Thermal fuse production quality management method and system

Abstract

The invention relates to the technical field of thermal fuse production quality management and control, and particularly discloses a thermal fuse production quality management method and system, and the method comprises the steps: carrying out the deviation analysis of historical sampling data and real data, associating historical defect positioning cases, judging whether there is a sampling deviation judgment-defect positioning imbalance problem, and if there is a sampling deviation judgment-defect positioning imbalance problem, judging whether there is a defect positioning imbalance problem; the method comprises the following steps: dividing stable and fluctuation sub-batches according to equipment parameter stable time periods, extracting and clustering fluctuation sub-batch parameter fluctuation characteristics, calculating a fluctuation coefficient of each fluctuation sub-batch group, constructing a dynamic sampling proportion mapping model based on the fluctuation coefficients, realizing stratified sampling, accurately calculating a reject ratio according to scenes, and determining a defect positioning range. The defect rate estimation accuracy and the defect positioning decision scientificity are improved, the excessive defect positioning cost and the defect positioning insufficiency risk are reduced, and accurate defect positioning and quality closed-loop management and control of the whole process of temperature fuse production are achieved.

Description

Technical Field

[0001] This invention relates to the field of quality control technology for the production of thermal fuses, and specifically to a method and system for quality management of the production of thermal fuses. Background Technology

[0002] As a core safety component in electronic devices, home appliances, and automotive electronics, thermal fuses directly determine the safety of end products through their critical performance characteristics, such as melting temperature accuracy and overload withstand capability. Failure can lead to serious risks such as fires and equipment damage. Therefore, the critical performance verification of thermal fuses must be achieved through destructive testing, such as heating to melting and overload impact. This characteristic makes it impossible for the industry to conduct full inspection of all products; quality control can only be achieved through sampling inspection, with a typical sampling rate of only 5%-10%.

[0003] However, sampling inspection of thermal fuses has limitations: on the one hand, a qualified sample cannot fully represent the quality of the entire batch of products, because the sample may not cover high-risk areas such as raw material switching transitions and periods of equipment parameter fluctuations. When a few defective products appear at the end, it is impossible to accurately determine whether they are occasional individual defects such as impurities in a single raw material or systemic batch defects such as raw material batch defects or process fluctuations. On the other hand, when defects are found during sampling, due to the lack of an accurate method for estimating the defect rate of the entire batch of products, it is impossible to determine whether the defect range is a local sub-batch or the entire batch. The defect location range can only be blindly expanded according to the production time period and equipment number. This can either lead to insufficient defect location due to incomplete coverage, causing safety risks at the end, or cause a large number of qualified products to be scrapped due to excessive defect location, significantly increasing production costs.

[0004] In existing technologies, the production positioning of thermal fuses is limited to the level of basic data recording. There is no correlation model between sampling data and actual quality status, no quantitative correction mechanism for sampling deviation, and no accurate sampling and defect location decision based on risk stratification. This leads to a dilemma of imbalance between sampling bias and defect location, which seriously restricts the quality control level and market competitiveness of thermal fuse manufacturers.

[0005] Therefore, the present invention provides a method and system for production quality management of thermal fuses. Summary of the Invention

[0006] The purpose of this invention is to provide a production quality management method and system for thermal fuses to solve the aforementioned background problems.

[0007] The objective of this invention can be achieved through the following technical solution: a method for quality management in the production of thermal fuses, characterized by comprising the following steps: Deviation analysis is performed on the historical defect location range and the actual defect range of defective thermal fuse products to identify unbalanced defect locations. By comparing and analyzing the sampling data corresponding to the unbalanced defect locations with the actual data, sampling unbalanced defect locations are identified. Statistical analysis is performed on the sampling unbalanced defect locations to determine whether there is an imbalance in sampling defect locations. If present, the current batch of temperature fuses is divided into stable sub-batches and fluctuating sub-batches according to the stable period of equipment parameters. The equipment parameter fluctuation characteristics of each fluctuating sub-batch are extracted and clustered to obtain multiple fluctuating sub-batch groups. The fluctuation coefficient of each fluctuating sub-batch group is calculated. A dynamic mapping model of fluctuation coefficient and sampling ratio is constructed based on the fluctuation coefficient of each fluctuation sub-batch group, and the sampling ratio of stable sub-batch and each fluctuation sub-batch group is calculated for stratified sampling detection. If defective products are found during sampling inspection, the actual defect rate is calculated using the stratified confidence interval method based on the sampling defect rate, and the defect location range is determined. If defective products are not found during sampling inspection, but are found at the terminal, the defect location range is determined through analysis of related sub-batches.

[0008] Furthermore, the identification method for locating the sampling imbalance defect is as follows: For any historical defect location: If new defective products still exist in the terminal of the unlocated area after defect location, it is considered insufficient defect location. If the actual defect rate of the products with defect location does not meet the requirements, it is considered excessive defect location. Defect location with insufficient or excessive defect location will be marked as unbalanced defect location. If defect localization is insufficient and the sampling rate and defect range are judged too low, or if defect localization is excessive and the sampling rate and defect range are judged too high, then the unbalanced defect localization is called sampling imbalance defect localization.

[0009] Furthermore, the sampling method for determining the defect rate and defect range is as follows: The sampling defect rate is calculated by proportionally dividing the number of defective samples by the sample size. The true defect rate is calculated by proportionally dividing the actual total number of defective products by the total number of historical batches. If the sampling defect rate is less than the actual defect rate, then the sampling rate underestimates the defect rate; conversely, if the sampling rate is greater than the actual defect rate, then the sampling rate overestimates the defect rate. The over-coverage rate is obtained by calculating the ratio of the portion of the defect location range without actual defects to the total number of defect location ranges in the sampling judgment. If the over-coverage rate meets the requirements, the sampling will overestimate the defect range; conversely, if the over-coverage rate does not meet the requirements, the sampling will underestimate the defect range.

[0010] Furthermore, the method for determining whether there is a sampling defect location imbalance is as follows: The proportion of sampling imbalance defect locations within all historical defect locations is calculated in the historical period. If the proportion of sampling imbalance defect locations meets the requirements, then there is a sampling imbalance phenomenon.

[0011] Furthermore, the method for dividing the stable sub-batch and the fluctuating sub-batch is as follows: Real-time monitoring of equipment parameters during the production process of temperature fuses; marking sub-batches produced during periods of stable equipment parameters as stable sub-batches, and marking sub-batches produced during periods of fluctuating equipment parameters as fluctuating sub-batches. The criteria for determining a stable period of equipment parameters are that both the fluctuation and duration of the equipment parameters meet the requirements. When the fluctuation of any equipment parameter does not meet the requirements, it is considered a period of equipment parameter fluctuation.

[0012] Furthermore, the process of extracting the parameter fluctuation features of each fluctuation sub-batch and performing clustering is as follows: For all fluctuating sub-batches, extract the fluctuation characteristics of the equipment parameters: fluctuation amplitude, fluctuation frequency, drift rate, and duration; The parameter fluctuation characteristics of the fluctuation sub-batch are standardized and used as clustering features to perform clustering, ultimately resulting in multiple fluctuation sub-batch groups.

[0013] Furthermore, the calculation process for the volatility coefficient of each volatility sub-batch group is as follows: For any one of the parameters—fluctuation amplitude, fluctuation frequency, drift rate, and duration—the fluctuation characteristics are as follows: The defect rate when the historical parameter fluctuation characteristics are abnormal is statistically analyzed. The defect rate of the parameter fluctuation characteristics is normalized and used as a weight. For any sub-batch in any fluctuation sub-batch group, the degree of fluctuation is obtained by weighting. For any fluctuation sub-batch group, the average fluctuation degree of each fluctuation sub-batch within the fluctuation sub-batch group is calculated as the fluctuation coefficient of the fluctuation sub-batch group.

[0014] Furthermore, the construction process of the fluctuation coefficient-sampling ratio dynamic mapping model and the calculation method of the sampling ratio are as follows: Calculate the risk contribution of stable sub-batch and fluctuating sub-batch separately, which is the product of defect rate and product quantity; The total sampling proportion of the stable sub-batch is obtained by multiplying the fixed overall proportion by the risk contribution of the stable sub-batch, and the total sampling proportion of the volatile sub-batch is obtained by multiplying the fixed overall proportion by the risk contribution of the volatile sub-batch. For stable sub-lots, the total sampling proportion of stable sub-lots is evenly distributed to each stable sub-lot; For any fluctuation sub-batch group: The sampling proportion of the fluctuation coefficient is calculated by proportionally to the sum of the fluctuation coefficients of all fluctuation sub-batches, and then multiplied by the total sampling proportion of the fluctuation sub-batches to obtain the sampling proportion of the fluctuation sub-batch group. The degree of fluctuation of each fluctuation sub-batch within the fluctuation sub-batch group is used as the sampling proportion coefficient. The sampling proportion of each fluctuation sub-batch is obtained by multiplying the sampling proportion of the fluctuation sub-batch group by the sampling proportion coefficient of each fluctuation sub-batch.

[0015] Furthermore, the method for determining the defect location range is as follows: If defective products are found during sampling inspection, the ratio of the number of defective products to the sample size is calculated to obtain the basic sampling defect rate. The defective product belongs to the sub-batch. If the defective product belongs to a stable sub-batch, the basic sampling defect rate is the actual defect rate, and the defect location range is the stable sub-batch to which the defective product belongs. If the defective product belongs to a fluctuating sub-batch, for any fluctuating sub-batch group: Calculate the ratio of the actual defect rate to the sampling defect rate of all fluctuating sub-batches in each fluctuating sub-batch group within the historical period, and take the average value as the historical deviation coefficient. The risk correction factor K is obtained by adding the relative weights to the historical deviation coefficients; The defect rate of the entire fluctuation sub-batch is estimated by using the stratified confidence interval method combined with the risk correction factor. The defect location range is all fluctuation sub-batches in the fluctuation sub-batch group. If the sampling inspection is qualified, but the terminal is defective, only the defective product belongs to the fluctuating sub-batch and the adjacent sub-batch in the same fluctuating sub-batch group.

[0016] A production quality management system for thermal fuses includes the following modules: Defect Location Imbalance Judgment Module: Analyzes the deviation between the historical defect location range and the actual defect range of defective temperature fuse products, identifies imbalanced defect locations, and identifies sampling imbalanced defect locations by comparing and analyzing the sampling data corresponding to the imbalanced defect locations with the actual data. Statistical analysis is performed on the sampling imbalanced defect locations to determine whether there is a sampling defect location imbalance phenomenon. Sub-batch segmentation module: If it exists, the current batch of temperature fuses is divided into stable sub-batches and fluctuating sub-batches according to the stable period of equipment parameters. The fluctuation characteristics of equipment parameters of each fluctuating sub-batch are extracted and clustered to obtain multiple fluctuating sub-batch groups. The fluctuation coefficient of each fluctuating sub-batch group is calculated. Sampling and inspection module: Construct a dynamic mapping model of fluctuation coefficient-sampling ratio based on the fluctuation coefficient of each fluctuation sub-batch group, and calculate the sampling ratio of stable sub-batch and each fluctuation sub-batch group to perform stratified sampling and inspection; Defect localization range determination module: If sampling inspection finds defective products, the actual defect rate is calculated using the stratified confidence interval method based on the sampling defect rate, and the defect localization range is determined. If sampling inspection does not find defective products, but defective products are found at the terminal, the defect localization range is determined through analysis of associated sub-batch.

[0017] The beneficial effects of this invention are as follows: By analyzing historical data deviations, refining sub-batch divisions, clustering fluctuations, mapping dynamic sampling ratios, and processing results across different scenarios, the system systematically solves the sampling bias problem caused by destructive testing of temperature fuses, achieving accurate determination of defect ratios, ranges, and defect types, and completely eliminating the dilemma of insufficient or excessive defect location. By integrating the current risk weights of fluctuating sub-batches with historical sampling bias experience through risk correction factors, and combining them with the stratified confidence interval method, the error in defect rate estimation is reduced, providing accurate data support for defect location decisions. Within fluctuating sub-batches, sampling is stratified by risk, while stable sub-batches are reduced to release resources. The sampling ratio in high-risk areas is increased. This reduces the cost of destructive testing while ensuring the defect detection rate. It also accurately locks the defect location range to the sub-batch level, avoids blind defect location in large batches, reduces the rate of over-defect location, and eliminates the terminal security risks caused by insufficient defect location. Attached Figure Description

[0018] The invention will now be further described with reference to the accompanying drawings.

[0019] Figure 1 This is a flowchart of a production quality management method for a temperature fuse as described in Embodiment 1 of the present invention; Figure 2 This is a flowchart of the process for determining whether there is a sampling defect location imbalance in Embodiment 1 of the present invention; Figure 3 This is a functional block diagram of a production quality management system for a temperature fuse according to Embodiment 2 of the present invention; Detailed Implementation

[0020] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0021] Example 1: Please refer to Figure 1As shown in the embodiment of the present invention, a production quality management method for thermal fuses addresses the defect range localization of thermal fuses. Through deviation analysis between historical sampling data and actual data, and by associating historical defect localization cases, it determines whether there is an imbalance between sampling bias and defect localization. If so, it divides the equipment into stable and fluctuating sub-batches according to the stable period of the equipment parameters, extracts the fluctuation characteristics of the fluctuating sub-batches and clusters them, calculates the fluctuation coefficient of each fluctuating sub-batch group, and constructs a dynamic sampling ratio mapping model based on the fluctuation coefficient to achieve stratified sampling. It accurately calculates the defect rate and determines the defect localization range according to different scenarios: when sampling is defective, it combines a risk correction factor to estimate the defect rate; when sampling is qualified but the terminal is defective, it determines the individual / batch defect. This improves the accuracy of defect rate estimation and the scientific nature of defect localization decisions, reduces the cost of excessive defect localization and the risk of insufficient defect localization, and realizes full-process production quality management of thermal fuses. Specifically, it includes the following steps: Step 1: Analyze the deviation between the historical defect location range and the actual defect range of the faulty thermal fuse to identify the location of unbalanced defects. By comparing and analyzing the sampling data corresponding to the location of unbalanced defects with the actual data, identify the location of sampling unbalanced defects. Perform statistical analysis on the location of sampling unbalanced defects to determine whether there is an imbalance in the location of sampling defects. In step one, the identification process for locating the imbalance defect includes: Obtain historical manufacturing defect location data for thermal fuses, including: Sampling data: Sampling time, sampling quantity, production period covered by the sampling, and number of defects for each historical batch. Real data: Production period, packaging batch, raw material batch, total number of actual defects, and total number of products in each historical batch of defective end products; Defect location data: Sampling judgment criteria for each defect location case (e.g., sampling defect rate of 3%, defect judgment range is the whole batch), defect location range (e.g., defect location of the whole batch / defect location of a part of the time period), number of defects located, full inspection defect rate after defect location, and number of new defects in terminals in areas where defects were not located. Production data: Raw material batches for each historical batch (solids by packaging, liquids by silo feeding), and periods of equipment parameter fluctuations; The first point to clarify is that for any historical defect location, it is necessary to determine whether there is an imbalance in defect location (insufficient or excessive defect location), specifically as follows: Based on historical defect location data, the defect location problem types are defined as follows: if new defective products still exist in the unlocated areas after defect location, it is considered insufficient defect location; if the actual defect rate of the products with defect location is less than the threshold, it is considered excessive defect location. Defects that are under- or over-located are marked as unbalanced defect locations. Secondly, it should be noted that the accuracy of the sampling's assessment of the defect rate and scope is determined as follows: For any sampling corresponding to the location of an imbalance defect: To determine whether the sampling method accurately assesses the defect rate: The sampling defect rate is calculated by proportionally dividing the number of defective samples by the sample size. The true defect rate is calculated by proportionally dividing the actual total number of defective products by the total number of historical batches. Calculate the deviation between the sampling defect rate and the actual defect rate, and compare it with the preset deviation. If the deviation is greater than or equal to the preset deviation, the sampling is inaccurate in judging the defect rate. Understandably, the preset deviation is the critical value for determining whether the sampling is accurate in judging the proportion of defects, and it is set based on historical data. Determination of deviation direction: If the sampling defect rate is less than the actual defect rate, it is determined that the sampling rate is too low (not fully reflecting the actual risk). If the sampling defect rate is greater than the actual defect rate, it is determined that the sampling rate is too high (overestimating the risk). To determine whether the sampling method accurately identifies the defective range, the following steps are taken: Identify the production period, raw material batch, and packaging batch corresponding to the sampling to obtain the sampling coverage. For example, the core records of this sampling can be retrieved from historical sampling data, including: sampling time, sampling quantity, sampling operation record, and production status data at the time of sampling. Based on the sampling time, pipeline transfer delay, and sub-batch division rules, the production period covered by the sampling is deduced in reverse: The actual production period for the sampled products is 9:20-9:35 (the production time of the products in the temporary storage area = sampling time - transmission delay). Corresponding sub-batches: The sub-batches within this time period are S7 (9:20-9:30, stable sub-batches, no fluctuation in equipment parameters) and S8 (9:30-9:35, fluctuating sub-batches, slight fluctuations in annealing temperature during sampling, but not exceeding the threshold). The final production period covered by the sampling was determined to be 9:20-9:35 on [Date] (corresponding to sub-batches S7 and S8). Based on the production status records at the time of sampling, the batch of raw materials used in the sampling sample was identified: Based on the packaging batch information of the sample source in the sampling operation record, determine the packaging batches covered by the sampling. The sampling coverage is obtained through integration: The information will be integrated to form the complete coverage of this sampling: Production period: 9:20-9:35 on [Date] (corresponding to sub-batch S7 and S8); Raw material batches: Alloy wire A1 (A1-001), temperature-sensitive resin B1 (B1-004), ceramic shell C1 (C1-003); Packaging batch: P3; By locating defective products at the terminal and combining full-volume testing results, the production period, raw material batch, and packaging batch where the actual defective products are concentrated can be identified, thus determining the true range of defects. The coverage matching degree is obtained by calculating the ratio of the overlap between the sampling coverage and the actual defect range to the actual defect range. The proportion of the portion of the defect location range without actual defects in the sampling judgment to the total number of defect location ranges in the sampling judgment is calculated to obtain the over-coverage; The omission coverage rate is obtained by calculating the ratio of the portion of the true defect range that was not covered by the sampling to the total amount of the true defect range. If the coverage matching degree is less than the preset matching degree, and the over-coverage rate or the omission rate is greater than or equal to the preset coverage rate, then the sampling will not accurately determine the poor range. Understandably, the preset coverage rate is a critical value for determining the inaccurate range of judgment, and is set based on historical data; For any single imbalance defect location: If the imbalance defect location type is insufficient defect location, and the sampling judgment of the defect ratio is too small, the omission rate of the defect range judgment is greater than or equal to the preset coverage rate; Alternatively, if the imbalance defect location type is defect over-location, and the sampling judgment on the defect ratio is too large, and the judgment on the defect range is over-coverage greater than or equal to the preset coverage; The location of the imbalance defect will then be marked as the location of the sampling imbalance defect. In step one, the process of determining whether there is a sampling defect location imbalance includes: The proportion of sampling imbalance defect locations within all historical defect locations is statistically analyzed to obtain the sampling imbalance defect location percentage. This percentage is then compared with a preset percentage. If the sampling imbalance defect location percentage is greater than or equal to the preset percentage, then there is a sampling defect location imbalance phenomenon. Understandably, the preset percentage is the critical value for determining whether there is an imbalance between sampling bias and defect location, and it is set in conjunction with the company's quality objectives. It should be noted that the purpose of determining whether there is an imbalance in defect location due to inaccurate sampling is to: quantify the sampling judgment bias (proportion / range) by backtracking historical data, establish the causal relationship between the bias and the imbalance in defect location, and determine whether subsequent optimization plans need to be initiated. Step 2: If it exists, divide the current batch of temperature fuses into stable sub-batches and fluctuating sub-batches according to the stable period of equipment parameters, and cluster them by extracting the fluctuation characteristics of equipment parameters of each fluctuating sub-batchens to obtain multiple fluctuating sub-batch groups, and calculate the fluctuation coefficient of each fluctuating sub-batch group. In step two, the process of dividing the current batch of temperature fuse production process into stable sub-batches and fluctuating sub-batches according to the stable period of production parameters includes: It should be noted that the core of sub-batch division is based on the raw material feeding batch and the stable equipment parameter period as the boundary. Through dual-dimensional constraints, it ensures that each sub-batch meets the requirements of the same batch of raw materials + the same stable process, so as to achieve precise binding of raw materials-process-product and avoid batch crossover (such as mixing different raw materials) and expansion of defect location range (such as misjudging related unrelated process periods). The criteria for determining a stable period of equipment parameters are that the fluctuation of equipment parameters (key parameters affecting the fuse characteristics) is ≤ a preset threshold and the duration is ≥ 5 minutes, which is considered a stable period of equipment. Equipment parameters include: stretching speed, annealing temperature, coating thickness, curing temperature, welding temperature, pressure, time, workshop temperature and humidity, tin plating temperature, etc. When any device parameter fluctuates beyond the threshold, the current sub-batch is immediately split. After the parameter stabilizes and remains stable for 3 minutes, a new sub-batch is started. The basic granularity is 5-10 minutes per sub-batch. When the equipment parameters are stable, the interval is split in 10-minute intervals (reducing management complexity). When the parameters fluctuate frequently, the interval is automatically adjusted to 5-minute intervals (improving the accuracy of fluctuation detection). Sub-batches produced during periods of stable equipment parameters are marked as stable sub-batches, and sub-batches produced during periods of fluctuating equipment parameters are marked as fluctuating sub-batches. In step two, the process of extracting the parameter fluctuation features of each fluctuation sub-batch and performing clustering includes: For all fluctuating sub-batches, extract the fluctuation characteristics of equipment parameters, including: Fluctuation range: the maximum deviation of the parameter; Fluctuation frequency: The number of sampling points exceeding the threshold; Drift rate: the amount of parameter change per unit time; Duration: The total duration of fluctuations exceeding the threshold; The parameter fluctuation characteristics of the fluctuation sub-batch are standardized and used as clustering features. The clustering process includes: The number of clusters K is determined using the elbow method, with K starting from 1. The sum of squared errors (SSE) corresponding to each K is calculated, and the K-SSE curve is plotted. The optimal value of K is the point where the SSE drops sharply and then flattens out. K clustering features are randomly selected as initial cluster centers. The Euclidean distance between each clustering feature and the K cluster centers is calculated, and the clustering features are assigned to the clusters with the closest Euclidean distance. After all clustering features are assigned, the center of each cluster is recalculated. The assignment and update are repeated until the change in the cluster center is less than or equal to the preset threshold. The clustering ends and multiple fluctuating sub-batch groups are finally obtained. In step two, the calculation process for the volatility coefficient of each volatility sub-batch group includes: For any one of the parameters—fluctuation amplitude, fluctuation frequency, drift rate, and duration—the fluctuation characteristics are as follows: The defect rate when the historical parameter fluctuation characteristics are abnormal is statistically analyzed. The defect rate of these four core characteristics is normalized and used as weights. For any sub-batch in any fluctuation sub-batch group, the degree of fluctuation is obtained by weighting. For any fluctuation sub-batch group, calculate the average fluctuation degree of each fluctuation sub-batch within the fluctuation sub-batch group as the fluctuation coefficient of the fluctuation sub-batch group. It should be noted that the role of sub-batch refinement and fluctuation clustering is to solve the problem of incomplete sampling coverage caused by the extensive management of large batches in the past, to reduce the sampling unit from large batches to sub-batches, and at the same time to achieve risk stratification through fluctuation clustering, so that sampling resources are tilted towards high-risk areas. Step 3: Construct a dynamic mapping model of volatility coefficient-sampling ratio based on the volatility coefficient of each volatility sub-batch group, and calculate the sampling ratio of stable sub-batch and each volatility sub-batch group to perform stratified sampling detection; In step three, the construction process of the volatility coefficient-sampling ratio dynamic mapping model includes: It should be noted that the construction logic of the volatility coefficient-sampling ratio dynamic mapping model is as follows: under a fixed overall ratio, the stable sub-batch decreases and the volatile sub-batch increases; the higher the volatility coefficient, the higher the sampling ratio. Based on historical data, the risk contribution of stable sub-batch and fluctuating sub-batch is calculated separately, which is the product of the defect rate and the product quantity. The total sampling proportion of the stable sub-batch is obtained by multiplying the fixed overall proportion by the risk contribution of the stable sub-batch, and the total sampling proportion of the volatile sub-batch is obtained by multiplying the fixed overall proportion by the risk contribution of the volatile sub-batch. For stable sub-lots, the total sampling proportion of stable sub-lots is evenly distributed to each stable sub-lot; For volatile sub-batches, the sampling ratio is allocated according to the volatility coefficient of each volatile sub-batch group, specifically as follows: The allocation logic is as follows: the volatility coefficient is converted into a risk weight. The higher the volatility coefficient, the greater the weight, and the larger the sample size allocated. For any fluctuation sub-batch group: The relative weight of the volatility sub-batch group is obtained by proportionally calculating the volatility coefficient to the sum of the volatility coefficients of all volatility sub-batches. The sampling proportion of the fluctuation sub-batch group is obtained by multiplying the total sampling proportion of the fluctuation sub-batch group by the relative weight of the fluctuation sub-batch group. Within the fluctuating sub-batch group, the sampling ratio is further allocated according to the degree of fluctuation of each fluctuating sub-batch; The degree of fluctuation of each fluctuation sub-batch within the fluctuation sub-batch group is used as the sampling proportion coefficient. The sampling proportion of each fluctuation sub-batch is obtained by multiplying the sampling proportion of the fluctuation sub-batch group by the sampling proportion coefficient of each fluctuation sub-batch. It should be noted that the purpose of dynamic sampling ratio calculation is to shift sampling resources from low-risk stable sub-batches to high-risk volatile sub-batches under a fixed overall ratio, thereby improving the efficiency of sampling in covering the true risk and reducing sampling bias. Step 4: If defective products are found during sampling inspection, the actual defect rate is calculated using the stratified confidence interval method based on the sampling defect rate, and the defect location range is determined. If defective products are not found during sampling inspection, but are found at the terminal, the defect location range is determined through analysis of related sub-batches. In step four, the process of calculating the defect rate and determining the defect location range using the stratified confidence interval method based on the sampling defect rate includes: If defective products are found during sampling inspection, the ratio of the number of defective products to the sample size is calculated to obtain the basic sampling defect rate. The defective product belongs to the sub-batch. If the defective product belongs to a stable sub-batch, the basic sampling defect rate is the actual defect rate, and the defect location range is the stable sub-batch to which the defective product belongs. If the defective product belongs to a fluctuating sub-batch, a risk correction factor is calculated based on the fluctuation coefficient of the fluctuating sub-batch group and the deviation between the historical sampling defect rate and the actual defect rate, specifically: For any fluctuation sub-batch group: Calculate the ratio of the actual defect rate to the sampling defect rate of all fluctuating sub-batches in each fluctuating sub-batch group within the historical period, and take the average value as the historical deviation coefficient. The risk correction factor K is obtained by adding the relative weights of the volatility sub-batch groups to the historical deviation coefficients; Understandably, the physical meaning of the risk correction factor is: to quantify the superposition effect of the current volatility risk intensity of the volatility sub-batch and the historical sampling bias. Essentially, it is the total correction coefficient of the sampling failure rate minus the actual failure rate. By integrating the relative risk weight of the current volatility sub-batch group with the experience of the historical sampling data covering the true risk of the group, it achieves accurate correction of the basic sampling failure rate, ensuring that the corrected actual failure rate not only matches the true risk level of the current volatility sub-batch, but also makes up for the systematic underestimation / overestimation bias that may exist in historical sampling. The defect rate of the fully fluctuating sub-batch was estimated using the stratified confidence interval method. The formula is , where n 不合格 Let n be the number of defective samples. 抽 Where K is the sampling size, and K is the risk correction factor for the fluctuating sub-batch group; The defect location scope includes all fluctuation sub-batches in the fluctuation sub-batch group described in the fluctuation sub-batch description; In step four, the process of determining whether a defect is an individual defect or a batch defect includes: If the sampling inspection is qualified, the terminal will malfunction: Extract the unique identifier of the defective product at the terminal, associate it with its sub-batch, and count the proportion of defective products in the same sub-batch or the same fluctuating sub-batch group to the total number of defective products at the terminal, so as to clarify the distribution concentration of defective products. Based on the characteristics of individual defects being randomly distributed and unrelated to systematic fluctuations, and batch defects being concentrated and strongly correlated with systematic fluctuations, the determination of whether a defect is an individual defect or a batch defect is as follows: Related defect to sub-batch: If the defect rate of the virtual sub-batch to which the defect belongs is less than or equal to the preset defect rate, and the defect is distributed in three or more non-adjacent sub-batches, and the sub-batch is a stable sub-batch, then it is an individual defect. Conversely, if the defect rate of the sub-batch to which the defect belongs is greater than the preset defect rate, and the same type of defect occurs in two or more sub-batches within the same historical batch, and the sub-batch is a fluctuating sub-batch, then it is a batch defect. In step four, the process of determining the defect location range includes: Only for products with poor defect localization and the fluctuating sub-batch to which the defect belongs, plus two adjacent sub-batches in the same fluctuating sub-batch group, prioritize high-risk products for defect localization through non-destructive full inspection screening, and avoid expanding to stable sub-batch or low-fluctuation sub-batch; It should be noted that the role of sampling result processing and defect location decision-making is to: solve the problems of sampling defects - accurate estimation of defect rate and sampling qualified terminals defects - defect type determination in different scenarios, and ultimately achieve the matching of defect location range with actual risk and eliminate defect location imbalance.

[0022] The technical solution and advantages of this application embodiment are as follows: Deviation analysis is performed on the historical defect location range and the actual defect range of defective thermal fuses to identify unbalanced defect locations. Furthermore, by comparing and analyzing the sampling data corresponding to the unbalanced defect locations with the actual data, sampling imbalance defect locations are identified. Statistical analysis is performed on the sampling imbalance defect locations to determine whether there is an imbalance in sampling defect location. If so, the current batch of thermal fuses is divided into stable sub-batches and fluctuating sub-batches according to the stable period of equipment parameters. Clustering is performed by extracting the fluctuation characteristics of equipment parameters from each fluctuating sub-batchens to obtain multiple fluctuating sub-batchens groups, and the fluctuation coefficient of each fluctuating sub-batchens group is calculated. A dynamic mapping model of fluctuation coefficient-sampling ratio is constructed based on the fluctuation coefficient of each fluctuating sub-batchens group, and the sampling ratio of the stable sub-batches and each fluctuating sub-batchens group is calculated for stratified sampling inspection. If sampling inspection finds defective products, the actual defect rate is calculated using the stratified confidence interval method based on the sampling defect rate, and the defect location range is determined. If sampling inspection does not find defective products, but the terminal finds defective products, the defect location range is determined through analysis of associated sub-batches. This application analyzes the discrepancy between historical sampling data and real data, correlates historical defect location cases, and determines whether there is an imbalance between sampling bias and defect location. If so, it divides stable and fluctuating sub-batches according to the stable period of equipment parameters, extracts the fluctuation characteristics of fluctuating sub-batches and clusters them, calculates the fluctuation coefficient of each fluctuating sub-batch group, and constructs a dynamic sampling ratio mapping model based on the fluctuation coefficient to achieve stratified sampling. It accurately calculates the defect rate and determines the defect location range according to different scenarios: when the sampling is defective, the defect rate is estimated by combining risk correction factors; when the sampling is qualified but the terminal is defective, the individual / batch defect is determined. This improves the accuracy of defect rate estimation and the scientific nature of defect location decision-making, reduces the cost of over-defect location and the risk of under-defect location, and realizes accurate location and closed-loop quality control throughout the entire process of temperature fuse production.

[0023] Example 2: Please refer to Figure 3 As shown in the embodiment of the present invention, a production quality management system for thermal fuses includes the following modules: Defect Location Imbalance Judgment Module: Analyzes the deviation between the historical defect location range and the actual defect range of defective temperature fuse products, identifies imbalanced defect locations, and identifies sampling imbalanced defect locations by comparing and analyzing the sampling data corresponding to the imbalanced defect locations with the actual data. Statistical analysis is performed on the sampling imbalanced defect locations to determine whether there is a sampling defect location imbalance phenomenon. Sub-batch segmentation module: If it exists, the current batch of temperature fuses is divided into stable sub-batches and fluctuating sub-batches according to the stable period of equipment parameters. The fluctuation characteristics of equipment parameters of each fluctuating sub-batch are extracted and clustered to obtain multiple fluctuating sub-batch groups. The fluctuation coefficient of each fluctuating sub-batch group is calculated. Sampling and inspection module: Construct a dynamic mapping model of fluctuation coefficient-sampling ratio based on the fluctuation coefficient of each fluctuation sub-batch group, and calculate the sampling ratio of stable sub-batch and each fluctuation sub-batch group to perform stratified sampling and inspection; Defect localization range determination module: If sampling inspection finds defective products, the actual defect rate is calculated using the stratified confidence interval method based on the sampling defect rate, and the defect localization range is determined. If sampling inspection does not find defective products, but defective products are found at the terminal, the defect localization range is determined through analysis of associated sub-batch.

[0024] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the present invention should still fall within the scope of the present invention.

Claims

1. A method for quality management in the production of thermal fuses, characterized in that: Includes the following steps: Deviation analysis is performed on the historical defect location range and the actual defect range of defective thermal fuse products to identify unbalanced defect locations. By comparing and analyzing the sampling data corresponding to the unbalanced defect locations with the actual data, sampling unbalanced defect locations are identified. Statistical analysis is performed on the sampling unbalanced defect locations to determine whether there is an imbalance in sampling defect locations. If present, the current batch of temperature fuses is divided into stable sub-batches and fluctuating sub-batches according to the stable period of equipment parameters. The equipment parameter fluctuation characteristics of each fluctuating sub-batch are extracted and clustered to obtain multiple fluctuating sub-batch groups. The fluctuation coefficient of each fluctuating sub-batch group is calculated. A dynamic mapping model of fluctuation coefficient and sampling ratio is constructed based on the fluctuation coefficient of each fluctuation sub-batch group, and the sampling ratio of stable sub-batch and each fluctuation sub-batch group is calculated for stratified sampling detection. If defective products are found during sampling inspection, the actual defect rate is calculated using the stratified confidence interval method based on the sampling defect rate, and the defect location range is determined. If defective products are not found during sampling inspection, but are found at the terminal, the defect location range is determined through analysis of related sub-batches.

2. The production quality management method for a thermal fuse according to claim 1, characterized in that: The identification method for locating sampling imbalance defects is as follows: For any historical defect location: If new defective products still exist in the terminal of the unlocated area after defect location, it is considered insufficient defect location. If the actual defect rate of the products with defect location does not meet the requirements, it is considered excessive defect location. Defect location with insufficient or excessive defect location will be marked as unbalanced defect location. If defect localization is insufficient and the sampling rate and defect range are judged too low, or if defect localization is excessive and the sampling rate and defect range are judged too high, then the unbalanced defect localization is called sampling imbalance defect localization.

3. The production quality management method for a thermal fuse according to claim 2, characterized in that: The sampling method for determining the defect rate and defect range is as follows: The sampling defect rate is calculated by proportionally dividing the number of defective samples by the sample size. The true defect rate is calculated by proportionally dividing the actual total number of defective products by the total number of historical batches. If the sampling defect rate is less than the actual defect rate, then the sampling rate underestimates the defect rate; conversely, if the sampling rate is greater than the actual defect rate, then the sampling rate overestimates the defect rate. The over-coverage rate is obtained by calculating the ratio of the portion of the defect location range without actual defects to the total number of defect location ranges in the sampling judgment. If the over-coverage rate meets the requirements, the sampling will overestimate the defect range; conversely, if the over-coverage rate does not meet the requirements, the sampling will underestimate the defect range.

4. The production quality management method for a thermal fuse according to claim 3, characterized in that: The method for determining whether there is a sampling defect location imbalance is as follows: The proportion of sampling imbalance defect locations within all historical defect locations is calculated in the historical period. If the proportion of sampling imbalance defect locations meets the requirements, then there is a sampling imbalance phenomenon.

5. The production quality management method for a thermal fuse according to claim 1, characterized in that: The method for dividing the stable sub-batch and the fluctuating sub-batch is as follows: Real-time monitoring of equipment parameters during the production process of temperature fuses; marking sub-batches produced during periods of stable equipment parameters as stable sub-batches, and marking sub-batches produced during periods of fluctuating equipment parameters as fluctuating sub-batches. The criteria for determining a stable period of equipment parameters are that both the fluctuation and duration of the equipment parameters meet the requirements. When the fluctuation of any equipment parameter does not meet the requirements, it is considered a period of equipment parameter fluctuation.

6. The production quality management method for a thermal fuse according to claim 5, characterized in that: The process of extracting the parameter fluctuation features of each fluctuation sub-batch and performing clustering is as follows: For all fluctuating sub-batches, extract the fluctuation characteristics of the equipment parameters: fluctuation amplitude, fluctuation frequency, drift rate, and duration; The parameter fluctuation characteristics of the fluctuation sub-batch are standardized and used as clustering features to perform clustering, ultimately resulting in multiple fluctuation sub-batch groups.

7. The production quality management method for a thermal fuse according to claim 6, characterized in that: The calculation process for the volatility coefficient of each volatility sub-batch group is as follows: For any one of the parameters—fluctuation amplitude, fluctuation frequency, drift rate, and duration—the fluctuation characteristics are as follows: The defect rate when the historical parameter fluctuation characteristics are abnormal is statistically analyzed. The defect rate of the parameter fluctuation characteristics is normalized and used as a weight. For any sub-batch in any fluctuation sub-batch group, the degree of fluctuation is obtained by weighting. For any fluctuation sub-batch group, the average fluctuation degree of each fluctuation sub-batch within the fluctuation sub-batch group is calculated as the fluctuation coefficient of the fluctuation sub-batch group.

8. The production quality management method for a thermal fuse according to claim 1, characterized in that: The construction process of the volatility coefficient-sampling ratio dynamic mapping model and the calculation method of the sampling ratio are as follows: Calculate the risk contribution of stable sub-batch and fluctuating sub-batch separately, which is the product of defect rate and product quantity; The total sampling proportion of the stable sub-batch is obtained by multiplying the fixed overall proportion by the risk contribution of the stable sub-batch, and the total sampling proportion of the volatile sub-batch is obtained by multiplying the fixed overall proportion by the risk contribution of the volatile sub-batch. For stable sub-lots, the total sampling proportion of stable sub-lots is evenly distributed to each stable sub-lot; For any fluctuation sub-batch group: The sampling proportion of the fluctuation coefficient is calculated by proportionally to the sum of the fluctuation coefficients of all fluctuation sub-batches, and then multiplied by the total sampling proportion of the fluctuation sub-batches to obtain the sampling proportion of the fluctuation sub-batch group. The degree of fluctuation of each fluctuation sub-batch within the fluctuation sub-batch group is used as the sampling proportion coefficient. The sampling proportion of each fluctuation sub-batch is obtained by multiplying the sampling proportion of the fluctuation sub-batch group by the sampling proportion coefficient of each fluctuation sub-batch.

9. A method for production quality management of a thermal fuse according to claim 8, characterized in that: The method for determining the defect location range is as follows: If defective products are found during sampling inspection, the ratio of the number of defective products to the sample size is calculated to obtain the basic sampling defect rate. The defective product belongs to the sub-batch. If the defective product belongs to a stable sub-batch, the basic sampling defect rate is the actual defect rate, and the defect location range is the stable sub-batch to which the defective product belongs. If the defective product belongs to a fluctuating sub-batch, for any fluctuating sub-batch group: Calculate the ratio of the actual defect rate to the sampling defect rate of all fluctuating sub-batches in each fluctuating sub-batch group within the historical period, and take the average value as the historical deviation coefficient. The risk correction factor K is obtained by adding the relative weights to the historical deviation coefficients; The defect rate of the entire fluctuation sub-batch is estimated by using the stratified confidence interval method combined with the risk correction factor. The defect location range is all fluctuation sub-batches in the fluctuation sub-batch group. If the sampling inspection is qualified, but the terminal is defective, only the defective product belongs to the fluctuating sub-batch and the adjacent sub-batch in the same fluctuating sub-batch group.

10. A production quality management system for thermal fuses, characterized in that: Includes the following modules: Defect Location Imbalance Judgment Module: Analyzes the deviation between the historical defect location range and the actual defect range of defective temperature fuse products, identifies imbalanced defect locations, and identifies sampling imbalanced defect locations by comparing and analyzing the sampling data corresponding to the imbalanced defect locations with the actual data. Statistical analysis is performed on the sampling imbalanced defect locations to determine whether there is a sampling defect location imbalance phenomenon. Sub-batch segmentation module: If it exists, the current batch of temperature fuses is divided into stable sub-batches and fluctuating sub-batches according to the stable period of equipment parameters. The fluctuation characteristics of equipment parameters of each fluctuating sub-batch are extracted and clustered to obtain multiple fluctuating sub-batch groups. The fluctuation coefficient of each fluctuating sub-batch group is calculated. Sampling and inspection module: Construct a dynamic mapping model of fluctuation coefficient-sampling ratio based on the fluctuation coefficient of each fluctuation sub-batch group, and calculate the sampling ratio of stable sub-batch and each fluctuation sub-batch group to perform stratified sampling and inspection; Defect localization range determination module: If sampling inspection finds defective products, the actual defect rate is calculated using the stratified confidence interval method based on the sampling defect rate, and the defect localization range is determined. If sampling inspection does not find defective products, but defective products are found at the terminal, the defect localization range is determined through analysis of associated sub-batch.

Images