A method for evaluating electromagnetic vulnerability of a vehicle engine system

Abstract

The application relates to a kind of vehicle engine system electromagnetic vulnerability evaluation methods, belong to electromagnetic environment effect evaluation field.Establish the engine system bayesian network fault model, get the triangular fuzzy number corresponding to the fuzzy expression of the failure correlation between each node in the bayesian network fault model through semantic contrast table, calculate fuzzy conditional probability, convert it into interval grey number conditional probability;Determine electromagnetic stress probability density function and component sensitivity threshold probability density function, calculate component layer interval grey number failure probability, and calculate engine system interval grey number failure probability in combination with interval grey number conditional probability.Affirmative effect is: determine the weight of each sensor through fuzzy analytic hierarchy process, solve the shortcoming that traditional analytic hierarchy process is more subjective, combine grey system theory and fuzzy theory with bayesian network, can more accurately express the uncertain information in the whole evaluation process and carry out failure probability calculation.

Description

Technical Field

[0001] This invention relates to the field of electromagnetic environment effect assessment, and in particular to a method for assessing the electromagnetic vulnerability of vehicle engine systems under strong electromagnetic pulse irradiation. Background Technology

[0002] With the rapid development of electronic information technology, electronic devices are being used more and more extensively. The engine system, as the core of a car, has seen an increase in the number of sensors and actuators to enhance its control capabilities. Electromagnetic pulse radiation fields can cause varying degrees of interference and even damage to the electronic equipment in the engine system, seriously affecting the safe operation of the engine. Therefore, assessing the electromagnetic vulnerability of vehicle engine systems is of great significance for improving the adaptability of vehicles to complex electromagnetic environments.

[0003] The paper "Electromagnetic Pulse Sensitivity Assessment of Vehicle Engine Systems Based on Hierarchical Bayesian Networks" by Sun Xiaoying, Hu Zezheng, et al. defines the voltage peak induced in the component's cables by a strong electromagnetic pulse when a component malfunctions as the sensitivity threshold of that component. Thirty sets of data were collected, and after Kolmogorov-Smirnov (KS) testing, the sensitivity threshold test data were deemed to follow a mean of μ. y The standard deviation is σ y The results follow a normal distribution. However, even under the same test conditions, the component sensitivity threshold test results still exhibit a certain degree of randomness. In practical engineering, it is difficult to obtain accurate and sufficient component failure data for calculating the failure probability. Summary of the Invention

[0004] This invention provides a method for assessing the electromagnetic vulnerability of a vehicle engine system. In the case of insufficient or inaccurate data, it introduces grey system theory and fuzzy theory, and combines them with Bayesian networks to assess the electromagnetic vulnerability of the engine system. This method can better express the uncertain information of the entire assessment process and calculate the failure probability.

[0005] The technical solution adopted by this invention includes the following steps:

[0006] Step 1: Based on the composition and working principle of the engine system, establish a Bayesian network fault model of the engine system and clarify the fault logic relationship between each subsystem and component in the Bayesian network fault model;

[0007] Step 2: Obtain the triangular fuzzy number corresponding to the fuzzy expression of the failure correlation between nodes in the Bayesian network fault model through the semantic lookup table, calculate the fuzzy conditional probability, and convert this fuzzy conditional probability into the interval gray number conditional probability.

[0008] Step 3: Determine the electromagnetic stress probability density function and the component sensitivity threshold probability density function, calculate the component layer interval gray number failure probability, and combine it with the interval gray number conditional probability in Step 2 to calculate the engine system interval gray number failure probability.

[0009] In step 1 of this invention, the method for constructing the Bayesian network fault model of the engine system is as follows:

[0010] Divide the engine system into n s There are three subsystems, each consisting of one sensor group and one actuator, wherein the sensor group of the i-th subsystem includes m... i There are n sensors, 1 ≤ i ≤ n s The Bayesian network fault model of the engine system is divided into four layers: system layer, subsystem layer, component layer, and electromagnetic stress layer. Node S(1,1) corresponds to the entire engine system, and node S(2,i) corresponds to the i-th subsystem of the engine system. The i-th subsystem S(2,i) includes sensor group S(3,i.1) and actuator S(3,i.2). The j-th sensor in the i-th sensor group S(3,i.1) is S(3,i.1.j), and the corresponding electromagnetic stress is S(4,i.1.j). The electromagnetic stress corresponding to the i-th actuator S(3,i.2) is S(4,i.2).

[0011] The semantic lookup table in step 2 of this invention is as follows:

[0012] The failure correlation between nodes is categorized into {low, relatively low, moderate, relatively high, high}, and the triangular fuzzy number corresponding to low node failure correlation is A. a = (a1, b1, c1), the triangular fuzzy number corresponding to the low correlation of node failure is A. b = (a2, b2, c2), the triangular fuzzy number generally corresponding to the node failure correlation is A. c = (a3, b3, c3), the triangular fuzzy number corresponding to the high correlation of node failure is A. d = (a4, b4, c4), the triangular fuzzy number corresponding to high correlation of node failure is A. e = (a5, b5, c5).

[0013] The method for determining the conditional probability of the gray number in step 2 of this invention is as follows:

[0014] For any two nodes with a failure logic relationship, the triangular fuzzy numbers corresponding to the fuzzy representation of their failure correlation obtained through a semantic lookup table are A1, A2, ..., A m Fuzzy conditional probability A cpt The calculation formula is:

[0015]

[0016] Among them, A a and A b For example, the addition of triangular fuzzy numbers is defined as follows: A a +A b = (a1+a2,b1+b2,c1+c2), where the fuzzy conditional probability A is... cpt =(a av ,b av ,c av Convert to interval gray number conditional probability

[0017]

[0018]

[0019] The method for determining the electromagnetic stress probability density function of a strong electromagnetic pulse in step 3 of this invention is as follows:

[0020] A simulation model was established in the CST microwave studio based on the vehicle's cable routing and mechanical structure to simulate the cable-coupled electromagnetic stress under strong electromagnetic pulse irradiation. The incident angle, azimuth angle, and polarization angle were sampled at equal intervals, with a step size of a0 for all three angles within the 0-90° range. The obtained electromagnetic stress simulation data was fitted to a mean value of μ using MATLAB. y The standard deviation is σ y The normal distribution function g(y) is the electromagnetic stress probability density function.

[0021] The method for determining the component sensitivity threshold probability density function in step 3 of this invention is as follows:

[0022] The operating status of sensors and actuators is monitored by a monitoring system. A strong electromagnetic pulse (ESP) irradiation test is conducted on the engine system, gradually increasing the ESP irradiation field intensity until the component malfunctions. The test data is recorded, with n sets of test data recorded for each component. The component sensitivity threshold test data follows a normal distribution with a mean μ and a standard deviation σ. Due to the randomness of the component sensitivity threshold test data, even under the same test conditions, the mean and standard deviation of the fitted normal distribution will not be exactly the same. Therefore, a 99% confidence interval estimate is performed on the mean μ, denoted as σ. The expression is:

[0023]

[0024] in, S is the sample mean, S is the sample standard deviation, and t is the sample mean. 0.005 (n-1) represents a t-distribution with n-1 degrees of freedom;

[0025] Since the mean value of the sensitivity threshold for calculating component failure probability plays a major role, assuming the standard deviation remains constant, let μ be... min for The minimum value, μ max for The maximum value of f(x) is the component sensitivity threshold probability density function. min (x) is a function with mean μ min The component sensitivity threshold probability density function with standard deviation S, f max (x) is a function with mean μ max The component sensitivity threshold probability density function with standard deviation S.

[0026] The method for calculating the component layer interval gray number failure probability in step 3 of this invention is as follows:

[0027] (1) Calculation of sensor range gray number failure probability

[0028] The interval gray number failure probability of the sensor is calculated based on the electromagnetic stress probability density function g(y) and the component sensitivity threshold probability density function f(x). This is achieved through g(y) and f(x). min (x) Calculate the upper bound of the sensor interval gray number failure probability, denoted as P. max (S(3,i.1.j)), through g(y) and f max (x) Calculate the lower bound of the sensor interval gray number failure probability, denoted as P. min (S(3,i.1.j)), the failure probability of the gray number in the sensor interval is denoted as...

[0029] (2) Calculation of the failure probability of gray number in the sensor group interval

[0030] Since each sensor in the sensor group is independent and does not affect the others, the interval gray number failure probability of the sensor group can be calculated by weighted summing of the interval gray number failure probabilities of each sensor. The weight of each sensor is determined using fuzzy hierarchical analysis, as follows:

[0031] 1) Construct the judgment matrix

[0032] The 0.1–0.9 scaling method is used for the i-th sensor group S(3,i.1)(1≤i≤n). s Construct the judgment matrix R = (r uv )m i ×m i r uv (1≤u,v≤m i ) represents the relative importance of the u-th sensor compared to the v-th sensor in the i-th sensor group S(3,i.1);

[0033] 2) Consistency Transformation

[0034] Perform a consistency transformation on the judgment matrix to obtain the fuzzy consistent judgment matrix R. c =(f uv )m i ×m i R c The value f in row u and column v uv The calculation formula is:

[0035]

[0036] in To determine the summation of the u-th row of the matrix, r v Similarly, m i The number of sensors in the i-th sensor group is also the order of the judgment matrix and the fuzzy consistency judgment matrix;

[0037] 3) Calculate sensor weights

[0038] The formula for calculating the weight of the u-th sensor in the i-th sensor group is:

[0039]

[0040] Where α is inversely proportional to the difference in weights, when When the weight difference is greatest, take...

[0041] 4) Calculation of the failure probability of the gray number interval

[0042] The probability of gray number failure in the i-th sensor group interval The calculation formula is:

[0043]

[0044] in For the j-th sensor in the i-th sensor group a The weight of each sensor;

[0045] (3) Calculation of the failure probability of the actuator range gray number

[0046] The failure of actuator S(3,i.2) is caused by two reasons: ① the electromagnetic stress experienced by the actuator causes failure; ② the failure of the actuator is caused by sensor failure. When either one occurs, the actuator will be in a failed state.

[0047] The method for calculating the interval gray number failure probability of actuator failure due to electromagnetic stress is the same as the method for calculating the interval gray number failure probability of sensors, denoted as...

[0048] Actuator range gray number failure probability The calculation formula is:

[0049]

[0050] The method for calculating the engine system interval gray number failure probability in step 3 of this invention is as follows:

[0051] (1) Calculation of the failure probability of the gray number in the subsystem interval:

[0052]

[0053] Where S(2,q1),…,S(2,q) r The subsystem is the subsystem whose failure will cause the S(2,i) subsystem to fail;

[0054] (2) Calculation of the failure probability of the engine system in the range of gray numbers:

[0055]

[0056] The interval gray number failure probability of an engine system is a set composed of the interval gray number failure probabilities of subsystems and the interval gray number conditional probabilities. Determining the range of values ​​can be transformed into an extreme value problem within the range of the interval gray number failure probability and the interval gray number conditional probability of the subsystem. The method for determining the range of values ​​is similar.

[0057] The beneficial effects of this invention are as follows: Addressing the randomness of component sensitivity threshold test results and the ambiguity in understanding the correlation between node failures in the Bayesian network fault model, grey system theory and fuzzy theory are introduced. Interval estimation is used to determine the interval grey number of the mean component sensitivity threshold, representing the uncertainty of the mean threshold. Fuzzy theory is used to describe the uncertainty in understanding the correlation between node failures in the Bayesian network fault model. The weight of each sensor is determined using fuzzy hierarchical analysis, which to some extent overcomes the inherent subjectivity of traditional hierarchical analysis. Combining grey system theory and fuzzy theory with Bayesian networks allows for a more accurate expression of uncertain information throughout the evaluation process and the calculation of failure probabilities. Attached Figure Description

[0058] Figure 1 This is a flowchart of the present invention;

[0059] Figure 2 This is a Bayesian network fault model diagram of the engine system. Detailed Implementation

[0060] Includes the following steps:

[0061] Step 1: Based on the composition and working principle of the engine system, establish a Bayesian network fault model of the engine system and clarify the fault logic relationship between each subsystem and component in the Bayesian network fault model;

[0062] Step 2: Obtain the triangular fuzzy number corresponding to the fuzzy expression of the failure correlation between nodes in the Bayesian network fault model through the semantic lookup table, calculate the fuzzy conditional probability, and convert this fuzzy conditional probability into the interval gray number conditional probability.

[0063] Step 3: Determine the electromagnetic stress probability density function and the component sensitivity threshold probability density function, calculate the component layer interval gray number failure probability, and combine it with the interval gray number conditional probability in Step 2 to calculate the engine system interval gray number failure probability.

[0064] In step 1 of this invention, the method for constructing the Bayesian network fault model of the engine system is as follows:

[0065] Divide the engine system into n s There are three subsystems, each consisting of one sensor group and one actuator, wherein the sensor group of the i-th subsystem includes m... i There are n sensors, 1 ≤ i ≤ n s The Bayesian network fault model of the engine system is divided into four layers: system layer, subsystem layer, component layer, and electromagnetic stress layer. Node S(1,1) corresponds to the entire engine system, and node S(2,i) corresponds to the i-th subsystem of the engine system. The i-th subsystem S(2,i) includes sensor group S(3,i.1) and actuator S(3,i.2). The j-th sensor in the i-th sensor group S(3,i.1) is S(3,i.1.j), and the corresponding electromagnetic stress is S(4,i.1.j). The electromagnetic stress corresponding to the i-th actuator S(3,i.2) is S(4,i.2).

[0066] The semantic lookup table in step 2 of this invention is as follows:

[0067] The failure correlation between nodes is categorized into {low, relatively low, moderate, relatively high, high}, and the triangular fuzzy number corresponding to low node failure correlation is A. a = (a1, b1, c1), the triangular fuzzy number corresponding to the low correlation of node failure is A. b = (a2, b2, c2), the triangular fuzzy number generally corresponding to the node failure correlation is A. c = (a3, b3, c3), the triangular fuzzy number corresponding to the high correlation of node failure is A. d = (a4, b4, c4), the triangular fuzzy number corresponding to high correlation of node failure is A. e = (a5, b5, c5).

[0068] The method for determining the conditional probability of the gray number in step 2 of this invention is as follows:

[0069] For any two nodes with a failure logic relationship, the triangular fuzzy numbers corresponding to the fuzzy representation of their failure correlation obtained through a semantic lookup table are A1, A2, ..., A m Fuzzy conditional probability A cpt The calculation formula is:

[0070]

[0071] Among them, A a and A b For example, the addition of triangular fuzzy numbers is defined as follows: A a +A b = (a1+a2,b1+b2,c1+c2), where the fuzzy conditional probability A is... cpt =(a av ,b av ,c av Convert to interval gray number conditional probability

[0072]

[0073]

[0074] The method for determining the electromagnetic stress probability density function of a strong electromagnetic pulse in step 3 of this invention is as follows:

[0075] A simulation model was built in the CST microwave studio based on the vehicle's mechanical structure (cable routing and body). The simulation of electromagnetic stress coupled to the cable under strong electromagnetic pulse irradiation was performed. The incident angle, azimuth angle, and polarization angle were sampled at equal intervals, with a step size of a0 for all three angles within the range of 0-90°. The obtained electromagnetic stress simulation data was fitted using MATLAB to obtain a mean value of μ. y The standard deviation is σ y The normal distribution function g(y) is the electromagnetic stress probability density function.

[0076] The method for determining the component sensitivity threshold probability density function in step 3 of this invention is as follows:

[0077] The operating status of sensors and actuators is monitored through a monitoring system. A strong electromagnetic pulse (ESP) irradiation test is conducted on the engine system, gradually increasing the ESP irradiation field intensity until the component malfunctions. The test data is recorded, with n sets of test data recorded for each component. The component sensitivity threshold test data follows a normal distribution with a mean μ and a standard deviation σ. Due to the randomness of the component sensitivity threshold test data, even under the same test conditions, the mean and standard deviation of the fitted normal distribution will not be exactly the same. Therefore, a 99% confidence interval estimate is performed on the mean μ, denoted as σ. The expression is:

[0078]

[0079] in, S is the sample mean, S is the sample standard deviation, and t is the sample mean. 0.005 (n-1) represents a t-distribution with n-1 degrees of freedom;

[0080] Since the mean value of the sensitivity threshold for calculating component failure probability plays a major role, assuming the standard deviation remains constant, let μ be... min for The minimum value, μ max for The maximum value of f(x) is the component sensitivity threshold probability density function. min (x) is a function with mean μ min The component sensitivity threshold probability density function with standard deviation S, f max (x) is a function with mean μ max The component sensitivity threshold probability density function with standard deviation S.

[0081] The method for calculating the component layer interval gray number failure probability in step 3 of this invention is as follows:

[0082] (1) Calculation of sensor range gray number failure probability

[0083] The interval gray number failure probability of the sensor is calculated based on the electromagnetic stress probability density function g(y) and the component sensitivity threshold probability density function f(x). This is achieved through g(y) and f(x). min (x) Calculate the upper bound of the sensor interval gray number failure probability, denoted as P. max (S(3,i.1.j)), through g(y) and f max (x) Calculate the lower bound of the sensor interval gray number failure probability, denoted as P. min (S(3,i.1.j)), the failure probability of the gray number in the sensor interval is denoted as...

[0084] (2) Calculation of the failure probability of gray number in the sensor group interval

[0085] Since each sensor in the sensor group is independent and does not affect the others, the interval gray number failure probability of the sensor group can be calculated by weighted summing of the interval gray number failure probabilities of each sensor. The weight of each sensor is determined using fuzzy hierarchical analysis, as follows:

[0086] 1) Construct the judgment matrix

[0087] The 0.1–0.9 scaling method is used for the i-th sensor group S(3,i.1)(1≤i≤n). s Construct the judgment matrix R = (r uv )m i ×m i r uv (1≤u,v≤m i ) represents the relative importance of the u-th sensor compared to the v-th sensor in the i-th sensor group S(3,i.1);

[0088] 2) Consistency Transformation

[0089] Perform a consistency transformation on the judgment matrix to obtain the fuzzy consistent judgment matrix R. c =(f uv )m i ×m i R c The value f in row u and column v uv The calculation formula is:

[0090]

[0091] in To determine the summation of the u-th row of the matrix, r v Similarly, m i The number of sensors in the i-th sensor group is also the order of the judgment matrix and the fuzzy consistency judgment matrix;

[0092] 3) Calculate sensor weights

[0093] The formula for calculating the weight of the u-th sensor in the i-th sensor group is:

[0094]

[0095] Where α is inversely proportional to the difference in weights, when When the weight difference is greatest, take...

[0096] 4) Calculation of the failure probability of the gray number interval

[0097] The probability of gray number failure in the i-th sensor group interval The calculation formula is:

[0098]

[0099] in For the j-th sensor in the i-th sensor group a The weight of each sensor;

[0100] (3) Calculation of the failure probability of the actuator range gray number

[0101] The failure of actuator S(3,i.2) is caused by two reasons: ① the electromagnetic stress experienced by the actuator causes failure; ② the failure of the actuator is caused by sensor failure. When either one occurs, the actuator will be in a failed state.

[0102] The method for calculating the interval gray number failure probability of actuator failure due to electromagnetic stress is the same as the method for calculating the interval gray number failure probability of sensors, denoted as...

[0103] Actuator range gray number failure probability The calculation formula is:

[0104]

[0105] The method for calculating the engine system interval gray number failure probability in step 3 of this invention is as follows:

[0106] (1) Calculation of the failure probability of the gray number in the subsystem interval:

[0107]

[0108] Where S(2,q1),…,S(2,q) r The subsystem is the subsystem whose failure will cause the S(2,i) subsystem to fail;

[0109] (2) Calculation of the failure probability of the engine system in the range of gray numbers:

[0110]

[0111] The interval gray number failure probability of an engine system is a set composed of the interval gray number failure probabilities of subsystems and the interval gray number conditional probabilities. Determining the range of values ​​can be transformed into an extreme value problem within the range of the interval gray number failure probability and the interval gray number conditional probability of the subsystem. The method for determining the range of values ​​is similar.

[0112] The present invention will be further illustrated below with specific experimental examples, including the following steps:

[0113] Step 1: Based on the composition and working principle of the engine system, establish a Bayesian network fault model of the engine system and clarify the fault logic relationships between subsystems and components in the Bayesian network fault model.

[0114] like Figure 2 As shown, the engine system S(1,1) is divided into a fuel supply subsystem S(2,1) and a fuel injection subsystem S(2,2). The fuel supply subsystem includes a sensor group AS(3,1.1) and a fuel metering valve S(3,1.2). Sensor group A includes a crankshaft position sensor S(3,1.1.1), a camshaft position sensor S(3,1.1.2), and a fuel rail pressure sensor S(3,1.1.3). The fuel injection subsystem includes a sensor group BS(3,2.1) and an injector S(3,2.2). Sensor group B includes a crankshaft position sensor S(3,2.1.1), a camshaft position sensor S(3,2.1.2), and an accelerator pedal position sensor S(3,2.1.2). Sensors S(3,2.1.3), coolant temperature sensor S(3,2.1.4), intake air temperature and pressure sensor S(3,2.1.5), and atmospheric pressure sensor S(3,2.1.6) are located in sensor group A. Node S(4,1.1.j1) (1≤j1≤3) represents the electromagnetic stress corresponding to the sensor in sensor group A, and node S(4,2.1.j2) (1≤j2≤6) represents the electromagnetic stress corresponding to the sensor in sensor group B. S(4,1.2) and S(4,2.2) represent the electromagnetic stress corresponding to the fuel metering valve and injector. The electronic control unit's metal casing provides good shielding against electromagnetic pulses, and it has corresponding internal filtering circuits; therefore, it is considered an insensitive device. Thus, the electronic control unit is not considered in the Bayesian network fault model.

[0115] Step 2: Obtain the triangular fuzzy number corresponding to the fuzzy expression of the failure correlation between nodes in the Bayesian network failure model through the semantic lookup table, calculate the fuzzy conditional probability, and convert this fuzzy conditional probability into the interval gray number conditional probability.

[0116] The semantic lookup table is as follows:

[0117] The failure correlation between nodes is categorized into {low, relatively low, moderate, relatively high, high}, and the triangular fuzzy number corresponding to low node failure correlation is A. a = (0,0,0.25), the triangular fuzzy number corresponding to the low correlation of node failure is A. b = (0, 0.25, 0.5), the triangular fuzzy number generally corresponding to the node failure correlation is A. c = (0.25, 0.5, 0.75), the triangular fuzzy number corresponding to the high correlation of node failure is A. d = (0.5, 0.75, 1), the triangular fuzzy number corresponding to high correlation of node failure is A. e = (0.75, 1, 1);

[0118] The method for determining the conditional probability of the gray number in the interval is as follows:

[0119] For any two nodes with a failure logic relationship, the triangular fuzzy numbers corresponding to the fuzzy representation of their failure correlation obtained through a semantic lookup table are A1, A2, ..., A m Fuzzy conditional probability A cpt The calculation formula is:

[0120]

[0121] Among them, A a and A b To illustrate the definition of triangular fuzzy number addition, let's take A as an example. a +A b = (a1+a2,b1+b2,c1+c2)

[0122] The fuzzy conditional probability A cpt =(a av ,b av ,c av Convert to interval gray number conditional probability

[0123]

[0124]

[0125] Step 3: Determine the electromagnetic stress probability density function and the component sensitivity threshold probability density function, calculate the component layer interval gray number failure probability, and combine it with the interval gray number conditional probability in Step 2 to calculate the engine system interval gray number failure probability.

[0126] The method for determining the electromagnetic stress probability density function of the strong electromagnetic pulse is as follows:

[0127] A simulation model was built in the CST microwave studio based on the vehicle's mechanical structure (cable routing and body) to simulate the electromagnetic stress coupled to the cables under strong electromagnetic pulse irradiation. The incident angle, azimuth angle, and polarization angle were sampled at equal intervals, with each angle sampled at 18° increments within the range of 0-90°. The obtained electromagnetic stress simulation data was fitted using MATLAB to obtain a mean value of μ. y The standard deviation is σ y The normal distribution function g(y) is the electromagnetic stress probability density function.

[0128] The method for determining the component sensitivity threshold probability density function is as follows:

[0129] The operating status of sensors and actuators is monitored through a monitoring system. A strong electromagnetic pulse (EMP) irradiation test is conducted on the engine system, gradually increasing the EMP irradiation field intensity until the component malfunctions, and the test data is recorded. Thirty sets of test data are recorded for each component. The component sensitivity threshold test data follow a normal distribution with a mean μ and a standard deviation σ. Due to the randomness of the component sensitivity threshold test data, even under the same test conditions, the mean and standard deviation of the fitted normal distribution will not be exactly the same. Therefore, a 99% confidence interval estimate is performed on the mean μ, denoted as σ. The expression is

[0130]

[0131] in S is the sample mean, S is the sample standard deviation, and t is the sample mean. 0.005 (29) is a t-distribution with 29 degrees of freedom;

[0132] Since the mean of the sensitivity threshold for calculating component failure probability plays a major role, we assume the standard deviation remains constant. Let μ be... min for The minimum value, μ max for The maximum value of f(x) is given by f(x), which is the component sensitivity threshold probability density function. min (x) is a function with mean μ min The component sensitivity threshold probability density function with standard deviation S, f max (x) is a function with mean μ max The component sensitivity threshold probability density function with standard deviation S.

[0133] (I) Calculation of failure probability of gray number in component layer interval

[0134] 1. Calculation of sensor range gray number failure probability

[0135] The formula for calculating the failure probability P(S(3,i.1.j)) of the sensor is as follows:

[0136]

[0137]

[0138] P(S(3,i.1.j))=P(AEME)P(EMS|AEME)P(Comp|EMS)

[0139] Here, AEME (Ambient electromagnetic environment) refers to the electromagnetic environment of a strong electromagnetic pulse in the engine compartment, EMS (Electromagnetic stress) refers to electromagnetic stress, and Comp refers to the component.max =μ y -3σ y Let x be the upper bound of the electromagnetic stress probability density function g(y). min =μ x -3σ x P(AEME) is the lower bound of the probability density function f(x) for the sensor sensitivity threshold, and P(AEME) = 1.

[0140] Through g(y) and f min (x) calculates the upper bound of the component interval gray number failure probability, and uses g(y) and f max (x) Calculate the lower bound of the gray number failure probability of the component interval. The gray number failure probability of each sensor interval in sensor group A is denoted as... The failure probability of gray number in each sensor interval of sensor group B is denoted as:

[0141] 2. Calculation of the failure probability of gray number in the sensor group interval

[0142] Since each sensor in the sensor group is independent and does not affect the others, the interval gray number failure probability of the sensor group can be calculated by weighted summation of the interval gray number failure probabilities of each sensor. The weight of each sensor is determined using fuzzy hierarchical analysis. The steps are as follows:

[0143] (1) Construct the judgment matrix

[0144] The 0.1–0.9 scaling method is used for the i-th sensor group S(3,i.1)(1≤i≤n). s Construct the judgment matrix R = (r uv )m i ×m i r uv (1≤u,v≤m i ) represents the relative importance of the u-th sensor compared to the v-th sensor in the i-th sensor group S(3,i.1).

[0145] r uv Let r represent the relative importance of the u-th sensor S(3, i.1.u) compared to the v-th sensor S(3, i.1.v) in sensor groups A and B. uv =0.5 indicates that S(3,i.1.u) and S(3,i.1.v) are equally important, r uv =0.6 indicates that S(3,i.1.u) is slightly more important than S(3,i.1.v), r uv =0.7 indicates that S(3,i.1.u) is significantly more important than S(3,i.1.v), r uv =0.8 indicates that S(3,i.1.u) is much more important than S(3,i.1.v), r uv=0.9 indicates that S(3,i.1.u) is extremely important than S(3,i.1.v);

[0146] Judgment matrix for sensor group A:

[0147]

[0148] Sensor group B judgment matrix:

[0149]

[0150] (2) Consistency Transformation

[0151] Perform a consistency transformation on the judgment matrix to obtain the fuzzy consistent judgment matrix R. c =(f uv )m i ×m i R c The value f in row u and column v uv The calculation formula is:

[0152]

[0153] in To determine the summation of the u-th row of the matrix, r u Similarly, m i Let be the number of sensors in the i-th sensor group, and also the order of the judgment matrix and the fuzzy consistency judgment matrix.

[0154] (3) Calculate sensor weights

[0155] The formula for calculating the weight of the u-th sensor in the i-th sensor group is:

[0156]

[0157] Where α is inversely proportional to the difference in weights, when When the weight difference is greatest, take...

[0158] (4) Calculation of failure probability of gray number in interval

[0159] The probability of gray number failure in sensor group A is:

[0160]

[0161] in, Let represent the sensor weights in sensor group A.

[0162] The probability of gray number failure in sensor group B is:

[0163]

[0164] in, The sensor weights are those in sensor group B.

[0165] 3. Calculation of the failure probability of the actuator range gray number

[0166] The failure of actuator S(3,i.2) is caused by two reasons: ① electromagnetic stress experienced by the actuator; ② sensor failure. When either of these occurs, the actuator will be in a failed state.

[0167] The method for calculating the interval gray number failure probability of fuel metering valve and injector failure due to electromagnetic stress is similar to the method for calculating the interval gray number failure probability of sensors, denoted as...

[0168] The failure probability of the fuel metering valve interval ash count is:

[0169]

[0170] The failure probability of the injector zone ash number is:

[0171]

[0172] (II) The method for calculating the failure probability of the engine system interval gray number is as follows:

[0173] 1. Calculation of failure probability of gray number in subsystem interval

[0174] The probability of failure of the fuel subsystem range ash number is:

[0175]

[0176] The failure probability of the fuel injection subsystem range is:

[0177]

[0178] 2. Calculation of failure probability of engine system range gray number

[0179]

[0180] The failure probability of the engine system range gray number is determined by and The set that constitutes. Determining the range of values ​​can be transformed into... and Extreme value problems within the range of values. and The same principle applies to determining the range of values.

Claims

1. A method for assessing the electromagnetic vulnerability of a vehicle engine system, characterized in that, Includes the following steps: Step 1: Based on the composition and working principle of the engine system, establish a Bayesian network fault model of the engine system and clarify the fault logic relationship between each subsystem and component in the Bayesian network fault model; Step 2: Obtain the triangular fuzzy number corresponding to the fuzzy expression of the failure correlation between nodes in the Bayesian network fault model through the semantic lookup table, calculate the fuzzy conditional probability, and convert this fuzzy conditional probability into the interval gray number conditional probability. Step 3: Determine the electromagnetic stress probability density function and the component sensitivity threshold probability density function, calculate the component layer interval gray number failure probability, and combine it with the interval gray number conditional probability in Step 2 to calculate the engine system interval gray number failure probability. The method for calculating the failure probability of gray number in the component layer interval is as follows: (1) Calculation of sensor range gray number failure probability: According to the electromagnetic stress probability density function and component sensitivity threshold probability density function Calculate the interval gray number failure probability of the sensor by... and The upper bound for calculating the failure probability of the sensor interval gray number is denoted as . ,pass and The lower bound for calculating the failure probability of the sensor interval gray number is denoted as . The failure probability of the gray number in the sensor range is denoted as: ; (2) Calculation of the failure probability of the gray number in the sensor group interval: Since each sensor in the sensor group is independent and does not affect the others, the interval gray number failure probability of the sensor group can be calculated by weighted summation of the interval gray number failure probabilities of each sensor. The weight of each sensor is determined using fuzzy hierarchical analysis, and the steps are as follows: 1) Construct the judgment matrix: The 0.1~0.9 scaling method is used to scale the i-th sensor group S(3,i.1)( Construct the judgment matrix , This represents the i-th sensor group S(3,i.1) in which the i-th sensor is... The sensor and the first The relative importance of each sensor compared to the others; 2) Consistency Transformation: Perform a consistency transformation on the judgment matrix to obtain a fuzzy consistent judgment matrix. , No. Line number Column values The calculation formula is: ； in To determine the matrix of the first Seeking peace Similarly, The number of sensors in the i-th sensor group is also the order of the judgment matrix and the fuzzy consistency judgment matrix; 3) Calculate sensor weights: In the i-th sensor group, the th The formula for calculating the weight of each sensor is: ； in It is inversely proportional to the degree of difference in weights, when When the weight difference is greatest, take... ; 4) Calculation of the failure probability of the gray number interval: The probability of gray number failure in the i-th sensor group interval The calculation formula is: ； in For the j-th sensor in the i-th sensor group a The weight of each sensor; (3) Calculation of the failure probability of the actuator range gray number: Actuator Failure can be caused by two reasons: ① electromagnetic stress on the actuator; ② sensor failure. When either of these occurs, the actuator will be in a failed state. The method for calculating the interval gray number failure probability of actuator failure due to electromagnetic stress is the same as the method for calculating the interval gray number failure probability of sensors, denoted as... ; Actuator range gray number failure probability The calculation formula is: ； The method for calculating the interval gray number failure probability of the engine system is as follows: (1) Calculation of the failure probability of the gray number interval in the subsystem: ； ； in The failure of a subsystem can lead to Subsystems that have failed; (2) Calculation of the failure probability of the engine system in the range of gray numbers: ； The interval gray number failure probability of an engine system is a set composed of the interval gray number failure probabilities of subsystems and the interval gray number conditional probabilities. Determining the range of values ​​can be transformed into an extreme value problem within the range of the interval gray number failure probability and the interval gray number conditional probability of the subsystem. , , The method for determining the range of values ​​is similar.

2. The method for assessing the electromagnetic vulnerability of a vehicle engine system according to claim 1, characterized in that, In step 1, the method for constructing the Bayesian network fault model of the engine system is as follows: Divide the engine system into There are three subsystems, each consisting of one sensor group and one actuator, wherein the sensor group of the i-th subsystem includes m... i One sensor, The Bayesian network fault model of the engine system is divided into four layers: system layer, subsystem layer, component layer and electromagnetic stress layer. Node S(1,1) corresponds to the entire engine system, and node S(2,i) corresponds to the i-th subsystem of the engine system. The i-th subsystem S(2,i) includes a sensor group S(3,i.1) and an actuator S(3,i.2). The j-th sensor in the i-th sensor group S(3,i.1) is S(3,i.1.j), and the corresponding electromagnetic stress is S(4,i.1.j). The electromagnetic stress corresponding to the i-th actuator S(3,i.2) is S(4,i.2).

3. The method for assessing the electromagnetic vulnerability of a vehicle engine system according to claim 1, characterized in that, The semantic lookup table in step 2 is as follows: The failure correlation between nodes is categorized into {low, relatively low, moderate, relatively high, high}. The triangular fuzzy number corresponding to low node failure correlation is... The triangular fuzzy number corresponding to low correlation of node failure is The triangular fuzzy number typically corresponds to the correlation of node failures. The triangular fuzzy number corresponding to a high correlation between node failures is: The triangular fuzzy number corresponding to high correlation of node failure is .

4. The method for assessing the electromagnetic vulnerability of a vehicle engine system according to claim 1, characterized in that, The method for determining the conditional probability of the gray number in step 2 is as follows: For any two nodes with a failure logic relationship, the triangular fuzzy number corresponding to the fuzzy expression of their failure correlation obtained through a semantic lookup table is: Fuzzy conditional probability The calculation formula is: ； Among them, with and For example, the definition of triangular fuzzy number addition is: , Fuzzy conditional probability Convert to interval gray number conditional probability ; 。 5. The method for assessing the electromagnetic vulnerability of a vehicle engine system according to claim 1, characterized in that, The method for determining the electromagnetic stress probability density function of the strong electromagnetic pulse in step 3 is as follows: A simulation model was built in the CST microwave studio based on the vehicle's cable routing and mechanical structure. The simulation examined the electromagnetic stress coupled to the cables under strong electromagnetic pulse irradiation, sampling the incident angle, azimuth angle, and polarization angle at equal intervals. The incident angle, azimuth angle, and polarization angle were all sampled within the range of 0-90°. Equal-interval sampling was performed with a step size, and the obtained electromagnetic stress simulation data was fitted to a mean value using MATLAB. Standard deviation is normal distribution function That is, the electromagnetic stress probability density function.

6. The method for assessing the electromagnetic vulnerability of a vehicle engine system according to claim 1, characterized in that, The method for determining the component sensitivity threshold probability density function in step 3 is as follows: The operating status of sensors and actuators is monitored by a monitoring system. A strong electromagnetic pulse (EMP) irradiation test is conducted on the engine system, gradually increasing the EMP irradiation field intensity until the component malfunctions. Test data is recorded, with n sets of test data recorded for each component. The component sensitivity threshold test data follows the mean. The standard deviation is The normal distribution is not perfectly consistent with the component sensitivity threshold test data due to the randomness of the data. Even under the same test conditions, the mean and standard deviation of the fitted normal distribution will not be exactly the same. Therefore, the mean... Perform 99% confidence interval estimation, denoted as , The expression is: ； in, The sample mean. The standard deviation of the sample is 1. It is a t-distribution with n-1 degrees of freedom; Since the mean value of the sensitivity threshold for calculating component failure probability plays a major role, assuming the standard deviation remains constant, let be... for The minimum value, for The maximum value, Let be the probability density function of the component sensitivity threshold. The mean is The standard deviation is The component sensitivity threshold probability density function, The mean is The standard deviation is The component sensitivity threshold probability density function.

Images