Carrier rocket engine fault diagnosis method suitable for strong wind area

Abstract

The invention provides a carrier rocket engine fault diagnosis method suitable for a strong wind area, and the method comprises the steps: employing a deep neural network to estimate the high-altitude wind borne by a carrier rocket according to the flight state of the carrier rocket; according to the estimated high-altitude wind information, a deep neural network is adopted to forecast high-altitude wind at the rocket fault diagnosis moment; according to the forecasted high-altitude wind information, the flight state quantity change of the carrier rocket under the influence of the high-altitude wind is estimated; taking the flight state quantity change of the carrier rocket under the influence of the high-altitude wind as a compensation value to obtain the flight state change without the influence of the high-altitude wind; and taking the flight state change without influence of high-altitude wind as a diagnosis factor, and performing engine fault diagnosis by adopting a deep neural network. The flight state deviation of the carrier rocket under the influence of the high-altitude wind is estimated, and the problems that the influence of the high-altitude wind on rocket flight is difficult to estimate and the effect of the high-altitude wind is difficult to consider in a fault diagnosis system in the prior art are solved.

Description

Technical Field

[0001] The invention belongs to the technical field of launch vehicle fault diagnosis, and in particular relates to a launch vehicle engine fault diagnosis method suitable for strong wind areas. Background Art

[0002] Rocket engines are the power source of launch vehicles. They have complex working principles and harsh working environments, which pose a risk of failure. Rocket engine failures leading to a drop in the thrust of launch vehicles is one of the important reasons for launch vehicle failures. New launch vehicles in various countries around the world all contain a large number of engines, which increases the probability of engine failure.

[0003] The dynamic model of a launch vehicle is a complex time-varying nonlinear differential equation. When the input parameters are known, the flight state can be obtained through numerical simulation. However, fault diagnosis is the inverse process of identifying the input from the flight state change, which is difficult to achieve. In addition, the rocket will be affected by high-altitude wind during flight, and its impact on the movement of the launch vehicle is similar to that of the fault condition. How to compensate for the high-altitude wind interference that is difficult to measure accurately is the key to improving the accuracy of fault diagnosis. Among the existing methods for fault diagnosis of power systems based on flight dynamics, most of the research has been conducted on the vacuum flight stage without the influence of aerodynamic forces. In the research considering aerodynamic forces, no mention is made on how to improve the accuracy of fault diagnosis when there is the influence of high-altitude wind. Relevant research is needed. Summary of the invention

[0004] In order to overcome the deficiencies in the prior art, the inventors have conducted intensive research and provided a method for diagnosing launch vehicle engine faults suitable for windy areas, which estimates the deviation of the launch vehicle's flight state under the influence of high-altitude wind, so as to solve the problems in the prior art that it is difficult to estimate the impact of high-altitude wind on rocket flight and it is difficult to consider the effect of high-altitude wind in the fault diagnosis system.

[0005] The technical solution provided by the present invention is as follows:

[0006] In a first aspect, a method for diagnosing a launch vehicle engine fault applicable to a high wind area comprises:

[0007] Based on the flight status of the launch vehicle, a deep neural network is used to estimate the high-altitude wind it is subject to;

[0008] Based on the estimated high-altitude wind information, a deep neural network is used to predict the high-altitude wind that the rocket is subject to during fault diagnosis.

[0009] According to the predicted high-altitude wind information, estimate the change of the launch vehicle's flight state under the influence of high-altitude wind;

[0010] The change of the flight state quantity of the launch vehicle under the influence of high-altitude wind is taken as the compensation value, and the compensation value is subtracted from the actual flight state quantity to obtain the flight state change without the influence of high-altitude wind;

[0011] The flight status changes without the influence of high-altitude wind are taken as diagnostic factors, and deep neural networks are used for engine fault diagnosis.

[0012] In combination with the first aspect, in the step of estimating the high-altitude winds to which the launch vehicle is subjected using a deep neural network based on its flight status, the input of the deep neural network is a time-correlated sequence including altitude, pitch angle deviation, yaw angle deviation, normal apparent acceleration deviation, and lateral apparent acceleration deviation.

[0013] In combination with the first aspect, in the step of estimating the high-altitude wind received by the launch vehicle using a deep neural network according to the flight state of the launch vehicle, the output of the deep neural network includes a highly correlated sequence of high-altitude wind speed information:

[0014]

[0015] The subscript WE is the abbreviation of wind estimation, which means wind estimation, ΔV Lx , ΔV Ly are the actual flight speed of the rocket V L With the flight speed V on the standard trajectory L * The difference between the components along the x and y axes in the local horizontal coordinate system; V Wx 、V Wy are the velocity components of the high-altitude wind along the x and y axes in the rectangular coordinate system, respectively.

[0016] In combination with the first aspect, in the step of using a deep neural network to predict the high-altitude wind encountered by the rocket at the time of fault diagnosis based on the estimated high-altitude wind information, the input of the deep neural network is a time-correlated sequence including altitude, high-altitude wind speed in a polar coordinate system, and high-altitude wind direction in a polar coordinate system.

[0017] In combination with the first aspect, in the step of using a deep neural network to predict the high-altitude wind that the rocket will be subject to at the time of fault diagnosis based on the estimated high-altitude wind information, the output of the deep neural network is the high-altitude wind information at the time of fault diagnosis after the end moment of the input high-altitude wind sequence, including the high-altitude wind speed in the polar coordinate system and the high-altitude wind direction in the polar coordinate system.

[0018] In combination with the first aspect, in the step of estimating the change in the flight state quantity of the carrier rocket under the influence of high-altitude wind based on the forecasted high-altitude wind information, a deep neural network is used to estimate the change in the flight state quantity of the carrier rocket under the influence of high-altitude wind; the input of the deep neural network is a time-correlation sequence containing altitude and high-altitude wind speed information, the starting point of the time-correlation sequence is consistent with the starting time of the input sequence during high-altitude wind estimation or is within the time period of the input sequence during high-altitude wind estimation, and the end point of the time-correlation sequence is the moment of fault diagnosis.

[0019] In combination with the first aspect, in the step of estimating the changes in the flight state quantities of the carrier rocket under the influence of high-altitude wind based on the predicted high-altitude wind information, the output of the deep neural network is a time-correlation sequence including pitch angle deviation, yaw angle deviation, normal apparent acceleration deviation, and lateral apparent acceleration deviation, and the starting and ending points of the time-correlation sequence are consistent with the starting and ending points of the input time-correlation sequence.

[0020] In combination with the first aspect, in the step of using a deep neural network to diagnose engine faults using the flight state change excluding the influence of high-altitude wind as a diagnostic factor, the input of the deep neural network is a time-correlated sequence including the rocket roll angle deviation, flight time, the residual thrust coefficient of the engine, the actual lateral apparent acceleration deviation of the rocket minus the lateral apparent acceleration deviation caused by high-altitude wind, the actual normal apparent acceleration deviation of the rocket minus the normal apparent acceleration deviation caused by high-altitude wind, the actual pitch angle deviation of the rocket minus the pitch angle deviation caused by high-altitude wind, and the actual yaw angle deviation of the rocket minus the yaw angle deviation caused by high-altitude wind. The starting point of the time correlation sequence is consistent with the starting time of the input sequence when the high-altitude wind is estimated or is within the time period of the input sequence when the high-altitude wind is estimated, and the end point of the time correlation sequence is the fault diagnosis time.

[0021] In combination with the first aspect, the residual thrust coefficient of the engine is determined by a deep neural network, and the input of the deep neural network includes: the height of the rocket at the time of fault diagnosis, the flight speed, the apparent acceleration in the x-direction in the rocket body coordinate system, and the ratio of the actual apparent acceleration of the rocket in the x-direction in the rocket body coordinate system to the apparent acceleration in the standard trajectory.

[0022] In a second aspect, a launch vehicle engine fault diagnosis device suitable for use in high wind areas comprises:

[0023] one or more processors;

[0024] a storage device for storing one or more programs,

[0025] When the one or more programs are executed by the one or more processors, the one or more processors implement the launch vehicle engine fault diagnosis method suitable for high wind areas described in the first aspect.

[0026] In a third aspect, a readable storage medium stores a computer program, which, when executed by a processor, implements the method for diagnosing faults of a launch vehicle engine suitable for windy areas as described in the first aspect.

[0027] In a fourth aspect, a computer program product is provided, comprising: a computer program (also referred to as code, or instruction), which, when executed, executes the method for diagnosing launch vehicle engine faults applicable to high wind areas as described in the first aspect.

[0028] A method for diagnosing a fault of a launch vehicle engine applicable to a high wind area provided by the present invention has the following beneficial effects:

[0029] (1) The present invention provides a method for fault diagnosis of launch vehicle engines suitable for strong wind areas. It uses the "end-to-end" characteristics of deep learning to decouple the fault diagnosis problem under strong high-altitude wind interference into sub-problems such as high-altitude wind prediction and compensation, fault location diagnosis, and residual thrust ratio diagnosis, thereby improving the accuracy of fault diagnosis under high-altitude wind interference;

[0030] (2) The present invention provides a method for diagnosing faults in launch vehicle engines suitable for windy areas. The method uses a deep learning method to establish a mapping relationship between high-altitude wind and changes in the flight state of the launch vehicle, which provides more accurate input for fault diagnosis and improves the accuracy of fault diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 The present invention is a flow chart of a method for diagnosing a launch vehicle engine fault applicable to windy areas;

[0032] Figure 2 The "encoder-decoder" structure for the upper wind disturbance state deviation;

[0033] Figure 3 This is the engine fault diagnosis flow chart;

[0034] Figure 4 When k = 0, the engine fault location diagnosis result;

[0035] Figure 5 When k = 0.4, the engine fault location diagnosis result;

[0036] Figure 6 Engine fault location diagnosis results when k=0.8. DETAILED DESCRIPTION

[0037] The following detailed description of the present invention will make the features and advantages of the present invention more clear and explicit.

[0038] The word “exemplary” is used exclusively herein to mean “serving as an example, example, or illustration.” Any embodiment described herein as “exemplary” is not necessarily to be construed as preferred or advantageous over other embodiments.

[0039] The present invention provides a method for diagnosing faults of a launch vehicle engine suitable for use in strong wind areas. Figure 1 As shown, the following steps are included:

[0040] Step 1: Estimate and forecast the high-altitude wind during the launch vehicle flight.

[0041] The high-altitude winds of the launch vehicle are predicted using deep learning methods. The method is to first estimate the high-altitude winds that the launch vehicle has been exposed to based on its flight status over a period of time, and then predict the high-altitude winds that the rocket will be exposed to in the future based on the estimated high-altitude winds. Different deep neural networks are used to estimate and predict the high-altitude winds.

[0042] Step 1.1, upper air wind estimation

[0043] High-altitude wind estimation uses a deep neural network, and the input of the deep neural network includes pitch angle deviation, yaw angle deviation, normal apparent acceleration deviation, and lateral apparent acceleration deviation. The response process of the rocket to wind interference is a dynamic process, and the change of flight state has serial correlation. The high-altitude wind is identified by the flight state deviation over a period of time. Since the high-altitude wind changes with altitude, the altitude H is added to the input to represent the sequence information. The input design of the deep neural network is:

[0044]

[0045] The subscript WE is the abbreviation of wind estimation, t0 is the starting time of each state quantity in the sequence, and t f is the end time. H is the height, is the pitch angle deviation, Δψ is the yaw angle deviation, Δa wy is the apparent acceleration deviation in the y direction (normal direction) in the rocket body coordinate system, Δa wz It is the apparent acceleration deviation in the z direction (lateral) in the rocket body coordinate system.

[0046] The output of the deep neural network should be the high-altitude wind information, which is expressed in rectangular coordinate form and the deviation of the rocket's flight speed in the horizontal plane is subtracted as the output:

[0047]

[0048] Where, ΔV Lx , ΔV Ly are the actual flight speed of the rocket V LWith the flight speed V on the standard trajectory L * The difference between the components along the x and y axes in the local horizontal coordinate system; V Wx 、V Wy are the velocity components of the high-altitude wind along the x and y axes in the rectangular coordinate system, respectively.

[0049] The wind estimation deep neural network structure is preferably but not limited to the structure in Table 1:

[0050] Table 1. Deep neural network structure for high-altitude wind estimation

[0051]

[0052] Among them, FCNN represents fully connected neural network, LSTM represents long short-term memory neural network, and elu is the activation function.

[0053] Step 1.2, Upper Air Wind Forecast

[0054] The input of the high-altitude wind forecast deep neural network is the high-altitude wind change within a certain altitude range and the corresponding altitude, and the output is the high-altitude wind at a certain altitude in the future. The real wind field can be used directly as input during training, and the online estimated high-altitude wind can be used as input in simulation and flight.

[0055] The input design is:

[0056]

[0057] The subscript WP is the abbreviation of wind prediction, which refers to wind forecast. f are the time points corresponding to the starting and ending heights in the H sequence during the rocket flight. W is the wind speed of the upper air in the polar coordinate system; A W is the wind direction of the upper air in the polar coordinate system.

[0058] The output of the deep neural network is the terminal time t of the input high-altitude wind sequence f The high-altitude wind after Δt is:

[0059]

[0060] The structure of the high-altitude wind forecast deep neural network is preferably but not limited to the structure in Table 2:

[0061] Table 2 Structure of deep neural network for high-altitude wind forecast

[0062]

[0063] Step 2: Determination of the change in the rocket flight state under the influence of high-altitude wind

[0064] High-altitude wind interference is external interference. The future evolution of high-altitude wind has nothing to do with the movement of the rocket, but the influence of high-altitude wind will be reflected in the movement of the rocket. When compensating for high-altitude wind interference, the existing high-altitude wind should be estimated first, and then the high-altitude wind at the time of fault diagnosis should be predicted by the estimated high-altitude wind result. Finally, the flight state deviation caused by the predicted and estimated high-altitude wind solution is used to compensate for the fault diagnosis. According to the characteristics of the high-altitude wind interference compensation algorithm, the high-altitude wind is used as the "code" to design an "encoder-decoder". Its structure is as follows: Figure 2 As shown, the high-altitude wind estimation and forecast encoder is a high-altitude wind estimation and high-altitude wind forecast deep neural network. FDD As the time for fault diagnosis, at t0-t fault No fault occurs during the time period t0-t fault The deviation of the flight state quantity within the time period is used as the input of the high-altitude wind encoder to calculate t0-t fault The high-altitude wind in the time period is then predicted using the high-altitude wind forecast deep neural network. FDD The momentary upper wind and t0-t fault The high-altitude winds in the time period are spliced ​​as the input of the flight state deviation decoder. Finally, the decoder obtains t0-t FDD Flight status deviation within a time period.

[0065] The inputs used to diagnose the position of the faulty engine include apparent acceleration, attitude, swing angle, flight time, and remaining thrust, among which apparent acceleration and attitude angle are flight state deviations that need to be compensated. The input of the state deviation decoder is the high-altitude wind sequence including altitude, x-axis wind speed, and y-axis wind speed:

[0066]

[0067] The output of the state deviation decoder is the flight state deviation value corresponding to the input high-altitude wind, which can be expressed as:

[0068]

[0069] in, is the pitch angle deviation caused by high-altitude wind, Δψ is the yaw angle deviation caused by high-altitude wind, Δa wy is the apparent acceleration deviation in the y direction caused by the high-altitude wind in the rocket body coordinate system, Δa wz It is the apparent acceleration deviation in the z direction caused by the high-altitude wind in the rocket body coordinate system.

[0070] The structure of the state deviation decoder (deep neural network) is preferably but not limited to the structure in Table 3:

[0071] Table 3 State deviation decoder structure

[0072]

[0073] Step 3, Engine Fault Diagnosis

[0074] Step 3.1 Engine Residual Thrust Coefficient Diagnosis

[0075] The residual thrust coefficient of the engine is diagnosed by a deep neural network method. The input of the deep neural network is:

[0076] I k =[HV a wx ρ] T

[0077] Among them, H is the current altitude of the rocket, V is the current flight speed of the rocket, and a wx is the corrected apparent acceleration in the x-direction in the rocket body coordinate system, and ρ is the ratio of the actual apparent acceleration of the rocket in the x-direction in the rocket body coordinate system to the apparent acceleration in the standard trajectory.

[0078] The output of the deep neural network is the residual thrust coefficient k.

[0079] The structure of the deep neural network is preferably but not limited to Table 4:

[0080] Table 4 Residual thrust coefficient diagnosis deep neural network structure

[0081]

[0082] Among them, Sigmoid is the activation function.

[0083] Step 3.2 Engine fault location diagnosis

[0084] The engine fault location is diagnosed through a deep neural network, and the input of the deep neural network is:

[0085]

[0086] Where Δγ is the roll angle deviation, t is the flight time, k is the residual thrust coefficient, Δa' wy Δa' is the value obtained by deducting the apparent acceleration deviation in the y direction caused by the high-altitude wind from the actual y-direction apparent acceleration deviation of the rocket in the rocket body coordinate system. wz It is the value obtained by deducting the apparent acceleration deviation in the z direction caused by the high-altitude wind from the actual apparent acceleration deviation in the z direction of the rocket in the rocket body coordinate system. It is the actual pitch angle deviation of the rocket minus the pitch angle deviation caused by the high-altitude wind, and Δψ' is the actual yaw angle deviation of the rocket minus the yaw angle deviation caused by the high-altitude wind.

[0087] The output is the engine fault position n.

[0088] The deep neural network structure is preferably but not limited to Table 5:

[0089] Table 5 Engine fault location diagnosis deep neural network structure

[0090]

[0091] Fault diagnosis flowchart see Figure 3 , where NET k Represents the residual thrust coefficient diagnostic deep neural network, NET n Represents the deep neural network for engine fault location diagnosis.

[0092] A simulation of a launch vehicle with 7 engines was conducted, and the engine fault location diagnosis results are as follows: Figures 4 to 6 As shown in the figure. Without interference compensation, the engine fault location diagnosis result did not change with the change of the implanted fault condition, and it was always the No. 2 engine fault. The high-altitude wind interference had a great impact on the rocket flight state, and the impact was greater than the engine fault. After interference compensation, the diagnosis result of the engine fault location was significantly improved. In each fault condition, accurate engine fault location diagnosis results could be obtained after a certain diagnosis delay.

[0093] The present invention also provides a launch vehicle engine fault diagnosis device suitable for strong wind areas, comprising:

[0094] one or more processors;

[0095] a storage device for storing one or more programs,

[0096] When the one or more programs are executed by the one or more processors, the one or more processors implement the launch vehicle engine fault diagnosis method suitable for high wind areas described in the first aspect.

[0097] The present invention also provides a readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the launch vehicle engine fault diagnosis method applicable to high wind areas as described in the first aspect.

[0098] The present invention also provides a computer program product, which includes: a computer program (also referred to as code, or instruction), which, when executed, executes the launch vehicle engine fault diagnosis method applicable to high wind areas described in the first aspect.

[0099] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions can be transmitted from a website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (digital subscriber line, DSL) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, server or data center.

[0100] Those of ordinary skill in the art will appreciate that the devices and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0101] The present invention has been described in detail above in conjunction with specific implementations and exemplary examples, but these descriptions cannot be understood as limiting the present invention. Those skilled in the art understand that, without departing from the spirit and scope of the present invention, a variety of equivalent substitutions, modifications or improvements may be made to the technical solution of the present invention and its implementation methods, all of which fall within the scope of the present invention. The scope of protection of the present invention shall be subject to the attached claims.

[0102] The contents not described in detail in the specification of the present invention belong to the common knowledge of those skilled in the art.

Claims

1. A method for diagnosing launch vehicle engine faults suitable for use in high wind areas, characterized in that: include: Based on the flight status of the launch vehicle, a deep neural network is used to estimate the high-altitude wind it is subject to; Based on the estimated high-altitude wind information, a deep neural network is used to predict the high-altitude wind that the rocket is subject to during fault diagnosis. According to the predicted high-altitude wind information, estimate the change of the launch vehicle's flight state under the influence of high-altitude wind; The change of the flight state quantity of the launch vehicle under the influence of high-altitude wind is taken as the compensation value, and the compensation value is subtracted from the actual flight state quantity to obtain the flight state change without the influence of high-altitude wind; The flight status changes without the influence of high-altitude wind are taken as diagnostic factors, and deep neural networks are used for engine fault diagnosis.

2. The method for diagnosing a launch vehicle engine fault in a windy area according to claim 1, characterized in that: In the step of estimating the high-altitude wind to which the launch vehicle is subjected using a deep neural network according to the flight status of the launch vehicle, the input of the deep neural network is a time-correlated sequence including altitude, pitch angle deviation, yaw angle deviation, normal apparent acceleration deviation, and lateral apparent acceleration deviation.

3. The method for diagnosing a launch vehicle engine fault suitable for use in a high wind area according to claim 1, characterized in that: In the step of estimating the high-altitude wind received by the launch vehicle using a deep neural network according to the flight state of the launch vehicle, the output of the deep neural network includes a highly correlated sequence of high-altitude wind speed information: The subscript WE is the abbreviation of wind estimation, which means wind estimation, ΔV Lx , ΔV Ly are the actual flight speed of the rocket V L With the flight speed V on the standard trajectory L * The difference between the components along the x and y axes in the local horizontal coordinate system; V Wx 、V Wy are the velocity components of the high-altitude wind along the x and y axes in the rectangular coordinate system, respectively.

4. The method for diagnosing a launch vehicle engine fault suitable for use in a high wind area according to claim 1, characterized in that: In the step of using a deep neural network to predict the high-altitude wind that the rocket will experience at the time of fault diagnosis based on the estimated high-altitude wind information, the input of the deep neural network is a time-correlated sequence including altitude, high-altitude wind speed in a polar coordinate system, and high-altitude wind direction in a polar coordinate system.

5. The method for diagnosing a launch vehicle engine fault suitable for use in a high wind area according to claim 1, characterized in that: In the step of using a deep neural network to predict the high-altitude wind that the rocket will experience at the time of fault diagnosis based on the estimated high-altitude wind information, the output of the deep neural network is the high-altitude wind information at the time of fault diagnosis, including the high-altitude wind speed in the polar coordinate system and the high-altitude wind direction in the polar coordinate system.

6. The method for diagnosing a launch vehicle engine fault suitable for use in a high wind area according to claim 1, characterized in that: In the step of estimating the change in the flight state quantity of the carrier rocket under the influence of the high-altitude wind based on the high-altitude wind information obtained by forecast, a deep neural network is used to estimate the change in the flight state quantity of the carrier rocket under the influence of the high-altitude wind; the input of the deep neural network is a time correlation sequence containing altitude and high-altitude wind speed information, the starting point of the time correlation sequence is consistent with the starting time of the input sequence during the high-altitude wind estimation or is within the time period of the input sequence during the high-altitude wind estimation, and the end point of the time correlation sequence is the fault diagnosis time.

7. The method for diagnosing a launch vehicle engine fault suitable for use in a strong wind region according to claim 6, characterized in that: In the step of estimating the change in the flight state quantity of the carrier rocket under the influence of high-altitude wind based on the predicted high-altitude wind information, the output of the deep neural network is a time-correlation sequence including pitch angle deviation, yaw angle deviation, normal apparent acceleration deviation, and lateral apparent acceleration deviation, and the starting and ending points of the time-correlation sequence are consistent with the starting and ending points of the input time-correlation sequence.

8. The method for diagnosing a launch vehicle engine fault suitable for use in a strong wind region according to claim 1, characterized in that: In the step of using the flight state change without the influence of high-altitude wind as a diagnostic factor and using a deep neural network to diagnose engine faults, the input of the deep neural network is a time-correlated sequence including the rocket roll angle deviation, flight time, the residual thrust coefficient of the engine, the value of the rocket's actual lateral apparent acceleration deviation minus the lateral apparent acceleration deviation caused by the high-altitude wind, the value of the rocket's actual normal apparent acceleration deviation minus the normal apparent acceleration deviation caused by the high-altitude wind, the rocket's actual pitch angle deviation minus the pitch angle deviation caused by the high-altitude wind, and the rocket's actual yaw angle deviation minus the yaw angle deviation caused by the high-altitude wind. The starting point of the time correlation sequence is consistent with the starting time of the input sequence when estimating the high-altitude wind or is within the time period of the input sequence when estimating the high-altitude wind, and the end point of the time correlation sequence is the fault diagnosis time.

9. The method for diagnosing a launch vehicle engine fault suitable for use in a strong wind region according to claim 8, characterized in that: The residual thrust coefficient of the engine is determined by a deep neural network, and the input of the deep neural network includes: the height of the rocket at the time of fault diagnosis, the flight speed, the apparent acceleration in the x direction in the rocket body coordinate system, and the ratio of the actual apparent acceleration of the rocket in the x direction in the rocket body coordinate system to the apparent acceleration in the standard trajectory.

10. A computer program product, characterized in that The computer program product comprises: a computer program, which, when being executed, executes the method for diagnosing faults of a launch vehicle engine applicable to a high wind region as claimed in any one of claims 1 to 9.

Images