Catenary part fault propagation risk assessment method based on evidence theory

Abstract

The invention discloses a contact network part fault propagation risk assessment method based on an evidence theory. The method comprises the following steps: acquiring historical fault associated data of a contact network part; defining a contact network part fault association relationship and distinguishing the contact network part fault association relationship into a direct fault association relationship and an indirect fault association relationship; constructing a fault associated data set under different mechanical connection relationships; constructing a fault incidence relation risk coefficient model; calculating a risk coefficient of a fault association relationship under different mechanical coupling relationships; constructing a part fault propagation risk assessment model based on an evidence theory; the fault propagation risk levels of all the parts under different mechanical connection relations are evaluated; according to the method, the risk degree of the fault incidence relation between the contact network parts can be evaluated, the fault propagation risk characteristics of the contact network parts under different mechanical coupling relations can be revealed, and a basis is provided for formulating a fault prevention strategy.

Description

Technical Field

[0001] The invention relates to the technical field of electrified railway traction power supply, and in particular to a method for evaluating the risk of fault propagation of contact network components based on evidence theory. Background Art

[0002] As the speed and frequency of high-speed railway trains continue to increase, higher and higher requirements are placed on the safety performance of the contact network. Surveys show that most of the faults in the traction power supply system come from the contact network. Once the contact network system collapses, the economic losses and social impacts will be catastrophic. However, the current maintenance strategy generally follows the idea of ​​planned maintenance and emergency maintenance, which will consume a lot of manpower and material resources. Therefore, the formulation of a fault prevention strategy for the contact network system is undoubtedly of great significance and prospect, which can not only improve the safety of the system, but also reduce the long-term maintenance cost. Since the contact network is a complex system composed of multiple components, the failure of one component may spread to other components. Therefore, revealing the fault propagation risk characteristics of components will become a key step in formulating fault prevention strategies.

[0003] At present, in the research on the fault propagation risk characteristics of contact network components, the reliability of key contact network components is mainly modeled by simulation modeling, and reasonable maintenance strategies are proposed based on the evaluation of system status. However, the simulation modeling method has great limitations. The first is the timeliness problem. Simulation modeling can often only reflect the information of impending failure in the system. However, when the fault information is detected, the equipment has usually suffered a certain degree of damage and cannot completely avoid its failure. Secondly, there is the limitation of the scope of simulation modeling. Due to the complexity of the contact network system, reliability modeling cannot effectively cover all components, which raises questions about the accuracy of the evaluation results. Therefore, in order to solve the above problems, it is of great theoretical significance and engineering value to propose a method for fault propagation risk assessment of contact network components based on evidence theory from a data-driven perspective. Summary of the invention

[0004] In order to overcome the defects of the prior art and reveal the fault propagation risk characteristics of components under different mechanical coupling relationships, the present invention constructs a component fault propagation risk trust function model based on evidence theory, including constructing an identification framework, a basic probability distribution function, and evidence fusion between different evidence sources, and then calculating the fault propagation risk trust function value of each component and the direct fault propagation and indirect fault propagation risk trust function values, revealing the fault propagation risk characteristics of each component, providing a basis for formulating fault prevention strategies, and improving the accuracy and reliability of fault propagation risk assessment.

[0005] The purpose of the present invention can be achieved by the following technical solutions:

[0006] The method for assessing the risk of fault propagation of overhead contact network components based on evidence theory includes the following steps:

[0007] S1: Obtain the historical fault correlation data set of contact network components;

[0008] S2: Based on the fault correlation data set, define the fault correlation relationship, including direct fault correlation relationship and indirect fault correlation relationship;

[0009] S3: Constructing direct fault association relationship dataset and indirect fault association relationship dataset;

[0010] S4: Construct a credibility model of fault correlation and an unreliability model of fault correlation;

[0011] S5: Calculating a risk coefficient of a fault association relationship based on the credibility model and the uncredibility model;

[0012] S6: Taking the risk coefficient of the fault association relationship as the evidence source, constructing an identification framework of a component fault propagation risk trust function model based on evidence theory and a basic probability distribution function of the identification framework, and fusing the basic probability distribution functions of different evidence sources;

[0013] S7: Based on the evidence fusion, calculate the fault propagation risk trust function value of each component and the direct fault propagation and indirect fault propagation risk trust function values, and evaluate the fault propagation risk level of each component.

[0014] Furthermore, based on the fault correlation data set, a specific method for defining the fault correlation relationship of the contact network components includes:

[0015] For any two components of the overhead line i and l j , if component l i The time when the failure occurs is t i , within a certain time window T, component l j At time t j A fault also occurs when t j ∈[t i ,t i +T], then they are defined to have a fault association relationship l i → j .

[0016] Furthermore, a specific method for defining the direct fault association relationship and the indirect fault association relationship includes:

[0017] For two components i and l j Fault association relationship i →j , if component l i and l j There is a direct mechanical connection, defining the fault association relationship l i → j is the direct fault association relationship, expressed as

[0018] For two components i and l j Fault association relationship i → j , if component l i and l j There is no direct mechanical connection, define the fault association relationship l i → j is the indirect fault association relationship, expressed as

[0019] Furthermore, the specific steps of constructing the credibility and uncredibility model of the fault association relationship include:

[0020] S41: For a fault association relationship l i → k , its residual credibility S(li→lk) is as shown in formula (S-1):

[0021]

[0022] In formula (1), P(l i ), P(l k ) and P(l i → k ) represent l i , l j and l i → j The frequency of occurrence in the fault correlation data set, P(l j / l i ) indicates that in component l i Faulty parts j Frequency of failures;

[0023] S42: Based on the residual credibility described in S41, as shown in formula (S-2), a credibility model of the fault association relationship is established:

[0024]

[0025] S43: Based on the residual credibility described in S41, as shown in formula (S-3), an unreliability model of the fault association relationship is established:

[0026]

[0027] Furthermore, for the fault association relationship l i → j , as shown in formula (S-4), calculate the fault association risk coefficient CCI (l i → j ):

[0028] CCI(l i → j )=C(l i → j )-I(l i → j )(S-4).

[0029] Furthermore, the specific steps of constructing the identification framework of the trust function model of component failure propagation risk based on evidence theory include:

[0030] S61: Identification frame Θ={H1,H2,...H M} is a problem consisting of all the assumptions H i The exhaustive set is composed of all hypotheses that are mutually exclusive, where M represents the number of elements contained in the recognition framework Θ. A subset of the recognition framework Θ is called a proposition. The power set of Θ is 2 Θ Consists of all subsets of Θ that contain 2 M elements, 2 of which Θ ={φ,{H1},{H2},...,{H M},{H1,H2},...,Θ};

[0031] S62: Put the parts l i Using the association relationship of the preceding faulty parts as a source of evidence, an identification framework Θ is constructed according to the method described in S61;

[0032] Among them, the power set of the recognition frame Θ2 Θ ={φ,A1,A2,Ω}, Proposition A1 is {support component l i Fault propagation}, Proposition A2 is {not supporting component l i of fault propagation}.

[0033] Furthermore, the specific method of constructing the basic probability distribution function of the recognition framework includes:

[0034] S71: For the recognition framework Θ, as shown in formula (S-5), construct the basic probability distribution function m: 2 Θ →[0,1]:

[0035]

[0036] In formula (S-5), if A i is the focal element, m(A i ) represents the evidence pair identification framework in proposition A i the level of support;

[0037] S72: According to the method described in S71, we use the risk coefficient CCI (l i → r ) to construct the basic probability distribution function m(A i ), it should be noted that the correlation risk coefficient RC(l i → r )∈[-1,1], and m(A i )∈[0,1], so we need to i → r ) is normalized, as shown in formula (S-6), to construct the basic probability distribution function m(A i ):

[0038]

[0039] Furthermore, the specific method of fusing the basic probability distribution functions of the different evidence sources includes:

[0040] S81: Based on the basic probability distribution functions of the different evidence sources, a trust coefficient α is first assigned to each evidence source before evidence fusion is performed, as shown in formulas (S-7) to (S-8), and a calculation method for discounting the basic probability distribution function of each evidence source is:

[0041]

[0042]

[0043] In formulas (S-7) to (S-8), and They represent the discounted basic probability distribution functions of the rth and sth evidence sources, respectively, and α r ∈[0,1] and α s ∈[0,1], indicating the degree of trust in the rth and sth sources of evidence. The larger the α, the higher the degree of trust in the source of evidence.

[0044] S82: According to the Discounting and Dempster evidence fusion rule, different sources of evidence are fused as shown in formula (S-9):

[0045]

[0046] In formula (S-9), m rsrepresents the basic probability distribution function after the r-th evidence source and the s-th evidence source are fused, represents the Discounting and Dempster evidence fusion rule, It is called the conflict coefficient, which is used to evaluate the degree of conflict between different sources of evidence. The larger k is, the greater the conflict is, and k≠1.

[0047] S83: According to Bayesian theory, m rs The value of (Ω) is proportional to m rs (A1) and m rs (A2), as shown in equations (S-10) to (S-11), define a new index M rs (A i )={m(A1),m(A2),m(Ω)}:

[0048]

[0049]

[0050] In formulas (S-10) to (S-11), M rs (A1) indicates that the r-th evidence source and the s-th evidence source are combined to support the proposition A1 {support component l i The degree of support for fault propagation}, M rs (A2) indicates that the r-th evidence source and the s-th evidence source are combined to support the proposition A2 {not supporting component l i The degree of support for fault propagation.

[0051] Furthermore, the specific method for calculating the fault propagation risk trust function value of each component includes:

[0052] According to the evidence fusion method, the component l i The fault association relationship dataset is used as the evidence source for evidence fusion to obtain M 12...N-1 (A i ), when considering different mechanical connections between components, the direct and indirect fault association datasets are used as evidence sources for evidence fusion to obtain M D 12...N-1 (A i ) and M I 12...N-1 (A i ), as shown in formula (S-12), calculate the fault propagation risk trust function value of components under different mechanical connection relationships:

[0053]

[0054] In formula (S-12), N represents the number of corresponding evidence sources. In addition, the larger the risk trust function value Bel is, the higher the risk of component propagation failure under the corresponding mechanical connection relationship is.

[0055] Compared with the prior art, the present invention has the following technical effects:

[0056] (1) The present invention can analyze the fault propagation risk characteristics under different mechanical coupling relationships;

[0057] (2) The present invention can process fault information from different evidence sources and reduce the impact of unreliable evidence by discounting the evidence, thereby improving the accuracy and reliability of fault propagation risk assessment. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 It is the overall flow chart of the present invention;

[0059] Figure 2 is the risk factor of the association relationship based on the premise of the failure of the load-bearing cable;

[0060] Figure 3 is the risk trust function value of the contact network components under different mechanical connection relationships; where, Figure 3 (a) Risk trust function value for direct fault propagation risk assessment; Figure 3 (b) Evaluate the risk trust function value for indirect fault propagation risk; Figure 3 (c) Risk trust function value for fault propagation risk assessment. DETAILED DESCRIPTION

[0061] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the invention, all other embodiments obtained by ordinary technicians in this field without creative work, any modifications, equivalent substitutions, improvements, etc., should be included in the protection scope of the present invention.

[0062] like Figure 1 As shown in the figure, the fault propagation risk assessment method for contact network components based on evidence theory includes the following steps:

[0063] S1: Obtain the historical fault correlation data set of contact network components;

[0064] S2: Based on the fault association data set, define the fault association relationship, including direct fault association relationship and indirect fault association relationship.

[0065] Specifically include: for any two parts of the contact network l i and lj , if component l i The time when the failure occurs is t i , within a certain time window T, component l j At time t j A fault also occurs when t j ∈[t i ,t i +T], then they are defined to have a fault association relationship l i → j ;

[0066] Based on the fault association relationship i → j , if component l i and l j If there is a direct mechanical connection, then a direct fault association relationship exists between them.

[0067] Based on the fault association relationship i → j , if component l i and l j If there is no direct mechanical connection, an indirect fault relationship is defined between them.

[0068] S3: define a direct fault association relationship data set and an indirect fault association relationship data set;

[0069] S4: Construct a fault correlation risk coefficient model based on credibility and unreliability:

[0070] S41: For a fault association relationship l i → k , its residual credibility S(li→lk) is as shown in formula (S-1):

[0071]

[0072] In formula (S-1), P(l i ), P(l k ) and P(l i → k ) represent l i , l j and l i → j The frequency of occurrence in the fault correlation data set, P(l j / l i ) indicates that in component l i Faulty parts j Frequency of failures;

[0073] S42: Based on the residual credibility described in S41, as shown in formula (S-2), a credibility model of the fault association relationship is established:

[0074]

[0075] S43: Based on the residual credibility described in S41, as shown in formula (S-3), an unreliability model of the fault association relationship is established:

[0076]

[0077] S5: Based on the credibility model C(l i → j ) and the unreliability model I(l i → j ), for the fault association relationship l i → j , as shown in formula (S-4), calculate the risk coefficient CCI (l i → j );

[0078] CCI(l i → j )=C(l i → j )-I(l i → j )(S-4)

[0079] S6: Using the risk coefficient of the fault association relationship as a source of evidence, an identification framework of a trust function model for component fault propagation risk based on evidence theory is constructed, specifically including:

[0080] S61: Identification frame Θ={H1,H2,...H M} is a problem consisting of all the assumptions H i The exhaustive set of all hypotheses is mutually exclusive, where M represents the number of elements contained in the recognition frame Θ. A subset of the recognition frame Θ is called a proposition, and the power set of Θ is 2 Θ Consists of all subsets of Θ that contain 2 M elements, 2 of which Θ ={φ,{H1},{H2},...,{H M},{H1,H2},...,Θ};

[0081] S62: Put the parts l i Using the association relationship of the preceding faulty parts as a source of evidence, an identification framework Θ is constructed according to the method described in S61;

[0082] Among them, the power set of the recognition frame Θ2Θ ={φ,A1,A2,Ω}, Proposition A1 is {support component l i Fault propagation}, Proposition A2 is {not supporting component l i of fault propagation}.

[0083] S7: Constructing the basic probability distribution function of the recognition framework, the specific method includes:

[0084] S71: For the recognition framework Θ, as shown in formula (S-5), construct the basic probability distribution function m: 2 Θ →[0,1]:

[0085]

[0086] In formula (S-5), if A i is the focal element, m(A i ) represents the evidence pair identification framework in proposition A i the level of support;

[0087] S72: According to the method described in S71, we use the risk coefficient CCI (l i → r ) to construct the basic probability distribution function m(A i ), it should be noted that the correlation risk coefficient RC(l i → r )∈[-1,1], and m(A i )∈[0,1], so we need to i → r ) is normalized, as shown in formula (S-6), to construct the basic probability distribution function m(A i ):

[0088]

[0089] S8: Evidence fusion is performed on the basic probability distribution functions of the different evidence sources. The specific method includes:

[0090] S81: Based on the basic probability distribution functions of the different evidence sources, a trust coefficient α is first assigned to each evidence source before evidence fusion is performed, as shown in formulas (S-7) to (S-8), and a calculation method for discounting the basic probability distribution function of each evidence source is:

[0091]

[0092]

[0093] In formulas (S-7) to (S-8), and They represent the discounted basic probability distribution functions of the rth and sth evidence sources, respectively, and α r ∈[0,1] and α s ∈[0,1], indicating the degree of trust in the rth and sth sources of evidence. The larger the α, the higher the degree of trust in the source of evidence.

[0094] S82: According to the Discounting and Dempster evidence fusion rule, different sources of evidence are fused as shown in formula (S-9):

[0095]

[0096] In formula (S-9), m rs represents the basic probability distribution function after the r-th evidence source and the s-th evidence source are fused, represents the Discounting and Dempster evidence fusion rule, It is called the conflict coefficient, which is used to evaluate the degree of conflict between different sources of evidence. The larger k is, the greater the conflict is, and k≠1.

[0097] S83: According to Bayesian theory, m rs The value of (Ω) is distributed proportionally to m rs (A1) and m rs (A2), as shown in equations (S-10) to (S-11), define a new index M rs (A i )={m(A1),m(A2),m(Ω)}:

[0098]

[0099]

[0100] In formulas (S-10) to (S-11), M rs (A1) indicates that the r-th evidence source and the s-th evidence source are combined to support the proposition A1 {support component l i The degree of support for fault propagation}, M rs (A2) indicates that the r-th evidence source and the s-th evidence source are combined to support the proposition A2 {not supporting component l i The degree of support for fault propagation.

[0101] S9: Based on the evidence fusion, the specific method for calculating the fault propagation risk trust function value of each component includes:

[0102] According to the evidence fusion method, the component li The fault association relationship dataset is used as the evidence source for evidence fusion to obtain M 12...N-1 (A i ). When considering different mechanical connections between components, the direct and indirect fault association datasets are used as evidence sources for evidence fusion to obtain M D 12...N-1 (A i ) and M I 12 ...N-1(A i ). As shown in formula (S-12), the fault propagation risk trust function value of components under different mechanical connection relationships is calculated:

[0103]

[0104] In formula (S-12), N represents the number of corresponding evidence sources. In addition, the larger the risk trust function value Bel is, the higher the risk of component propagation failure under the corresponding mechanical connection relationship is.

[0105] Using the method proposed in the present invention, taking Chengdu Railway Bureau as an example, the names and numbers of its contact network parts are shown in Table 1.

[0106] Table 1

[0107]

[0108] By obtaining the historical fault correlation data of contact network components of Chengdu Railway Bureau, a fault correlation relationship dataset under different mechanical connection relationships is established, a risk trust function model based on evidence theory is constructed, and the fault propagation risk trust function value of each component is calculated.

[0109] like Figure 2 As shown, taking component 2 - load-bearing cable as an example, formulas (S-1)-(S-3) are used to calculate the credibility and unreliability of the direct fault correlation relationship and the indirect fault correlation relationship based on the contact line fault, and then formula (S-4) is used to calculate the risk coefficient of the fault correlation relationship under the two mechanical connection relationships.

[0110] analyze Figure 2 It can be known that Figure 2 The solid line in the middle represents a direct fault correlation, and the dotted line represents an indirect fault correlation). When component 2-the load-bearing cable fails, the fault correlation with the highest risk is 2→9, which requires special attention. In addition, although the fault correlation 2→10 is an indirect fault correlation, its risk is also high. Therefore, it is necessary to pay attention to this high-risk fault correlation without direct mechanical connection.

[0111] like Figure 3As shown, the risk coefficient of each fault correlation relationship is calculated using formulas (S-1) to (S-4). This is used as a source of evidence, and the fault propagation risk trust function value of each component and the risk trust function value under different mechanical connection relationships are calculated using formulas (S-5) to (S-12).

[0112] analyze Figure 3 (a) Figure 3 (b) Figure 3 (c) It can be seen that the fault propagation risk trust function values ​​of components 2, 11, and 12 are the highest, and there is a high risk of propagating the fault to other components after a fault occurs. In addition, components 2 and 11 present high risks in both direct fault propagation and indirect fault propagation, and should be taken seriously. In addition, the direct fault propagation risk trust function value of component 4 is the highest, but its indirect fault propagation risk trust function value is the lowest. Therefore, after component 4 fails, more attention should be paid to the status of components that are directly mechanically connected to it to avoid the propagation of the fault.

[0113] Through the above analysis of the historical fault data of contact network components, it can be seen that the method proposed in the present invention can be used to analyze the fault propagation risk characteristics under different mechanical coupling relationships, and can process fault information from different evidence sources. By discounting the evidence, the impact of unreliable evidence can be reduced, thereby improving the accuracy and reliability of fault propagation risk assessment.

Claims

1. A method for assessing the risk of fault propagation of contact network components based on evidence theory, characterized in that: The evaluation method includes the following steps: S1: Obtain the historical fault correlation data set of contact network components; S2: Based on the fault correlation data set, define the fault correlation relationship, including direct fault correlation relationship and indirect fault correlation relationship; S3: Constructing direct fault association relationship dataset and indirect fault association relationship dataset; S4: Construct a credibility model of fault correlation and an unreliability model of fault correlation; S5: Calculating a risk coefficient of a fault association relationship based on the credibility model and the uncredibility model; S6: Taking the risk coefficient of the fault association relationship as the evidence source, constructing an identification framework of a component fault propagation risk trust function model based on evidence theory and a basic probability distribution function of the identification framework, and fusing the basic probability distribution functions of different evidence sources; S7: Based on the evidence fusion, calculate the fault propagation risk trust function value of each component and the direct fault propagation and indirect fault propagation risk trust function values, and evaluate the fault propagation risk level of each component.

2. The method for evaluating the risk of fault propagation of contact network components based on evidence theory according to claim 1 is characterized in that: Based on the fault association dataset, for the contact network components l i and l j , define the fault association relationship of contact network components l i → j .

3. The method for evaluating the risk of fault propagation of contact network components based on evidence theory according to claim 2 is characterized in that: The specific method of defining the direct fault association relationship and the indirect fault association relationship includes: The contact network components i and l j If there is a direct mechanical connection, define the fault association relationship l i → j is the direct fault association relationship, expressed as l i ——D→l j ; The contact network components i and l j If there is no direct mechanical connection, define the fault association l i → j is the indirect fault association relationship, expressed as 4. The method for evaluating the risk of fault propagation of contact network components based on evidence theory according to claim 3 is characterized in that: For fault association relationship i → j , calculate the residual credibility S(l i → k ), establish the reliability model S(l i → k ), and the unreliability model of fault correlation I(l i → k ), calculate the risk coefficient of the fault association relationship as shown in formula (1): CCI(l i →l j )=C(l i →l j )-I(l i →l j ) (1)。 5. The method for evaluating the risk of fault propagation of contact network components based on evidence theory according to claim 4 is characterized in that: The parts l i As the correlation relationship of the preceding faulty parts is used as the evidence source, the identification framework Θ is constructed, where the power set of the identification framework Θ is 2 Θ ={φ,A1,A2,Ω}, Proposition A1 is {support component l i Fault propagation}, Proposition A2 is {not supporting component l i of fault propagation}.

6. The method for evaluating the risk of fault propagation of contact network components based on evidence theory according to claim 5 is characterized in that: The specific method of constructing the basic probability distribution function of the recognition framework includes: S61: Based on the recognition framework Θ, as shown in formula (2), construct the basic probability distribution function m:2 Θ →[0,1]: In formula (2), if m(A i )>0, then A i is the focal element, m(A i ) is the evidence pair identification framework for proposition A i the level of support; S62: Risk coefficient CCI based on fault correlation relationship (l i → r ), as shown in formula (3), construct the basic probability distribution function m(A i ):

7. The method for evaluating the risk of fault propagation of contact network components based on evidence theory according to claim 6 is characterized in that: The specific methods of fusing the basic probability distribution functions of different evidence sources include: S71: According to the basic probability distribution function, a trust coefficient α is assigned to each evidence source, as shown in formulas (4) to (5), and the basic probability distribution function of each evidence source is discounted: In formulas (4) to (5), and are the discounted basic probability distribution functions of the rth and sth evidence sources, respectively, and α r and α s Respectively represent the degree of trust in the rth source of evidence and the sth source of evidence, α r ∈[0,1],α s ∈[0,1]; S72: According to the Discounting and Dempster evidence fusion rule, as shown in formula (6), different evidence sources are fused: In formula (6), m rs is the basic probability distribution function after the evidence fusion of the rth evidence source and the sth evidence source, is the Discounting and Dempster evidence fusion rule, It is called the conflict coefficient, which is used to evaluate the degree of conflict between different sources of evidence. The larger k is, the greater the conflict is, and k≠1; S73: According to Bayesian theory, m rs The value of (Ω) is proportional to m rs (A1) and m rs (A2), as shown in equations (7) to (8), define a new index M rs (A i )={m(A1),m(A2),m(Ω)}: In formulas (7) to (8), M rs (A1) is the evidence fusion of the r-th evidence source and the s-th evidence source for proposition A1{support component l i The degree of support for fault propagation}, M rs (A2) is the result of evidence fusion between the rth and sth evidence sources for proposition A2 {does not support component l i The degree of support for fault propagation.

8. The method for evaluating the risk of fault propagation of contact network components based on evidence theory according to claim 7 is characterized in that: According to the evidence fusion method, the component l i All evidence sources of fault propagation are fused to obtain M 12...N-1 (A i ), as shown in formula (9), calculate the fault propagation risk trust function value of the component: Bel(A1)=M 12...N-1 (A1) (9) In formula (9), N represents the number of components in the system.

9. The method for evaluating the risk of fault propagation of contact network components based on evidence theory according to claim 8 is characterized in that: The direct fault association relationship dataset and the indirect fault association relationship dataset are used as evidence sources respectively, and M is obtained after evidence fusion. D 12...N-1 (A i ) and M I 12...N-1 (A i ), as shown in formula (10), calculate the fault propagation risk trust function value of components under different mechanical connection relationships:

Images