A method for constructing a supplier quality access evaluation model based on multiple levels of indicators

Abstract

This invention discloses a method for constructing a supplier quality access evaluation model based on multi-level indicators. The method includes constructing a multi-level quality access evaluation indicator system; collecting multi-source heterogeneous supplier quality data and classifying it according to primary indicators; verifying data authenticity to obtain data credibility coefficients; calculating the dynamic weights of each tertiary evaluation indicator and the scores of secondary classification indicators; calculating the dynamic weights of each secondary classification indicator and the scores of primary indicators; calculating the feedback coefficient and the dynamic weights of each primary indicator; combining the data credibility coefficient and the feedback coefficient to weightedly calculate the primary indicator scores to obtain a comprehensive quality access score; and mapping the comprehensive quality access score to intervals according to preset hierarchical decision rules to generate supplier quality access level assessment results and corresponding management decision schemes. This method provides enterprises with a scientific, objective, and traceable quantitative basis for supplier access, improving the overall quality level of the supply chain.

Description

Technical Field

[0001] This invention relates to the field of supplier quality management and evaluation technology, and in particular to a method for constructing a supplier quality access evaluation model based on multi-level indicators. Background Technology

[0002] With the increasing specialization of global supply chains and intensifying market competition, the quality capabilities of suppliers have become a key factor affecting the final quality of enterprise products, cost control, and brand reputation. In the supplier introduction stage, building a scientific, systematic, and quantifiable quality access evaluation system to select partners with the ability to continuously and stably provide qualified products and services from the source is the primary step for enterprises to implement supply chain quality management and prevent quality risks.

[0003] Existing supplier evaluation methods primarily rely on expert experience-based weighting or fixed-weight models, which suffer from the following technical shortcomings: First, the authenticity of data lacks a verification mechanism, making it difficult to effectively identify data fraud, misrepresentation, or misuse of expired data. Second, the weight allocation is static and rigid, failing to dynamically optimize based on the actual statistical correlation between indicators and quality performance, the completeness of historical data, or the structural conformity of indicators with the standard indicator system. Furthermore, it neglects the dynamic correlation and synergistic effects between indicators, lacks quantitative analysis of the mutual influence between dimensions such as quality, technology, and cost, and treats service quality as a simple parallel dimension, failing to reflect its multiplier effect as a soft capability on hard indicators. Finally, the evaluation results are susceptible to interference from low-reliability data, lacking an adaptive correction mechanism based on data reliability, leading to biased evaluation results. Therefore, this invention proposes a supplier quality access evaluation model construction method based on multi-level indicators. By constructing multi-level indicators, verifying data authenticity, calculating dynamic weights, and constructing feedback coefficients to coordinately adjust the weighted results of the indicators, the scientific nature, resistance to manipulation, and risk prediction capabilities of supplier access evaluation are improved. This provides enterprises with quantifiable, traceable, and verifiable technical means to control supply chain quality risks from the source and select partners that best meet their quality strategies. Summary of the Invention

[0004] The purpose of this invention is to provide a method for constructing a supplier quality access evaluation model based on multi-level indicators.

[0005] To achieve the above objectives, the present invention is implemented according to the following technical solution: This invention includes the following steps: Construct a multi-level quality access evaluation index system, collect multi-source heterogeneous data on supplier quality, classify them according to primary indicators, and verify the authenticity of the data to obtain the data credibility coefficient. Based on the mutual information entropy between each level 3 evaluation indicator and the actual quality results in the historical supplier database, as well as the historical frequency of occurrence, the dynamic weight of each level 3 evaluation indicator is calculated, and the weighted calculation is used to obtain the score value of the secondary classification indicator. Based on the structural and numerical similarity between each secondary classification indicator and its corresponding standard secondary classification indicator, the dynamic weight of each secondary classification indicator is calculated, and the score of the primary indicator is obtained by weighted calculation. The feedback coefficient is calculated based on service quality and collaboration capability data. The sensitivity of mutual influence between primary indicators and the actual supply frequency are calculated based on the historical supplier database. The dynamic weight of each primary indicator is calculated. The scores of the primary indicators are weighted and calculated by combining the data credibility coefficient and the feedback coefficient to obtain the comprehensive quality access score. Based on the preset hierarchical decision-making rules, the comprehensive score of quality access is mapped to a range to generate the supplier quality access level assessment results and corresponding management decision-making schemes; The multi-level quality access evaluation index system includes primary indicators, secondary classification indicators, and tertiary evaluation indicators; The primary indicators include quality control capability, technology R&D capability, cost control capability, and service quality and collaboration capability; the secondary indicators of quality control capability include quality system, process quality control, and quality assurance traceability; the secondary indicators of technology R&D capability include advanced technology capability, testing technology development, and R&D system; the secondary indicators of cost control capability include cost system and delivery status. The service quality and collaboration capabilities are not subject to secondary classification indicators or tertiary evaluation indicators.

[0006] Furthermore, the method for obtaining the data credibility coefficient includes: A multi-level quality access evaluation index system is constructed; the multi-level quality access evaluation index system includes primary indicators, secondary classification indicators, and tertiary evaluation indicators. The multi-source heterogeneous data on supplier quality is collected and categorized according to primary indicators; the multi-source heterogeneous data includes supplier-submitted materials, on-site audit materials, third-party verification data, and historical reputation. The internal logical consistency of various primary indicator data is checked to obtain the self-consistency coefficient; the logical consistency check includes timestamp continuity check, numerical range rationality check, and text description self-consistency check. Extract comparable key fields from various primary indicator data, and calculate the semantic similarity and numerical deviation of primary indicator data of the same type and other primary indicator data on the same field to obtain the cross-validation coefficient. The timeliness coefficient is calculated based on the data collection time of various primary indicators; The reliability coefficient of similar primary indicator data is calculated using the self-consistency coefficient, cross-validation coefficient, and timeliness coefficient. The expression is as follows: ; ; ; in for The reliability coefficient of the first-level indicator data. , , To incorporate weights for credibility, The self-consistency coefficient. The cross-validation coefficients are... This is the timeliness coefficient. for Initial values ​​for the credibility weights of first-level indicator data. for Class and Semantic similarity of class-level indicator data The numerical deviation between the two types of data. The maximum allowable deviation threshold, The time decay coefficient, The interval between data collection time and the current time. This is the data half-life parameter.

[0007] Furthermore, the method for obtaining the secondary classification index score includes: All three-level evaluation indicators are quantified and assigned a percentage-based value to obtain the three-level evaluation indicator score; Based on the mutual information entropy between each level 3 evaluation indicator and the actual quality results in the historical supplier database, and their historical frequency of occurrence, the dynamic weight of each level 3 evaluation indicator is calculated, as expressed in the following expression: ; ; in Three-level evaluation indicators Dynamic weights, This is the relevance preference coefficient. Three-level evaluation indicators Compared with actual quality results Mutual information entropy, The maximum mutual information entropy value, Three-level evaluation indicators The effective collection rate is the ratio of the number of valid samples to the total number of samples in the historical evaluation data. Three-level evaluation indicators Historical values The set, For the actual quality results of the supplier Historical values The set, Indicators of Level 3 Evaluation Indicators Values And the actual quality results of the supplier Values The joint probability density, To consider only the three-level evaluation indicators Values Marginal probability density, To consider only the actual quality results of the supplier Values of; The dynamic weights of the tertiary evaluation indicators under the same type of secondary classification indicators are normalized, and the scores of the secondary classification indicators are calculated in combination with the scores of the tertiary evaluation indicators.

[0008] Furthermore, the method for obtaining the primary indicator score includes: A tertiary evaluation index system is determined under each secondary classification index. The structural similarity between the secondary classification index and the corresponding standard secondary classification index system is calculated based on a tree structure matching algorithm. The structural similarity includes the index field completeness rate, hierarchical correspondence rate, and logical association matching degree. Calculate the score similarity between each secondary classification indicator and the corresponding standard secondary classification indicator's tertiary evaluation indicator score to obtain the numerical similarity of the secondary classification indicators. Based on the structural similarity and numerical similarity of the secondary classification indicators, calculate the dynamic weights of the secondary classification indicators. The expression is as follows: ; ; in Secondary classification indicators Dynamic weights, This is the structural similarity preference coefficient. For the similarity of the secondary classification indicators, For the numerical similarity of secondary classification indicators, Secondary classification indicators The three-level evaluation index scoring vector, Standard secondary classification indicators The three-level evaluation index scoring vector, Allowable deviation bandwidth; The score of the primary indicator is calculated by weighting the scores of the secondary classification indicators.

[0009] Furthermore, the method for calculating the feedback coefficient includes: Service quality and collaboration capability data are quantified and assigned a percentage score to obtain feedback indicator scores. These feedback indicator scores are then weighted and fused to obtain the overall service quality and collaboration capability score. A feedback coefficient is calculated based on the service quality and collaboration capability score, expressed as follows: ; in For feedback coefficients, To adjust the strength coefficient, The service quality and collaboration capabilities are scored. This serves as a benchmark score for service quality and collaborative capabilities.

[0010] 7. Furthermore, the method for obtaining the comprehensive quality access score includes: Extract historical time-series change data for each primary indicator from the historical supplier database, and calculate the partial effect sensitivity and common change sensitivity among the primary indicators respectively. The expressions are as follows: ; ; in The partial effect sensitivity represents the score of the primary indicator when other indicators remain constant. Changes in the scoring of primary indicators Influence coefficient, For common change sensitivity, the score of the primary indicator is represented. and Simultaneous changes in the scoring of primary indicators The coupling influence coefficient, For historical sample index, For historical sample size, For indicator functions, For change detection threshold, To control the threshold; The comprehensive impact index is calculated based on the partial effect sensitivity and the common change sensitivity. The basic weight of each primary indicator is calculated by extracting the actual supply and usage frequency of each primary indicator. The expression is as follows: ; ; in As a primary indicator Dynamic weights, As a primary indicator The comprehensive impact index As a primary indicator The actual frequency of supply and use; The overall quality access score is obtained by weighting the scores of the primary indicators by combining the data credibility coefficient and the feedback coefficient. The expression is as follows: ; ; in The overall score for quality access is calculated as follows: For feedback coefficients, For the credibility adjustment function, The reliability coefficient for the primary indicator data of service quality and collaboration capability. This is a set of primary indicator categories, including quality control capability, technological research and development capability, and cost control capability. for Category 1 indicator score, This is the credibility attenuation coefficient. for The reliability coefficient of the first-level indicator data. This is the credibility threshold.

[0011] The beneficial effects of this invention are: This invention is a method for constructing a supplier quality access evaluation model based on multi-level indicators. Compared with existing technologies, this invention has the following technical advantages: This invention constructs a multi-dimensional data credibility verification model that includes self-consistency coefficient, cross-validation coefficient, and timeliness coefficient to quantify the confidence of supplier quality data, and introduces an exponential decay penalty mechanism for data below the credibility threshold to ensure the reliability and timeliness of the evaluation basis data. This invention constructs a multi-level indicator system and achieves a scientific, dynamic, and standardized approach to weight allocation by calculating dynamic weights based on mutual information entropy, historical occurrence frequency, structural similarity, numerical similarity, sensitivity, and actual supply and usage frequency. This invention designs service quality and collaboration capability as collaboration adjustment coefficients rather than simple parallel items. It constructs feedback coefficients by comparing their deviations from industry benchmarks, and adjusts the overall score of hard indicators as a whole, thus avoiding the selection of high-risk suppliers who are "technically strong but have poor service" or "falsify data but cover up service deficiencies". This invention enables the evaluation system to adapt to changes in data quality, technological iterations, and business scenario evolution by real-time calculation of data credibility coefficients and dynamic weighting driven by historical data. Combined with preset hierarchical decision-making rules and dynamic review mechanisms, it forms a quality management closed loop of access evaluation, credibility monitoring, dynamic level adjustment, and continuous improvement tracking, thereby achieving refined and data-driven management and control of the supplier's entire lifecycle. Attached Figure Description

[0012] Figure 1 This is a flowchart illustrating the steps of constructing a supplier quality access evaluation model based on multi-level indicators according to the present invention. Detailed Implementation

[0013] The present invention will be further described below through specific embodiments. The illustrative embodiments and descriptions herein are used to explain the present invention, but are not intended to limit the present invention.

[0014] The present invention provides a method for constructing a supplier quality access evaluation model based on multi-level indicators, comprising the following steps: like Figure 1 As shown, this embodiment includes the following steps: Construct a multi-level quality access evaluation index system, collect multi-source heterogeneous data on supplier quality, classify them according to primary indicators, and verify the authenticity of the data to obtain the data credibility coefficient. Based on the mutual information entropy between each level 3 evaluation indicator and the actual quality results in the historical supplier database, as well as the historical frequency of occurrence, the dynamic weight of each level 3 evaluation indicator is calculated, and the weighted calculation is used to obtain the score value of the secondary classification indicator. Based on the structural and numerical similarity between each secondary classification indicator and its corresponding standard secondary classification indicator, the dynamic weight of each secondary classification indicator is calculated, and the score of the primary indicator is obtained by weighted calculation. The feedback coefficient is calculated based on service quality and collaboration capability data. The sensitivity of mutual influence between primary indicators and the actual supply frequency are calculated based on the historical supplier database. The dynamic weight of each primary indicator is calculated. The scores of the primary indicators are weighted and calculated by combining the data credibility coefficient and the feedback coefficient to obtain the comprehensive quality access score. Based on the preset hierarchical decision-making rules, the comprehensive score of quality access is mapped to a range to generate the supplier quality access level assessment results and corresponding management decision-making schemes; The multi-level quality access evaluation index system includes primary indicators, secondary classification indicators, and tertiary evaluation indicators; The primary indicators include quality control capability, technology R&D capability, cost control capability, and service quality and collaboration capability; the secondary indicators of quality control capability include quality system, process quality control, and quality assurance traceability; the secondary indicators of technology R&D capability include advanced technology capability, testing technology development, and R&D system; the secondary indicators of cost control capability include cost system and delivery status. The service quality and collaboration capabilities are not subject to secondary classification indicators or tertiary evaluation indicators.

[0015] In this embodiment, the method for obtaining the data credibility coefficient includes: A multi-level quality access evaluation index system is constructed; the multi-level quality access evaluation index system includes primary indicators, secondary classification indicators, and tertiary evaluation indicators. The multi-source heterogeneous data on supplier quality is collected and categorized according to primary indicators; the multi-source heterogeneous data includes supplier-submitted materials, on-site audit materials, third-party verification data, and historical reputation. The internal logical consistency of various primary indicator data is checked to obtain the self-consistency coefficient; the logical consistency check includes timestamp continuity check, numerical range rationality check, and text description self-consistency check. Extract comparable key fields from various primary indicator data, and calculate the semantic similarity and numerical deviation of primary indicator data of the same type and other primary indicator data on the same field to obtain the cross-validation coefficient. The timeliness coefficient is calculated based on the data collection time of various primary indicators; The reliability coefficient of similar primary indicator data is calculated using the self-consistency coefficient, cross-validation coefficient, and timeliness coefficient. The expression is as follows: ; ; ; in for The reliability coefficient of the first-level indicator data. , , To incorporate weights for credibility, The self-consistency coefficient. The cross-validation coefficients are... This is the timeliness coefficient. for Initial values ​​for the credibility weights of first-level indicator data. for Class and Semantic similarity of class-level indicator data The numerical deviation between the two types of data. The maximum allowable deviation threshold, The time decay coefficient, The interval between data collection time and the current time. For data half-life parameters; In practical evaluation, taking the access evaluation scenario of a high-end logic chip packaging and testing service provider (OSAT) as an example, a multi-level quality access evaluation index system was determined: The primary indicator, Quality A-Control Capability, includes secondary classification indicators: A1 Quality System, A2 Process Quality Control, and A3 Quality Assurance and Traceability. A1 Quality System includes tertiary evaluation indicators: ISO 9001 certification effectiveness (A11), IATF 16949 automotive-grade certification effectiveness (A12), and ISO 26262 functional safety certification (A13). A2 Process Quality Control includes tertiary evaluation indicators: Wafer incoming inspection pass rate (A21), CP test yield (A22), FT test yield (A23), and packaging appearance defect rate (A24). A3 Quality Assurance and Traceability includes tertiary evaluation indicators: Batch traceability completeness rate (A31), reliability test pass rate (A32), and timeliness of closed-loop handling of quality anomalies (A33). The first-level indicator, quality B, is technology R&D capability, which includes B1 advanced technology capability, B2 testing technology development, and B3 R&D system. Among them, B1 advanced technology capability includes the third-level evaluation indicators of advanced packaging process coverage B11, minimum package pitch process capability B12, and high-density pin packaging capability B13. B2 testing technology development includes the third-level evaluation indicators of test program development cycle B21, test coverage B22, and multi-site parallel testing capability B23. B3 R&D system includes the third-level evaluation indicators of R&D investment ratio B31, number of packaging process patents B32, and core technology self-controllability rate B33. The primary indicator C - cost control capability includes the secondary classification indicators C1 cost system and C2 delivery status. The C1 cost system includes the tertiary evaluation indicators of packaging and testing price competitiveness index C11, wafer-level packaging cost optimization capability C12, and testing cost efficiency C13. The C2 delivery status includes the tertiary evaluation indicators of on-time delivery rate C21, capacity scale matching degree C22, emergency order response capability C23, and logistics and warehousing conditions C24. The primary indicator D - Service Quality and Collaboration Capability - does not have secondary classification indicators and is directly evaluated based on three dimensions: after-sales response and technical support (timeliness of technical support team response and FA report provision), information sharing and collaboration (ability of MES system to connect with client ERP system / completeness of API interface and real-time visualization of production data), and historical customer evaluation (customer repurchase rate, overall customer satisfaction score, Net Promoter Score (NPS), and quality complaint rate). The following four channels are used to collect multi-source heterogeneous data on supplier quality: (1) Suppliers submit materials by uploading them through the supplier portal system (including quality certification certificates, CP / FT test yield reports for the past 6 months, packaging process capability lists, R&D investment details and scanned copies of patent certificates); (2) On-site audit materials are collected by auditors carrying mobile audit terminals to the packaging and testing plant (in accordance with IATF16949 and VDA). 6.3 Standard implementation, including cleanroom environment collection, production line equipment calibration, batch traceability drills, and on-site inventory of work-in-process (WIP); (3) Third-party verification data is imported through third-party platforms (by connecting to TianXcha / QiXcha databases via API interface to obtain business registration information and legal risks; by verifying the legal status of invention patents through the XX patent database interface; by querying laboratory accreditation qualifications through the XX Conformity Assessment Committee website; by importing reliability test reports issued by third-party authoritative testing institutions through the LIMS system); (4) Historical reputation and customer evaluation are obtained through the enterprise SRM (supplier relationship management) system (the past 3 years of service history and historical order delivery records of the packaging and testing plant) and survey questionnaires (issue NPS survey questionnaires to 3-5 IC design companies (historical customers) that the current packaging and testing service provider is cooperating with); The heterogeneous multi-source data on supplier quality were categorized according to primary indicators, and the data authenticity was verified (the initial credibility weights for the four categories of primary indicator data were set to 0.25 / 0.3 / 0.25 / 0.2, the time decay coefficient was 0.693, and the half-life parameter was 180 days). The self-consistency coefficient, cross-validation coefficient, and timeliness coefficient of the four categories of primary indicator data (essentially, all tertiary evaluation indicators are divided into four categories) were calculated to be 0.9 / 0.95 / 0.8 / 0.80, 1 / 1 / 0.667 / 0.667, and 0.95 / 0.97 / 0.788 / 0.707, respectively. The credibility fusion weight was set to 0.4 / 0.4 / 0.2, and the credibility coefficients of the four categories of primary indicator data were calculated to be 0.947 / 0.974 / 0.72 / 0.88. Among them, the credibility of the cost control capability data was lower than the credibility threshold, and a penalty coefficient was introduced to correct the corresponding primary indicator quality score in the final evaluation.

[0016] In this embodiment, the method for obtaining the secondary classification index score includes: All three-level evaluation indicators are quantified and assigned a percentage-based value to obtain the three-level evaluation indicator score; Based on the mutual information entropy between each level 3 evaluation indicator and the actual quality results in the historical supplier database, and their historical frequency of occurrence, the dynamic weight of each level 3 evaluation indicator is calculated, as expressed in the following expression: ; ; in Three-level evaluation indicators Dynamic weights, This is the relevance preference coefficient. Three-level evaluation indicators Compared with actual quality results Mutual information entropy, The maximum mutual information entropy value, Three-level evaluation indicators The effective collection rate is the ratio of the number of valid samples to the total number of samples in the historical evaluation data. Three-level evaluation indicators Historical values The set, For the actual quality results of the supplier Historical values The set, Indicators of Level 3 Evaluation Indicators Values And the actual quality results of the supplier Values The joint probability density, To consider only the three-level evaluation indicators Values Marginal probability density, To consider only the actual quality results of the supplier Values of; The dynamic weights of the tertiary evaluation indicators under the same type of secondary classification indicators are normalized, and the scores of the secondary classification indicators are calculated in combination with the scores of the tertiary evaluation indicators. In actual evaluation, only the dynamic weights of the tertiary evaluation indicators and the scores of the secondary classification indicators under the three primary indicators of quality control capability, technology research and development capability, and cost control capability are calculated. First, establish a standard numerical reference system, including industry standard values ​​for each of the three levels of evaluation indicators. Optimal value and worst value For positive indicators, a percentage score is used. ,in These are the three-level evaluation index values; negative indicators are scored out of 100. For interval indicators, Time rating Take 100, otherwise decrease linearly according to the degree of deviation, where This is the minimum value of the three-level evaluation indicators. This represents the maximum value of the three-level evaluation indicators; for verification indicators, 100 is taken if verification is passed, and 0 is taken if verification is not passed. Maximum mutual information entropy value The maximum value of the mutual information entropy among all three-level evaluation indicators is taken as the normalization benchmark, and the mutual information entropy of each indicator is mapped to the interval [0,1]. The correlation preference coefficient is used to balance the relative importance of "indicator-performance correlation" and "data completeness" in the weight allocation, and the value range is [0,1]. When the historical data quality is high and the missing rate is low, the correlation preference coefficient can be increased to 0.7-0.8 (emphasizing the priority of predictive ability). When it is in the early stage of data accumulation or when the natural missing rate of some indicators is high, the correlation preference coefficient can be reduced to 0.4-0.5 (ensuring the priority of data availability). In the field of IC packaging and testing, for key quality indicators (such as A23 FT test yield), the correlation preference coefficient is set to 0.8. For forward-looking innovative indicators (such as B13 high-density pin packaging capability), due to the scarcity of historical data, the correlation preference coefficient is set to 0.6. The sum of the dynamic weights of the tertiary evaluation indicators under the same type of secondary classification indicators is not equal to 1, so normalization is required to ensure that the sum of the dynamic weights of the tertiary evaluation indicators under the same type of secondary classification indicators is equal to 1 (ensuring that the score of the secondary classification indicator is out of 100). Taking the dynamic weight calculation of the three-level evaluation indicators B-technology R&D capability / B1 advanced technology capability / minimum packaging pitch process capability B12 as an example, the maximum mutual information entropy value is taken. Given a high historical data missing rate / effective data collection rate of 0.6 and a correlation preference coefficient of 0.7, calculate the dynamic weights of the three-level evaluation indicators for minimum package pitch process capability. Normalization is performed based on the sum of dynamic weights of all tertiary evaluation indicators under the B1 advanced technology capability secondary classification indicator, which is 2.5, to obtain... Similarly, the dynamic weights of all tertiary indicator numerators are calculated, and the corresponding secondary classification indicator scores are calculated in combination with the scores of the corresponding tertiary evaluation indicators.

[0017] In this embodiment, the method for obtaining the primary indicator score includes: A tertiary evaluation index system is determined under each secondary classification index. The structural similarity between the secondary classification index and the corresponding standard secondary classification index system is calculated based on a tree structure matching algorithm. The structural similarity includes the index field completeness rate, hierarchical correspondence rate, and logical association matching degree. Calculate the score similarity between each secondary classification indicator and the corresponding standard secondary classification indicator's tertiary evaluation indicator score to obtain the numerical similarity of the secondary classification indicators. Based on the structural similarity and numerical similarity of the secondary classification indicators, calculate the dynamic weights of the secondary classification indicators. The expression is as follows: ; ; in Secondary classification indicators Dynamic weights, This is the structural similarity preference coefficient. For the similarity of the secondary classification indicators, For the numerical similarity of secondary classification indicators, Secondary classification indicators The three-level evaluation index scoring vector, Standard secondary classification indicators The three-level evaluation index scoring vector, Allowable deviation bandwidth; The score of the primary indicator is calculated by weighting the scores of the secondary classification indicators; In actual evaluation, only the scores of the three primary indicators—quality control capability, technology research and development capability, and cost control capability—and the corresponding dynamic weights of the secondary indicators are calculated. Taking the dynamic weight calculation of the secondary classification index of B-technology R&D capability / B1 advanced technology capability as an example, The structural similarity of the secondary classification indicators is calculated based on a tree-structure matching algorithm and is quantitatively evaluated through the following three dimensions: The cardinality of the standard field set is 3, and the supplier has submitted complete technical capability proof documents for indicators B11, B12, and B13, resulting in a field completeness rate of 1; the standard structure is B1 (secondary) → B11 / B12 / B13 (tertiary, parallel relationship), and in the supplier's submitted materials, B11, B12, and B13 are all directly attached to B1. The tree edit distance algorithm verifies that the actual structure is a complete subtree isomorphism of the standard structure, with a level correspondence rate of 1; based on knowledge of the IC packaging and testing field, the logical relationship between the three indicators is verified. The standard logic is that "B11, B12, and B13 should be parallel evaluation dimensions, jointly constituting the technical picture of advanced packaging technology capabilities, without any causal or subordinate relationship." The supplier clearly distinguishes the technical scope of the three indicators in their materials, resulting in a logical association matching degree of 1, and the weighted average structural similarity is taken as 1. The scores of the tertiary evaluation indicators B11 / B12 / B13 under the secondary classification indicators are 85 / 90 / 78 respectively. According to the standard numerical reference system of percentage-based quantitative assignment, the standard score corresponding to each tertiary evaluation indicator is 100 points. Taking the allowable deviation bandwidth as 30, the numerical similarity is calculated to be 0.739. In evaluating the technical capabilities of IC packaging and testing suppliers, the completeness (structure) of the indicator system is more critical than subtle differences in technical level (specific scores) in its impact on access decisions. Therefore, a structural similarity preference coefficient of 0.6 is set, and the dynamic weights of the secondary classification indicators for B1 advanced technical capabilities are calculated. Similarly, calculate the dynamic weights of the remaining secondary classification indicators, and combine them with the corresponding secondary classification indicator scores to calculate the corresponding primary indicator scores.

[0018] In this embodiment, the method for calculating the feedback coefficient includes: Service quality and collaboration capability data are quantified and assigned a percentage score to obtain feedback indicator scores. These feedback indicator scores are then weighted and fused to obtain the overall service quality and collaboration capability score. A feedback coefficient is calculated based on the service quality and collaboration capability score, expressed as follows: ; in For feedback coefficients, To adjust the strength coefficient, The service quality and collaboration capabilities are scored. This serves as a benchmark score for service quality and collaborative capabilities. In actual evaluations, service quality and collaboration capabilities data correspond to three dimensions of evaluation data, including after-sales response and technical support (timeliness of technical support team response, timeliness of FA report provision), information sharing and collaboration (MES system and client ERP system integration capabilities / API interface completeness, real-time visualization of production data), and historical customer evaluations (customer repurchase rate, overall customer satisfaction score, Net Promoter Score (NPS), and quality complaint rate). When assigning a percentage-based quantitative score to obtain feedback indicators: After-sales response and technical support dimensions: The technical support team's response timeliness score is calculated by using a segmented linear quantification based on the average time from submission to the first response of technical consultation tickets over the past 12 months; the FA report delivery timeliness score is calculated by using an exponential decay quantification based on the average time from submission to report delivery of quality anomaly FA requests over the past 12 months. Information sharing and collaboration dimension: API interface completeness score is quantified by the Jaccard similarity coefficient between the standard set of connected fields and the actual set of connected fields; production data real-time visualization score is quantified by the proportion of data packets with real-time data latency less than the threshold in the data return logs of the past 30 days; Historical customer evaluation dimensions: Customer repurchase rate score is quantified by the percentage of customers who continue to place orders in the second and subsequent years among customers who have cooperated for the past 3 years; overall customer satisfaction score is determined by scaling the "service quality" dimension in the annual supplier performance evaluation proportionally; Net Promoter Score (NPS) score is determined by mapping the NPS value range from [-100, +100] to the [0, 100] interval; quality complaint rate score is determined by statistically analyzing the customer quality complaint PPM value over the past 12 months and scoring it in reverse (tolerance thresholds vary by industry). When performing layered weighted fusion, the weighted scores for after-sales response and technical support are 0.6 / 0.4, for information sharing and collaboration are 0.5 / 0.5, and for historical customer evaluations are 0.25 / 0.25 / 0.3 / 0.2. The scores for the three dimensions (88 / 85 / 78.25 points) are weighted according to a weight of 0.3 / 0.3 / 0.4 to obtain a service quality and collaboration capability score of 83.2 points. Taking the benchmark score of 75 points for service quality and collaboration capability in the IC packaging and testing industry, the adjustment intensity coefficient is 0.15, and the calculated feedback coefficient is 1.016.

[0019] 8. Furthermore, the method for obtaining the comprehensive quality access score includes: Extract historical time-series change data for each primary indicator from the historical supplier database, and calculate the partial effect sensitivity and common change sensitivity among the primary indicators respectively. The expressions are as follows: ; ; in The partial effect sensitivity represents the score of the primary indicator when other indicators remain constant. Changes in the scoring of primary indicators Influence coefficient, For common change sensitivity, the score of the primary indicator is represented. and Simultaneous changes in the scoring of primary indicators The coupling influence coefficient, For historical sample index, For historical sample size, For indicator functions, For change detection threshold, To control the threshold; The comprehensive impact index is calculated based on the partial effect sensitivity and the common change sensitivity. The basic weight of each primary indicator is calculated by extracting the actual supply and usage frequency of each primary indicator. The expression is as follows: ; ; in As a primary indicator Dynamic weights, As a primary indicator The comprehensive impact index As a primary indicator The actual frequency of supply and use; The overall quality access score is obtained by weighting the scores of the primary indicators by combining the data credibility coefficient and the feedback coefficient. The expression is as follows: ; ; in The overall score for quality access is calculated as follows: For feedback coefficients, For the credibility adjustment function, The reliability coefficient for the primary indicator data of service quality and collaboration capability. This is a set of primary indicator categories, including quality control capability, technological research and development capability, and cost control capability. for Category 1 indicator score, This is the credibility attenuation coefficient. for The reliability coefficient of the first-level indicator data. This is the credibility threshold; In practical assessments, the change detection threshold Take 5, control threshold Take 3, in the indicator function A change in the absolute value of a Level 1 indicator score greater than 5 is considered a valid change (indicator function takes 1, otherwise 0); in the indicator function The absolute value of the change in the score of the primary indicator used as a control variable should be less than 3; in the indicator function The first-level indicator coupling score of the other two control variables is greater than 5, which is considered a valid change; Historical time-series change data of each primary indicator were extracted from 50 sets of historical supplier databases. The partial effect sensitivity and common change sensitivity among the primary indicators were calculated respectively. The comprehensive influence index of each primary indicator was calculated to be 0.47, 0.353, and 0.342. The actual supply usage frequency of each primary indicator was extracted (0.35, 0.4, and 0.25). The basic weight of each primary indicator was calculated to be 0.4205, 0.3609, and 0.2186. Based on the reliability coefficients of the four primary indicators calculated above (0.941 / 0.974 / 0.72 / 0.88), the corresponding reliability adjustment function is calculated as 1 / 1 / 0.942 / 1. Based on the scores of the primary indicators of quality control capability, technology research and development capability (82 / 78 / 75) and cost control capability and the feedback coefficient (1.016), the comprehensive score for quality access is calculated to be 79.3 points. Based on the preset hierarchical decision-making rules, the comprehensive score for quality access (79.3 points) is mapped to a range to generate the supplier quality access level assessment result (Level B) and the corresponding management decision plan (access is granted, but improvement requirements are required for the shortcomings in cost control capabilities). The specific tiered decision-making rules are as follows: Grade A (Excellent, >85 points): Priority access, strategic cooperation may be considered; Grade B (Good, 70-85 points): Access granted, but improvement requirements must be put forward for the shortcomings; Grade C (Acceptable, 60-70 points): Restricted access (e.g., only for non-critical materials, small-batch trial production), strict improvement plans must be formulated and tracked; Grade D (Unacceptable, <60 points or with veto items): Access denied.

[0020] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for constructing a supplier quality access evaluation model based on multi-level indicators, characterized in that, Includes the following steps: S1. Construct a multi-level quality access evaluation index system, collect multi-source heterogeneous data on supplier quality, classify them according to primary indicators, and verify the authenticity of the data to obtain the data credibility coefficient. S2. Based on the mutual information entropy between each level 3 evaluation indicator and the actual quality results in the historical supplier database, as well as the historical frequency of occurrence, calculate the dynamic weight of each level 3 evaluation indicator, and obtain the score value of the secondary classification indicator by weighted calculation. S3. Based on the structural and numerical similarity between each secondary classification indicator and the corresponding standard secondary classification indicator, calculate the dynamic weight of each secondary classification indicator, and calculate the primary indicator score by weighted calculation. S4. Calculate the feedback coefficient based on service quality and collaboration capability data, calculate the mutual influence sensitivity and actual supply frequency among primary indicators based on the historical supplier database, calculate the dynamic weight of each primary indicator, and combine the data credibility coefficient and feedback coefficient to calculate the weighted score of the primary indicator scores to obtain the comprehensive quality access score. S5. Based on the preset hierarchical decision-making rules, perform interval mapping on the comprehensive score of quality access to generate the supplier quality access level assessment results and corresponding management decision-making schemes. The multi-level quality access evaluation index system includes primary indicators, secondary classification indicators, and tertiary evaluation indicators; The primary indicators include quality control capability, technology R&D capability, cost control capability, and service quality and collaboration capability; the secondary indicators of quality control capability include quality system, process quality control, and quality assurance traceability; the secondary indicators of technology R&D capability include advanced technology capability, testing technology development, and R&D system; the secondary indicators of cost control capability include cost system and delivery status. The service quality and collaboration capabilities are not subject to secondary classification indicators or tertiary evaluation indicators.

2. The method for constructing a supplier quality access evaluation model based on multi-level indicators according to claim 1, characterized in that, The method for obtaining the data credibility coefficient includes: A multi-level quality access evaluation index system is constructed; the multi-level quality access evaluation index system includes primary indicators, secondary classification indicators, and tertiary evaluation indicators. The multi-source heterogeneous data on supplier quality is collected and categorized according to primary indicators; the multi-source heterogeneous data includes supplier-submitted materials, on-site audit materials, third-party verification data, and historical reputation. The internal logical consistency of various primary indicator data is checked to obtain the self-consistency coefficient; the logical consistency check includes timestamp continuity check, numerical range rationality check, and text description self-consistency check. Extract comparable key fields from various primary indicator data, and calculate the semantic similarity and numerical deviation of primary indicator data of the same type and other primary indicator data on the same field to obtain the cross-validation coefficient. The timeliness coefficient is calculated based on the data collection time of various primary indicators; The reliability coefficient of similar primary indicator data is calculated using the self-consistency coefficient, cross-validation coefficient, and timeliness coefficient. The expression is as follows: ； ； ； in for The reliability coefficient of the first-level indicator data. , , To incorporate weights for credibility, The self-consistency coefficient. The cross-validation coefficients are... This is the timeliness coefficient. for Initial values ​​for the credibility weights of first-level indicator data. for Class and Semantic similarity of class-level indicator data The numerical deviation between the two types of data. The maximum allowable deviation threshold, The time decay coefficient, The interval between data collection time and the current time. This is the data half-life parameter.

3. The method for constructing a supplier quality access evaluation model based on multi-level indicators according to claim 1, characterized in that, The method for obtaining the secondary classification index score includes: All three-level evaluation indicators are quantified and assigned a percentage-based value to obtain the three-level evaluation indicator score; Based on the mutual information entropy between each level 3 evaluation indicator and the actual quality results in the historical supplier database, and their historical frequency of occurrence, the dynamic weight of each level 3 evaluation indicator is calculated, as expressed in the following expression: ； ； in Three-level evaluation indicators Dynamic weights, This is the relevance preference coefficient. Three-level evaluation indicators Compared with actual quality results Mutual information entropy, The maximum mutual information entropy value, Three-level evaluation indicators The effective collection rate is the ratio of the number of valid samples to the total number of samples in the historical evaluation data. Three-level evaluation indicators Historical values The set, For the actual quality results of the supplier Historical values The set, Indicators of Level 3 Evaluation Indicators Values And the actual quality results of the supplier Values The joint probability density, To consider only the three-level evaluation indicators Values Marginal probability density, To consider only the actual quality results of the supplier Values of; The dynamic weights of the tertiary evaluation indicators under the same type of secondary classification indicators are normalized, and the scores of the secondary classification indicators are calculated in combination with the scores of the tertiary evaluation indicators.

4. The method for constructing a supplier quality access evaluation model based on multi-level indicators according to claim 1, characterized in that, The method for obtaining the primary indicator score includes: A tertiary evaluation index system is determined under each secondary classification index. The structural similarity between the secondary classification index and the corresponding standard secondary classification index system is calculated based on a tree structure matching algorithm. The structural similarity includes the index field completeness rate, hierarchical correspondence rate, and logical association matching degree. Calculate the score similarity between each secondary classification indicator and the corresponding standard secondary classification indicator's tertiary evaluation indicator score to obtain the numerical similarity of the secondary classification indicators. Based on the structural similarity and numerical similarity of the secondary classification indicators, calculate the dynamic weights of the secondary classification indicators. The expression is as follows: ； ； in Secondary classification indicators Dynamic weights, This is the structural similarity preference coefficient. For the similarity of the secondary classification indicators, For the numerical similarity of secondary classification indicators, Secondary classification indicators The three-level evaluation index scoring vector, Standard secondary classification indicators The three-level evaluation index scoring vector, Allowable deviation bandwidth; The score of the primary indicator is calculated by weighting the scores of the secondary classification indicators.

5. The method for constructing a supplier quality access evaluation model based on multi-level indicators according to claim 1, characterized in that, The method for calculating the feedback coefficient includes: Service quality and collaboration capability data are quantified and assigned a percentage score to obtain feedback indicator scores. These feedback indicator scores are then weighted and fused to obtain the overall service quality and collaboration capability score. A feedback coefficient is calculated based on the service quality and collaboration capability score, expressed as follows: ； in For feedback coefficients, To adjust the strength coefficient, The service quality and collaboration capabilities are scored. This serves as a benchmark score for service quality and collaborative capabilities.

6. The method for constructing a supplier quality access evaluation model based on multi-level indicators according to claim 1, characterized in that, The method for obtaining the comprehensive score for quality access includes: Extract historical time-series change data for each primary indicator from the historical supplier database, and calculate the partial effect sensitivity and common change sensitivity among the primary indicators respectively. The expressions are as follows: ； ； in The partial effect sensitivity represents the score of the primary indicator when other indicators remain constant. Changes in the scoring of primary indicators Influence coefficient, For common change sensitivity, the score of the primary indicator is represented. and Simultaneous changes in the scoring of primary indicators The coupling influence coefficient, For historical sample index, For historical sample size, For indicator functions, For change detection threshold, To control the threshold; The comprehensive impact index is calculated based on the partial effect sensitivity and the common change sensitivity. The basic weight of each primary indicator is calculated by extracting the actual supply and usage frequency of each primary indicator. The expression is as follows: ； ； in As a primary indicator Dynamic weights, As a primary indicator The comprehensive impact index As a primary indicator The actual frequency of supply and use; The overall quality access score is obtained by weighting the scores of the primary indicators by combining the data credibility coefficient and the feedback coefficient. The expression is as follows: ； ； in The overall score for quality access is calculated as follows: For feedback coefficients, For the credibility adjustment function, The reliability coefficient for the primary indicator data of service quality and collaboration capability. This is a set of primary indicator categories, including quality control capability, technological research and development capability, and cost control capability. for Category 1 indicator score, This is the credibility attenuation coefficient. for The reliability coefficient of the first-level indicator data. This is the credibility threshold.

Images