Improved FMEA method based on interval type-2 fuzzy set and DEMATEL-VIKOR mixed multi-criterion decision

Abstract

According to the improved FMEA method based on the interval type-2 fuzzy set and the DEMATEL-VIKOR mixed multi-criterion decision, on the basis of traditional FMEA, a more perfect IMS key hardware risk assessment model is constructed by improving related methods. A related failure mode is obtained through a fault tree analysis method. According to the invention, the interval type-2 fuzzy set and the FMEA are combined to solve the problem of inaccurate RPN value in the traditional FMEA; calculating comprehensive weights of experts by utilizing a similarity aggregation method and combining personal backgrounds of the experts; in the aspect of risk factors, a combined weight method is adopted, so that the analysis precision can be improved, and the contribution of each factor to the overall risk can be evaluated more comprehensively by comprehensively considering the weights of a plurality of indexes; the DEMATEL-VIKOR hybrid multi-criterion decision-making method for preference relationship analysis fully considers the relevance between factors and the comprehensive influence of multiple criteria, and provides relatively comprehensive and visual decision-making support.

Description

Technical Field

[0001] The present invention belongs to the field of risk assessment and management, especially in the application of industrial systems or complex mechanical systems. The invention mainly emphasizes improving the accuracy and reliability of risk assessment, and solves some limitations existing in traditional FMEA by combining different technologies and methods, such as the inaccuracy of RPN value (risk priority number) and the subjectivity of expert evaluation. In addition, it also considers the interaction and comprehensive impact between risk factors, thereby providing a more comprehensive and systematic risk assessment. Background Art

[0002] Risk assessment and failure prevention are crucial in the management of modern industrial systems and complex mechanical equipment. Traditional Failure Mode and Effects Analysis (FMEA) methods are widely used to identify potential failure modes and their potential impact on system performance. However, traditional FMEA faces a series of challenges, including the inaccurate calculation of risk priority number (RPN) and the lack of ability to consider the interaction between failure modes. In addition, the FMEA process often relies on the subjective judgment of experts, which may lead to inconsistency and unreliability of risk assessment results.

[0003] In order to overcome these limitations, a variety of methods and technologies have emerged in recent years, aiming to improve the accuracy and systematicity of risk assessment. For example, the introduction of fuzzy logic theory provides an effective means to deal with uncertainty and ambiguity. Relative interval type-2 fuzzy sets provide a more flexible and detailed method to express and deal with uncertainty in expert evaluation. At the same time, multi-criteria decision-making methods such as the DEMATEL-VIKOR hybrid method enhance the comprehensiveness and intuitiveness of the decision-making process by considering the correlation between factors and the combined influence of multiple criteria. In addition, the application of advanced technologies such as similarity aggregation method and risk factor combination weight method has further improved the accuracy and reliability of the evaluation process.

[0004] As industrial systems become increasingly complex, there is a growing demand for more efficient and accurate risk assessment methods. This requires the development of new risk assessment models that can not only address the limitations of traditional FMEA methods, but also adapt to the complexity and dynamics of modern industrial environments. This invention was created in this context and aims to provide a comprehensive and advanced risk assessment solution to meet the challenges in modern industrial systems. Summary of the invention

[0005] 1. Purpose: The purpose of this invention is to build a more efficient and accurate risk assessment model by integrating and improving multiple risk assessment tools and methods, so as to better serve the risk management needs of modern industrial systems.

[0006] 2. Technical solution:

[0007] S1. Establish a FMEA team consisting of K experts to analyze the potential failure modes of IMS key hardware. i (i=1, 2, ..., m) are identified to construct a set of IMS key hardware failure modes.

[0008] S2. Based on the IMS key hardware failure mode set, the traditional three risk factors (severity S, occurrence O, detection D) are expanded to four (severity S, occurrence O, detection D, economy E).

[0009] S3. Establish interval type-2 fuzzy sets, which are interval type-2 trapezoidal fuzzy numbers. The relative consistency of each expert is calculated based on the expert's scoring results. Describe the evaluation criteria of the four risk factors under the corresponding linguistic variables (Table 1), and produce the type-2 fuzzy numbers corresponding to the linguistic variables (Table 2);

[0010] Furthermore, K experts scored based on the evaluation table (Table 1), and the evaluation results of severity, occurrence, detection, and economy of each expert in each failure mode were set as

[0011] Each value in the expert scoring results (Table 11) is converted into an interval type-2 trapezoidal fuzzy number in the corresponding linguistic variable:

[0012] set up is an interval type-2 fuzzy number, if The upper membership function and the lower membership function are two trapezoidal type 1 fuzzy numbers, then It is called trapezoidal interval type-2 fuzzy number; the reference point in the domain and the height of the upper and lower membership functions of the trapezoidal interval type-2 fuzzy number are used to characterize the trapezoidal interval type-2 fuzzy number.

[0013] Representation elements In the upper trapezoidal membership function The membership value in Representation elements In the following trapezoidal membership function The membership value in

[0014] Combined with expert background weight Relative agreement with experts The combined weight of the j-th risk factor of the k-th expert under the i-th failure mode can be calculated as follows:

[0015]

[0016] S4. Collect experts’ personal background scores Combined with similarity aggregation method to calculate the consistency of expert evaluation results Derive the expert's personal weights for each risk factor for each failure mode That is, the weight of each expert in each risk factor.

[0017] S5, then follow the And the evaluation results The corresponding type II trapezoidal fuzzy number weighted average is used to obtain the comprehensive fuzzy evaluation matrix F of severity, occurrence, detection and economy under each failure mode. ij ) m×4 ; m represents the number of failure modes, 4 represents the number of risk factors;

[0018] S6. Use DEMETAL to calculate the influence and impact of each failure mode;

[0019] Furthermore, the comprehensive evaluation value of severity is modified to obtain the comprehensive evaluation matrix

[0020] S7. Use the analytic hierarchy process to calculate the expert risk factor impact score results to obtain the subjective weight of the risk factor, and then combine it with the objective weight of the risk factor to obtain the final combined weight of the risk factor.

[0021] S8. Calculate the distance between fuzzy numbers based on relative preference relationship analysis and rank the failure modes using the VIKOR method. Relevant personnel formulate corresponding prevention and response measures based on the ranking;

[0022] Furthermore, in S3, the language evaluation terms include five levels: very low, low, average, high and very high. The five levels of the language evaluation terms correspond to 1 point, 2 points, 3 points, 4 points and 5 points, respectively, as shown in Table 1, and the corresponding type-2 fuzzy numbers are shown in Table 2.

[0023] Table 1 Language assessment terms

[0024]

[0025] Table 2 Type-2 fuzzy numbers corresponding to linguistic variables

[0026]

[0027] Furthermore, in S4, the expert weight The calculation includes the following steps:

[0028] S41. In this patent, the expert background score standard table is established based on the survey indicators of educational background, professional title, years of work experience, and research direction / position, where the total score of the kth expert background is It is obtained by adding up the scores of each indicator. The formula for calculating the expert background weight is:

[0029]

[0030] S42. Use the distance formula between interval trapezoidal fuzzy numbers to calculate the distance d between the evaluation result of the kth expert and the evaluation results of other experts. ij (k, u), the calculation formula is as follows:

[0031]

[0032] Then according to Calculate the average distance between the evaluation results of the kth expert and other experts u represents other experts except the kth expert,

[0033] represents the interval trapezoidal fuzzy number corresponding to the evaluation result of the kth expert,

[0034] represents the interval trapezoidal fuzzy number corresponding to the evaluation results of other experts among the K experts;

[0035] S43, according to Calculate the average agreement of the kth expert;

[0036] S44, according to The relative consistency of each expert is calculated using the normalization method

[0037] S45. Combined with expert background weight Relative agreement with experts The combined weight of the j-th risk factor of the k-th expert under the i-th failure mode can be calculated as follows:

[0038]

[0039] Among them, α and β are relaxation factors of background weight and relative consistency, respectively, reflecting the relative importance between weights, satisfying 0≤α≤1, 0≤β≤1 and α+β=1. Generally, α=β=0.5.

[0040] Furthermore, in S6, the steps of determining the influence degree and the influence degree of the failure mode are as follows:

[0041] S61. Experts analyze the influence relationship between failure modes and obtain the direct correlation matrix between failure modes

[0042]

[0043] Among them Indicates the kth expert's evaluation of failure mode FM i For failure mode FM j If there is no impact, the score is 0;

[0044] S62. Calculate the direct association aggregation matrix

[0045]

[0046] Direct Link Aggregation Matrix Elements in By weight of each expert Aggregate

[0047]

[0048] After aggregation, defuzzification is performed to obtain a clear and directly related aggregation matrix

[0049]

[0050] S63. Calculate the normalized direct association aggregation matrix D. The common method for normalizing the direct association aggregation matrix is ​​based on the maximum value of the sum of the row vector elements of the matrix.

[0051]

[0052] for The elements in Represents the maximum value of the sum of elements in each row;

[0053] S64. Calculate the total incidence matrix T, where I is the identity matrix

[0054]

[0055] Where D n Represents the multiplication of n D matrices;

[0056] S65. Determine the influence and impact of failure mode

[0057] Define the sum of the i-th row of matrix T as R i , the sum of the jth column is defined as C j ,

[0058] Ri Indicates the failure mode FM i Direct and indirect effects on other failure modes, C j Indicates the failure mode FM i Direct and indirect effects of other failure modes

[0059]

[0060] Let i = j, influence R i +C i Expressed failure mode FM i The total degree of being affected by other failure modes and affecting other failure modes; the degree of being affected R i -C i Expressed failure mode FM i The difference between the degree of being affected by other failure modes and the degree of affecting other failure modes;

[0061] S66, R i and C i Normalize them separately and get and Assume that the expert has a good understanding of the failure mode FM i The initial comprehensive severity S assessment value is

[0062] After DEMATEL analysis, the corrected failure severity is

[0063] Furthermore, in S7, the scoring criteria and scoring table for the relative importance of risk factors are as shown in Tables 3 and 4.

[0064] Table 3. Scoring criteria for relative importance of risk factors

[0065]

[0066] Table 4 Risk factor relative importance score table

[0067]

[0068]

[0069] When determining the risk factor weights, the analytic hierarchy process is first used to calculate the expert risk factor impact score results to obtain the risk factor subjective weights;

[0070] That is, use the scoring criteria mentioned in Table 3 to compare the risk factors in Table 4 in pairs, and then summarize and average the scoring results of each group. Similarly, set each expert to have the same weight, use the hierarchical analysis method to calculate the subjective weight ω1 of the risk factor, and combine it with the objective weight ω2 of the risk factor. The following formula is used to finally obtain the combined weight ω of the risk factor, where α is usually 0.5, where ω2=1 / K

[0071] ω=α×ω1+(1-α)×ω2

[0072] Furthermore, in S8, the priority ranking of failure modes is calculated by the VIKOR method using relative preference relationship analysis:

[0073] S81. Comprehensive evaluation matrix calculated based on expert evaluation results (Table 11) Find the positive ideal solution in the failure mode Negative ideal solution

[0074] in, represents the severity of the ith failure mode after being aggregated by experts and corrected by the DEMATEL method, They respectively represent the occurrence, detection and economy of the i-th failure mode after expert aggregation and correction by the DEMATEL method.

[0075] S82, Assumption interval type-2 fuzzy number and They are The upper and lower bounds of membership functions; and B L They are We want to calculate the upper and lower bounds of the membership function. Relative to Preference in, and They are respectively arrive and from arrive The preference score is calculated as follows:

[0076]

[0077] Then calculate the group benefit B of each failure mode i and the maximum individual regret T i as follows:

[0078]

[0079] Among them, ω O ,ω S ,ω D and ω E , both refer to the combined weights of risk factors.

[0080] S83. Calculate the comprehensive VIKOR index Q for each failure mode i .

[0081] S84. Defuzzified result Q i Sorting, if the following two conditions are met, the failure mode FM (1) The optimal compromise solution.

[0082] (1)Q(FM (2) )-Q(FM (1) )≥1 / (n-1), where FM (1) is the Q value of the best solution in the sorting result, FM (2) is the Q value of the suboptimal solution in the sorting result, and n is the total number of failure modes;

[0083] (2) The best solution B i Value and T i The value must also be optimal, so that the solution with the smallest Q value can be ensured to be the optimal solution. If the above two conditions cannot be met at the same time, a set of compromise solutions can be obtained, which are as follows:

[0084] (3) If only condition (1) is met, a set of compromise solutions is obtained: FM (1) ,FM (2) All are compromise solutions;

[0085] (4) If only condition (2) is met, a set of compromise solutions is obtained: FM (1) ,...,FM (n) , that is, they are the most important failure modes;

[0086] Compared with the prior art, the present invention has the following beneficial effects:

[0087] 1. The present invention adds economic degree on the basis of traditional severity, occurrence and detection degree, which can more comprehensively reflect the economic losses that may be caused by the failure mode and improve the comprehensiveness of the assessment.

[0088] 2. The present invention calculates expert weights by combining the expert's personal background and similarity aggregation method, which effectively solves the problem of inconsistent expert subjective evaluation results.

[0089] 3. The present invention uses the DEMATEL method to calculate the influence relationship between failure modes, clarify the influence and influence of each failure mode, so as to better optimize risk decision-making.

[0090] 4. The present invention adopts the DEMATEL-VIKOR hybrid decision-making method, which not only considers the comprehensive impact of multiple risk factors, but also provides more intuitive and comprehensive decision support through a ranking method.

[0091] 5. The present invention uses a combined weight method of risk factors and a strategy for correcting severity evaluation values ​​to make the risk assessment model more adaptable in a dynamic environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0092] Figure 1 It is a schematic diagram of the method proposed by the present invention. DETAILED DESCRIPTION

[0093] The present invention is further described below in conjunction with embodiments.

[0094] This embodiment analyzes and evaluates the turbine engine on the aircraft. In this embodiment, the FMEA team is composed of expert one, senior manager of information technology project k1, expert two, intelligent manufacturing system implementation consultant k2, expert three, technical manager k3, and expert four, university scholar k4 in the field of intelligent manufacturing.

[0095] First, we identify the risks of turbine engines by conducting a top-down hierarchical analysis of turbine engines, combined with brainstorming, expert interviews, and the organization and summary of relevant literature. We divide them into six modules: turbocharger system, cooling system, electronic control system, fuel supply system, airframe structure and support system, and lubrication system. Then, we analyze each system in detail, identify the key failure modes of each system, and evaluate their possible impacts. For ease of presentation, these failure modes are numbered FM i (i=1, 2, ..., 19) indicates that the relevant failure modes and their consequences and causes are listed in Table 5, Table 6, Table 7, Table 8, Table 9 and Table 10.

[0096] Table 5 Failure modes and consequences of turbine engine systems

[0097]

[0098]

[0099] Table 6 Failure modes and consequences of cooling system

[0100]

[0101] Table 7 Failure modes and consequences of electronic control systems

[0102]

[0103] Table 8 Failure modes and consequences of fuel supply system

[0104]

[0105] Table 9 Failure modes and consequences of airframe structure and support systems

[0106]

[0107] Table 10 Failure modes and consequences of lubrication system

[0108]

[0109] Each expert of the FMEA team uses the scores of the five linguistic variables to evaluate the risk factors of the failure mode according to the evaluation criteria corresponding to the linguistic variables in Table 1, and the expert evaluation table is shown in Table 11. Substitution is performed based on the type-2 fuzzy number of the fuzzy linguistic variable.

[0110] Table 11 Expert scoring table

[0111]

[0112] The scores of each expert are collected, i.e., Table 11, and then the language variables (1 to 5) in the expert score table are converted into type II interval fuzzy numbers according to the conversion relationship in Table 2. Since each expert has different professional backgrounds and personal judgments, the personal background scores of the experts are collected. Then calculate the distance between the evaluation results of the expert and other experts. For example, the interval type-2 fuzzy numbers corresponding to the evaluation results O of the occurrence degree of FM1 of the first expert and the second expert are [(1,2,4,5; 1,1), (1.2,2.2,3.8,4.8; 0.8,0.8)], [(1,1,1,1; 1,1), (1,1,1,1; 1,1)] respectively;

[0113] Using the formula

[0114]

[0115] The distance between the two people can be calculated as d(1,2)=27.63. Similarly, the distances between the first expert and the third and fourth experts are d(1,3)=27.63 and d(1,4)=27.63 respectively. The average distance is Average Agreement Similarly, the average agreement of other experts is calculated as

[0116] Then, the consistency of the expert evaluation results is calculated by combining the similarity aggregation method. Derive the expert's personal weights for each risk factor for each failure mode In order to obtain more accurate evaluation results.

[0117] The expert weight matrix is ​​obtained (Table 12).

[0118] Table 12 Expert weight table

[0119]

[0120] The scoring results of each risk factor of each failure mode are summarized (Table 11), and the type II trapezoidal fuzzy numbers corresponding to the scoring results of the four experts are compared with the expert weights. The weighted average is used to obtain the comprehensive fuzzy evaluation matrix F = (f ij ) m×4 (Table 13). Table 13 is a comprehensive fuzzy evaluation matrix of an expert.

[0121]

[0122] Then use DEMETAL to calculate each failure mode FM i of and i Indicates the failure mode FM i Direct and indirect effects on other failure modes, C j Indicates the failure mode FM i Affected directly and indirectly by other failure modes, R i and C i Normalize them separately to get and use The comprehensive evaluation value of severity is modified to obtain the modified comprehensive fuzzy evaluation matrix (Table 15).

[0123] Among them, we need to calculate the direct correlation aggregation matrix. Since we divide the failure modes into failure modes in different systems, and different systems have basically no impact on each other, we can analyze the impact between failure modes in different systems. For example, in the electronic control system, its direct correlation aggregation matrix is ​​Table 14.

[0124] Table 14 Electronic control system direct association aggregation matrix

[0125]

[0126] The values ​​in Table 14 are obtained by aggregating the ratings of four experts. For example, the impact of FM8 on FM9 is calculated by the weight of each expert. And the evaluation results are aggregated, that is, in, is the type II trapezoidal fuzzy number corresponding to the four experts’ rating results of the impact of FM8 on FM9, The weight representing the influence of the k-th expert on FM8 on FM9 is calculated in the same way as the expert weight method for calculating the comprehensive fuzzy evaluation matrix F above, that is, it can be obtained by calculating the relative consistency of the scoring results of each expert.

[0127] When determining the weight of risk factors, we first use the analytic hierarchy process to calculate the expert risk factor impact score results to obtain the risk factor subjective weight. Use the scoring criteria mentioned in Table 3 to compare the risk factors in Table 4 in pairs, and then summarize and weight the score results of each group.

[0128] Similarly, each expert is set to have the same weight. According to the comprehensive scores of the experts in the comprehensive score table of the relative importance of risk factors given by the experts (Table 16), the total score of S is calculated as 1+0.875+1=2.875, where the score is obtained by taking the average of the scores of the above experts, for example Similarly, the total scores of O, D, and E are 2, 1.768, and 1.6 respectively. After normalization, their respective subjective weights are (0.386, 0.262, 0.203, and 0.149).

[0129] The “—” in the expert scoring table of the relative importance of risk factors means that experts do not need to score. When calculating the comprehensive score matrix, the score in the lower left corner corresponds to the value in the upper right corner, where 1 and 0 correspond to each other, and the remaining values ​​are reciprocals of each other. For example, the influence of O on S is 0, which corresponds to the influence of S on O of 1, and the influence of D on S is 1.413, the reciprocal of the influence of S on D of 0.875.

[0130]

[0131] Table 16 Comprehensive score table of relative importance of risk factors

[0132]

[0133] Combined with the objective weights of the risk factors, each expert is usually given the same objective weight, i.e. (0.25, 0.25, 0.25, 0.25), and the final combined weight of the risk factors is calculated to be 0.5×(0.386, 0.262, 0.203, 0.149)+0.5×(0.205, 0.25, 0.25, 0.25)=(0.318, 0.256, 0.226, 0.200).

[0134] The VIKOR method is used to determine the positive and negative ideal values ​​in the failure mode of the comprehensive evaluation matrix, and the comprehensive evaluation matrix is ​​defuzzified to obtain the clear values ​​of each risk factor S, O, D, and E corresponding to each failure mode, as shown in Table 17. Then the maximum group benefit value B is calculated. i and the minimum individual regret value T i , according to B i 、T i Calculate the comprehensive index of each failure mode The B value of each failure mode is obtained. i 、T i , Q values, as shown in Table 18.

[0135] Table 17 Explicit value table of each failure mode corresponding to each risk factor S, O, D, E

[0136]

[0137]

[0138] Table 18 B of each failure mode i 、T i , Q value

[0139]

[0140] Where Q(FM (2) )-Q(FM (1) )=0.203≥0.056, it can be seen that FM14 meets condition (1), and the T of FM14 i Value or B i The value is also optimal compared to other failure modes. It can be seen that FM14 also satisfies condition (2), so FM14 is the compromise solution to this problem, that is, FM14 is the failure mode with the highest risk.

[0141] In addition, as shown in Table 18, high-risk failure modes mainly include FM1, FM2, FM12, FM13, FM14 and FM17, which require the most resources for monitoring, prevention and correction; medium-risk failure modes mainly include FM3, FM6, FM7, FM15, FM16, FM18 and FM19, which require medium resources to monitor and prevent related systems and equipment; low-risk failure modes mainly include FM4, FM5, FM8, FM9, FM10 and FM11, which only require a small amount of resources for monitoring and prevention.

[0142] Through example analysis, the present invention is applicable and effective in risk assessment and management, and can better serve the risk management needs of modern industrial systems.

[0143] The above is only a specific embodiment of the present invention and does not limit the present invention in any way. Any technical personnel in the relevant technical field, without departing from the scope of the technical solution of the present invention, makes any form of modification or equivalent replacement to the technical solution and content disclosed in the present invention, which does not depart from the content of the technical solution of the present invention and still falls within the protection scope of the present invention.

Claims

1. An improved FMEA method based on interval type-2 fuzzy sets and DEMATEL-VIKOR hybrid multi-criteria decision making, characterized in that: The following steps are involved: S1. Establish a FMEA team consisting of K experts to analyze the potential failure modes of IMS key hardware. i Identify and build a set of IMS key hardware failure modes; S2, expand the traditional three risk factors to four; S3. Establish interval type-2 fuzzy sets: Describe the evaluation criteria of the four risk factors under the corresponding language variables and make a table of evaluation criteria corresponding to the language variables. K experts score based on the evaluation table. Suppose the evaluation results of severity, occurrence, detection and economy of each expert under each failure mode are The expert's rating results are converted into interval type-2 trapezoidal fuzzy numbers in the corresponding linguistic variables; S4. Collect experts’ personal background scores Combined with similarity aggregation method to calculate the consistency of expert evaluation results Derive the expert's personal weights for each risk factor for each failure mode S5, then follow the The evaluation results are The corresponding type II trapezoidal fuzzy number weighted average is used to obtain the comprehensive fuzzy evaluation matrix of severity, occurrence, detection and economy under each failure mode. S6. Calculate the influence and the affected degree of each failure mode by using DEMETAL, and modify the comprehensive evaluation value of the severity in the comprehensive fuzzy evaluation matrix F; S7. Use the analytic hierarchy process to calculate the expert risk factor impact score results to obtain the risk factor subjective weight, and then combine the objective weight of the risk factor to obtain the final risk factor combined weight; S8. Calculate the distance between fuzzy numbers based on relative preference relationship analysis and sort the failure modes using the VIKOR method. Relevant personnel formulate corresponding prevention and response measures based on the sorting.

2. The improved FMEA method based on interval type-2 fuzzy sets and DEMATEL-VIKOR hybrid multi-criteria decision making as claimed in claim 1, characterized in that: In step S2, economic degree is added to the traditional risk factors.

3. The improved FMEA method based on interval type-2 fuzzy sets and DEMATEL-VIKOR hybrid multi-criteria decision making as claimed in claim 1, characterized in that: The interval type-2 fuzzy set in step S3 includes five linguistic variables, including five levels of very low VL, low L, general M, high H, and very high VH, and the corresponding evaluation language evaluation numbers are 1, 2, 3, 4, and 5.

4. The improved FMEA method based on interval type-2 fuzzy sets and DEMATEL-VIKOR hybrid multi-criteria decision making as claimed in claim 1, characterized in that: Expert weights in step S4 The calculation includes the following steps: S41. Establish an expert background score standard table, where the total score of the kth expert background is It is obtained by adding up the scores of each indicator to calculate the expert background weight The formula is: S42. Use the distance formula between interval trapezoidal fuzzy numbers to calculate the distance d between the evaluation result of the kth expert and the evaluation results of other experts among the K experts. ij (k, u), the calculation formula is as follows: Then according to Calculate the average distance between the evaluation result of the kth expert and other experts among the K experts u represents other experts except the kth expert; S43, according to Calculate the average agreement; S44, according to The relative consistency of each expert is calculated using the normalization method S45. Combined with expert background weight Relative agreement with experts The combined weight of the j-th risk factor of the k-th expert under the i-th failure mode can be calculated as follows: Where α and β are relaxation factors of background weight and relative consistency, respectively, 0≤α≤1, 0≤β≤1 and α+β=1.

5. The improved FMEA method based on interval type-2 fuzzy sets and DEMATEL-VIKOR hybrid multi-criteria decision making as claimed in claim 1, characterized in that: The steps of determining the influence degree and the influence degree of the failure mode in step S6 are as follows: S61. Experts analyze the influence relationship between failure modes and obtain the direct correlation matrix between failure modes: Among them Indicates the kth expert's evaluation of failure mode FM i For failure mode FM j If there is no impact, the score is 0; S62. Calculate the direct association aggregation matrix: Direct Link Aggregation Matrix Elements in By weight of each expert Aggregate After aggregation, defuzzification is performed to obtain a clear and directly related aggregation matrix S63. Calculate the normalized direct association aggregation matrix D. The common method for normalizing the direct association aggregation matrix is ​​based on the maximum value of the sum of the row vector elements of the matrix. for The elements in Represents the maximum value of the sum of elements in each row; S64. Calculate the total incidence matrix T, where I is the identity matrix: Where D n Represents the multiplication of n D matrices; S65. Determine the influence and impact of the failure mode, and define the sum of the i-th row of the matrix T as R i , the sum of the columns is defined as C j , R i Indicates the failure mode FM i Direct and indirect effects on other failure modes, C j Indicates the failure mode FM i Directly and indirectly affected by other failure modes: Let i = j, influence R i +C i Expressed failure mode FM i The total degree of being affected by other failure modes and affecting other failure modes, the degree of influence R i -C i Expressed failure mode FM i The difference between the degree of being affected by other failure modes and the degree of affecting other failure modes; S66, R i and C i Normalize them separately and get and Assume that the expert has a good understanding of the failure mode FM i The initial severity S assessment value is After DEMATEL analysis, the corrected failure severity is 6. The improved FMEA method based on interval type-2 fuzzy sets and DEMATEL-VIKOR hybrid multi-criteria decision making as claimed in claim 1, characterized in that: In step S8, relative preference relationship analysis is used to calculate the priority ranking of failure modes through the VIKOR method, and the steps are as follows: S81. Comprehensive evaluation matrix calculated based on expert evaluation results Find the positive ideal solution in the failure mode Negative ideal solution S82. Calculate the group benefit B for each failure mode i and the maximum individual regret T i ; S83. Calculate the comprehensive VIKOR index Q for each failure mode i ; S84. Defuzzified result Q i Sort.

Images