Intelligent cable joint detection method and system

Abstract

This application provides an intelligent cable joint inspection method and system, belonging to the field of intelligent inspection technology. The method includes: acquiring processing parameters and quality inspection data of the current process component during cable joint manufacturing; determining a first defect probability of the component component based on the processing parameters and the corresponding defect assessment model; determining a second defect probability of the component component based on the quality inspection data; determining a third defect probability of the component component based on the first and second defect probabilities; and transferring the component component to the next process if the third defect probability is less than or equal to a preset first probability threshold. The intelligent cable joint inspection method and system provided by this application can achieve early detection and control of cable joint defects.

Description

Technical Field

[0001] This application belongs to the field of intelligent detection technology, and more specifically, relates to an intelligent cable joint detection method and system. Background Technology

[0002] Cable joints are core components in cable laying, connecting two sections of cable and enabling electrical continuity between the cable conductors. They are widely used in power, communication, and other fields, and are critical nodes in cable lines. However, cable joints are also weak points in cable lines. Insulation and sealing failures at cable joints can easily lead to short circuits, leakage, and even fires and power outages, directly affecting the safe and stable operation of power / communication systems. Therefore, cable joints need to undergo comprehensive testing before leaving the factory to promptly identify defects and mitigate safety hazards.

[0003] Current cable factory inspections are mostly conducted on finished cable products, which leads to a delay in defect detection and increases subsequent rework costs. Summary of the Invention

[0004] The purpose of this application is to provide an intelligent cable joint inspection method and system to enable early detection and control of cable joint defects.

[0005] A first aspect of this application provides an intelligent cable joint detection method, comprising: During the cable connector manufacturing process, obtain the processing parameters of the current process and the quality inspection data of the parts manufactured in the current process; Based on the processing parameters of the current process and the defect assessment model corresponding to the current process, the first defect probability of the part in the current process is determined; The probability of a second defect in the part manufactured in the current process is determined based on the quality inspection data. The third defect probability of the part manufactured in the current process is determined based on the first defect probability and the second defect probability. If the probability of the third defect is less than or equal to the preset first probability threshold, the part in the current process will be transferred to the next process.

[0006] A second aspect of this application provides an intelligent cable joint detection system, comprising: The data acquisition module is used to acquire the processing parameters of the current process and the quality inspection data of the parts produced in the current process during the cable joint processing. The first probability calculation module is used to determine the first defect probability of the part in the current process based on the processing parameters of the current process and the defect assessment model corresponding to the current process. The second probability calculation module is used to determine the second defect probability of the part manufactured in the current process based on the quality inspection data. The third probability calculation module is used to determine the third defect probability of the part in the current process based on the first defect probability and the second defect probability. The process control module is used to transfer the current process part to the next process when the probability of the third defect is less than or equal to a preset first probability threshold.

[0007] A third aspect of this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the steps of the intelligent cable connector detection method described above.

[0008] A fourth aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the intelligent cable joint detection method described above.

[0009] The beneficial effects of the intelligent cable joint detection method and system provided in this application are as follows: This embodiment evaluates the defect probability of the part in the current process based on the processing parameters of the current process to obtain a first defect probability, which can predict processing defects caused by deviations in processing parameters in advance. Based on the quality inspection data of the part in the current process, the defect probability of the part in the current process is evaluated to obtain a second defect probability, which can realize a true representation of the defect probability of the part in the current process. On this basis, the first defect probability and the second defect probability are comprehensively considered to determine the third defect probability of the part in the current process. The third defect probability is used as the final defect probability evaluation result, which can improve the accuracy of defect probability evaluation.

[0010] Furthermore, the current process part is automatically transferred to the next process only when the probability of the third defect in the current process part is less than or equal to the preset first probability threshold. This can prevent defective parts from entering subsequent processes and causing defects to accumulate and amplify, thereby enabling early detection and control of cable joint defects. Attached Figure Description

[0011] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 A schematic flowchart of an intelligent cable joint detection method provided in an embodiment of this application; Figure 2This is a structural block diagram of an intelligent cable joint detection system provided in an embodiment of this application; Figure 3 This is a schematic block diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0013] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0014] To make the objectives, technical solutions, and advantages of this application clearer, the following description will be provided in conjunction with the accompanying drawings and specific embodiments.

[0015] Please refer to Figure 1 , Figure 1 This is a flowchart illustrating an embodiment of the intelligent cable joint detection method provided in this application. The method can be executed by an electronic device and may include: S101: Obtain the processing parameters of the current process and the quality inspection data of the parts produced in the current process during the cable connector processing.

[0016] In this embodiment, the cable connector manufacturing process includes multiple steps such as wire stripping, crimping, insulation layer restoration, and sealing. During the current step, the processing parameters can be acquired through the production line's data acquisition system, such as crimping force, crimping depth, and concentricity in the crimping step, and filling pressure, sealant temperature, and pressure holding time in the sealing step. Simultaneously, after the current step is completed, the quality of the finished part (i.e., the part from the current step) can be inspected using specialized production line testing equipment such as a dimensional measuring instrument, a withstand voltage tester, or a surface flatness tester. This yields quality inspection data for the part from the current step, such as cutting dimensional deviations in the cutting step, heat shrinkage rate in the insulation layer restoration step, or withstand voltage test values ​​in the finished product step.

[0017] S102: Based on the processing parameters of the current process and the defect assessment model corresponding to the current process, determine the probability of the first defect of the part in the current process.

[0018] In this embodiment, the processing parameters of the current process can reflect the process status of the part being processed in the current process. Based on the processing parameters of the current process, the processing defects caused by the deviation of the processing parameters can be predicted in advance. Therefore, the first defect probability of the part in the current process can be determined based on the processing parameters of the current process.

[0019] Specifically, considering the differences in processing technology, core parameters, quality inspection indicators, and defect types among different processing steps of cable joints, this embodiment trains a defect assessment model for each step based on the historical processing parameters and corresponding historical first defect probabilities for each step. Then, by inputting the processing parameters of the current step into the corresponding defect assessment model, the first defect probability of the part manufactured in the current step can be obtained. The defect assessment model can be implemented based on existing decision tree models, random forest models, support vector machine models, or convolutional neural network models.

[0020] S103: Determine the probability of a second defect in the part manufactured in the current process based on quality inspection data.

[0021] In this embodiment, the quality inspection data can directly characterize the quality of the part manufactured in the current process. Specifically, a mapping relationship between the quality inspection data and the second defect probability can be established in advance based on the historical data corresponding to the quality inspection data and the second defect probability, respectively. This mapping relationship can be in tabular form, and the second defect probability can be obtained by looking up the table based on the quality inspection data of the part manufactured in the current process.

[0022] Taking the crimping process as an example, the crimping size of the conductor (the outer diameter of the terminal after crimping) directly determines the contact resistance between the conductor and the terminal. Excessive dimensional deviation can easily lead to defects such as overheating and poor contact. The process standard value is φ12.0mm, with an allowable deviation of ±0.2mm. The crimping size deviation = actual measured outer diameter - process standard value. By statistically analyzing the historical data of the crimping size deviation and the corresponding second defect probability, the mapping relationship between the crimping size deviation and the second defect probability can be obtained, as shown in Table 1 below.

[0023] Table 1 - Mapping relationship between crimping dimensional deviation and second defect probability

[0024] It should be noted that S102 can be executed before S103, after S103, or simultaneously. Figure 1 The execution method described is merely an example and is not intended to be limiting.

[0025] S104: Determine the third defect probability of the part in the current process based on the first defect probability and the second defect probability.

[0026] In this embodiment, the third defect probability of the part in the current process can be determined by comprehensively considering the first defect probability and the second defect probability, and the third defect probability is used as the final defect probability evaluation result.

[0027] For example, determining the third defect probability of the part in the current process based on the first defect probability and the second defect probability may include: If the first defect probability is greater than the preset second probability threshold, or the second defect probability is greater than the preset second probability threshold, the maximum probability between the first defect probability and the second defect probability shall be used as the third defect probability of the part in the current process. If the first defect probability is less than or equal to the second defect probability, and both the first and second defect probabilities are less than or equal to the preset second probability threshold, the first and second defect probabilities are weighted and summed to obtain the third defect probability of the part manufactured in the current process.

[0028] In the embodiment, a second defect probability, such as 0.6, can be preset. When the first defect probability is greater than the preset second probability threshold, or the second defect probability is greater than the preset second probability threshold, it indicates that there is a clear defect risk in the current process part. At this time, the maximum probability between the first defect probability and the second defect probability can be used as the third defect probability of the current process part to strictly control the processing quality.

[0029] When the first defect probability is less than and the second defect probability is both less than or equal to the preset second probability threshold, it indicates that there is no obvious defect risk in the part manufactured in the current process. At this time, the first defect probability and the second defect probability can be combined by weighted summation to improve the accuracy of defect probability assessment.

[0030] S105: If the probability of the third defect is less than or equal to the preset first probability threshold, the part in the current process will be transferred to the next process.

[0031] In this embodiment, a second defect probability, such as 0.5, can be preset. If the third defect probability of the current process part is less than or equal to the preset first probability threshold, it indicates that the quality of the current process part is qualified, and the current process part is automatically transferred to the next process to continue the cable connector processing flow. If the third defect probability of the current process part is greater than the preset first probability threshold, it indicates that the quality of the current process part is unqualified, the transfer to the next process is stopped, and the staff is reminded to re-inspect / rework / scrap the current process part.

[0032] In the cable joint processing, the quality status of each process affects each other. If defects in the preceding process are not detected in time, they will accumulate in subsequent processes, eventually leading to the scrapping of the finished product. Moreover, the rework cost will increase exponentially as the process progresses. In this embodiment, when the quality of the part in the current process is unqualified, it can be dealt with directly in the current process, thereby avoiding defective parts from entering subsequent processes and causing defects to accumulate and amplify.

[0033] As can be seen from the above, this embodiment evaluates the defect probability of the part in the current process based on the processing parameters of the current process to obtain the first defect probability, which can predict processing defects caused by deviations in processing parameters in advance; evaluates the defect probability of the part in the current process based on the quality inspection data of the part in the current process to obtain the second defect probability, which can realize the true representation of the defect probability of the part in the current process; on this basis, the third defect probability of the part in the current process is determined by comprehensively considering the first defect probability and the second defect probability, and the third defect probability is used as the final defect probability evaluation result, which can improve the accuracy of defect probability evaluation.

[0034] Furthermore, the current process part is automatically transferred to the next process only when the probability of the third defect in the current process part is less than or equal to the preset first probability threshold. This can prevent defective parts from entering subsequent processes and causing defects to accumulate and amplify, thereby enabling early detection and control of cable joint defects.

[0035] In one embodiment of this application, the cable joint processing includes multiple steps, and the method for determining the defect assessment model corresponding to each step includes: Obtain multiple training samples corresponding to each process; for each process, each training sample includes multiple historical processing parameters of that process and the corresponding historical first defect probability; For each process, based on multiple training samples corresponding to that process, the correlation strength between each historical processing parameter of that process and the corresponding first defect probability is calculated; the correlation strengths of each of the multiple historical processing parameters are arranged in descending order to obtain the correlation strength change sequence corresponding to that process. Calculate the sample uniformity of multiple training samples corresponding to each process; Processes with sample uniformity greater than or equal to the uniformity threshold are classified as first-class processes, and processes with sample uniformity less than the uniformity threshold are classified as second-class processes. For each first type of process, a preset initialization vector is used as the initial weight vector of the pre-trained decision tree model. Based on multiple training samples corresponding to the first type of process, the weight vector of the pre-trained decision tree model is iterated multiple times to obtain the defect assessment model corresponding to the first type of process. For each second-type process, from multiple first-type processes, the first-type process with the highest similarity between its corresponding correlation strength change sequence and the correlation strength change sequence of the second-type process is selected as a reference process, and the defect assessment model corresponding to the reference process is used as the reference assessment model. Based on the weight vector of the reference assessment model, the initial weight vector of the pre-trained decision tree model is determined, and the weight vector of the pre-trained decision tree model is iterated multiple times based on multiple training samples corresponding to the second-type process to obtain the defect assessment model corresponding to the second-type process.

[0036] In this embodiment, multiple processes in the cable joint manufacturing process can all use a pre-trained decision tree model as the basic framework, and train a defect assessment model for each process based on multiple training samples corresponding to each process. Specifically, for each process, each training sample includes multiple historical processing parameters for that process and the corresponding historical first defect probability.

[0037] Specifically, for each process, based on multiple training samples corresponding to that process, the Pearson correlation coefficient (absolute value) or mutual information value between each historical processing parameter of that process and the corresponding first defect probability can be calculated as the correlation strength between each historical processing parameter of that process and the corresponding first defect probability. Arranging the correlation strength of all historical processing parameters in descending order yields the correlation strength change sequence corresponding to that process. The correlation strength change sequence can intuitively characterize the importance ranking of each processing parameter of that process to the first defect probability. The higher the correlation strength of a certain processing parameter, the greater its contribution to the output of the first defect probability, and the greater its corresponding weight in the defect assessment model.

[0038] Simultaneously, the sample uniformity of multiple training samples corresponding to each process can be calculated, and the multiple processes can be divided into a first type of process and a second type of process based on the magnitude of the sample uniformity. For example, a uniformity threshold (e.g., 0.7) can be preset, and processes with a corresponding sample uniformity greater than or equal to the uniformity threshold can be classified as first type processes, while processes with a corresponding sample uniformity less than the uniformity threshold can be classified as second type processes.

[0039] For the first type of process, since the training samples for this process have a high uniformity, a general model training method can be used. Starting from a preset initialization vector (e.g., an all-zero vector), the defect assessment model is trained. That is, the preset initialization vector (e.g., an all-zero vector) is used as the initial weight vector of the pre-trained decision tree model. Based on multiple training samples corresponding to the first type of process, the weight vector of the pre-trained decision tree model is iterated multiple times to obtain the defect assessment model corresponding to the first type of process.

[0040] For the second type of process, due to the poor uniformity of the training samples, if the training process still starts from a preset initialization vector (e.g., an all-zero vector), it is easy for the training process to get stuck in a local optimum. To avoid the above problem, this embodiment considers that if the correlation strength change sequences of two different processes are similar, the distribution law of the importance of the processing parameters of the two processes is consistent. Therefore, from multiple first-type processes, the first-type process with the highest similarity between its correlation strength change sequence and the correlation strength change sequence of the second-type process can be selected as the reference process. The defect evaluation model corresponding to the reference process is used as the reference evaluation model. The distribution law of "processing parameters with high correlation strength corresponding to high weights" in the reference evaluation model is transferred to the second-type process. The weight vector of the reference evaluation model is used as the basis for setting the initial weight vector of the pre-trained decision tree model corresponding to the second-type process. This allows the pre-trained decision tree model to start iterating from "an initial state that fits its own process law", avoiding the blindness of iterating from zero, and thus effectively avoiding the training process getting stuck in a local optimum.

[0041] It should be noted that the method in this embodiment can also be combined with other methods to further avoid the training process from getting stuck in local optima. For example, L1 or L2 regularization mechanisms can be introduced during the model iterative training phase. By adding a penalty term for the parameter weights in the loss function, the excessive increase of the weight values ​​can be suppressed, and the model can be prevented from overfitting to local sample features.

[0042] As can be seen from the above, this embodiment divides multiple processes into a first type of process and a second type of process based on the sample uniformity of each process. For the second type of process with poor sample uniformity, the reference process with the highest similarity of the correlation strength change sequence is matched, and the weight vector of the reference evaluation model corresponding to the reference process is used to determine the initial weight vector of the pre-trained decision tree model. This can effectively avoid the problem that the model training process of the second type of process is prone to getting trapped in local optima due to uneven sample distribution.

[0043] In one embodiment of this application, the weight vector of the reference evaluation model includes reference weights corresponding to multiple historical processing parameters of the reference process. The reference weights corresponding to each historical processing parameter correspond one-to-one with the correlation strengths corresponding to each historical processing parameter. The correlation strengths corresponding to the multiple historical processing parameters of the reference process are multiple reference correlation strengths corresponding to the reference evaluation model. Specifically, for each second type of process, the initial weight vector of the pre-trained decision tree model is determined based on the weight vector of the reference evaluation model, including: For each historical processing parameter of the second type of process, based on the correlation strength corresponding to the historical processing parameter, select the two reference correlation strengths that are closest to the correlation strength corresponding to the historical processing parameter from multiple reference correlation strengths corresponding to the reference evaluation model; perform data fitting based on the reference weights corresponding to the two reference correlation strengths to obtain the initial weights corresponding to the historical processing parameter. The initial weights corresponding to the various historical processing parameters of the second type of process are used as the initial weight vectors of the pre-trained decision tree model.

[0044] In this embodiment, for each second type of process, after determining the reference process corresponding to the second type of process, the correlation strengths corresponding to multiple historical processing parameters of the reference process can be used as multiple reference correlation strengths corresponding to the reference evaluation model. For each historical processing parameter of the second type of process, the difference between the correlation strength corresponding to the historical processing parameter and each reference correlation strength can be calculated. The two reference correlation strengths with the smallest absolute value of the difference are selected as the two reference correlation strengths closest to the correlation strength corresponding to the historical processing parameter. Based on the reference weights corresponding to the two reference correlation strengths, data fitting can be performed to obtain the initial weight corresponding to the historical processing parameter.

[0045] For example, two reference association strengths and their corresponding two reference weights constitute two reference data points. and ,in, Indicates the reference correlation strength of the first reference data point. This represents the reference weight of the first reference data point. This indicates the reference correlation strength of the second reference data point. This represents the reference weight of the second reference data point. A linear equation can be fitted based on the two reference data points. Based on this, the correlation strength corresponding to historical processing parameters is used as the independent variable. Substituting these values ​​into the linear equation above, we can obtain the initial weights corresponding to the historical processing parameters. value).

[0046] Using the same method, the initial weights corresponding to all historical processing parameters in the second type of process can be obtained. The initial weights corresponding to all historical processing parameters are then summarized and used as the initial weight vector of the pre-trained decision tree model corresponding to the process.

[0047] As can be seen from the above, for each historical processing parameter, this embodiment determines the initial weight corresponding to the historical processing parameter based on the two reference association strengths that are closest to the correlation strength of the historical processing parameter. This can make the setting of the initial weight closely match the actual influence of the processing parameter on the first defect probability, thereby accelerating the model convergence.

[0048] In one embodiment of this application, for each second type of process, from a plurality of first type processes, the first type of process with the highest similarity between its corresponding correlation strength change sequence and the correlation strength change sequence of the second type of process is selected, including: The cumulative distance between the correlation strength change sequence of the second type of process and the correlation strength change sequence of each first type of process is calculated based on the dynamic time warping algorithm. The first type of process with the smallest cumulative distance is selected as the first type of process with the highest similarity.

[0049] In this embodiment, traditional distance algorithms (such as Euclidean distance) require that the two sequences to be matched have the same length and can only be matched one-to-one at fixed positions. In this embodiment, the lengths of the sequences corresponding to the changes in association strength for different processes are not necessarily the same, that is, the lengths of the two sequences to be matched may be different. Traditional distance algorithms cannot be adapted to the distance calculation scenario of this embodiment.

[0050] Dynamic Time Warping (DTW) does not force the elements of two sequences to match in fixed positions. Instead, it uses dynamic programming to find an optimal matching path and ultimately calculates the overall cumulative distance between the two sequences. This flexible matching characteristic of DTW can solve the matching problem of inconsistent sequence lengths of correlation strength between different processes in cable jointing, ensuring that even if the number of processing parameters differs, the overall difference between sequences can be accurately quantified.

[0051] Therefore, this embodiment calculates the distance between the association strength change sequence of the second type of process and the association strength change sequence of each first type of process based on the dynamic time warping algorithm. Specifically, in actual use, the dynamic time warping calculation module in the open-source library can be called. The two sequences to be matched are input into the dynamic time warping calculation module to obtain the cumulative distance between the two sequences to be matched. This cumulative distance is used as the similarity between the two sequences to be matched. The smaller the cumulative distance, the higher the similarity.

[0052] In one embodiment of this application, the step of calculating the sample uniformity of multiple training samples corresponding to each process includes: Obtain the value range of each processing parameter in this process; The value range of each processing parameter is divided into multiple value intervals; For each processing parameter, the number of training samples in each value interval of the processing parameter is counted, and the relative standard deviation of the number of training samples corresponding to each of the multiple value intervals is calculated. The distribution uniformity of the processing parameter is determined based on the relative standard deviation. The distribution uniformity is negatively correlated with the relative standard deviation. Among the distribution uniformities corresponding to multiple processing parameters, the smallest distribution uniformity is taken as the sample uniformity of multiple training samples corresponding to that process.

[0053] In this embodiment, according to the process design requirements, each processing parameter has a corresponding design range, which is used as the value range for each processing parameter. Based on this, for each processing parameter, the value range can be divided into multiple (e.g., 10) non-overlapping value intervals using an equal-interval partitioning method. The number of samples within each value interval is counted, resulting in a data sequence composed of multiple sample numbers. The relative standard deviation of this data sequence is calculated and used as the relative standard deviation for that processing parameter. The relative standard deviation characterizes the dispersion of the number of samples within each value interval relative to the mean. The larger the relative standard deviation, the greater the dispersion of the number of samples within each value interval, and the smaller the sample uniformity. Therefore, the distribution uniformity of each processing parameter can be determined based on the relative standard deviation corresponding to each processing parameter.

[0054] For example, the uniformity of distribution for each processing parameter can be calculated using the following formula: ; in, This represents the uniformity of the distribution of the i-th processing parameter. This represents the relative standard deviation corresponding to the i-th processing parameter.

[0055] For each process, based on the distribution uniformity of each processing parameter of that process, and considering that the overall quality of the training samples depends on the processing parameter with the least uniform distribution, the minimum distribution uniformity is selected as the sample uniformity of the multiple training samples corresponding to that process.

[0056] As can be seen from the above, this embodiment determines the distribution uniformity of each processing parameter based on the relative standard deviation. On this basis, for each process, the smallest distribution uniformity is selected as the sample uniformity of multiple training samples corresponding to that process, so that the quantification result of the sample uniformity is consistent with the actual process.

[0057] Corresponding to the intelligent cable joint detection method in the above embodiment, Figure 2 This is a structural block diagram of an intelligent cable joint detection system provided in one embodiment of this application. For ease of explanation, only the parts relevant to the embodiment of this application are shown. References Figure 2The intelligent cable joint detection system 20 includes: a data acquisition module 21, a first probability calculation module 22, a second probability calculation module 23, a third probability calculation module 24, and a process control module 25.

[0058] Among them, the data acquisition module 21 is used to acquire the processing parameters of the current process and the quality inspection data of the parts in the current process during the cable joint processing. The first probability calculation module 22 is used to determine the first defect probability of the part in the current process based on the processing parameters of the current process and the defect evaluation model corresponding to the current process. The second probability calculation module 23 is used to determine the second defect probability of the part in the current process based on the quality inspection data. The third probability calculation module 24 is used to determine the third defect probability of the part in the current process based on the first defect probability and the second defect probability. The process control module 25 is used to transfer the current process part to the next process when the probability of the third defect is less than or equal to the preset first probability threshold.

[0059] In one embodiment of this application, the cable connector processing includes multiple steps, and the first probability calculation module 22 is specifically used for: Obtain multiple training samples corresponding to each process; for each process, each training sample includes multiple historical processing parameters of that process and the corresponding historical first defect probability; For each process, based on multiple training samples corresponding to that process, the correlation strength between each historical processing parameter of that process and the corresponding first defect probability is calculated; the correlation strengths of each of the multiple historical processing parameters are arranged in descending order to obtain the correlation strength change sequence corresponding to that process. Calculate the sample uniformity of multiple training samples corresponding to each process; Processes with sample uniformity greater than or equal to the uniformity threshold are classified as first-class processes, and processes with sample uniformity less than the uniformity threshold are classified as second-class processes. For each first type of process, a preset initialization vector is used as the initial weight vector of the pre-trained decision tree model. Based on multiple training samples corresponding to the first type of process, the weight vector of the pre-trained decision tree model is iterated multiple times to obtain the defect assessment model corresponding to the first type of process. For each second-type process, from multiple first-type processes, the first-type process with the highest similarity between its corresponding correlation strength change sequence and the correlation strength change sequence of the second-type process is selected as a reference process, and the defect assessment model corresponding to the reference process is used as the reference assessment model. Based on the weight vector of the reference assessment model, the initial weight vector of the pre-trained decision tree model is determined, and the weight vector of the pre-trained decision tree model is iterated multiple times based on multiple training samples corresponding to the second-type process to obtain the defect assessment model corresponding to the second-type process.

[0060] In one embodiment of this application, the weight vector of the reference evaluation model includes reference weights corresponding to multiple historical processing parameters of the reference process, with each reference weight corresponding to a historical processing parameter corresponding to a one-to-one correlation strength, and the correlation strengths corresponding to the multiple historical processing parameters of the reference process being multiple reference correlation strengths corresponding to the reference evaluation model; for each second type of process, the first probability calculation module 22 is further used for: For each historical processing parameter of the second type of process, based on the correlation strength corresponding to the historical processing parameter, select the two reference correlation strengths that are closest to the correlation strength corresponding to the historical processing parameter from multiple reference correlation strengths corresponding to the reference evaluation model; perform data fitting based on the reference weights corresponding to the two reference correlation strengths to obtain the initial weights corresponding to the historical processing parameter. The initial weights corresponding to the various historical processing parameters of the second type of process are used as the initial weight vectors of the pre-trained decision tree model.

[0061] In one embodiment of this application, for each second type of process, the first probability calculation module 22 is further used for: The cumulative distance between the correlation strength change sequence of the second type of process and the correlation strength change sequence of each first type of process is calculated based on the dynamic time warping algorithm. The first type of process with the smallest cumulative distance is selected as the first type of process with the highest similarity.

[0062] In one embodiment of this application, for each process, the first probability calculation module 22 is further used for: Obtain the value range of each processing parameter in this process; The value range of each processing parameter is divided into multiple value intervals; For each processing parameter, the number of training samples in each value interval of the processing parameter is counted, and the relative standard deviation of the number of training samples corresponding to each of the multiple value intervals is calculated. The distribution uniformity of the processing parameter is determined based on the relative standard deviation. The distribution uniformity is negatively correlated with the relative standard deviation. Among the distribution uniformities corresponding to multiple processing parameters, the smallest distribution uniformity is taken as the sample uniformity of multiple training samples corresponding to that process.

[0063] In one embodiment of this application, the second probability calculation module 23 is specifically used for: Based on the quality inspection data, a preset mapping relationship is found to obtain the second defect probability of the part in the current process; wherein, the mapping relationship is used to characterize the correspondence between the quality inspection data and the second defect probability.

[0064] In one embodiment of this application, the third probability calculation module 24 is specifically used for: If the first defect probability is greater than the preset second probability threshold, or the second defect probability is greater than the preset second probability threshold, the maximum probability between the first defect probability and the second defect probability shall be used as the third defect probability of the part in the current process. If the first defect probability is less than or equal to the second defect probability, and both the first and second defect probabilities are less than or equal to the preset second probability threshold, the first and second defect probabilities are weighted and summed to obtain the third defect probability of the part manufactured in the current process.

[0065] See Figure 3 , Figure 3 This is a schematic block diagram of an electronic device provided according to an embodiment of this application. Figure 3 The electronic device 300 in this embodiment may include one or more processors 301, one or more input devices 302, one or more output devices 303, and one or more memories 304. The processors 301, input devices 302, output devices 303, and memories 304 communicate with each other via a communication bus 305. The memories 304 store computer programs, including program instructions. The processors 301 execute the program instructions stored in the memories 304. Specifically, the processors 301 are configured to invoke the program instructions to perform the functions of each module / unit in the above-described device embodiments, for example... Figure 2 The functions of the data acquisition module 21, the first probability calculation module 22, the second probability calculation module 23, the third probability calculation module 24, and the process control module 25 are shown.

[0066] It should be understood that, in the embodiments of this application, the processor 301 may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0067] Input device 302 may include a touchpad, a fingerprint sensor (for collecting the user's fingerprint information and fingerprint orientation information), a microphone, etc., and output device 303 may include a display (LCD, etc.), a speaker, etc.

[0068] The memory 304 may include read-only memory and random access memory, and provides instructions and data to the processor 301. A portion of the memory 304 may also include non-volatile random access memory. For example, the memory 304 may also store preset constants such as a first probability threshold and a second probability threshold.

[0069] In specific implementations, the processor 301, input device 302, and output device 303 described in the embodiments of this application can execute the implementation method described in the intelligent cable joint detection method provided in the embodiments of this application, or they can execute the implementation method of the electronic device described in the embodiments of this application, which will not be repeated here.

[0070] In another embodiment of this application, a computer-readable storage medium is provided. This computer-readable storage medium stores a computer program, which includes program instructions. When executed by a processor, the program instructions implement all or part of the processes in the methods described above. Alternatively, the computer program can instruct related hardware to complete the process. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include any entity or device capable of carrying computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0071] The computer-readable storage medium can be an internal storage unit of the electronic device in any of the foregoing embodiments, such as a hard disk or memory of the electronic device. The computer-readable storage medium can also be an external storage device of the electronic device, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the electronic device. Furthermore, the computer-readable storage medium can include both internal and external storage units of the electronic device. The computer-readable storage medium is used to store computer programs and other programs and data required by the electronic device. The computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

[0072] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this application.

[0073] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the electronic devices and units described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0074] In the several embodiments provided in this application, it should be understood that the disclosed electronic devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces or units, or it may be an electrical, mechanical, or other form of connection.

[0075] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of the embodiments of this application, depending on actual needs.

[0076] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0077] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An intelligent cable joint detection method, characterized in that, include: During the cable connector manufacturing process, obtain the processing parameters of the current process and the quality inspection data of the parts manufactured in the current process; Based on the processing parameters of the current process and the defect assessment model corresponding to the current process, the first defect probability of the part in the current process is determined; The probability of a second defect in the part manufactured in the current process is determined based on the quality inspection data. The third defect probability of the part manufactured in the current process is determined based on the first defect probability and the second defect probability. If the probability of the third defect is less than or equal to the preset first probability threshold, the part in the current process will be transferred to the next process.

2. The intelligent cable joint detection method as described in claim 1, characterized in that, The cable joint manufacturing process involves multiple steps, and the method for determining the defect assessment model for each step includes: Obtain multiple training samples corresponding to each process; for each process, each training sample includes multiple historical processing parameters of that process and the corresponding historical first defect probability; For each process, based on multiple training samples corresponding to that process, the correlation strength between each historical processing parameter of that process and the corresponding first defect probability is calculated; the correlation strengths of each of the multiple historical processing parameters are arranged in descending order to obtain the correlation strength change sequence corresponding to that process. Calculate the sample uniformity of multiple training samples corresponding to each process; The process with a corresponding sample uniformity greater than or equal to the uniformity threshold is classified as the first type of process, and the process with a corresponding sample uniformity less than the uniformity threshold is classified as the second type of process. For each of the first type of process, a preset initialization vector is used as the initial weight vector of the pre-trained decision tree model. Based on multiple training samples corresponding to the first type of process, the weight vector of the pre-trained decision tree model is iterated multiple times to obtain the defect assessment model corresponding to the first type of process. For each of the second type of processes, from multiple first type processes, the first type process with the highest similarity between the corresponding correlation strength change sequence and the correlation strength change sequence of the second type process is selected as a reference process, and the defect assessment model corresponding to the reference process is used as a reference assessment model; the initial weight vector of the pre-trained decision tree model is determined based on the weight vector of the reference assessment model, and the weight vector of the pre-trained decision tree model is iterated multiple times based on multiple training samples corresponding to the second type of process to obtain the defect assessment model corresponding to the second type of process.

3. The intelligent cable joint detection method as described in claim 2, characterized in that, The weight vector of the reference evaluation model includes the reference weights corresponding to each of the multiple historical processing parameters of the reference process. The reference weights corresponding to each historical processing parameter correspond one-to-one with the correlation strengths corresponding to each historical processing parameter. The correlation strengths corresponding to each of the multiple historical processing parameters of the reference process are the multiple reference correlation strengths corresponding to the reference evaluation model. Specifically, for each of the second type of processes, determining the initial weight vector of the pre-trained decision tree model based on the weight vector of the reference evaluation model includes: For each historical processing parameter of the second type of process, based on the correlation strength corresponding to the historical processing parameter, select the two reference correlation strengths that are closest to the correlation strength corresponding to the historical processing parameter from the multiple reference correlation strengths corresponding to the reference evaluation model; perform data fitting based on the reference weights corresponding to the two reference correlation strengths to obtain the initial weights corresponding to the historical processing parameter; The initial weights corresponding to the various historical processing parameters of the second type of process are used as the initial weight vectors of the pre-trained decision tree model.

4. The intelligent cable joint detection method as described in claim 2, characterized in that, For each of the second type of processes, selecting the first type of process from a plurality of first type processes whose corresponding correlation strength change sequence has the highest similarity to the correlation strength change sequence of that second type of process includes: The cumulative distance between the correlation strength change sequence of the second type of process and the correlation strength change sequence of each first type of process is calculated based on the dynamic time warping algorithm. The first type of process with the smallest cumulative distance is selected as the first type of process with the highest similarity.

5. The intelligent cable joint detection method as described in claim 2, characterized in that, For each process, the step of calculating the sample uniformity of multiple training samples corresponding to that process includes: Obtain the value range of each processing parameter in this process; The value range of each processing parameter is divided into multiple value intervals; For each processing parameter, the number of training samples in each value interval of the processing parameter is counted, and the relative standard deviation of the number of training samples corresponding to each of the multiple value intervals is calculated. The distribution uniformity of the processing parameter is determined based on the relative standard deviation; wherein, the distribution uniformity is negatively correlated with the relative standard deviation. Among the distribution uniformities corresponding to multiple processing parameters, the smallest distribution uniformity is taken as the sample uniformity of multiple training samples corresponding to that process.

6. The intelligent cable joint detection method as described in claim 1, characterized in that, Determining the second defect probability of the part manufactured in the current process based on the quality inspection data includes: Based on the quality inspection data, a preset mapping relationship is found to obtain the second defect probability of the part manufactured in the current process; wherein, the mapping relationship is used to characterize the correspondence between the quality inspection data and the second defect probability.

7. The intelligent cable joint detection method as described in claim 1, characterized in that, Determining the third defect probability of the part manufactured in the current process based on the first defect probability and the second defect probability includes: If the first defect probability is greater than a preset second probability threshold, or the second defect probability is greater than a preset second probability threshold, the maximum probability between the first defect probability and the second defect probability shall be taken as the third defect probability of the part manufactured in the current process. If the first defect probability is less than and the second defect probability is less than or equal to a preset second probability threshold, the first defect probability and the second defect probability are weighted and summed to obtain the third defect probability of the part manufactured in the current process.

8. An intelligent cable joint detection system, characterized in that, include: The data acquisition module is used to acquire the processing parameters of the current process and the quality inspection data of the parts produced in the current process during the cable joint processing. The first probability calculation module is used to determine the first defect probability of the part in the current process based on the processing parameters of the current process and the defect assessment model corresponding to the current process. The second probability calculation module is used to determine the second defect probability of the part manufactured in the current process based on the quality inspection data. The third probability calculation module is used to determine the third defect probability of the part in the current process based on the first defect probability and the second defect probability. The process control module is used to transfer the current process part to the next process when the probability of the third defect is less than or equal to a preset first probability threshold.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 7.

Images