Wiring harness crimping quality intelligent detection and self-adaptive control method and wiring harness crimping quality intelligent detection and self-adaptive control system

Abstract

The invention provides a wire harness crimping quality intelligent detection and self-adaptive control method and system, and the method comprises the steps: calculating a stress distribution value of a crimping point through the obtained internal metal flow distribution characteristics in combination with a mechanical strength simulation algorithm, and obtaining a mechanical strength evaluation index; if the mechanical strength evaluation index is lower than a preset threshold value, triggering an alarm signal, extracting deviation type data from the alarm signal, and judging a quality judgment result; generating a feedback control instruction according to the judged quality judgment result, transmitting the feedback control instruction to crimping equipment to adjust crimping force parameters, and obtaining optimized crimping process data; updating a real-time monitoring database by adopting the obtained optimized crimping process data, and determining a geometric parameter reference value of a subsequent crimping batch; and matching the determined geometric parameter reference value with a vehicle safety requirement to obtain a final quality verification report.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of information technology, and in particular to a kind of intelligent detection and adaptive control method and system for wire harness crimping quality. BACKGROUND

[0002] As the nerve network of modern automobile electrical system, automobile wire harness carries the core function of vehicle power transmission and signal communication, and its crimping quality is directly related to the safety and reliability of the vehicle. With the increasing degree of automotive electrification, the precision requirement of wire harness crimping process is becoming increasingly stringent, and any small crimping defect may cause vehicle electrical system failure.

[0003] Current automobile wire harness crimping mainly relies on traditional mechanical crimping equipment combined with manual detection method. This method has obvious shortcomings when faced with complex and variable crimping environment. Traditional detection methods cannot realize real-time monitoring of the crimping process, and can only be checked by sampling after crimping is completed. This lag detection method is difficult to find and correct problems in the crimping process in time. More importantly, manual detection is limited by the experience level and visual ability of the detection personnel, and lacks accurate judgment ability for key quality indicators such as metal flow state inside the crimping point and the tightness of the wire and terminal combination.

[0004] The core of crimping quality lies in the accurate control of crimping geometric parameters, including crimping height, crimping width and wire insertion depth. Small deviations in these geometric parameters will directly affect the contact resistance and mechanical strength of the crimping point, and thus affect the reliability of the entire electrical connection. However, the plastic deformation behavior of metal materials in the crimping process is extremely complex, and slight changes in wire material, terminal shape and crimping force will cause significant differences in crimping geometry. For example, when the crimping height is too small, the wire and terminal do not contact fully, and the contact resistance increases; while the crimping height is too large, it may cause wire damage or terminal deformation. This complex geometric deformation process makes it difficult for traditional detection methods to accurately capture and quantify the key features of crimping quality.

[0005] Therefore, how to establish a detection system that can accurately identify crimping geometric parameters and internal bonding state in real time, and realize accurate measurement and quality judgment of key dimensions such as crimping height, width and insertion depth, has become a key problem in the development of automobile wire harness precision crimping technology. SUMMARY

[0006] The present application provides a kind of intelligent detection and adaptive control method for wire harness crimping quality, mainly including: The original image data of wire harness crimping point is obtained by image acquisition device, and the edge detection algorithm is used to process the image data to extract the crimping geometric profile features, and the crimping height and crimping width values are obtained; The extracted crimping height and crimping width values are compared with a preset standard range, and if the crimping height exceeds a preset threshold, the height deviation state is marked, and whether the insertion depth meets the requirements is judged; A deep learning model is used to analyze the conductor and terminal combination image from the marked height deviation state to obtain internal metal flow distribution characteristics and determine the potential increase risk of contact resistance; The stress distribution value of the crimping point is calculated by combining the obtained internal metal flow distribution characteristics with a mechanical strength simulation algorithm to obtain a mechanical strength evaluation index; If the mechanical strength evaluation index is lower than a preset threshold, an alarm signal is triggered, deviation type data is extracted from the alarm signal, and a quality determination result is judged; According to the judged quality determination result, a feedback control instruction is generated and transmitted to the crimping equipment to adjust the crimping force parameter, and the optimized crimping process data is obtained; The obtained optimized crimping process data is used to update the real-time monitoring database to determine the geometric parameter reference value of the subsequent crimping batch; The determined geometric parameter reference value is matched with the vehicle safety requirements to obtain a final quality verification report.

[0007] The present application provides a wire harness crimping quality intelligent detection and adaptive control system, mainly comprising: An image acquisition and processing module is used to obtain original image data for wire harness crimping points through an image acquisition device, and an edge detection algorithm is used to process the image data to extract crimping geometric contour features to obtain crimping height and crimping width values; A geometric feature comparison module is used to compare the extracted crimping height and crimping width values with a preset standard range, and if the crimping height exceeds a preset threshold, the height deviation state is marked, and whether the insertion depth meets the requirements is judged; A deep learning analysis module is used to analyze the conductor and terminal combination image from the marked height deviation state using a deep learning model to obtain internal metal flow distribution characteristics and determine the potential increase risk of contact resistance; A mechanical strength evaluation module is used to calculate the stress distribution value of the crimping point by combining the obtained internal metal flow distribution characteristics with a mechanical strength simulation algorithm to obtain a mechanical strength evaluation index; An alarm triggering module is used to trigger an alarm signal if the mechanical strength evaluation index is lower than a preset threshold, extract deviation type data from the alarm signal, and judge the quality determination result; A feedback control module is used to generate a feedback control instruction according to the judged quality determination result, transmit it to the crimping equipment to adjust the crimping force parameter, and obtain the optimized crimping process data; a data updating module, configured to update the real-time monitoring database with the obtained optimized crimping process data, and determine the geometric parameter reference value of the subsequent crimping batch; a quality verification module, configured to match the determined geometric parameter reference value with vehicle safety requirements, and obtain a final quality verification report.

[0008] The technical scheme provided by the embodiment of the present application can include the following beneficial effects: The present application discloses a kind of harness crimping quality intelligent detection and adaptive control method, for the technical problem that internal metal flow state is difficult to monitor in real time and predict contact resistance risk in traditional crimping process, obtains crimping point original data by image acquisition equipment, uses edge detection algorithm to extract geometric contour feature to obtain crimping height and width value, when detecting height deviation state, using deep learning model analysis wire and terminal combination image, obtain internal metal flow distribution characteristics and determine contact resistance potential increase risk, combined with mechanical strength simulation algorithm calculation stress distribution value obtains mechanical strength evaluation index, if index is lower than preset threshold, then trigger alarm and extract deviation type data to determine quality, according to determination result generates feedback control instruction adjustment crimping equipment parameter, crimping process data is updated to real-time monitoring database after optimization, determine the geometric parameter reference value of subsequent batch and match with vehicle safety requirements to generate final quality verification report, realize the closed-loop management of crimping quality intelligent detection, risk prediction and adaptive control. BRIEF DESCRIPTION OF DRAWINGS

[0009] Figure 1 It is a flow chart of the harness crimping quality intelligent detection and adaptive control method of the present application.

[0010] Figure 2 It is a schematic diagram of the harness crimping quality intelligent detection and adaptive control method of the present application.

[0011] Figure 3 It is another schematic diagram of the harness crimping quality intelligent detection and adaptive control method of the present application.

[0012] Figure 4 It is a structural schematic diagram of the harness crimping quality intelligent detection and adaptive control system of the present application. DETAILED DESCRIPTION

[0013] The technical scheme of the present application will be described clearly and completely below in conjunction with embodiments. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0014] AsFigures 1-4 The embodiment of the wire harness crimping quality intelligent detection and adaptive control method and system can specifically include: S101, acquiring original image data of the wire harness crimping point through an image acquisition device, processing the image data by an edge detection algorithm to extract crimping geometric profile features, and obtaining crimping height and crimping width values.

[0015] The crimping size data is verified by an image data analysis algorithm to determine whether the crimping size data meets the preset specification requirements, and a verification result is obtained. According to the verification result, a support vector machine algorithm is used to classify the quality of the crimping point to determine the quality grade of the crimping point. If the quality grade of the crimping point is lower than the preset threshold, the first image data is enhanced by an image processing algorithm to obtain second image data. The second image data is reprocessed by a Canny edge detection algorithm to extract enhanced geometric profile features, and an updated profile feature set is obtained. According to the updated profile feature set, the least square method is used to re-fit the geometric shape of the crimping point profile to obtain updated geometric parameters. The crimping height value and the crimping width value are calculated by the updated geometric parameters to obtain updated crimping size data. According to the updated crimping size data, the support vector machine algorithm is used to evaluate the quality of the crimping point again to determine the final quality grade.

[0016] For example, an industrial-grade CCD camera is used to capture images of the wire harness crimping point in combination with a macro lens. The camera resolution is set to 2048x1536 pixels, the exposure time is 5 milliseconds, and the gain value is set to 128. This ensures that the crimping point occupies at least 800x600 pixel area in the image to ensure subsequent measurement accuracy. The acquired original RGB image is first converted to a grayscale image using a weighted average method with a ratio of 0.299R+0.587G+0.114B. To eliminate noise interference, a 5x5 Gaussian filter is used to smooth the image with a standard deviation of 1.2. Then the Canny edge detection algorithm is applied to extract the crimping profile, with a low threshold of 50 and a high threshold of 150, and the gradient direction is calculated using the Sobel operator. The edge breakpoints are filled by morphological closing operation, and a 3x3 rectangular structural element is used for 2 iterations. The maximum connected region is identified as the main profile of the crimping point using the contour tracking algorithm, and the minimum bounding rectangle of the contour is calculated to obtain the crimping geometric parameters. Based on the pixel calibration coefficient of 0.025 mm per pixel, the pixel coordinates are converted to actual physical dimensions to obtain a crimping height of 2.85 mm and a crimping width of 4.12 mm. The crimping quality is evaluated by calculating the ratio of the contour area to the standard ellipse area, and the product is determined as qualified when the ratio is within the range of 0.85 to 1.15.

[0017] S102, compare the extracted crimping height and crimping width values with the preset standard range, if the crimping height exceeds the preset threshold, mark it as a height deviation state, and determine whether the insertion depth meets the requirements.

[0018] According to the comparison of the crimping height value and the preset height range, if it exceeds, mark the height deviation state. Obtain the deviation amplitude data according to the height deviation state. Use linear regression algorithm to predict the insertion depth adjustment from the deviation amplitude data. Correct the original insertion depth value according to the insertion depth adjustment to obtain the corrected depth value. Compare the corrected depth value with the preset depth range to determine the depth compliance state. If the depth compliance state does not meet the requirements, trigger the secondary acquisition signal to obtain updated image data. Use support vector machine algorithm to classify the crimping quality level from the updated image data to determine the final level.

[0019] For example, a crimping quality judgment database is established, the standard crimping height range is set to 2.5 to 3.2 mm, and the crimping width range is set to 3.8 to 4.5 mm. When the detected crimping height value exceeds the upper limit of 3.2 mm, the system automatically triggers the height deviation warning mechanism and marks it as "over-pressing state". A multi-layer neural network classifier is used to comprehensively evaluate the crimping parameters. The input layer includes three characteristic values of crimping height, width, and crimping force. The hidden layer is set to 64 neurons, the activation function uses ReLU function, and the learning rate is set to 0.001. The internal structure image of the crimping point is obtained by X-ray perspective detection technology, the tube voltage is set to 80kV, the tube current is set to 5mA, and the exposure time is set to 200ms. Detect whether the wire insertion depth reaches more than 75% of the terminal length. Use gray value distribution histogram analysis method to identify the interface between the wire and the terminal. When the gray value difference is greater than 30, it is determined as an effective interface, the insertion depth is calculated as 78.5% of the total length of the terminal, i.e. 5.34 mm, which meets the insertion depth requirement. A crimping quality scoring model is established, and a weighted summation algorithm is used to calculate the comprehensive score, where the height deviation weight is 0.4, the width deviation weight is 0.3, and the insertion depth weight is 0.3. When the comprehensive score is less than 0.7, the rework process is automatically triggered.

[0020] S103, use a deep learning model to analyze the wire and terminal combination image from the marked height deviation state, obtain the internal metal flow distribution characteristics, and determine the potential risk of increased contact resistance.

[0021] The height deviation state marking data is acquired, from which the conductor terminal combination image sample is screened out, and an image data set is established. A convolutional neural network is used to extract features of the combination image, identify the texture pattern of internal metal flow, and the density change area. According to the metal flow texture pattern, a flow continuity index is calculated, and if the continuity index is lower than a preset threshold, it is marked as a flow uneven state. Through pixel gray value analysis of the flow uneven state area, the effective coverage rate value of the metal contact area is obtained. A support vector machine algorithm is used to classify the resistance risk level from the contact area coverage rate value, and determine three categories of low risk, medium risk and high risk. According to the resistance risk level, a preset resistance increment range is matched, and the expected change amplitude value of the contact resistance is calculated. If the expected change amplitude exceeds the safe resistance range, a risk warning signal is generated, and the corresponding conductor terminal position coordinate information is recorded.

[0022] For example, a convolutional neural network model is constructed to perform deep feature extraction on the labeled height deviation crimping sample. The network architecture uses ResNet-50 as the backbone network, the input layer receives a 512x512 pixel crimping cross-section image, and the metal deformation texture features are extracted through three convolutional layer combinations. The convolution kernel size is set to 3x3, the step is 1, and the padding method is same padding. During model training, the Adam optimizer is used, the initial learning rate is set to 0.0005, the batch size is 32, and the training rounds are 150 rounds. For metal flow distribution feature recognition, a semantic segmentation algorithm is used to divide the crimping area into three key areas: copper wire deformation area, terminal wall thickness change area, and gap distribution area. The metal flow degree is quantified by calculating the pixel density distribution difference of each area. When the copper wire deformation area ratio is less than 65%, it indicates that the metal flow is insufficient, and the reduction of the contact area will lead to an increase in the contact resistance. A resistance prediction regression model is established, taking the metal flow area ratio, terminal wall thickness change rate, and gap density as input features. The support vector regression algorithm is used to predict the contact resistance value. Model verification shows that when the metal flow area ratio is 62.3%, the wall thickness change rate is 18.7%, and the gap density is 0.24, the predicted contact resistance is 2.8 milliohms, which exceeds the standard threshold of 2.0 milliohms. The system automatically marks the high-risk crimping point and generates a quality warning report.

[0023] S104, calculate the stress distribution value of the crimping point by combining the acquired internal metal flow distribution feature with the mechanical strength simulation algorithm, and obtain the mechanical strength evaluation index.

[0024] The three-dimensional geometric coordinate data and material attribute parameters of the crimping point are acquired, a finite element grid model of the crimping point is established, and the grid nodes and unit structures are divided. According to the metal flow distribution characteristic data, the material density variation region is set, and the non-uniform material attribute distribution state is defined in the finite element model. The finite element analysis method is used to apply the preset load condition to the crimping point model, the stress tensor component numerical value of each grid node is calculated, and the overall stress field distribution matrix of the crimping point is obtained. If the node stress value exceeds the material yield strength threshold value, the node is marked as a high stress risk area. The maximum principal stress and minimum principal stress numerical values of the crimping point are calculated through the stress field distribution matrix data, and the stress concentration coefficient and stress gradient change rate parameters are obtained. According to the stress concentration coefficient, the weak position distribution of the crimping point is judged, and the coordinate range of the potential failure area is determined. The fatigue life prediction model is used to analyze the material response characteristics under the stress cycle loading condition, and the fatigue damage accumulation rate of the crimping point under the alternating stress action is calculated. If the fatigue damage accumulation rate exceeds the preset safety threshold value, it is judged that the crimping point has a fatigue failure risk. According to the stress distribution numerical value and the fatigue damage parameter, a mechanical strength comprehensive evaluation index system is established, the strength safety factor and reliability probability numerical value of the crimping point are calculated. The mechanical performance grade of the crimping point is classified through the strength safety factor numerical value, and three strength grades of excellent, qualified and unqualified are determined. The stress response curve data of the crimping point under different load working conditions are acquired, the stress amplitude variation trend and load transmission path characteristics are analyzed. According to the load transmission path, the structure design parameters of the crimping point are optimized, the contact area and crimping depth numerical values are adjusted, and the stress distribution uniformity is improved.

[0025] Exemplarily, based on the extracted metal flow distribution feature data, the system uses a finite element analysis algorithm to construct a crimping point three-dimensional stress field model, and converts the density distribution data of the metal flow area into material attribute parameter input for simulation calculation. The algorithm first divides the crimping area into a grid, generating a finite element grid containing 8642 tetrahedral elements, with each grid element size controlled within 0.05 millimeters to ensure calculation accuracy. According to the metal flow feature recognition result, the system sets the elastic modulus of the copper wire to 110 GPa and the Poisson's ratio to 0.34, and sets the elastic modulus of the terminal material to 200 GPa and the Poisson's ratio to 0.30. In the mechanical strength simulation process, the algorithm applies boundary conditions based on the crimping process parameters, sets the crimping force to 1200 Newton, and sets the crimping depth to 1.2 millimeters. The Newton-Raphson iteration method is used to solve the nonlinear equation set. The calculation shows that the maximum equivalent stress inside the crimping point is 285 MPa, the average stress is 156 MPa, and the stress concentration coefficient is 1.83. The system further analyzes the stress distribution uniformity, and calculates the stress standard deviation to be 42.7 MPa. When the stress distribution unevenness exceeds 30%, it indicates that there is a risk of local stress concentration in the crimping point. Combined with the material yield strength threshold of 220 MPa, the system determines that the current crimping point mechanical strength safety factor is 0.77, which is lower than the standard requirement of 1.2, and automatically generates a quality evaluation report of insufficient mechanical strength.

[0026] S105, if the mechanical strength evaluation index is lower than the preset threshold value, an alarm signal is triggered, deviation type data is extracted from the alarm signal, and a quality judgment result is determined.

[0027] The resistance-temperature correlation database is established by acquiring the crimping point resistance measurement data and temperature change curve, and recording the resistance value change law of the crimping point under different working conditions. If the resistance value exceeds the preset normal range, the abnormal detection module is triggered, and the crimping point is marked as an abnormal state of electrical performance. According to the marked state output by the abnormal detection module, the decision tree algorithm is used to analyze the cause type of resistance abnormality, and to judge whether the abnormality is caused by poor contact, material aging or environmental factors. The abnormal cause classification result and the corresponding severity level value are obtained. By starting the corresponding quality evaluation process through the severity level value, the surface oxide layer thickness data and the contact surface roughness parameters of the crimping point are collected. If the oxide layer thickness exceeds the limit value of the material standard specification, it is determined that the crimping point has surface quality defects. According to the surface quality defect determination result, the microstructure scanning image data of the crimping point is obtained, and the grain boundary distribution state and the phase change area range are analyzed. The image segmentation technology is used to extract the defect area contour coordinates, and the defect area proportion and distribution density parameters are calculated. By the defect area proportion parameter, a quality level classification model is established to divide the quality of the crimping point into four levels: excellent, good, qualified and unqualified. If the quality level is determined to be unqualified, a quality abnormality report is generated, and the defect position coordinates and repair suggestion scheme are recorded. According to the defect position coordinates in the quality abnormality report, the remaining service life prediction value and the maintenance cycle suggestion time of the crimping point are calculated. The life prediction data and maintenance suggestion parameters are obtained, and the crimping point state monitoring database record is updated. The historical record data in the state monitoring database is used to analyze the quality change trend and failure mode evolution law of the crimping point. According to the evolution law, a preventive maintenance strategy is established to determine the optimal detection frequency and replacement time node.

[0028] For example, the system establishes a multi-level threshold decision mechanism based on the mechanical strength evaluation results, sets the critical strength safety factor as 1.0 as the first-level alarm trigger condition, when the calculated safety factor 0.77 is lower than the threshold, the system immediately starts the Kalman filter algorithm to analyze the trend of the strength data in the next 5 sampling periods, and finds that the strength decay rate reaches 15.3% per cycle. The alarm signal is sampled by the digital signal processor at a frequency of 2kHz, and the fast Fourier transform algorithm is used to analyze the signal spectrum characteristics, and the main frequency component is identified at 450Hz, with an amplitude of 3.2V, corresponding to the type of serious mechanical strength shortage. The system further classifies the alarm signal using a pattern recognition algorithm, and classifies the deviation type into three categories of material defects, process parameter abnormalities and structural deformation through a support vector machine classifier. The current signal feature vector [3.2, 450, 0.77] is calculated by the trained classification model, and the probability of belonging to the process parameter abnormality category is 0.89. Based on the Bayesian decision theory, the system combines the statistical analysis of 4500 samples in the historical quality database, and calculates that the current crimping point qualified probability is only 0.23, which is far lower than the quality standard requirement of 0.95, and finally determines that the crimping point is unqualified product, and generates a complete quality judgment report containing deviation type code WP-002, risk level high, and recommended treatment measures.

[0029] S106, generate feedback control instructions according to the quality judgment result, and transmit to the crimping equipment to adjust the crimping force parameter, and obtain the optimized crimping process data.

[0030] The quality judgment result data is obtained, the quality level is converted into a corresponding force adjustment coefficient by using a mapping table, and a feedback control instruction containing a device address and a parameter correction value is generated. According to the device address information in the feedback control instruction, the control instruction is transmitted to the parameter adjustment module of the target crimping device through the industrial communication protocol, triggering the device to perform real-time adjustment of the crimping force parameter. The correction value data received by the parameter adjustment module is used to update the force control register value of the crimping device, and if the correction value exceeds the safe operating range of the device, it is limited within the preset boundary value to determine the final crimping force execution parameter. The crimping force execution parameter drives the hydraulic system or pneumatic system of the device to adjust the mechanical pressure applied by the crimping head, and the adjusted crimping head displacement sensor data and pressure sensor data are obtained. According to the pressure sensor data and displacement sensor data, the Kalman filter algorithm is used to process the sensor noise signal, and the actual force curve and crimping depth change trajectory during crimping are calculated. The actual force curve data is analyzed for deviation from the preset standard crimping process curve, and if the deviation is within the allowable error range, the current crimping cycle is recorded as a success state, and a crimping process record containing a timestamp and process parameters is generated. The process parameter data in the crimping process record is used to update the device operating state database, establish a correlation table between the quality judgment result and the device parameter adjustment effect, and form a closed-loop control of historical data accumulation.

[0031] For example, after the system receives the quality judgment result, it immediately starts the adaptive control algorithm to generate accurate feedback control instructions. The controller uses the PID algorithm to calculate the crimping force correction value. When it is detected that the current crimping force is 2850N while the standard value should be 3200N, the system calculates a force increment of 350N through a proportional coefficient of 0.65, an integral coefficient of 0.12, and a differential coefficient of 0.08. The control instruction is transmitted to the crimping device at a rate of 500kbps through the CAN bus protocol. After the device receives the hexadecimal instruction code 0x4A2F, the servo motor drive system automatically adjusts the hydraulic cylinder pressure from the current 8.5MPa to 9.8MPa. The crimping head displacement sensor monitors the depth of depression in real time, and when the depth is adjusted from the original 1.2mm to the standard 1.45mm, the force sensor synchronously collects new crimping force data of 3185N, close to the target value. The system uses the least squares method to perform fitting analysis on the adjusted 10 groups of continuous crimping data, and calculates that the crimping force standard deviation is reduced to 15.2N, which is significantly improved compared to the adjustment before 42.8N. The temperature compensation algorithm fine-tunes the crimping parameters according to the environmental temperature of 23.5℃, and finally obtains the optimized crimping process data containing crimping force 3185N, holding time 2.3 seconds, and springback amount 0.08mm. The data is uploaded to the quality management system through the Ethernet protocol for closed-loop verification.

[0032] S107, update the real-time monitoring database with the obtained optimized crimping process data to determine the geometric parameter reference value of the subsequent crimping batch.

[0033] The optimized crimping process data is obtained, the crimping process data is analyzed by the data analysis module, the geometric parameter information of the crimping depth, the contact area and the deformation amount is extracted, the mean value and the standard deviation of each parameter are calculated by using the statistical analysis method according to the numerical range in the geometric parameter information, the parameter is marked as an abnormal data point if the parameter fluctuation exceeds the preset tolerance range, and an effective geometric parameter set is obtained. The process record table in the real-time monitoring database is updated through the effective geometric parameter set, the association index of the geometric parameter and the current batch identification is established, and the data record entry containing the time stamp and the batch number is obtained. The batch number information in the data record entry is used to query the geometric parameter statistical result of the historical crimping batch, the parameter similarity score between different batches is calculated by using the weighted average algorithm, and the selection basis of the reference batch is determined. The corresponding geometric parameter reference value is extracted from the historical process data according to the selection basis of the reference batch, the default reference value is used if the similarity score of the current batch and the reference batch is lower than the threshold value, and the geometric parameter reference configuration suitable for the subsequent batch is obtained. The parameter template file containing the crimping depth reference, the contact area reference and the deformation amount reference is generated through the geometric parameter reference configuration, the mapping table structure of the batch identification and the reference value is established, and the standardized geometric parameter reference data set is obtained. The reference value update instruction is sent to the parameter management module of the crimping equipment by using the standardized geometric parameter reference data set, the value refreshing operation of the internal reference parameter register of the equipment is triggered, and the geometric parameter control standard during the execution of the subsequent crimping batch is determined.

[0034] Exemplary, the data acquisition module automatically acquires the optimized crimping process data, including the contact resistance value 0.85 mΩ, the crimping depth 1.45 mm, the contact area 12.8 mm² and other key geometric parameters, the system uses Kalman filtering algorithm to preprocess the collected noise data, and the filtering gain is set to 0.75, effectively removing high-frequency interference signals. The database update engine uses the time stamp index mechanism to write the newly acquired parameter values into the PostgreSQL real-time monitoring database, and the update frequency is set to 50 times per second, ensuring the real-time and integrity of the data. The system analyzes the historical crimping data trend based on the Bayesian inference algorithm, and calculates that the confidence interval of the contact resistance is 0.82-0.88 mΩ, and the control range of the crimping depth is 1.42-1.48 mm. The parameter reference value determination module uses the weighted average algorithm to process the crimping data of the last 100 batches, and the weight coefficient is set to decrease by time, and the latest data weight is 1.0, and the weight decreases by 0.02 for each batch. The algorithm calculates that the subsequent batch contact resistance reference value is 0.84 mΩ, the crimping depth reference value is 1.46 mm, and the contact area reference value is 12.5 mm². The system also establishes a dynamic threshold model, which automatically adjusts the tolerance range of the reference value according to the batch difference of the material and the change of the environment. When it is detected that the deviation of the diameter of the new batch of copper wire exceeds 0.05 mm, the system automatically corrects the crimping depth reference value to 1.48 mm, ensuring the stability and consistency of the subsequent crimping quality.

[0035] S108, match the determined geometric parameter reference value with the vehicle safety requirements to obtain a final quality verification report.

[0036] A crimping quality specification document in a vehicle safety standard database is acquired, geometric parameter limit information in the safety standard is extracted through a document parser, a standard parameter set containing minimum value, maximum value and tolerance range is established, and a structured safety standard data table is obtained. According to the parameter limit range in the safety standard data table, a numerical comparison algorithm is used to perform item-by-item matching verification on the geometric parameter reference value, and if the reference value exceeds the standard limit range, it is marked as an unqualified item, and matching result data containing compliance state identification is obtained. Through the compliance state information in the matching result data, a preset risk assessment rule library is queried, the comprehensive risk level score is calculated according to the number and severity of unqualified items, and the quality risk classification standard of the current batch is determined. Using the quality risk classification standard, the actual measurement data of the corresponding batch is extracted from the detection data management system, the deviation value and stability index of each geometric parameter are calculated through statistical analysis method, and a quantitative quality evaluation result set is obtained. According to the deviation value distribution in the quality evaluation result set, a predefined audit judgment logic module is called, if the deviation value is within the acceptable range, a qualified judgment is generated, otherwise an unqualified judgment is generated, and the final quality audit conclusion is determined. Through the quality audit conclusion and the risk level information, the corresponding verification report format file is obtained by accessing the report template library, the matching result, the evaluation data and the audit conclusion are filled in according to the template structure, and the quality verification document containing the timestamp and batch identification is generated. Using the structured data in the quality verification document, the document content is converted into a standardized output format through a report generation engine, an association index between the verification result and the batch record is established, and a complete final quality verification report is obtained.

[0037] Exemplarily, the quality verification system automatically calls a vehicle safety standard database, extracts technical requirements for a wire harness connector in the ISO 26262 functional safety standard, including key indicators such as a maximum allowable contact resistance of 0.95 mΩ, a minimum pulling force of 150 N, and an insulation resistance of not less than 1000 MΩ. The matching algorithm uses a multi-dimensional similarity calculation method to compare the contact resistance reference value of the current batch of 0.84 mΩ with the upper limit of the safety standard of 0.95 mΩ, and calculates the safety margin to be 11.6%, meeting the reliability requirements of vehicle-grade products. The system uses a fuzzy logic reasoning engine to evaluate the safety of the crimping depth reference value of 1.46 mm, maps the depth parameter to a safety level scoring system by establishing a membership function model, and calculates that the safety score of the current parameter is 8.7 (full score 10). The temperature cycle test simulation module is based on a finite element analysis algorithm to predict the thermal expansion coefficient change of the crimping point under the condition of -40°C to 125°C, and calculates that the contact area changes in the range of ±0.3 mm² under extreme temperature, verifying the stability of the reference value of 12.5 mm² in the full temperature range. The vibration durability evaluation algorithm uses the rainflow counting method to analyze the stress cycle of the crimping point during vehicle operation, and combines the Weibull distribution model to predict the fatigue life of the connector to be 150,000 kilometers, exceeding the vehicle design life requirement. The final quality verification report generator integrates all evaluation results, calculates the comprehensive quality index to be 92.3% using the analytic hierarchy process, and automatically generates a standardized verification report containing parameter compliance, safety margin analysis, reliability prediction, and the like.

[0038] The present application provides a wire harness crimping quality intelligent detection and adaptive control system, mainly comprising: An image acquisition and processing module is configured to acquire original image data of a wire harness crimping point through an image acquisition device, and to process the image data using an edge detection algorithm to extract crimping geometric profile features, and to obtain crimping height and crimping width values. A geometric feature comparison module is configured to compare the extracted crimping height and crimping width values with a preset standard range, and to mark a height deviation state if the crimping height exceeds a preset threshold, and to determine whether the insertion depth meets the requirements. A deep learning analysis module is configured to analyze the wire and terminal combination image from the marked height deviation state using a deep learning model, to obtain internal metal flow distribution features, and to determine the potential risk of increased contact resistance. A mechanical strength evaluation module is configured to calculate stress distribution values of the crimping point by combining the obtained internal metal flow distribution features with a mechanical strength simulation algorithm, and to obtain mechanical strength evaluation indicators. An alarm triggering module is configured to trigger an alarm signal if the mechanical strength evaluation indicators are lower than a preset threshold, to extract deviation type data from the alarm signal, and to determine the quality determination result. The feedback control module is used for generating a feedback control instruction according to the quality judgment result, transmitting the feedback control instruction to the crimping equipment to adjust a crimping force parameter, and obtaining optimized crimping process data. The data updating module is used for updating the real-time monitoring database by using the obtained optimized crimping process data, and determining a geometric parameter reference value of a subsequent crimping batch. The quality verification module is used for matching the determined geometric parameter reference value with vehicle safety requirements to obtain a final quality verification report.

[0039] The above content is merely an example and description of the structure of the present application. Those skilled in the art can make various modifications or supplements to the described specific embodiments or use similar ways to replace them, as long as they do not deviate from the structure of the present application or exceed the scope defined by the present claims.

Claims

1. A wire harness crimping quality intelligent detection and adaptive control method, characterized in that, The method comprises: acquiring original image data for a wire harness crimping point through an image acquisition device, processing the image data using an edge detection algorithm to extract crimping geometric profile features, and obtaining crimping height and crimping width values; comparing the extracted crimping height and crimping width values with a preset standard range, and if the crimping height exceeds a preset threshold, marking as a height deviation state and determining whether the insertion depth meets the requirements; using a deep learning model to analyze the wire and terminal combination image from the marked height deviation state, obtaining internal metal flow distribution features, and determining the potential increase risk of contact resistance; calculating the stress distribution value of the crimping point through the obtained internal metal flow distribution features combined with a mechanical strength simulation algorithm, and obtaining a mechanical strength evaluation index; if the mechanical strength evaluation index is lower than a preset threshold, triggering an alarm signal, extracting deviation type data from the alarm signal, and determining a quality determination result; generating a feedback control instruction according to the determined quality determination result, transmitting the feedback control instruction to the crimping device to adjust the crimping force parameter, and obtaining optimized crimping process data; updating the real-time monitoring database using the obtained optimized crimping process data, and determining the geometric parameter reference value of the subsequent crimping batch; matching the determined geometric parameter reference value with vehicle safety requirements to obtain a final quality verification report.

2. The method of claim 1, wherein the method further comprises: The method comprises: verifying the crimping size data through an image data analysis algorithm, determining whether the crimping size data meets the preset specification requirements, and obtaining a verification result; according to the verification result, using a support vector machine algorithm to classify the quality of the crimping point, and determining the quality grade of the crimping point; if the quality grade of the crimping point is lower than a preset threshold, enhancing the first image data through an image processing algorithm to obtain second image data; reprocessing the second image data using a Canny edge detection algorithm to extract enhanced geometric profile features, and obtaining an updated profile feature set; according to the updated profile feature set, using a least squares method to re-fit the geometric shape of the crimping point profile, and obtaining updated geometric parameters; calculating the crimping height value and the crimping width value through the updated geometric parameters, and obtaining updated crimping size data; according to the updated crimping size data, using a support vector machine algorithm to re-evaluate the quality of the crimping point, and determining a final quality grade.

3. The method of claim 1, wherein the method further comprises: The method comprises: comparing the crimping height value with a preset height range, and if it exceeds, marking a height deviation state; obtaining deviation amplitude data according to the height deviation state; using a linear regression algorithm to predict the insertion depth adjustment amount from the deviation amplitude data; correcting the original insertion depth value according to the insertion depth adjustment amount to obtain a corrected depth value; comparing the corrected depth value with a preset depth range to determine the depth compliance state; If the depth meets the state of non-compliance, a secondary acquisition signal is triggered to obtain updated image data; The support vector machine algorithm is used to classify the crimping quality level from the updated image data to determine the final level.

4. The method of claim 1, wherein the method further comprises: The depth learning model is used to analyze the wire and terminal combination image from the labeled height deviation state, obtain the internal metal flow distribution characteristics, and determine the potential increase risk of contact resistance, including: Labeled height deviation state data is obtained, from which wire terminal combination image samples are screened out to establish an image data set; A convolutional neural network is used to extract features from the combination image, identify the texture pattern and density change area of the internal metal flow, and calculate the flow continuity index according to the metal flow texture pattern. If the continuity index is lower than the preset threshold, it is marked as a flow uneven state. Through pixel gray value analysis of the flow uneven state area, the effective coverage rate value of the metal contact area is obtained. The support vector machine algorithm is used to classify the resistance risk level from the contact area coverage rate value to determine three categories of low risk, medium risk, and high risk. According to the resistance risk level, the expected change amplitude value of the contact resistance is calculated by matching the preset resistance increment range. If the expected change amplitude exceeds the safe resistance range, a risk warning signal is generated, and the corresponding wire terminal position coordinate information is recorded.

5. The method of claim 1, wherein, The stress distribution value of the crimping point is calculated by combining the obtained internal metal flow distribution characteristics with the mechanical strength simulation algorithm to obtain the mechanical strength evaluation index, including: Three-dimensional geometric coordinate data and material attribute parameters of the crimping point are obtained to establish a crimping point finite element grid model, and the grid nodes and unit structures are divided. According to the metal flow distribution characteristic data, the material density change area is set, and the non-uniform material attribute distribution state is defined in the finite element model. The finite element analysis method is used to apply a preset load condition to the crimping point model to calculate the stress tensor component values of each grid node and obtain the overall stress field distribution matrix of the crimping point. If the node stress value exceeds the material yield strength threshold, the node is marked as a high stress risk area. The maximum principal stress and minimum principal stress values of the crimping point are calculated through the stress field distribution matrix data to obtain the stress concentration coefficient and stress gradient change rate parameters. According to the stress concentration coefficient, the weak position distribution of the crimping point is determined to determine the potential failure area coordinate range. The fatigue life prediction model is used to analyze the material response characteristics under stress cycle loading conditions to calculate the fatigue damage accumulation rate of the crimping point under alternating stress. If the fatigue damage accumulation rate exceeds the preset safety threshold, it is determined that the crimping point has a fatigue failure risk. A mechanical strength comprehensive evaluation index system is established according to the stress distribution values and fatigue damage parameters to calculate the strength safety factor and reliability probability values of the crimping point. The mechanical performance level of the crimping point is classified by the strength safety factor value to determine three strength levels of excellent, qualified, and unqualified. The stress response curve data of the crimping point under different load conditions are obtained to analyze the stress amplitude variation trend and load transmission path characteristics. According to the load transmission path, the structure design parameters of the crimping point are optimized, the contact area and crimping depth values are adjusted, and the stress distribution uniformity is improved.

6. The method of claim 1, wherein the method further comprises: The mechanical strength evaluation index below the preset threshold triggers an alarm signal, and deviation type data is extracted from the alarm signal to determine the quality determination result, including: Obtain the crimping point resistance measurement data and temperature change curve, establish the resistance-temperature correlation database, and record the resistance value change law of the crimping point under different working conditions; If the resistance value exceeds the preset normal range, the abnormal detection module is triggered, and the crimping point is marked as an electrical performance abnormal state; According to the marking state output by the abnormal detection module, the decision tree algorithm is used to analyze the cause type of resistance abnormality, and it is judged that the abnormality is caused by poor contact, material aging or environmental factors; Obtain the abnormal cause classification result and the corresponding severity level value; Start the corresponding quality evaluation process through the severity level value, collect the surface oxide layer thickness data and contact surface roughness parameters of the crimping point; If the oxide layer thickness exceeds the material standard specification limit, it is determined that the crimping point has surface quality defects; According to the surface quality defect determination result, obtain the microstructure scanning image data of the crimping point, analyze the grain boundary distribution state and phase change area range; Use image segmentation technology to extract the defect area contour coordinates, calculate the defect area ratio and distribution density parameters; Through the defect area ratio parameter, a quality grade classification model is established to divide the crimping point quality into four grades: excellent, good, qualified and unqualified; If the quality grade is determined to be unqualified, generate a quality abnormality report, record the defect position coordinates and repair proposal; According to the defect position coordinates in the quality abnormality report, calculate the remaining service life prediction value and maintenance cycle suggestion time of the crimping point; Obtain the life prediction data and maintenance suggestion parameters, and update the crimping point state monitoring database record; Use the historical record data in the state monitoring database to analyze the quality change trend and failure mode evolution law of the crimping point; According to the evolution law, establish a preventive maintenance strategy to determine the optimal detection frequency and replacement timing node.

7. The method of claim 1, wherein the method further comprises: The feedback control instruction is generated according to the quality determination result, and is transmitted to the crimping equipment to adjust the crimping force parameter, and the optimized crimping process data is obtained, including: Obtain the quality determination result data, convert the quality grade to the corresponding force adjustment coefficient using a mapping table, and generate a feedback control instruction containing the device address and parameter correction value; According to the device address information in the feedback control instruction, the control instruction is transmitted to the parameter adjustment module of the target crimping equipment through the industrial communication protocol, triggering the device to perform real-time adjustment of the crimping force parameter; Use the correction value data received by the parameter adjustment module to update the force control register value of the crimping equipment, and if the correction value exceeds the safe operation range of the equipment, limit it within the preset boundary value to determine the final crimping force execution parameter; Drive the equipment hydraulic system or pneumatic system through the crimping force execution parameter to adjust the mechanical pressure applied by the crimping head, and obtain the adjusted crimping head displacement sensor data and pressure sensor data; According to the pressure sensor data and displacement sensor data, use the Kalman filter algorithm to process the sensor noise signal, calculate the actual force curve and crimping depth change trajectory in the crimping process; The actual force curve data is acquired, deviation analysis is performed on the preset standard crimping process curve, if the deviation value is within the allowable error range, the current crimping cycle is recorded as a successful state, and a crimping process record containing a timestamp and process parameters is generated; The process parameter data in the crimping process record is used to update the equipment operation state database, establish an association table of quality judgment results and equipment parameter adjustment effects, and form a closed-loop control historical data accumulation.

8. The method of claim 1, wherein the method further comprises: The acquired optimized crimping process data is used to update the real-time monitoring database, and the geometric parameter reference values of subsequent crimping batches are determined, including: The optimized crimping process data is acquired, the crimping process data is structurally analyzed by a data analysis module, and the geometric parameter information of the crimping depth, contact area and deformation amount is extracted; According to the value range in the geometric parameter information, the mean value and standard deviation of each parameter are calculated by a statistical analysis method, if the parameter fluctuation exceeds the preset tolerance range, it is marked as an abnormal data point, and an effective geometric parameter set is obtained; Through the effective geometric parameter set, the process record table in the real-time monitoring database is updated, the association index of the geometric parameters and the current batch identification is established, and a data record entry containing a timestamp and a batch number is obtained; The batch number information in the data record entry is used to query the geometric parameter statistical results of the historical crimping batches, the parameter similarity scores between different batches are calculated by a weighted average algorithm, and the selection basis of the reference batch is determined; According to the selection basis of the reference batch, the corresponding geometric parameter reference values are extracted from the historical process data, if the similarity score of the current batch and the reference batch is lower than the threshold value, the default reference value is used, and the geometric parameter reference configuration suitable for subsequent batches is obtained; Through the geometric parameter reference configuration, a parameter template file containing the crimping depth reference, contact area reference and deformation amount reference is generated, a mapping table structure of batch identification and reference value is established, and a standardized geometric parameter reference data set is obtained; The standardized geometric parameter reference data set is used to send a reference value update instruction to the parameter management module of the crimping equipment, trigger the value refreshing operation of the internal reference parameter register of the equipment, and determine the geometric parameter control standard when the subsequent crimping batch is executed.

9. The method of claim 1, wherein the method further comprises: The determined geometric parameter reference value is matched with the vehicle safety requirements to obtain a final quality verification report, including: The crimping quality specification document in the vehicle safety standard database is acquired, the geometric parameter limit value information in the safety standard is extracted by a document parser, a standard parameter set containing the minimum value, maximum value and tolerance range is established, and a structured safety standard data table is obtained; According to the parameter limit value range in the safety standard data table, the geometric parameter reference value is matched and verified item by item by a numerical comparison algorithm, if the reference value exceeds the standard limit value range, it is marked as an illegal item, and matching result data containing a compliance state identification is obtained; Through the compliance state information in the matching result data, the preset risk assessment rule library is queried, the comprehensive risk level score is calculated according to the number and severity of the illegal items, and the quality risk classification standard of the current batch is determined; The actual measurement data corresponding to the batch is extracted from the detection data management system by using the mass risk classification standard, the deviation value and stability index of each geometric parameter are calculated by statistical analysis method, and a set of quantitative quality evaluation results is obtained; According to the deviation value distribution in the quality evaluation result set, the pre-defined audit judgment logic module is called, if the deviation value is within the acceptable range, the qualified judgment is generated, otherwise the unqualified judgment is generated, and the final quality audit conclusion is determined; Through the quality audit conclusion and the risk level information, the corresponding verification report format file is accessed from the report template library, the matching results, evaluation data and audit conclusion are filled in the template structure, and the quality verification document containing timestamp and batch identification is generated; The structured data in the quality verification document is used to convert the document content into a standardized output format through the report generation engine, establish the association index of the verification result and the batch record, and obtain the complete final quality verification report.

10. A wire harness crimping quality intelligent detection and adaptive control system, characterized in that, The system comprises: An image acquisition and processing module is configured to acquire original image data for a wire harness crimping point through an image acquisition device, process the image data using an edge detection algorithm to extract crimping geometric contour features, and obtain crimping height and crimping width values. A geometric feature comparison module is configured to compare the extracted crimping height and crimping width values with a preset standard range, and if the crimping height exceeds the preset threshold, mark it as a height deviation state and determine whether the insertion depth meets the requirements. A deep learning analysis module is configured to analyze the wire and terminal combination image from the marked height deviation state using a deep learning model, obtain internal metal flow distribution features, and determine the potential increase risk of contact resistance. A mechanical strength evaluation module is configured to calculate stress distribution values of the crimping point by combining the obtained internal metal flow distribution features with a mechanical strength simulation algorithm, and obtain a mechanical strength evaluation index. An alarm triggering module is configured to trigger an alarm signal if the mechanical strength evaluation index is lower than a preset threshold, extract deviation type data from the alarm signal, and determine a quality judgment result. A feedback control module is configured to generate a feedback control instruction according to the determined quality judgment result, transmit the feedback control instruction to the crimping device to adjust the crimping force parameter, and obtain optimized crimping process data. A data updating module is configured to update a real-time monitoring database using the obtained optimized crimping process data, and determine geometric parameter reference values for subsequent crimping batches. A quality verification module is configured to match the determined geometric parameter reference values with vehicle safety requirements, and obtain a final quality verification report.

Images