Defective part reprocessing decision-making method based on multi-target balance

Abstract

The invention provides a defective part reprocessing decision-making method based on multi-target balance, which comprises the following steps: acquiring real-time defect rate data and part distribution information from production, analyzing a defect rate change trend and a part quantity fluctuation condition, and setting a standard defect threshold according to an industrial standard and historical data; if the defect rate exceeds a standard defect threshold value, identifying and fitting part performance parameters to obtain an average performance curve, calculating slope change and abrupt change amplitude of the curve, and determining time and influence range of a variation point; acquiring a factory processing load, combining the priority distribution sequence with the target part grouping scheme and the factory processing load to obtain a reprocessing flow configuration scheme, and identifying a balance point of resource consumption and a reprocessing coverage rate through simulation; and acquiring a real-time ratio of the reprocessing coverage rate and the resource consumption corresponding to the balance point, comparing the real-time ratio with a historical optimal ratio, and if the real-time ratio is not smaller than the historical optimal value, determining that a performance bottleneck risk exists, and analyzing a risk cause.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and particularly to a reprocessing decision-making method for defective parts based on multi-objective balance. Background Art

[0002] In the fields of manufacturing and industrial engineering, a reprocessing decision-making method for defective parts based on multi-objective balance is directly related to the optimization of production efficiency, resource utilization rate, and overall performance. With the increasing pursuit of high quality and high efficiency in the global manufacturing industry, how to achieve dynamic optimization of reprocessing in the face of a large number of defective parts has become one of the core issues for enhancing the competitiveness of enterprises. Research in this field not only affects production cost control but also has a profound impact on sustainable development and resource recycling. However, existing reprocessing decision-making methods often expose limitations when dealing with parts with a high defect rate. Traditional methods mostly rely on static role allocation and fixed resource scheduling, lacking the ability to dynamically respond to performance mutations and increased resource competition. When a large number of unqualified parts flood in, these methods are difficult to effectively adjust the part grouping strategy and the factory matching priority, resulting in a decline in reprocessing coverage and increased resource waste. Especially in scenarios with a high defect rate, the mutation points of the average performance curve are insufficiently identified, making the optimization algorithm unable to adapt to the changing load requirements in a timely manner. The core challenges faced by the research focus on several key technical factors: one is the performance mutation caused by the increase in the proportion of parts with a high defect rate, how to accurately detect the mutation points and adjust the strategy accordingly; the second is how the allocation algorithm dynamically balances the part grouping and the factory matching priority under the condition of intensified resource competition; the third is that the impact of these adjustments on the overall reprocessing coverage and the balance of resource consumption has not been fully quantified. These factors lead to performance bottlenecks and resource allocation imbalances in reprocessing when facing a large number of defective parts, thus forming unique technical problems. Therefore, how to design a multi-objective balanced mutation point detection mechanism that can dynamically adjust the part grouping strategy and the factory matching priority when a large number of parts with a high defect rate enter, and optimize the balance between reprocessing coverage and resource consumption has become the key problem that needs to be solved urgently in this research. Summary of the Invention

[0003] The present invention provides a reprocessing decision-making method for defective parts based on multi-objective balance, mainly including:

[0004] Obtain real-time defect rate data and part distribution information from production, analyze the change trend of the defect rate and the fluctuation of the part quantity, and set a standard defect threshold according to industry standards and historical data;

[0005] If the defect rate exceeds the standard defect threshold, identify and fit the part performance parameters, obtain the average performance curve, calculate the slope change and mutation amplitude of the curve, and determine the time and influence range of the mutation point;

[0006] Dynamically cluster parts based on the defect rate and average performance curve to obtain the part grouping corresponding to the mutation point, and determine the target part grouping. The parts include internal parts and cross-plant parts. Allocate internal parts according to the processing capabilities of each process and the real-time defect rate, and allocate cross-plant parts according to geographical location, transportation cost, processing capabilities, and real-time defect rate;

[0007] Identify the resource allocation status and processing capabilities of the factory after implementing the target plan, dynamically allocate parts among internal multi-processes according to the real-time production load, obtain the factory matching priority based on historical production data and real-time production load, and allocate cross-plant parts according to the factory matching priority;

[0008] Obtain the processing load of the factory, combine the priority allocation order with the target part grouping plan and the factory processing load to obtain the reprocessing process configuration plan, and identify the balance point between resource consumption and reprocessing coverage through simulation;

[0009] Obtain the real-time ratio of reprocessing coverage and resource consumption corresponding to the balance point, compare it with the historical optimal ratio. If it is worse than the historical optimal value, determine that there is a risk of performance bottleneck and analyze the risk causes;

[0010] Dynamically update the part grouping strategy and factory matching priority according to the risk causes to obtain the optimized reprocessing process configuration, obtain the optimized defect rate change and performance curve, verify through cycles whether resource consumption is reduced and coverage is improved, judge the optimization stable state, and form an executable improvement plan.

[0011] Furthermore, obtain real-time defect rate data and part distribution information from production, analyze the changing trend of the defect rate and the fluctuation of the number of parts, and set the standard defect threshold according to industry standards and historical data, including: obtain real-time part location information and production quality record form data from the production manufacturing execution, preprocess the data of part numbers, station numbers, production times, and inspection results, remove outliers and missing values, and obtain the defect data set by associating the defect occurrence station and time point through the real-time traceability code of the part. For the preprocessed defect data set, classify and count according to station number, part type, and defect type, and calculate the real-time defect rate data of each station by dividing the number of defects by the total production volume within that time period. Use an automatic detection device to randomly inspect the products at each station on the production line at a preset frequency, associate and compare the obtained detection data with the real-time defect rate data, and calculate the defect detection rate index through the Pearson correlation coefficient. Weight the defect data according to the defect detection rate index, and use the exponential smoothing method to perform time series analysis on the weighted defect rate data to obtain the changing trend curve of the defect rate. Extract historical defect records from the product quality management database, identify the outlier interval of the defect rate through box plot analysis, combine industry standards to set the defect rate warning threshold, and establish the defect rate monitoring baseline. Collect the position distribution data of the parts at each station detection point, calculate the part distribution density through the density clustering algorithm, and analyze the correlation between the station production capacity and the defect rate in combination with the defect rate changing trend curve.

[0012] Furthermore, if the defect rate exceeds the standard defect threshold, identify and fit the part performance parameters to obtain the average performance curve, calculate the slope change and mutation amplitude of the curve, and determine the time and influence range of the mutation point, including: obtain the measurement data of performance parameters such as size, weight, surface finish, hardness, and temperature from the part detection equipment. For the stations where the defect rate exceeds the preset threshold range, use sensors to collect the original data of the part performance parameters in real time, and perform normalization processing on the original data to obtain the standardized performance parameter set. Remove outliers from the standardized performance parameter set, calculate the mean and standard deviation of each parameter through the sliding window method, identify the outliers according to the three-standard-deviation principle, and obtain the performance parameter time series after removing the abnormal data. For the performance parameter time series, use the least squares method to calculate the fitting curve equation, calculate the parameter change rate of adjacent time points through the curve fitting equation, and obtain the parameter slope sequence. Set the slope change threshold according to the historical performance parameter fluctuation range, traverse the parameter slope sequence, and mark it as a mutation point when the slope difference between adjacent two points exceeds the set threshold, and record the corresponding moment of the mutation point. With the mutation point moment as the center, extend to the front and back time series, calculate the deviation between the parameter value and the parameter value at the mutation point moment, and record the time point when the deviation value falls back within the threshold range. Use the clustering method to group the recorded time points, calculate the time interval of each group of time points, and determine the start and end time points of the influence range through the density distribution of the time interval.

[0013] Further, dynamic clustering of parts is performed based on the defect rate and the average performance curve to obtain the part grouping corresponding to the mutation points, and the target part grouping is determined. The parts include internal parts and cross-plant parts. The internal parts are allocated according to the processing capabilities of each process and the real-time defect rate, and the cross-plant parts are allocated according to the geographical location, transportation cost, processing capabilities and real-time defect rate, including: calculating the Euclidean distance between the slope value of the part performance curve and the defect rate data by using the k-means clustering method, generating a distance sequence by sorting the distance values from small to large, and performing initial grouping of the parts through the minimum distance threshold to obtain a mutation part identification table. Grouping the parts in the mutation part identification table, calculating the fluctuation amplitude of the performance curve through the curve fitting equation, and performing secondary division of the part grouping in combination with the size of the real-time defect rate to obtain a list of target part groupings. Extracting the accuracy inspection records, operation duration records, and fault repair records of the processing equipment from the production management database, calculating the equipment production capacity index through the processing accuracy qualification rate, equipment operation duration, and mean time between failures, and obtaining a process-level processing capacity evaluation table. For the internally processed parts, reading the equipment production capacity index and the real-time defect rate data, calculating the remaining processing duration of each process, and constructing a process production capacity allocation network through the maximum flow network algorithm to obtain an internal part allocation plan. Obtaining the inter-factory transportation distance data from the geographical information database, calculating the inter-factory logistics cost matrix in combination with the transportation cost per kilometer, and generating an inter-plant processing cost matrix through the factory production capacity index and the real-time defect rate. Using the multi-objective programming algorithm to allocate the cross-plant parts, constructing a constraint equation set with the logistics cost matrix and the processing cost matrix, and obtaining the optimal allocation plan for the cross-plant parts through genetic iterative operation.

[0014] Furthermore, obtain the real-time defect rate data of internal parts and the transportation costs of cross-factory parts. Generate an initial parts grouping based on the real-time defect rate data. Identify the process processing capacity vectors in the initial parts grouping according to the equipment status of each factory. Generate a target internal parts allocation plan based on the process processing capacity vectors. Perform data fusion on the target internal parts allocation plan and the cross-factory parts allocation plan to obtain a global parts scheduling instruction set, including: obtaining part processing records and quality inspection data from the production execution database, calculating the real-time defect rate of internal parts, extracting the cross-factory part transportation distance and freight rate standard from the logistics management database, dividing the defect rate less than 1% into a low-risk group, the defect rate between 1% and 3% into a medium-risk group, and the defect rate greater than 3% into a high-risk group through a three-layer grouping rule to generate an initial parts grouping table. Collect the operation status data of processing equipment in each factory from the equipment management database, assign weights of 0.3 to the proportion of continuous operation duration of the equipment, 0.4 to the qualified rate of equipment precision inspection, and 0.3 to the equipment maintenance interval duration, and calculate the comprehensive equipment status score through weighted summation. Construct a process processing capacity vector based on the comprehensive equipment status score, with the vector dimension including processing precision index, processing beat index, and equipment stability index, and use the least squares method to calculate the process equipment status eigenvalue. For each part identifier in the internal parts grouping table, calculate the matching degree between the part processing requirements and the process processing capacity vector through Euclidean distance, and select the process equipment with the highest matching degree as the target process. Use the priority sorting method to preferentially allocate the process equipment with the best processing capacity to the high-risk group parts, allocate the process equipment with medium processing capacity to the medium-risk group parts, and allocate the remaining process equipment to the low-risk group parts to obtain an internal parts allocation plan. Read the cross-factory parts allocation plan from the factory collaboration platform, establish a correspondence table between internal parts and cross-factory parts through part code association, and sort the parts according to the parts in the risk level correspondence table. Generate a processing priority sequence according to the part risk level sorting result, and generate a process scheduling time table corresponding to the priority sequence to form a global parts scheduling instruction data set.

[0015] Further, identify the resource allocation status and processing capacity of the factory after implementing the target plan, dynamically allocate parts among internal multi-processes according to the real-time production load, obtain the factory matching priority based on historical production data and real-time production load, and allocate cross-factory parts according to the factory matching priority, including: obtaining process scheduling data and equipment operation records from the production management platform, calculating the equipment processing load using a 4-hour fixed time window, obtaining the equipment load rate by dividing the number of online processed parts by the equipment standard production capacity, and generating a process equipment load data table. Calculate the remaining available working hours for each equipment in the load data table, and use the linear programming method to calculate the upper and lower limits of equipment production capacity according to the process sequence specified in the part processing process regulations, obtaining a process-level remaining production capacity data table. Extract the actual beat time records of each process in the past 30 days from the historical production database, calculate the process standard processing beat through the weighted average method, and generate a process processing capacity score table according to the deviation value between the standard beat and the actual beat. Normalize the data in the process processing capacity score table, combine the real-time load rate calculation results, and construct a process production capacity prediction curve through a multi-layer perceptron to obtain a process-level production capacity prediction data table. According to the process production capacity prediction data, use the minimum cost flow algorithm to allocate internal parts among multiple processes, and determine the priority allocation order by calculating the ratio of the unit processing duration of the part to the remaining production capacity of the process. Obtain the production equipment configuration list from the factory resource database, summarize the equipment production capacity and actual load data of each factory, calculate the comprehensive factory ability score through the analytic hierarchy process, and generate a factory matching priority sequence. Use the maximum flow algorithm to allocate cross-factory parts, divide the high-priority factory group and the low-priority factory group according to the factory priority sequence, and generate a cross-factory part allocation plan according to the principle of proximity.

[0016] Further, obtain the factory processing load, combine the priority allocation order with the target part grouping scheme and the factory processing load to obtain a reprocessing process configuration scheme, and identify the balance point between resource consumption and reprocessing coverage rate through simulation, including: obtaining the factory real-time processing load and equipment operation data from the manufacturing execution database, using the moving average method to statistically calculate the operation rate and production capacity utilization rate of process equipment, dividing the process load into high-load intervals, medium-load intervals, and low-load intervals according to the production capacity utilization rate, and generating a factory load distribution data table. Extract the operation rate, production capacity utilization rate, and maintenance interval duration of process equipment as characteristic parameters according to the priority sequence in the part grouping scheme, and use the random forest algorithm to calculate the correlation score between the priority sequence and the load characteristics. Sort the correlation scores, screen the matching processes by setting a correlation threshold, and generate a reprocessing process path table according to the processing sequence constraints between processes. Use a discrete event simulator to construct a reprocessing process scenario, input the process equipment parameters, processing cycle time, and line change time into the simulation environment to generate a simulation operation data set. Extract the equipment operation duration and maintenance interval duration from the simulation operation data set, and calculate the resource consumption rate curve by dividing the cumulative operation time by the standard operation time. Statistically calculate the part processing records during the execution of the reprocessing process, and calculate the coverage rate curve by dividing the number of completed parts by the planned number. Use the piecewise regression method to fit the resource consumption rate curve and the coverage rate curve, and calculate the process parameter combination corresponding to the intersection of the curves. Generate a reprocessing balance configuration scheme including equipment configuration, processing cycle time, and line change time according to the process parameter settings at the intersection point.

[0017] Further, classify the reprocessing process configuration schemes to obtain a set of configuration schemes. Obtain the resource consumption data and reprocessing coverage rates of different configuration schemes from the set of configuration schemes. If the resource consumption data exceeds the preset consumption threshold, adjust the configuration schemes to obtain an optimized set of configuration schemes. According to the optimized set of configuration schemes, use the linear regression algorithm to predict the relationship between resource consumption and reprocessing coverage rate to obtain the balance point data, including: extract the process parameters, equipment configuration, and processing rhythm in the reprocessing process configuration scheme, use the k-means clustering algorithm to calculate the Euclidean distance between the schemes, and divide them into high similarity groups, medium similarity groups, and low similarity groups according to the distance value size to generate a configuration scheme grouping table. Read the equipment operation records of each configuration scheme from the production execution database, calculate the equipment utilization rate according to the single-shift standard working hours, calculate the resource consumption index through the ratio of the cumulative equipment operation time to the maintenance interval, and calculate the reprocessing coverage rate index by dividing the number of online completed parts by the planned number. Obtain the resource consumption baseline value from the historical production database, set the resource consumption threshold to 1.2 times the baseline value, and when the resource consumption index exceeds the threshold, give priority to adjusting the processing rhythm parameters. If it still exceeds the threshold after the rhythm adjustment, increase the number of equipment, and calculate the production capacity redundancy through the product of the number of equipment and the rhythm time. Use the genetic algorithm to optimize the adjustment scheme, set the production capacity utilization rate as the fitness function, and perform iterative calculations with the constraint condition that the resource consumption index is less than the threshold. Extract the resource consumption index and coverage rate index from the optimized configuration scheme, use the linear regression algorithm to fit the two sets of index data to generate a consumption rate curve and a coverage rate curve. Calculate the intersection coordinates of the two curves by the least squares method to obtain the resource consumption value and coverage rate value corresponding to the balance point. Use the cross-validation method to test the balance point data, verify the stability of the balance point parameter combination through historical data, and generate a balance parameter verification report.

[0018] Further, obtain the reprocessing coverage rate and the real-time ratio of resource consumption corresponding to the balance point, and compare them with the historical optimal ratio. If it is worse than the historical optimal value, it is determined that there is a risk of performance bottleneck, and the risk causes are analyzed, including: reading the reprocessing data record corresponding to the balance point from the production execution database, calculating the coverage rate value by dividing the number of completed parts by the planned quantity, calculating the resource consumption value by dividing the equipment operation duration by the standard man-hour, and obtaining the real-time performance ratio by dividing the coverage rate by the resource consumption. Extract the production records of the most recent 30 days from the historical database, sort the daily performance ratios, select the top 10% of the records as the optimal ratio samples, and use the support vector regression method to establish a performance prediction curve. Set the warning threshold based on the performance prediction curve. When the real-time performance ratio is lower than the warning threshold, it is marked as the risk warning state, and it is divided into three levels: mild warning, moderate warning, and severe warning according to the degree of deviation from the threshold. For the warning data, extract the operation parameters from the equipment management database, including processing accuracy, operation duration, and maintenance interval, normalize the parameters, and generate the equipment state feature vector. Use the hierarchical clustering method to group the equipment state feature vectors, identify the abnormal parameter combinations by calculating the inter-class distance, and generate the parameter anomaly feature table. Calculate the confidence level of parameter anomaly and performance reduction based on the association rule algorithm, and screen the high-risk parameter combinations through the confidence level threshold. Classify the high-risk parameter combinations according to equipment accuracy class, equipment life class, and equipment maintenance class, and generate the risk cause classification table. Use the decision tree algorithm to construct a risk warning model, verify the risk causes in combination with the historical failure data, and output the risk warning report.

[0019] Furthermore, dynamically update the part grouping strategy and factory matching priority according to the risk causes, obtain the optimized reprocessing flow configuration, obtain the optimized defect rate change and performance curve, verify through iteration whether the resource consumption is reduced and the coverage rate is increased, determine the optimized stable state, and form an executable improvement plan, including: extracting the equipment status characteristics and performance reduction records from the risk cause database, setting three grade criteria of high-risk threshold, medium-risk threshold, and low-risk threshold according to the historical performance data, using a deep learning algorithm to conduct risk prediction and classification on the parts, and generating an updated part grouping table. Score the equipment operation status, set the equipment status weight at 0.4, the processing accuracy weight at 0.4, and the maintenance cycle weight at 0.2, and calculate the comprehensive factory score through weighted summation. Sort the equipment based on the comprehensive factory score, select the top 30% of the equipment with the highest scores as the highest-priority group, the middle 40% of the equipment with the scores as the medium-priority group, and the remaining equipment as the lowest-priority group. According to the priority grouping result, use a genetic algorithm to optimize the process parameters, with the product of the equipment load rate and the production capacity utilization rate as the optimization objective function. Verify the optimized process parameters through the production capacity calculation module, calculate the production capacity matching degree between each process, and generate a new processing beat and line change cycle parameter table. Monitor the optimized production data, and use a 4-hour fixed time window to statistically analyze the defect rate change and performance curve fluctuation. Calculate the reduction ratio of resource consumption and the increase ratio of coverage rate. When the index fluctuation is less than 5% within three consecutive monitoring cycles, it is determined that the stable state is reached. Use the cross-validation method to test the stable state data, verify the stability of the optimization plan by calculating the index fluctuation range of the verification samples, and generate an improvement plan verification report.

[0020] The technical solution provided by the embodiment of the present invention may include the following beneficial effects:

[0021] The present invention discloses a reprocessing decision-making method for defective parts based on multi-objective balance. By obtaining the production real-time defect rate data and part distribution information, analyzing the defect rate change trend and setting a standard threshold, when the defect rate exceeds the standard, identify and fit the part performance parameters, and determine the mutation point and its influence range. Dynamically cluster and group the parts according to the defect rate and performance curve, and allocate internal and cross-factory parts based on factors such as process processing capacity, geographical location, and transportation cost. The present invention also identifies the balance point between resource consumption and reprocessing coverage rate through simulation, dynamically updates the part grouping strategy and factory matching priority, and forms an executable improvement plan. This method can effectively reduce resource consumption, improve the reprocessing coverage rate, and realize the intelligent configuration of the part reprocessing process. Description of the Drawings

[0022] Figure 1 It is a flowchart of a reprocessing decision-making method for defective parts based on multi-objective balance of the present invention.

[0023] Figure 2 This is a schematic diagram of a reprocessing decision-making method for defective parts based on multi-objective balance of the present invention. Detailed implementation manners

[0024] The technical solutions of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0025] In the embodiments of the present invention, for the convenience of understanding, relevant terms need to be briefly described: Defective parts refer to products that do not meet the standards due to dimensional tolerances, surface defects or other quality problems during the production process, and need to be repaired through reprocessing to restore their functions. Defective parts can be divided into internal parts and cross-factory parts, corresponding to parts that are repaired within the factory or need to be transported to other factories for reprocessing respectively. Variation points refer to time points or states where significant changes in performance occur due to sudden increases in defect rates or abnormal part distributions, usually identified by changes in the slope or mutation amplitude of the performance curve. The reprocessing coverage rate refers to the proportion of the number of defective parts successfully reprocessed to the total number of defective parts, which is a key indicator for measuring reprocessing efficiency.

[0026] In the embodiments of the present invention, by collecting production data in real time, analyzing the trend of defect rates, and dynamically optimizing the scheduling strategy, the performance bottleneck problem of traditional methods in high-defect-rate scenarios is solved. The following combines Figure 1 , and details the technical solutions of the embodiments of the present invention.

[0027] S101. Obtain real-time defect rate data and part distribution information from production, analyze the change trend of the defect rate and the fluctuation of the number of parts, and set a standard defect threshold according to industry standards and historical data.

[0028] In the embodiments of the present invention, when starting the reprocessing process of defective parts in production, first obtain real-time data from the manufacturing execution and analyze it to determine the change trend of the defect rate and the part distribution characteristics. The specific operations include the following steps:

[0029] Such as Figure 2 , S1011. Obtain real-time part position information and production quality record data from the production manufacturing execution, associate the defective occurrence station and time point through the part traceability code, and generate a defect data set.

[0030] In an embodiment of the present invention, production and manufacturing execution records the processing status and quality information of parts at each station. Through the unique traceability code of the part, the station it passes through on the production line, the processing time and the inspection results are associated to generate a defect data set containing the station and time point where the defect occurred. For example, in the automobile engine assembly line, the parts of the cylinder processing station are marked with barcodes, and the processing parameters and quality inspection results of each part are recorded. If a certain station detects a dimensional deviation, the quality inspector enters the unqualified information through the data acquisition terminal, and generates a corresponding defect data set based on the traceability code. This step does not place too many restrictions on the specific method of data collection, which can be determined by technical personnel according to the actual scenario.

[0031] S1012. Preprocess the defect data set, remove outliers and missing values, classify and count according to the workstation number and defect type, and calculate the real-time defect rate data of each workstation.

[0032] In an embodiment of the present invention, the collected defect data set is cleaned to remove records with missing part numbers, wrong station numbers or incomplete inspection results. Then, group statistics are performed according to the station number and defect type, and the real-time defect rate of each station is calculated by dividing the number of defects by the total production in the time period. For example, if a station produces 500 products on a certain day, 10 of which are judged to be defective due to dimensional deviation, then the real-time defect rate of the station is 2%. This step provides a reliable data basis for subsequent trend analysis through classification statistics.

[0033] S1013. Use an automatic detection device to obtain workstation detection data, calculate the defect detection rate index through the Pearson correlation coefficient, and perform weighted processing on the defect rate data.

[0034] In an embodiment of the present invention, an automatic detection device is used to perform random inspections on products at each station of the production line at a preset frequency, and after obtaining the detection data, it is correlated and compared with the real-time defect rate data. By calculating the Pearson correlation coefficient, a defect detection rate index is generated to evaluate the reliability of the detection results. For example, a three-coordinate measuring machine is used for dimensional inspection at the cylinder processing station, and one piece is randomly inspected for every 50 pieces, and the inspection data is uploaded to the quality management in real time. If the detection rate is lower than 90%, the adjustment of the detection parameters will be triggered. When weighted processing is performed on the defect rate data, different weights are assigned according to the severity of the defect type, such as the weight of critical dimension deviation is 1.5, and the weight of surface defects is 1.0.

[0035] S1014. Use exponential smoothing method to perform time series analysis on the weighted defect rate data to generate a defect rate change trend curve.

[0036] In the embodiments of the present invention, exponential smoothing method is used for time series analysis of the weighted defect rate data to generate a trend curve reflecting the change of the defect rate. Higher weights are assigned to recent data. For example, the weight coefficient of the data within the most recent week is 0.3, and the weights of historical data decay over time. Through the trend curve, the change range of the defect rate within the next 24 hours can be predicted, providing a basis for subsequent change point detection. This step does not overly limit the specific parameters of the smoothing algorithm and can be adjusted according to actual requirements.

[0037] S1015. Identify the outlier interval of the defect rate through box plot analysis, and set the standard defect threshold in combination with industry standards and historical data.

[0038] In the embodiments of the present invention, defect records for the past three months are extracted from the product quality management database, and box plot analysis is used to identify the outlier interval of the defect rate. For example, the normal defect rate of a certain workstation fluctuates between 0.5% and 1.5%, and the upper limit is the quartile plus 1.5 times the interquartile range. If the real-time defect rate exceeds 2%, an alarm will be triggered. In combination with industry standards and historical data, the standard defect threshold is set as the baseline for subsequent change point detection. This step ensures the rationality of the threshold through outlier analysis.

[0039] S1016. Analyze the position distribution data of parts at each workstation, calculate the part distribution density through density clustering algorithm, and analyze the correlation between workstation production capacity and defect rate in combination with the defect rate change trend curve.

[0040] In the embodiments of the present invention, the real-time position information of parts on the production line is collected through workstation sensors, the adjacent part spacing and production rhythm are calculated, and the density clustering algorithm is used to generate the part distribution density. For example, the inspection rhythm of an assembly workstation is 90 seconds. If the part spacing is less than 85 seconds due to too fast feeding speed at the upstream workstation, the production rhythm will be adjusted to ensure inspection quality. In combination with the defect rate change trend curve, analyze the correlation between workstation production capacity and defect rate, providing data support for dynamic optimization of part grouping.

[0041] In the embodiments of the present invention, through the above steps, it is possible to efficiently collect real-time defect rate data and part distribution information from production, and set the standard defect threshold through trend analysis and outlier detection. These operations ensure the accuracy of subsequent change point detection and part grouping strategies, can significantly improve the reprocessing coverage rate, optimize the resource utilization efficiency, and meet the urgent needs of the manufacturing industry for high-quality production.

[0042] S102. If the defect rate exceeds the standard defect threshold, collect the part performance parameters from production, fit and generate the average performance curve, analyze the slope change and mutation amplitude of the curve, and determine the time point and influence range of the change point.

[0043] In an embodiment of the present invention, when the defect rate at a certain station during production exceeds a preset standard defect threshold, a process for collecting and analyzing performance parameters will be triggered. By fitting the part performance curve and detecting its mutation characteristics, the variation point and its influence range can be accurately located, providing a basis for part grouping and reprocessing optimization, and ensuring the ability to quickly respond to abnormal fluctuations in high defect rate scenarios. The embodiment of the present invention does not limit the specific type or collection method of performance parameters, which can be determined by technicians according to the actual production scenario.

[0044] S1021. Obtain multi-dimensional performance parameter data from the part detection device, perform normalization processing on the data, and generate a standardized performance parameter set.

[0045] In an embodiment of the present invention, multi-dimensional performance parameters are collected in real time from the part detection device through a distributed sensor network, such as key indicators like size, weight, surface finish, hardness, and temperature. Taking the processing of an automotive engine piston as an example, the collected parameters include piston outer diameter, pin hole coaxiality, ring groove runout value, and surface roughness. When the defect rate at a certain station exceeds 2.5%, the full inspection program is started, and the sensor collects raw data at a high frequency. Since the dimensions of each parameter are different, such as the size in millimeters and the hardness in HRC, the maximum-minimum normalization method is used to map all parameter values to a unified range of 0 to 1, generating a standardized performance parameter set. This step eliminates the dimension difference through normalization processing.

[0046] S1022. For the standardized performance parameter set, use the sliding window method to calculate the local mean and standard deviation, and eliminate abnormal data according to the three-sigma principle to generate a performance parameter time series.

[0047] In an embodiment of the present invention, outlier processing is performed on the standardized performance parameter set to ensure data reliability. The sliding window method is used, with a window of 20 data points, to calculate the local mean and standard deviation of each parameter. If a parameter value deviates from the mean by more than three standard deviations, it is determined as an outlier and eliminated. For example, the jump value generated by equipment jitter in the piston outer diameter parameter sequence will be removed. After eliminating the outliers, a continuous performance parameter time series is generated, providing high-quality data for subsequent curve fitting. This step does not strictly limit the window size or the outlier determination standard, which can be adjusted according to the production environment.

[0048] S1023. Based on the performance parameter time series, use the least squares method to fit the average performance curve, calculate the parameter change rate at adjacent time points, and generate a parameter slope sequence.

[0049] In an embodiment of the present invention, the least squares method is used to perform curve fitting on the time series of performance parameters to generate an average performance curve reflecting the changing trend of the parameters over time. Taking the outer diameter of the piston as an example, the fitting curve describes the fluctuation law of the dimensional parameters during the production process. At intervals of 10 seconds, the parameter change rate between adjacent time points of the curve is calculated to form a parameter slope sequence. Under normal production conditions, the change rate of the outer diameter of the piston usually fluctuates within ±0.002 mm / s. This step quantifies the dynamic characteristics of parameter changes through fitting and slope calculation, laying a foundation for the detection of mutation points.

[0050] S1024. Analyze the parameter slope sequence, determine whether the slope difference between adjacent time points exceeds a preset threshold, mark the mutation point and record its time point.

[0051] In an embodiment of the present invention, according to the statistics of historical performance parameter fluctuations, the slope change threshold is set to 5 times the normal fluctuation range. For example, the slope change threshold for the outer diameter of the piston is 0.01 mm / s. Traverse the parameter slope sequence, calculate the slope difference between adjacent time points. If the difference exceeds the threshold, mark this time point as a mutation point and record its moment. For example, when the slope of the outer diameter of the piston suddenly changes to 0.012 mm / s at a certain moment, mark this point as a mutation point.

[0052] S1025. Extend forward and backward from the mutation point in the time series, calculate the parameter deviation, determine the start and end time points of the influence range, and confirm the time interval through cluster analysis.

[0053] In an embodiment of the present invention, with the marked mutation point as the center, extend forward and backward in the time series, and calculate the deviation between the parameter values of each time point and the parameter value of the mutation point. When the deviation value falls back to the normal fluctuation range, record the corresponding time point. For example, when the deviation of the outer diameter of the piston mutation point falls back to ±0.002 mm / s, record this moment as the boundary of the influence range. Use the density clustering algorithm to group multiple mutation points and boundary time points, and calculate the density distribution of the time intervals of each group. For example, if 5 dense mutation points are detected during the period from 9:30 to 9:45, through density distribution analysis, determine the influence range to be from 9:25 to 9:50, indicating that there are significant abnormalities in the production process during this time period.

[0054] In an embodiment of the present invention, through the above steps, it is possible to efficiently identify the mutation points of performance parameters in high defect rate scenarios, and determine their time points and influence ranges. This step makes full use of the dynamic characteristics of multi-dimensional performance parameters, and through normalization, outlier removal, curve fitting and cluster analysis, ensures the accuracy and robustness of mutation point detection. The obtained results can provide reliable data support for subsequent part grouping and factory matching optimization, significantly improve the adaptability and efficiency of reprocessing, and ultimately achieve a higher quality production process.

[0055] S103. Dynamically cluster parts based on defect rate data and average performance curves, generate part groupings corresponding to variation points, determine the target part grouping, which includes internal parts and cross-plant parts, optimize the part allocation plan in combination with process processing capabilities, real-time defect rates, geographical locations, and transportation costs, and generate a global part scheduling instruction set through equipment status evaluation and matching degree calculation.

[0056] In the embodiment of the present invention, when the defect rate detected at a certain station during production exceeds the standard defect threshold, the defect rate data and performance curves will be used to group parts by means of dynamic clustering, distinguish internal parts and cross-plant parts, and optimize the allocation plan according to factors such as processing capabilities and transportation costs. At the same time, a global scheduling instruction set is generated through equipment status evaluation and processing requirement matching to ensure the efficiency and stability of the reprocessing process.

[0057] S1031. Based on the slope of the part performance curve and defect rate data, use the k-means clustering method to calculate the Euclidean distance, generate a variation part identification table, and determine the target part grouping list through curve fluctuation analysis.

[0058] In the embodiment of the present invention, taking the slope value of the part performance curve and the real-time defect rate as characteristic parameters, use the k-means clustering method to calculate the Euclidean distance between parts, generate a distance sequence, and perform initial grouping according to the preset minimum distance threshold to obtain a variation part identification table. Taking the processing of automobile engine cylinder blocks as an example, when 15 out of 100 cylinder blocks in a certain batch have dimensional deviations, and the slope of the performance curve reaches 0.015 mm / s, far exceeding 0.002 mm / s of normal batches, this batch is marked as a variation part group through a distance threshold of 0.01. Subsequently, the least squares method is used to fit the performance curve of the variation part group to calculate the fluctuation amplitude. If the fluctuation amplitude reaches 0.08 mm, significantly higher than 0.02 mm of normal batches, and the real-time defect rate is 5%, secondary clustering is performed in combination with the fluctuation amplitude and defect rate to divide the parts into high-risk groups and medium-risk groups, where the parts in the high-risk group have a fluctuation amplitude exceeding 0.05 mm and a defect rate greater than 3%. This step ensures the accuracy and pertinence of part grouping through clustering and fluctuation analysis.

[0059] S1032. Extract the operation records of processing equipment from the production management database, calculate the equipment production capacity index, construct a process production capacity allocation network, and generate an internal part allocation plan.

[0060] In an embodiment of the present invention, accuracy detection records, operation duration records, and fault repair records of processing equipment are obtained from a production management database, and an equipment production capacity index is calculated. For example, the monthly accuracy qualification rate of a certain machining center is 98.5%, the cumulative operation time is 480 hours, and the average trouble-free operation time is 96 hours. According to the accuracy qualification rate, operation duration, and fault interval time, the calculated production capacity index is 0.92, which is higher than the benchmark value of 0.85. For internal parts, real-time defect rate data and remaining processing durations of each process are read. Taking the rough machining process as an example, the remaining production capacities of three machining centers are 120 hours, 80 hours, and 60 hours respectively, and the defect rates are 1.2%, 1.5%, and 1.8% respectively. Using the maximum flow network algorithm, a process production capacity allocation network is constructed, and 80% of high-risk parts are allocated to the machining center with the lowest defect rate to ensure processing quality. This step maximizes the internal processing efficiency by quantifying equipment production capacity and optimizing allocation.

[0061] S1033. Combine geographical information and cost data to optimize the cross-factory part allocation plan.

[0062] In an embodiment of the present invention, the transportation distances between factories are extracted from a geographical information database, and a logistics cost matrix is calculated in combination with the unit transportation cost. For example, the distances between the main factory and two cooperative factories are 120 kilometers and 180 kilometers respectively, and the unit transportation cost is 5 yuan / km, generating a corresponding logistics cost matrix. At the same time, according to the production capacity indexes of 0.88 and 0.85 of the cooperative factories, and the real-time defect rates of 2.1% and 2.3%, a processing cost matrix is generated. Using the multi-objective programming algorithm, a constraint equation set including logistics cost and processing cost is constructed, and through genetic algorithm iteration optimization, under the conditions that the transportation cost does not exceed 1 million yuan and the production capacity utilization rate is not less than 85%, the allocation plan is determined: the cooperative factory with a shorter distance undertakes 60% of the cross-factory parts, the remote factory undertakes 25%, and the main factory retains 15% of the key part processing.

[0063] S1034. Based on part processing records and equipment status data, generate a global part scheduling instruction set through risk level grouping and processing capacity matching.

[0064] In the embodiments of the present invention, part processing records are extracted from the production execution database, including quality inspection data and defect rate data. Parts are divided into a low-risk group with a defect rate less than 1%, a medium-risk group with a defect rate of 1% to 3%, and a high-risk group with a defect rate greater than 3% according to a three-layer grouping rule, and an initial part grouping table is generated. Taking the cylinder block processing as an example, among 100 cylinder blocks in a certain batch, the defect rate reaches 4.2%. Among them, 25 pieces enter the high-risk group, 50 pieces enter the medium-risk group, and 25 pieces enter the low-risk group. The operation status data of the processing equipment is collected from the equipment management database, and the comprehensive equipment status score is calculated through the weight of the continuous operation duration ratio of 0.3, the weight of the accuracy inspection qualification rate of 0.4, and the weight of the maintenance interval duration of 0.3. For example, for a certain machining center, the continuous operation ratio is 85%, the accuracy qualification rate is 98%, and the fault-free operation time is 96 hours, and the score is 0.92. An operation processing capacity vector is constructed based on the score, including three dimensions of machining accuracy, machining beat, and equipment stability, and the operation equipment status eigenvalue is calculated by the least squares method. For each group of parts, the Euclidean distance between the part processing requirements and the operation processing capacity vector is calculated, and the operation equipment with the highest matching degree is selected. The parts in the high-risk group are preferentially allocated to the equipment with the highest status score, the medium-risk group is allocated to the medium-level equipment, and the low-risk group is allocated to the remaining equipment. The internal and cross-factory part allocation schemes are integrated, and a processing priority sequence is generated by associating part codes, and a global scheduling instruction set including processing equipment, time periods, and operation sequences is generated.

[0065] In the embodiments of the present invention, through the above steps, it is possible to realize dynamic clustering of parts based on the defect rate and performance curve, optimize the internal and cross-factory part allocation, and generate a global scheduling instruction set. This method significantly improves the adaptability of reprocessing and the resource utilization efficiency through multi-dimensional data analysis and algorithm optimization, and provides stable support for the production process in high-defect-rate scenarios.

[0066] S104. Based on the factory resource status and processing capacity after the execution of the target allocation plan, combined with the real-time production load and historical data, dynamically optimize the part allocation among internal multi-operations, and determine the matching priority through the factory comprehensive ability assessment, and allocate cross-factory parts to improve the reprocessing efficiency.

[0067] In the embodiments of the present invention, after the execution of the target allocation plan, by real-time monitoring the factory resource status and processing capacity, combined with the production load data and historical production records, dynamically adjust the part allocation strategy among internal multi-operations, and generate a matching priority sequence according to the factory comprehensive ability assessment to optimize the cross-factory part allocation. The embodiments of the present invention do not strictly limit the data collection frequency or algorithm parameters, which can be adjusted by technicians according to the actual production scenario.

[0068] S1041. Obtain process scheduling data and equipment operation records from the production management platform, calculate the equipment load rate, generate a process-level load data table, and determine the upper and lower limits of equipment capacity through a linear programming method.

[0069] In an embodiment of the present invention, the process scheduling data and equipment operation records are extracted from the production management platform, and the ratio of the number of online processed parts to the standard capacity of the equipment is calculated with 4 hours as a fixed time window to generate equipment load rate data. For example, on the automobile engine cylinder production line, the standard capacity of a rough processing equipment is 8 pieces per hour, and 6 pieces are currently processed, with a load rate of 75%. The load rate data is summarized into a process-level load data table. According to the process sequence specified in the process regulations, such as the cylinder body processing needs to go through rough processing, fine processing and honing in sequence, the linear programming method is used to calculate the upper and lower limits of the capacity of each process equipment in combination with the equipment operation status and load rate. For example, the remaining working hours of the rough processing equipment are 120 minutes, the fine processing equipment is 180 minutes, and the honing equipment is 150 minutes, and the process-level remaining capacity data table is generated. This step provides an accurate basis for subsequent allocation by quantifying the load rate and capacity range.

[0070] S1042. Extract historical production data, calculate the standard processing rhythm of the process, generate a processing capability score based on the rhythm deviation, and predict the process capacity trend through a multi-layer perceptron.

[0071] In an embodiment of the present invention, the actual cycle time records of the process in the past 30 days are extracted from the historical production database, and the weighted average method is used to calculate the standard processing cycle of the process. For example, the standard cycle of a certain finishing equipment is 15 minutes / piece, the actual cycle is an average of 16.5 minutes / piece, and the deviation rate is 10%. The processing capacity score is calculated to be 0.90. Similarly, the scores of the roughing and honing processes are 0.95 and 0.92, respectively. The scoring data is normalized, combined with the real-time load rate, and input into the multi-layer perceptron model to predict the process capacity trend in the next 4 hours. The model takes the load rate, processing capacity score and historical capacity data as input, and the output shows that the capacity utilization rate of the roughing process is expected to reach 85%, the finishing process 78%, and the honing process 70%. This step enhances the foresight of the allocation decision through beat analysis and prediction.

[0072] S1043. Use the minimum cost flow algorithm to optimize the internal multi-process parts allocation and determine the priority allocation order based on the ratio of unit processing time to remaining capacity.

[0073] In the embodiment of the present invention, for the parts to be processed, such as 50 cylinder blocks, the minimum cost flow algorithm is used for multi-process allocation. Calculate the ratio of the unit processing duration of each part in each process to the remaining production capacity of the process. The ratio for rough machining is 0.15, for finish machining is 0.12, and for honing is 0.10. Based on the ratio, 20 cylinder blocks are preferentially allocated to the honing process, 18 to the finish machining process, and 12 to the rough machining process to minimize the processing cost and balance the process load. This step ensures efficient and balanced part allocation through the minimum cost flow algorithm, optimizing the resource utilization among internal processes.

[0074] S1044. Generate a factory matching priority sequence through factory equipment configuration and comprehensive capacity assessment, and optimize the cross-factory part allocation using the maximum flow algorithm.

[0075] In the embodiment of the present invention, obtain the production equipment configuration list from the factory resource database, summarize the equipment production capacity and real-time load data of each factory, and calculate the comprehensive capacity score of the factory through the analytic hierarchy process. For example, the main factory is equipped with 4 finish machining centers and 6 honing machines, with a score of 0.95; cooperative factory A is equipped with 3 finish machining centers and 4 honing machines, with a score of 0.88; cooperative factory B is equipped with 2 finish machining centers and 3 honing machines, with a score of 0.82. Generate a factory matching priority sequence according to the scores, and allocate cross-factory parts using the maximum flow algorithm. The main factory preferentially processes high-precision cylinder blocks, cooperative factory A undertakes medium-precision parts, and cooperative factory B processes ordinary-precision parts. Combining the principle of proximity allocation, optimize the transportation path to ensure the cost-effectiveness of cross-factory allocation. This step realizes the efficient collaboration of cross-factory resources through comprehensive capacity assessment and the maximum flow algorithm.

[0076] S105. Combine the factory real-time processing load, the target part grouping scheme, and the priority sequence to generate a reprocessing flow configuration scheme, and identify the balance point between resource consumption and reprocessing coverage.

[0077] In the embodiment of the present invention, generate a reprocessing flow configuration scheme by integrating the factory real-time processing load data, the target part grouping scheme, and the priority sequence, and use simulation technology and optimization algorithms to analyze the balance point between the resource consumption rate and the reprocessing coverage. The embodiment of the present invention does not strictly limit the data collection frequency or algorithm parameters, and can be flexibly adjusted according to actual production requirements.

[0078] S1051. Obtain the operation data of process equipment from the manufacturing execution database, divide the load intervals, and use the random forest algorithm to extract the correlation between the load characteristics and the priority sequence to generate a reprocessing flow path table.

[0079] In an embodiment of the present invention, the start-up rate and production capacity utilization rate data of process equipment are extracted from the manufacturing execution database. The sliding average method is used to calculate the real-time load status, and the process load is divided into a high load interval above 80%, a medium load interval of 60% to 80%, and a low load interval below 60% according to the production capacity utilization rate. Taking the machining of an automotive engine cylinder block as an example, the start-up rate of the rough machining process is 85%, and the production capacity utilization rate is 75%, belonging to the medium load interval. The start-up rate, production capacity utilization rate, and maintenance interval duration of the process equipment are extracted as characteristic parameters and input into the random forest algorithm to calculate the correlation score between the priority sequence and the load characteristics. For example, the correlation degree of rough machining is 0.82, that of finish machining is 0.88, and that of honing is 0.75. The correlation degree threshold is set to 0.80, and the rough machining and finish machining are selected as key processes. According to the process sequence constraint, a reprocessing flow path table is generated to ensure that the rough machining is executed prior to the finish machining. This step provides accurate input for the simulation through feature extraction and path planning.

[0080] S1052. Use a discrete event simulator to construct a reprocessing flow scenario, calculate the resource consumption rate and coverage rate curve, and determine the balance point parameters through piecewise regression analysis.

[0081] In an embodiment of the present invention, a discrete event simulator is adopted. Based on the reprocessing flow path table, process parameters such as machining cycle time and line change time are input to construct a virtual production scenario. Taking the cylinder block machining as an example, the rough machining cycle time is set to 15 minutes, and the line change time is 30 minutes; the finish machining cycle time is 20 minutes, and the line change time is 45 minutes. After the simulation runs for 8 hours, the rough machining center accumulatively runs for 420 minutes, and the standard running time is 480 minutes, with a resource consumption rate of 87.5%; the finish machining center accumulatively runs for 400 minutes, and the resource consumption rate is 83.3%. During the same period, 28 pieces are completed in rough machining, and the planned number is 30 pieces, with a coverage rate of 93.3%; 24 pieces are completed in finish machining, and the coverage rate is 80%. By using the piecewise regression method to fit the resource consumption rate curve and the coverage rate curve, it is found that the values of both curves are close to 85% at the intersection point at the 6th hour. The intersection point parameters are extracted: the rough machining cycle time is adjusted to 18 minutes, the finish machining cycle time is 22 minutes, and the line change time is shortened to 40 minutes. This step accurately locates the balance point between resource consumption and coverage rate through simulation and regression analysis.

[0082] S1053. Use the k-means clustering algorithm to classify the configuration scheme, calculate the resource consumption and coverage rate indicators based on the equipment operation records, and optimize the configuration scheme through the genetic algorithm.

[0083] In the embodiment of the present invention, process parameters, equipment configuration, and processing rhythm in the reprocessing flow configuration plan are extracted, and the Euclidean distance between the plans is calculated using the k-means clustering algorithm. The plans are divided into a high similarity group, a medium similarity group, and a low similarity group. For example, plans with a distance less than 0.2 are classified into the high similarity group. Equipment operation records are obtained from the production execution database, and equipment utilization rate and resource consumption indicators are calculated. Taking 8 hours per single shift as the standard, a certain equipment operates for 6.5 hours, and the utilization rate is 81.25%; the cumulative operation is 240 hours, the maintenance interval is 48 hours, and the resource consumption indicator is 5. It is planned to reprocess 100 cylinder blocks, and 85 are completed, with a coverage rate of 85%. If the resource consumption indicator exceeds 1.2 times the historical average of 4.8, that is, 5.76, the beat time is preferentially adjusted, for example, from 15 minutes to 18 minutes. If it still exceeds the threshold, one more device is added, and the production capacity redundancy is calculated to be 6%. Genetic algorithm optimization is adopted, and the production capacity utilization rate of 85% is set as the fitness function. After 200 iterations, the optimal plan is obtained: 2 devices, a beat of 16 minutes, and the resource consumption indicator drops to 5.5. This step ensures the efficiency of the configuration plan through clustering and optimization.

[0084] S1054. Fit the relationship between resource consumption and coverage rate through linear regression, determine the balance point, verify the parameter stability, and generate an optimized configuration plan.

[0085] In the embodiment of the present invention, resource consumption indicators and coverage rate indicators are extracted from the optimized configuration plan, and linear regression algorithms are used for fitting to generate a resource consumption curve y = 0.8x + 1.2 and a coverage rate curve y = -0.6x + 4.5. The intersection coordinates 2.3, 3.0 are calculated by the least squares method, indicating that when the resource consumption indicator is 2.3, the coverage rate reaches 90%. The cross-validation method is adopted, and 50 groups of historical configuration plans are randomly selected to verify the stability of the balance point. The results show that the resource consumption of 90% of the samples fluctuates within ±0.2, and the coverage rate fluctuates within ±5%

[0086] inside. Based on the balance point parameters, an optimized configuration plan is generated: 2 devices, a processing beat of 16 minutes, and a line change time of 30 minutes. This step ensures the stability and repeatability of the configuration plan through regression analysis and verification, providing reliable support for efficient reprocessing.

[0087] S106. Calculate the real-time performance ratio of reprocessing coverage rate to resource consumption based on the balance point data, compare it with the historical optimal performance ratio, identify the risk of performance bottlenecks and analyze their causes, and optimize the reprocessing process.

[0088] In the embodiment of the present invention, by using the reprocessing data corresponding to the balance point, the real-time performance ratio of the coverage rate and resource consumption is calculated and compared with the historical optimal value. If it is lower than the warning threshold, the potential performance bottleneck risk is identified. Through equipment status analysis and risk classification, the root cause of the abnormality is traced, providing a basis for optimizing the reprocessing process. This step significantly improves the production stability in high defect rate scenarios through a data-driven risk warning mechanism, combined with multi-dimensional feature analysis and historical data verification. The embodiment of the present invention does not strictly limit the data collection method or threshold setting, which can be adjusted according to the actual production environment.

[0089] S1061. Extract the balance point data from the production execution database, calculate the coverage rate and resource consumption rate, generate the real-time performance ratio, and compare it with the historical optimal value.

[0090] In the embodiment of the present invention, the reprocessing data corresponding to the balance point is obtained from the production execution database. The coverage rate is calculated by dividing the number of completed parts by the planned quantity, and the resource consumption rate is calculated by dividing the equipment operation duration by the standard working hours. The real-time performance ratio is obtained by dividing the coverage rate by the resource consumption rate. Taking the processing of automobile engine cylinder blocks as an example, for a certain batch, 100 cylinder blocks are planned for reprocessing, and 85 are actually completed, with a coverage rate of 85%; the standard working hours are 8 hours, and the actual operation is 6.8 hours, with a resource consumption rate of 85% and a performance ratio of 1.0. The production records of the past 30 days are extracted from the historical database, and the samples with the top 10% performance ratio are selected, with an average optimal value of 1.25. If the real-time performance ratio is lower than the historical average of 1.05, further analysis is triggered. This step quantifies the reprocessing efficiency through the performance ratio, providing a data basis for risk identification.

[0091] S1062. Establish a performance prediction model using the support vector regression method, set the warning threshold, mark the risk level, and extract the equipment status feature vector.

[0092] In the embodiment of the present invention, based on the historical production records, a performance prediction curve is constructed using the support vector regression method. The input parameters include the daily performance ratio, equipment operation duration, and maintenance interval. The prediction curve reflects the normal fluctuation range of the performance ratio, and accordingly, the warning threshold is set: less than 0.9 is a mild warning, less than 0.85 is a moderate warning, and less than 0.8 is a severe warning. For example, when the performance ratio of a machining center drops to 0.82, a moderate warning is triggered. The operation parameters are extracted from the equipment management database, including the machining accuracy of 0.92 being lower than the normal value of 0.95, the continuous operation duration of 180 hours exceeding the maintenance interval of 160 hours, and the remaining tool life of 10 hours approaching the replacement cycle. After normalization processing, the parameters generate the equipment status feature vector. This step accurately locates the risk status through the prediction model and threshold classification.

[0093] S1063. Use hierarchical clustering to analyze the device status feature vectors, identify abnormal parameter combinations, and mine the risk causes through association rules.

[0094] In the embodiment of the present invention, the hierarchical clustering method is used to group the device status feature vectors, calculate the Euclidean distance between classes, and identify abnormal parameter combinations. If the feature vector of a certain machining center reaches 0.35 from the normal working condition, exceeding the threshold of 0.3, it is marked as abnormal. Further, the association rule algorithm is used to calculate the confidence level of parameter abnormality and performance degradation. For example, the confidence level of continuous operation duration exceeding the limit and performance degradation is 85%, and the confidence level of insufficient tool life and machining accuracy degradation is 90%. The abnormal parameter combinations are classified into device accuracy classes such as dimensional tolerance, device life classes such as spindle wear, and device maintenance classes such as abnormal lubricating oil pressure, generating a parameter abnormality feature table. This step clarifies the source of the risk causes through clustering and association analysis.

[0095] S1064. Build a risk early warning model based on the decision tree algorithm, verify the risk classification, and generate a risk cause report to guide the device maintenance decision.

[0096] In the embodiment of the present invention, a risk early warning model is built using the decision tree algorithm, and the abnormal parameter feature table and historical failure data are input to classify and verify the risk types. For example, a certain machining center has both device life risks such as tool wear and maintenance risks such as insufficient coolant concentration. Historical data shows that similar parameter combinations have caused batch quality problems. Generate a risk cause report, and recommend immediately maintaining the device, replacing the tool, and adjusting the coolant parameters to restore the machining accuracy to above 0.95. This step provides accurate guidance for maintenance decisions through decision tree verification and report generation, avoiding the expansion of quality problems.

[0097] S107. Dynamically adjust the part grouping strategy and factory matching priority according to the risk cause analysis results, optimize the reprocessing flow configuration, and through real-time monitoring of the defect rate and performance curve changes, cyclically verify the reduction of resource consumption and the improvement of coverage rate, confirm the stable state of the optimization plan, and generate an executable improvement plan.

[0098] In the embodiment of the present invention, based on the analysis results of the risk cause database, the part grouping strategy and factory matching priority are dynamically updated, generating an optimized reprocessing flow configuration plan, and through real-time data monitoring and cyclic verification, ensuring the reduction of resource consumption while improving the reprocessing coverage rate, and finally forming a stable improvement plan, realizing the intelligent optimization of the production process in high defect rate scenarios, and significantly improving the production efficiency and resource utilization rate. The embodiment of the present invention does not strictly limit the monitoring period or threshold standard, and can be flexibly adjusted according to actual production needs.

[0099] S1071. Extract the equipment status and performance degradation records from the risk cause database, set the risk threshold, use the deep learning algorithm to predict the part risk classification, and update the part grouping strategy.

[0100] In the embodiment of the present invention, obtain the equipment status characteristics and performance degradation records from the risk cause database, and set the risk threshold based on historical data: a performance degradation exceeding 10% is a high risk, 5% to 10% is a medium risk, and within 5% is a low risk. Taking the machining of an automotive engine cylinder block as an example, a certain machining center has been continuously operating for 180 hours beyond the maintenance period, and the machining accuracy has dropped to 0.92, which is lower than the standard value of 0.95, triggering a high-risk warning. Using the deep learning algorithm, with the equipment operation duration, machining accuracy, and maintenance interval as inputs, predict the risk distribution of 100 cylinder blocks in a certain batch. The results are 35 high-risk, 45 medium-risk, and 20 low-risk, and generate an updated part grouping table.

[0101] S1072. Calculate the comprehensive score of the factory based on the equipment operation status, machining accuracy, and maintenance cycle, divide the equipment priority levels, and optimize the process parameters to maximize the product of the load rate and the production capacity utilization rate.

[0102] In the embodiment of the present invention, score the equipment operation status, set the operation status weight at 0.4, the machining accuracy weight at 0.4, and the maintenance cycle weight at 0.2, and calculate the comprehensive score of the factory through weighted summation. The operation status score is based on the continuous operation duration, failure interval, and maintenance plan execution rate; the machining accuracy score includes the dimension qualification rate, surface roughness, and geometric accuracy; the maintenance cycle score covers the equipment life, lubrication status, and cleanliness. For example, the comprehensive scores of 5 machining centers in a certain workshop are 0.92, 0.88, 0.85, 0.82, and 0.78 respectively. Classify the equipment with the top 30% of the scores, that is, 0.92 and 0.88, into the high-priority group, the middle 40% of 0.85 and 0.82 into the medium-priority group, and the rest into the low-priority group. For high-priority equipment, use the genetic algorithm to optimize the process parameters, with the product of the equipment load rate and the production capacity utilization rate as the objective function. For example, the original cycle time of a certain equipment is 15 minutes, the load rate is 85%, and the utilization rate is 80%. After optimization, the cycle time is adjusted to 18 minutes, the load rate drops to 75%, and the utilization rate rises to 85%, and the product increases from 0.68 to 0.74.

[0103] S1073. Verify the optimized process parameters, generate new machining cycle times and line change cycles, and evaluate the defect rate and performance curve changes through real-time monitoring.

[0104] In the embodiment of the present invention, the optimized process parameters are verified through the production capacity calculation module to evaluate the production capacity matching degree among various processes. For example, the optimized daily production capacities of 5 devices are 26 pieces, 24 pieces, 22 pieces, 20 pieces, and 18 pieces respectively, and the production capacity difference among processes is controlled within 10%. The line change cycle is adjusted from 4 hours to 6 hours, the line change frequency is reduced, and the production continuity is improved. The production data is monitored using a 4-hour time window. The defect rate in the first cycle is reduced from 4.2% to 3.5%, and the performance is improved by 15%; the defect rate in the second cycle is 3.3%, and the performance is improved by 18%; the defect rate in the third cycle is 3.4%, and the performance is improved by 16%. This step ensures the effectiveness of the optimization plan through real-time monitoring and parameter verification.

[0105] S1074. Calculate the reduction ratio of resource consumption and the increase ratio of coverage rate, confirm the stable state, and generate an improvement plan verification report through cross-verification.

[0106] In the embodiment of the present invention, it is calculated that the resource consumption after optimization is reduced by 12% and the coverage rate is increased by 16%, and it is verified whether the index fluctuation is less than 5% within 3 consecutive monitoring cycles. The results show that the fluctuations in each cycle do not exceed the threshold, and it is confirmed that the production enters a stable state. Using the cross-verification method, 10 groups of historical optimization plans are randomly selected for comparison, and the index fluctuation range of the current plan is better than the historical average level. The optimized parameter combination includes a beat of 18 minutes and a line change cycle of 6 hours, which is recorded in the improvement plan library for reference in subsequent production. This step ensures the reliability and repeatability of the optimization plan through stability and verification analysis.

[0107] Obviously, those skilled in the art can make various changes and modifications to the embodiments of the present application without departing from the spirit and scope of the embodiments of the present application. Thus, if these modifications and variations of the embodiments of the present application fall within the scope of the claims of the present application and their equivalent technologies, the present application is also intended to include these changes and modifications.

Claims

1. A reprocessing decision-making method for defective parts based on multi-objective balance, characterized in that The method includes: Obtain the real-time defect rate and part distribution information, set a standard defect threshold. If the defect rate exceeds the standard defect threshold, identify and fit the part performance parameters to obtain the average performance curve, calculate the slope change and mutation amplitude of the curve, and determine the time and influence range of the mutation point; Based on the defect rate and performance curve, perform dynamic clustering on the parts to obtain part groups, determine the target part group. The parts include internal parts and cross-factory parts. Allocate the internal parts according to the processing capacity and defect rate, and allocate the cross-factory parts according to the geographical location, cost, processing capacity, and real-time defect rate; Identify the resource consumption of the factory after implementing the target part group, allocate the internal parts according to the load, and combine the historical data to obtain the factory matching priority, and allocate the cross-factory parts accordingly; Obtain the processing load, combine the priority allocation order with the part grouping scheme and the load to obtain the reprocessing process configuration scheme, and simulate the reprocessing process configuration scheme to identify the balance point between resource consumption and coverage; Obtain the coverage rate and resource consumption ratio corresponding to the balance point, compare it with the historical optimal ratio. If it is worse than the historical optimal value, determine that there is a risk of system performance bottleneck and analyze the risk causes; Dynamically update the part grouping strategy and factory matching priority according to the risk causes to obtain the optimized reprocessing process configuration, obtain the optimized defect rate change and performance curve, verify whether the consumption is reduced and the coverage rate is increased, judge the stable state, and form an executable improvement plan.

2. The method according to claim 1, characterized in that, The obtaining the real-time defect rate and part distribution information, and setting the standard defect threshold includes: Obtain the part position information and quality record form data, and obtain the defect station data set by associating with the part traceability code; Perform station number classification statistics on the defect station data set, and obtain the station real-time defect rate data by dividing the number of defects by the total production volume; Use the automatic detection device to obtain the station detection data, and calculate the defect detection rate index through the Pearson correlation coefficient; Weight the defect rate data according to the defect detection rate index, obtain the defect rate change trend curve through the exponential smoothing method, and identify the outlier interval of the defect rate through the box plot analysis method.

3. The method according to claim 1, characterized in that, The if the defect rate exceeds the standard defect threshold, identify and fit the part performance parameters to obtain the average performance curve, calculate the slope change and mutation amplitude of the curve, and determine the time and influence range of the mutation point includes: Obtain the measurement data of the part detection equipment and perform normalization processing to obtain the standardized performance parameter set; For the standardized performance parameter set, use the sliding window method to calculate the parameter mean and standard deviation, and eliminate the abnormal data points according to the mean and standard deviation according to the three-standard-deviation principle to obtain the performance parameter time series; According to the performance parameter time series, calculate the fitting curve equation by the least squares method, and calculate the parameter change rate of adjacent time points according to the fitting curve equation to obtain the parameter slope sequence; For the parameter slope sequence, judge whether the slope difference between adjacent two points exceeds the preset threshold. If it exceeds the preset threshold, mark it as a mutation point, and use the clustering method to group the mutation points to obtain the start and end time points of the influence range.

4. The method according to claim 1, characterized in that The dynamic clustering of parts based on the defect rate and performance curve to obtain part groupings, determine the target part grouping. The parts include internal parts and cross-plant parts. Allocate internal parts according to processing capabilities and defect rates, and allocate cross-plant parts according to geographical location, cost, processing capabilities, and real-time defect rates, including: Calculate the Euclidean distance based on the slope value of the part performance curve and defect rate data, and obtain the mutant part identification table by comparing the Euclidean distance with the preset minimum distance threshold; For the part groups in the mutant part identification table, calculate the fluctuation amplitude of the performance curve using the curve fitting equation, and obtain the target part grouping list through the correlation analysis of the fluctuation amplitude and the real-time defect rate; Obtain the accuracy inspection records, operation duration records, and fault repair records of the processing equipment according to the target part grouping list, and calculate the equipment production capacity index through the recorded data; For the equipment production capacity index and real-time defect rate data, construct a process production capacity allocation network to obtain the internal part allocation plan.

5. The method according to claim 4, characterized in that It also includes: Obtain the real-time defect rate data of internal parts and the transportation costs of cross-plant parts, generate an initial part grouping according to the real-time defect rate data, identify the process processing capacity vectors in the initial part grouping according to the equipment status of each factory, generate a target internal part allocation plan according to the process processing capacity vectors, and perform data fusion on the target internal part allocation plan and the cross-plant part allocation plan to obtain the global part scheduling instruction set, specifically including: Obtain the part processing records according to the production execution database records. The part processing records contain quality inspection data and defect rate data, and obtain the part risk level grouping table through the three-layer grouping rules; Obtain the processing equipment operation status data using the equipment management database, and obtain the comprehensive equipment status score through the weighted calculation of three indicators: the proportion of continuous equipment operation duration, the qualified rate of precision inspection, and the maintenance interval duration; Construct a process processing capacity vector according to the comprehensive equipment status score. The process processing capacity vector includes processing accuracy indicators, processing cycle indicators, and equipment stability indicators, and obtain the process equipment status eigenvalue through the least squares method; Calculate the matching degree between the part processing requirements and the process processing capacity vector, and allocate process equipment corresponding to the part risk level grouping table to obtain the part processing allocation plan; Establish a correspondence table between internal parts and cross-plant parts through part code association, sort the parts according to the parts in the risk level correspondence table, generate a processing priority sequence according to the part risk level sorting result, and generate a process scheduling time table corresponding to the priority sequence to form a global part scheduling instruction data set.

6. The method according to claim 1, characterized in that, Identify the resource consumption of the factory after executing the target part grouping, allocate internal parts according to the load, and obtain the factory matching priority in combination with historical data, and allocate cross-plant parts accordingly, including: Obtain the process scheduling data and equipment operation records from the production management platform, and obtain the equipment load rate data by calculating the number of on-line processed parts divided by the standard production capacity of the equipment; According to the equipment load rate data and the process sequence specified in the process specification, calculate the upper limit value and lower limit value of the equipment production capacity using the linear programming method; The actual operation cycle time record of the process is processed by the weighted average method to obtain the standard processing cycle of the process, and the process processing capacity score is generated according to the deviation value between the standard processing cycle of the process and the actual operation cycle time record of the process. Multi-process allocation is performed on internal parts, and the priority allocation order is determined according to the ratio of the unit processing duration of the parts to the remaining production capacity of the process.

7. The method according to claim 1, characterized in that, The processing load is obtained, and the priority allocation order is combined with the part grouping scheme and the load to obtain the reprocessing flow configuration scheme. By simulating the reprocessing flow configuration scheme, the balance point of the resource consumption rate and the coverage rate is identified, including: The operation rate data and the production capacity utilization rate data of the process equipment in the manufacturing execution database are obtained, and the high-load interval data, the medium-load interval data, and the low-load interval data are obtained by dividing according to the production capacity utilization rate data. The random forest algorithm is used to extract features from the operation rate data of the process equipment, the production capacity utilization rate data, and the maintenance interval duration data to obtain the correlation score between the priority sequence and the load characteristics. An association threshold is set according to the correlation score, and the matching process data is screened through the association threshold. A reprocessing flow path table is generated based on the processing sequence constraint between processes. A discrete event simulator is used to construct a reprocessing flow scenario. The reprocessing flow path table is input into the simulation environment. The resource consumption rate curve is obtained by dividing the cumulative running time by the standard running time, and the coverage rate curve is obtained by dividing the number of completed parts by the planned number. The intersection parameters of the resource consumption rate curve and the coverage rate curve are calculated by using the piecewise regression method.

8. The method according to claim 7, wherein It also includes: The reprocessing flow configuration schemes are classified to obtain a set of configuration schemes. The resource consumption data and the reprocessing coverage rate of different configuration schemes are obtained from the set of configuration schemes. If the resource consumption data exceeds the preset consumption threshold, the configuration scheme is adjusted to obtain an optimized set of configuration schemes. According to the optimized set of configuration schemes, the linear regression algorithm is used to predict the relationship between the resource consumption and the reprocessing coverage rate to obtain the balance point data, specifically including: The process parameters, equipment configuration, and processing cycle in the reprocessing flow configuration scheme are obtained, and the configuration scheme grouping table is obtained by using the k-means clustering algorithm according to the process parameters. The equipment operation records are obtained from the production execution database according to the configuration scheme grouping table, and the equipment utilization rate and the resource consumption index are calculated by using the equipment operation records and the standard working hours of a single shift. The optimization calculation of the equipment utilization rate is performed, and the optimized configuration scheme is obtained with the constraint that the resource consumption index is less than the preset threshold. The resource consumption index and the coverage rate index are extracted for the optimized configuration scheme, and the resource consumption index and the coverage rate index are fitted to obtain the resource consumption value and the coverage rate value corresponding to the balance point.

9. The method according to claim 1, wherein The ratio of the coverage rate and the resource consumption corresponding to the balance point is obtained and compared with the historical optimal ratio. If it is worse than the historical optimal value, it is determined that there is a risk of system performance bottleneck, and the risk causes are analyzed, including: Obtain the ratio of the number of completed parts to the planned number to get the coverage rate value, obtain the ratio of the equipment operation duration to the standard man-hour to get the resource consumption value, and divide the coverage rate value by the resource consumption value to get the real-time performance ratio; Extract production records from the historical database, and establish a performance prediction curve using the support vector regression method based on the production records; If the real-time performance ratio is lower than the warning threshold, mark it as a risk warning state, and extract operation parameters from the equipment management database according to the warning state to obtain the equipment status feature vector; Use the hierarchical clustering method to group the equipment status feature vectors, and identify abnormal parameter combinations by calculating the inter-class distance to obtain the parameter anomaly feature table.

10. The method according to claim 1, characterized in that Dynamically update the part grouping strategy and factory matching priority according to the risk causes, obtain the optimized reprocessing process configuration, obtain the optimized defect rate change and performance curve, verify whether the consumption is reduced and the coverage rate is increased, judge the stable state, and form an executable improvement plan, including: Obtain the equipment status feature records and performance reduction records from the risk cause database, and set high-risk thresholds, medium-risk thresholds, and low-risk thresholds according to historical performance data; According to the equipment status feature records and the performance reduction records, use the deep learning algorithm to obtain the part risk prediction classification results; For the equipment operation status records, calculate the comprehensive factory score by using the status weight value, accuracy weight value, and maintenance cycle weight value; Perform a three-level priority division on the equipment according to the comprehensive factory score, and calculate the optimized value of the process parameters that maximizes the product of the equipment load rate and the production capacity utilization rate.

Images