A method and system for tracing defects in the manufacturing process of watch buttons

Abstract

This invention discloses a method and system for tracing defects in the manufacturing process of watch buttons. The method includes: acquiring process values ​​and quality values; preprocessing the quality values ​​to obtain a performance dataset; evaluating the performance dataset based on multiple regression analysis to obtain evaluation results; constructing nodes and performing traceability analysis based on process values ​​and quality values ​​to obtain a node graph and mapping table; constructing a model based on the node graph, mapping table, and evaluation results to obtain a prediction model; predicting defects based on the prediction model to obtain prediction results; classifying the defect types in the prediction results based on a clustering algorithm to obtain a standard defect library; identifying features based on the standard defect library to obtain traceability identification codes; and associating real-time process quality data with traceability identification codes to obtain defect traceability information. This method can improve the efficiency of defect handling by constructing a node graph and establishing a correspondence between processes and quality indicators.

Description

Technical Field

[0001] The present invention relates to the technical fields of intelligent manufacturing and quality control, and particularly to a method and system for defect traceability in the processing of watch buttons. Background Art

[0002] Various defects are likely to occur during the processing of watch buttons, which will seriously affect the user experience of the buttons and the product quality. At present, there are many technical problems in defect traceability. First, there are no scientific quantitative standards and testing methods for the performance indicators of buttons such as response speed and durability, resulting in the inability to accurately evaluate the button quality. Second, the processing of buttons involves multiple processes, and there are many nodes where defects occur. Existing quality management means are difficult to accurately locate the specific links and reasons causing the defects. Third, the types of defects are diverse and the severity levels are different. There is an urgent need to establish a defect classification system and design a reasonable sorting algorithm to clarify the priority order of defect handling. Finally, a large amount of defect data records are scattered, lacking a unified management platform and traceability mechanism, which affects the rapid response and closed-loop processing of defect problems. There is an urgent need to build a complete management system for defect information collection, storage, analysis, and traceability, to achieve refined control of the entire life cycle of defects, and to accurately retrieve defect information and quickly locate problems with the help of traceability codes, so as to continuously improve the processing technology of watch buttons and comprehensively ensure product quality.

[0003] In an existing technology, the specific implementation includes: sampling and inspecting the buttons through traditional quality inspection means to detect their performance indicators such as response speed and durability. These inspection means will include manual pressing tests, simple mechanical test equipment, etc., and the inspection results are usually recorded in paper or spreadsheets for subsequent analysis and processing.

[0004] However, in the existing technology, the quality management means are difficult to accurately locate the specific links and reasons causing the defects, the types of defects are diverse and the severity levels are different, lacking a unified defect classification system and priority sorting algorithm, resulting in a low response speed and processing efficiency for defect problems. Summary of the Invention

[0005] The present invention provides a method and system for defect traceability in the processing of watch buttons to solve the problems in the existing technology that the quality management means are difficult to accurately locate the specific links and reasons causing the defects, the types of defects are diverse and the severity levels are different, lacking a unified defect classification system and priority sorting algorithm, resulting in a low response speed and processing efficiency for defect problems.

[0006] In the first aspect, to solve the above technical problems, the present invention provides a method for defect traceability in the processing of watch buttons, including:

[0007] Obtain process values ​​and quality values; wherein, the quality values ​​include force values, displacement values, and resistance values;

[0008] The force value, displacement, and resistance value are preprocessed to obtain a performance dataset;

[0009] The performance dataset was evaluated using a multivariate regression analysis method, and the evaluation results were obtained.

[0010] Based on the process value and the quality value, node construction and traceability analysis are performed to obtain a node diagram and a mapping relationship table;

[0011] A prediction model is obtained by constructing a model based on the node graph, the mapping table, and the evaluation results.

[0012] Defect prediction is performed based on the prediction model to obtain the prediction results;

[0013] The defect types in the prediction results are classified based on a clustering algorithm to obtain a standard defect library; feature identification is performed based on the standard defect library to obtain a traceability identification code;

[0014] By associating and mapping real-time process quality data with the traceability identifier, defect traceability information can be obtained.

[0015] In one possible implementation of the first aspect, the preprocessing of the force value, the displacement, and the resistance value to obtain a performance dataset includes:

[0016] Synchronous calibration is performed based on the force value, the displacement, and the resistance value to obtain a calibration dataset;

[0017] The calibration dataset is time-series aligned to obtain a sequence dataset;

[0018] Feature extraction is performed on the sequence dataset to obtain a multi-dimensional feature dataset;

[0019] The multi-dimensional feature data is classified using a clustering algorithm to obtain a performance dataset.

[0020] In one possible implementation of the first aspect, the performance evaluation of the performance dataset based on a multivariate regression analysis method to obtain the evaluation result includes:

[0021] A model was constructed for the performance dataset based on the multiple regression analysis method to obtain an initial correlation model;

[0022] Based on the initial association model, dynamic factors and interaction factors are incorporated to expand the dimensions, resulting in an extended association model.

[0023] Based on the extended correlation model, feature analysis is performed on the real-time process quality data to obtain feature values;

[0024] Weight coefficients are obtained by calculating the weights based on the eigenvalues.

[0025] The performance dataset is optimized based on the weighting coefficients to obtain an optimized set of process parameters;

[0026] Performance evaluation is performed based on the optimized process parameter set to obtain the evaluation results.

[0027] In one possible implementation of the first aspect, the step of calculating weights based on the feature values ​​to obtain weight coefficients includes:

[0028] The weighting coefficients are calculated using the following formula:

[0029]

[0030] in, Indicates the first The weighting coefficients of each indicator Indicates the first The characteristic values ​​of each indicator This represents the total number of indicators.

[0031] In one possible implementation of the first aspect, the step of constructing nodes and performing traceability analysis based on the process values ​​and quality values ​​to obtain a node graph and a mapping table includes:

[0032] Data cleaning and format conversion are performed based on the process values ​​and quality values ​​to obtain standardized data;

[0033] Based on a pre-defined node relationship model, the standardized data is stored in a distributed repository to obtain a node graph;

[0034] An anomaly detection analyzer is used to perform anomaly analysis on the node graph. If there are abnormal data points in the node graph, quality tracing and root cause analysis are performed on the abnormal data points to obtain a mapping relationship table.

[0035] In one possible implementation of the first aspect, the step of obtaining a traceability identifier code by feature identification based on the standard defect library includes:

[0036] The weights of the standard defect library are calculated based on the analytic hierarchy process to obtain the index weight values.

[0037] Based on the weight values ​​of the aforementioned indicators, priority scores are calculated for the defects to obtain a score value.

[0038] The scores are sorted from highest to lowest to obtain the defect processing sequence;

[0039] The defect processing sequence is encoded based on pre-stored defect encoding rules to obtain a traceability identifier code.

[0040] In one possible implementation of the first aspect, the step of associating and mapping real-time process quality data with the traceability identifier to obtain defect traceability information includes:

[0041] The judgment is made based on the real-time process quality data and the preset threshold range; if the real-time process quality data exceeds the preset threshold range, it is determined that there is a defect source.

[0042] The defect source is retrieved based on the traceability identifier code to obtain the defect information storage location and timestamp;

[0043] The defect information storage location and the timestamp are output to obtain defect tracing information.

[0044] Secondly, the present invention provides a system for tracing the source of defects in the manufacturing process of watch buttons, comprising:

[0045] The data acquisition module is used to acquire process values ​​and quality values; wherein, the quality values ​​include force values, displacement values, and resistance values.

[0046] The data cleaning module is used to preprocess the force value, the displacement, and the resistance value to obtain a performance dataset;

[0047] The performance evaluation module is used to perform performance evaluation on the performance dataset based on the multivariate regression analysis method and obtain the evaluation results.

[0048] The node construction and analysis module is used to construct nodes and perform traceability analysis based on the process values ​​and quality values, and obtain a node diagram and a mapping relationship table.

[0049] The model building module is used to build a predictive model based on the node graph, the mapping table, and the evaluation results.

[0050] The defect prediction module is used to predict defects based on the prediction model and obtain prediction results.

[0051] The defect classification module is used to classify the defect types in the prediction results based on a clustering algorithm to obtain a standard defect library;

[0052] The defect coding module is used to perform feature identification based on the standard defect library to obtain a traceability identification code;

[0053] The output module is used to associate and map real-time process quality data with the traceability identifier to obtain defect traceability information.

[0054] In one possible implementation of the second aspect, the preprocessing of the force value, the displacement, and the resistance value to obtain a performance dataset includes:

[0055] Synchronous calibration is performed based on the force value, the displacement, and the resistance value to obtain a calibration dataset;

[0056] The calibration dataset is time-series aligned to obtain a sequence dataset;

[0057] Feature extraction is performed on the sequence dataset to obtain a multi-dimensional feature dataset;

[0058] The multi-dimensional feature data is classified using a clustering algorithm to obtain a performance dataset.

[0059] In one possible implementation of the second aspect, the performance evaluation of the performance dataset based on the multivariate regression analysis method to obtain the evaluation results includes:

[0060] A model was constructed for the performance dataset based on the multiple regression analysis method to obtain an initial correlation model;

[0061] Based on the initial association model, dynamic factors and interaction factors are incorporated to expand the dimensions, resulting in an extended association model.

[0062] Based on the extended correlation model, feature analysis is performed on the real-time process quality data to obtain feature values;

[0063] Weight coefficients are obtained by calculating the weights based on the eigenvalues.

[0064] The performance dataset is optimized based on the weighting coefficients to obtain an optimized set of process parameters;

[0065] Performance evaluation is performed based on the optimized process parameter set to obtain the evaluation results.

[0066] In one possible implementation of the second aspect, the step of calculating weights based on the feature values ​​to obtain weight coefficients includes:

[0067] The weighting coefficients are calculated using the following formula:

[0068]

[0069] in, Indicates the first The weighting coefficients of each indicator Indicates the first The characteristic values ​​of each indicator This represents the total number of indicators.

[0070] In one possible implementation of the second aspect, the step of constructing nodes and performing traceability analysis based on the process values ​​and quality values ​​to obtain a node graph and a mapping relationship table includes:

[0071] Data cleaning and format conversion are performed based on the process values ​​and quality values ​​to obtain standardized data;

[0072] Based on a pre-defined node relationship model, the standardized data is stored in a distributed repository to obtain a node graph;

[0073] An anomaly detection analyzer is used to perform anomaly analysis on the node graph. If there are abnormal data points in the node graph, quality tracing and root cause analysis are performed on the abnormal data points to obtain a mapping relationship table.

[0074] In one possible implementation of the second aspect, the step of obtaining a traceability identifier code by feature identification based on the standard defect library includes:

[0075] The weights of the standard defect library are calculated based on the analytic hierarchy process to obtain the index weight values.

[0076] Based on the weight values ​​of the aforementioned indicators, priority scores are calculated for the defects to obtain a score value.

[0077] The scores are sorted from highest to lowest to obtain the defect processing sequence;

[0078] The defect processing sequence is encoded based on pre-stored defect encoding rules to obtain a traceability identifier code.

[0079] In one possible implementation of the second aspect, the step of associating and mapping real-time process quality data with the traceability identifier to obtain defect traceability information includes:

[0080] The judgment is made based on the real-time process quality data and the preset threshold range; if the real-time process quality data exceeds the preset threshold range, it is determined that there is a defect source.

[0081] The defect source is retrieved based on the traceability identifier code to obtain the defect information storage location and timestamp;

[0082] The defect information storage location and the timestamp are output to obtain defect tracing information.

[0083] Thirdly, the present invention also provides an electronic device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor executes the computer program to implement the method for tracing defects in the manufacturing process of watch buttons as described in any of the above.

[0084] Fourthly, the present invention also provides a computer-readable storage medium comprising a stored computer program, wherein, when the computer program is executed, it controls the device containing the computer-readable storage medium to perform the method for tracing the source of defects in the manufacturing process of watch buttons as described in any one of the above.

[0085] Compared with the prior art, the present invention has the following beneficial effects:

[0086] This invention discloses a method for tracing defects in the manufacturing process of watch buttons, including obtaining process values ​​and quality values; wherein, the quality values ​​include force values, displacement values, and resistance values; preprocessing the force values, displacement values, and resistance values ​​to obtain a performance dataset; performing performance evaluation on the performance dataset based on a multivariate regression analysis method to obtain evaluation results; constructing nodes and performing traceability analysis based on the process values ​​and quality values ​​to obtain a node graph and a mapping relationship table; constructing a model based on the node graph, the mapping relationship table, and the evaluation results to obtain a prediction model; predicting defects based on the prediction model to obtain prediction results; classifying the defect types in the prediction results based on a clustering algorithm to obtain a standard defect library; identifying features based on the standard defect library to obtain a traceability identifier code; and associating real-time process quality data with the traceability identifier code to obtain defect traceability information. This invention establishes a performance index quantification model based on physical characteristics, collects key response time and pressure change data, and constructs a multi-dimensional performance index dataset. It employs multivariate regression analysis to establish a correlation model between key performance and process parameters, identifying key influencing factors. This invention deploys online testing equipment at each processing node to collect process parameters and quality data in real time, constructing a node diagram and the correspondence between processes and quality indicators. Based on the testing results, defects are classified and a standard defect library is established. A defect priority evaluation model is constructed using the analytic hierarchy process (AHP) to generate a list of disposal orders. This invention also designs a defect tracking mechanism based on traceability codes, realizing a full lifecycle information chain for defects, and implements a quality data analysis and early warning system to provide decision support for quality improvement. This invention can effectively improve key performance, reduce defect rates, and achieve full-process control of product quality. Attached Figure Description

[0087] Figure 1 This is a schematic flowchart of a method for tracing the source of defects in the manufacturing process of watch buttons provided in the first embodiment of the present invention;

[0088] Figure 2 This is a schematic diagram of the system structure for tracing the source of defects in the watch button manufacturing process provided in the second embodiment of the present invention. Detailed Implementation

[0089] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0090] Reference Figure 1 The first embodiment of the present invention provides a method for tracing the source of defects in the manufacturing process of watch buttons, including the following steps:

[0091] S1, obtain process value and quality value; wherein, the quality value includes force value, displacement and resistance value;

[0092] S2, preprocess the force value, the displacement, and the resistance value to obtain a performance dataset;

[0093] S3, The performance dataset is evaluated based on the multivariate regression analysis method to obtain the evaluation results;

[0094] S4. Based on the process value and the quality value, perform node construction and traceability analysis to obtain a node diagram and a mapping relationship table;

[0095] S5. Based on the node graph, the mapping relationship table, and the evaluation results, a model is constructed to obtain a prediction model;

[0096] S6, perform defect prediction based on the prediction model to obtain the prediction result;

[0097] S7. Based on the clustering algorithm, classify the defect types in the prediction results to obtain a standard defect library;

[0098] S8. Based on the standard defect database, feature identification is performed to obtain the traceability identification code;

[0099] S9. Associate and map the real-time process quality data with the traceability identifier to obtain defect traceability information.

[0100] In step S1, process values ​​and quality values ​​are obtained; wherein, the quality values ​​include force value, displacement amount and resistance value.

[0101] For example, raw data is obtained by acquiring the response time and pressure change data of a button through sensors. A multi-channel data acquisition system is fundamental to button performance evaluation. Such a system typically includes a force sensor, a displacement sensor, and a resistance measurement device, used to acquire the force, displacement, and resistance values ​​during button pressing, respectively. For instance, a piezoelectric force sensor can be used to measure the pressing force, a photoelectric encoder to measure displacement, and a high-precision resistance measurement circuit to measure the change in button resistance. Data synchronization calibration is a crucial step to ensure the accuracy of the acquired data. Different sensors may have slight differences in response time, requiring calibration through methods such as timestamp alignment or interpolation. For example, assuming the force sensor has a sampling rate of 1000Hz and the displacement sensor is 500Hz, linear interpolation can be used to match the displacement data to the time point of the force data. Time series alignment unifies data from different dimensions onto the same time axis. This can be achieved through resampling or interpolation. For example, unifying all data to a 10ms time interval facilitates subsequent analysis and feature extraction. Multi-dimensional feature extraction extracts meaningful features from the raw data. For button performance, the following features can be extracted: maximum force, force rise time, force fall time, displacement curve slope, and resistance change rate. These features comprehensively reflect the tactile and responsive characteristics of the buttons. Constructing a performance evaluation system involves transforming these extracted features into quantifiable performance indicators.

[0102] S2, the force value, the displacement and the resistance value are preprocessed to obtain the performance dataset.

[0103] In step S2, the force value, the displacement, and the resistance value are preprocessed to obtain a performance dataset.

[0104] In step S2 above, the preprocessing of the force value, the displacement, and the resistance value to obtain the performance dataset further includes the following steps:

[0105] S21, perform synchronization calibration based on the force value, the displacement, and the resistance value to obtain a calibration dataset;

[0106] S22, perform time series alignment on the calibration dataset to obtain a sequence dataset;

[0107] S23, Perform feature extraction based on the sequence dataset to obtain a multi-dimensional feature dataset;

[0108] S24. Classify the multi-dimensional feature data based on the clustering algorithm to obtain the performance dataset.

[0109] It should be noted that the specific implementation process in steps S21 to S24 above includes: acquiring the force, displacement, and resistance values ​​during the button pressing process using a multi-channel data acquisition system; performing synchronization calibration on the acquired force, displacement, and resistance values; aligning the calibrated force, displacement, and resistance values ​​over time; extracting multi-dimensional features from the time-series aligned data; constructing a performance index evaluation system based on the extracted multi-dimensional features; classifying the data in the performance index evaluation system using a clustering algorithm; and determining that the multi-dimensional performance index dataset has been successfully constructed if the classification results meet a preset threshold.

[0110] For example, a multi-channel data acquisition system is used to acquire the force, displacement, and resistance values ​​during button pressing. Since different sensors have different response times, synchronization calibration is required to ensure the accuracy of the acquired data. For instance, the sampling rates of piezoelectric force sensors, photoelectric encoders, and high-precision resistance measurement circuits may differ. In this case, timestamp alignment or interpolation methods can be used to calibrate the data acquired by different sensors, thus obtaining a calibration dataset. Next, in step S22, the calibrated dataset is time-series aligned. To facilitate subsequent analysis and feature extraction, data from different dimensions needs to be unified onto the same time axis. For example, resampling or interpolation can be used to unify all data to a 10ms time interval, thus obtaining a sequence dataset. Then, in step S23, feature extraction is performed based on the sequence dataset. From the button pressing data, multi-dimensional features that comprehensively reflect the button's tactile and responsive characteristics are extracted. Features such as maximum force, force rise time, force fall time, displacement curve slope, and resistance change rate constitute a multi-dimensional feature dataset.

[0111] Finally, step S24 involves classifying the multi-dimensional feature data using a clustering algorithm. Algorithms such as K-means or hierarchical clustering can be used. Assuming 10 performance metrics are extracted, cluster analysis can categorize button performance into different types such as "light touch," "firm touch," and "balanced," thus obtaining the performance dataset.

[0112] In step S3, the performance dataset is evaluated based on the multivariate regression analysis method to obtain the evaluation results.

[0113] In step S3 above, the performance evaluation of the performance dataset based on the multivariate regression analysis method to obtain the evaluation results further includes the following steps:

[0114] S31, The performance dataset is modeled based on the multivariate regression analysis method to obtain an initial correlation model;

[0115] S32, Based on the initial association model, dynamic factors and interaction factors are incorporated to expand the dimensions, resulting in an extended association model;

[0116] S33, Perform feature analysis on the real-time process quality data according to the extended correlation model to obtain feature values;

[0117] S34, Calculate the weights based on the eigenvalues ​​to obtain the weight coefficients;

[0118] It should be noted that the weighting coefficients are calculated using the following formula:

[0119]

[0120] in, Indicates the first The weighting coefficients of each indicator Indicates the first The characteristic values ​​of each indicator This represents the total number of indicators.

[0121] S35, Optimize the performance dataset according to the weighting coefficients to obtain an optimized process parameter set;

[0122] S36, Perform performance evaluation based on the optimized process parameter set to obtain the evaluation results.

[0123] In one specific embodiment, steps S31 to S36 above are implemented as follows: Using a multiple regression analysis method, data is obtained from button performance indicators and processing parameters to establish an initial correlation model; dynamic factors and interactive factors are incorporated into the initial correlation model to expand its dimensions, resulting in an extended correlation model; through the extended correlation model, the relationship between performance indicators and processing parameters is analyzed to identify key influencing factors; based on the key influencing factors, the weight coefficients of each factor are calculated to obtain the weight allocation results. If the weight coefficient is greater than a preset threshold, the factor is marked as a significant influencing factor; based on the significant influencing factors, the processing parameters are optimized to obtain an optimized set of processing parameters. Using the optimized set of processing parameters, the button performance indicators are re-evaluated to obtain the final performance evaluation results.

[0124] In this embodiment, the dimensional expansion specifically refers to incorporating dynamic and interactive factors to broaden the model's dimensions in order to make it more realistic and improve its accuracy. Dynamic factors include the impact of ambient temperature changes on material properties. Even with the same materials and processing techniques, button performance may differ under varying ambient temperatures. For example, in high-temperature environments, material flexibility may increase, affecting button response speed and force. Interactive factors consider the combined effects of various process parameters, such as the interaction between material hardness and processing pressure. When material hardness is high, the impact of processing pressure on button performance may differ from that when material hardness is low, indicating a complex interaction between the two. Incorporating these dynamic and interactive factors into the initial correlation model yields the expanded correlation model.

[0125] For example, the deployment of distributed data acquisition systems can cover key inspection points on the production line, such as raw material feeding, processing, and finished product inspection. Taking automobile manufacturing as an example, in the body welding process, sensors can be installed on the robotic welding torch, welding power source, and workpiece fixture to collect process values ​​such as welding current, voltage, and speed in real time, while weld quality data is collected at the weld inspection station. This data is streamed in real time via industrial Ethernet or 5G networks to ensure timeliness and integrity. Data cleaning and format conversion are crucial steps in ensuring data quality. In the automobile manufacturing scenario, data problems encountered include missing values, outliers, and inconsistent formats. For example, welding current data may show zero values ​​or outliers outside the normal range due to sensor malfunctions, and these data need to be processed through interpolation or rejection methods. Simultaneously, different devices output data in different formats, which need to be uniformly converted to a standard format, such as JSON or CSV, for subsequent analysis. The construction of a node relationship model is fundamental to understanding the process flow. In automobile manufacturing, each processing step can be considered a node, with material flow and information transmission between steps serving as connections between nodes. For example, the vehicle painting process after welding can be structured as a chain of nodes: "welding-grinding-pretreatment-primer-topcoat-drying," with each node containing corresponding process parameters and quality indicators. The mapping table between processes and quality is the core of quality traceability. Taking engine manufacturing as an example, a mapping relationship can be established between cylinder block machining process parameters (such as cutting speed and feed rate) and cylinder block quality indicators (such as surface roughness and dimensional accuracy). This mapping relationship helps to quickly pinpoint the causes of quality problems and improve production efficiency. The application of anomaly detection analyzers can promptly identify and resolve production issues. In electronic product manufacturing, if PCB board soldering quality abnormalities are found, retrospective analysis can reveal whether the cause is insufficient solder paste printing thickness or an improper reflow soldering temperature profile. This analysis can guide process improvements and prevent similar problems from recurring.

[0126] In step S4, node construction and traceability analysis are performed based on the process value and the quality value to obtain a node diagram and a mapping relationship table.

[0127] In step S4 above, the step of constructing nodes and performing traceability analysis based on the process value and the quality value to obtain a node diagram and a mapping relationship table further includes the following steps:

[0128] S41, perform data cleaning and format conversion based on the process value and quality value to obtain standardized data;

[0129] S42, Based on the preset node relationship model, the standardized data is stored in a distributed repository to obtain a node graph;

[0130] S43, perform anomaly analysis on the node graph based on the anomaly detection analyzer. If there are abnormal data points in the node graph, perform quality tracing and root cause analysis on the abnormal data points to obtain a mapping relationship table.

[0131] In one specific embodiment, steps S41 to S43 above are implemented as follows: A distributed data collector acquires process values ​​and quality values ​​from each detection point, and transmits them to a processor in real-time for data cleaning and format conversion. Based on a preset node relationship model, the cleaned process values ​​and quality values ​​are stored in a distributed repository to construct a complete node graph. Process parameter sets and quality indicator sets are extracted from the repository, and a mapping relationship table between processes and quality is established based on preset mapping rules. If abnormal data points exist in the node graph, an anomaly detection analyzer is activated to perform quality tracing and root cause analysis on the abnormal data. For example, in a process quality monitoring system, when an abnormal data point appears in the node graph, the anomaly detection analyzer is immediately activated to perform quality tracing and root cause analysis. The system first performs a preliminary classification of the abnormal data according to preset rules, such as determining whether it belongs to abnormal process parameters or abnormal product quality indicators. If it is an abnormal process parameter, the system traces back along the production process according to the previously constructed mapping relationship table between processes and quality, combined with the process data stored in the node graph, to examine the changes in process parameters before the abnormal data point. For example, acquiring defect data through a quality inspection system is a crucial aspect of product quality management. Taking the automotive manufacturing industry as an example, in the body welding process, laser scanners and image recognition systems can capture weld defect information. This raw data typically contains a large amount of noise and redundant information, requiring preprocessing. Preprocessing involves steps such as data cleaning, noise reduction, and standardization to ensure the accuracy of subsequent analysis. Extracting defect features is fundamental to identifying and classifying defects. Features in welding defects include weld size, shape, and depth. Edge detection algorithms can extract weld contours and calculate area and perimeter; depth image analysis can measure weld depth. These features constitute a multidimensional vector describing the defect. Clustering algorithms can group defects based on the similarity of their features. The commonly used K-means algorithm can classify weld defects into several main types, such as porosity, cracks, and incomplete penetration. By analyzing the clustering results, the distribution of various defects can be obtained, such as porosity accounting for 30%, cracks for 15%, and incomplete penetration for 20%. The determination of defect severity is usually based on preset standards. For example, for weld cracks, a length of less than 1 mm is considered minor, 1-3 mm is moderate, and more than 3 mm is severe. This classification helps optimize resource allocation and prioritize the most serious problems.

[0132] In step S5, a model is constructed based on the node graph, the mapping relationship table, and the evaluation results to obtain a prediction model.

[0133] In one specific implementation, a predictive model for process parameters and quality indicators is established using regression analysis based on a mapping table and node diagram. Based on the predictive model and real-time collected process values, quality indicators are predicted online, generating quality early warning reports. The predicted results are compared with measured quality values, and the accuracy of the predictive model is continuously improved through iterative optimization.

[0134] For example, the establishment and application of predictive models are an effective means of achieving proactive quality control. In the steel smelting process, regression models can be established based on historical data to correlate process parameters such as furnace temperature and charge quantity with quality indicators such as steel strength and purity.

[0135] In step S6, defect prediction is performed based on the prediction model to obtain the prediction result.

[0136] For example, by inputting current process parameters in real time, the model can predict the final product quality. If the prediction results are unsatisfactory, the process parameters can be adjusted promptly to avoid the generation of defective products. Continuous optimization of the prediction model is key to maintaining its effectiveness. By comparing the predicted results with the actual quality inspection results, the model's prediction error can be calculated. If the error exceeds a preset threshold, the model can be retrained using new data to continuously improve its accuracy. This dynamic optimization mechanism allows the model to adapt to changes in production conditions, such as equipment wear and tear, and variations in raw material batches.

[0137] In step S7, the defect types in the prediction results are classified based on a clustering algorithm to obtain a standard defect library.

[0138] In one feasible approach, classifying the defect types in the prediction results using a clustering algorithm to obtain a standard defect library specifically includes: classifying the extracted defect features using a clustering algorithm to obtain a defect type distribution; determining the severity level of each type of defect based on the defect type distribution and a preset severity standard; matching preset processing schemes to defect types with different severity levels; integrating defect features, defect types, severity, and processing schemes into structured data and storing it in the standard defect library; optimizing the parameter settings of the clustering algorithm using data from the standard defect library; and updating the defect classification model based on the optimized clustering algorithm to improve classification accuracy.

[0139] In step S8, feature identification is performed based on the standard defect library to obtain the traceability identification code.

[0140] In step S8 above, the step of obtaining the traceability identifier code by performing feature identification based on the standard defect library further includes the following steps:

[0141] S81, The weights of the standard defect library are calculated based on the analytic hierarchy process to obtain the index weight values;

[0142] S82, calculate the priority score of the defect based on the weight value of the indicator, and obtain the score value;

[0143] S83, Sort the scores from high to low to obtain the defect processing sequence;

[0144] S84, the defect processing sequence is encoded based on the pre-stored defect encoding rules to obtain the traceability identifier code.

[0145] It should be noted that the specific implementation of steps S81 to S84 above includes: Based on the raw defect data obtained from the standard defect database, including defect frequency, impact rating, and repair cost estimation, a third-order weight matrix is ​​constructed using the analytic hierarchy process (AHP) to calculate the weight values ​​of each indicator. The priority score for each defect is calculated using the weight values, and defects are sorted from highest to lowest score to generate a defect processing sequence. For quality data from different processes, a unified defect coding rule is established, mapping defect features to unique identifiers. Quality data is categorized and stored according to the defect codes, establishing a structured defect information database and setting up an effective indexing mechanism. For defect information retrieval needs, an inverted index technique is used to construct a mapping relationship between defect features and storage locations. During the retrieval process, if defect features are input, the inverted index quickly locates the defect information storage location and queries relevant data. Newly generated defect data is coded according to the defect coding rules and updated to the defect information database to maintain data consistency. For example, deploying data acquisition equipment on the production line is crucial for achieving quality control. Taking automobile manufacturing as an example, sensors and scanning equipment are installed at key process points such as body welding, painting, and final assembly to collect quality data such as vehicle identification number (VIN), number of weld points, and coating thickness in real time. This data is linked to the process points to form a product quality file. When judging defect sources by preset thresholds, a weld strength below 800N or a coating thickness deviation exceeding ±0.1mm can be defined as a defect. Once an anomaly is detected, the system immediately records the defect information and timestamp, providing a basis for subsequent analysis. Defect propagation chain analysis is crucial for improving production efficiency. Taking body welding defects as an example, if they are not detected and corrected in time, they can lead to a chain reaction, such as uneven paint surfaces in subsequent painting processes and abnormal noises from doors in final assembly processes. By establishing a mapping relationship between defect propagation chains and remediation points, timely intervention can be made before problems escalate. The application of association rule algorithms helps optimize defect remediation strategies. For example, analysis shows a strong correlation between welding current and weld strength; adjusting welding parameters can effectively reduce the defect rate. This method not only improves the targeting of remediation but also enables defect traceability and prevention. Constructing a time-series data model of the entire defect lifecycle clearly demonstrates the complete process from defect generation to resolution. For example, recording the discovery time, rework time, and re-inspection time of welding defects ensures the integrity of the information chain, aiding in the analysis of defect handling efficiency and identification. Defect classification and statistics are crucial for optimizing production processes. Defects can be categorized according to type (e.g., welding, painting, assembly) and severity (minor, moderate, severe), generating a priority list. This helps in the rational allocation of resources, prioritizing the handling of defects with greater impact. The application of multi-source data fusion technology further enhances the completeness of the defect lifecycle information chain.By integrating process point data (such as welding station number), quality data (weld strength), defect source information (poor welding), and transfer chain data (affecting subsequent spraying quality), a comprehensive quality control system is formed.

[0146] In step S9, the real-time process quality data is associated and mapped with the traceability identifier to obtain defect traceability information.

[0147] In step S9 above, the step of associating and mapping real-time process quality data with the traceability identifier to obtain defect traceability information further includes the following steps:

[0148] S91, a judgment is made based on the real-time process quality data and the preset threshold range; if the real-time process quality data exceeds the preset threshold range, it is determined that there is a defect source;

[0149] S92, the defect source is retrieved according to the traceability identifier code to obtain the defect information storage location and timestamp;

[0150] S93, output the storage location of the defect information and the timestamp to obtain defect tracing information.

[0151] For example, in automobile manufacturing, sensors and scanning equipment are installed at key process points such as body welding, painting, and final assembly to collect quality data such as vehicle identification number (VIN), number of weld points, and coating thickness in real time. This data is linked to the process points to form a product quality file. When judging the source of defects using preset thresholds, a weld strength below 800N or a coating thickness deviation exceeding ±0.1mm can be defined as a defect. Once an anomaly is detected, the system immediately records the defect information and timestamp, providing a basis for subsequent analysis. Defect propagation chain analysis is crucial for improving production efficiency. Taking body welding defects as an example, if they are not detected and corrected in time, they can lead to a chain reaction, such as uneven paint surfaces in subsequent painting processes and abnormal noises from doors in final assembly processes. By establishing a mapping relationship between the defect propagation chain and the remediation points, timely intervention can be made before the problem escalates. The application of association rule algorithms helps optimize defect remediation strategies. For example, analysis reveals a strong correlation between welding current and weld strength; adjusting welding parameters can effectively reduce the defect rate. This method not only improves the targeting of remediation but also enables defect traceability and prevention. Constructing a time-series data model of the entire defect lifecycle clearly demonstrates the complete process from defect generation to resolution. For example, recording the discovery time, rework time, and re-inspection time of welding defects ensures the integrity of the information chain, aiding in the analysis of defect handling efficiency and identification. Defect classification and statistics are crucial for optimizing production processes. Defects can be categorized according to type (e.g., welding, painting, assembly) and severity (minor, moderate, severe), generating a priority list. This helps in the rational allocation of resources, prioritizing the handling of defects with greater impact. The application of multi-source data fusion technology further enhances the completeness of the defect lifecycle information chain. Integrating process point data (e.g., welding station number), quality data (weld strength), defect source information (poor welding), and transmission chain data (affecting subsequent coating quality) forms a comprehensive quality control system.

[0152] In summary, this invention discloses a method for tracing defects in the manufacturing process of watch buttons, including obtaining process values ​​and quality values; wherein the quality values ​​include force values, displacement values, and resistance values; preprocessing the force values, displacement values, and resistance values ​​to obtain a performance dataset; performing performance evaluation on the performance dataset based on a multivariate regression analysis method to obtain evaluation results; constructing nodes and performing traceability analysis based on the process values ​​and quality values ​​to obtain a node graph and a mapping relationship table; constructing a model based on the node graph, the mapping relationship table, and the evaluation results to obtain a prediction model; predicting defects based on the prediction model to obtain prediction results; classifying the defect types in the prediction results based on a clustering algorithm to obtain a standard defect library; identifying features based on the standard defect library to obtain a traceability identifier code; and associating real-time process quality data with the traceability identifier code to obtain defect traceability information. This invention establishes a performance index quantification model based on physical characteristics, collects key response time and pressure change data, and constructs a multi-dimensional performance index dataset. It employs multivariate regression analysis to establish a correlation model between key performance and process parameters, identifying key influencing factors. This invention deploys online testing equipment at each processing node to collect process parameters and quality data in real time, constructing a node diagram and the correspondence between processes and quality indicators. Based on the testing results, defects are classified and a standard defect library is established. A defect priority evaluation model is constructed using the analytic hierarchy process (AHP) to generate a list of disposal orders. This invention also designs a defect tracking mechanism based on traceability codes, realizing a full lifecycle information chain for defects, and implements a quality data analysis and early warning system to provide decision support for quality improvement. This invention can effectively improve key performance, reduce defect rates, and achieve full-process control of product quality.

[0153] Reference Figure 2 The second embodiment of the present invention provides a system for tracing the source of defects in the manufacturing process of watch buttons, comprising:

[0154] The data acquisition module 101 is used to acquire process values ​​and quality values; wherein, the quality values ​​include force values, displacement values, and resistance values;

[0155] The data cleaning module 102 is used to preprocess the force value, the displacement and the resistance value to obtain a performance dataset;

[0156] Performance evaluation module 103 is used to perform performance evaluation on the performance dataset based on the multivariate regression analysis method to obtain evaluation results;

[0157] The node construction and analysis module 104 is used to construct nodes and perform traceability analysis based on the process value and quality value to obtain a node diagram and a mapping relationship table.

[0158] Model building module 105 is used to build a model based on the node graph, the mapping relationship table and the evaluation results to obtain a prediction model;

[0159] Defect prediction module 106 is used to predict defects based on the prediction model and obtain prediction results;

[0160] The defect classification module 107 is used to classify the defect types in the prediction results based on a clustering algorithm to obtain a standard defect library;

[0161] Defect coding module 108 is used to perform feature identification based on the standard defect library to obtain a traceability identification code;

[0162] The output module 109 is used to associate and map real-time process quality data with the traceability identifier code to obtain defect traceability information.

[0163] It should be noted that the system for tracing defects in the manufacturing process of watch buttons provided in this embodiment of the invention is used to execute all the process steps of the method for tracing defects in the manufacturing process of watch buttons described in the above embodiment. The working principles and beneficial effects of the two are one-to-one, so they will not be described again.

[0164] This invention also provides an electronic device. The electronic device includes a processor, a memory, and a computer program stored in the memory and executable on the processor, such as a program for tracing defects in the manufacturing process of watch buttons. When the processor executes the computer program, it implements the steps in the above-described methods for tracing defects in the manufacturing process of watch buttons, for example... Figure 1 Step S1 is shown. Alternatively, when the processor executes the computer program, it implements the functions of each module / unit in the above-described device embodiments, such as the data acquisition module.

[0165] For example, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in the electronic device.

[0166] The electronic device may be a desktop computer, laptop, handheld computer, or smart tablet, etc. The electronic device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that the above components are merely examples of electronic devices and do not constitute a limitation on the electronic device. It may include more or fewer components than described above, or combine certain components, or different components. For example, the electronic device may also include input / output devices, network access devices, buses, etc.

[0167] The processor can be a Central Processing Unit (CPU), or 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. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the electronic device, connecting all parts of the electronic device via various interfaces and lines.

[0168] The memory can be used to store the computer programs and / or modules. The processor implements various functions of the electronic device by running or executing the computer programs and / or modules stored in the memory and by calling data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0169] Wherein, if the modules / units integrated in the electronic device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. 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 the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electrical carrier signals and telecommunication signals.

[0170] It should be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and 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 modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the device embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0171] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A method for tracing the source of defects in the manufacturing process of watch buttons, characterized in that, Executed by a computer, including: Obtain process values ​​and quality values; wherein, the quality values ​​include force values, displacement values, and resistance values; The force value, displacement, and resistance value are preprocessed to obtain a performance dataset; The performance dataset was evaluated using a multivariate regression analysis method, and the evaluation results were obtained. Based on the process value and the quality value, node construction and traceability analysis are performed to obtain a node diagram and a mapping relationship table; A prediction model is obtained by constructing a model based on the node graph, the mapping table, and the evaluation results. Defect prediction is performed based on the prediction model to obtain the prediction results; The defect types in the prediction results are classified based on a clustering algorithm to obtain a standard defect library; feature identification is performed based on the standard defect library to obtain a traceability identification code; By associating and mapping real-time process quality data with the traceability identifier, defect traceability information can be obtained.

2. The method for tracing defects in the manufacturing process of watch buttons according to claim 1, characterized in that, The step of preprocessing the force value, the displacement, and the resistance value to obtain a performance dataset includes: Synchronous calibration is performed based on the force value, the displacement, and the resistance value to obtain a calibration dataset; The calibration dataset is time-series aligned to obtain a sequence dataset; Feature extraction is performed on the sequence dataset to obtain a multi-dimensional feature dataset; The multi-dimensional feature data is classified using a clustering algorithm to obtain a performance dataset.

3. The method for tracing defects in the manufacturing process of watch buttons according to claim 1, characterized in that, The performance evaluation of the performance dataset based on the multivariate regression analysis method yields the following evaluation results: A model was constructed for the performance dataset based on the multiple regression analysis method to obtain an initial correlation model; Based on the initial association model, dynamic factors and interaction factors are incorporated to expand the dimensions, resulting in an extended association model. Based on the extended correlation model, feature analysis is performed on the real-time process quality data to obtain feature values; Weight coefficients are obtained by calculating the weights based on the eigenvalues. The performance dataset is optimized based on the weighting coefficients to obtain an optimized set of process parameters; Performance evaluation is performed based on the optimized process parameter set to obtain the evaluation results.

4. The method for tracing defects in the manufacturing process of watch buttons according to claim 3, characterized in that, The step of calculating weights based on the feature values ​​to obtain weight coefficients includes: The weighting coefficients are calculated using the following formula: in, Indicates the first The weighting coefficients of each indicator Indicates the first The characteristic values ​​of each indicator This represents the total number of indicators.

5. The method for tracing defects in the manufacturing process of watch buttons according to claim 1, characterized in that, The step of constructing nodes and performing traceability analysis based on the process values ​​and quality values ​​to obtain a node graph and mapping relationship table includes: Data cleaning and format conversion are performed based on the process values ​​and quality values ​​to obtain standardized data; Based on a pre-defined node relationship model, the standardized data is stored in a distributed repository to obtain a node graph; An anomaly detection analyzer is used to perform anomaly analysis on the node graph. If there are abnormal data points in the node graph, quality tracing and root cause analysis are performed on the abnormal data points to obtain a mapping relationship table.

6. The method for tracing defects in the manufacturing process of watch buttons according to claim 1, characterized in that, The step of identifying the source identification code based on the standard defect database includes: The weights of the standard defect library are calculated based on the analytic hierarchy process to obtain the index weight values. Based on the weight values ​​of the aforementioned indicators, priority scores are calculated for the defects to obtain a score value. The scores are sorted from highest to lowest to obtain the defect processing sequence; The defect processing sequence is encoded based on pre-stored defect encoding rules to obtain a traceability identifier code.

7. The method for tracing defects in the manufacturing process of watch buttons according to claim 1, characterized in that, The step of associating and mapping real-time process quality data with the traceability identifier to obtain defect traceability information includes: The judgment is made based on the real-time process quality data and the preset threshold range; if the real-time process quality data exceeds the preset threshold range, it is determined that there is a defect source. The defect source is retrieved based on the traceability identifier code to obtain the defect information storage location and timestamp; The defect information storage location and the timestamp are output to obtain defect tracing information.

8. A system for tracing defects in the manufacturing process of watch buttons, used to implement the method for tracing defects in the manufacturing process of watch buttons as described in any one of claims 1 to 7, characterized in that, include: The data acquisition module is used to acquire process values ​​and quality values; wherein, the quality values ​​include force values, displacement values, and resistance values. The data cleaning module is used to preprocess the force value, the displacement, and the resistance value to obtain a performance dataset; The performance evaluation module is used to perform performance evaluation on the performance dataset based on the multivariate regression analysis method and obtain the evaluation results. The node construction and analysis module is used to construct nodes and perform traceability analysis based on the process values ​​and quality values, and obtain a node diagram and a mapping relationship table. The model building module is used to build a predictive model based on the node graph, the mapping table, and the evaluation results. The defect prediction module is used to predict defects based on the prediction model and obtain prediction results. The defect classification module is used to classify the defect types in the prediction results based on a clustering algorithm to obtain a standard defect library; The defect coding module is used to perform feature identification based on the standard defect library to obtain a traceability identification code; The output module is used to associate and map real-time process quality data with the traceability identifier to obtain defect traceability information.

9. An electronic device, characterized in that, The device includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor executes the computer program to implement the method for tracing defects in the manufacturing process of watch buttons as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein, when the computer program is executed, it controls the device containing the computer-readable storage medium to perform the method for tracing defects in the manufacturing process of watch buttons as described in any one of claims 1 to 7.

Images