Intelligent quality management method, electronic device, and storage medium
By using intelligent quality management methods and electronic devices, and leveraging time-dependent risk priority index algorithms and AI algorithms, the problems of low efficiency and data clutter in traditional quality management have been solved. This has enabled automated auditing and problem handling, thereby improving production efficiency and customer satisfaction.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHENZHEN FULIAN FUGUI PRECISION INDUSTRY CO LTD
- Filing Date
- 2021-06-21
- Publication Date
- 2026-06-09
AI Technical Summary
Traditional quality management suffers from inefficiency due to paper records, disorganized data, untimely updates, and an inability to quickly identify repetitive and high-risk issues, leading to a lack of timely improvement and a lack of mechanisms for detecting repetitive problems, which affects customer satisfaction.
Intelligent quality management methods are adopted, using a time-dependent risk priority index algorithm to generate heat maps, formulate intelligent audit scheduling algorithms, and combine AI and machine learning algorithms to recommend solutions and preventive measures. The audit plan and problem handling are automated through electronic devices.
It has improved production management efficiency, reduced labor costs, enhanced customer satisfaction, enabled the rapid identification and resolution of quality issues, and achieved intelligent quality management.
Smart Images
Figure CN115577890B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to quality management methods, and more particularly to an intelligent quality management method and electronic device. Background Technology
[0002] In manufacturing, quality management plays an indispensable role in factory production. Generally, during quality control audits, audit items are first classified and categorized. Then, internal special audit plans are developed to monitor the quality status of the production line. Problems discovered are uploaded to the system in paper form or manually to record, track, and consider improvement, prevention, and subsequent improvement measures. This forms a Closed-Loop Corrective Action (CLCA) cycle, which not only ensures quality control but also continuously improves customer satisfaction.
[0003] Traditional audit processes rely on paper-based communication, resulting in issues such as "lack of integrated information management," "low efficiency," and "disorganized and outdated data." Furthermore, manually scheduling audit plans cannot quickly identify repetitive or high-risk issues or ensure accurate scheduling. There is also no mechanism for detecting repetitive issues, leading to numerous blind spots and an inability to promptly address recurring problems. Summary of the Invention
[0004] In view of the above, it is necessary to provide an intelligent quality management method, electronic device and storage medium that can reduce labor costs, improve management and production efficiency, effectively implement countermeasures to address pain points and improve customer satisfaction.
[0005] This invention provides an intelligent quality management method applied to electronic devices. The method includes: calculating demand data and parameter configurations based on audit-related data and a time-dependent risk priority index formula; generating a heat map risk interface based on the demand data and parameter configuration; formulating an intelligent audit scheduling algorithm through the heat map risk interface to automatically generate an audit plan; notifying audit units to check the audit time and audit team, and executing audit procedures and selecting multiple problem points according to the audit plan; providing intelligent root cause category suggestions for the multiple problem points; providing intelligent corrective and preventive action suggestions for the multiple problem points based on the root cause category suggestions to obtain optimal improvement and preventive measures; and implementing improvement actions for each audit unit according to the improvement actions to solve the problem points, and implementing preventive actions for each audit unit according to the preventive measures.
[0006] This invention also provides an electronic device, including a risk management module, an intelligent audit scheduling module, a processing module, an intelligent recommendation module, and an execution module. The risk management module calculates demand data and parameter configurations based on audit-related data and a time-dependent risk priority index formula, and generates a heatmap risk interface based on the demand data and parameter configurations. The intelligent audit scheduling module formulates an intelligent audit scheduling algorithm through the heatmap risk interface to automatically generate an audit plan. The processing module notifies the audit unit to check the audit time and audit team, and executes the audit procedure and selects multiple problem points according to the audit plan. The intelligent recommendation module provides intelligent root cause category suggestions for the multiple problem points, and provides intelligent corrective and preventive action suggestions for the multiple problem points based on the root cause category suggestions to obtain the best improvement and prevention strategies. The execution module executes improvement actions for each audit unit according to the improvement strategies to solve the problem points, and executes preventive actions for each audit unit according to the prevention strategies.
[0007] This invention also provides a storage medium storing a computer program, which, when executed, implements the steps of the aforementioned intelligent quality management method.
[0008] The intelligent quality management method, electronic device, and storage medium of this invention establish an intelligent audit recommendation system. It uses the Time Dependent Risk Priority Number (RPN) algorithm to highlight risk pain points. Combined with the time-dependent RPN, an original audit scheduling algorithm automatically generates audit plans. It uses AI Chinese-English word segmentation model, bag-of-words model, and text exploration algorithm combined with convolutional neural network algorithm and machine learning algorithm to recommend solutions and preventive measures to provide engineers with implementation and improvement. Attached Figure Description
[0009] Figure 1 This is a flowchart of the steps of the intelligent quality management method according to an embodiment of the present invention.
[0010] Figure 2 This is a flowchart of the steps of the intelligent auditing and scheduling plan method according to an embodiment of the present invention.
[0011] Figure 3 This is a schematic diagram of the intelligent recommendation implementation process according to an embodiment of the present invention.
[0012] Figure 4 This is a schematic diagram illustrating the data cleaning / preprocessing implementation flow of the intelligent recommendation algorithm according to an embodiment of the present invention.
[0013] Figure 5 This is a schematic diagram of the CA / PA ranking calculation logic in the intelligent recommendation algorithm of this invention.
[0014] Figure 6 This is a schematic diagram of the hardware architecture of the electronic device according to an embodiment of the present invention.
[0015] Figure 7 This is a functional block diagram of an electronic device according to an embodiment of the present invention.
[0016] Figure 8 This is a functional block diagram of the intelligent audit and scheduling module in an embodiment of the present invention.
[0017] Explanation of main component symbols
[0018] Electronic devices 200 processor 210 Memory 220 Intelligent Quality Management System 230 Heatmap Risk Management Module 310 Intelligent audit scheduling module 320 Processing module 330 Intelligent recommendation module 340 Execution module 350 Select Unit 3210 Computing unit 3220 Judgment Unit 3230 Setting unit 3240
[0019] The following detailed description, in conjunction with the accompanying drawings, will further illustrate the present invention. Detailed Implementation
[0020] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.
[0021] Numerous specific details are set forth in the following description to provide a thorough understanding of the invention. The described embodiments are merely some, not all, of the embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.
[0022] 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 description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.
[0023] It should be noted that the descriptions involving "first," "second," etc., in this invention are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, features defined with "first" or "second" may explicitly or implicitly include at least one of the stated features. Furthermore, the technical solutions of the various embodiments can be combined with each other, but only on the basis of being achievable by those skilled in the art. When the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by this invention.
[0024] The advantages of the intelligent quality management method in this invention are as follows:
[0025] (1) Establish a quality status heatmap risk platform: After calculating the factory's quality status using the Time Dependent Risk Priority Number (RPN) algorithm, the risk heatmap is displayed to enable managers to quickly grasp the on-site situation and target weaknesses.
[0026] (2) Develop intelligent audit scheduling plans: Based on recent factory weaknesses and the integration of periodic audit projects, audit scheduling algorithms automatically generate audit plans, enabling every aspect of production to be reviewed and improved; and
[0027] (3) Establish an intelligent recommendation system: Past quality management data is organized and revitalized using Artificial Intelligence (AI) natural language processing technology, and neural network technology is added to establish an intelligent quality module. This intelligent quality module learns from the experience of all past data and incorporates expert knowledge judgment, enabling it to intelligently recommend suitable solutions to quality problems.
[0028] The intelligent quality management method of this invention can reduce labor costs, improve management and production efficiency, effectively address pain points and implement countermeasures, and enhance customer satisfaction.
[0029] Figure 1 This is a flowchart of the intelligent quality management method according to an embodiment of the present invention, applied to electronic devices. The order of the steps in the flowchart can be changed, and some steps can be omitted, depending on different requirements.
[0030] Establish a Heatmap risk platform for quality status
[0031] This phase includes at least steps S11 and S12, which involve statistical analysis of audit data, recording quality risk issues within the factory into the system, and establishing a risk management interface to present risk hotspots using data covering the past seven periods plus the current period. The Heatmap risk platform can integrate problem categories and details from multiple factories, along with corresponding time-dependent RPNs and Process Failure Mode and Effects Analysis (PFMEA). It uses color-coding to display problem severity, Pareto charts to present the root causes of problems, and integrates statistical data using tags such as "Problem Category," "Manufacturing Division," and "Recurring Problems," allowing managers to grasp all quality risk issues within the group in the shortest possible time.
[0032] Step S11: Calculate the required data and parameter configuration based on audit-related data and the Time Dependent Risk Priority Number (RPN) formula. The Time Dependent RPN algorithm is constructed primarily using Python-Pandas, Python-Numpy, and C++, and an Application Programming Interface (API) service is built using Django.
[0033] Existing common RPN formulas only reflect the current severity of factory problems and cannot detect or punish items with higher past problem severity. Therefore, this invention incorporates time weights into the formula and designs severity-related parameters according to user needs to calculate the time-dependent RPN algorithm of this invention. The time-dependent RPN algorithm formula of this invention is as follows:
[0034]
[0035] Where Security(S) represents severity, Occurrence(O) represents incidence, Detection(D) represents detectability, i represents quarters (i = 1, ..., 8), N represents questions (j = 1, ..., N), and k represents event types (k = 1, ..., K). i,j (Event Type), where m is the number of quality events, m = 1,...,M i,j,k .
[0036] For example, in the third quarter of fiscal year FY21, the table shows that issue 4.7.1 corresponds to 2 quality incidents, with the incident type being Quality Incident, and m = 2. In the fourth quarter of fiscal year FY21, the table shows that issue 4.7.3 corresponds to 3 quality incidents, with the incident type being Internal Progressive Audit, and m = 3.
[0037] The time-dependent RPN algorithm combines severity, occurrence rate, detectability, and PFMEA score to form an original algorithm that achieves an innovative indicator that can weigh the severity of a problem based on "time relationship", "problem category", "number of occurrences" and "risk".
[0038] Step S12: Generate a heatmap risk interface based on the required data and parameter configuration. The heatmap interface and other visualization interfaces are constructed by connecting the database PostgreSQL and front-end web application frameworks such as HTML, CSS, and Vue.js via API.
[0039] Develop intelligent audit scheduling plan
[0040] This stage includes at least step S13.
[0041] Step S13: Develop an intelligent audit scheduling algorithm through the heat map risk interface to automatically generate an audit plan.
[0042] Intelligent audit scheduling algorithms are developed based on Heatmap quality information, customer audit specifications, and innovative time-dependent RPN algorithms to automatically generate audit plans.
[0043] Figure 2 This is a flowchart of the steps of the intelligent auditing and scheduling plan method according to an embodiment of the present invention.
[0044] Step S201: Select N audit units and M problem points that need to be audited in the current time period.
[0045] Step S202: Calculate the RPN values for K problems, where the K problems include the current time period plus the previous seven time periods.
[0046] Step S203: Calculate the RPN value of the problem point for each audit unit.
[0047] Step S204: Select the audit unit whose RPN value ranks before X.
[0048] Step S205: Determine whether the current time period is the first week of the cycle.
[0049] Step S206: If it is not the first week in the cycle, determine whether there are duplicate audit units selected in step S204.
[0050] In step S207, if there are duplicate audit units, the duplicate audit units are removed, and then the process returns to step S204.
[0051] If in step S205 the current time period is the first week of the cycle, or if in step S206 there are no duplicate audit units, then the selection of N audit units is complete. Next, from the M problem points, select the top Y largest RPN problem points (step S208) and the next Z largest RPN problem points (step S209), where Y + Z = M. The top Y largest problem points are the most serious; duplicate problem points are acceptable, and they must be audited weekly. The next Z largest problem points are the next most serious, and the principle is to avoid duplicate audits within a single round.
[0052] Step S210: Determine whether the current time period is the first week of the cycle.
[0053] Step S211: If it is not the first week of the cycle, determine whether there are duplicate problem points among the selected problem points.
[0054] In step S212, if there are duplicate problem points, remove the duplicate problem points and then return to step S208.
[0055] Step S213: After selecting the top Y largest RPN problem points in step S208, or if the current time period is the first week of the cycle in step S210, or if there are no duplicate problem points in step S211, then the selection of M problem points is complete. Next, it is determined whether a quality incident has occurred within a preset time (e.g., one month).
[0056] Step S214: If a quality event occurs, the selected problem point is set as a cross-audit unit project.
[0057] Step S215: If no quality event occurs, assign L problem points with higher RPN values to each audit unit.
[0058] Step S216: Proceed to the audit for the next time period.
[0059] Step S14: Notify the auditing unit to check the audit time and audit team, and execute the audit procedures and select problem points according to the audit plan, including: generating new quality cases in the database, filling in descriptions of finding and identifying root causes, etc.
[0060] Intelligent recommendation algorithm
[0061] This stage includes at least steps S15 and S16.
[0062] Step S15: Provide intelligent root cause category suggestions for the problem points.
[0063] Step S16: Based on the root cause category, intelligent corrective and preventive action suggestions are made for the multiple problem points to obtain the best improvement and prevention strategies.
[0064] Figure 3 This is a schematic diagram illustrating the implementation process of intelligent recommendation in an embodiment of the present invention.
[0065] The improvement and prevention strategies for historical audit data are stored in a "textual description" and "mixed Chinese and English" format. Therefore, this embodiment of the invention uses Chinese-English word segmentation model technology combined with a custom dictionary as the basis, and then uses bag-of-words model combined with text exploration algorithm to keep the training data in the best state. Finally, the model is trained and classified through convolutional neural network, deep learning algorithm and machine learning algorithm, and finally the best improvement and prevention strategies are recommended.
[0066] First, data cleaning and preprocessing operations are performed on the raw data of quality issues, including data labeling, Chinese-English word segmentation model, and the establishment of a bag of quality keywords.
[0067] Figure 4 This is a schematic diagram illustrating the data cleaning / preprocessing implementation flow of the intelligent recommendation algorithm according to an embodiment of the present invention.
[0068] Word segmentation for a mixed Chinese-English quality database is a complex problem. This invention first employs a Hidden Markov Model (HMM) for word segmentation, adjusting the segmentation results and removing redundant words to establish a unique quality dictionary. Based on expert knowledge, the most suitable quality Chinese-English word segmentation model is obtained; finally, word segmentation is performed again using this model, adjusting the results and removing redundant words to establish the final quality bag-of-vocabulary model.
[0069] Based on each past quality issue record, standardized categories are established according to the issue type, the work section where the issue occurred, and the root cause of the issue. Then, each issue is assigned a corresponding label, so that historical data can be classified and a semantic classification database is established.
[0070] Regarding the training model update in intelligent recommendation algorithms, the present invention selects the best-performing Convolutional Neural Network (CNN) model as the basis for prediction and classification after experiments and tests. It also incorporates preprocessing and parameter tuning training, and designs various different parameter combinations to retrain and adjust the model to improve the accuracy of prediction.
[0071] In the Convolutional Neural Network (CNN) architecture, this invention employs two convolutional layers paired with a global max pooling layer, preserving key features while reducing parameters and computational cost. Simultaneously, a fully connected (dense) layer is used to weight and sum the previously designed features, mapping the learned "distributed feature representation" onto the sample label space, and then followed by the Sofmax function to present the scores of each category probabilistically. The cross-entropy function is used on the model's loss function to calculate the distance between the predicted probability distribution and the probability distribution of the true answer, optimizing the model's predictions. The optimization function employs a modified gradient descent method, namely the Adam algorithm, dynamically adjusting the learning rate of each parameter based on the first-order and second-order gradient matrix estimates of each parameter from the loss function. Furthermore, momentum (the moving average of the parameters) is used to improve traditional gradient descent, promoting dynamic adjustment of hyperparameters.
[0072] After adjusting the parameters, the accuracy was further improved to 91.01%. This model was ultimately used as the prediction category for quality problems and became an important category basis for recommending corrective actions (CA) and preventive actions (PA) in the future.
[0073] Regarding the CA / PA ranking calculation in model training of intelligent recommendation algorithms, improvement strategies (CA) and prevention strategies (PA) are the most important parts of recommendation. They can not only activate past experience in handling problems, but also guide newcomers to take countermeasures to deal with problems.
[0074] The key to this ranking calculation is to identify the descriptions most similar to the current problem from a database of past countermeasures and recommend corresponding countermeasures. Therefore, in this embodiment of the invention, evaluation formulas for Improvement Countermeasures (CA) and Prevention Countermeasures (PA) are developed, giving lower weight to problems with high recurrence rates and higher weight to countermeasures that can be effectively solved. Furthermore, the evaluation scores of CA and PA by professionals in the relevant fields are incorporated, and the sum of these two scores is used as the score for each countermeasure in the past. When recommending CA and PA, this ranking score is combined with keyword links to make the final recommendation.
[0075] Figure 5 This is a schematic diagram of the CA / PA ranking calculation logic in the intelligent recommendation algorithm of this invention.
[0076] Regarding the predictive model in intelligent recommendation algorithms, to intelligently recommend the best improvement strategies (CA) and prevention strategies (PA), the discovered quality problems must first be categorized by topic. After establishing the categories, the system incorporates information corresponding to the same category into the discrimination, performing keyword searches and sentence similarity searches. Items with "high keyword relevance" and "high sentence similarity" from the database are ranked first. Next, the ranking within each group is calculated by combining expert knowledge evaluation scores. Finally, improvement strategies and prevention strategies are recommended based on the ranking scores. These improvement strategies and prevention strategies are provided for user reference, and users can freely edit the statements to ensure they effectively solve the problem and are stored in the database as important recommended strategies again, enriching the data breadth.
[0077] Step S17: After obtaining the best improvement measures (CA) and prevention measures (PA), implement improvement actions for each audit unit according to the improvement measures (CA) to solve the problem points, and implement prevention actions for each audit unit according to the prevention measures (PA) to avoid the same problem from happening again in the future.
[0078] Figure 6 This is a schematic diagram of the hardware architecture of an electronic device according to an embodiment of the present invention. The electronic device 200, for example, but not limited to, a smart device B, can communicate with and be connected to the processor 210, memory 220, and intelligent quality management system 230 via a system bus. Figure 6 Only the electronic device 200 with components 210-230 is shown; however, it should be understood that it is not required to implement all of the components shown, and more or fewer components may be implemented instead.
[0079] The memory 220 includes at least one type of readable storage medium, including flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 220 may be an internal storage unit of the electronic device 200, such as the hard disk or memory of the electronic device 200. In other embodiments, the memory may also be an external storage device of the electronic device 200, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc. Of course, the memory 220 may also include both internal storage units and external storage devices of the electronic device 200. In this embodiment, the memory 220 is typically used to store the operating system and various application software installed on the electronic device 200, such as the program code of the intelligent quality management system 230. Furthermore, the memory 220 can also be used to temporarily store various types of data that have been output or will be output.
[0080] In some embodiments, the processor 210 may be a central processing unit (CPU), controller, microcontroller, microprocessor, or other data processing chip. The processor 210 is typically used to control the overall operation of the electronic device 200. In this embodiment, the processor 210 is used to run program code stored in the memory 220 or process data, for example, to run the intelligent quality management system 230.
[0081] It should be noted that, Figure 2 The electronic device 200 is merely an example. In other embodiments, the electronic device 200 may also include more or fewer components, or have different component configurations.
[0082] Figure 7 This is a functional block diagram of an electronic device according to an embodiment of the present invention, which is used to execute an intelligent quality management method. The intelligent quality management method of this embodiment can be implemented by a computer program stored in a storage medium, such as memory 220 in the electronic device 200. When the computer program implementing the method of the present invention is loaded into memory 220 by processor 210, it drives processor 210 of the electronic device 200 to execute the intelligent quality management method of this embodiment.
[0083] The electronic device 200 of this invention includes a Heatmap risk management module 310, an intelligent audit and scheduling module 320, a processing module 330, an intelligent recommendation module 340, and an execution module 350.
[0084] Establish a Heatmap risk platform for quality status
[0085] In this phase, audit data is statistically analyzed, and quality risk issues within the factory are recorded one by one in the system. A risk management interface is established to present risk hotspots, covering all data from the past seven periods plus the current period. The Heatmap risk platform can integrate problem categories and details from multiple factories, along with corresponding time-dependent RPNs and Process Failure Mode and Effects Analysis (PFMEA). It uses color-coding to display problem severity, Pareto charts to present the root causes of problems, and integrates statistical data using tags such as "Problem Category," "Manufacturing Division," and "Recurring Problems," allowing managers to grasp all quality risk issues within the group in the shortest possible time.
[0086] Heatmap's risk management module 310 calculates required data and parameter configurations based on audit-related data and the Time Dependent RPN (RPN) formula. The Time Dependent RPN algorithm is primarily constructed using Python-Pandas, Python-Numpy, and C++, and an application programming interface (API) service is built using Django.
[0087] Existing common RPN formulas only reflect the current severity of factory problems and cannot detect or punish items with higher past problem severity. Therefore, this invention incorporates time weights into the formula and designs severity-related parameters according to user needs to calculate the time-dependent RPN algorithm of this invention. The time-dependent RPN algorithm formula of this invention is as follows:
[0088]
[0089] Where Security(S) represents severity, Occurrence(O) represents incidence, Detection(D) represents detectability, i represents quarters (i = 1, ..., 8), N represents questions (j = 1, ..., N), and k represents event types (k = 1, ..., K). i,j (Event Type), and m = 1,...,M i,j,k .
[0090] For example, in the third quarter of fiscal year FY21, the table shows that issue 4.7.1 corresponds to 2 quality incidents, with the incident type being Quality Incident, and m = 2. In the fourth quarter of fiscal year FY21, the table shows that issue 4.7.3 corresponds to 3 quality incidents, with the incident type being Internal Progressive Audit, and m = 3.
[0091] The time-dependent RPN algorithm combines severity, occurrence rate, detectability, and PFMEA score to form an original algorithm that achieves an innovative indicator that can weigh the severity of a problem based on "time relationship", "problem category", "number of occurrences" and "risk".
[0092] The Heatmap risk management module 310 generates a heatmap risk interface based on the required data and parameter configuration. It connects to the PostgreSQL database via API and front-end web application frameworks such as HTML, CSS, and Vue.js to construct the heatmap interface and other visualization interfaces.
[0093] Develop intelligent audit scheduling plan
[0094] The intelligent audit scheduling module 320 formulates an intelligent audit scheduling algorithm through the heat map risk interface to automatically generate an audit plan.
[0095] Intelligent audit scheduling algorithms are developed based on Heatmap quality information, customer audit specifications, and innovative time-dependent RPN algorithms to automatically generate audit plans.
[0096] Figure 8 This is a functional block diagram of the intelligent audit scheduling module according to an embodiment of the present invention. 320 of this embodiment includes a selection unit 3210, a calculation unit 3220, a judgment unit 3230, and a setting unit 3240.
[0097] Unit 3210 selects N audit units and M problem points that need to be audited in the current time period.
[0098] The calculation unit 3220 calculates the RPN value of K issues, where K issues include the current time period plus the previous seven time periods, and calculates the RPN value of the issue point for each audit unit.
[0099] Select unit 3210 to select the audit unit whose RPN value ranks first X.
[0100] The judgment unit 3230 determines whether the current time period is the first week of the cycle. If it is not the first week of the cycle, the judgment unit 3230 determines whether there are duplicate audit units among the selected audit units. If there are duplicate audit units, the duplicate audit units are removed.
[0101] If the current time period is the first week of the cycle, and there are no duplicate audit units, then the selection of N audit units is complete. Next, unit 3210 selects the top Y largest RPN (Recurring Problem Points) and the next Z largest RPN (Recurring Problem Points) from the M problem points, where Y + Z = M. The top Y largest problem points are the most serious; duplicate problem points are acceptable, and they must be audited weekly. The next Z largest problem points are the next most serious, and the principle is to avoid duplicate audits within a single round.
[0102] The judgment unit 3230 determines whether the current time period is the first week of the cycle. If it is not the first week of the cycle, it determines whether there are duplicate problem points among the selected problem points. If there are duplicate problem points, it removes the duplicate problem points.
[0103] After selecting the top Y problem points of the RPN, or if the current time period is the first week of the cycle, or if there are no duplicate problem points, then the selection of M problem points is complete. Next, the judgment unit 3230 determines whether a quality event has occurred within a default time (e.g., one month).
[0104] If at least one quality event occurs, the setting unit 3240 sets the selected problem point as a cross-audit unit project. If no quality event occurs, the setting unit 3240 assigns L problem points with higher RPN values to each audit unit, and then proceeds to the next audit cycle.
[0105] Processing module 330 notifies the auditing unit of the audit time and auditing team, and executes the audit procedures and selects problem points according to the audit plan, including: generating new quality cases in the database, filling in descriptions of finding and identifying root causes, etc.
[0106] Intelligent recommendation algorithm
[0107] The intelligent recommendation module 340 provides intelligent root cause category suggestions for the problem points, and provides intelligent corrective and preventive behavior suggestions for the multiple problem points based on the root cause category suggestions, so as to obtain the best improvement and prevention strategies.
[0108] Figure 3 This is a schematic diagram illustrating the implementation process of intelligent recommendation in an embodiment of the present invention.
[0109] The improvement and prevention strategies for historical audit data are stored in a "textual description" and "mixed Chinese and English" format. Therefore, this embodiment of the invention uses Chinese-English word segmentation model technology combined with a custom dictionary as the basis, and then uses bag-of-words model combined with text exploration algorithm to keep the training data in the best state. Finally, the model is trained and classified through convolutional neural network, deep learning algorithm and machine learning algorithm, and finally the best improvement and prevention strategies are recommended.
[0110] First, data cleaning and preprocessing operations are performed on the raw data of quality issues, including data labeling, Chinese-English word segmentation model, and the establishment of a bag of quality keywords.
[0111] Figure 4 This is a schematic diagram illustrating the data cleaning / preprocessing implementation flow of the intelligent recommendation algorithm according to an embodiment of the present invention.
[0112] Word segmentation for a mixed Chinese-English quality database is a complex problem. This invention first employs a Hidden Markov Model (HMM) for word segmentation, adjusting the segmentation results and removing redundant words to establish a unique quality dictionary. Based on expert knowledge, the most suitable quality Chinese-English word segmentation model is obtained; finally, word segmentation is performed again using this model, adjusting the results and removing redundant words to establish the final quality bag-of-vocabulary model.
[0113] Based on each past quality issue record, standardized categories are established according to the issue type, the work section where the issue occurred, and the root cause of the issue. Then, each issue is assigned a corresponding label, so that historical data can be classified and a semantic classification database is established.
[0114] Regarding the training model update in intelligent recommendation algorithms, the present invention selects the best-performing Convolutional Neural Network (CNN) model as the basis for prediction and classification after experiments and tests. It also incorporates preprocessing and parameter tuning training, and designs various different parameter combinations to retrain and adjust the model to improve the accuracy of prediction.
[0115] In the Convolutional Neural Network (CNN) architecture, this invention employs two convolutional layers paired with a global max pooling layer, preserving key features while reducing parameters and computational cost. Simultaneously, a fully connected (dense) layer is used to weight and sum the previously designed features, mapping the learned "distributed feature representation" onto the sample label space, and then followed by the Sofmax function to present the scores of each category probabilistically. The cross-entropy function is used on the model's loss function to calculate the distance between the predicted probability distribution and the probability distribution of the true answer, optimizing the model's predictions. The optimization function employs a modified gradient descent method, namely the Adam algorithm, dynamically adjusting the learning rate of each parameter based on the first-order and second-order gradient matrix estimates of each parameter from the loss function. Furthermore, momentum (the moving average of the parameters) is used to improve traditional gradient descent, promoting dynamic adjustment of hyperparameters.
[0116] After adjusting the parameters, the accuracy was further improved to 91.01%. This model was ultimately used as the prediction category for quality problems and became an important category basis for recommending corrective actions (CA) and preventive actions (PA) in the future.
[0117] Regarding the CA / PA ranking calculation in model training of intelligent recommendation algorithms, improvement strategies (CA) and prevention strategies (PA) are the most important parts of recommendation. They can not only activate past experience in handling problems, but also guide newcomers to take countermeasures to deal with problems.
[0118] The key to this ranking calculation is to identify the descriptions most similar to the current problem from a database of past countermeasures and recommend corresponding countermeasures. Therefore, in this embodiment of the invention, evaluation formulas for Improvement Countermeasures (CA) and Prevention Countermeasures (PA) are developed, giving lower weight to problems with high recurrence rates and higher weight to countermeasures that can be effectively solved. Furthermore, the evaluation scores of CA and PA by professionals in the relevant fields are incorporated, and the sum of these two scores is used as the score for each countermeasure in the past. When recommending CA and PA, this ranking score is combined with keyword links to make the final recommendation.
[0119] Figure 5 This is a schematic diagram of the CA / PA ranking calculation logic in the intelligent recommendation algorithm of this invention.
[0120] Regarding the predictive model in intelligent recommendation algorithms, to intelligently recommend the best improvement strategies (CA) and prevention strategies (PA), the discovered quality problems must first be categorized by topic. After establishing the categories, the system incorporates information corresponding to the same category into the discrimination, performing keyword searches and sentence similarity searches. Items with "high keyword relevance" and "high sentence similarity" from the database are ranked first. Next, the ranking within each group is calculated by combining expert knowledge evaluation scores. Finally, improvement strategies and prevention strategies are recommended based on the ranking scores. These improvement strategies and prevention strategies are provided for user reference, and users can freely edit the statements to ensure they effectively solve the problem and are stored in the database as important recommended strategies again, enriching the data breadth.
[0121] After obtaining the best improvement measures (CA) and prevention measures (PA), the execution module 350 implements improvement actions for each audit unit according to the improvement measures (CA) to solve the problem points, and implements prevention actions for each audit unit according to the prevention measures (PA) to prevent the same problem from recurring in the future.
[0122] If the modules / units integrated in the electronic device 200 are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, disks, optical discs, computer memory, read-only memory, random access memory, electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electrical carrier signals and telecommunication signals.
[0123] It is understood that the module division described above is merely a logical functional division, and other division methods may be used in actual implementation. Furthermore, the functional modules in the various embodiments of this application can be integrated into the same processing unit, or each module can exist physically separately, or two or more modules can be integrated into the same unit. The integrated modules described above can be implemented in hardware or in a combination of hardware and software functional modules.
[0124] For those skilled in the art, other corresponding changes or adjustments can be made to the technical solutions and concepts provided in the embodiments of the present invention in combination with actual needs, and all such changes and adjustments should fall within the protection scope of the claims of the present invention.
Claims
1. An intelligent quality management method applied in electronic devices, characterized in that, The method includes: The required data and parameter configuration are calculated based on relevant audit data and the time-dependent risk priority index formula, wherein: the time-dependent risk priority index formula is: ； in, Security(S) For seriousness, For incidence rate, For detection rate, i represents the quarter. N represents the problem. k represents the event type. And m is the number of quality events. ; A heat map risk interface is generated based on the required data and the parameter configuration. The intelligent audit scheduling algorithm is formulated through the heat map risk interface to automatically generate audit plans, which also includes: Select N audit units and M problem points that need to be audited in the current time period; Calculate the time-dependent RPN values for K problems; Calculate the RPN value of the problem points for each of the N audit units; Select the audit unit whose RPN value ranks first X; Determine if the current time period is the first week of a cycle; If it is not the first week of the cycle, determine whether the selected audit unit has duplicate audit units; If there are duplicate audit units, the duplicate audit units will be removed. If the current time period is the first week of the cycle, or there are no repeated audit units, it means that the selection of the N audit units has been completed. Then, the top Y largest RPN problem points and the second largest Z largest RPN problem points are selected from the M problem points, Y+Z=M. Determine if the current time period is the first week of a cycle; If it is not the first week of the cycle, determine whether there are duplicate problem points among the selected problem points; If there are duplicate issues, remove the duplicate issues. After selecting the top Y problem points of the RPN, or if the current time period is the first week of the cycle, or if there are no repeated problem points, it means that the selection of the M problem points is completed. Then it is determined whether any quality event has occurred within the preset time. If at least one quality incident occurs, the selected issue point will be set as a cross-audit unit project; If no quality incident occurs, assign L problem points with higher RPN values to each audit unit; and The audit will proceed to the next time period. Notify the auditing unit to check the audit time and audit team, and execute the audit procedures according to the audit plan and select multiple issue points; Intelligent root cause category suggestions are provided for the aforementioned multiple problem points; Based on the aforementioned root cause categories, intelligent corrective and preventative behavioral suggestions are recommended for the multiple problem areas to achieve optimal improvement and prevention strategies; and Based on the improvement measures, each audit unit shall implement improvement actions to address the problem points, and based on the prevention measures, each audit unit shall implement prevention actions.
2. The intelligent quality management method as described in claim 1, characterized in that, Also includes: The audit process includes generating new quality cases and completing descriptions of finding and identifying root causes.
3. The intelligent quality management method as described in claim 1, characterized in that, Also includes: The heatmap risk interface is constructed by connecting the database PostgreSQL and the front-end HTML, CSS, and Vue.js web application framework through the application programming interface.
4. The intelligent quality management method as described in claim 1, characterized in that, Also includes: Data cleaning and preprocessing, model training, and prediction model building are performed on the raw data of the multiple problem points. The data cleaning and preprocessing includes data labeling, Chinese-English word segmentation model, and the establishment of a quality keyword bag. The model training includes training the model, updating and improving countermeasures, and calculating the ranking of preventive countermeasures; and The establishment of the prediction model includes prediction of the main causes of quality problems and recommendations for improvement and prevention strategies.
5. The intelligent quality management method as described in claim 4, characterized in that, The ranking calculation of improvement and prevention measures also includes: Find the descriptions most similar to the current problem from the database of past countermeasures and recommend corresponding countermeasures; The evaluation score for ranking the improvement and prevention measures is W1+W2, where W1 is the score for reducing duplicate attributes and W2 is the effectiveness of the professional rating.
6. An electronic device, characterized in that, include: The risk management module is used to calculate demand data and parameter configurations based on audit-related data and a time-dependent risk priority index formula, and to generate a heatmap risk interface based on the demand data and parameter configurations. The time-dependent risk priority index formula is: ； in, Security(S) For seriousness, For incidence rate, For detection rate, i represents the quarter. N represents the problem. k represents the event type. And m is the number of quality events. ; The intelligent audit scheduling module is used to formulate an intelligent audit scheduling algorithm through the heat map risk interface to automatically generate an audit plan; Select Unit: Used to select N audit units and M problem points that need to be audited in the current time period; A calculation unit is used to calculate the time-dependent RPN values of K issues, and to calculate the RPN value of each issue point of the N audit units; The selection unit selects audit units whose RPN values rank before X. The judgment unit is used to determine whether the current time period is the first week of the cycle. If it is not the first week of the cycle, it determines whether the selected audit unit has duplicate audit units. If there are duplicate audit units, the duplicate audit units are removed. If the current time period is the first week of the cycle, or there are no repeated audit units, it means that the selection of the N audit units has been completed. Then, the top Y largest RPN problem points and the second largest Z largest RPN problem points are selected from the M problem points, Y+Z=M. The judgment unit determines whether the current time period is the first week of the cycle. If it is not the first week of the cycle, it determines whether there are duplicate problem points among the selected problem points. If there are duplicate problem points, it removes the duplicate problem points. After selecting the problem points of the first Y largest RPN, or when the current time period is the first week of the cycle, or if there are no duplicate problem points, it indicates that the selection of the M problem points is completed, and it determines whether any quality event occurs within the preset time. The setting unit is used to set the selected problem points as cross-audit unit projects if at least one quality event occurs, and if no quality event occurs, to assign L problem points with higher RPN values to each audit unit, and then proceed to the audit of the next time period. The processing module is used to notify the auditing unit of the audit time and auditing team, and to execute the auditing procedures and select multiple problem points according to the audit plan. The intelligent recommendation module is used to provide intelligent root cause category suggestions for the multiple problem points, and based on the root cause category suggestions, to provide intelligent corrective and preventive behavior suggestions for the multiple problem points, so as to obtain the best improvement and prevention strategies; and The execution module is used to perform improvement actions on each audit unit to address the problem points according to the improvement countermeasures, and to perform preventive actions on each audit unit according to the preventive countermeasures.
7. A storage medium storing at least one computer instruction thereon, characterized in that, The instructions are loaded and executed by the processor as described in any one of claims 1-5, representing the intelligent quality management method.