Train air brake system random fault injection method and injection system

Abstract

The application relates to the technical field of fault analysis, and discloses a random fault injection method and an injection system for a train air brake system, which comprises the following steps: a simulation model of a target air brake system is constructed; all historical fault information is classified to obtain fault degree information and fault type information, and a fault degree state transition diagram and a fault type state transition diagram of the target air brake system are drawn; a fault type Markov chain is constructed according to the fault type state transition diagram, and a fault degree Markov chain is constructed according to the fault degree state transition diagram; a state transition probability matrix is determined, and a steady-state distribution is solved; a probability distribution of all fault information is set according to the steady-state distribution, and all fault information is randomly extracted in layers based on the probability distribution; and the extracted fault information is injected into the simulation model; the application solves the problem that the existing fault injection method cannot conveniently and effectively perform random fault injection on the train air brake system.

Description

Technical Field

[0001] This invention relates to the field of fault analysis technology, and in particular to a method and system for injecting random faults into a train air braking system. Background Technology

[0002] With the large-scale operation of high-speed trains, the safety of their operation has become an increasingly important concern. The air braking system is one of the key systems ensuring the safe operation of high-speed trains, but it is also a frequent source of failures. Therefore, it is necessary to conduct fault research on the air braking system to provide strong data support for subsequent fault detection and diagnosis research. Currently, most fault studies on air braking systems use random fault injection to simulate faults. This typically involves fitting a large amount of raw sample data to obtain a load distribution function, and then randomly sampling fault parameters based on this function to complete the random fault injection. However, constructing the fault distribution function is a cumbersome process, requiring data processing and parameter estimation. Furthermore, current fault research lacks raw fault data for the air braking system, making comprehensive and accurate fault distribution function fitting impossible. Therefore, existing fault injection methods cannot conveniently and effectively inject random faults into train air braking systems. Summary of the Invention

[0003] This invention provides a method and system for injecting random faults into a train air braking system, in order to solve the problem that existing fault injection methods cannot conveniently and effectively inject random faults into the train air braking system.

[0004] To achieve the above objectives, the present invention employs the following technical solution:

[0005] In a first aspect, the present invention provides a method for injecting random faults into a train air braking system, comprising:

[0006] A simulation model of the target air braking system is constructed, and all fault information of the target air braking system is obtained to build a fault model library.

[0007] All fault information is classified to obtain fault severity information and fault type information. A fault severity state set is determined based on the fault severity information, and a fault type state set is determined based on the fault type information.

[0008] Based on the set of fault severity states, a fault severity state transition diagram of the target air brake system is drawn; based on the set of fault type states, a fault type state transition diagram of the target air brake system is drawn.

[0009] Construct a fault type Markov chain based on the fault type state transition diagram, and construct a fault degree Markov chain based on the fault degree state transition diagram.

[0010] The state transition probability matrix is ​​determined based on the fault type Markov chain and the fault degree Markov chain, and the steady-state distribution is obtained by solving the state transition probability matrix.

[0011] Based on the steady-state distribution, a probability distribution for all fault information is set, and based on this probability distribution, all fault information is randomly sampled in a hierarchical manner, and the sampled fault information is injected into the simulation model.

[0012] Secondly, embodiments of the present invention provide a random fault injection system for a train air braking system, including a processor and a memory;

[0013] Memory is used to store computer programs.

[0014] When a processor executes a program stored in memory, it implements any of the steps of the method described in the first aspect.

[0015] Beneficial effects:

[0016] The random fault injection method for train air brake systems provided by this invention classifies all fault information of the target air brake system into fault severity information and fault type information, and determines the fault severity state set and fault type state set. A fault severity state transition diagram is drawn using the fault severity information state set, and a fault type state transition diagram is drawn using the fault type state set, thereby constructing a fault type Markov chain and a fault state Markov chain, and determining the state transition probability matrix. A steady-state distribution is obtained through the state transition probability matrix, and the probability distribution of all fault information can be set according to the steady-state distribution. Based on this probability distribution, fault information is randomly injected into the simulation model of the target air brake system for fault analysis. Through the above steps, the fault distribution function can be fitted more comprehensively and accurately, thus quickly, conveniently, and effectively injecting fault information randomly into the simulation model to achieve the effect of fault injection. Attached Figure Description

[0017] Figure 1 This is a flowchart of a preferred embodiment of the random fault injection method for a train air brake system according to the present invention;

[0018] Figure 2 This is a fault-degree Markov chain state transition diagram of a preferred embodiment of the present invention;

[0019] Figure 3 This is a fault type Markov chain state transition diagram of a preferred embodiment of the present invention;

[0020] Figure 4 This is a schematic diagram of the random fault injector structure of the train air brake system according to a preferred embodiment of the present invention;

[0021] Figure 5 This is a schematic diagram of the user setting interface of the random fault injector of the train air braking system built in the GUI according to a preferred embodiment of the present invention;

[0022] Figure 6 In a preferred embodiment of the present invention, a fault type Markov chain state transition diagram is generated in the GUI according to user settings;

[0023] Figure 7 In a preferred embodiment of the present invention, a fault-level Markov chain state transition diagram is generated in the GUI according to user settings. Detailed Implementation

[0024] The technical solution of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] Unless otherwise defined, the technical or scientific terms used in this invention shall have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Similarly, the terms "an" or "a" and similar terms do not indicate a quantity limitation, but rather indicate the presence of at least one. The terms "connected" or "linked" and similar terms are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. "Up," "down," "left," "right," etc., are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship also changes accordingly.

[0026] Please see Figure 1-7 :

[0027] Example 1:

[0028] This embodiment provides a method for injecting random faults into a train air braking system, including the following steps:

[0029] Step 1: Construct a fault model library, classify fault information according to fault type and fault severity, and determine the fault type state set and fault severity state set.

[0030] Step 2: Based on the system's operating behavior under fault conditions, draw the system's fault type state transition diagram and fault severity state transition diagram.

[0031] Step 3: Construct mathematical expressions for the fault degree Markov chain and the fault type Markov chain, determine the state transition probability matrix, and find its steady-state distribution.

[0032] Step 4: Set the probability distribution of fault information in the model library according to the steady-state distribution, perform hierarchical random fault extraction according to the distribution, and inject the corresponding faults into the simulation model.

[0033] Specifically, in step one, a fault model library containing seven fault types and four fault degrees was established for the train air brake system, including components such as the EP valve, train pipe, brake cylinder, and sensors, as shown in Table 1. The total number of fault models in the fault model library is denoted as N, and the calculation formula is as follows:

[0034]

[0035] in, The number of fault models representing the degree of the j-th fault type of the i-th fault.

[0036] Table 1

[0037]

[0038]

[0039] The fault types include normal operation, EP valve solenoid valve leakage, EP valve spring failure, EP valve pipeline blockage, sensor gain failure, brake cylinder air leakage failure, and train pipe air leakage failure. These states sequentially form the fault type state set.

[0040] Fault severity is categorized into healthy, minor, moderate, and severe faults, with these states forming a set of fault severity states in that order.

[0041] Each of the above fault types includes the above four fault levels.

[0042] Specifically, in step two, the principles for constructing the fault type state transition diagram and the fault severity state transition diagram of the train air braking system are as follows:

[0043] State transition diagram for fault type:

[0044] ① When studying the state transition of a single fault in a system, each fault state and healthy state is reachable from each other and connected to each other, while different fault states are not reachable from each other. The state transition diagram should satisfy the following equation:

[0045]

[0046] ② When a single fault state transition occurs, different fault states are unreachable. The state transition diagram should satisfy the following equation:

[0047]

[0048] State transition diagram for fault severity:

[0049] ① Ignoring system repairability, faults always tend to worsen, and the state transition diagram should satisfy the following equation:

[0050]

[0051] ② Assuming the system is repairable, any fault state can be directly restored to a healthy state. The state transition diagram should satisfy the following equation:

[0052]

[0053] The constructed fault severity state transition diagram and fault severity state transition diagram are shown below. Figure 1 and Figure 2 .

[0054] Specifically, in step three, a Markov chain is defined as a set of discrete random variables with Markov properties, which has the unique "memoryless property" that means that after the random variable in step n is given, the random variable in step n+1 is conditionally independent of the other random variables.

[0055] For a set of random variables X = {X_{n} in a probability space, whose exponent is a one-dimensional countable set, ... n : n>0}, the values ​​of the random variable are all contained in the countable set S={s0,s1,...,s m The countable set S is called the state set. The homogeneity of a homogeneous Markov chain is reflected in the fact that the transition probability depends only on the state before and after the transition, and is independent of the two time points.

[0056] Therefore, a homogeneous Markov chain can be expressed as follows:

[0057] P(X n+1 |X n ..., X1) = P(X n+1 |X n )=P(X1|X0) (6)

[0058] For any time n, define π i Indicate its probability distribution, Π i，j The mathematical expression for its one-step transition probability is as follows:

[0059] π i =P(X) n =s i ), s i ∈S (7)

[0060] Π i，j =P(X) n+1 =s j |X n =s i ), s i s j ∈S (8)

[0061] A Markov chain has a unique steady-state solution when all objects satisfy the following properties: recurrence, aperiodicity, and pairwise connectivity. This steady-state solution is independent of the initial state and can be expressed as:

[0062]

[0063] For the fault severity Markov chain of the EP valve solenoid valve leakage fault type: Define H d =(Π) i，j (i, j = 0, 1, 2, 3) is the state transition probability matrix of the fault-level Markov chain, expressed as follows:

[0064]

[0065] Define π d For the stationary probability distribution of the fault-prone Markov chain, π d0 , π d1 , π d2 , π d3 Let represent the steady-state probabilities of four states: healthy, minor fault, moderate fault, and severe fault, respectively. Then we have:

[0066] π d =[π d0 , π d1 , π d2 , π d3 (11)

[0067] π d H d =π d (12)

[0068]

[0069] Solving equations (10) and (12) simultaneously yields π. di The specific values ​​of (i = 0, 1, 2, 3) are [0.7993, 0.0740, 0.0685, 0.0582], which is the steady-state solution.

[0070] For the Markov chain of fault types: Besides the normal operating state, there are six fault types: EP valve solenoid leakage, EP valve spring failure, EP valve pipeline blockage, sensor failure, brake cylinder air leakage, and train pipe air leakage. The state transition probability matrix H of the fault type Markov chain can be constructed using the steady-state probability distribution of the fault severity Markov chain. t .

[0071] From equations (10) to (12), it can be seen that the steady-state probability of the Markov chain maintaining a healthy state for the fault type EP valve solenoid valve leakage is π. d0 The total steady-state probability of a failure is 1-π. d0 .

[0072] Define λ i (i = 1, 2, ..., 6) represents the failure rate of the i-th failure type, i.e., from state... Transition to state The probability of leakage in the EP valve solenoid valve, classified as the first type of failure, is expressed by the following formula:

[0073]

[0074] in, This represents the first type of failure, namely the failure level corresponding to leakage in the EP valve solenoid valve, and the steady-state probability of the Markov chain maintaining a healthy state.

[0075] Define μ i (i = 1, 2, ..., 6) represents the state from Transition to state The probability of.

[0076] Therefore, the state transition probability matrix H of the Markov chain of fault types is constructed. t as follows:

[0077]

[0078] Define π t Let π be the stationary probability distribution of the Markov chain of fault types. t0 , π t1 , ..., π t6 Let the steady-state probabilities of seven states occur: normal operation, EP valve solenoid valve leakage, EP valve spring failure, EP valve pipeline blockage, sensor failure, brake cylinder air leakage, and train pipe air leakage. Then:

[0079] π t =[π t0 , π t1 , ..., π t6(16)

[0080] π t H t =π t (17)

[0081]

[0082] Solving equations (15) and (17) simultaneously yields π. ti The specific values ​​of (i = 0, 1, ..., 6) [0.6767, 0.0529, 0.0876, 0.0408, 0.0272, 0.0529, 0.0619] are the steady-state solutions.

[0083] Based on the obtained steady-state distribution π d and π t Set the probability distribution of fault information in the model library, perform hierarchical random fault extraction according to the distribution, and inject the corresponding faults into the simulation model.

[0084] Example 2

[0085] This embodiment provides a method for random injection of composite faults.

[0086] The fault types in Example 1 are all single fault types. However, in actual research, due to the complexity and coupling of the system, multiple faults often occur simultaneously, i.e., compound faults. Compound faults refer to the simultaneous occurrence of multiple faults of different types or multiple faults of the same type originating from different fault sources, or the mutual influence of faults occurring in components of multiple subsystems of the system, resulting in new fault characteristics.

[0087] For fault injection, the essence of composite faults is the simultaneous injection of multiple different faults, either of different types or from different devices. Random fault injection of composite faults involves simultaneously extracting multiple fault models according to a probability distribution and injecting them into the simulation model. The method is as follows:

[0088] Assuming the set of fault type states where a compound fault occurs is still [value missing] This means selecting multiple fault types from EP valve solenoid valve leakage, EP valve spring failure, EP valve pipeline blockage, sensor failure, brake cylinder air leakage failure, and train pipe air leakage failure to form a composite fault.

[0089] The steady-state distribution of the Markov chain of fault types obtained by the above method is π. t = [0.6767, 0.0529, 0.0876, 0.0408, 0.0272, 0.0529, 0.0619]. Assuming that the occurrence of various individual faults is independent of each other, the probability of a compound fault occurring is calculated using joint probability.

[0090] Taking a compound fault, which is composed of two different types of faults combined in pairs, as an example, we define... For composite fault types and The probability of is calculated using the following formula:

[0091]

[0092] The probabilities of composite faults composed of different types of faults, calculated according to formula (18), are shown in Table 2:

[0093] Table 2

[0094]

[0095]

[0096] The severity of the aforementioned composite faults is not uniformly set. Instead, the probability distribution of the severity of each fault type is determined according to its own characteristics. For details, please refer to the fault severity Markov chain method mentioned above, which will not be elaborated here.

[0097] Example 3

[0098] Please see Figure 4 This application provides a random fault injector for a train air brake system, including: a user setting interface, a random fault injection control unit, and a train air brake system model library. The user setting interface is connected to the random fault injection control unit and the train air brake system model library, respectively.

[0099] The user setting interface is used to set the Markov chain state transition matrix parameters for the fault type and the Markov chain state transition matrix parameters for the fault degree of the train air braking system; and to display the Markov chain state transition diagram and steady-state probability for the fault type and fault degree.

[0100] The random fault injection control unit is used to perform random fault injection based on the fault information.

[0101] The train air brake system model library is used to store the train air brake system fault models.

[0102] In one example, the user interface is set up as follows: This interface is built in the Matlab GUI. Its function is to allow users to set the parameters of the Markov chain state transition matrix for the fault type and the Markov chain state transition matrix for the fault degree through the editable text boxes on the interface, and to display the Markov chain state transition diagram and steady-state probability for the fault type and fault degree; after confirmation, random fault injection begins.

[0103] The fault injection control unit transmits user-defined random fault injection information to various modules in Simulink, including the preloading module, random fault generation module, steady-state probability calculation module, and model switching module. The steady-state probability calculation module calculates the steady-state probability based on the fault type and severity Markov chain state transition matrix parameters in the user-defined interface and displays it. The random fault generation module randomly generates fault parameters based on the probability distribution of each fault parameter generated by the steady-state probability calculation module. The preloading module preloads specific fault models based on the output signal of the random fault generation module and disconnects the remaining fault models to save simulation resources. The model switching module performs model switching based on the model switching control signal, completing the switch from the normal model of the train air braking system to various fault models.

[0104] Train air brake system model library: The normal train air brake system model and various types of train air brake system fault models generated above are built and stored in Amesim and Simulink software to form a train air brake system valve model library for model switching.

[0105] Please see Figure 5 Following the prompts in the Introduction diagram, set the parameters for the fault type Markov chain state transition matrix in the lower left corner. See Example 1 for specific parameters. After setting, click the OK button. A fault type Markov chain state transition diagram drawn according to the user-set parameters will then pop up (see Example 1). Figure 6 The parameters are then set for the fault severity Markov chain state transition matrix in the lower right corner. See Example 1 for specific parameters. After setting, click the OK button; a fault severity Markov chain state transition diagram drawn according to the user-set parameters will then pop up (see Example 1). Figure 7 The system calculates the steady-state probability of the Markov chain and the fault severity level. After confirmation, click the "Start Fault injection" button to begin random fault injection. The fault injection control unit then calculates the steady-state probability of the Markov chain in the steady-state probability calculation module based on the user-defined parameters. The random fault generation module then randomly generates fault parameters based on the probability distribution of each fault parameter generated by the steady-state probability calculation module. The preloading module preloads a specific fault model based on the output signal of the random fault generation module and disconnects the remaining fault models. The model switching module performs model switching based on the model switching control signal. Upon receiving the model switching signal, the switch is switched from the normal model to the corresponding fault model, completing the random fault injection.

[0106] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.

Claims

1. A method for random fault injection in a train air brake system, the method comprising: include: A simulation model of the target air braking system is constructed, and all historical fault information of the target air braking system is obtained to build a fault model library. classifying all historical fault information to obtain fault degree information and fault type information, and determining a fault degree state set according to the fault degree information , determining a fault type state set according to the fault type information ; Based on the set of fault severity states, a fault severity state transition diagram of the target air brake system is drawn; based on the set of fault type states, a fault type state transition diagram of the target air brake system is drawn. Construct a fault type Markov chain based on the fault type state transition diagram, and construct a fault degree Markov chain based on the fault degree state transition diagram. The state transition probability matrix is ​​determined based on the fault type Markov chain and the fault degree Markov chain, and the steady-state distribution is obtained by solving the state transition probability matrix. The probability distribution of all fault information is set according to the steady-state distribution, and all fault information is randomly sampled in layers based on this probability distribution. The sampled fault information is then injected into the simulation model. The step of drawing a fault state transition diagram of the target air braking system based on the fault state set includes: When the fault becomes irreparable, it progresses towards a more severe state. At this point, the fault state transition diagram satisfies the following equation: ； wherein represents a first i level of failure condition, represents a first j level of failure condition, When the fault is repairable, the normal state tends to recover to a healthy state. At this point, the fault state transition diagram satisfies the following equation: ； wherein, represents a first i degree fault condition, represents a healthy state, The step of drawing a fault type state transition diagram of the target air braking system based on the fault type state set includes: Different fault types are reachable from health, and the fault type state transition diagram satisfies the following equation: ； wherein, represents a first i type of failure state, represents a healthy state, When different fault types are unreachable, the fault type state transition diagram satisfies the following equation: ； in, Indicates the first i Types of fault states Indicates the first j Types of faults k Indicates the number of fault types. l The number of faults indicates the degree of failure.

2. The random fault injection method for a train air braking system according to claim 1, characterized in that, The step of acquiring all fault information of the target air braking system to construct a fault model library includes: Based on the EP valve, train pipe, brake cylinder, and sensors of the target air braking system, a system is established that includes... k Types of faults and l A fault model library for various fault levels; The total number of fault models in the fault model library is denoted as N, and the calculation formula for the fault model library is as follows: ； in, Indicates the first i The first type of fault j The number of fault models of medium severity k Indicates the number of fault types. l The number of fault levels is represented by N, where N represents the total number of fault models in the fault model library.

3. The random fault injection method for a train air braking system according to claim 1, characterized in that, Determining the fault severity state set based on the fault severity information includes: Fault severity information includes health and l The fault severity levels are grouped into fault severity states, and these fault severity states are combined into a fault severity state set, the expression of which is as follows: ； in, Represents a set of fault severity states. Indicates the first l Level of fault status, It is an integer greater than or equal to 0; Determining the fault type status set based on the fault type information includes: Fault type information includes normal operating status and k The fault type states are composed of various fault types, and these fault type states are grouped into a fault type state set, the expression of which is as follows: ； in, Represents a set of fault type states. Indicates the first k Various fault levels For each of the above fault types, the integer is greater than or equal to 0. l The degree of failure.

4. The random fault injection method for a train air braking system according to claim 1, characterized in that, The method further includes: A Markov chain is defined as a set of discrete random variables that possess the Markov property, that is, given the first... n After the first random variable, the first n + 1 The random variable of the step is conditionally independent of the other random variables; For a set of random variables in a probability space whose exponent is a one-dimensional countable set. The values ​​that a random variable can take are all contained in the countable set. This countable set S is called the state set; Therefore, a homogeneous Markov chain can be expressed as follows: ； Among them, for any time n ,definition Indicate its probability distribution, The probability of a one-step transition is expressed as follows: ； ； in, express The probability, Indicates the first i a state, Indicates from the first j state to the first i The one-step transition probability of a state. Indicates the first j a state, A Markov chain has a unique steady-state solution when all objects satisfy the following properties: recurrence, aperiodicity, and pairwise connectivity. This steady-state solution is independent of the initial state and can be expressed as: 。 5. The random fault injection method for a train air braking system according to claim 1 or 4, characterized in that, The step of constructing a fault-degree Markov chain based on the fault-degree state transition diagram includes: definition Let be the state transition probability matrix of a fault-prone Markov chain, when hour, The expression is as follows: (1) in, Indicates from state Transition to state The probability of failure is denoted as the failure rate. Indicates from state Transition to state The probability of this is denoted as the repair rate; definition For the stationary probability distribution of the fault-level Markov chain, Let represent the steady-state probabilities of healthy and l-level failures, respectively. Then we have: (2) (3) (4) Solving equations (1) and (4) simultaneously yields the result. The specific value of , i.e. the steady-state solution.

6. The random fault injection method for a train air braking system according to claim 1 or 4, characterized in that, The step of constructing a fault type Markov chain based on the fault type state transition graph includes: definition For the occurrence of the first The probability of a fault type, i.e., from state Transition to state The probability is expressed as follows: (5) in, Indicates the first The failure severity corresponding to each type of failure and the steady-state probability of the Markov chain maintaining a healthy state. definition Indicates from state Transition to state The probability of; This allows us to construct the state transition probability matrix of the Markov chain with fault types. as follows: (6) definition For the stationary probability distribution of the Markov chain of fault types, Given the steady-state probabilities of normal operation and k different fault states, we have: (7) (8) (9) Solving equations (6) and (8) simultaneously yields the result. The specific value of , i.e. the steady-state solution.

7. The random fault injection method for a train air braking system according to claim 1, characterized in that, The step of injecting the extracted fault information into the simulation model includes: Define the fault type state set as The obtained fault type Markov chain steady-state distribution is as follows: When individual faults occur independently, the probability of a compound fault is calculated using joint probability. When a compound fault is composed of two pairs of different types of faults, the definition is... For composite fault types and The probability of is calculated using the following formula: ； in, Indicates a composite fault type as and The probability, Indicates the first i Types of faults Indicates the first j Types of faults Indicates occurrence i The probability of each type of failure Indicates occurrence j The probability of each type of failure.

8. A random fault injection system for a train air braking system, characterized in that, Including processor and memory; Memory, used to store computer programs; A processor, when executing a program stored in memory, implements the steps of the method described in any one of claims 1-7.

Images