System theory process analysis and quantification method and system for an aeronautical power system

Abstract

The application discloses a kind of system theory process analysis quantification method and system of aviation power system, belong to the field of civil aviation safety analysis, unsafe control behavior of influencing aviation power system is as final leaf node, cause scene is as intermediate leaf node, cause factor information is as root node, establishes aviation power system bayesian network model;Expert scoring is carried out to the occurrence probability of root node unsafe problem;Weight distribution is carried out to each expert score result, and the weight of the weight of each comparison data column after correction of modified reference data column is obtained, and then the modified expert scoring result is obtained;The root node of the modified expert scoring result is fused, and the occurrence probability of the fused root node is obtained and substituted into aviation power system bayesian network model, to obtain the posterior probability of each final leaf node, and then the system theory process analysis quantification result is obtained.The method can reduce the subjective influence of researchers on data processing process.

Description

Technical Field

[0001] This invention relates to the field of aviation safety analysis technology, and more specifically to a system and method for quantitative analysis of the theoretical process of aerospace propulsion systems based on a combination of grey relational analysis theory, DS evidence theory and Bayesian network model. Background Technology

[0002] In the field of civil aviation, safety analysis of aircraft propulsion systems is essential. Systems-Theoretic Process Analysis (STPA) shifts the focus of safety analysis from fault perspective to control perspective. By analyzing hazard sources, it establishes safety constraints on component behavior, inter-component interactions and communication, external conditions, and anti-interference conditions. However, STPA can only perform qualitative analysis and cannot provide quantitative analysis for practical problems.

[0003] Bayesian Networks (BNs) leverage the transitive properties of graph structures to model causal relationships between variables, and then use Bayesian theory to solve for the joint and marginal distributions of variables. They are inference models that organically combine probability theory and graph theory, and can be used for STPA quantitative analysis. In the process of using Bayesian Networks for quantification, it is necessary to first obtain the prior probabilities of the root nodes. When empirical data is lacking, the prior probabilities can be obtained from expert experience.

[0004] In existing technologies, qualitative evaluation results from expert questionnaires are converted into fuzzy numbers using fuzzy theory to obtain prior probabilities. However, since the options in the expert questionnaires are qualitative indicators, experts cannot directly provide quantitative evaluations. The number of qualitative options is limited, and the probability values ​​after fuzzy conversion have a large range. Furthermore, the correspondence between qualitative options and fuzzy numbers is determined subjectively by researchers, which is limited by their experience level and is highly subjective.

[0005] In summary, existing analytical methods for aerospace propulsion systems do not fully utilize all the experience and information of experts, are greatly influenced by the subjective opinions of researchers, and obtain inaccurate probability values, which in turn leads to inaccurate quantitative analysis results. Summary of the Invention

[0006] To address the problems existing in the above-mentioned fields, this invention proposes a system and method for quantitative analysis of the theoretical process of aero-engine systems. This method can solve the technical problem that existing aero-engine system analysis methods do not fully utilize all the experience information of experts and are greatly influenced by the subjective opinions of researchers, resulting in inaccurate probability values ​​and thus inaccurate analysis and quantitative results.

[0007] To address the aforementioned technical problems, this invention discloses a method for the quantitative analysis of the theoretical processes of an aero-engine system, comprising the following steps:

[0008] Conduct system theoretical process analysis on the aero-engine system to determine the unsafe control behaviors, causal scenarios, and causal factors affecting the aero-engine system;

[0009] By taking unsafe control behaviors as final leaf nodes, causative scenarios as intermediate leaf nodes, and causative factor information as root nodes, the flow of node relationships is obtained through system theoretical process analysis, and a Bayesian network model of the aerospace power system is constructed.

[0010] Experts scored the probability of root node insecurity issues, generating reference data columns and comparison data columns respectively.

[0011] Based on the reference data column and each comparison data column, obtain the absolute value of the difference between the reference data column and each comparison data column; based on the maximum and minimum values ​​of the absolute values, calculate the correlation coefficient of the reference data column to obtain the weight of the corrected reference data column; based on the weight of the corrected reference data column, adjust the weight of each comparison data column to obtain the corrected weight of each comparison data column; based on the corrected reference data column and the weight of each comparison data column, calculate the discount rate of each expert weight to obtain the corrected expert scoring result.

[0012] The root nodes of the corrected expert scoring results are fused to obtain the probability of occurrence of the fused root nodes. This probability is then substituted into the Bayesian network model of the aerospace propulsion system to obtain the posterior probability of each final leaf node, thereby obtaining the quantitative results of the theoretical process analysis of the system.

[0013] Preferably, the construction of the Bayesian network model of the aerospace propulsion system includes the following steps:

[0014] Unsafe control behaviors are transformed into final leaf nodes in the Bayesian network model of aerospace power systems, causal scenarios are transformed into intermediate leaf nodes in the Bayesian network model of aerospace power systems, and causal factors are transformed into root nodes in the Bayesian network model of aerospace power systems.

[0015] Based on the process analysis and reasoning of system theory, the flow of node relationships is obtained, and the nodes of the Bayesian network model of the aerospace power system are connected to construct the Bayesian network structure of the aerospace power system.

[0016] Preferably, obtaining the absolute value of the difference includes the following steps:

[0017] The scores from each expert are used as data columns, and the identification framework is X = (A, B, C, ...), with n experts.

[0018] The scores from the most authoritative experts are selected as the reference data column X0, and the scores from other experts are selected as the comparison data column X. i Where i = 1, 2, ..., n-1;

[0019] Subtract X from X0 respectively i Obtain the absolute value Δ of the difference between the reference data column and each comparison data column. i (A)

[0020] Preferably, calculating the correlation coefficient of the reference data column includes the following steps:

[0021] Select the maximum value MaxΔ and the minimum value MinΔ among the absolute values ​​of the differences:

[0022]

[0023]

[0024] Calculate the correlation coefficient δ i (A):

[0025]

[0026] In the formula, ρ is the resolution coefficient, and its value range is [0, 1].

[0027] Preferably, the weighting of the corrected reference data column includes the following steps:

[0028] The correlation coefficient δ of each reference data column is calculated separately. i (A), δ i (B), δ i (C), ... summed, we get δ i :

[0029] δ i =δ i (A)+δ i (B)+δ i (C)+…

[0030] δ i After normalization, we obtain δ i ':

[0031]

[0032] Preferably, obtaining the corrected weights of each comparison data column includes the following steps:

[0033] The reference data column is normalized to obtain the weight ω0 of the most authoritative expert;

[0034] Let the weight ω0 of the most authoritative expert be 0.5, then we obtain the corrected weights ω of each comparison data column.i :

[0035]

[0036] Preferably, obtaining the corrected expert scoring result includes the following steps:

[0037] Calculate the discount rate α for each expert's weight. i for:

[0038]

[0039] Discount rate based on the weight of each expert α i The scores from the experts were revised to obtain new scores from the experts regarding recognition framework A, i.e., the revised scores for the reference data column:

[0040] X, i (A)=1-α i +α i X i (A)

[0041] Similarly, the new scores of each element within the same recognition framework are normalized to obtain the corrected scores for each comparison data column.

[0042] Preferably, obtaining the posterior probability of each final leaf node includes the following steps:

[0043] According to Dempster's rule of composition in the DS evidence theory, the expert scores of the root nodes are fused together. The formula for Dempster's rule of composition is as follows:

[0044]

[0045] In the formula, m(A) represents the probability of occurrence of the fused recognition frame A. i (A) is the score of A after the i-th expert has been weighted;

[0046] Substituting all the probability data of the fused recognition framework into the Dempster synthesis rule formula, we obtain the Bayesian network prior probability of each node.

[0047] The prior probabilities of each node in the Bayesian network are input into the Bayesian network model of the aerospace propulsion system, and the posterior probabilities of each final leaf node are obtained using the inference function.

[0048] Preferably, it also includes a system theory process analysis and quantification system for aero-engine systems, comprising:

[0049] The Bayesian network model construction module for aero-engine systems is used to perform system theoretical process analysis on aero-engine systems, identify unsafe control behaviors, causal scenarios, and causal factors affecting the aero-engine system; it uses unsafe control behaviors as final leaf nodes, causal scenarios as intermediate leaf nodes, and causal factor information as root nodes, and obtains the node relationship flow through system theoretical process analysis to construct a Bayesian network model of the aero-engine system.

[0050] The expert scoring result correction module is used to score the probability of root node insecurity issues by experts, generating a reference data column and comparison data columns. Based on the reference data column and each comparison data column, the absolute value of the difference between the reference data column and each comparison data column is obtained. Based on the maximum and minimum values ​​of the absolute values, the correlation coefficient of the reference data column is calculated to obtain the corrected weight of the reference data column. Based on the corrected weight of the reference data column, the weights of each comparison data column are corrected to obtain the corrected weights of each comparison data column. Based on the corrected weights of the reference data column and each comparison data column, the discount rate of each expert weight is calculated to obtain the corrected expert scoring result.

[0051] The analysis and quantification result acquisition module is used to fuse the root nodes of the corrected expert scoring results, obtain the occurrence probability of the fused root nodes, and substitute them into the Bayesian network model of the aerospace power system to obtain the posterior probability of each final leaf node, thereby obtaining the analysis and quantification results of the system theoretical process.

[0052] Compared with the prior art, the present invention has the following beneficial effects:

[0053] This invention overcomes the technical shortcomings of existing analytical methods for aero-engine systems, which fail to fully utilize all expert experience and are heavily influenced by researchers' subjectivity, resulting in inaccurate probability values ​​and consequently inaccurate quantitative analysis results. This new method treats unsafe control behaviors affecting the aero-engine system as final leaf nodes, causative scenarios as intermediate leaf nodes, and causative factor information as root nodes. Through system theoretical process analysis, it obtains the flow of node relationships and constructs a Bayesian network model of the aero-engine system. This model directly obtains expert quantitative evaluations, exhibits high accuracy in expert probability scoring, and eliminates any range of probability values. By assigning expert scores to the probability of root node insecurity issues, reference data columns and comparison data columns are generated. The absolute values ​​of the differences between the reference data columns and each comparison data column are obtained. Based on the maximum and minimum values ​​of the absolute values, the correlation coefficient of the reference data column is calculated to correct the expert scores. By correcting the scores given by each expert and fusing the root nodes of the corrected expert scores, the process of subjectively determining the correspondence between qualitative options and fuzzy numbers in existing technologies can be eliminated. This reduces the subjective influence of researchers on data processing and results in more accurate posterior probabilities. Attached Figure Description

[0054] Figure 1 This is a flowchart of the system theoretical process analysis and quantification method for aero-engine systems proposed in this invention;

[0055] Figure 2 This is a diagram of the hybrid hydrogen internal combustion engine-electric aircraft propulsion system architecture of the present invention;

[0056] Figure 3 The propulsion system Bayesian network model of this invention;

[0057] Figure 4 represents the posterior probability of each root node in the Bayesian network model of this invention. Detailed Implementation

[0058] The following will refer to the appendices in the embodiments of the present invention. Figure 1-4 The technical solutions in the embodiments of the present invention will be clearly and completely described. It should be understood that the terminology used in the present invention is only for describing particular implementation methods and is not intended to limit the present invention.

[0059] like Figure 1 As shown, this embodiment of the invention provides a method for the quantitative analysis of the theoretical processes of an aero-engine system, comprising the following steps:

[0060] S1: Perform system theoretical process analysis on the aero-engine system to determine the unsafe control action (UCA), causative scenarios, and causative factors that affect the STPA of the aero-engine system; take the unsafe control action as the final leaf node, the causative scenario as the intermediate leaf node, and the causative factor information as the root node, and obtain the node relationship flow through system theoretical process analysis to construct a Bayesian network model of the aero-engine system.

[0061] S2: Experts score the probability of root node insecurity issues, generating reference data columns and comparison data columns respectively;

[0062] S3: Based on the generated reference data column and each comparison data column, obtain the absolute value of the difference between the reference data column and each comparison data column; based on the maximum and minimum values ​​of the absolute values, calculate the correlation coefficient of the reference data column to obtain the weight of the corrected reference data column; based on the weight of the corrected reference data column, adjust the weight of each comparison data column to obtain the corrected weight of each comparison data column; based on the corrected reference data column and the weight of each comparison data column, calculate the discount rate of each expert weight to obtain the corrected expert scoring result.

[0063] S4: The corrected expert scores are fused to obtain the probability of the occurrence of the root node. This probability is then substituted into the Bayesian network model of the aerospace propulsion system to obtain the posterior probability of each final leaf node, thereby obtaining the quantitative results of the theoretical process analysis of the system.

[0064] In step S1, a Bayesian network model of the aerospace propulsion system is constructed, including the following steps:

[0065] Unsafe control behaviors are transformed into final leaf nodes in the Bayesian network model of the aerospace power system, causal scenarios are transformed into intermediate leaf nodes in the Bayesian network model of the aerospace power system, and causal factors are transformed into root nodes in the Bayesian network model of the aerospace power system.

[0066] Based on the process analysis and reasoning of system theory, the flow of node relationships is obtained, and the nodes of the Bayesian network model of the aerospace power system are connected to construct the Bayesian network model of the aerospace power system.

[0067] In step S2, experts in the field are invited to score the probability of root node insecurity issues.

[0068] In step S3, the absolute value of the difference is obtained, including the following steps:

[0069] The scores from each expert are used as data columns, and the identification framework is X = (A, B, C, ...), with n experts.

[0070] The scores from the most authoritative experts are selected as the reference data column X0, and the scores from other experts are selected as the comparison data column X. i Where i = 1, 2, ..., n-1;

[0071] Subtract X from X0 respectively i Obtain the absolute value Δ of the difference between the reference data column and each comparison data column. i (A)

[0072] Calculating the correlation coefficient of the reference data column includes the following steps:

[0073] Select the maximum value MaxΔ and the minimum value MinΔ among the absolute values ​​of the differences:

[0074]

[0075]

[0076] Calculate the correlation coefficient δ i (A):

[0077]

[0078] In the formula, ρ is the resolution coefficient, and its value range is [0, 1].

[0079] To obtain the corrected weights of the reference data column, the following steps are involved:

[0080] The correlation coefficient δ of each reference data column is calculated separately. i (A), δ i (B), δ i (C), ... summed, we get δ i :

[0081] δ i =δ i (A)+δ i (B)+δ i (C)+…

[0082] δ i After normalization, we obtain δ i ':

[0083]

[0084] To obtain the corrected weights for each comparison data column, the following steps are involved:

[0085] The reference data column is normalized to obtain the weight ω0 of the most authoritative expert;

[0086] Let the weight ω0 of the most authoritative expert be 0.5, then we obtain the corrected weights ω of each comparison data column. i :

[0087]

[0088] To obtain the revised expert scores, the following steps are required:

[0089] Calculate the discount rate α for each expert's weight. i for:

[0090]

[0091] Discount rate based on the weight of each expert α i The scores from the experts were revised to obtain new scores from the experts regarding recognition framework A, i.e., the revised scores for the reference data column:

[0092] X, i (A)=1-α i +α i X i (A)

[0093] Similarly, the new scores of each element within the same recognition framework are normalized to obtain the corrected scores for each comparison data column.

[0094] In step S4, the posterior probability of each final leaf node is obtained, including the following steps:

[0095] According to Dempster's rule of composition in the DS evidence theory, the expert scores of the root nodes are fused together. The formula for Dempster's rule of composition is as follows:

[0096]

[0097] In the formula, m(A) represents the probability of occurrence of the fused recognition frame A. i (A) is the score of A after the i-th expert has been weighted;

[0098] Substituting all the probability data of the fused recognition framework into the Dempster synthesis rule formula, we obtain the Bayesian network prior probability of each node.

[0099] The prior probabilities of each node in the Bayesian network are input into the Bayesian network model of the aerospace propulsion system, and the posterior probabilities of each final leaf node are obtained using the inference function.

[0100] This application also proposes a system theory process analysis and quantification system for aero-engine systems, including:

[0101] The Bayesian network model construction module for aero-engine systems is used to perform system theoretical process analysis on aero-engine systems, identify unsafe control behaviors, causal scenarios, and causal factors affecting the aero-engine system; it uses unsafe control behaviors as final leaf nodes, causal scenarios as intermediate leaf nodes, and causal factor information as root nodes, and obtains the node relationship flow through system theoretical process analysis to construct a Bayesian network model of the aero-engine system.

[0102] The expert scoring result correction module is used to score the probability of root node insecurity issues by experts, generating a reference data column and comparison data columns. Based on the reference data column and each comparison data column, the absolute value of the difference between the reference data column and each comparison data column is obtained. Based on the maximum and minimum values ​​of the absolute values, the correlation coefficient of the reference data column is calculated to obtain the corrected weight of the reference data column. Based on the corrected weight of the reference data column, the weights of each comparison data column are corrected to obtain the corrected weights of each comparison data column. Based on the corrected weights of the reference data column and each comparison data column, the discount rate of each expert weight is calculated to obtain the corrected expert scoring result.

[0103] The analysis and quantification result acquisition module is used to fuse the root nodes of the corrected expert scoring results, obtain the occurrence probability of the fused root nodes, and substitute them into the Bayesian network model of the aerospace power system to obtain the posterior probability of each final leaf node, thereby obtaining the analysis and quantification results of the system theoretical process.

[0104] Example

[0105] To verify the feasibility of the method proposed in this application, this application takes a hybrid hydrogen internal combustion engine-electric aircraft propulsion system as an example, such as... Figure 2 The diagram shows a safety analysis of a hybrid hydrogen internal combustion engine-electric aircraft propulsion system. The system consists of: a flight control system, a hydrogen internal combustion engine and its controller, an electric motor and its controller, a power battery pack and sensor signals, a transmission system, and display instruments.

[0106] The STPA quantitative analysis method for aero-engine systems provided in this application, based on a combination of grey relational theory, DS evidence theory, and Bayesian network model, includes the following steps performed in sequence:

[0107] Step 1: Export unsafe control behavior (UCA), causative scenarios, and causative factors from STPA.

[0108] The causative scenarios describe the triggering factors that may lead to unsafe control behaviors and dangers. Taking the UCA2-1 "Engine control system failed to issue drive commands to the hydrogen internal combustion engine controller as expected" in the hybrid hydrogen internal combustion engine-electric aircraft propulsion system as an example, the causative scenarios derived from its STPA analysis are shown in Table 1.

[0109] Table 1. Causative Scenarios Derived from STPA Analysis

[0110]

[0111] The details of the causative factors for the causative scenarios derived from the STPA analysis in Table 1 are shown in Table 2.

[0112] Table 2 shows the causative factors of the causative scenarios derived from the STPA analysis.

[0113]

[0114]

[0115] Step 2: Based on the UCA, causal scenarios, and causal factors derived from the STPA of the hybrid hydrogen internal combustion engine-electric aircraft propulsion system, and using the transformation rules in Step S1 above, establish a Bayesian network model of the hybrid hydrogen internal combustion engine-electric aircraft propulsion system, such as... Figure 3 As shown.

[0116] Step 3: Invite five domain experts to score the probability of insecurity issues occurring in the root node. Detailed scores are shown in the table below (unit: 1*10). -6 ).

[0117] Table 3 Expert Scoring Results

[0118] F1 0.65 0.07 0.59 0.16 0.78 F2 0.76 0.21 0.96 0.48 0.45 F3 0.96 0.10 0.62 0.58 0.19 F4 0.54 0.56 0.05 0.88 0.51 F5 0.45 0.67 0.45 0.56 0.28 F6 0.25 0.78 0.41 0.33 0.08 F7 0.69 0.48 0.44 0.50 0.31 F8 0.63 0.28 0.57 0.69 0.44 F9 0.48 0.64 0.63 0.26 0.42 F10 0.12 0.71 0.26 0.68 0.53

[0119] Step 4: Based on the grey relational theory mentioned above, i.e., step S3, the scoring results of each expert in Table 3 are corrected.

[0120] Assuming that expert A has the most engineering experience among the 5 experts, we select expert A's score as the reference data column and the scores of the other 4 experts as the comparison data column.

[0121] Subtract each comparison data column from the reference data column to obtain the absolute value of the difference between the reference data column and each comparison data column. The maximum and minimum values ​​of the absolute difference are 0 and 0.86, respectively.

[0122] The correlation coefficients were calculated and normalized to obtain the weights of each expert, as shown in Table 4 below.

[0123] Table 4 shows the revised reference data column and the weights of each comparison data column.

[0124] A 0.5 B 0.11 C 0.14 D 0.12 E 0.13

[0125] After correction using the weighting formula, the corrected scores from each expert are shown in Table 5 below.

[0126] Table 5 shows the revised scores from each expert.

[0127] A 0.65 0.76 0.96 0.54 0.45 0.25 0.69 0.63 0.48 0.12 B 0.47 0.48 0.48 0.50 0.51 0.52 0.50 0.49 0.51 0.51 C 0.51 0.53 0.51 0.47 0.50 0.49 0.50 0.51 0.51 0.48 D 0.48 0.50 0.51 0.52 0.50 0.49 0.50 0.51 0.48 0.51 E 0.52 0.50 0.48 0.50 0.48 0.47 0.49 0.50 0.49 0.50

[0128] Step 5: According to Dempster's rule of synthesis in DS evidence theory, the expert scores of the five root nodes are fused to obtain the probability of occurrence of the fused root nodes as shown in Table 6 below, which serves as the prior probability table of the Bayesian network.

[0129] Table 6. Probability of Root Node Occurrence After Merging

[0130] Prior probability 0.63 0.77 0.94 0.53 0.44 0.23 0.67 0.63 0.48 0.12

[0131] Step 6: Input the prior probabilities of each node into the Bayesian network, and use the Bayesian network inference function to obtain the posterior probabilities of each final leaf node, such as... Figure 4 As shown, this is to enable timely preventative measures to be taken against unsafe control behaviors in actual use.

[0132] The method proposed in this application can directly obtain quantitative evaluations from experts, and the experts' probability scores are highly accurate and do not have a wide range of probability values. Furthermore, by eliminating the step in existing technologies where the correspondence between qualitative options and fuzzy numbers needs to be subjectively determined by researchers, the method proposed in this invention reduces the subjective influence of researchers on data processing.

[0133] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

[0134] Furthermore, unless otherwise stated, all technical and scientific terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. All references to this specification are incorporated by way of citation to disclose and describe methods relating to those references. In the event of any conflict with any incorporated reference, the content of this specification shall prevail.

Claims

1. A quantitative method for system theoretical process analysis of an aero-engine system, characterized in that, Includes the following steps: Conduct system theoretical process analysis on the aero-engine system to determine the unsafe control behaviors, causal scenarios, and causal factors affecting the aero-engine system; By taking unsafe control behaviors as final leaf nodes, causative scenarios as intermediate leaf nodes, and causative factor information as root nodes, the flow of node relationships is obtained through system theoretical process analysis, and a Bayesian network model of the aerospace power system is constructed. Experts scored the probability of root node insecurity issues, generating reference data columns and comparison data columns respectively. Based on the reference data column and each comparison data column, obtain the absolute value of the difference between the reference data column and each comparison data column; based on the maximum and minimum values ​​of the absolute values, calculate the correlation coefficient of the reference data column to obtain the weight of the corrected reference data column; Based on the corrected weights of the reference data columns, the weights of each comparison data column are adjusted to obtain the corrected weights of each comparison data column. Based on the revised reference data column and the weights of each comparison data column, the revised expert scoring results are obtained by calculating the discount rate of each expert weight. The root nodes of the corrected expert scoring results are fused to obtain the probability of occurrence of the fused root nodes. This probability is then substituted into the Bayesian network model of the aerospace power system to obtain the posterior probability of each final leaf node, thereby obtaining the quantitative results of the theoretical process analysis of the system. Obtaining the posterior probability of each final leaf node includes the following steps: According to Dempster's rule of composition in the DS evidence theory, the expert scores of the root nodes are fused together. The formula for Dempster's rule of composition is as follows: In the formula, m (A) represents the probability of occurrence of the fused recognition framework A. m i (A) is the first i The scores for A by the experts after weighting; Substituting all the probability data of the fused recognition framework into the Dempster synthesis rule formula, we obtain the Bayesian network prior probability of each node. The prior probabilities of each node in the Bayesian network are input into the Bayesian network model of the aerospace propulsion system, and the posterior probabilities of each final leaf node are obtained using the inference function.

2. The method for system theoretical process analysis and quantification of aero-engine systems according to claim 1, characterized in that, The construction of the Bayesian network model of the aerospace propulsion system includes the following steps: Unsafe control behaviors are transformed into final leaf nodes in the Bayesian network model of aerospace power systems, causal scenarios are transformed into intermediate leaf nodes in the Bayesian network model of aerospace power systems, and causal factors are transformed into root nodes in the Bayesian network model of aerospace power systems. Based on the process analysis and reasoning of system theory, the flow of node relationships is obtained, and the nodes of the Bayesian network model of the aerospace power system are connected to construct the Bayesian network structure of the aerospace power system.

3. The method for system theoretical process analysis and quantification of aero-engine systems according to claim 2, characterized in that, Obtaining the absolute value of the difference between the reference data column and each comparison data includes the following steps: Using the scores from the experts as data columns, the recognition framework is as follows: X =( A , B , C The number of experts is... n ; The scores from the most authoritative experts were selected as the reference data series. X 0, with the scores from the remaining experts used as comparative data. X i ,in i =1,2,…, n -1; use X Subtract 0 respectively X i Obtain the absolute value of the difference between the reference data column and each comparison data column. Δ i ( A ).

4. The method for system theoretical process analysis and quantification of aero-engine systems according to claim 3, characterized in that, The calculation of the correlation coefficient of the reference data column includes the following steps: Select the maximum value among the absolute values ​​of the differences. MaxΔ and minimum value MinΔ : Calculate the correlation coefficient δ i ( A ): In the formula, ρ The resolution coefficient takes values ​​in the range [0, 1].

5. The method for system theoretical process analysis and quantification of aero-engine systems according to claim 4, characterized in that, The weights of the corrected reference data columns include the following steps: Correlation coefficients for each reference data column δ i ( A ), δ i ( B ), δ i ( C ), ...accumulate, to get δ i : Will δ i After normalization, we get δ i ': 。 6. The method for quantitative analysis of the theoretical process of an aero-engine system according to claim 5, characterized in that, The process of obtaining the corrected weights of each comparison data column includes the following steps: The reference data columns are normalized to obtain the weights of the most authoritative experts. ω 0; Assign weights to the most authoritative experts. ω 0 is set to 0.5, thus correcting the weights of each comparison data column. ω i : 。 7. The method for system theoretical process analysis and quantification of aero-engine systems according to claim 6, characterized in that, The process of obtaining the revised expert scoring results includes the following steps: Calculate the discount rate for each expert's weight. a i for: Discount rates based on the weight of each expert a i The scores from the experts were revised to obtain their opinions on the recognition framework. A The new scoring, i.e., the revised scoring of the reference data column, is as follows: Similarly, the new scores of each element within the same recognition framework are normalized to obtain the corrected scores for each comparison data column.

8. A system for quantitative analysis of the theoretical process of an aero-engine system, characterized in that, include: The Bayesian network model construction module for aero-engine systems is used to perform system theoretical process analysis on aero-engine systems and determine information on unsafe control behaviors, causal scenarios, and causal factors affecting aero-engine systems. By taking unsafe control behaviors as final leaf nodes, causative scenarios as intermediate leaf nodes, and causative factor information as root nodes, the flow of node relationships is obtained through system theoretical process analysis, and a Bayesian network model of the aerospace power system is constructed. The expert scoring result correction module is used to score the probability of root node insecurity issues by experts, generating a reference data column and each comparison data column; based on the reference data column and each comparison data column, the absolute value of the difference between the reference data column and each comparison data column is obtained; based on the maximum and minimum values ​​of the absolute values, the correlation coefficient of the reference data column is calculated to obtain the weight of the corrected reference data column. Based on the corrected weights of the reference data columns, the weights of each comparison data column are adjusted to obtain the corrected weights of each comparison data column. Based on the revised reference data column and the weights of each comparison data column, the revised expert scoring results are obtained by calculating the discount rate of each expert weight. The analysis and quantification result acquisition module is used to fuse the root nodes of the corrected expert scoring results, obtain the probability of occurrence of the fused root nodes, and substitute them into the Bayesian network model of the aerospace power system to obtain the posterior probability of each final leaf node, thereby obtaining the analysis and quantification results of the system theoretical process. Obtaining the posterior probability of each final leaf node includes the following steps: According to Dempster's rule of composition in the DS evidence theory, the expert scores of the root nodes are fused together. The formula for Dempster's rule of composition is as follows: In the formula, m ( A () is the fused recognition framework A The probability of occurrence, m i ( A ) is the first i After the experts allocated the weights, regarding A The rating; Substituting all the probability data of the fused recognition framework into the Dempster synthesis rule formula, we obtain the Bayesian network prior probability of each node. The prior probabilities of each node in the Bayesian network are input into the Bayesian network model of the aerospace propulsion system, and the posterior probabilities of each final leaf node are obtained using the inference function.

Images