Method and device for analyzing human factors in PMA manufacturing process of civil aviation product

Abstract

The invention discloses a civil aviation product PMA manufacturing process human factor analysis method and device, and belongs to the technical field of computer-aided analysis, and the method comprises the steps: S1, constructing a civil aviation product PMA manufacturing professional vocabulary library; s2, establishing a production line analysis model by using a civil aviation product PMA manufacturing specialized vocabulary library and a Jieba word library; s3, analyzing upper and lower layer factors in the production line analysis model by using chi-square test and yield ratio, and performing factor weight distribution based on conditional probability; s4, DEMATEL analysis is carried out; and S5, TOPSIS comprehensive evaluation sorting is carried out, firstly, the influence degree Ci, the affected degree Di, the centrality Mi and the reason degree Ri are regarded as attributes of each scene, then a standard matrix is calculated, and finally the optimal solution distance, the worst solution distance and the score of the ith influence factor of the target object are calculated. According to the method, the problem of complex coupling between human factors and production line data in the PMA manufacturing process of civil aviation products can be solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of computer-aided analysis, and particularly relates to a method and device for analyzing human factors in the manufacturing process of PMA for civil aviation products. Background Art

[0003] Currently, the research on human factors in civil aircraft mainly focuses on identifying and analyzing error sources, improving operator skills and training effects, enhancing human-machine interaction design, and promoting teamwork. Research shows that human factors of operators in MRO enterprises are the main causes of many aviation accidents. The statistical results of Kharoufah et al. indicate that human factors have become the most important factor in civil aviation safety accidents, surpassing mechanical failures. To analyze and control human factors, Professor James Reason proposed a model in 1990 to describe the mechanism from organizational behavior to unsafe behavior. Wiegmann and Scott A Shappell further developed the Reason model and proposed the HFACS model, which solved the long-term separation problem between theory and practical application and became a widely accepted human factor classification model in aviation accident investigations. Wang Jiaxu et al. used the HFACS model to analyze human errors in the flight tests of avionics experimental aircraft, demonstrating the advantages of the model in analyzing cases. Wu Yaogang calculated accident causes through chi-square tests based on the HFACS model, revealing the accident occurrence path. In 2001, the US Navy proposed new indicators based on the HFACS model and developed the HFACS-ME model for civil aviation maintenance personnel, focusing on the analysis of human factors in aircraft airframe maintenance work. Rashid et al. statistically analyzed human error events in aircraft maintenance work through case analysis and the HFACS-ME model, revealing multi-level human factor error causes.

[0004] From the perspective of theoretical research results, the accident causes and human factor analysis models in the civil aviation field are mainly analyzed and summarized based on human error cases in civil aircraft maintenance work, forming corresponding theoretical models. With the expansion of work types, many MRO enterprises have expanded their business scope and developed the PMA manufacturing capacity for civil aviation products by using the existing work systems, personnel, and knowledge and skills. Although there are a large number of human errors in this field, there is currently a lack of corresponding analysis models, resulting in insufficient analysis and control of human factors in the manufacturing process of PMA for civil aviation products.

[0005] The DEMATEL method was initially proposed by research teams such as Gabus in the United States in the 1970s and has gained wide recognition, especially in the fields of complex systems and decision science. This method can reveal the causal relationships between various factors in a complex system, providing a systematic perspective to help decision-makers understand the essence of the problem. TOPSIS is a method widely used in multi-criteria decision analysis, which ranks options by evaluating the distances between each option and the ideal solution and the negative ideal solution. To address the issues that the DEMATEL method relies on expert subjective judgments and opinions, which may lead to result biases and inconsistencies, and that the TOPSIS method highly depends on the weights of various criteria, which may introduce subjectivity and affect the reliability of decision-making results, it is particularly important to develop a method and device for analyzing human factors in the PMA manufacturing process of civil aviation products. Summary of the Invention

[0006] In view of the deficiencies of the prior art, the present invention provides a method and device for analyzing human factors in the PMA manufacturing process of civil aviation products, which are used to solve the complex coupling problem between human factors and production line data in the PMA manufacturing of civil aviation products.

[0007] To achieve the above invention purpose, the first object of the present invention is to provide a method for analyzing human factors in the PMA manufacturing process of civil aviation products, including the following steps:

[0008] S1. Construct a professional vocabulary library for the PMA manufacturing of civil aviation products;

[0009] S2. Establish a production line analysis model using the professional vocabulary library for the PMA manufacturing of civil aviation products and the Jieba word library;

[0010] S3. Analyze the upper and lower layer factors in the production line analysis model using chi-square test and concession ratio, and allocate factor weights based on conditional probability;

[0011] S4. DEMATEL analysis; first, establish an initial relationship matrix based on the results of chi-square test and concession ratio analysis, then calculate the normal matrix N according to the initial relationship matrix, calculate the comprehensive influence matrix T according to the normal matrix N, and finally, by analyzing the comprehensive influence matrix T, calculate the influence degree C i , the influenced degree D i , the centrality M i and the reason degree R i ;

[0012] S5. TOPSIS comprehensive evaluation and ranking. First, take the influence degree C i , the influenced degree D i , the centrality M i and the reason degree R iTreat them as the attributes of each scenario, then calculate the standard matrix, and finally calculate the optimal solution distance, the worst solution distance of the target object, and the score of the i-th influencing factor.

[0013] Preferably, the construction of the professional vocabulary library for civil aviation product PMA manufacturing includes: sorting out the professional vocabulary of civil aviation product PMA manufacturing, combining the content of relevant documents, and collecting and summarizing from five aspects of people, machines, materials, methods, and environment to obtain the professional vocabulary library for civil aviation product PMA manufacturing, and supplementing the professional vocabulary library for civil aviation product PMA manufacturing into the basic vocabulary library to obtain a new professional vocabulary library for civil aviation product PMA manufacturing.

[0014] Preferably, the establishment of the production line analysis model includes: first adding the professional vocabulary library for civil aviation product PMA manufacturing to the Jieba word library, and then using the Python Jieba word segmentation method to extract the vocabulary information in the historical error cases of MRO enterprises to obtain a keyword library; subsequently, based on the keyword library, conduct a questionnaire survey for different positions of personnel, statistically screen the results of the questionnaire survey, and count the consistency of each vocabulary. When the consistency exceeds 50%, the vocabulary is retained, otherwise it is eliminated. After statistical screening, five-level indicators are obtained; finally, the items of the five-level indicators are added to the HFACS-ME simplified model to form a production line analysis model.

[0015] Preferably, the factor weight assignment based on conditional probability includes: first using chi-square test to analyze whether there is a significant causal relationship between the upper and lower layer factors in the production line analysis model, and then adjusting the concession ratio based on conditional probability factors to reflect the influence coefficient between the upper and lower layer factors of the production line analysis model.

[0016] Preferably, the DEMATEL analysis includes: establishing an initial relationship matrix. First, based on the results of chi-square test and concession ratio analysis, analyze the influence degree between the factors in the production line analysis model, then divide the influence degree into several levels, and finally generate factor A i For factor A j The direct relationship matrix H of the influence degree = [h ij n×n , where n is the number of rows and columns of the direct relationship matrix H, i and j represent the i-th row and the j-th column in the direct relationship matrix H, and h ij represents the element in the i-th row and the j-th column of the direct relationship matrix H;

[0017] Calculate the normal matrix N:

[0018]

[0019] Calculate the comprehensive influence matrix T;

[0020] ​

[0021] I is an n×n identity matrix;

[0022] By analyzing the comprehensive influence matrix T, calculate the influence degree C of each factor i , the influenced degree D i , the centrality M i and the reason degree R i ; Construct a factor influence relationship graph: Use the centrality M i as the horizontal axis and the reason degree R i as the vertical axis to draw a scatter plot.

[0023] Preferably: The influence degree C i and the influenced degree D i are respectively obtained from the sum of the elements in the i-th row and the i-th column of the comprehensive influence matrix T; The centrality M i is calculated by adding C i and D i , and the reason degree R i is obtained by subtracting D i from C i . The higher the value of M i , the more important the status of this factor. The positive or negative value of R i is used to judge the nature of the factor: when R i >0, this factor is the influence source; when R i is less than 0, this factor is the influenced object.

[0024] Preferably: The calculation formulas for the influence degree C i , the influenced degree D i , the centrality M i and the reason degree R i are as follows:

[0025]

[0026] h ji represents the element in the j-th row and the i-th column of the direct relationship matrix H;

[0027] M i =C i +D i ;

[0028] R i =C i -D i .

[0029] Preferably: The influence degree C i , the influenced degree D i , the centrality M i and the reason degree R iRegarding the attributes of each scenario, an original matrix X is established, and the original matrix X is standardized to obtain a standard matrix Z;

[0030] The formula for calculating the standard matrix is as follows:

[0031]

[0032] Where X ij represents the element in the i-th row and j-th column of the original matrix X;

[0033] The optimal solution distance of the target object The formula is:

[0034]

[0035] Where m is the number of scenarios in the original matrix X, represents the maximum value of the j-th column in the standard matrix Z, and z ij represents the element in the i-th row and j-th column of the standard matrix Z;

[0036] The worst solution distance of the target object The formula is:

[0037]

[0038] Where, represents the minimum value of the j-th column in the standard matrix Z, and z ij represents the element in the i-th row and j-th column of the standard matrix Z

[0039] The score Si of the i-th influencing factor i The calculation formula is:

[0040]

[0041] Where, represents the optimal solution distance of the target object, represents the worst solution distance of the target object.

[0042] The second object of the present invention is to provide a device for analyzing human factors in the manufacturing process of civil aviation product PMA, including:

[0043] A vocabulary library construction module for constructing a professional vocabulary library for the manufacturing of civil aviation product PMA;

[0044] A model establishment module for establishing a production line analysis model using the professional vocabulary library for the manufacturing of civil aviation product PMA and the Jieba vocabulary library;

[0045] A weight assignment module for analyzing the upper and lower layer factors in the production line analysis model using chi-square test and concession ratio, and assigning factor weights based on conditional probability;

[0046] Analysis module, DEMATEL analysis; First, based on the results of chi-square test and concession ratio analysis, an initial relationship matrix is established. Then, according to the initial relationship matrix, the normal matrix N is calculated. According to the normal matrix N, the comprehensive influence matrix T is calculated. Finally, by analyzing the comprehensive influence matrix T, the influence degree C of each factor is calculated. i , influenced degree D i , centrality M i and reason degree R i ;

[0047] Sorting module, TOPSIS comprehensive evaluation and sorting. First, the influence degree C i , influenced degree D i , centrality M i and reason degree R i are regarded as the attributes of each scenario. Then, the standard matrix is calculated. Finally, the optimal solution distance, worst solution distance of the target object and the score of the i-th influencing factor are calculated.

[0048] The third object of the present invention is to provide a computer-readable storage medium, which stores a computer program, and when the program is executed by a processor, the above-mentioned method for analyzing human factors in the manufacturing process of civil aviation product PMA is realized.

[0049] The fourth object of the present invention is to provide a computer program product, including a computer program, and when the computer program is executed by a processor, the above-mentioned method for analyzing human factors in the manufacturing process of civil aviation product PMA is realized.

[0050] The advantages and positive effects of the present application are:

[0051] The present invention constructs a production line analysis model suitable for analyzing the manufacturing of civil aviation product PMA. By adopting two analysis and calculation methods of DEMATEL (Decision Making Trial and Evaluation Laboratory method) and TOPSIS (Technique for Order Preference by Similarity to Ideal Solution), the causal path leading to human error events is successfully identified. The establishment of this model aims to put forward practical suggestions to reduce the errors caused by human factors in the manufacturing process, so as to improve the production quality and safety of civil aviation product PMA.

[0052] Based on the HFACS (Human Factors Analysis and Classification System) model, the present invention further subdivides production line indicators, coordinately applies the Decision Making Trial and Evaluation Laboratory method and the Technique for Order Preference by Similarity to Ideal Solution to complete the importance analysis of human factors. Through this method, the present invention can identify the causal transmission path of events that may cause manufacturing errors and give effective risk prevention suggestions accordingly. In addition, based on chi-square test and odds ratio analysis, the present invention realizes the quantitative division of the influence degree between human factors. Using the DEMATEL method, the present invention determines the influence and being-influenced degree between factors, and combines with the TOPSIS method to identify and rank the most significant key indicators. To verify the practicability and accuracy of this analysis model, the present invention takes 225 human factor error events of the parts manufacturing production line of an MRO enterprise as samples for in-depth analysis and method verification. The research results show that: during the parts manufacturing process, the occurrence of human factor errors is not an isolated event, but there are significant causal relationships with the management level, the preconditions of unsafe behaviors, and unsafe behaviors. Therefore, this analysis model is proven to be an effective tool applicable to the analysis and optimization of human factors in the PMA manufacturing of civil aviation products. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0054] Figure 1 It is a flowchart provided by a preferred embodiment of the present invention;

[0055] Figure 2 It is a flowchart for establishing a production line analysis model provided by a preferred embodiment of the present invention;

[0056] Figure 3 It is a comprehensive ranking diagram of the optimal solution based on TOPSIS provided by a preferred embodiment of the present invention;

[0057] Figure 4 It is a comprehensive ranking diagram of human factor indicators in aviation parts provided by a preferred embodiment of the present invention;

[0058] Figure 5 It is a risk causal path diagram provided by a preferred embodiment of the present invention. Specific implementation mode

[0059] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0060] For the first embodiment, please refer to Figure 1 , a method for analyzing human factors in the manufacturing process of civil aviation product PMA, including the following steps:

[0061] S1. Construct a professional vocabulary library for the manufacturing of civil aviation product PMA; the construction of the professional vocabulary library for the manufacturing of civil aviation product PMA includes: sorting out the professional vocabulary for the manufacturing of civil aviation product PMA, combining the content of relevant documents, collecting and summarizing from five aspects of people, machines, materials, methods, and environment to obtain a professional vocabulary library for the manufacturing of civil aviation product PMA, and supplementing the professional vocabulary library for the manufacturing of civil aviation product PMA into the basic vocabulary library to obtain a new professional vocabulary library for the manufacturing of civil aviation product PMA.

[0062] S2. Establish a production line analysis model by using the professional vocabulary library for the manufacturing of civil aviation product PMA and the Jieba word library; please refer to Figure 2 , the establishment of the production line analysis model includes: first adding the professional vocabulary library for the manufacturing of civil aviation product PMA to the Jieba word library, then using the Python Jieba word segmentation method to extract the vocabulary information in the historical error cases of MRO enterprises to obtain a keyword library; subsequently, based on the keyword library, conduct a questionnaire survey for different positions of personnel, statistically screen the results of the questionnaire survey, and count the consistency of each vocabulary. When the consistency exceeds 50%, the vocabulary is retained, otherwise it is excluded. After statistical screening, five-level indicators are obtained; finally, the entries of the five-level indicators are added to the HFACS-ME simplified model to form a production line analysis model.

[0063] S3. Analyze the upper and lower layer factors in the production line analysis model by using chi-square test and concession ratio, and allocate factor weights based on conditional probability; the factor weight allocation based on conditional probability includes: first using the chi-square test to analyze whether there is a significant causal relationship between the upper and lower layer factors in the production line analysis model, and then adjusting the concession ratio based on the conditional probability factor to reflect the influence coefficient between the upper and lower layer factors of the production line analysis model.

[0064] S4, DEMATEL analysis; First, based on the results of chi-square test and concession ratio analysis, an initial relationship matrix is established. Then, the normal matrix N is calculated according to the initial relationship matrix, and the comprehensive influence matrix T is calculated according to the normal matrix N. Finally, by analyzing the comprehensive influence matrix T, the influence degree C of each factor is calculated. i , the influenced degree D i , the centrality M i and the reason degree R i ; The DEMATEL analysis includes:

[0065] Establish the initial relationship matrix. First, based on the results of chi-square test and concession ratio analysis, analyze the interaction between factors in the production line analysis model. Then, divide the influence degree into several levels. Finally, generate factor A i For factor A j The direct relationship matrix H of the influence degree = [h ij n×n , where n is the number of rows and columns of the direct relationship matrix H, ij represents the i-th row and the j-th column in the direct relationship matrix H, and h ij represents the element in the i-th row and the j-th column of the direct relationship matrix H;

[0066] Calculate the normal matrix N:

[0067] Calculate the comprehensive influence matrix T;

[0068] By analyzing the comprehensive influence matrix T, calculate the influence degree C i , the influenced degree D i , the centrality M i and the reason degree R i ;

[0069] Construct the factor influence relationship graph: Take the centrality M i as the horizontal axis and the reason degree R i as the vertical axis to draw a scatter plot.

[0070] S5, TOPSIS comprehensive evaluation and ranking. First, regard the influence degree C i , the influenced degree D i , the centrality M i and the reason degree R i as the attributes of each scenario. Then, calculate the standard matrix. Finally, calculate the optimal solution distance, the worst solution distance of the target object, and the score of the i-th influencing factor.

[0071] To further understand the concept of the present invention, the above embodiments are elaborated in detail below:

[0072] ​The HFACS model provides a systematic method for analyzing and classifying human errors in index classification. However, there is no unified classification method for the third-level and fourth-level indicators. Therefore, relevant indicators and items need to be selected or combined according to the characteristics of different industries and jobs as needed, and the relative importance of various factors in the manufacturing environment and the potential interaction effects between different levels and categories need to be considered. This application focuses on two aspects of research:

[0073] A. Refine and expand the existing HFACS model to form an analysis model applicable to the manufacturing process of civil aircraft parts;

[0074] B. Select appropriate calculation methods to quantify the analysis results of the model and obtain the causal path of human error events;

[0075] Analysis of the HFACS-ME simplified model

[0076] Combined with the characteristics of the work in MRO enterprises, factors with low occurrence probability in the HFACS model are eliminated, and the HFACS theoretical framework is recombined to form the HFACS-ME simplified model, as shown in Table 1. This simplified model integrates and refines the HFACS model, forming a more guiding HFACS-ME simplified model. The HFACS-ME simplified model contains a total of 3 secondary indicators, 6 tertiary indicators, and 14 quaternary indicators. Considering that the production line of civil aircraft parts in MRO enterprises is established by relying on the existing work system, personnel, knowledge and skills of the enterprise.

[0077] Table 1 is the professional vocabulary library for the PMA manufacturing process of civil aviation products

[0078]

[0079] Construction of the professional vocabulary library for the PMA manufacturing of civil aviation products. Word segmentation is the first step in natural language processing, and the word segmentation dictionary is the premise for accurate analysis. Special domain dictionaries need to be established for special fields. In the present invention, the commonly used professional vocabulary in the PMA manufacturing work of civil aviation products is sorted out, and combined with relevant documents issued by the Civil Aviation Administration, such as the content in the "Production Approval and Supervision Procedures" (AP-21-AA-2019-31), from the five aspects of people, machines, materials, methods, and environment, through the method of automatic computer collection or manual collection, a professional vocabulary library for the PMA manufacturing of civil aviation products is formed, and this professional vocabulary library for the PMA manufacturing of civil aviation products is supplemented to the Jieba word library, thereby improving the accuracy of word segmentation.

[0080] Establishment of the production line analysis model. In this invention, the Python Jieba word segmentation method is used to extract the vocabulary information from the "Compilation of Error Cases" of a certain MRO enterprise from 2019 to 2022. In order to correctly extract professional vocabulary, the vocabulary in Table 1 needs to be added to the Jieba word library. Finally, 122 pieces of vocabulary information are extracted. After obtaining the vocabulary information, the questionnaire method is used to collect the opinions of personnel in each position in the PMA production line of civil aviation products. A total of 9 experienced production personnel, quality personnel, engineers, and production managers participated in the questionnaire survey. The opinions of personnel at all levels on the classification of 122 pieces of vocabulary information are collected, and the unanimous opinions that exceed half are retained. For the vocabulary that has not formed a unanimous opinion or is unanimously considered not to belong to any classification, it is excluded. Finally, 98 five-level index contents are obtained. The specific process is shown in Figure 2 。

[0081] The entries of the five-level indicators are added to the HFACS-ME simplified model, and finally the production line analysis model is formed, as shown in Table 2.

[0082] Table 2 is the production line analysis model

[0083]

[0084]

[0085] Factor weight assignment based on conditional probability. The chi-square test (X 2 test) is a hypothesis testing method based on the chi-square distribution, which is mainly applicable to statistically measure the deviation between the actual observed value and the theoretical value, and can be used to calculate whether there is a causal relationship between two variables.

[0086] First, establish the hypothesis events H0 and H1;

[0087] H0: There is no significant causal relationship between the upper and lower level factors in the production line analysis model;

[0088] H1: There is a significant causal relationship between the upper and lower level factors in the production line analysis model;

[0089] The test statistic X 2 value of two factors can be expressed as:

[0090]

[0091] In the formula, f i represents the sample frequency, np i represents the theoretical frequency. The larger the X 2 value, the more the actual value differs from the theoretical value. By querying the chi-square distribution quantile table, the corresponding chi-square 2 value can be obtained. 2The probability value corresponding to the value. The smaller the probability value, the lower the likelihood of event H0.

[0092] Secondly, to more intuitively reflect the magnitude of the influence coefficients between the upper and lower layer factors in the production line analysis model, a judgment is made through the odds ratio (OR value) analysis. The odds ratio analysis is a risk statistical indicator OR used to evaluate a specific result, and its calculation method can be expressed as:

[0093]

[0094] In the formula, P A and P B respectively represent the probabilities of the occurrence of the upper and lower layer factors in the model. The larger the OR value, the greater the mutual influence between the two factors, and the more relevant they are to each other. The specific meaning analysis is shown in Table 3. At the same time, a basis for the degree of factor influence is constructed based on the magnitude of the OR value. According to the theory of statistics and the calculation data in the maintenance work of civil aviation aircraft using the HFACS model, when and only when the chi-square probability is less than 0.05 and the OR value is greater than 1, the H0 hypothesis can be rejected and H1 can be accepted, thereby determining that there is an obvious causal relationship for this factor.

[0095] The OR value is a symmetric index measuring the association strength between two binary variables, but it cannot distinguish the direction of influence between variables. To solve this problem, the present invention introduces conditional probability weights to quantify the directional influence between variables through asymmetric weight allocation. Define the directional conditional probability to reflect the asymmetric dependence relationship between variables, calculate the conditional probability and define the weight of the conditional probability. The conditional probability weight of A i for A j can be expressed as:

[0096]

[0097] In the formula, P(A j =1|A i =1) represents the conditional probability that A j and A i occur simultaneously, and P(A j =1|A i =0) represents the conditional probability that A j occurs and A i does not occur.

[0098] To be closer to the actual causal direction, the OR value is directionally adjusted. The adjusted OR i of A j for A i→j can be expressed as:

[0099]

[0100] In the formula, the numerator Wi→j Reflect A i For A j The driving strength, with denominator W j→i Reflect A j For A i The feedback strength. If the drive is stronger than the feedback (W i→j >W j→i ), then OR i→j >OR.

[0101] When A i and A j are significantly correlated, the degree of influence between factors is constructed based on the adjusted OR value, combining statistics and human factor analysis, providing more comprehensive and in-depth insights for decision-making in complex human factors.

[0102] Table 3 is the OR value analysis table

[0103] Meaning analysis Chi-square probability OR value Influence degree This group of factors has obvious relevance Less than 0.05 Greater than 1 1，2，3，4 This group of factors has no relevance Less than 0.05 Less than 1 0 This group of factors has no relevance Greater than 0.05 Greater than 1 0 This group of factors has no relevance Greater than 0.05 Less than 1 0

[0104] DEMATEL analysis. DEMATEL is a complex system analysis method based on graph theory and matrix theory. This method can evaluate the mutual influence between elements in the system and identify key factors. Through DEMATEL analysis, the degree of influence and the degree of being influenced of each factor can be determined, and at the same time, their central position and causal relationship in the system can be revealed. In the analysis of human factors in aircraft component manufacturing, the application of the DEMATEL method mainly includes the following steps:

[0105] (1), Establish the initial relationship matrix. Based on the results of chi-square test and odds ratio analysis, evaluate the interaction between each factor in Table 2. Using a 5-level quantification scale, the degree of influence is divided into significant (4 points), large (3 points), medium (2 points), slight (1 point), and no influence (0 points). In this way, the direct relationship matrix H = [h i for factor A j on factor A ij n×n .

[0106] (2), Calculate the normal matrix N and the comprehensive influence matrix T. Use formula (5) to normalize the initial matrix to obtain the normal matrix N. Subsequently, use formula (6) to calculate the overall correlation effect between elements in the system to form the comprehensive influence matrix T.

[0107]

[0108] In the formula, I is the n×n order identity matrix.

[0109] (3), Calculate the influence degree C of each factor i, Influence Degree D i , Centrality M i and Cause Degree R i . By analyzing the comprehensive influence matrix T, the following indicators are calculated: Influence Degree C i and Influence Degree D i are respectively obtained from the sum of the elements in the i-th row and j-th column of the T matrix; System Centrality M i is calculated from C i plus D i , while Cause Degree R i is obtained by subtracting D i from C i . The higher the value of M i , the more important the factor is in the system. The positive or negative value of R i can be used to judge the nature of the factor: when R i > 0, the factor is the influencing source; when R i is less than 0, the factor is the affected object. These indicators together reveal the roles and importance degrees of each factor in the system.

[0110]

[0111] h ji represents the element in the j-th row and i-th column of the direct relationship matrix H;

[0112] The centrality M i

[0113] M i = C i + D i ; (9)

[0114] The cause degree R i

[0115] R i = C i - D i (10)

[0116] (4) Construct a factor influence relationship graph. Using centrality M i as the horizontal axis and cause degree R i as the vertical axis, draw a scatter plot. This visualization method can clearly show the status and role of each factor in the system, and helps to identify the core influencing factors.

[0117] TOPSIS comprehensive evaluation ranking. TOPSIS is a commonly used multi-attribute decision analysis method, aiming to rank by comprehensively considering the distances between each alternative solution and the ideal solution and the negative ideal solution, and select the best solution. The advantage of this ranking method is to rank the target data by comparing the distance between the existing feasibility and the optimal solution, without being affected by samples or indicators. The influence degree, being influenced degree, centrality, and reason degree calculated according to the aforementioned DEMATEL method are regarded as the attributes of each scenario.

[0118] Please refer to Figure 3 , which lists the optimal solution distances of each target object the worst solution distance and the comprehensive score S of this factor i .

[0119] Please refer to Figure 4 , and rank according to the comprehensive score S of each factor i from high to low to obtain Figure 4 the rankings of each risk factor in

[0120] Please refer to Figure 5 , connect the factors with significant correlation relationships in the order of high reason degree to low reason degree, and pay attention to the factors with high centrality and reason degree.

[0121] Regard the influence degree C i , the being influenced degree D i , the centrality M i and the reason degree R i as the attributes of each scenario, establish the original matrix X, and perform standardization processing on X to obtain the standard matrix Z.

[0122] The formula for calculating the standard matrix is as follows:

[0123]

[0124] where, X ij represents the element in the i-th row and j-th column of the original matrix X.

[0125] The optimal solution distance of the target object The formula is:

[0126]

[0127] where, m is the number of scenarios in the original matrix X, represents the maximum value of the j-th column in the standard matrix Z, and z ij represents the element in the i-th row and j-th column of the standard matrix Z.

[0128] The worst solution distance of the target object The formula is as follows:

[0129]

[0130] Where m is the number of scenarios in the original matrix X, represents the minimum value of the j-th column in the standard matrix Z, and z ij represents the element in the i-th row and j-th column of the standard matrix Z.

[0131] The score S of the i-th influencing factor i The calculation formula is as follows:

[0132]

[0133] Where, represents the optimal solution distance of the target object, represents the worst solution distance of the target object.

[0134] The following is a detailed analysis and elaboration through specific examples:

[0135] Sample data collection and analysis

[0136] In this embodiment, 225 pieces of human error event statistical data during the implementation of civil aviation product PMA manufacturing in a certain MRO enterprise are selected. The data sources include: error cases found during quality inspection or handover in daily work, error information feedback from product use departments, problems recorded during safety supervision, and various problems supplemented and recorded through interviews, inquiries, etc. The data content details multiple pieces of information such as the occurrence time, event overview, responsible department, and number of defects of each human error event. This data covers all work links and various workshop types in the manufacturing process, and the recorded information includes all data content from the initial stage to the stable stage of manufacturing, which can completely reflect the entire process of this MRO enterprise in implementing civil aviation product PMA manufacturing, ensuring the integrity and rationality of subsequent conclusions.

[0137] By matching the event overview content in each case with the five-level indicators of the production line analysis model, the occurrence frequency of human error in each case can be obtained. The statistical results are shown in Table 4.

[0138] Table 4 is the statistical table of human error events in the workshop

[0139]

[0140] From the statistical conclusions in Table 4, it can be seen that among all unsafe behaviors, the proportions of attention and memory errors and habitual violations are the highest, at 74.2% and 30.2% respectively, which are the key problems in the current manufacturing process of this component. Therefore, to solve these two problems, it is necessary to analyze the causal factors at all levels that lead to these problems, and by optimizing and improving each causal factor, interrupt the causal path of human error events, thereby reducing the occurrence frequency.

[0141] Chi-square test and odds ratio analysis calculation;

[0142] Establish hypothesis events H0 - H15: Among the 4 fourth-level indicators at the management level and the 4 fourth-level indicators of the preconditions of unsafe behaviors, there is no significant causal relationship between any two; establish hypothesis events H16 - H39: Among the 4 fourth-level indicators of the preconditions of unsafe behaviors and the 6 fourth-level indicators of unsafe behaviors, there is no significant causal relationship between any two; establish hypothesis events H40 - H63: Among the 4 fourth-level indicators at the management level and the 6 fourth-level indicators of unsafe behaviors, there is no significant causal relationship between any two.

[0143] Use the chi-square test and odds ratio analysis method to calculate the data in Table 5, and statistically analyze the results. The rejected hypothesis events are shown in Table 5 below, indicating that there is a significant causal relationship between any two of this group of indicators, and there is no significant causal relationship for the remaining items. Since no statistically significant association is found between indicators A10, A11, and A14 and other indicators, this invention will not analyze them. Rank the influence degrees according to the adjusted OR value. Among them, when the OR value is between (0 - 5), the influence degree is set to 1; when the OR value is between (5 - 10), the influence degree is set to 2; when the OR value is between (10 - 35), the influence degree is set to 3; when the OR value > 35, the influence degree is set to 4; when there is no significant causal relationship between any two, it is set to 0.

[0144] Table 5 shows the chi-square test and OR value of human error events in the workshop (only including the items that reject the original hypothesis)

[0145]

[0146] According to the calculation results in Table 5, it can be seen that at the management level and the precondition level of unsafe behaviors, the mutual influence of improper operation on the working environment, and improper supervision on physical and mental state and limitations is the most significant, so the influence level is set to 4; at the precondition level of unsafe behaviors and the unsafe behavior level, the influence of team coordination on knowledge and rules is relatively large, and the influence level is set to 4; at the management level and the unsafe behavior level, the influence between improper operation and insufficient knowledge and rules is relatively large, and the influence level is set to 3. The above relationships indicate that these preconditions play a key role in human errors in actual operations.

[0147] The results calculated based on the DEMATEL method (shown in Table 6) reveal that the centrality of improper operation and working environment is significantly higher than other risk factors, indicating their crucial positions in the risk system. High centrality means that these factors not only have a significant impact on other risk factors but are also vulnerable to the influence of other factors. Therefore, they should be the core focus of risk management. The cause degrees of improper operation, team coordination, and habitual violations are positive, indicating that these factors are the main driving factors in the risk conduction chain and important roots for inducing other risk events. Therefore, it is necessary to take targeted measures to fundamentally eliminate or alleviate the influence of these factors to achieve effective risk control. On the contrary, the cause degrees of factors such as insufficient resources, improper organizational processes, physical and mental states and limitations, inadequate work preparation, lack of knowledge and rules, and working environment are negative, indicating that they more often manifest as result-type risk factors, and their occurrence is often the result of being driven by the aforementioned cause-type risk factors. Therefore, in risk management practice, attention should be focused on strengthening the protection of these factors and reducing the possibility of their exposure to the risk environment, thereby reducing the overall risk level.

[0148] Table 6 shows the analysis results of the secondary indicators of influencing factors based on DEMATEL and TOPSIS

[0149]

[0150] The comprehensive analysis results based on DEMATEL and TOPSIS jointly provide multi-dimensional perspectives and methodological support for the analysis of the cause path mechanism of human error events. The following cause paths of human error events can be inferred: As the core driving factor, improper operation may, by affecting the working environment, trigger intermediate links such as attention and memory errors and habitual violations, and ultimately lead to the occurrence of human error events. In addition, the lack of team coordination may also exacerbate the impact of improper operation and further increase the probability of human error events. Factors such as insufficient resources, improper organizational processes, physical and mental states and limitations, work preparation, and lack of knowledge and rules may, as potential background factors, indirectly promote the occurrence of human error events by affecting the above key links. The cause path is shown in Table 7.

[0151] In summary, in the risk management of PMA manufacturing of civil aviation products, key risk factors such as improper operation and working environment should be focused on, and targeted measures should be taken to fundamentally eliminate or alleviate the influence of factors such as improper operation, team coordination, and habitual violations. At the same time, attention should also be paid to the control of potential background factors such as insufficient resources and improper organizational processes to build a more perfect prevention system for human error events.

[0152] Table 7 shows the cause path of human error events

[0153] Serial number Management level Premise of unsafe behavior Unsafe behavior Path 1 <![CDATA[Improper operation A1]]> <![CDATA[Operating environment A8]]> <![CDATA[Note memory error A9]]> Path 2 <![CDATA[Improper operation A1]]> <![CDATA[Team coordination A7]]> <![CDATA[Insufficient knowledge rule A 12 > Path 3 <![CDATA[Improper operation A1]]> <![CDATA[Work preparation A6]]> <![CDATA[Habitual violation A 13 > Path 4 <![CDATA[Improper supervision A2]]> <![CDATA[Physical and mental state and limitations A5]]> <![CDATA[Attention to memory error A9]]> Path 5 <![CDATA[Insufficient resource A3]]> <![CDATA[Operating environment A8]]> <![CDATA[Attention to memory error A9]]> Path 6 <![CDATA[Improper organizational process A4]]> <![CDATA[Work Preparation A6]]> <![CDATA[Habitual violation A 13 >

[0154] Discussion and Suggested Measures

[0155] Based on the statistics of human error events occurring in the PMA manufacturing process of civil aviation products and the calculation results of the production line analysis model, the following conclusions and suggestions can be drawn:

[0156] From the individual level, identify the indicators with a relatively high probability of unsafe behavior and conduct education and training for relevant employees. According to the statistical results in Table 4, among all unsafe behaviors, the probability of attention and memory errors is 74.2%, and the probability of habitual violations is 30.2%, which are the two problems with the highest proportions. The staff can be trained intensively through the summarized violation information. Regularly collect and summarize the regulations that are easily violated to form a safety warning education material for warning and educating employees.

[0157] From the organizational level, as can be seen from Causal Paths 5 and 6 in Table 7, insufficient resources and improper organizational processes are both reasons for human error events. The resource allocation can be optimized by identifying bottleneck resources in the production process, expanding key resources, optimizing the production process flow, training employees in various positions with relevant professional knowledge, and reasonably planning the use of relevant resources.

[0158] From the management level, as can be seen from Causal Paths 1 - 4 in Table 7, the problems at the management level are mainly reflected in improper operation and improper supervision. According to Figure 4 the results, the factor of improper operation has the highest comprehensive ranking, and Causal Paths 1, 2, and 3 are all caused by this key problem of improper operation. Therefore, it is necessary to prevent the occurrence of unsafe behaviors from the aspects of management and process, restrict the manufacturing process from the operation level through the promulgation of systems and the standardization of processes, and comprehensively consider factors such as the working environment, communication and collaboration, and work preparation to build a multi-level and all-round risk prevention and control system to effectively prevent and control human error events.

[0159] In summary, the occurrence of unsafe behaviors is not isolated but is related to all aspects of the manufacturing process. In order to improve and optimize the manufacturing process, it is necessary to comprehensively improve from the individual, organizational, and management levels to achieve the purpose of preventing human errors in the manufacturing process.

[0160] In order to better control human error in the PMA manufacturing process of civil aviation products, the present invention analyzes and improves the HFACS model, and constructs a production line analysis model by adding five-level indicators. The influence degree relationship between factors is obtained through chi-square test and concession analysis, and DEMATEL and TOPSIS provide multi-dimensional perspectives and methodological support for the analysis of the cause path mechanism of human error events. According to the actual case analysis results, the production line analysis model can calculate the cause path of events that are likely to lead to human error, and obtain the path relationship between improper supervision and operation, working environment, physical and mental state and limitations of employees, and finally violations and errors during the manufacturing process, so as to better put forward improvement suggestions in the manufacturing process and achieve preventive measures to avoid accidents by cutting off the accident cause path.

[0161] The second embodiment is a device for analyzing human factors in the PMA manufacturing process of civil aviation products, which is used to implement the method of the first embodiment. The device includes:

[0162] A vocabulary library construction module for constructing a professional vocabulary library for the PMA manufacturing of civil aviation products;

[0163] A model establishment module for establishing a production line analysis model by using the professional vocabulary library for the PMA manufacturing of civil aviation products and the Jieba vocabulary library;

[0164] A weight distribution module for analyzing the upper and lower layer factors in the production line analysis model by using chi-square test and concession ratio, and performing factor weight distribution based on conditional probability;

[0165] An analysis module for DEMATEL analysis; first, based on the results of chi-square test and concession ratio analysis, an initial relationship matrix is established, then the normal matrix N is calculated according to the initial relationship matrix, the comprehensive influence matrix T is calculated according to the normal matrix N, and finally, by analyzing the comprehensive influence matrix T, the influence degree C of each factor is calculated i , the influenced degree D i , the centrality M i and the cause degree R i ;

[0166] A sorting module for TOPSIS comprehensive evaluation and sorting. First, the influence degree C i , the influenced degree D i , the centrality M i and the cause degree R i are regarded as the attributes of each scenario, then the standard matrix is calculated, and finally the optimal solution distance, the worst solution distance and the score of the i-th influencing factor of the target object are calculated.

[0167] The third embodiment is a computer-readable storage medium that stores a computer program, and when the program is executed by a processor, it implements the above-mentioned method for analyzing human factors in the manufacturing process of civil aviation product PMA.

[0168] The fourth embodiment is a computer program product, including a computer program, and when the computer program is executed by a processor, it implements the above-mentioned method for analyzing human factors in the manufacturing process of civil aviation product PMA.

[0169] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented in whole or in part in the form of a computer program product, the computer program product includes one or more computer instructions. When the computer program instructions are loaded or executed on a computer, the processes or functions described in the embodiments of the present invention are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by wire (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that the computer can access or a data storage device such as a server or data center that includes one or more integrated available media. The available medium can be a magnetic medium (such as a floppy disk, hard disk, magnetic tape), an optical medium (such as a DVD), or a semiconductor medium (such as a solid-state disk (SSD)).

[0170] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A method for analyzing human factors in the manufacturing process of civil aviation products PMA, characterized in that, Including: S1. Construct a professional vocabulary library for the PMA manufacturing of civil aviation products; S2. Establish a production line analysis model using the professional vocabulary library for the PMA manufacturing of civil aviation products and the Jieba word library; S3. Analyze the upper and lower level factors in the production line analysis model using chi-square test and concession ratio, and allocate factor weights based on conditional probability; S4, DEMATEL analysis; First, based on the results of chi-square test and concession ratio analysis, an initial relationship matrix is established. Then, the normal matrix N is calculated according to the initial relationship matrix, and the comprehensive influence matrix T is calculated based on the normal matrix N. Finally, by analyzing the comprehensive influence matrix T, the influence degree C of each factor is calculated i , the affected degree D i , the centrality M i , and the reason degree R i ; S5. TOPSIS comprehensive evaluation and ranking. First, regard the influence degree C i , the influenced degree D i , the centrality M i and the reason degree R i as the attributes of each scenario, then calculate the standard matrix, and finally calculate the optimal solution distance, the worst solution distance of the target object and the score of the i-th influencing factor.

2. The method for analyzing human factors in the PMA manufacturing process of civil aviation products according to claim 1, wherein The construction of the professional vocabulary library for the PMA manufacturing of civil aviation products includes: sorting out the professional vocabulary of the PMA manufacturing of civil aviation products, combining the content of relevant documents, collecting and summarizing from five aspects of people, machine, material, method, and environment to obtain the professional vocabulary library for the PMA manufacturing of civil aviation products, and supplementing the professional vocabulary library for the PMA manufacturing of civil aviation products into the basic vocabulary library to obtain a new professional vocabulary library for the PMA manufacturing of civil aviation products.

3. The method for analyzing human factors in the PMA manufacturing process of civil aviation products according to claim 2, characterized in that, The establishment of the production line analysis model includes: first, adding the professional vocabulary library for the PMA manufacturing of civil aviation products to the Jieba word library, then using the Python Jieba word segmentation method to extract the vocabulary information in the historical error cases of MRO enterprises to obtain a keyword library; subsequently, based on the keyword library, conducting a questionnaire survey for different positions of personnel, statistically screening the results of the questionnaire survey, and counting the consistency of each vocabulary. When the consistency exceeds 50%, the vocabulary is retained, otherwise it is excluded. After statistical screening, five-level indicators are obtained; finally, adding the entries of the five-level indicators into the HFACS-ME simplified model to form a production line analysis model.

4. The method for analyzing human factors in the PMA manufacturing process of civil aviation products according to claim 3, wherein The allocation of factor weights based on conditional probability includes: first, using the chi-square test to analyze whether there is a significant causal relationship between the upper and lower level factors in the production line analysis model, and then adjusting the concession ratio based on the conditional probability factor to reflect the influence coefficient between the upper and lower level factors of the production line analysis model.

5. The method for analyzing human factors in the PMA manufacturing process of civil aviation products according to claim 4, characterized in that The DEMATEL analysis includes: establishing an initial relationship matrix. First, based on the results of chi-square test and concession ratio analysis, analyze the influence degrees among various factors in the production line analysis model. Then, divide the influence degrees into several levels. Finally, generate factor A i For factor A j The direct relationship matrix H of the influence degree = [h ij n×n , where n is the number of rows and columns of the direct relationship matrix H, i and j represent the i-th row and the j-th column in the direct relationship matrix H, and h ij represents the element in the i-th row and the j-th column of the direct relationship matrix H;​ Calculate the normal matrix N: Calculate the comprehensive influence matrix T; I is an n×n order identity matrix; By analyzing the comprehensive influence matrix T, calculate the influence degree C of each factor i , the influenced degree D i , the centrality M i and the reason degree R i ; Construct a factor influence relationship graph: use the centrality M i as the horizontal axis and the reason degree R i as the vertical axis to draw a scatter plot 6. The method for analyzing human factors in the PMA manufacturing process of civil aviation products according to claim 4, wherein Influence degree C i and influenced degree D i are respectively obtained from the sum of the elements in the i-th row and the i-th column of the comprehensive influence matrix T; Centrality M i is obtained by adding D i to C i ; Cause degree R i is obtained by subtracting D i from C i . The higher the value of M i , the more important the status of this factor. The positive or negative value of R i is used to judge the nature of the factor: when R i > 0, this factor is the influence source; when R i is less than 0, this factor is the influenced object.

7. The method for analyzing human factors in the PMA manufacturing process of civil aviation products according to claim 5, wherein: Influence degree C i , Degree of being influenced D i , Centrality M i and Cause degree R i The calculation formulas are as follows: h ji represents the element in the \(j\)-th row and \(i\)-th column of the direct relationship matrix \(H\); M i = C i + D i ; R i = C i - D i .

8. The method for analyzing human factors in the PMA manufacturing process of civil aviation products according to claim 5, wherein: The influence degree C i and the influenced degree D i as well as the centrality M i and the reason degree R i are regarded as the attributes of each scenario, an original matrix X is established, and the original matrix X is standardized to obtain a standard matrix Z; The formula for calculating the standard matrix is as follows: Among them, X ij represents the element in the i-th row and j-th column of the original matrix X; Optimal solution distance of the target object The formula is: where m is the number of scenarios in the original matrix X, represents the maximum value of the j-th column in the standard matrix Z, z ij represents the element in the i-th row and j-th column of the standard matrix Z; The worst solution distance of the target object The formula is: Among them, represents the minimum value of the j-th column in the standard matrix Z; The score S of the i-th influencing factor i The calculation formula is as follows: Among them, represents the optimal solution distance of the target object, represents the worst solution distance of the target object.

9. An artificial factor analysis device for the PMA manufacturing process of a civil aviation product, characterized in that, Including: A vocabulary library construction module for constructing a professional vocabulary library for the PMA manufacturing of civil aviation products; A model establishment module for establishing a production line analysis model using the professional vocabulary library for the PMA manufacturing of civil aviation products and the Jieba word library; A weight allocation module for analyzing the upper and lower level factors in the production line analysis model using chi-square test and concession ratio, and allocating factor weights based on conditional probability; Analysis module, DEMATEL analysis; First, based on the results of chi-square test and concession ratio analysis, an initial relationship matrix is established. Then, the normal matrix N is calculated according to the initial relationship matrix, and the comprehensive influence matrix T is calculated according to the normal matrix N. Finally, by analyzing the comprehensive influence matrix T, the influence degree C of each factor is calculated i , the influenced degree D i , the centrality M i , and the reason degree R i ; Sorting module, TOPSIS comprehensive evaluation and sorting. First, consider the influence degree C i , the influenced degree D i , the centrality M i and the reason degree R i as the attributes of each scenario, then calculate the standard matrix, and finally calculate the optimal solution distance, the worst solution distance of the target object and the score of the i-th influencing factor.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the method for analyzing human factors in the PMA manufacturing process of civil aviation products according to any one of claims 1-8.

Images