Method for automatic detection technology of welding seam quality of narrow lap joint seam rolling welding machine

Abstract

The invention discloses a method of an automatic detection technology for the welding seam quality of a narrow-lap seam rolling welding machine, relates to the technical field of welding seam quality detection, and solves the technical problem that the product quality is affected due to the fact that the influence of subsequent welding plates on welded plates during layered welding is not considered in the prior art. The method comprises the following steps: collecting a front image and a side image of welding, correcting stage temperature time of a preheating stage by combining external environment data, setting a stage fault coefficient of each stage according to an ideal temperature of each stage and a difference between the stage temperature, and calculating a fault probability by combining the ideal temperature; obtaining a test result according to the detection result of the cooling front image and the prediction evaluation result, counting the prediction accuracy of the test result, comparing the prediction accuracy with a prediction threshold value, and updating a stage fault coefficient of misprediction; and generating detection standards of different hierarchies for the side image based on the detection result, and updating the detection standards according to the different hierarchies.

Description

Technical Field

[0001] The invention belongs to the field of weld quality detection, in particular to a method of automatic detection technology of weld quality of a narrow lap seam welding machine. Background Art

[0002] In the modern steel and metal processing industry, narrow lap seam welders are widely used because they can efficiently and continuously weld strips of various specifications. This equipment not only improves production efficiency, but also reduces production costs, giving companies an advantage in the fierce market competition. However, as market demand for product quality continues to increase, weld quality, as a key factor affecting the overall performance of the product, has become a technical problem that needs to be solved urgently.

[0003] The invention patent with application number CN2024106927804 discloses a weld quality detection system and method based on welding data. The invention obtains a welding trajectory by performing image recognition on the weld image, and judges the welding quality according to the welding trajectory; obtains a segmentation trajectory based on the welding trajectory, generates a welding satisfaction signal and a welding dissatisfaction signal according to the shape of the segmentation trajectory, analyzes and verifies the influencing factors of the generation of the welding dissatisfaction signal, and completes the quality judgment according to the verification signal; when judging the welding quality, this method judges according to the welding trajectory. When welding thicker plates, the welding trajectory is more complicated and layered welding is required. When layered welding is performed on subsequent layered plates, it may affect the plates that have been welded, causing secondary changes in the welded plates to cause failures, thereby affecting the quality of the final product.

[0004] The present invention provides a method for automatically detecting the weld quality of a narrow lap seam welder to solve the above technical problems. Summary of the invention

[0005] The present invention aims to solve at least one of the technical problems existing in the prior art; to this end, the present invention proposes a method for automatic detection technology of weld quality of a narrow lap welder, which is used to solve the technical problem that the prior art does not consider the influence of subsequent welded plates on the welded plates during layered welding, resulting in product quality being affected.

[0006] To achieve the above-mentioned object, the first aspect of the present invention provides a method for automatic detection technology of weld quality of narrow lap roll seam welder, comprising: an image collection module, a quality assessment module and a detection warning module;

[0007] Image collection module: used to collect front and side images of layered welding respectively;

[0008] Quality assessment module: used to divide the front image into a welding front image and a cooling front image, perform temperature analysis on the welding front image to obtain the real-time welding temperature; predict the welding quality based on the real-time welding temperature to obtain the prediction and assessment results; combine the prediction and assessment results with the cooling front image to inspect the welding quality to obtain the inspection results;

[0009] Detection and warning module: used to generate detection standards for side images based on inspection results and update them in real time; detect side images according to the detection standards and generate warning signals.

[0010] Preferably, the step of performing temperature analysis on the welding front image to obtain the real-time welding temperature includes:

[0011] Extract the welding front image;

[0012] The RGB color value of each pixel in the welding front image is extracted, the RGB color values ​​of all pixels in the welding front image are counted and the average value is calculated to obtain the image color value; the image color value is matched with the color-temperature mapping table to obtain the corresponding real-time welding temperature.

[0013] It should be noted that the front image of the welding is collected by an infrared thermal imager, and the infrared thermal imager has a color mapping table configured therewith, which can match the corresponding temperature according to the color of the image.

[0014] Preferably, the prediction of welding quality based on the real-time welding temperature to obtain a prediction evaluation result includes:

[0015] Extract the real-time welding temperature and record the corresponding temperature time; collect external environmental data through external sensors;

[0016] Extract the stage temperature from the historical database, divide the temperature time to obtain the stage temperature time; wherein the temperature stage includes the preheating stage, the heating stage and the welding stage;

[0017] The stage temperature and stage temperature time of each temperature stage are integrated to obtain the temperature stage data; the prediction and evaluation results are obtained by combining the external environment data with the temperature stage data prediction.

[0018] Preferably, the prediction and evaluation results obtained by combining the external environment data with the temperature stage data include:

[0019] Extracting external environment data and temperature stage data; wherein the external environment data is the external temperature outside the weld during welding;

[0020] Set the ideal preheating temperature, mark the stage temperature of the preheating stage as the preheating temperature, and obtain the time correction coefficient by the ratio of the difference between the preheating temperature and the external temperature to the difference between the preheating temperature and the ideal preheating temperature; multiply the stage temperature time of the preheating stage by the time correction coefficient to obtain the adjusted preheating time; wherein, the ideal preheating temperature is the temperature that ensures the production quality of the plate of this type after preheating;

[0021] Set the ideal heating temperature and the ideal welding temperature, integrate the ideal preheating temperature, the ideal heating temperature and the ideal welding temperature to get the ideal temperature; through the formula GZL = α1 × | ln(YRW-BZW-1) | / YRT×100%+α2× | ln(JRW-BJW-1) | / JRT×100%+α3× | ln(HJW-BHW-1) | / HJT×100% to calculate the failure probability GZL; wherein, YRW represents the preheating temperature, BZW represents the ideal preheating temperature, YRT represents the preheating time, JRW represents the stage temperature of the heating stage, BJW represents the ideal heating temperature, JRT represents the stage temperature time of the heating stage, HJW represents the stage temperature of the welding stage, BHW represents the ideal welding temperature, and HJT represents the stage temperature time of the welding stage; α1 represents the preheating failure coefficient, α2 represents the heating failure coefficient, and α3 represents the welding failure coefficient; α1<α2<α3, the preheating failure coefficient, the heating failure coefficient and the welding failure coefficient are set based on the influence of the difference between the stage temperature at each stage and the corresponding ideal temperature on the occurrence of the failure;

[0022] The failure probability is combined with the corresponding weld front image to obtain the prediction evaluation result.

[0023] It should be noted that the ideal temperature is the temperature at which the best production quality is achieved in each stage, and the stage temperature is the temperature actually achieved in each stage.

[0024] Preferably, the preheating failure coefficient, the heating failure coefficient and the welding failure coefficient are set based on the influence of the difference between the stage temperature of each stage and the corresponding ideal temperature on the occurrence of the failure, including:

[0025] S1: extraction stage temperature and corresponding ideal temperature;

[0026] S2: extract the influence temperature and failure temperature of each stage from the historical database; mark the preheating failure coefficient, heating failure coefficient and welding failure coefficient as stage failure coefficients;

[0027] S3: Mark the difference between the stage temperature of each stage and the corresponding ideal temperature as the corresponding stage temperature difference; determine whether the stage temperature is greater than the fault temperature; if yes, normalize the ratio of the difference between the fault temperature and the stage temperature to the difference between the fault temperature and the ideal temperature as the stage fault coefficient; if no, jump to S4;

[0028] S4: Determine whether the stage temperature is greater than the influence temperature; if yes, normalize the ratio of the difference between the influence temperature and the stage temperature to the difference between the influence temperature and the ideal temperature and use it as the stage fault coefficient; if no, mark the stage fault coefficient as 0.

[0029] During the welding process, different levels of temperature have different effects on failures. The impact temperature is the temperature that affects the welding quality of the weld at each stage, and the failure temperature is the temperature that directly causes failures in the welding quality of the weld at each stage.

[0030] Preferably, the inspection of welding quality by combining the prediction evaluation result with the cooling front image to obtain the inspection result includes:

[0031] Extract the prediction evaluation results and the cooling frontal image;

[0032] Extract the weld image features in the cooling front image, extract the fault image features of the weld fault and the corresponding fault type from the historical database; determine whether the weld image features match the fault image features; if yes, mark the corresponding weld image features as weld fault information, and mark the corresponding cooling front image as a fault image; if no, mark the corresponding cooling front image as a non-fault image;

[0033] Setting a fault judgment threshold; when the cooling front image is detected as a fault image, judging whether the fault probability is greater than the fault judgment threshold; if yes, marking the corresponding inspection result as a verification fault; if no, marking the corresponding inspection result as an incorrect verification fault; wherein the fault judgment threshold is set based on the degree of judging the image as a fault;

[0034] A prediction threshold is set, and the prediction results of all corresponding layers are counted to obtain the prediction accuracy; whether the prediction accuracy is greater than the prediction threshold is determined; if yes, the corresponding prediction result is marked as an accurate prediction; if no, the corresponding prediction result is marked as an incorrect prediction, and the stage failure coefficient is updated based on the prediction accuracy; wherein the prediction threshold is set based on the staff's requirements for product qualification.

[0035] Preferably, the updating of the stage failure coefficient based on the prediction accuracy includes:

[0036] Extract prediction accuracy and stage failure coefficient;

[0037] The prediction difference between the prediction accuracy and the prediction threshold is calculated; the product of the prediction difference plus one and the stage failure coefficient is used as the updated failure coefficient.

[0038] Preferably, the generating of the detection standard for the side image based on the inspection result and updating it in real time includes:

[0039] A1: Extract the inspection result of the front image;

[0040] A2: Number the front image in layer order; when the layer completion signal is detected, assign the matching side image layer number based on the number of the front image, and determine whether the inspection result of the front image is a verification fault; if yes, mark the corresponding verification fault in the side image and jump to A3; if no, mark the corresponding layer in the side image as a fault-free layer and jump to A4;

[0041] A3: Integrate the layer number, fault type and weld fault information to obtain layer fault data; mark all layer fault data as side image fault data, and after the detection is completed, add the new layer fault data to the side image fault data to complete the fault detection standard update;

[0042] A4: Determine whether the number of the front image is 1; if yes, mark the corresponding layer level as the first-level layer, and mark the corresponding detection standard as the first-level detection standard; if no, add the detection standard of the corresponding layer to the detection standard corresponding to the previous level level layer to complete the trouble-free detection standard update;

[0043] A5: Integrate the fault detection standard update with the no-fault detection standard update to complete the detection standard update.

[0044] When welding thicker plates, multiple layers of welding are required. When performing multiple layers of welding, the plates that are welded first will overlap with the plates that are welded subsequently. This may cause a fault that was not originally prominent to overlap with other faults and generate new faults. Or, the heat generated when welding subsequent plates may affect the plates that have been welded and cause new faults.

[0045] Preferably, the updating of the fault detection standard is integrated with the updating of the no-fault detection standard to complete the updating of the detection standard, including:

[0046] The fault detection standard update and the no-fault detection standard update are numbered according to the layered number to obtain the fault detection number, and the fault detection standard update and the no-fault detection standard update are respectively added to the detection standard map of the side image according to the fault detection number to complete the detection standard update.

[0047] Preferably, the step of detecting the side image according to the detection standard and generating a warning signal comprises:

[0048] After completing the update of the detection standard, extract the side image features of the side image in real time;

[0049] Determine whether the side image features are consistent with the detection standards; if yes, mark the side detection status of the corresponding side image as normal; if no, mark the side detection status of the corresponding side image as abnormal, and generate a warning signal.

[0050] Compared with the prior art, the present invention has the following beneficial effects:

[0051] 1. The present invention collects the front image and the side image of welding respectively, obtains the real-time temperature during welding according to the front image of welding, obtains the stage temperature time based on the stage temperature division, integrates the stage temperature and the stage temperature time to obtain the temperature stage data, corrects the stage temperature time of the preheating stage in combination with the external environment data, sets the stage failure coefficient of each stage according to the difference between the ideal temperature of each stage and the stage temperature, and calculates the failure probability in combination with the preset ideal temperature, which can more accurately reflect the temperature changes of each stage in the welding process, more accurately reflect the influence of the temperature of each stage on the generation of failures, and correct the influence of the external environment on the preheating stage; detects and identifies the cooling front image, compares it with the prediction evaluation result, obtains the inspection result, and statistically compares the prediction accuracy with the preset prediction threshold, updates the stage failure coefficient of the wrong prediction based on the prediction accuracy, generates different layered detection standards for the side image based on the inspection result, can more comprehensively evaluate the quality of the weld, discover potential problems, realize double verification and comprehensive evaluation, and improve the detection accuracy.

[0052] 2. The present invention compares the stage temperature difference between the stage temperature and the ideal temperature of each stage with the fault temperature and the influencing temperature, and selects different stage fault coefficient acquisition methods to set the stage fault coefficient according to the comparison results, and calculates the failure probability after welding is completed in combination with the stage temperature and the ideal temperature, extracts the cooling front image and detects the fault, and compares the failure probability with the preset fault judgment threshold according to the detection result to verify the prediction accuracy, statistically layers the prediction results to obtain the prediction accuracy, compares the prediction accuracy with the preset prediction threshold, and updates the stage failure coefficient corresponding to the wrong prediction based on the prediction accuracy, which can more accurately reflect the failure risk of each stage and further improve the accuracy of welding quality detection. If the prediction accuracy is greater than the prediction threshold, it means that the welding image during welding can be used to accurately predict the possibility of failure after cooling, and timely adjustments can be made to ensure production quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0054] Figure 1 The following is a flowchart of an embodiment of the present invention.

[0055] Figure 2 The figure is a complete flow chart for obtaining the stage failure coefficient in one embodiment of the present invention.

[0056] Figure 3 A complete flow chart for generating detection standards in one embodiment of the present invention. DETAILED DESCRIPTION

[0057] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0058] See also Figure 1-Figure 3 , the first aspect of the present invention provides a method for automatic detection technology of weld quality of narrow lap roll seam welder, including: an image collection module, a quality assessment module and a detection warning module;

[0059] Image collection module: used to collect front and side images of layered welding respectively.

[0060] Exemplarily, an infrared thermal imager is used to capture a frontal image of the weld during welding, and a CCD camera is used to capture a side image of the weld and a frontal image of the weld after cooling.

[0061] Quality assessment module: used to divide the front image into a welding front image and a cooling front image, perform temperature analysis on the welding front image to obtain the real-time welding temperature; predict the welding quality based on the real-time welding temperature to obtain a prediction assessment result; and inspect the welding quality by combining the prediction assessment result with the cooling front image to obtain an inspection result.

[0062] Exemplarily, the welding front image and the cooling front image are denoised and contrast enhanced, and an image processing algorithm is used to traverse all pixels in the front image and extract the red channel value, green channel value and blue channel value of the pixels, respectively. The average value of the red channel value, the average value of the green channel value and the average value of the blue channel value of all pixels are calculated, respectively, and the image color value of the image is obtained by combining them. The image color value is matched with the color-temperature mapping table of the infrared thermal imager to obtain the corresponding real-time welding temperature.

[0063] It should be noted that during the welding process, the welding temperature often needs to reach several hundred or several thousand degrees. If a CCD camera is used to shoot, the light generated during welding will affect the image quality of the image, so an infrared thermal imager is used to collect the image of the weld; when welding the plate, the plate will show different colors at different temperatures; for example, when the temperature of the plate reaches 200 to 300 degrees, the plate will show initial oxidation, and will appear light yellow to golden yellow. When the temperature reaches 400 to 500 degrees, the oxide layer of the plate will further thicken, and a dark blue will appear on the surface of the plate.

[0064] The temperature time corresponding to the real-time welding temperature is recorded, and the external temperature of the welding room is collected by an external temperature sensor set in the welding room; the stage temperature of the preheating stage is 300 degrees, the stage temperature of the heating stage is 500 degrees, and the stage temperature of the welding stage is 700 degrees extracted from the historical database, and the stage temperature time of the preheating stage when reaching 300 degrees, the stage temperature time of the heating stage when reaching 500 degrees, and the stage temperature time of the welding stage when reaching 700 degrees are recorded respectively in this embodiment, and the stage temperature and stage temperature time of each temperature stage are integrated to obtain temperature stage data.

[0065] In this embodiment, the ideal preheating temperature is set to 200 degrees, the ideal heating temperature is set to 600 degrees, and the ideal welding temperature is set to 800 degrees; the ratio of the difference between the preheating temperature and the external temperature to the difference between the preheating temperature and the ideal preheating temperature is calculated to obtain the time correction coefficient; and the stage temperature of the preheating stage is multiplied by the time correction coefficient to obtain the adjusted preheating time; the temperature time of the preheating stage is corrected in combination with the external environment data, so that the prediction is closer to the actual working conditions, the error is reduced, the consistency and reliability of the welding quality are ensured, and the defective rate is reduced.

[0066] During the preheating stage, the external temperature has a certain influence on the preheating time and the quality of preheating. In a low temperature environment, the heat supply required for preheating will increase, and it may cause uneven heating of the plate during preheating. Compared with a low temperature environment, the preheating time required in a high temperature environment will be shortened. Eliminating the influence of the external environment on the preheating time can better control the temperature and improve the welding quality and production efficiency. In the heating stage and the welding stage, the temperature is higher and the preheating has been completed, and the influence of the external temperature on it can be ignored.

[0067] Extract the influencing temperature and fault temperature of each stage from the historical database; mark the preheating fault coefficient, heating fault coefficient and welding fault coefficient as stage fault coefficients; calculate the difference between the stage temperature of the preheating stage and the ideal temperature to obtain the preheating stage temperature difference, calculate the difference between the stage temperature of the heating stage and the ideal temperature to obtain the heating stage temperature difference, calculate the difference between the stage temperature of the welding stage and the corresponding ideal temperature to obtain the welding stage temperature difference, and judge whether the stage temperature of the preheating stage is greater than the fault temperature. In this embodiment, the fault temperature of the preheating stage is 350 degrees, and the influencing temperature is 300 degrees. It is judged that the stage temperature is less than the fault temperature, and whether the stage temperature is greater than the influencing temperature is further judged. It is judged that the stage temperature is less than the influencing temperature, which means that in this embodiment, the temperature of the preheating stage will not affect the welding quality, nor will it cause welding failures. The influence of external factors on the preheating stage can be eliminated, and the actual preheating time in the preheating stage can be obtained more accurately, thereby ensuring the accuracy of the process flow and improving the production quality.

[0068] Through the formula GZL = α1 × | ln(YRW-BZW-1) | / YRT×100%+α2×ln(JRW-BJW-1) | / JRT×100%+α3× | ln(HJW-BHW-1) | / HJT×100% calculates the fault probability GZL; extracts shape features and texture features in the cooling front image through image processing technology and combines them to obtain weld image features, extracts fault image features of weld faults and corresponding fault types from the historical database, and determines whether the weld image features match the fault image features. If the weld image features match the fault image features, the corresponding weld image features are marked as weld fault information, and the corresponding cooling front image is marked as a fault image, indicating that a fault has occurred in the weld; in this embodiment, the fault judgment threshold is set to 70%. When the cooling front image is detected as a fault image, the prediction result is verified to determine whether the fault probability is greater than the fault judgment threshold; if the fault probability is greater than the fault judgment threshold, it means that the prediction is accurate, and the corresponding inspection result is marked as a verification fault; if the fault probability is less than the fault judgment threshold, it means that the prediction is wrong, and the corresponding inspection result is marked as an incorrect verification fault.

[0069] The prediction accuracy is obtained by statistically analyzing the prediction results. In this embodiment, the prediction threshold is set to 80%, and it is determined whether the prediction accuracy is greater than the prediction threshold. If the prediction accuracy is greater than the prediction threshold, it means that the result after cooling can be accurately predicted according to the welding image, and the welding front image collected in the welding stage can predict the result after the welding is completed. The equipment parameters can be adjusted in advance to improve the production quality, timely discover and correct welding defects, and reduce the defective rate. If the prediction accuracy is less than the prediction threshold, it means that the current prediction of the final result is inaccurate, and the fault coefficient needs to be adjusted to ensure the prediction accuracy. The prediction threshold is subtracted from the prediction accuracy to obtain the prediction difference, the prediction difference is added by one and multiplied with the preheating fault coefficient, the heating fault coefficient and the welding fault coefficient to obtain the updated preheating fault coefficient, the updated heating fault coefficient and the updated welding fault coefficient, and the updated preheating fault coefficient, the updated heating fault coefficient and the updated welding fault coefficient are integrated to obtain the updated fault coefficient; the updated fault probability is recalculated according to the updated fault coefficient, the updated fault probability is statistically updated to obtain the prediction accuracy again, and the inspection is re-performed.

[0070] Detection and warning module: used to generate detection standards for side images based on inspection results and update them in real time; detect side images according to the detection standards and generate warning signals.

[0071] Exemplarily, the front image is numbered in the order of layers. In this embodiment, the plate needs to be divided into three layers for welding. When each layer of layer detection is completed, a corresponding layer completion signal is generated; the side image layer number corresponding to the front image number is assigned, and the front image is inspected in turn to determine whether the inspection result of the front image corresponding to the first layer is a verification fault. In this embodiment, the inspection result of the front image of the first layer is a verification fault, then the weld fault information of the verification fault is marked in the side image, and the layer number, fault type and weld fault information are integrated to obtain layer fault data. In this embodiment, it is obtained that in the area of ​​the first layer, there is residual slag with a height exceeding 20 mm on the normal welding plane; the second and third layer inspection results are continued to be inspected. The detection is carried out, and the layered fault data of all layers are marked as side image fault data. The new layered fault data is added to the side image fault data in the layer order to complete the update of the fault detection standard. The fault of each layer is superimposed. The welding fault of the layer after the previous welding is completed may cause the subsequent layer welding to fail due to the accumulation of faults during the subsequent layer welding. For example, the fine bubbles that appear in the first layer are superimposed on the fine bubbles that appear in the second layer. Due to the superposition of fine bubbles between the first layer and the second layer, the third layer could have been welded normally during welding, but the welding plane at the superposition position of the faults of the first layer and the second layer is raised, making it impossible to weld the third layer normally during welding, resulting in a welding failure.

[0072] In some other preferred embodiments, if the inspection result of the front image corresponding to the layer is an incorrect verification fault, the corresponding side image is marked as a fault-free layer, and it is determined whether the front image is numbered 1. If the front image is numbered 1, it means that the layer in the corresponding side image is the first layer, and there is no welding fault in the first layer. Therefore, the welding image feature of the first layer is used as the first-level detection standard. In subsequent inspections, the first layer is inspected with the corresponding first-level detection standard. If the image feature of the first layer detected later is different from the first-level detection standard, it means that during subsequent welding, the first layer was damaged for the second time, resulting in a fault. The reason may be that when other layers are welded, the heat energy generated causes the first layer to be heated for the second time, resulting in cracks and other faults. The front images are inspected in turn. If the front image number is not 1, it means that the front image is a subsequent layer. The inspection standard of the corresponding layer is added to the inspection standard corresponding to the previous layer to complete the update of the fault-free detection standard.

[0073] In multiple layers, different layers may have inspection results that are verification faults or that are not verification faults. The layer number is integrated with the corresponding fault detection standard and the corresponding no-fault detection standard, and the detection standard of the side image is updated.

[0074] After each update of the detection standard, the side image is detected in real time, the side image features of the side image are extracted in real time, and the side image features are compared with the detection standard. If the side image features are different from the detection standard, it means that after the welding of the layer is completed, the layer has undergone secondary changes, and an early warning signal is generated.

[0075] Part of the data in the above formula is calculated by removing the dimension and taking its numerical value. The formula is a formula closest to the actual situation obtained by software simulation of a large amount of collected data; the preset parameters and preset thresholds in the formula are set by technical personnel in this field according to actual conditions or obtained through simulation of a large amount of data.

[0076] Working principle of the present invention:

[0077] The present invention collects the front image and the side image of welding respectively through an infrared thermal imager and a CCD camera, obtains the real-time welding temperature by matching the image color value of the front image of welding with a color mapping table; obtains the stage temperature time based on the stage temperature division; corrects the stage temperature time of the preheating stage in combination with external environment data, sets the stage failure coefficient, and calculates the failure probability in combination with the ideal temperature; obtains the inspection result according to the detection result of the cooling front image and the prediction evaluation result, calculates the prediction accuracy of the inspection result and compares it with the prediction threshold, and updates the stage failure coefficient of the wrong prediction; generates different layered inspection standards for the side image based on the inspection result, and updates the inspection standard according to the different layers.

[0078] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A method for automatic detection of weld quality of narrow lap seam welder, characterized in that: include: Image collection module, quality assessment module and detection and early warning module; Image collection module: used to collect front and side images of layered welding respectively; Quality assessment module: used to divide the front image into a welding front image and a cooling front image, perform temperature analysis on the welding front image to obtain the real-time welding temperature; predict the welding quality based on the real-time welding temperature to obtain the prediction and assessment results; combine the prediction and assessment results with the cooling front image to inspect the welding quality to obtain the inspection results; Detection and warning module: used to generate detection standards for side images based on inspection results and update them in real time; The side image is detected according to the detection standard to generate a warning signal.

2. The method of automatic detection technology of weld quality of narrow lap seam welder according to claim 1 is characterized in that: The step of performing temperature analysis on the welding front image to obtain the real-time welding temperature includes: Extract the welding front image; The RGB color value of each pixel in the welding front image is extracted, the RGB color values ​​of all pixels in the welding front image are counted and the average value is calculated to obtain the image color value; the image color value is matched with the color-temperature mapping table to obtain the corresponding real-time welding temperature.

3. The method of automatic detection technology of weld quality of narrow lap seam welder according to claim 1 is characterized in that: The prediction of welding quality based on the real-time welding temperature to obtain a prediction evaluation result includes: Extract the real-time welding temperature and record the corresponding temperature time; collect external environmental data through external sensors; Extract the stage temperature from the historical database, divide the temperature time to obtain the stage temperature time; wherein the temperature stage includes the preheating stage, the heating stage and the welding stage; The stage temperature and stage temperature time of each temperature stage are integrated to obtain the temperature stage data; the prediction and evaluation results are obtained by combining the external environment data with the temperature stage data prediction.

4. The method of automatic detection technology of weld quality of narrow lap seam welder according to claim 3 is characterized in that: The prediction and evaluation results obtained by combining the external environment data with the temperature stage data include: Extracting external environment data and temperature stage data; wherein the external environment data is the external temperature outside the weld during welding; Set the ideal preheating temperature, mark the stage temperature of the preheating stage as the preheating temperature, and obtain the time correction coefficient by the ratio of the difference between the preheating temperature and the external temperature to the difference between the preheating temperature and the ideal preheating temperature; multiply the stage temperature time of the preheating stage by the time correction coefficient to obtain the adjusted preheating time; wherein, the ideal preheating temperature is the temperature that ensures the production quality of the plate of this type after preheating; Set the ideal heating temperature and the ideal welding temperature, integrate the ideal preheating temperature, the ideal heating temperature and the ideal welding temperature to get the ideal temperature; through the formula GZL = α1 × | ln(YRW-BZW-1) | / YRT×100%+α2×|ln(JRW-BJW-1)| / JRT×100%+α3×|ln(HJW-BHW-1) | / HJT×100% to calculate the failure probability GZL; wherein, YRW represents the preheating temperature, BZW represents the ideal preheating temperature, YRT represents the preheating time, JRW represents the stage temperature of the heating stage, BJW represents the ideal heating temperature, JRT represents the stage temperature time of the heating stage, HJW represents the stage temperature of the welding stage, BHW represents the ideal welding temperature, and HJT represents the stage temperature time of the welding stage; α1 represents the preheating failure coefficient, α2 represents the heating failure coefficient, and α3 represents the welding failure coefficient; α1<α2<α3, the preheating failure coefficient, the heating failure coefficient and the welding failure coefficient are set based on the influence of the difference between the stage temperature at each stage and the corresponding ideal temperature on the occurrence of the failure; The failure probability is combined with the corresponding weld front image to obtain the prediction evaluation result.

5. The method of automatic detection technology of weld quality of narrow lap seam welder according to claim 4 is characterized in that: The preheating failure coefficient, heating failure coefficient and welding failure coefficient are set based on the influence of the difference between the stage temperature of each stage and the corresponding ideal temperature on the occurrence of failure, including: S1: extraction stage temperature and corresponding ideal temperature; S2: extract the influence temperature and failure temperature of each stage from the historical database; mark the preheating failure coefficient, heating failure coefficient and welding failure coefficient as stage failure coefficients; S3: Mark the difference between the stage temperature of each stage and the corresponding ideal temperature as the corresponding stage temperature difference; determine whether the stage temperature is greater than the fault temperature; if yes, normalize the ratio of the difference between the fault temperature and the stage temperature to the difference between the fault temperature and the ideal temperature as the stage fault coefficient; if no, jump to S4; S4: Determine whether the stage temperature is greater than the influence temperature; if yes, normalize the ratio of the difference between the influence temperature and the stage temperature to the difference between the influence temperature and the ideal temperature and use it as the stage fault coefficient; if no, mark the stage fault coefficient as 0.

6. The method of automatic detection technology of weld quality of narrow lap seam welder according to claim 1 is characterized in that: The welding quality is inspected by combining the prediction evaluation result with the cooling front image to obtain the inspection result, including: Extract the prediction evaluation results and the cooling frontal image; Extract the weld image features in the cooling front image, extract the fault image features of the weld fault and the corresponding fault type from the historical database; determine whether the weld image features match the fault image features; if yes, mark the corresponding weld image features as weld fault information, and mark the corresponding cooling front image as a fault image; if no, mark the corresponding cooling front image as a non-fault image; Setting a fault judgment threshold; when the cooling front image is detected as a fault image, judging whether the fault probability is greater than the fault judgment threshold; if yes, marking the corresponding inspection result as a verification fault; if no, marking the corresponding inspection result as an incorrect verification fault; wherein the fault judgment threshold is set based on the degree of judging the image as a fault; A prediction threshold is set, and the prediction results of all corresponding layers are counted to obtain the prediction accuracy; whether the prediction accuracy is greater than the prediction threshold is determined; if yes, the corresponding prediction result is marked as an accurate prediction; if no, the corresponding prediction result is marked as an incorrect prediction, and the stage failure coefficient is updated based on the prediction accuracy; wherein the prediction threshold is set based on the staff's requirements for product qualification.

7. The method of automatic detection technology of weld quality of narrow lap seam welder according to claim 6 is characterized in that: The updating of the stage failure coefficient based on the prediction accuracy includes: Extract prediction accuracy and stage failure coefficient; The prediction difference between the prediction accuracy and the prediction threshold is calculated; the product of the prediction difference plus one and the stage failure coefficient is used as the updated failure coefficient.

8. The method of automatic detection technology of weld quality of narrow lap seam welder according to claim 1 is characterized in that: The generating of the detection standard for the side image based on the inspection result and updating it in real time includes: A1: Extract the inspection result of the front image; A2: Number the front image in layer order; when the layer completion signal is detected, assign the matching side image layer number based on the number of the front image, and determine whether the inspection result of the front image is a verification fault; if yes, mark the corresponding verification fault in the side image and jump to A3; if no, mark the corresponding layer in the side image as a fault-free layer and jump to A4; A3: Integrate the layer number, fault type and weld fault information to obtain layer fault data; mark all layer fault data as side image fault data, and after the detection is completed, add the new layer fault data to the side image fault data to complete the fault detection standard update; A4: Determine whether the number of the front image is 1; if yes, mark the corresponding layer level as the first-level layer, and mark the corresponding detection standard as the first-level detection standard; if no, add the detection standard of the corresponding layer to the detection standard corresponding to the previous level level layer to complete the trouble-free detection standard update; A5: Integrate the fault detection standard update with the no-fault detection standard update to complete the detection standard update.

9. The method of automatic detection technology of weld quality of narrow lap seam welder according to claim 8 is characterized in that: The updating of the detection standard is completed by integrating the updating of the fault detection standard with the updating of the non-fault detection standard, including: The fault detection standard update and the no-fault detection standard update are numbered according to the layered number to obtain the fault detection number, and the fault detection standard update and the no-fault detection standard update are respectively added to the detection standard map of the side image according to the fault detection number to complete the detection standard update.

10. The method of automatic detection technology of weld quality of narrow lap seam welder according to claim 1 is characterized in that: The detecting the side image according to the detection standard to generate the warning signal includes: After completing the update of the detection standard, extract the side image features of the side image in real time; Determine whether the side image features are consistent with the detection standards; if yes, mark the side detection status of the corresponding side image as normal; if no, mark the side detection status of the corresponding side image as abnormal, and generate a warning signal.

Images