A method and system for quality and health management of new commercial vehicle projects

By constructing a standard quality dataset and performing dimensionless standardization and target weight calculation, combined with hierarchical early warning and multidimensional traceability algorithms, the problem of distorted quality assessment results in the development of new vehicle projects has been solved, realizing systematic and intelligent quality control throughout the entire life cycle, and effectively identifying and preventing potential risks.

CN123117794APending Publication Date: 2026-11-17JAINGXI ISUZU AUTOMOBILE CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611312760.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-27
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing technologies lack systematic and intelligent quality control throughout the entire lifecycle of new vehicle development projects, resulting in distorted quality assessment results, difficulty in achieving data interoperability and collaborative analysis, inability to effectively identify potential quality risks, and recurring similar problems.

Method used

By collecting raw quality data from new vehicle projects, a raw quality database is constructed. After preprocessing, a standard quality dataset is generated. Dimensionless standardization and target weight calculation are performed using function and weighting methods. Combined with hierarchical early warning and multidimensional source tracing algorithms, the root causes of quality anomalies are located.

Benefits of technology

It has achieved standardized collection and full-scale storage of quality data, quantitatively characterized the overall quality and health level of projects, identified potential risks in advance, shifted to proactive prevention and control, reduced the recurrence of similar problems, and supported systematic and intelligent quality management throughout the entire life cycle.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123117794A_ABST
    Figure CN123117794A_ABST
Patent Text Reader

Abstract

This invention provides a method and system for quality health management of new commercial vehicle projects. The method includes: collecting raw quality data from new commercial vehicle projects to construct a raw project quality database; preprocessing the raw project quality database to generate a standard quality dataset; performing dimensionless standardization on all quality indicators in the standard quality dataset using a function method; calculating the target weights of each quality indicator using a weighted method; performing weighted summation on the target weights of each quality indicator to calculate the evaluation result; determining the project quality status level based on the evaluation result; implementing graded early warning based on the project quality status level; and locating the root cause of quality anomalies using a multi-dimensional source tracing algorithm. This invention can significantly improve the efficiency of project quality management.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of project management technology, and in particular to a method and system for quality and health management of new commercial vehicle projects. Background Technology

[0002] The development of new vehicle projects is a complex systems engineering project encompassing planning, simulation, prototyping, testing, evaluation, and mass production implementation. The quality of each stage directly determines the project's quality, R&D cycle, and production stability. With the rapid iteration of new energy and electrification technologies in the automotive industry, the complexity of vehicle R&D has significantly increased, quality control risks have intensified, and influencing factors have become more diverse, highlighting the inadequacies of traditional quality management models.

[0003] Currently, the industry generally adopts a phased and fragmented quality management model. Quality data from each R&D stage are independent, making it difficult to achieve data interoperability and collaborative analysis, and a systematic and dynamic control system has not been formed. Quality assessments often rely on manual experience or fixed-weight models, which suffer from problems such as single dimensions, fixed weights, and strong subjectivity. At the same time, there is a lack of standardized processing mechanisms for multi-dimensional indicators, and heterogeneous indicators such as pass rate, number of defects, and rectification effect cannot be effectively integrated, resulting in distorted quality health evaluation results and difficulty in accurately quantifying the project quality level.

[0004] Existing publicly available technologies mostly target the manufacturing stage or a single testing phase, failing to cover the entire R&D process. Furthermore, they lack a quantitative evaluation system based on objective algorithms, making them ill-suited for managing massive amounts of R&D quality data. In addition, existing solutions often involve reactive rectification after problems occur, resulting in insufficient exposure of quality issues during development and recurring similar problems. This hinders the achievement of systematic and intelligent quality control throughout the entire lifecycle, necessitating a more comprehensive quality health management solution for new vehicle projects. Summary of the Invention

[0005] Based on this, the purpose of this invention is to provide a method and system for quality and health management of new commercial vehicle projects, in order to solve the problems that existing technologies mostly involve passive rectification after problems occur, insufficient quality exposure during the development process, repeated occurrence of similar problems, and difficulty in achieving systematic and intelligent quality control throughout the entire life cycle.

[0006] The first aspect of the present invention proposes: A method for quality and health management of new commercial vehicle projects, wherein the method includes: Collect raw quality data from new automotive projects to construct a raw quality database for the projects, and preprocess the raw quality database to generate a standard quality dataset. The dimensionless standardization of all quality indicators in the standard quality dataset is performed using a function method, and the target weights of each quality indicator are calculated using a weighting method. The target weights of each of the quality indicators are weighted and solved to calculate the evaluation results, and the project quality status level is determined based on the evaluation results. A graded early warning is executed based on the project's quality status level, and the root cause of the quality anomaly is located through a multi-dimensional tracing algorithm.

[0007] The beneficial effects of this invention are as follows: This solution constructs an original quality database for new automotive projects by collecting original quality data, and generates a standard quality dataset after preprocessing, enabling standardized collection and full-scale storage of quality data; based on a preset quality health assessment model, it completes dimensionless standardization of quality indicators and calculation of target weights, obtains assessment results and determines the quality status level through weighted solution, quantitatively characterizes the overall quality health level of the project, and identifies potential quality risks in advance; combined with a hierarchical early warning mechanism and a multi-dimensional tracing algorithm to locate the root cause of quality anomalies, it can shift quality control from passive post-event rectification to proactive pre-event prevention, fully expose quality hazards in the development process, reduce the recurrence of similar problems, and effectively support the systematic and intelligent quality control of the entire life cycle of automotive projects.

[0008] Furthermore, the steps of performing dimensionless standardization on all quality indicators in the standard quality dataset using a function method and calculating the target weights of each quality indicator using a weighting method include: An original evaluation matrix is ​​constructed based on the standard quality dataset, and positive and negative indicators are determined in the standard quality dataset based on the original evaluation matrix. The positive and negative indicators were normalized using a function method to obtain the standard efficacy score. Based on the standard efficacy score, the standard deviation, conflict coefficient, and information content of each quality indicator are calculated using a weighting method to obtain the target weight of each quality indicator.

[0009] Furthermore, the step of preprocessing the original quality database of the project to generate a standard quality dataset includes: The original quality data is mapped with stage labels according to the pre-set milestone nodes of commercial vehicle development, and quality data fragments corresponding to each development stage are generated. For each quality data segment, retrieve the confidence interval of the quality parameters of the benchmark project of the same stage and platform, and identify and remove abnormal data in the current segment based on the confidence interval; After removing outlier data, the field formats of each quality data segment are standardized and the units are aligned. The data is then categorized and reorganized according to the quality indicator categories to generate a standard quality dataset.

[0010] Furthermore, the step of performing a weighted solution process on the target weights of each of the quality indicators to calculate the evaluation result includes: Based on the vehicle BOM hierarchy and quality transfer path, a transmission correlation matrix between various quality indicators is constructed; Based on the aforementioned transmission correlation matrix, the target weights of each quality indicator are transmitted and corrected to obtain the corrected weights corresponding to each quality indicator. The dimensionless standardized values ​​of each quality indicator are weighted and summed with their corresponding corrected weights to obtain the final evaluation result.

[0011] Furthermore, the step of executing a graded early warning based on the project quality status level includes: By continuously collecting project evaluation results for multiple preset statistical periods, the current trend and rate of change in project quality and health are obtained through fitting. By combining the project quality status level with the rate of change, a corresponding early warning level is matched; Based on the matched early warning level, initiate the corresponding level of quality control response process.

[0012] 6. The method for quality and health management of new commercial vehicle projects according to claim 1, wherein the step of locating the root cause of quality anomalies using a multi-dimensional tracing algorithm includes: Using the quality indicator that triggers the anomaly as the top-level root node, the quality indicator is decomposed from top to bottom according to the hierarchy of whole vehicle-assembly-parts-process to form a quality indicator decomposition tree; The deviation correlation between the indicators of each lower-level node in the decomposition tree and the root node anomaly is calculated sequentially, and lower-level nodes with deviation correlation higher than the preset threshold are selected as candidate anomaly nodes. By retrospectively verifying the production process data and testing data corresponding to each candidate abnormal node, the root cause node of the quality abnormality and the corresponding deviation parameter are located.

[0013] Furthermore, the step of sequentially calculating the deviation correlation degree between the indicators of each lower-level node in the decomposition tree and the root node anomaly, and selecting lower-level nodes with deviation correlation degrees higher than a preset threshold as candidate anomaly nodes includes: Extract the quality data of each lower-level node indicator within two preset time windows before and after the anomaly occurs, and calculate the deviation of the corresponding indicator. Based on the hierarchical distance between each lower-level node and the root node, a corresponding hierarchical attenuation coefficient is introduced, and the deviation correlation degree of each lower-level node is calculated in combination with the deviation amplitude. The lower-level nodes are sorted from high to low according to the deviation correlation degree, and the nodes that are ranked first and whose deviation correlation degree is higher than the preset threshold are selected as candidate abnormal nodes.

[0014] The second aspect of the present invention proposes: A quality and health management system for new commercial vehicle projects, wherein the system includes: The processing module is used to collect raw quality data of new automobile projects to build a raw quality database for the project, and to preprocess the raw quality database to generate a standard quality dataset. The calculation module is used to perform dimensionless standardization of all quality indicators in the standard quality dataset using a function method, and to calculate the target weight of each quality indicator using a weighting method. The evaluation module is used to perform weighted solution processing on the target weights of each of the quality indicators to calculate the evaluation results and determine the project quality status level based on the evaluation results. The source tracing module is used to execute graded early warnings based on the project's quality status level and to locate the root cause of quality anomalies through a multi-dimensional source tracing algorithm.

[0015] Furthermore, the steps of performing dimensionless standardization on all quality indicators in the standard quality dataset using a function method and calculating the target weights of each quality indicator using a weighting method include: An original evaluation matrix is ​​constructed based on the standard quality dataset, and positive and negative indicators are determined in the standard quality dataset based on the original evaluation matrix. The positive and negative indicators were normalized using a function method to obtain the standard efficacy score. Based on the standard efficacy score, the standard deviation, conflict coefficient, and information content of each quality indicator are calculated using a weighting method to obtain the target weight of each quality indicator.

[0016] Furthermore, the step of preprocessing the original quality database of the project to generate a standard quality dataset includes: The original quality data is mapped with stage labels according to the pre-set milestone nodes of commercial vehicle development, and quality data fragments corresponding to each development stage are generated. For each quality data segment, retrieve the confidence interval of the quality parameters of the benchmark project of the same stage and platform, and identify and remove abnormal data in the current segment based on the confidence interval; After removing outlier data, the field formats of each quality data segment are standardized and the units are aligned. The data is then categorized and reorganized according to the quality indicator categories to generate a standard quality dataset.

[0017] Furthermore, the step of performing a weighted solution process on the target weights of each of the quality indicators to calculate the evaluation result includes: Based on the vehicle BOM hierarchy and quality transfer path, a transmission correlation matrix between various quality indicators is constructed; Based on the aforementioned transmission correlation matrix, the target weights of each quality indicator are transmitted and corrected to obtain the corrected weights corresponding to each quality indicator. The dimensionless standardized values ​​of each quality indicator are weighted and summed with their corresponding corrected weights to obtain the final evaluation result.

[0018] Furthermore, the step of executing a graded early warning based on the project quality status level includes: By continuously collecting project evaluation results for multiple preset statistical periods, the current trend and rate of change in project quality and health are obtained through fitting. By combining the project quality status level with the rate of change, a corresponding early warning level is matched; Based on the matched early warning level, initiate the corresponding level of quality control response process.

[0019] Furthermore, the step of locating the root cause of quality anomalies using a multidimensional source tracing algorithm includes: Using the quality indicator that triggers the anomaly as the top-level root node, the quality indicator is decomposed from top to bottom according to the hierarchy of whole vehicle-assembly-parts-process to form a quality indicator decomposition tree; The deviation correlation between the indicators of each lower-level node in the decomposition tree and the root node anomaly is calculated sequentially, and lower-level nodes with deviation correlation higher than the preset threshold are selected as candidate anomaly nodes. By retrospectively verifying the production process data and testing data corresponding to each candidate abnormal node, the root cause node of the quality abnormality and the corresponding deviation parameter are located.

[0020] Furthermore, the step of sequentially calculating the deviation correlation degree between the indicators of each lower-level node in the decomposition tree and the root node anomaly, and selecting lower-level nodes with deviation correlation degrees higher than a preset threshold as candidate anomaly nodes includes: Extract the quality data of each lower-level node indicator within two preset time windows before and after the anomaly occurs, and calculate the deviation of the corresponding indicator. Based on the hierarchical distance between each lower-level node and the root node, a corresponding hierarchical attenuation coefficient is introduced, and the deviation correlation degree of each lower-level node is calculated in combination with the deviation amplitude. The lower-level nodes are sorted from high to low according to the deviation correlation degree, and the nodes that are ranked first and whose deviation correlation degree is higher than the preset threshold are selected as candidate abnormal nodes.

[0021] The third aspect of the present invention proposes: A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the quality and health management method for new commercial vehicle projects as described above.

[0022] The fourth aspect of the present invention proposes: A readable storage medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the quality and health management method for new commercial vehicle projects as described above.

[0023] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0024] Figure 1 A flowchart of the method for managing the quality and health of new commercial vehicle projects provided in the first embodiment of the present invention; Figure 2 The structural block diagram of the commercial vehicle new project quality and health management system provided in the third embodiment of the present invention.

[0025] The following detailed description, in conjunction with the accompanying drawings, will further illustrate the present invention. Detailed Implementation

[0026] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Several embodiments of the invention are illustrated in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.

[0027] It should be noted that when a component is said to be "fixed to" another component, it can be directly on the other component or there may be an intervening component. When a component is said to be "connected to" another component, it can be directly connected to the other component or there may be an intervening component. The terms "vertical," "horizontal," "left," "right," and similar expressions used in this document are for illustrative purposes only.

[0028] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the specification of this invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0029] Please see Figure 1 The image shows a method for managing the quality and health of new commercial vehicle projects provided in the first embodiment of the present invention. This method can combine a graded early warning mechanism with a multi-dimensional traceability algorithm to locate the root cause of quality anomalies. It can shift quality control from passive rectification after the fact to proactive prevention and control before the fact, fully expose quality hazards in the development process, reduce the recurrence of similar problems, and effectively support the systematic and intelligent quality control of the entire life cycle of automobile projects.

[0030] Specifically, this embodiment provides: A method for quality and health management of new commercial vehicle projects, wherein the method includes: Step S10: Collect raw quality data of new vehicle projects to build a raw quality database for the project, and preprocess the raw quality database to generate a standard quality dataset. It should be noted that the project collects raw quality data from the entire lifecycle of new automotive projects. Data sources cover design review records, DV / PV bench test reports, and simulation analysis results from the R&D side; incoming inspection data and supplier quality ratings for components from the procurement side; first-pass yield rates, assembly process quality control data, and rework / repair records from the manufacturing side; road test failure lists and durability test results from the testing side; and early user feedback and warranty claim data from the market side. By integrating with multiple business systems such as PLM, MES, QMS, and DMS, a complete raw quality database is constructed, achieving unified collection of multi-source heterogeneous quality data. Based on this, the raw database undergoes preprocessing operations such as cleaning, standardization, verification, and completion to eliminate data noise, unify statistical definitions, complete missing fields, and remove outlier samples, generating a standardized quality dataset with a consistent structure and definitions. This step is the foundation of the entire health management system. Only when the data is consistent in scope and reliable in quality can the subsequent evaluation results have reference value. It fundamentally solves the pain points of traditional quality management, such as data being scattered across different departments, inconsistent statistical standards, inability to make horizontal comparisons, and difficulty in comprehensive evaluation. Taking this light truck project as an example, the original data was scattered across 11 business systems, with a total of 287 quality fields. After preprocessing, it was organized into 6 major categories and 32 core quality indicators, forming a unified standard dataset, which laid a solid data foundation for subsequent evaluation.

[0031] Step S20: The dimensionless standardization of all quality indicators in the standard quality dataset is completed using the function method, and the target weight of each quality indicator is calculated using the weighting method. It's important to note that the process begins with a function-based approach to standardize all quality indicators in the standard quality dataset using dimensionless methods. This eliminates differences in dimensions, magnitudes, and performance across different indicators, ensuring the additivity and comparability of various quality indicators such as pass rate, number of failures, and financial losses. Next, an objective weighting method is used to calculate the target weights for each quality indicator. Based on the inherent fluctuations and correlations of the data, this objectively differentiates the impact of different indicators on overall vehicle quality, avoiding evaluation biases and human interference caused by subjective expert weighting. Standardization and weighting are core prerequisites for quantitative assessment. They transform scattered quality indicators into a unified dimension that can be weighted and calculated, allowing multi-dimensional quality information to be integrated into a single health score, providing a scientific quantitative basis for subsequent comprehensive evaluation.

[0032] Step S30: Perform weighted solution processing on the target weights of each of the quality indicators to calculate the evaluation results, and determine the project quality status level based on the evaluation results; It's important to note that the weighted calculation process for each quality indicator involves weighting the target weights and then deeply integrating the standardized indicator scores with their corresponding weights to calculate the overall project quality health assessment result. Based on a preset threshold range, the corresponding project quality status level is determined, typically categorized into four levels: Excellent, Good, Warning, and Abnormal. This weighted assessment integrates multi-dimensional, fragmented quality information into a single health score, providing a comprehensive overview of the project's quality status. The level classification makes the quality status intuitive and easy to judge, allowing the project team to quickly grasp the overall quality situation and providing a clear and quantifiable basis for subsequent management decisions. For example, in the DV testing phase of this light truck project, the weighted health score was 76, corresponding to the "Warning" level, indicating a certain risk to the overall project quality and requiring close monitoring.

[0033] Step S40: Execute a graded early warning based on the project quality status level, and locate the root cause of the quality anomaly through a multi-dimensional tracing algorithm.

[0034] It's important to note that tiered early warnings are implemented based on the assessed project quality status level. Different levels correspond to different response levels, control intensities, and resource allocations, achieving differentiated and precise risk management. Simultaneously, a multi-dimensional source tracing algorithm decomposes quality indicators layer by layer from top to bottom, pinpointing the deep-seated root causes and specific deviation parameters of quality anomalies. This stage serves as the final exit point in the management loop. Tiered early warnings enable proactive risk control, nipping problems in the bud and preventing cost losses and delays caused by escalation. Precise source tracing provides clear direction for rectification, addressing the problem of "only issuing warnings, not providing solutions" in traditional quality management. This forms a complete management loop from risk discovery to root cause identification and rectification implementation.

[0035] Second Embodiment Furthermore, the steps of performing dimensionless standardization on all quality indicators in the standard quality dataset using a function method and calculating the target weights of each quality indicator using a weighting method include: An original evaluation matrix is ​​constructed based on the standard quality dataset, and positive and negative indicators are determined in the standard quality dataset based on the original evaluation matrix. The positive and negative indicators were normalized using a function method to obtain the standard efficacy score. Based on the standard efficacy score, the standard deviation, conflict coefficient, and information content of each quality indicator are calculated using a weighting method to obtain the target weight of each quality indicator.

[0036] It's important to note that an original evaluation matrix is ​​constructed based on a standard quality dataset. The rows of the matrix correspond to different statistical periods or sample batches, while the columns correspond to various quality indicators. This organizes the scattered quality data into a structured matrix form, providing a standard data structure for subsequent batch calculations. Furthermore, based on the quality attributes of the indicators, positive and negative indicators are clearly distinguished in the standard quality dataset: positive indicators, such as first-pass yield of vehicles, component test pass rate, mean time between failures (MTBF), design review pass rate, and supplier quality rating, represent higher quality; negative indicators, such as the number of failures per thousand vehicles, the number of defects per unit, rework and repair rate, quality loss costs, market complaint rate, and downtime, represent lower quality. Distinguishing between positive and negative indicators is a necessary prerequisite for normalization, as the two types of indicators have completely opposite directions of advantage and disadvantage. Corresponding normalization logic must be used to ensure the consistent meaning of the scores after standardization and to avoid evaluation distortion caused by directional inconsistencies. For example, if the negative number of thousands of faults is normalized using positive logic, the absurd result of more faults resulting in a higher score will occur, directly causing the evaluation to become completely invalid.

[0037] The efficacy coefficient method is used as the core function to normalize both positive and negative indicators, resulting in a standardized efficacy score that falls within the [0,1] interval. A higher score indicates better quality performance. For positive indicators, the industry benchmark value (satisfactory value) and the industry minimum value (unacceptable value) are used to linearly map the actual value to the efficacy score. The calculation formula is: Positive Efficacy Score = (Actual Value - Unacceptable Value) / (Satisfactory Value - Unacceptable Value). For negative indicators, the reverse mapping is used, with the calculation formula: Negative Efficacy Score = (Unacceptable Value - Actual Value) / (Unacceptable Value - Satisfactory Value). Through function-based normalization, quality indicators with different dimensions and magnitudes, such as percentage-based pass rates, number of defects, monetary losses, and mileage-based failure intervals, are all transformed into standardized efficacy scores with unified dimensions. This provides a comparable and weighted mathematical basis, completely solving the problem of directly calculating multiple types of quality indicators. Taking this light truck project as an example, the first-pass yield of the whole vehicle is a positive indicator, with a benchmark value of 95%, a minimum value of 75%, and a current actual value of 88%, resulting in an efficiency score of 0.65. The number of faults per thousand vehicles is a negative indicator, with a benchmark value of 12, a minimum value of 40, and a current actual value of 22, resulting in an efficiency score of 0.64. Since the scores of the two indicators are on the same dimension, they can be directly weighted.

[0038] Based on the calculated standard efficacy score, the CRITIC objective weighting method is used to calculate the standard deviation, conflict coefficient, and information content of each quality indicator. These three factors are then combined to obtain the target weight for each quality indicator. This weighting logic is entirely based on the statistical characteristics of the data itself, unlike subjective weighting methods such as the analytic hierarchy process (AHP) that rely on expert experience, effectively avoiding weight bias caused by human judgment. Specifically, the standard deviation reflects the degree of variation of the indicator; indicators with greater fluctuations have stronger discriminative power and better reflect the quality differences between different projects and stages, thus receiving higher weights. The conflict coefficient is calculated based on the correlation coefficient between indicators; indicators with lower correlation contain more independent information and less redundant information, resulting in higher weights. Information content combines the standard deviation and conflict coefficient to measure the overall contribution of the indicator. The weights calculated by combining these three factors objectively reflect the actual contribution of each quality indicator to the health assessment, making the assessment results more closely reflect the true quality status of the project. Taking this project as an example, the weight distribution of 32 indicators was calculated: the braking system failure rate had the highest weight at 0.092, because it fluctuates greatly, is strongly correlated with safety, and has a lot of independent information; the interior and exterior defects had the lowest weight at 0.021, because they fluctuate little and have a weak impact on the core quality of the whole vehicle; the core indicators such as power system reliability, transmission system failure rate, and frame strength all had high weights. The weight distribution fully conforms to the core priorities of commercial vehicle quality control, verifying the rationality of objective weighting.

[0039] Furthermore, the step of preprocessing the original quality database of the project to generate a standard quality dataset includes: The original quality data is mapped with stage labels according to the pre-set milestone nodes of commercial vehicle development, and quality data fragments corresponding to each development stage are generated. For each quality data segment, retrieve the confidence interval of the quality parameters of the benchmark project of the same stage and platform, and identify and remove abnormal data in the current segment based on the confidence interval; After removing outlier data, the field formats of each quality data segment are standardized and the units are aligned. The data is then categorized and reorganized according to the quality indicator categories to generate a standard quality dataset.

[0040] It's important to note that, following the pre-defined milestones in the GVDP (Geometric Development Process) for commercial vehicle development, including key milestones such as PA (Project Approval), CC (Completion Control) scheme freezing, prototype vehicle rollout, Design Verification (DV) completion, LR (Production Verification) completion, small-batch trial production, and LS (Start of Production) (SOP), the raw quality data is mapped with stage labels to generate corresponding quality data segments for each development stage. The quality characteristics of new commercial vehicle projects exhibit distinct stage-specific features: the R&D stage is primarily characterized by design defects and system compatibility issues, with failures concentrated at the principle and structural levels; the prototyping stage focuses on process adaptation and tooling accuracy issues, with failures concentrated at the dimensional deviation and assembly consistency levels; and the trial production stage primarily addresses component consistency and assembly stability issues, with failures concentrated at the component batch differences and personnel operation fluctuations levels. The quality benchmarks and control priorities differ significantly across stages. Mixing data from different stages for evaluation would severely distort the data and fail to reflect the true stage-specific quality level. By segmenting the data by milestones, quality data corresponds one-to-one with the development stage, allowing subsequent quality analysis to accurately match project progress and providing stronger management focus and guidance. For example, in the DV test phase of this light truck project, all the data in the segments were quality data related to bench tests and prototype production, which were strongly correlated with R&D and design, and the evaluation results directly guided design optimization; after entering the PV phase, the data switched to data related to production line trial assembly and reliability road tests, and the evaluation results guided the improvement of process and procurement quality.

[0041] For each quality data segment, the confidence interval of the quality parameters from the benchmark project on the same platform at the same stage is retrieved. Based on this confidence interval, outlier data in the current segment is identified and removed. Commercial vehicle development generally adopts a platform-based development strategy. Historical quality data from mature models on the same platform has been validated by a large number of samples and has strong reference value. The 95% confidence interval calculated based on large historical sample data represents the normal fluctuation range of this type of indicator at that stage. If a quality data point in the current project significantly exceeds the confidence interval, it is often due to abnormal factors such as data entry errors, occasional special operating conditions, or atypical test samples, and lacks general representativeness. Including such data in the evaluation would interfere with the objectivity of the overall health assessment. Filtering outliers through the benchmark confidence interval effectively purifies the dataset, eliminates non-systematic interference, and allows the evaluation results to truly reflect the normal quality level of the project. Taking this project as an example, the 95% confidence interval for the number of failures in the CC phase of the previous generation of light trucks on the same platform was [14, 26]. The number of failures in a certain batch of test data in the current project was 42. After verification, this batch was a special test sample for extreme working conditions, which is not a routine verification working condition. Therefore, it was judged as abnormal data and removed to avoid raising the overall failure level and causing the health score to be falsely low. In addition, the pass rate data of a certain batch of parts was marked as 120%, which was an input error. It was also removed as an outlier to ensure the accuracy of the dataset.

[0042] After removing outlier data, the quality data segments were standardized in terms of field format and dimensional alignment: Data coding rules were unified, mapping material codes and indicator names from different systems to a unified standard; the statistical period was standardized, clarifying whether it was daily, weekly, or monthly; and the unit of measurement was standardized, unifying heterogeneous units submitted by different suppliers and departments into industry standard units, eliminating data format differences between different departments, suppliers, and systems. Then, according to quality indicator categories, the data was categorized and reorganized into six major categories: design quality, component incoming quality, vehicle assembly quality, testing and verification quality, and market feedback quality, ultimately generating a well-structured standard quality dataset. Field standardization addressed the pain point of inconsistent data dimensionality from multiple sources. For example, defect data submitted by different suppliers might be counted by "number of pieces" or "number of times," but standardizing it to "defects per thousand pieces" allows for cross-supplier horizontal comparisons. Categorization and reorganization resulted in a clear and well-organized data structure, facilitating subsequent batch calling and weight calculation by category in the evaluation model. After this processing step, the originally scattered, heterogeneous, and noisy raw data is transformed into a standard dataset with standardized structure, unified caliber, and reliable quality, laying a solid data foundation for subsequent health assessments.

[0043] Furthermore, the step of performing a weighted solution process on the target weights of each of the quality indicators to calculate the evaluation result includes: Based on the vehicle BOM hierarchy and quality transfer path, a transmission correlation matrix between various quality indicators is constructed; Based on the aforementioned transmission correlation matrix, the target weights of each quality indicator are transmitted and corrected to obtain the corrected weights corresponding to each quality indicator. The dimensionless standardized values ​​of each quality indicator are weighted and summed with their corresponding corrected weights to obtain the final evaluation result.

[0044] It's important to note that a transmission correlation matrix is ​​constructed based on the vehicle's BOM hierarchy and quality transmission path, linking various quality indicators. Commercial vehicles are typical hierarchical complex mechanical systems, exhibiting a bottom-up hierarchical transmission characteristic for quality risks: quality defects in lower-level components propagate upwards to the assembly system, and problems in the assembly system further propagate to the vehicle level, ultimately manifesting as anomalies in overall vehicle quality indicators. Traditional weighted assessments treat each indicator as an independent entity, ignoring this transmission effect and easily underestimating the impact of critical defects at the lower levels on overall vehicle quality, leading to inflated health scores that fail to accurately reflect potential risks. The transmission correlation matrix, based on the BOM's hierarchical relationships and the strength of quality transmission, quantifies the influence coefficients between indicators at different levels, clearly depicting the quality transmission path from "components—assemblies—systems—vehicle," providing a quantitative structural basis for subsequent weight adjustments. Taking the braking system of this light truck project as an example, the constructed transmission matrix contains a three-level transmission relationship: the friction coefficient deviation of the brake pads in the bottom layer is transmitted to the braking performance index of the middle layer brake assembly, with a transmission coefficient of 0.78; the braking performance deviation of the brake assembly is transmitted to the braking distance index of the top layer vehicle, with a transmission coefficient of 0.82; at the same time, the brake pad deviation is also indirectly transmitted to multiple upper-level indicators such as braking noise and brake wear, resulting in a significant comprehensive transmission effect. Each transmission coefficient in the matrix is ​​calibrated based on historical fault data and engineering experience, quantifying the degree of influence of lower-level indicators on upper-level indicators.

[0045] Based on the transmission correlation matrix, the original target weights of each quality indicator are corrected to obtain the corrected weights for each quality indicator. The core logic of the correction is: if a quality problem in a lower-level indicator will propagate upwards and cause anomalies in multiple higher-level indicators, then its actual quality impact is far greater than its original weight, and the corrected weight needs to be increased accordingly; conversely, if an indicator only affects local functions and has weak transmission, the corrected weight is appropriately reduced. For example, the brake pad friction coefficient indicator has an original weight of only 0.048, but it not only affects the braking performance of the braking assembly, but also propagates and affects multiple higher-level indicators such as vehicle braking distance, braking stability, braking noise, and brake pad wear life. The comprehensive transmission effect amplifies its quality importance. After correction by the transmission matrix, the weight increases to 0.112, more than doubling its weight proportion, and more realistically reflects the actual impact of this indicator on the overall vehicle quality and safety. Another example is the interior storage box size deviation indicator, which only affects the interior appearance and has no further transmission effect. After correction, the weight is reduced from 0.023 to 0.018. The weights after transmission correction are no longer isolated single-indicator weights, but rather comprehensive weights that incorporate hierarchical transmission effects, which better reflect the true transmission pattern of vehicle quality risks.

[0046] The dimensionless standardized values ​​of each quality indicator are multiplied one by one by their corresponding corrected weights, and then the results are weighted and summed to obtain the final project quality health assessment result. The score is usually converted to a percentage system, which is more in line with daily management practices. The weighted calculation after transmission correction retains the fairness of the original objective weighting while incorporating the objective laws of quality hierarchy transmission, avoiding the limitation of the traditional weighted method of "seeing the trees but not the forest," and allowing the final health score to more accurately reflect the true quality risk level of the project. Taking the CC stage of this light truck project as an example, after the standardized scores of 32 indicators are weighted and summed with the corrected weights, the final health score is 76 points, corresponding to the "warning" level. If the original weights are used without transmission correction, the score will be artificially inflated to 81 points, corresponding to the "good" level, which will mask the potential risks of underlying components. The comparison shows that transmission correction effectively avoids the underestimation of risk and provides a more reliable quantitative basis for subsequent level determination and warning decisions.

[0047] Furthermore, the step of executing a graded early warning based on the project quality status level includes: By continuously collecting project evaluation results for multiple preset statistical periods, the current trend and rate of change in project quality and health are obtained through fitting. By combining the project quality status level with the rate of change, a corresponding early warning level is matched; Based on the matched early warning level, initiate the corresponding level of quality control response process.

[0048] It's important to note that project evaluation results are collected continuously over multiple preset statistical periods, typically on a weekly basis. Weekly health scores for 4 to 6 consecutive weeks are extracted, and linear or exponential curve fitting is used to obtain the current trend and rate of change in project quality health. A single-period health score only reflects the current static quality status and cannot reflect the overall quality trend. Quality risks in project development often evolve gradually: some projects may currently be in acceptable condition, but their quality deteriorates rapidly, and without timely intervention, they will soon slip into the abnormal range; other projects may have low current scores but are on a continuous improvement trend, with overall risk manageable. Introducing trends and rates of change expands the evaluation perspective from "single-point status" to "dynamic trends," enabling a more comprehensive grasp of the evolution of project quality and providing data support for early warning. Taking the end of the CC phase of this light truck project as an example, the health scores for the six consecutive weeks were 85, 83, 81, 79, 77, and 76, respectively. Linear fitting revealed a rate of change of 1.8 points per week, showing a continuous and rapid deterioration trend. If we only look at the warning level of 76 points for a single week, it is easy to think that the risk is still far away. However, considering the rate of deterioration, it can be predicted that if no intervention is taken, it will fall into the abnormal range in 3 weeks, and control measures must be initiated in advance.

[0049] Combining the project's quality status level and rate of change, a warning level matrix is ​​used to match corresponding warning levels. The warning judgment abandons the traditional single-threshold alarm mode, adopting a comprehensive two-dimensional judgment: for static levels already in the abnormal range, a high-level warning is directly triggered; for static levels still within the normal range, but the rate of deterioration exceeds a preset threshold, an escalation warning is triggered, issuing an early alert; for good status and continuous improvement, a low-level concern is maintained. This mechanism breaks the lag of traditional alarms, capturing early signals of quality deterioration and truly achieving early detection and early warning of risks. The project divides warning levels into four levels: Level 1 (General Concern), Level 2 (Yellow Warning), Level 3 (Orange Warning), and Level 4 (Red Warning), with corresponding judgment matrices: Excellent and stable or improving health status is Level 1 Concern; good health status but rapidly deteriorating, or health status warning and slowly deteriorating, is Level 2 Yellow Warning; health status warning and rapidly deteriorating, or health status abnormal and slowly deteriorating, is Level 3 Orange Warning; abnormal health status and rapidly deteriorating, is Level 4 Red Warning. In the above case, the health score of 76 points is at the warning level, and the rate of deterioration is relatively fast. Therefore, it was matched as a level three orange warning, which was issued two weeks earlier than the single-dimensional judgment.

[0050] Based on the matched early warning level, the corresponding quality control response process is initiated. Different levels correspond to different response entities, meeting mechanisms, rectification requirements, and resource allocations to ensure that risks and resources are matched. Level 1 (General Attention) is monitored by the project quality team, with weekly reviews of quality trends; no specific rectification is required. Level 2 (Yellow Warning) involves the project manager leading a special quality meeting to develop a rectification plan, clarify responsibilities and closing points, and track rectification progress every two weeks. Level 3 (Orange Warning) involves the business unit's quality director, coordinating cross-departmental resources to promote rectification, holding weekly progress meetings, and including it in the project's key supervision items. Level 4 (Red Warning) triggers a project-level quality halt mechanism, suspending relevant milestone reviews, establishing a special task force, and resuming work only after a thorough review and complete resolution of the problem. This tiered response mechanism ensures that early warnings are not merely vague reminders but deeply integrated into the company's management processes, ensuring that different levels of risk are matched with corresponding control resources. This avoids both overreacting to minor issues and wasting management resources, as well as preventing insufficient response to high-risk issues from escalating the problem. After the orange alert was triggered, the business unit's quality director immediately organized a special meeting with the R&D, procurement, and process departments. They formulated 12 corrective measures for the core issues of the braking and transmission systems, and made it clear that optimization and verification would be completed within two weeks. This effectively curbed the downward trend in quality and avoided project delays.

[0051] Furthermore, the step of locating the root cause of quality anomalies using a multidimensional source tracing algorithm includes: Using the quality indicator that triggers the anomaly as the top-level root node, the quality indicator is decomposed from top to bottom according to the hierarchy of whole vehicle-assembly-parts-process to form a quality indicator decomposition tree; The deviation correlation between the indicators of each lower-level node in the decomposition tree and the root node anomaly is calculated sequentially, and lower-level nodes with deviation correlation higher than the preset threshold are selected as candidate anomaly nodes. By retrospectively verifying the production process data and testing data corresponding to each candidate abnormal node, the root cause node of the quality abnormality and the corresponding deviation parameter are located.

[0052] It's important to note that the quality indicator decomposition tree is constructed by using the top-level quality indicator that triggers the anomaly as the root node and breaking it down layer by layer from top to bottom according to the hierarchy of "vehicle—system—assembly—component—process." Commercial vehicles are typical hierarchical and complex systems; a quality anomaly in the vehicle inevitably corresponds to a problem in one or more assemblies or components at the lower level, ultimately traceable to a specific production or assembly process. The decomposition tree is built based on the vehicle's BOM structure and the quality formation path, clearly presenting the hierarchical relationship between top-level anomalies and lower-level factors. This provides a clear decomposition path for tracing the source, avoiding aimless, comprehensive investigations. Taking the "vehicle braking distance exceeding the standard" anomaly in this project as an example, the constructed decomposition tree has four layers: the top root node is the vehicle braking distance exceeding the standard; the first layer, system level, is decomposed into four major branches: braking system, tire system, suspension system, and load distribution; the second layer, assembly level, further decomposes the braking system into brake caliper assembly, brake disc assembly, brake pad assembly, brake pipeline assembly, and brake booster assembly; the third layer, component level, decomposes the brake pad assembly into friction block, steel backing, muffler, and accessories; the fourth layer, process level, decomposes the friction block into material preparation process, hot pressing process, heat treatment process, grinding process, and inspection process. The complete decomposition tree covers all possible influencing factors from the top-level symptom to the bottom-level process, providing a complete investigation framework for subsequent layer-by-layer screening.

[0053] The correlation between the indicators of each lower-level node in the decomposition tree and the root node anomaly is calculated sequentially. Lower-level nodes with a correlation higher than a preset threshold are selected as candidate anomaly nodes. The core purpose of calculating the correlation layer by layer is to quickly identify high-suspicious objects from numerous lower-level nodes, significantly narrowing the scope of investigation. The number of parts and processes in a complete vehicle project is enormous, often numbering in the tens of thousands. Investigating them one by one would be extremely inefficient. By using correlation metric sorting, nodes with large deviations and strong correlations can be prioritized, greatly improving traceability efficiency. In cases of excessive braking distance, the correlation degree of the four system nodes in the first layer was calculated: braking system 0.88, tire system 0.23, suspension system 0.17, and load distribution 0.12. With a threshold of 0.4, only the braking system was considered for further analysis. The correlation degree of the subordinate assembly nodes of the braking system was then calculated: brake pad assembly 0.82, brake caliper assembly 0.35, brake disc assembly 0.31, piping assembly 0.24, and booster assembly 0.29. Again, only the brake pad assembly was considered. Further analysis down to the component and process level revealed two highly correlated candidate nodes: the friction block heat treatment process and the friction material formulation. Through these three layers of screening, the scope of investigation was narrowed from hundreds of factors to just two core candidates, improving efficiency by tens of times.

[0054] The production process data and testing data corresponding to each candidate anomaly node were backtracked and verified. Original data such as process parameters, incoming inspection records, batch traceability information, and personnel and equipment records were retrieved to verify the causal relationship between each candidate node and the root node anomaly, ultimately pinpointing the root cause of the quality anomaly and its corresponding deviation parameters. Correlation screening is only a preliminary judgment based on data correlation; final confirmation requires causal verification using actual process data to avoid misinterpreting data correlation as causation and to ensure the accuracy of the tracing conclusion. In the above case, for the candidate node in the heat treatment process, the production process records and incoming inspection reports of the batch of brake pads were retrieved backtracked. It was found that the sintering temperature of this batch of heat treatment process was 15°C lower than the process requirement, and the holding time was 8 minutes shorter, resulting in a lower friction coefficient of the friction block, with the measured value being 28% lower than the design value. Simultaneously, bench test data verified that a lower friction coefficient directly leads to decreased braking performance and increased braking distance, with the deviation magnitude and the excessive braking distance magnitude highly matched, and the timeline also perfectly aligned. The investigation ultimately determined that insufficient temperature and inadequate holding time during the brake pad heat treatment process led to a substandard friction coefficient, which was the root cause of the vehicle's excessive braking distance. After identifying the cause, the purchasing department immediately required the supplier to rectify the heat treatment process and to rework and screen already delivered batches, quickly closing the quality issue in a closed loop. This two-step approach of "preliminary screening – retrospective verification" balanced traceability efficiency with accurate identification, quickly pinpointing the true cause of the quality anomaly and providing precise direction for developing a rectification plan.

[0055] Furthermore, the step of sequentially calculating the deviation correlation degree between the indicators of each lower-level node in the decomposition tree and the root node anomaly, and selecting lower-level nodes with deviation correlation degrees higher than a preset threshold as candidate anomaly nodes includes: Extract the quality data of each lower-level node indicator within two preset time windows before and after the anomaly occurs, and calculate the deviation of the corresponding indicator. Based on the hierarchical distance between each lower-level node and the root node, a corresponding hierarchical attenuation coefficient is introduced, and the deviation correlation degree of each lower-level node is calculated in combination with the deviation amplitude. The lower-level nodes are sorted from high to low according to the deviation correlation degree, and the nodes that are ranked first and whose deviation correlation degree is higher than the preset threshold are selected as candidate abnormal nodes.

[0056] It should be noted that the quality data of each lower-level node indicator is extracted within two preset time windows before and after the anomaly occurs. Using the normal period data before the anomaly as a benchmark, the deviation amplitude of the corresponding indicator during the anomaly period is calculated. The deviation amplitude reflects the degree of quality deviation of the node itself; the larger the deviation, the more significant the anomaly of the node and the stronger the potential correlation with the top-level anomaly. The time window setting needs to match the quality propagation cycle, ensuring both the inclusion of the complete anomaly evolution process and avoiding the introduction of too much interference data from irrelevant periods. The general setting for this project is: 4 weeks before the anomaly trigger is the normal benchmark window, and 2 weeks after the anomaly occurs is the anomaly comparison window, ensuring sufficient data volume and reflecting changes before and after the anomaly. In the case of excessive braking distance, for the brake pad friction coefficient indicator, the mean value of the normal benchmark window is 0.38, and the mean value of the anomaly window is 0.27. The deviation amplitude is calculated using the formula |abnormal mean value - normal mean value| / normal mean value, yielding a deviation amplitude of 28.9%. For the brake disc flatness indicator, the mean value of the normal window is 0.02mm, and the mean value of the anomaly window is 0.021mm, with a deviation amplitude of only 5%. The difference in deviation magnitude directly reflects the degree of anomaly of different nodes and is the basis for correlation calculation.

[0057] Based on the hierarchical distance between each lower-level node and the root node, a corresponding hierarchical attenuation coefficient is introduced, and the deviation correlation degree of each lower-level node is calculated in conjunction with the deviation amplitude. The core logic of hierarchical attenuation is that the closer the level is to the root node, the stronger the direct correlation with the top-level anomaly, the fewer intermediate transmission links, and the lower the uncertainty; the lower the level, the more intermediate transmission links, and the greater the uncertainty of indirect correlation. Therefore, it is necessary to correct the correlation degree through attenuation coefficients to avoid the local deviation of the lower-level nodes being excessively amplified as the main cause of the top-level anomaly. The attenuation rules set for this project are as follows: system-level nodes 1 level away from the root node, attenuation coefficient 0.9; assembly-level nodes 2 levels away, attenuation coefficient 0.7; component-level nodes 3 levels away, attenuation coefficient 0.55; and process-level nodes 4 levels away, attenuation coefficient 0.4. The correlation degree after hierarchical attenuation correction considers both the deviation degree of the node itself and the transmission uncertainty brought about by hierarchical distance. The correlation degree calculation is more in line with the actual quality transmission law and effectively avoids the misjudgment of "small deviation at the bottom level, high correlation degree". Taking the aforementioned brake pad friction coefficient as an example, it belongs to the component level, is 3 layers away from the root node, has an attenuation coefficient of 0.55, and a deviation amplitude of 0.289. The calculated deviation correlation degree is: deviation amplitude × (1 / layer distance correction factor) × attenuation coefficient, which finally yields 0.82, matching the actual correlation strength. If layer attenuation is not performed, directly using the deviation amplitude will overestimate the direct correlation degree of the bottom node, leading to a deviation in the tracing direction.

[0058] The lower-level nodes are sorted from highest to lowest based on their deviation correlation, and nodes with the highest deviation correlation and a preset threshold are selected as candidate anomaly nodes. This sorting and filtering mechanism prioritizes the nodes with the highest correlation in the source tracing work, conducting investigations in a "high-to-low, step-by-step verification" order. This minimizes the investigation path and improves efficiency while ensuring accuracy. A fallback threshold is set to prevent nodes with very low correlation from entering the candidate range, reducing ineffective investigations. In the assembly level of the braking distance case, the correlation of the five nodes from highest to lowest is as follows: brake pad assembly 0.82, brake caliper assembly 0.35, brake disc assembly 0.31, brake booster 0.29, and brake lines 0.24. With a correlation threshold of 0.4, only the brake pad assembly is included as a candidate, while the other nodes are temporarily excluded, directly reducing the investigation scope by 80%. If subsequent verification shows that the brake pads are not the root cause, the threshold is gradually relaxed to include the next highest-corresponding nodes, achieving a balance between efficiency and accuracy. The candidate node list obtained through this calculation logic can accurately pinpoint the most likely root cause of the anomaly, providing clear priority guidance for subsequent backtracking and verification. This significantly improves the operational efficiency and accuracy of the entire traceability process and enables the entire quality and health management system to form a complete management closed loop of "assessment-early warning-traceability-rectification," providing a systematic and implementable end-to-end solution for quality control of new commercial vehicle projects.

[0059] Please see Figure 2 The third embodiment of the present invention provides: A quality and health management system for new commercial vehicle projects, wherein the system includes: The processing module is used to collect raw quality data of new automobile projects to build a raw quality database for the project, and to preprocess the raw quality database to generate a standard quality dataset. The calculation module is used to perform dimensionless standardization of all quality indicators in the standard quality dataset using a function method, and to calculate the target weight of each quality indicator using a weighting method. The evaluation module is used to perform weighted solution processing on the target weights of each of the quality indicators to calculate the evaluation results and determine the project quality status level based on the evaluation results. The source tracing module is used to execute graded early warnings based on the project's quality status level and to locate the root cause of quality anomalies through a multi-dimensional source tracing algorithm.

[0060] Furthermore, the steps of performing dimensionless standardization on all quality indicators in the standard quality dataset using a function method and calculating the target weights of each quality indicator using a weighting method include: An original evaluation matrix is ​​constructed based on the standard quality dataset, and positive and negative indicators are determined in the standard quality dataset based on the original evaluation matrix. The positive and negative indicators were normalized using a function method to obtain the standard efficacy score. Based on the standard efficacy score, the standard deviation, conflict coefficient, and information content of each quality indicator are calculated using a weighting method to obtain the target weight of each quality indicator.

[0061] Furthermore, the step of preprocessing the original quality database of the project to generate a standard quality dataset includes: The original quality data is mapped with stage labels according to the pre-set milestone nodes of commercial vehicle development, and quality data fragments corresponding to each development stage are generated. For each quality data segment, retrieve the confidence interval of the quality parameters of the benchmark project of the same stage and platform, and identify and remove abnormal data in the current segment based on the confidence interval; After removing outlier data, the field formats of each quality data segment are standardized and the units are aligned. The data is then categorized and reorganized according to the quality indicator categories to generate a standard quality dataset.

[0062] Furthermore, the step of performing a weighted solution process on the target weights of each of the quality indicators to calculate the evaluation result includes: Based on the vehicle BOM hierarchy and quality transfer path, a transmission correlation matrix between various quality indicators is constructed; Based on the aforementioned transmission correlation matrix, the target weights of each quality indicator are transmitted and corrected to obtain the corrected weights corresponding to each quality indicator. The dimensionless standardized values ​​of each quality indicator are weighted and summed with their corresponding corrected weights to obtain the final evaluation result.

[0063] Furthermore, the step of executing a graded early warning based on the project quality status level includes: By continuously collecting project evaluation results for multiple preset statistical periods, the current trend and rate of change in project quality and health are obtained through fitting. By combining the project quality status level with the rate of change, a corresponding early warning level is matched; Based on the matched early warning level, initiate the corresponding level of quality control response process.

[0064] Furthermore, the step of locating the root cause of quality anomalies using a multidimensional source tracing algorithm includes: Using the quality indicator that triggers the anomaly as the top-level root node, the quality indicator is decomposed from top to bottom according to the hierarchy of whole vehicle-assembly-parts-process to form a quality indicator decomposition tree; The deviation correlation between the indicators of each lower-level node in the decomposition tree and the root node anomaly is calculated sequentially, and lower-level nodes with deviation correlation higher than the preset threshold are selected as candidate anomaly nodes. By retrospectively verifying the production process data and testing data corresponding to each candidate abnormal node, the root cause node of the quality abnormality and the corresponding deviation parameter are located.

[0065] Furthermore, the step of sequentially calculating the deviation correlation degree between the indicators of each lower-level node in the decomposition tree and the root node anomaly, and selecting lower-level nodes with deviation correlation degrees higher than a preset threshold as candidate anomaly nodes includes: Extract the quality data of each lower-level node indicator within two preset time windows before and after the anomaly occurs, and calculate the deviation of the corresponding indicator. Based on the hierarchical distance between each lower-level node and the root node, a corresponding hierarchical attenuation coefficient is introduced, and the deviation correlation degree of each lower-level node is calculated in combination with the deviation amplitude. The lower-level nodes are sorted from high to low according to the deviation correlation degree, and the nodes that are ranked first and whose deviation correlation degree is higher than the preset threshold are selected as candidate abnormal nodes.

[0066] The fourth embodiment of the present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the quality and health management method for new commercial vehicle projects as described above.

[0067] The fifth embodiment of the present invention provides a readable storage medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the quality and health management method for new commercial vehicle projects as described above.

[0068] In summary, the commercial vehicle new project quality health management method and system provided by the above embodiments of the present invention can combine a hierarchical early warning mechanism and a multi-dimensional traceability algorithm to locate the root cause of quality anomalies. It can shift quality control from passive rectification after the fact to proactive prevention and control before the fact, fully expose quality hazards in the development process, reduce the recurrence of similar problems, and effectively support the systematic and intelligent quality control of the entire life cycle of automobile projects.

[0069] It should be noted that the above modules can be functional modules or program modules, and can be implemented through software or hardware. For modules implemented through hardware, the above modules can reside in the same processor; or the above modules can be located in different processors in any combination.

[0070] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0071] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0072] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0073] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0074] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the appended claims.

Claims

1. A method for managing the quality health of a new whole vehicle program for a commercial vehicle, characterized by, The method includes: Collect raw quality data from new automotive projects to construct a raw quality database for the projects, and preprocess the raw quality database to generate a standard quality dataset. The dimensionless standardization of all quality indicators in the standard quality dataset is performed using a function method, and the target weights of each quality indicator are calculated using a weighting method. The target weights of each of the quality indicators are weighted and solved to calculate the evaluation results, and the project quality status level is determined based on the evaluation results. A graded early warning is executed based on the project's quality status level, and the root cause of the quality anomaly is located through a multi-dimensional tracing algorithm.

2. The method of claim 1, wherein: The steps of performing dimensionless standardization of all quality indicators in the standard quality dataset using a function method and calculating the target weights of each quality indicator using a weighting method include: An original evaluation matrix is ​​constructed based on the standard quality dataset, and positive and negative indicators are determined in the standard quality dataset based on the original evaluation matrix. The positive and negative indicators were normalized using a function method to obtain the standard efficacy score. Based on the standard efficacy score, the standard deviation, conflict coefficient, and information content of each quality indicator are calculated using a weighting method to obtain the target weight of each quality indicator.

3. The method of claim 1, wherein, The step of preprocessing the original quality database of the project to generate a standard quality dataset includes: The original quality data is mapped with stage labels according to the pre-set milestone nodes of commercial vehicle development, and quality data fragments corresponding to each development stage are generated. For each quality data segment, retrieve the confidence interval of the quality parameters of the benchmark project of the same stage and platform, and identify and remove abnormal data in the current segment based on the confidence interval; After removing outlier data, the field formats of each quality data segment are standardized and the units are aligned. The data is then categorized and reorganized according to the quality indicator categories to generate a standard quality dataset.

4. The method of claim 1, wherein, The step of weighted solving of the target weights of each of the quality indicators to calculate the evaluation result includes: Based on the vehicle BOM hierarchy and quality transfer path, a transmission correlation matrix between various quality indicators is constructed; Based on the aforementioned transmission correlation matrix, the target weights of each quality indicator are transmitted and corrected to obtain the corrected weights corresponding to each quality indicator. The dimensionless standardized values ​​of each quality indicator are weighted and summed with their corresponding corrected weights to obtain the final evaluation result.

5. The method of claim 1, wherein, The step of executing a graded early warning based on the project quality status level includes: By continuously collecting project evaluation results for multiple preset statistical periods, the current trend and rate of change in project quality and health are obtained through fitting. By combining the project quality status level with the rate of change, a corresponding early warning level is matched; Based on the matched early warning level, initiate the corresponding level of quality control response process.

6. The method of claim 1, wherein, The steps for locating the root cause of quality anomalies using a multidimensional source tracing algorithm include: Using the quality indicator that triggers the anomaly as the top-level root node, the quality indicator is decomposed from top to bottom according to the hierarchy of whole vehicle-assembly-parts-process to form a quality indicator decomposition tree; The deviation correlation between the indicators of each lower-level node in the decomposition tree and the root node anomaly is calculated sequentially, and lower-level nodes with deviation correlation higher than the preset threshold are selected as candidate anomaly nodes. By retrospectively verifying the production process data and testing data corresponding to each candidate abnormal node, the root cause node of the quality abnormality and the corresponding deviation parameter are located.

7. The method for managing the quality and health of new commercial vehicle projects according to claim 6, characterized in that, The step of sequentially calculating the deviation correlation between the indicators of each lower-level node in the decomposition tree and the root node anomaly, and selecting lower-level nodes with deviation correlation higher than a preset threshold as candidate anomaly nodes includes: Extract the quality data of each lower-level node indicator within two preset time windows before and after the anomaly occurs, and calculate the deviation of the corresponding indicator. Based on the hierarchical distance between each lower-level node and the root node, a corresponding hierarchical attenuation coefficient is introduced, and the deviation correlation degree of each lower-level node is calculated in combination with the deviation amplitude. The lower-level nodes are sorted from high to low according to the deviation correlation degree, and the nodes that are ranked first and whose deviation correlation degree is higher than the preset threshold are selected as candidate abnormal nodes.

8. A quality and health management system for a new commercial vehicle project, characterized in that, The system includes: The processing module is used to collect raw quality data of new automobile projects to build a raw quality database for the project, and to preprocess the raw quality database to generate a standard quality dataset. The calculation module is used to perform dimensionless standardization of all quality indicators in the standard quality dataset using a function method, and to calculate the target weight of each quality indicator using a weighting method. The evaluation module is used to perform weighted solution processing on the target weights of each of the quality indicators to calculate the evaluation results and determine the project quality status level based on the evaluation results. The source tracing module is used to execute graded early warnings based on the project's quality status level and to locate the root cause of quality anomalies through a multi-dimensional source tracing algorithm.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method for managing the quality and health of new commercial vehicle projects as described in any one of claims 1 to 7.

10. A readable storage medium having a computer program stored thereon, characterized in that, When executed by the processor, the program implements the method for managing the quality and health of new commercial vehicle projects as described in any one of claims 1 to 7.