torch moves when the operator welds the workpiece. The output current value refers to the current value output by the welding torch in real time. The moving position points are detected and obtained by a position sensor preset on the welding torch and stored. The output current value is detected and obtained after being output by the welding machine and stored. The welding parameter information includes the moving position points and the output current values. After retrieving the moving position points and the output current values through the welding parameter information, it is convenient for subsequent use.

[0105] Step S220: Perform curve analysis based on the output current value to form a current change curve corresponding to the time change process.

[0106] Among them, the current change curve refers to the curve of the current of the welding torch changing with time. By performing curve analysis on the output current value according to the time change, a current change curve corresponding to the change of the output current value during the time change process is formed.

[0107] Step S230: Select the segment where the current change curve is consistent with the preset no-current reference curve as the no-current segment.

[0108] Among them, the no-current reference curve refers to the reference curve corresponding to the situation when there is no current at the welding torch. The no-current reference curve is preset by the operator and obtained by querying in the database storing the no-current reference curve. The no-current segment refers to the segment corresponding to the situation when the current of the welding torch is zero during the time change. By comparing the current change curve with the preset no-current reference curve and selecting the segment that is consistent in the comparison as the no-current segment, it is convenient for subsequent use.

[0109] Step S240: Remove the no-current segment based on the current change curve and connect the remaining parts to form a current curve.

[0110] Among them, the current curve refers to the curve corresponding to the situation when there is current at the welding torch. By removing the no-current segment in the current change curve and connecting the remaining parts to form a current curve, it is convenient for subsequent use.

[0111] Step S250: Analyze the change situation according to the current curve to determine the working position point, and use the working position point as the welding position point.

[0112] Among them, the working position point refers to the position point where the welding torch is located when starting the welding work. By analyzing the change situation of the current curve, the working position point is determined and used as the welding position point, which is convenient for subsequent use.

[0113] In Figure 2 In step S250 shown, in order to further ensure the rationality of the working position point, it is necessary to perform a further separate analysis and calculation on the working position point. Specifically, it is described in detail through the Figure 3 steps shown.

[0114] Referring to Figure 3 , the method for determining the working position point includes the following steps:

[0115] Step S251: Select the moving position points corresponding to the peaks according to the current curve and use them as the suspected welding position points.

[0116] Among them, the suspected welding position points refer to the position points where welding is suspected to be carried out. By selecting the moving position points corresponding to the peak positions in the current curve and using the selected moving position points as the suspected welding position points, it is convenient for subsequent use.

[0117] Step S252: Based on the welding parameter information, retrieve the residence time value corresponding to the suspected welding position point.

[0118] Among them, the residence time value refers to the time length value corresponding to the residence of the welding torch at the suspected welding position point. The welding parameter information includes the residence time value, and the residence time value is obtained and stored by calculating the time points corresponding to the welding torch at two positions within a continuous time. Retrieving the residence time value through the welding parameter information is convenient for subsequent use.

[0119] Step S253: Based on the residence time value and the preset residence reference time value, use the suspected welding position points corresponding to the residence time values greater than the residence reference time value as the residence position points, and use the suspected welding position points corresponding to the residence time values not greater than the residence reference time value as the remaining position points.

[0120] Among them, the residence reference time value refers to the minimum time length value corresponding to the welding torch in the residence state, and the residence reference time value is retrieved from a database that is pre-set and stores the residence reference time value. The residence position points refer to the position points where the welding torch is in the residence state, and the remaining position points refer to the position points corresponding to the welding torch in the moving state.

[0121] By comparing the residence time values corresponding to each suspected welding position point with the preset residence reference time value, and using the positions corresponding to the values greater than it as the residence position points and the positions corresponding to the values not greater than it as the remaining position points, it is convenient for subsequent use.

[0122] Step S254: Calculate the distance value between two adjacent remaining position points and use it as the adjacent position distance value.

[0123] Among them, the adjacent position distance value refers to the distance value between two adjacent positions per unit time when the welding torch is in the moving state. By calculating the distance value between two adjacent remaining position points and using it as the adjacent position distance value, it is convenient for subsequent use.

[0124] Step S255: Determine the continuous working position points according to the falling situation of the adjacent position distance value within the preset distance reference interval.

[0125] Among them, the distance reference interval refers to the reference distance interval corresponding to two adjacent positions of the welding torch per unit time during welding work, and the distance reference interval is obtained by querying from a database that pre-sets and stores the distance reference interval. The continuous working position points refer to the position points where the welding torch is located during continuous welding work. By analyzing the falling situation of the adjacent position distance values with respect to the preset distance reference interval, and taking the remaining position points corresponding to when the adjacent position distance values are within the distance reference interval as the continuous working position points, it is convenient for subsequent use.

[0126] Step S256: Take both the staying position points and the continuous working position points as working position points.

[0127] Among them, by taking both the staying position points and the continuous working position points as working position points, the obtained working position points include both the positions where the welding torch is in a staying state and in a moving state, thereby improving the accuracy of the obtained working position points.

[0128] In Figure 1 In the step S400 shown, in order to further ensure the rationality of the quality requirement information, it is necessary to perform a further separate analysis and calculation on the quality requirement information, specifically through the Figure 4 steps shown for detailed description.

[0129] Referring to Figure 4 , the method for determining the quality requirement information includes the following steps:

[0130] Step S410: Identify the original image information according to the preset workpiece surface reference features and take the different parts as the gap image information.

[0131] Among them, the workpiece surface reference features refer to the reference features corresponding to the workpiece surface in the normal state. The workpiece surface reference features can be obtained by the operator's pre-input, or can be obtained by identifying the original image information and selecting the features of the area with the most similar parts as the workpiece surface reference features. The gap image information refers to the image information corresponding to the gap at the welding part of the workpiece.

[0132] By using the workpiece surface reference features to identify the original image information and taking the different parts that are different from the workpiece surface reference features in the original image information as the gap image information, it is convenient for subsequent use.

[0133] Step S420: Analyze and determine the gap depth value, gap width value, and gap inclination direction information according to the gap image information.

[0134] Among them, the gap depth value refers to the depth value corresponding to the gap to be welded in the workpiece, the gap width value refers to the width value corresponding to the gap to be welded in the workpiece, and the gap inclination direction information refers to the inclination direction information corresponding to the gap to be welded in the workpiece.

[0135] By analyzing the gap image information, the gap contour is identified, and based on the position corresponding to the gap contour, calculation and analysis are carried out to obtain the gap width value and the gap inclination direction information, and based on the brightness change of the inner and outer parts of the gap contour, the gap depth value is determined.

[0136] Step S430: Calculate the product value between the gap depth value and the gap width value and use it as the gap volume value.

[0137] Among them, the gap volume value refers to the volume value corresponding to the gap to be welded in the workpiece. By calculating the product value between the gap depth value and the gap width value and using it as the gap volume value, it is convenient for subsequent reference.

[0138] Step S440: Query and obtain the current welding rod specification used and use it as the welding rod specification information.

[0139] Among them, the welding rod specification information refers to the welding rod specification used for welding the workpiece at the current time. The welding rod specification information is obtained through pre-input by the operator and is determined by the clamping opening angle when the welding rod is clamped by the welding torch.

[0140] Step S450: Analyze and determine the required initial quality information based on the welding rod specification information and the gap volume value.

[0141] Among them, the required initial quality information refers to the initial information of the quality requirements such as the weld width and smoothness that the solder joint needs to meet after welding the gap on the workpiece.

[0142] By analyzing the welding rod specification information and the gap volume value, query the historical welding quality database based on the welding rod specification information and the gap volume value to determine the average quality that can be achieved when welding the gap corresponding to the gap volume value with the welding rod specification, and use it as the required initial quality information. The historical welding quality database refers to the quality database obtained after welding gaps of different volumes with different welding rod specifications at historical times. The historical welding quality database is obtained by uploading the data of the welding rod specification, gap volume, and final welding quality after each welding.

[0143] Step S460: Determine the inclination direction influence information corresponding to the gap inclination direction information according to the corresponding relationship between the gap inclination direction information and the preset inclination direction influence information.

[0144] Among them, the tilt direction influence information refers to the influence information on the required quality when the tilt direction of the gap on the workpiece affects the welding quality. The tilt direction influence information is obtained by querying from a database that has pre-set and stores gap tilt direction information and tilt direction influence information. When the angle corresponding to the gap tilt direction information is larger, the greater the influence on the quality of subsequent welding. At this time, the required quality needs to be correspondingly reduced.

[0145] Determine the tilt direction influence information by querying from the database through the gap tilt direction information, which is convenient for subsequent use.

[0146] Step S470: Adjust the required initial quality information based on the tilt direction influence information to form the required adjusted quality information, and use the required adjusted quality information as the quality requirement information.

[0147] Among them, the required adjusted quality information refers to the quality information that needs to be achieved after welding when the gap of the workpiece is affected by the gap tilt direction. The required initial quality information is adjusted by the tilt direction influence information to form the required adjusted quality information and use it as the quality requirement information, thereby improving the accuracy of the obtained quality requirement information.

[0148] In Figure 4 In step S430 shown, in order to further ensure the rationality of the gap volume value, it is necessary to perform a further separate analysis and calculation before calculating the gap volume value. Specifically, it is described in detail through the steps shown in Figure 5 shown.

[0149] Refer to Figure 5 , the steps before calculating the gap volume value include the following steps:

[0150] Step S431: Obtain the ambient light information corresponding to the current time.

[0151] Among them, the ambient light information refers to the light conditions of the environment where the workpiece is welded at the current time. The ambient light information is detected and obtained through a photosensitive sensor array preset on the electric welding machine or the welding mask worn by the operator.

[0152] Step S432: Retrieve the ambient brightness value and the ambient light direction information based on the ambient light information.

[0153] Among them, the environmental brightness value refers to the illumination brightness value of the environment where the workpiece is welded, and the environmental light direction information refers to the light irradiation direction information of the environment where the workpiece is welded. The environmental light information includes the environmental brightness value and the environmental light direction information, and both the environmental brightness value and the environmental light direction information are detected and stored by the photosensitive sensor array. The environmental brightness value and the environmental light direction information are retrieved through the environmental light information, so as to facilitate subsequent use.

[0154] Step S433: According to the corresponding relationship between the environmental brightness value and the preset depth brightness influence value, analyze and determine the depth brightness influence value corresponding to the environmental brightness value.

[0155] Among them, the depth brightness influence value refers to the influence degree value of the environmental brightness on the detection of the gap depth. The depth brightness influence value is obtained by querying from a database that has pre-set and stored different environmental brightness values and the corresponding depth brightness influence values. Querying and determining the depth brightness influence value from the database through the environmental brightness value facilitates subsequent use.

[0156] Step S434: Calculate the deviation angle between the environmental light direction information and the preset light reference direction information and use it as the light direction deviation angle value.

[0157] Among them, the light reference direction information refers to the light direction required for welding, and the light reference direction information is obtained by querying from a database that has pre-set and stored the light reference direction information. The light direction deviation angle value refers to the deviation angle value when the light direction is deviated.

[0158] By calculating the deviation angle between the environmental light direction information and the preset light reference direction information and using it as the light direction deviation angle value, it facilitates subsequent use.

[0159] Step S435: According to the corresponding relationship between the light direction deviation angle value and the preset depth light direction influence value, analyze and determine the depth light direction influence value corresponding to the light direction deviation angle value.

[0160] Among them, the depth light direction influence value refers to the influence degree value of the light direction in the welding environment on the detection of the gap depth. The depth light direction influence value is obtained by querying from a database that has pre-set and stored different light direction deviation angle values and the corresponding depth light direction influence values. Querying and determining the depth light direction influence value through the light direction deviation angle value facilitates subsequent use.

[0161] Step S436: Calculate the sum value between the depth brightness influence value and the depth light direction influence value and use it as the depth comprehensive influence value, and adjust the gap depth value based on the depth comprehensive influence value to form a new gap depth value.

[0162] Among them, the depth comprehensive influence value refers to the comprehensive influence degree value that affects the detection of the gap depth. By calculating the sum value between the depth brightness influence value and the depth illumination direction influence value as the depth comprehensive influence value, and obtaining the corresponding depth adjustment value based on the depth comprehensive influence value, and adjusting the gap depth value to form a new gap depth value, so as to improve the accuracy of the obtained gap depth value.

[0163] In Figure 1 In step S600 shown, in order to further ensure the rationality of the quality adjustment control information, it is necessary to perform a further separate analysis and calculation on the quality adjustment control information before, specifically through Figure 6 the steps shown for detailed description.

[0164] Referring to Figure 6 , the method for determining the quality adjustment control information includes the following steps:

[0165] Step S610: Analyze the deviation between the solder joint quality information and the quality requirement information and use it as the quality deviation information.

[0166] Among them, the quality deviation information refers to the deviation information when there is a deviation in the quality of the solder joint after welding the workpiece. By analyzing the deviation between the solder joint quality information and the quality requirement information and using the deviation situation as the quality deviation information, it is convenient for subsequent use.

[0167] Step S620: Retrieve the quality deviation type information and the type deviation value based on the quality deviation information.

[0168] Among them, the quality deviation type information refers to the type corresponding to the deviation in the solder joint quality, and the type deviation value refers to the specific deviation value corresponding to the deviation type of the solder joint quality. The quality deviation information includes the quality deviation type information and the type deviation value. By retrieving the quality deviation type information and the type deviation value through the quality deviation information, it is convenient for subsequent use.

[0169] Step S630: Determine the deviation type reference evaluation value corresponding to the quality deviation type information according to the corresponding relationship between the quality deviation type information and the preset deviation type reference evaluation value.

[0170] Among them, the deviation type reference evaluation value refers to the evaluation value of the deviation type of the solder joint quality under the unit deviation value. The deviation type reference evaluation value is obtained by querying from a database that has been preset and stores different quality deviation type information and the corresponding deviation type reference evaluation values. By querying and determining the deviation type reference evaluation value from the database through the quality deviation type information, it is convenient for subsequent use.

[0171] Step S640: Calculate the product value between the deviation type reference evaluation value and the type deviation value and use it as the single type evaluation value.

[0172] Among them, the single type evaluation value refers to the evaluation value corresponding to a single deviation type under the specific deviation value. By calculating the product value between the deviation type reference evaluation value and the type deviation value and using it as the single type evaluation value, it is convenient for subsequent use.

[0173] Step S650: Calculate the sum value of all single type evaluation values and use it as the comprehensive evaluation value.

[0174] Among them, the comprehensive evaluation value refers to the evaluation value after comprehensively evaluating all deviation types. By calculating the sum value of all single type evaluation values and using it as the comprehensive evaluation value, it is convenient for subsequent use.

[0175] Step S660: Input the comprehensive evaluation value into a preset quality deviation adjustment model for analysis to obtain comprehensive evaluation adjustment information, and use the comprehensive evaluation adjustment information as quality adjustment control information.

[0176] Among them, the quality deviation adjustment model is a model for controlling and adjusting the analysis when there is a deviation in the solder joint quality. The quality deviation adjustment model is obtained by pre-inputting a large amount of solder joint quality deviation data by the operator for neural network training. The comprehensive evaluation adjustment information refers to the control information for controlling the welding robot to adjust the quality of the solder joint according to the comprehensive evaluation value. By inputting the comprehensive evaluation value into the preset quality deviation adjustment model for analysis, the comprehensive evaluation adjustment information is obtained and used as the quality adjustment control information, thereby improving the accuracy of the obtained quality adjustment control information.

[0177] After Figure 6 shown in Step S650, in order to further ensure the rationality of the comprehensive evaluation value, it is necessary to perform further separate analysis and calculation after calculating the sum value of all single type evaluation values and using it as the comprehensive evaluation value. Specifically, it is described in detail through Figure 7 the steps shown.

[0178] Refer to Figure 7 , the steps after calculating the sum value of all single type evaluation values and using it as the comprehensive evaluation value include the following steps:

[0179] Step S651: Determine whether the comprehensive evaluation value is greater than the preset evaluation reference value. If it is, execute Step S652; if not, execute Step S653.

[0180] Among them, the evaluation reference value refers to the minimum evaluation value corresponding to the qualified solder joint quality. The evaluation reference value is retrieved from a database that has been preset and stores the evaluation reference value. By judging whether the comprehensive evaluation value is greater than the preset evaluation reference value, it is determined whether further evaluation of the solder joint quality is required.

[0181] Step S652: Continue to output the comprehensive evaluation value.

[0182] Among them, when the comprehensive evaluation value is greater than the preset evaluation reference value, it indicates that further evaluation of the solder joint quality is not required at this time. Therefore, the comprehensive evaluation value is continuously output.

[0183] Step S653: Analyze and determine the workpiece type information based on the image detection information.

[0184] Among them, the workpiece type information refers to the type to which the workpiece to be welded belongs. When the comprehensive evaluation value is not greater than the preset evaluation reference value, it indicates that further evaluation of the solder joint quality is required at this time. Therefore, the image detection information is identified and matched with the characteristics of each workpiece type stored in advance to obtain the workpiece type information.

[0185] Step S654: Analyze and determine the welding required angle range based on the workpiece type information and the welding position point.

[0186] Among them, the welding required angle range refers to the angle range in which the welding angle should be during welding so that the solder joint quality requirements can be met after welding for this type of workpiece. By querying the workpiece type information, the required angle ranges at different positions are obtained, and then the required angle range corresponding to the welding position point is queried and determined as the welding required angle range, which is convenient for subsequent use.

[0187] Step S655: Based on the welding parameter information, retrieve the clamping angle value between the welding torch and the welding electrode and the rotation angle value of the welding torch.

[0188] Among them, the clamping angle value refers to the angle value formed between the welding torch and the welding electrode when the welding torch clamps the welding electrode. The clamping angle value is detected and obtained by a pressure sensor preset in the welding torch and then stored. The rotation angle value refers to the angle value at which the welding torch rotates during the welding process. The rotation angle value is detected and obtained by a rotation angle sensor preset in the welding torch. The welding parameter information includes the clamping angle value and the rotation angle value. By retrieving the clamping angle value and the rotation angle value through the welding parameter information, it is convenient for subsequent use.

[0189] Step S656: Analyze and determine the welding angle value based on the clamping angle value and the rotation angle value.

[0190] Among them, the welding angle value refers to the angle value when welding the workpiece. Based on the rotation angle value and the clamping angle value, the included angle value between the current welding rod and the horizontal plane is calculated and used as the welding angle value for subsequent use.

[0191] Step S657: Determine the angle influence value according to the inclusion situation between the welding angle value and the welding required angle range, and add the angle influence value to the comprehensive evaluation value to form a new comprehensive evaluation value.

[0192] Among them, the angle influence value refers to the influence degree value of the welding angle on the quality of the solder joint. By analyzing the inclusion situation between the welding angle value and the welding required angle range, when included, the angle non-influence value is output and used as the angle influence value. When not included, the angle deviation value between the welding angle value and the welding required angle range is calculated, and the angle influence value is determined by querying in the database pre-set and storing the angle deviation value and the corresponding angle influence value. Then, the angle influence value is added to the comprehensive evaluation value to form a new comprehensive evaluation value, so as to improve the accuracy of the obtained comprehensive evaluation value and facilitate reminding the operator to adjust the welding angle.

[0193] After Figure 7 the step S657 shown, in order to further ensure the rationality of the comprehensive evaluation value, it is necessary to perform further separate analysis and calculation after adding the angle influence value to the comprehensive evaluation value to form a new comprehensive evaluation value, which is specifically described in detail through the following steps.

[0194] The steps after adding the angle influence value to the comprehensive evaluation value to form a new comprehensive evaluation value include the following steps:

[0195] Step S6571: Retrieve the type reference shape information based on the workpiece type information.

[0196] Among them, the type reference shape information refers to the reference shape corresponding to the workpiece type, and the type reference shape information is retrieved from the database storing the type reference shape information. Retrieving the type reference shape information through the workpiece type information facilitates subsequent use.

[0197] Step S6572: Analyze the image detection information based on the center of gravity of the type reference shape information to determine the shape center of gravity position point.

[0198] Among them, the shape center of gravity position point refers to the position point where the center of gravity of the shape of the workpiece type is located. By analyzing the image detection information based on the center of gravity of the type reference shape information, the center of gravity position of the workpiece in the image detection information is determined, and then the shape center of gravity position point is calculated according to the ratio value between the image and the actual, which is convenient for subsequent use.

[0199] Step S6573: Calculate the distance vector between the shape centroid position point and the welding position point and use it as the welding centroid vector value.

[0200] Among them, the welding centroid vector value refers to the vector value between the welding position and the workpiece centroid. By calculating the vector value between the shape centroid position point and the welding position point and using it as the welding centroid vector value, it is convenient for subsequent use.

[0201] Step S6574: Retrieve the important influence position area based on the type reference shape information.

[0202] Among them, the important influence position area refers to the position area in the shape to which the workpiece belongs that has an important impact on the subsequent overall quality. The important influence position area is retrieved from the database storing the important influence position area. The type reference shape information includes the important influence position area, and the important influence position area is retrieved through the type reference shape information.

[0203] Step S6575: Determine whether the welding position point is within the important influence position area. If it is, execute Step S6576; if not, execute Step S6577.

[0204] Among them, by determining whether the welding position point is within the important influence position area, it is thus determined whether the current welding position will affect the overall quality of the workpiece.

[0205] Step S6576: Analyze and determine the centroid influence value based on the welding centroid vector value and the type reference shape information, and add the centroid influence value to the comprehensive evaluation value to form a new comprehensive evaluation value.

[0206] Among them, the centroid influence value refers to the influence degree value of the welding quality on the centroid and the workpiece quality. When the welding position point is within the important influence position area, it indicates that the current welding position will affect the overall quality of the workpiece at this time. Therefore, by analyzing the welding centroid vector value and the type reference shape information, the centroid influence value is determined, and the centroid influence value is added to the comprehensive evaluation value to form a new comprehensive evaluation value, thereby improving the accuracy of the obtained comprehensive evaluation value.

[0207] In order to improve the accuracy of the obtained centroid influence value, the method for determining the centroid influence value includes:

[0208] Step S65761: Retrieve the type reference area value and the shape vector reference influence value based on the type reference shape information.

[0209] Among them, the type reference area value refers to the reference area value in the corresponding shape of the workpiece, and the shape vector reference influence value refers to the reference influence degree value generated by the unit vector on the centroid in the corresponding shape of the workpiece. The type reference area value and the shape vector reference influence value are retrieved through the type reference shape information for convenient subsequent use.

[0210] Step S65762: Retrieve the type reference weight value based on the workpiece type information.

[0211] Among them, the type reference weight value refers to the weight value per unit area corresponding to the type to which the workpiece belongs. The type reference weight value is obtained through the workpiece type information for convenient subsequent use.

[0212] Step S65763: Analyze and calculate the type reference area value, the shape vector reference influence value, the type reference weight value, the welding centroid vector value, and the preset solder joint reference mass value according to the preset centroid influence value calculation formula to determine the centroid influence value.

[0213] Among them, the centroid influence value calculation formula refers to the formula for calculating the centroid influence value, and the centroid influence value calculation formula is obtained after being pre-input by the operator. By using the centroid influence value calculation formula to analyze and calculate the type reference area value, the type reference thickness value, the type reference weight value, and the welding centroid vector value, the centroid influence value is determined, thereby improving the accuracy of the obtained centroid influence value.

[0214] The centroid influence value calculation formula is Q is the centroid influence value, a is the welding centroid vector value, σ is the shape vector reference influence value, g is the preset solder joint reference mass value, s is the type reference area value, and m is the type reference weight value. The solder joint reference mass value refers to the reference mass value corresponding to a single solder joint, and the solder joint reference mass value is obtained after being pre-input by the operator.

[0215] For example, when the welding centroid vector value and the shape vector reference influence value are in the same direction, and a = 2, σ = 0.1, g = 0.1, m = 2, d = 2, at this time the centroid influence value

[0216] Step S6577: Continue to output the comprehensive evaluation value.

[0217] Among them, when the welding position point is not located in the important influence position area, it means that the current welding position will not affect the overall quality of the workpiece at this time, so the comprehensive evaluation value is continued to be output.

[0218] Based on the same inventive concept, an embodiment of the present invention provides an intelligent welding robot control system based on a deep learning model, including:

[0219] An acquisition module for acquiring image detection information, welding parameter information, electrode specification information, and ambient light information;

[0220] A memory for storing a program of the intelligent welding robot control method based on a deep learning model as described in any one of Figures 1 to 7 the above;

[0221] A processor for loading and executing the program in the memory and implementing the intelligent welding robot control method based on a deep learning model as described in any one of Figures 1 to 7 the above;

[0222] Those skilled in the art can clearly understand that, for the convenience and conciseness of description, only the above division of each functional module is used as an example. In actual applications, the above functions can be allocated to different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. The specific working processes of the system, device, and unit described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0223] The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. Any technical solution falling within the concept of the present invention belongs to the protection scope of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements should also be regarded as within the protection scope of the present invention.

Claims

1. An intelligent welding robot control method based on a deep learning model, characterized in that, Including: Obtain image detection information during welding of the workpiece and welding parameter information of the welding torch; Analyze and determine the welding position point based on the welding parameter information; Based on the welding position point, determine the solder joint image information corresponding to the welding position point and the original image information when not welded from the image detection information; Analyze and determine the quality requirement information based on the original image information; Analyze and determine the solder joint quality information based on the solder joint image information; Based on the deviation between the solder joint quality information and the quality requirement information, determine the quality adjustment control information, and output the quality adjustment control information to the welding robot for welding adjustment.

2. The intelligent welding robot control method based on a deep learning model according to claim 1, wherein The method for determining the welding position point includes: Based on the welding parameter information, retrieve the moving position point of the welding torch and the output current value corresponding to the moving position point; Conduct a curve analysis based on the output current value to form a current change curve corresponding to the time change process; Select the segment where the current change curve is consistent with the preset no-current reference curve as the no-current segment; Based on the current change curve, remove the no-current segment and connect the remaining part to form a current curve; Analyze the change situation based on the current curve to determine the working position point, and use the working position point as the welding position point.

3. The intelligent welding robot control method based on a deep learning model according to claim 2, wherein The method for determining the working position point includes: Select the moving position point corresponding to the wave peak based on the current curve and use it as the suspected welding position point; Based on the welding parameter information, retrieve the dwell time value corresponding to the suspected welding position point; Based on the dwell time value and the preset dwell reference time value, use the suspected welding position point corresponding to the dwell time value greater than the dwell reference time value as the dwell position point, and use the suspected welding position point corresponding to the dwell time value not greater than the dwell reference time value as the remaining position point; Calculate the distance value between two adjacent remaining position points and use it as the adjacent position distance value; Based on the falling situation of the adjacent position distance value within the preset distance reference interval, determine the continuous working position points; Use both the dwell position point and the continuous working position points as the working position points.

4. The intelligent welding robot control method based on a deep learning model according to claim 3, characterized in that, The method for determining the quality requirement information includes: Based on the preset reference features of the workpiece surface, identify the original image information and use the different part as the gap image information; Analyze and determine the gap depth value, gap width value, and gap inclination direction information based on the gap image information; Calculate the product value of the gap depth value and the gap width value and use it as the gap volume value; Query and obtain the specification of the welding electrode currently in use and use it as the welding electrode specification information; Analyze and determine the initial required quality information based on the welding electrode specification information and the gap volume value; Based on the corresponding relationship between the gap inclination direction information and the preset inclination direction influence information, determine the inclination direction influence information corresponding to the gap inclination direction information; Based on the inclination direction influence information, adjust the initial required quality information to form the adjusted required quality information, and use the adjusted required quality information as the quality requirement information.

5. The intelligent welding robot control method based on a deep learning model according to claim 4, wherein, It also includes a step before calculating the gap volume value, specifically as follows: Obtain the environmental light information corresponding to the current time; Based on the environmental light information, retrieve the environmental brightness value and the environmental light direction information; Analyze and determine the depth luminance influence value corresponding to the ambient luminance value according to the corresponding relationship between the ambient luminance value and the preset depth luminance influence value; Calculate the deviation angle between the ambient light direction information and the preset light reference direction information and use it as the light direction deviation angle value; Analyze and determine the depth light direction influence value corresponding to the light direction deviation angle value according to the corresponding relationship between the light direction deviation angle value and the preset depth light direction influence value; Calculate the sum value between the depth luminance influence value and the depth light direction influence value and use it as the depth comprehensive influence value, and adjust the gap depth value based on the depth comprehensive influence value to form a new gap depth value.

6. The intelligent welding robot control method based on a deep learning model according to claim 1, characterized in that, The method for determining the quality adjustment control information includes: Analyze the deviation between the solder joint quality information and the quality requirement information and use it as the quality deviation information; Retrieve the quality deviation type information and the type deviation value based on the quality deviation information; Determine the deviation type reference evaluation value corresponding to the quality deviation type information according to the corresponding relationship between the quality deviation type information and the preset deviation type reference evaluation value; Calculate the product value between the deviation type reference evaluation value and the type deviation value and use it as the single type evaluation value; Calculate the sum value of all single type evaluation values and use it as the comprehensive evaluation value; Input the comprehensive evaluation value into the preset quality deviation adjustment model for analysis to obtain the comprehensive evaluation adjustment information, and use the comprehensive evaluation adjustment information as the quality adjustment control information.

7. The intelligent welding robot control method based on a deep learning model according to claim 6, characterized in that, It also includes the steps after calculating the sum value of all single type evaluation values and using it as the comprehensive evaluation value, specifically as follows: Judge whether the comprehensive evaluation value is greater than the preset evaluation reference value; If it is, continue to output the comprehensive evaluation value; If it is not, analyze and determine the workpiece type information according to the image detection information; Analyze and determine the welding requirement angle range according to the workpiece type information and the welding position point; Retrieve the clamping angle value between the welding torch and the welding electrode and the rotation angle value of the welding torch based on the welding parameter information; Analyze and determine the welding angle value according to the clamping angle value and the rotation angle value; Determine the angle influence value according to the falling situation between the welding angle value and the welding requirement angle range, and add the angle influence value to the comprehensive evaluation value to form a new comprehensive evaluation value.

8. The intelligent welding robot control method based on a deep learning model according to claim 7, characterized in that It also includes the steps after adding the angle influence value to the comprehensive evaluation value to form a new comprehensive evaluation value, specifically as follows: Retrieve the type reference shape information based on the workpiece type information; Analyze the image detection information based on the center of gravity of the type reference shape information to determine the shape center of gravity position point; Calculate the distance vector between the shape center of gravity position point and the welding position point and use it as the welding center of gravity vector value; Retrieve the important influence position area based on the type reference shape information; Judge whether the welding position point is within the important influence position area; If it is, analyze and determine the center of gravity influence value according to the welding center of gravity vector value and the type reference shape information, and add the center of gravity influence value to the comprehensive evaluation value to form a new comprehensive evaluation value; If it is not, continue to output the comprehensive evaluation value.

9. The intelligent welding robot control method based on a deep learning model according to claim 8, wherein The method for determining the center of gravity influence value includes: Retrieve the type reference area value and the shape vector reference influence value based on the type reference shape information; Retrieve the type reference weight value based on the workpiece type information; Analyze and calculate the type reference area value, shape vector reference influence value, type reference weight value, welding center of gravity vector value, and preset welding point reference mass value according to the preset center of gravity influence value calculation formula to determine the center of gravity influence value. The center of gravity influence value calculation formula is Q is the center of gravity influence value, a is the welding center of gravity vector value, σ is the shape vector reference influence value, g is the preset welding point reference mass value, s is the type reference area value, and m is the type reference weight value.

10. An intelligent welding robot control system based on a deep learning model, characterized in that, Include: An acquisition module, configured to acquire image detection information, welding parameter information, electrode specification information, and ambient light information; A memory, configured to store a program of the intelligent welding robot control method based on a deep learning model according to any one of claims 1 to 9; A processor, configured to load and execute the program in the memory and implement the intelligent welding robot control method based on a deep learning model according to any one of claims 1 to 9.

Images