A rocket engine fault diagnosis method, device, equipment and medium

Abstract

The application discloses a rocket engine fault diagnosis method, device, equipment and medium, and relates to the technical field of rocket engines, to solve the problem that the standard value is difficult to determine due to state drift and test difference of the rocket engine, and misdiagnosis of fault diagnosis is easily caused.A rocket engine fault diagnosis method comprises the following steps: obtaining measurement data; based on the measurement value of the covariant, a probability prediction neural network is used to predict the probability distribution of a target parameter; an abnormal threshold range is determined according to the probability distribution of the target parameter; it is judged whether the measurement value of the target parameter exceeds the abnormal threshold range, if yes, the measurement value of the target parameter at the last moment is input into an autoregressive neural network to obtain a standard value; a measurement residual is obtained according to the standard value and the measurement value of the target parameter, an observer is used to observe a fault factor, and fault detection is realized.The rocket engine fault diagnosis method provided by the application is used for real-time determination of the standard value, and the accuracy of rocket engine fault diagnosis is improved.

Description

Technical Field

[0001] This invention relates to the field of rocket engine technology, and in particular to a method, apparatus, equipment and medium for diagnosing rocket engine faults. Background Technology

[0002] Rocket engines are among the most prone to failure on launch vehicles, and serious engine malfunctions can affect launch reliability. Therefore, fault diagnosis is crucial to promptly identify engine failures and implement fault-tolerant control measures to minimize their impact on the launch mission.

[0003] Engines experience state drift and variations between different engine batches during operation. State drift causes the standard values ​​characterizing the engine's normal operating condition to change, and these standard values ​​also differ between different engine batches. However, existing engine fault diagnosis methods typically set the standard values ​​as the design values ​​or average values ​​of the measured parameters. This can lead to serious misdiagnosis of faults. Summary of the Invention

[0004] The purpose of this invention is to provide a method, device, equipment, and medium for diagnosing rocket engine faults, in order to solve the problem that rocket engines have state drift and differences in the number of launches, making it difficult to determine standard values ​​and easily leading to misdiagnosis of faults.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] On one hand, the present invention provides a method for diagnosing rocket engine faults, comprising:

[0007] Acquire engine measurement data; the measurement data includes covariates and target parameters;

[0008] Based on the measured values ​​of the covariates, a probability distribution of the predicted values ​​of the target parameters is obtained using a probabilistic prediction neural network.

[0009] The abnormal threshold range of the target parameter at the corresponding time is determined based on the probability distribution of the predicted values ​​of the target parameter;

[0010] Determine whether the measured value of the target parameter exceeds the abnormal threshold range. If it does, input the measured value of the target parameter at the previous time step into the autoregressive neural network of the target parameter to obtain the standard value of the target parameter.

[0011] The measurement residual of the target parameter is obtained by comparing the standard value of the target parameter with the measured value of the target parameter at the corresponding time.

[0012] Based on the measurement residuals, the fault factors are observed using an observer and an augmented state-space model of the rocket engine to obtain estimated values ​​of the fault factors, thereby achieving fault diagnosis of the engine. The augmented state-space model of the rocket engine is a model using the fault factors as system state parameters. Compared with existing technologies, this invention provides a rocket engine fault diagnosis method, including acquiring engine measurement data; using a probabilistic prediction neural network to predict the probability distribution of predicted values ​​of target parameters based on the measured values ​​of covariates; determining the abnormal threshold range of the target parameters at the corresponding time based on the probability distribution of the predicted values ​​of the target parameters; determining whether the measured value of the target parameters exceeds the abnormal threshold range; if it does, the engine startup is abnormal, and the measured value of the target parameters at the previous time is input into the autoregressive neural network of the target parameters to obtain the standard value of the target parameters; after determining the engine abnormality, the measured value of the target parameters at the moment before the engine abnormality is input into the autoregressive neural network of the target parameters to predict the target parameters at the next time. The predicted values ​​can mitigate the deviation of the autoregressive network, making the standard values ​​of the target parameters more accurate. The measurement residuals of the target parameters are obtained based on the standard values ​​and the measured values ​​at the corresponding time points. Based on these residuals, fault factors are observed using an observer and an augmented state-space model of the rocket engine to obtain estimated values, thus enabling fault diagnosis of the engine. The predictive neural network and the autoregressive network of the target parameters can predict the standard values ​​of the target parameters in real time, eliminating the impact of differences between different batches of the same type of engine and state drift during engine operation on the online observation of fault factors. The augmented state-space model of the rocket engine is used to observe the fault factors.

[0013] Secondly, the present invention also provides a rocket engine fault diagnosis device, comprising:

[0014] An engine measurement data acquisition module is used to acquire engine measurement data; the measurement data includes covariates and target parameters.

[0015] The probability distribution calculation module for the predicted value of the target parameter is used to predict the probability distribution of the predicted value of the target parameter based on the measured value of the covariate using a probability prediction neural network.

[0016] The abnormal threshold range determination module is used to determine the abnormal threshold range of the target parameter at the corresponding time based on the probability distribution of the predicted value of the target parameter.

[0017] The target parameter standard value determination module is used to determine whether the measured value of the target parameter exceeds the abnormal threshold range. If it does, the measured value of the target parameter at the previous moment is input into the autoregressive neural network of the target parameter to obtain the standard value of the target parameter.

[0018] The measurement residual calculation module is used to obtain the measurement residual of the target parameter based on the standard value of the target parameter and the measured value of the target parameter at the corresponding time.

[0019] The fault diagnosis module is used to observe the fault factors based on the measurement residuals through an observer and the augmented state space model of the rocket engine, obtain the estimated values ​​of the fault factors, and realize the fault diagnosis of the engine; the augmented state space model of the rocket engine is a model with the fault factors as system state parameters.

[0020] Compared with the prior art, the beneficial effects of the rocket engine fault diagnosis device provided by the present invention are the same as those of the rocket engine fault diagnosis method described in the above technical solution, and will not be repeated here.

[0021] Thirdly, the present invention also provides a rocket engine fault diagnosis device, including a processor and a communication interface coupled to the processor; the processor is used to run computer program instructions to implement the above-mentioned rocket engine fault diagnosis method.

[0022] Compared with the prior art, the beneficial effects of the rocket engine fault diagnosis device provided by the present invention are the same as the beneficial effects of the rocket engine fault diagnosis method described in the above technical solution, and will not be repeated here.

[0023] Fourthly, the present invention also provides a computer-readable storage medium, characterized in that it includes: instructions stored in the computer-readable storage medium, which, when executed, implement the above-mentioned rocket engine fault diagnosis method.

[0024] Compared with the prior art, the beneficial effects of the computer-readable storage medium provided by the present invention are the same as those of the rocket engine fault diagnosis method described in the above technical solution, and will not be repeated here. Attached Figure Description

[0025] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this invention, illustrate exemplary embodiments of the invention and are used to explain the invention, but do not constitute an undue limitation of the invention. In the drawings:

[0026] Figure 1 A flowchart of a rocket engine fault diagnosis method provided by the present invention;

[0027] Figure 2 The flowchart for online observation of fault factors provided by this invention;

[0028] Figure 3 This is a schematic diagram of the structure of a rocket engine fault diagnosis device provided by the present invention;

[0029] Figure 4 The present invention provides a structural block diagram of a rocket engine fault diagnosis device. Detailed Implementation

[0030] To facilitate a clear description of the technical solutions in the embodiments of the present invention, the terms "first" and "second" are used to distinguish identical or similar items with essentially the same function and effect. For example, the first threshold and the second threshold are merely used to distinguish different thresholds and do not limit their order. Those skilled in the art will understand that the terms "first" and "second" do not limit the quantity or execution order, and that the terms "first" and "second" are not necessarily different.

[0031] It should be noted that in this invention, the terms "exemplary" or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary" or "for example" in this invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0032] In this invention, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can represent: a, b, c, a combination of a and b, a combination of a and c, a combination of b and c, or a, b, and c, where a, b, and c can be single or multiple.

[0033] Before introducing the embodiments of the present invention, the relevant terms involved in the embodiments of the present invention are first defined as follows:

[0034] Kalman filtering is a highly efficient recursive filter (autoregressive filter) that can estimate the state of a dynamic system from a series of measurements that are not completely free of noise.

[0035] Object-oriented programming (OOP) is a software development methodology and a programming paradigm. It is contrasted with procedural programming. OOP approaches treat related data and methods as a whole, modeling the system from a higher level and more closely reflecting the natural operating patterns of things.

[0036] A probability distribution is a method used to describe the probabilistic patterns of a random variable's values, that is, the possible values ​​of a random variable and the probability of obtaining the corresponding values.

[0037] Staged combustion cycle engines, such as the main engines of the Space Shuttle, often exhibit state drift. State drift refers to the change in the engine's state over time during steady-state operation, with different engines showing different changes. The cause of this is unclear, even though the engine is functioning perfectly normally. Furthermore, due to accumulated manufacturing tolerances, rocket engines can exhibit differences in operating characteristics between different engines of the same model. Because of state drift and engine variations, standard values ​​for measurement parameters are difficult to determine, easily leading to serious misdiagnosis of engine faults.

[0038] To address the aforementioned problems, this invention provides a method, apparatus, equipment, and medium for diagnosing rocket engine faults. It employs time series analysis to determine standard values ​​for the fault diagnosis process in real time. Based on the residual between sensor measurements and standard values, an observer is used to monitor the changes in fault factors relative to the baseline value, thereby achieving online fault diagnosis of the engine. The following is a detailed description in conjunction with the accompanying drawings.

[0039] Figure 1 A flowchart of a rocket engine fault diagnosis method provided by the present invention is shown below. Figure 1 As shown, it includes the following steps:

[0040] Step 101: Obtain engine measurement data.

[0041] Measurement data may include multiple covariates and multiple target parameters; covariates are independent variables, also known as explanatory variables, which are not manipulated by the experimenter but still affect the experimental results; target parameters are dependent variables that have a linear relationship with the covariates. There is a mapping relationship between the target parameters and one or more covariates.

[0042] Step 102: Based on the measured values ​​of the covariates, use a probabilistic prediction neural network to predict the probability distribution of the predicted values ​​of the target parameters.

[0043] A probabilistic predictive neural network (PRN) is a probabilistic prediction model. Common PPNs include DeepAR, DeepState, and SSDNet. A PPN consists of a Structured State-Space Equation (SSM) and a Recurrent Neural Network (RNN), or a Structured State-Space Equation (SSM) and a Transformer network. At time t, the input to the PPN is the information H from the hidden layer of the RNN at time t-1. t-1 The probability distribution of the predicted target parameters at time t is generated by taking the measured values ​​of the covariates at time t. The probability distribution of the predicted target parameters refers to the probability distribution of the i target parameters at time t, expressed in terms of mean and variance; therefore, the prediction results are expressed in terms of mean and variance. The goal of establishing the probabilistic prediction neural network is to build the probability distribution of the target parameters of an engine model, given historical measurement data of the target parameters and covariates under normal operating conditions.

[0044] Before making predictions, it is necessary to establish an initial probability prediction neural network and train it. The specific method is as follows:

[0045] Establish an initial probability prediction neural network;

[0046] The given initial values ​​and the measured values ​​of the target parameter at time n and the covariate at time n+1 in the engine normal test data during the steady state phase are output to the initial probability prediction neural network to obtain the probability distribution of the predicted value of the target parameter at time n+1. The measured values ​​at time n+1 and the measured values ​​of the covariate at time n+2 are used as the input to the initial probability prediction neural network to obtain the probability distribution of the predicted value of the target parameter at time n+2.

[0047] The mean of the probability distribution of the predicted target parameter values ​​is compared with the measured value of the target parameter at the corresponding time. Based on the comparison result, the parameters in the initial probability prediction neural network are adjusted, and the probability prediction neural network is obtained after training.

[0048] Step 103: Determine the abnormal threshold range of the target parameter at the corresponding time based on the probability distribution of the predicted value of the target parameter.

[0049] The mean and variance of the predicted values ​​of the target parameters can be obtained from the probability distribution. The predicted value of the target parameter is most likely to be the mean, and the variance is the probability density. The anomaly threshold range refers to the range within which the engine is considered to have an anomaly. Anomalies are detected by checking this range; if the measured value of the target parameter exceeds this range, it indicates an engine anomaly. Specifically, the anomaly threshold range is obtained by expanding the mean of the target parameter in the probability distribution to both sides by a preset width.

[0050] Step 104: Determine whether the measured value of the target parameter exceeds the abnormal threshold range. If it does, input the measured value of the target parameter at the previous moment into the autoregressive neural network of the target parameter to obtain the standard value of the target parameter.

[0051] If the measured value of the target parameter is within the abnormal threshold range, it indicates that the engine is operating normally, and the prediction for the next moment is performed; if the measured value of the target parameter is outside the abnormal threshold range, the engine is abnormal, and the probability prediction neural network stops operating.

[0052] The process of establishing an autoregressive neural network for target parameters is as follows:

[0053] First, establish an initial autoregressive neural network;

[0054] Then, the measured value of the target parameter at time n-1 in the normal test data of the same engine model during the steady state phase is input into the initial autoregressive neural network to predict the predicted value of the target parameter at time n.

[0055] The measured value of the target parameter at time n in the normal test data of the same engine model during the steady state phase is compared with the predicted value of the target parameter at time n. The parameters of the initial autoregressive neural network are adjusted according to the comparison results, and the autoregressive neural network of the target parameter is obtained after training is completed, where n is a positive integer.

[0056] An autoregressive neural network (ARNN) for the target parameters is run to obtain the standard value of the target parameters. The initial value of the ARNN is the measurement data at the moment before the anomaly occurred. Specifically, when using the trained ARNN for the target parameters, assuming the transmitter anomaly occurs at time n, the measured value of the target parameter at time n-1 is input into the ARNN to predict the predicted value of the target parameter at time n. This predicted value is then used as the standard value of the target parameter at time n. Furthermore, to avoid data contamination caused by the fault, the output of the ARNN is used as the input to the neural network at the next time step to achieve autoregressive prediction.

[0057] Step 105: Obtain the measurement residual of the target parameter based on the standard value of the target parameter and the measured value of the corresponding target parameter.

[0058] The measurement residual is obtained by subtracting the standard value of the target parameter at time n+1 from the measured value of the target parameter at time n+1; at this time, the measured value of the target parameter at time n+1 is the measured value at the time of engine abnormality.

[0059] Step 106: Based on the measurement residuals, the fault factors are observed through the observer and the augmented state space model of the rocket engine to obtain the estimated values ​​of the fault factors, thereby realizing the fault diagnosis of the engine.

[0060] The augmented state-space model of the rocket engine is a model with the failure factor as the system state parameter.

[0061] Specifically, the steps for establishing an augmented state-space model of a rocket engine are as follows:

[0062] First, an object-oriented rocket engine model is established;

[0063] An object-oriented rocket engine model can be built using Simulink software. Specifically, based on the system dynamics equations, subsystem component-level models are established, including models of centrifugal pumps, turbines, piping, gas generators, thrust chambers, control systems, and physical property models. Then, these subsystem component-level models are connected to obtain the object-oriented rocket engine model.

[0064] In the object-oriented rocket engine model, fault factors representing the component-level models of each subsystem are constructed and linearized to obtain the discrete state-space model of the rocket engine.

[0065] The process of constructing fault factors is the process of fault factor injection. The following will use pipeline model fault injection and turbine model fault injection as examples for explanation.

[0066] Specifically, after a pipeline leak occurs, some fluid inside the pipeline flows into the atmosphere through the leak orifice. When constructing a fault factor characterizing the pipeline leak in the pipeline model, the leak orifice can be regarded as a throttling element, with the inlet pressure being the pressure inside the pipe at the leak orifice and the outlet pressure being atmospheric pressure. The flow in the front section of the cooling pipeline is described by a dynamic flow equation, and the flow in the rear section of the cooling pipeline and the throttling orifice is described by a static flow resistance equation.

[0067] The flow capacity equation in the flow equation of the front section of the cooling pipe is shown in formula (1):

[0068]

[0069] Where t is time, V is the volume of the cavity, and a is the speed of sound in the cavity. The compressibility is reflected, where p is the pressure inside the pipeline cavity. For inbound traffic, This refers to the outbound flow.

[0070] The flow resistance equation is shown in formula (2):

[0071]

[0072] Where R is the inertial flow resistance coefficient, ξ is the pipe flow resistance coefficient, and p i For the inlet pressure, p e This is due to export pressure.

[0073] The equations for the downstream section of the cooling pipe and the leakage hole are shown in formula (3):

[0074]

[0075] in, C is the flow rate through the leak orifice. d Let A be the flow coefficient, A be the flow area, and ρ be the fluid density.

[0076] When constructing fault factors characterizing health in the turbine model, efficiency offset is used as a representation of the health status of the LRE turbopump subsystem. Therefore, turbopump faults are injected through efficiency fault factors, which include the pump efficiency factor FF. T The turbine efficiency factor is shown in formula (4):

[0077]

[0078] Where, η T For the efficiency of the turbine, η P For pump efficiency, The efficiency of the turbine of a rocket engine under healthy conditions. The efficiency of the pump for the rocket engine under healthy conditions.

[0079] Fault injection modifies the object-oriented rocket engine model to enable it to simulate faults. The principle of fault injection is to inject common parameters that can significantly affect the engine's normal operation and state parameters. Fault injection can be achieved through multiplicative and additive fault methods based on fault mode analysis. Fault injection for other subsystem component-level models will not be elaborated upon.

[0080] The nonlinear discrete dynamic model of the rocket engine after the injection of the fault factor is shown in Equations (5) and (6):

[0081] x k+1 =f(k,u k ,x k ,p k )+e k (5)

[0082] y k =g(k,u k ,x k ,p k )+ε k (6)

[0083] Where k is the time step, and the value of k is 1, 2, 3, ..., x k These are system state parameters. For the oxygen flow rate of the generator, u k The flow coefficient of the flow regulator determines the generator fuel flow rate, y k For measurement parameters, p k For engine failure factors, e k and ε k For an uncorrelated zero-mean white noise sequence, e k The covariance matrix is ​​Q,ε k The covariance matrix is ​​R. k The parameters used to characterize engine health include pipeline leakage, turbopump efficiency factor and turbo efficiency factor, as well as other injected fault factors.

[0084] Linearizing the nonlinear discrete dynamic model of a rocket engine containing a fault factor at a certain reference point yields the discrete state-space model of the rocket engine, as shown in equations (7) and (8):

[0085] Δx k+1 =AΔx k +BΔu k +LΔp k +e k (7)

[0086] Δy k =CΔx k +DΔu k +MΔp k +ε k (8)

[0087] Where Δx = x - x0, Δy = y - y0, Δu = u - u0. x0 is the state parameter of the engine in the steady state, y0 is the measured parameter of the engine in the steady state, and u0 is the control variable of the engine in the steady state; Δx represents the residual of x, and Δy represents the residual of y. A is the state transition matrix, B is the control matrix, C is the output matrix, D is the feedforward matrix, L is the fault matrix of the state parameters, and M is the fault matrix of the measured parameters. I is the standard matrix.

[0088] The fault factors in the discrete state-space model of the rocket engine are expanded to system state parameters to obtain the augmented state-space model of the rocket engine.

[0089] Specifically, for the discrete state-space model of the rocket engine in formulas (7) and (8), the failure factor of the rocket engine is regarded as the system state parameter. Therefore, the failure factor can be expanded to the system state parameter, and the augmented state-space model of the rocket engine is obtained as shown in formulas (11) and (12):

[0090]

[0091] Δx aug,k =A aug Δx aug,k +B aug Δu aug,k +e aug,k

[0092]

[0093] Since this diagnostic method is designed for fault diagnosis in the steady-state region, the control variable Δu k =0 or Δu aug,k =0.

[0094] Based on the augmented state-space model of the rocket engine, an observer is used to observe the fault factors. The observer can be a Kalman filter. The specific process of Kalman filtering is as follows:

[0095] The measurement residuals are input into the augmented state space model of the rocket engine;

[0096] Using Kalman filtering, the state parameters of the augmented state space model of the rocket engine are predicted based on the state equations in the augmented state space model, and the mean and variance of the state parameters are obtained.

[0097] Based on the prediction results and the measurement residuals, the mean and variance of the state parameters are updated to obtain the estimated results of the state parameters; the estimated results include the estimated values ​​of the failure factors.

[0098] Specifically, the prediction equations for the Kalman filter are shown in equations (13), (14), and (15):

[0099]

[0100]

[0101]

[0102] in, Let be the prior of the state parameter Δx at time k+1, and P be the variance of the state parameter. For P aug In the prior at time k+1, A aug,k+1 Let be the Jacobian matrix of the process quantity at time k+1.

[0103] Update: The update equations for the Kalman filter are shown in equations (16) and (17):

[0104]

[0105]

[0106]

[0107] Where I is the standard matrix, K aug For Kalman gain. When K aug,k The smaller the value, the more accurate the predicted state value is to obtain the estimated state value. aug,k The larger the value, the more accurate the true state value becomes, and the more accurate the estimated state value becomes.

[0108] The specific process of prediction and update is as follows: First, given the initial state value Δx... aug,1 (0), Δp1(0) is predicted by formulas (13), (14) and (15) to obtain the state parameter Δx when k=1. aug The prior prediction results of the mean and variance of the state parameter Δp are obtained; then, based on formulas (16), (17) and (18), the state parameter Δx at k=1 is predicted according to the measurement residuals at the corresponding time. aug The state parameter Δx is obtained by updating the prior prediction result of the state parameter Δp and the state parameter Δx when k=1. aug The posterior results of the mean and variance of the state parameter Δp are used to complete one prediction and update process; the estimated results of the mean and variance of the state parameter at that time are obtained based on the posterior results; and the state parameter Δx at k=2 is predicted based on the posterior results at k=1. aug The prior predictions of the mean and variance of the state parameter Δp are obtained; then, based on the measurement residuals at the corresponding time, the state parameter Δx at k=2 is predicted. aug The state parameter Δx is updated by combining the prior prediction of the state parameter Δp with the state parameter Δp to obtain the state parameter Δx when k=2. aug The posterior result of the state parameter Δp, and so on, can be used to iteratively calculate the state parameter Δx at any time. aug The mean and variance of the state parameter Δp are estimated, and the estimated value of the fault factor is confirmed based on the estimated value of the fault factor. The fault condition of the rocket engine is judged based on the estimated value of the fault factor.

[0109] The specific implementation process can be combined with Figure 2 Please provide an explanation, such as Figure 2As shown, firstly, various parameters of the engine are measured. Based on the covariates and the measured values ​​of the target parameters in the measurement data, a probabilistic predictive neural network is used for time series analysis. According to the probability distribution of the predicted values ​​of the target parameters, the upper and lower thresholds of the abnormal threshold range of the target parameters at the corresponding time are determined. Anomaly detection is performed through the abnormal threshold range. Whether a fault has occurred is determined based on whether the measurement results of the target parameters exceed the threshold. If the engine is abnormal, the measured value of the target parameters at the time before the anomaly is input into the autoregressive neural network of the target parameters to obtain the standard value of the target parameters. The measurement residual of the target parameters is obtained based on the standard value of the target parameters and the measured value of the target parameters at the corresponding time. The measurement residual is used as input to the augmented state space model of the rocket engine, which uses the fault factor as the system state parameter. Kalman filtering technology is used to observe the fault factor and obtain the estimated value of the fault factor, thereby realizing the fault diagnosis of the rocket engine.

[0110] Based on the above-mentioned rocket engine fault diagnosis method and specific implementation process, it can be seen that the rocket engine fault diagnosis method provided by this invention can predict the standard value of the target parameter in real time by using a predictive neural network and the measured value of the target parameter. This eliminates the influence of differences between different batches of the same type of engine and state drift during engine operation on the online observation of fault factors. By observing the fault factors through the augmented state space model of the rocket engine, it solves the problems of poor interpretability and poor engineering practicality of existing data-driven methods.

[0111] The above mainly describes the solution provided by the embodiments of the present invention from the perspective of the interaction between various network elements. It is understood that, in order to achieve the above functions, it includes corresponding hardware structures and / or software modules for executing each function. Those skilled in the art should readily recognize that, in conjunction with the units and algorithm steps of the various examples described in the embodiments disclosed herein, the present invention can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0112] The embodiments of the present invention can divide functional modules according to the above method examples. For example, each function can be divided into its own functional module, or two or more functions can be integrated into one processing module. The integrated module can be implemented in hardware or as a software functional module. It should be noted that the module division in the embodiments of the present invention is illustrative and only represents one logical functional division; other division methods may be used in actual implementation.

[0113] When dividing functions into different modules, each with its own function, Figure 3 A schematic diagram of the structure of a rocket engine fault diagnosis device provided by the present invention is shown. Figure 3 As shown, the device includes:

[0114] Engine measurement data acquisition module 301 is used to acquire engine measurement data; the measurement data includes covariates and target parameters;

[0115] The probability distribution calculation module 302 for the predicted value of the target parameter is used to predict the probability distribution of the predicted value of the target parameter based on the measured value of the covariate using a probability prediction neural network.

[0116] The anomaly detection module 303 is used to determine the anomaly threshold range of the target parameter at a corresponding time based on the probability distribution of the predicted value of the target parameter;

[0117] The target parameter standard value determination module 304 is used to determine whether the measured value of the target parameter exceeds the abnormal threshold range. If it does, the measured value of the target parameter at the previous moment is input into the autoregressive neural network of the target parameter to obtain the standard value of the target parameter.

[0118] The measurement residual calculation module 305 is used to obtain the measurement residual of the target parameter based on the standard value of the target parameter and the measured value of the target parameter at the corresponding time.

[0119] The fault diagnosis module 306 is used to observe the fault factors based on the measurement residuals through an observer and the augmented state space model of the rocket engine, obtain the estimated values ​​of the fault factors, and realize the fault diagnosis of the engine; the augmented state space model of the rocket engine is a model with the fault factors as system state parameters.

[0120] Optionally, the device may further include:

[0121] The object-oriented rocket engine model building module is used to build object-oriented rocket engine models.

[0122] The rocket engine discrete state space model building module is used to construct fault factors representing the component-level models of each subsystem in the object-oriented rocket engine model, and to perform linearization processing to obtain the rocket engine discrete state space model.

[0123] The rocket engine augmented state space model establishment module is used to expand the dimension of the fault factors in the discrete state space model of the rocket engine into system state parameters, thereby obtaining the rocket engine augmented state space model.

[0124] Optionally, the fault diagnosis module 306 can be specifically used for:

[0125] The measurement residuals are input into the augmented state space model of the rocket engine;

[0126] Using Kalman filtering, the state parameters of the augmented state space model of the rocket engine are predicted based on the state equations in the augmented state space model, and the mean and variance of the state parameters are obtained.

[0127] Based on the prediction results and the measurement residuals, the mean and variance of the state parameters are updated to obtain the estimated results of the state parameters; the estimated results include the estimated values ​​of the failure factors.

[0128] The failure status of the rocket engine is determined based on the estimated value of the failure factor.

[0129] Optionally, the device may further include a probabilistic prediction neural network establishment module, which can be used for:

[0130] Establish an initial probability prediction neural network;

[0131] The given initial values ​​and the measured values ​​of the target parameter at time n and the covariate at time n+1 in the engine normal test data during the steady state phase are output to the initial probability prediction neural network to obtain the probability distribution of the predicted value of the target parameter at time n+1. The measured values ​​of the target parameter at time n+1 and the covariate at time n+2 are used as the inputs to the initial probability prediction neural network to obtain the probability distribution of the predicted value of the target parameter at time n+2.

[0132] The mean of the probability distribution of the predicted target parameter values ​​is compared with the measured value of the target parameter at the corresponding time. Based on the comparison result, the parameters in the initial probability prediction neural network are adjusted, and the probability prediction neural network is obtained after training.

[0133] Optionally, the device may further include an autoregressive neural network establishment module for target parameters, which can be used to: establish an initial autoregressive neural network;

[0134] The measured value of the target parameter at time n-1 in the normal test data of the same engine model during the steady state phase is input into the initial autoregressive neural network to predict the predicted value of the target parameter at time n.

[0135] The measured value of the target parameter at time n in the normal test data of the same engine model during the steady state phase is compared with the predicted value of the target parameter at time n. The parameters of the initial autoregressive neural network are adjusted according to the comparison results, and the autoregressive neural network of the target parameter is obtained after training is completed, where n is a positive integer.

[0136] Optionally, the subsystem component-level model includes a centrifugal pump model, a turbine model, a pipeline model, a gas generator model, a thrust chamber model, a control system model, and a physical property model.

[0137] Optionally, the observer is a Kalman filter.

[0138] When using integrated units Figure 4 This diagram illustrates the structural block diagram of a rocket engine fault diagnosis device provided by the present invention. Figure 4 As shown, the device includes a communication unit and a processing unit.

[0139] A communication unit / interface is used to acquire engine measurement data; the measurement data includes covariates and target parameters.

[0140] A processing unit / processor is used to predict the probability distribution of the target parameter prediction value using a probabilistic prediction neural network based on the measured value of the covariate.

[0141] The standard value of the target parameter is determined based on the probability distribution and the measured value of the target parameter;

[0142] The measurement residual of the target parameter is obtained based on the standard value and the measured value;

[0143] Based on the measurement residuals, the fault factors are observed by an observer and the augmented state space model of the rocket engine to obtain estimated values ​​of the fault factors, thereby achieving fault diagnosis of the engine; the augmented state space model of the rocket engine is a model with fault factors as system state parameters.

[0144] like Figure 4 As shown, the processor described above can be a general-purpose central processing unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits used to control the execution of the program of the present invention. The communication interface described above can be one or more. The communication interface can use any transceiver-like device for communicating with other devices or communication networks.

[0145] like Figure 4 As shown, the terminal device described above may also include a communication line. The communication line may include a path for transmitting information between the components described above.

[0146] Optional, such as Figure 4As shown, the terminal device may further include a memory. The memory stores computer execution instructions for implementing the present invention, and the execution is controlled by a processor. The processor executes the computer execution instructions stored in the memory, thereby implementing the method provided in the embodiments of the present invention.

[0147] like Figure 4 As shown, the memory can be read-only memory (ROM) or other types of static storage devices capable of storing static information and instructions, random access memory (RAM) or other types of dynamic storage devices capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed discs, laser discs, optical discs, digital universal discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited to these. The memory can exist independently and be connected to the processor via communication lines. The memory can also be integrated with the processor.

[0148] Optionally, the computer execution instructions in the embodiments of the present invention may also be referred to as application code, and the embodiments of the present invention do not specifically limit this.

[0149] In a specific implementation, as one example, such as Figure 4 As shown, a processor may include one or more CPUs, such as Figure 4 CPU0 and CPU1 in the CPU.

[0150] In a specific implementation, as one example, such as Figure 4 As shown, the terminal device may include multiple processors, such as Figure 4 The processors in the system. Each of these processors can be a single-core processor or a multi-core processor.

[0151] The methods disclosed in this invention can be applied to or implemented by a processor. The processor may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method can be completed by integrated logic circuits in the processor's hardware or by instructions in software form. The processor can be a general-purpose processor, a digital signal processor (DSP), an ASIC, a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this invention. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this invention can be directly embodied in the execution of a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory; the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above method.

[0152] On one hand, the present invention provides a computer-readable storage medium storing instructions that, when executed, are used to implement:

[0153] Acquire engine measurement data; the measurement data includes covariates and target parameters;

[0154] Based on the measured values ​​of the covariates, a probability distribution of the predicted values ​​of the target parameters is obtained using a probabilistic prediction neural network.

[0155] The abnormal threshold range of the target parameter at the corresponding time is determined based on the probability distribution of the predicted values ​​of the target parameter;

[0156] Determine whether the measured value of the target parameter exceeds the abnormal threshold range. If it does, input the measured value of the target parameter at the previous time step into the autoregressive neural network of the target parameter to obtain the standard value of the target parameter.

[0157] The measurement residual of the target parameter is obtained by comparing the standard value of the target parameter with the measured value of the target parameter at the corresponding time.

[0158] Based on the measurement residuals, the fault factors are observed by an observer and the augmented state space model of the rocket engine to obtain estimated values ​​of the fault factors, thereby achieving fault diagnosis of the engine; the augmented state space model of the rocket engine is a model with fault factors as system state parameters.

[0159] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer programs or instructions. When the computer program or instructions are loaded and executed on a computer, the processes or functions described in the embodiments of the present invention are performed entirely or partially. The computer can be a general-purpose computer, a special-purpose computer, a computer network, a terminal, a user equipment, or other programmable device. The computer program or instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another. For example, the computer program or instructions can be transferred from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium, such as a floppy disk, hard disk, or magnetic tape; it can also be an optical medium, such as a digital video disc (DVD); or it can be a semiconductor medium, such as a solid-state drive (SSD).

[0160] Although the invention has been described herein in conjunction with various embodiments, those skilled in the art will understand and implement other variations of the disclosed embodiments by reviewing the accompanying drawings, the disclosure, and the appended claims in carrying out the claimed invention. In the claims, the word "comprising" does not exclude other components or steps, and "a" or "an" does not exclude a plurality. A single processor or other unit can implement several functions listed in the claims. While different dependent claims may recite certain measures, this does not mean that these measures cannot be combined to produce good results.

[0161] Although the invention has been described in conjunction with specific features and embodiments, it is obvious that various modifications and combinations can be made therein without departing from the spirit and scope of the invention. Accordingly, this specification and drawings are merely exemplary descriptions of the invention as defined by the appended claims, and are considered to cover any and all modifications, variations, combinations, or equivalents within the scope of the invention. Clearly, those skilled in the art can make various alterations and modifications to the invention without departing from its spirit and scope. Thus, if such modifications and modifications of the invention fall within the scope of the claims and their equivalents, the invention is also intended to include such modifications and modifications.

Claims

1. A method of diagnosing a fault in a rocket engine, characterized by, include: Acquire engine measurement data; the measurement data includes covariates and target parameters; Based on the measured values ​​of the covariates, a probability distribution of the predicted values ​​of the target parameters is obtained using a probabilistic prediction neural network. The abnormal threshold range of the target parameter at the corresponding time is determined based on the probability distribution of the predicted values ​​of the target parameter; Determine whether the measured value of the target parameter exceeds the abnormal threshold range. If it does, input the measured value of the target parameter at the previous time step into the autoregressive neural network of the target parameter to obtain the standard value of the target parameter. The measurement residual of the target parameter is obtained by comparing the standard value of the target parameter with the measured value of the target parameter at the corresponding time. Based on the measurement residuals, the fault factors are observed by an observer and an augmented state-space model of the rocket engine to obtain estimated values ​​of the fault factors, thereby achieving fault diagnosis of the engine; the augmented state-space model of the rocket engine is a model with fault factors as system state parameters. The method of detecting engine faults by observing the fault factors based on the measurement residuals using an observer and the augmented state-space model of the rocket engine, and obtaining estimated values ​​of the fault factors, includes: The measurement residuals are input into the augmented state space model of the rocket engine; Using Kalman filtering, the state parameters of the augmented state space model of the rocket engine are predicted based on the state equations in the augmented state space model, and the mean and variance of the state parameters are obtained. Based on the prediction results and the measurement residuals, the mean and variance of the state parameters are updated to obtain the estimated results of the state parameters; the estimated results include the estimated values ​​of the failure factors. The failure status of the rocket engine is determined based on the estimated values ​​of the failure factors. The augmented state-space model of a rocket engine is shown in the formula: ； ； in, For time steps, The value can be 1, 2, 3... These are system state parameters. A Here is the state transition matrix. B For the control matrix, C For the output matrix, D It is a feedforward matrix. L The fault matrix represents the state parameters. M The fault matrix for the measured parameters, For a standard matrix, The flow coefficient of the flow regulator. For measurement parameters, Engine failure factor, and The sequence is an uncorrelated zero-mean white noise sequence. The covariance matrix is , The covariance matrix is .

2. The rocket engine fault diagnosis method according to claim 1, characterized in that, The step of determining whether the measured value of the target parameter exceeds the abnormal threshold range, and if so, inputting the measured value of the target parameter from the previous time step into the autoregressive neural network of the target parameter to obtain the standard value of the target parameter, further includes: Establish an initial autoregressive neural network; The measured value of the target parameter at time n-1 in the normal test data of the same engine model during the steady state phase is input into the initial autoregressive neural network to predict the predicted value of the target parameter at time n. The measured value of the target parameter at time n in the normal test data of the same engine model during the steady state phase is compared with the predicted value of the target parameter at time n. The parameters of the initial autoregressive neural network are adjusted according to the comparison results, and the autoregressive neural network of the target parameter is obtained after training is completed, where n is a positive integer.

3. The rocket engine fault diagnosis method according to claim 1, characterized in that, Before using the measured values ​​of the covariates to predict the probability distribution of the target parameter prediction values ​​using a probabilistic prediction neural network, the process further includes: Establish an initial probability prediction neural network; The given initial values ​​and the measured values ​​of the target parameter at time n and the covariate at time n+1 in the engine normal test data during the steady state phase are output to the initial probability prediction neural network to obtain the probability distribution of the predicted value of the target parameter at time n+1. The measured values ​​at time n+1 and the measured values ​​of the covariate at time n+2 are used as the input to the initial probability prediction neural network to obtain the probability distribution of the predicted value of the target parameter at time n+2. The mean of the probability distribution of the predicted target parameter values ​​is compared with the measured value of the target parameter at the corresponding time. Based on the comparison result, the parameters in the initial probability prediction neural network are adjusted, and the probability prediction neural network is obtained after training.

4. The rocket engine fault diagnosis method according to claim 1, characterized in that, Before the step of observing the fault factors based on the measurement residuals using an observer and an augmented state-space model of the rocket engine to obtain estimated values ​​of the fault factors and thus diagnose engine faults, the following steps are also included: Establish an object-oriented rocket engine model; In the object-oriented rocket engine model, fault factors representing the component-level models of each subsystem are constructed and linearized to obtain a discrete state-space model of the rocket engine. The fault factors in the discrete state-space model of the rocket engine are then expanded to system state parameters to obtain an augmented state-space model of the rocket engine.

5. The rocket engine fault diagnosis method according to claim 1, characterized in that, The observer is a Kalman filter.

6. The rocket engine fault diagnosis method according to claim 4, characterized in that, The subsystem component-level models include centrifugal pump models, turbine models, pipeline models, gas generator models, thrust chamber models, control system models, and physical property models.

7. A rocket engine fault diagnosis device, characterized in that, The apparatus used in the rocket engine fault diagnosis method according to any one of claims 1-6 includes: An engine measurement data acquisition module is used to acquire engine measurement data; the measurement data includes covariates and target parameters. The probability distribution calculation module for the predicted value of the target parameter is used to predict the probability distribution of the predicted value of the target parameter based on the measured value of the covariate using a probability prediction neural network. The anomaly detection module is used to determine the anomaly threshold range of the target parameter at a corresponding time based on the probability distribution of the predicted value of the target parameter; The target parameter standard value determination module is used to determine whether the measured value of the target parameter exceeds the abnormal threshold range. If it does, the measured value of the target parameter at the previous moment is input into the autoregressive neural network of the target parameter to obtain the standard value of the target parameter. The measurement residual calculation module is used to obtain the measurement residual of the target parameter based on the standard value of the target parameter and the measured value of the target parameter at the corresponding time. The fault diagnosis module is used to observe the fault factors based on the measurement residuals through an observer and the augmented state space model of the rocket engine, obtain the estimated values ​​of the fault factors, and realize the fault diagnosis of the engine; the augmented state space model of the rocket engine is a model with the fault factors as system state parameters.

8. A rocket engine fault diagnosis device, characterized in that, It includes a processor and a communication interface coupled to the processor; the processor is used to run computer program instructions to implement the rocket engine fault diagnosis method as described in any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, include: The computer-readable storage medium stores instructions that, when executed, implement the rocket engine fault diagnosis method according to any one of claims 1-6.

Images