Intelligent fault diagnosis method for thruster of on-orbit service aircraft

Abstract

The invention discloses an intelligent fault diagnosis method for a thruster of an on-orbit service aircraft, and the method comprises the following steps: building a mathematical model related to the control of the aircraft, and configuring the thruster of the aircraft; making a fault data set, wherein the fault data set comprises the output of a control system under various fault states in the rendezvous and docking process of the aircraft; a liquid neural network is constructed and trained, the liquid neural network comprises an input layer, a liquid layer and a classifier, and the input layer is used for integrating original time sequence data of a control system of the aircraft into a unified input vector; the liquid layer comprises N neurons and is used for carrying out dynamic modeling and time sequence feature extraction on an input sequence; the classifier is used for mapping to a fault category probability through an output layer; the real-time data output by the control system of the aircraft is obtained, the pre-trained liquid neural network is applied to perform fault diagnosis, and the method can perform fault diagnosis in real time.

Description

Technical Field

[0001] The invention relates to the technical field of spacecraft guidance, navigation and control, and in particular to an intelligent fault diagnosis method for an on-orbit service aircraft thruster. Background Art

[0002] The on-orbit service and maintenance system for spacecraft, a major national science and technology project, has entered the development stage. The aircraft is required to have the ability to transport and hand over supplies and can be used multiple times with heavy loads. The on-orbit service aircraft has the ability to control relative position and relative attitude in six degrees of freedom. This capability requires the configuration of several thrusters as actuators for real-time control. Thruster failures account for a relatively high proportion of the types of failures on the aircraft platform. Their failures have an impact on the success or failure of rendezvous and docking, and the implementation of on-orbit control, and even safety issues. When individual thrusters fail, the thrusters configured in the system generally have redundancy. The mission could have been completed by changing the thruster usage strategy, but they often exit the mission because they cannot perform fault isolation and system reconstruction. Therefore, real-time fault diagnosis of on-orbit service aircraft is an urgent problem to be solved.

[0003] To solve the above technical problems, a Chinese patent discloses a method for managing thruster redundancy mode, fault shielding and autonomous recovery CN112083645B, which provides a method for autonomously selecting and switching corresponding thruster modes according to thruster classification to perform fault diagnosis, including: setting the thruster switching sequence and determining the switching principle; setting the attitude control thruster autonomous switching polling and the number of switching rounds according to the thrust sequence from small to large, main to backup. This is a method based on autonomous testing, but it is impossible to perform real-time fault diagnosis and processing when the propulsion system is working.

[0004] Therefore, there is an urgent need for a method for real-time fault diagnosis of an on-orbit service vehicle thruster. Summary of the invention

[0005] The purpose of the present invention is to overcome the shortcomings of the prior art and propose an intelligent fault diagnosis method for thrusters of an on-orbit service aircraft, which can perform fault diagnosis of thrusters of an on-orbit service aircraft in real time. This method greatly enhances the reliability of the on-orbit service aircraft in relative motion control for rendezvous and docking and space manipulation, ensures its safe implementation of space missions and has strong autonomy.

[0006] In order to achieve the above object, the technical solution specifically adopted by the present invention is as follows:

[0007] An intelligent fault diagnosis method for an on-orbit servicing vehicle thruster comprises the following steps:

[0008] Step 1: Establish a mathematical model related to the control of the aircraft and configure the propeller of the aircraft, wherein the mathematical model includes the dynamic equation of the aircraft and the dynamic method of the flight environment, and the configuration includes the nominal thrust, thrust direction and installation position;

[0009] Step 2: Create a fault data set, where the fault data set includes the output of the control system under various fault states during the rendezvous and docking process of the aircraft;

[0010] Step 3: construct and train a liquid neural network, wherein the liquid neural network includes an input layer, a liquid layer, and a classifier. The input layer is used to integrate the original time series data of the control system of the aircraft into a unified input vector; the liquid layer includes N neurons, and the parameters of the neurons change in real time with the input and state to adaptively capture the time series features; the classifier is used to map to the fault category probability through the output layer;

[0011] Step 4: Obtain real-time data output by the aircraft's control system and apply the pre-trained liquid neural network for fault diagnosis.

[0012] Preferably, in step 2, when preparing the fault data set, fault conditions need to be set, and the fault conditions include the thruster's propulsion valve being unable to open, the propulsion valve being normally open and unable to be closed, thrust reduction, and thrust leakage.

[0013] Preferably, in step 2, when a fault condition is set, the thrust decreases by 30% and 50% respectively, and the thrust leakage leaks by 30% and 50% respectively.

[0014] Preferably, in step 2, the output of the control system under various fault conditions during the rendezvous and docking process of the aircraft includes three-axis relative attitude measurement data, three-axis relative position measurement data, three-axis relative attitude controller output and three-axis relative position controller output data.

[0015] Preferably, the method for extracting the timing characteristics of the liquid layer is: Represents the hidden state at time t. The input vector is projected into the hidden state space through the weight matrix, and the state at the previous moment is projected into the hidden state space through the hidden state weight matrix. Introducing the update of the current moment captures the dynamic continuity between states; as shown in the formula As shown, is the state at the previous moment Based on this, add a time interval and time constant The relevant adjustments and , integrating the current input with historical information to obtain a new state change. Contains the temporal characteristics of the input sequence.

[0016] Preferably, when the classifier predicts the probability of the fault category, the hidden state of the final time step T obtained by the liquid layer is Mapped to fault category probability through a fully connected layer.

[0017] Preferably, it is characterized in that in step 3, when training the liquid neural network, a cross entropy loss function and a weight decay function are used, and the parameters are updated by back propagation and gradient descent method.

[0018] Preferably, in step 3, a threshold is set, and when the failure probability predicted by the liquid neural network reaches the set threshold, the parameters are fixed.

[0019] Preferably, in step 4, the method of applying the pre-trained liquid neural network to perform real-time fault diagnosis is as follows: when the liquid neural network detects a thruster fault and N consecutive beats are all determined to be a certain fault, the faulty thruster is isolated and the thrusters are reallocated for control; because the fault recognition probability cannot reach 100%, there is still a probability of undetected faults, and the judgment of the safety corridor is introduced, and evacuation is carried out after exceeding the safety corridor, and the rendezvous and docking mission is abandoned.

[0020] As a preference, the safety corridor is an asymmetric axis with respect to the docking axis of the target aircraft for rendezvous and docking with the aircraft, and the half-angle is =8° cone, the apex of the cone in the docking corridor takes the origin of the target spacecraft docking surface coordinate system as the reference point. If the relative position exceeds the corridor area or the relative attitude is greater than 8 degrees, it is considered an abnormality and evacuation control is required to abandon the rendezvous and docking or space control mission.

[0021] The present invention has the following characteristics and beneficial effects:

[0022] By adopting the above technical scheme, the present invention proposes a fault diagnosis method based on liquid neural network, which can perform real-time fault diagnosis of thrusters in the process of rendezvous and docking of on-orbit service aircraft or space control. When a fault occurs, it can detect the fault in real time and isolate the faulty thruster, and reallocate the use strategy of the thruster. The experimental results show that the fault detection accuracy is very high and has excellent fault diagnosis performance. This improves the reliability of rendezvous and docking and space control tasks, and in order to eliminate a small number of undetected faults, the safety corridor is still used to judge flight anomalies, and evacuation control is performed when the safety corridor is exceeded to ensure the safety of the aircraft. Compared with traditional deep learning models (such as CNN, RNN, Transformer, etc.), the use of liquid neural networks shows advantages in dynamic adaptability, computational efficiency, robustness, etc. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 A schematic diagram of the distribution of the thrusters of the on-orbit servicing vehicle in an embodiment of the present invention;

[0024] Figure 2 A schematic diagram of a rendezvous track in an embodiment of the present invention;

[0025] Figure 3 A schematic diagram of a closed loop system in an embodiment of the present invention;

[0026] Figure 4 A schematic diagram of a liquid neural network in an embodiment of the present invention;

[0027] Figure 5 It is a diagram showing the effect of a specific implementation method of the present invention. DETAILED DESCRIPTION

[0028] The present invention is described in detail below in conjunction with specific embodiments. The following embodiments will help those skilled in the art to further understand the present invention, but are not intended to limit the present invention in any form. It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other.

[0029] Example 1

[0030] This embodiment provides an intelligent fault diagnosis method for thrusters of an on-orbit servicing aircraft. Figure 3 As shown, the following steps are included:

[0031] Step 1: Establish a mathematical model related to the control of the aircraft and configure the propeller of the aircraft. The mathematical model includes the dynamic equations of the aircraft and the dynamic method of the flight environment. The configuration includes the nominal thrust, thrust direction and installation position.

[0032] Specifically, the steps are as follows:

[0033] Step 1-1: Build a thruster model

[0034] In this embodiment, Figure 1 As shown in Table 1, the aircraft has a total of 13 thrusters, of which J1~J12 are cold air thrusters, each with a rated thrust of 1N; J13 is a Hall thruster with a rated thrust of 50mN. J1~J4 are installed on the satellite-X plane, with the thrust axis at an angle of 30° to the satellite Z axis or -Z axis; J5~J8 are installed on the satellite +X plane, with the thrust axis at an angle of 30° to the satellite Z axis or -Z axis; J9 and J10 are installed on the satellite-X plane, with the thrust axis parallel to the satellite -X axis; J11 and J12 are installed on the satellite +X plane, with the thrust axis parallel to the satellite +X axis; J13 is installed at the geometric center of the -X plane, with the thrust axis parallel to the satellite +X axis.

[0035] like Figure 2As shown, the origin of the satellite mechanical installation coordinate system is defined as the geometric center of the -X plane, +Xj is the flight direction, +Zj points to the center of the earth when it is stable relative to the earth, and the Yj axis forms a right-handed system with the Zj and Xj axes. The center of mass of the satellite in the Xj, Yj, and Zj directions is (1,0, 0) (tentative). Tracker fixed coordinate system , coordinate origin , Each axis coincides with the Xj, Yj, and Zj axes; Table 1 gives the positions of the intersections of the 13 thrusters and the mounting plane in the mechanical mounting coordinate system, as well as the angles between the jet axis and the three axes of the star. , It is time The control cycle is The startup duration of the i-th thruster under the setting of Describes the proportion of engine startup time in the control cycle. is the kth thruster Delayed instruction, It reflects the time delay characteristics of the engine from receiving the start-up command to the thrust. The resultant force and torque generated by the thruster (in the absence of faults) are given in the tracker fixed coordinate system, as shown in Table 1.

[0036] Combined force:

[0037] Resultant torque:

[0038]

[0039]

[0040] The thruster fixed direction vector is defined as , the nominal thrust is defined as ;

[0041] is the center of mass relative to the tracker geometry The position vector of the center, is the position vector of the thruster.

[0042] Table 1 Thruster installation position (relative to the mechanical installation coordinate system)

[0043]

[0044] Fault model of thruster

[0045] Refers to the A thruster. Simulating leakage failure

[0046] Refers to thruster full opening failure

[0047] Simulates efficiency loss failures.

[0048] Refers to thruster jam shutdown failure

[0049] In the absence of faults

[0050] definition in

[0051]

[0052] Step 1-2: Establish the aircraft control model

[0053]

[0054]

[0055]

[0056] Describes the relative motion between two spacecraft. Where n is the average orbital angular velocity of the target spacecraft. They are the three-axis relative position of the tracker relative to the target coordinate system and its velocity vector. Figure 2 Schematic diagram of the rendezvous between the tracker and the target. It is a reference coordinate system centered on the target. In the rendezvous motion of the near-circular orbit, the motion of the tracker is mainly caused by the earth's gravity and its thrust.

[0057] Rotation Matrix By Tracker and target The thrusters generate the force calculated from the attitude quaternion estimate from Coordinate system goes to Coordinate system. The relative position is measured by relative measurement equipment such as lidar equipment, and the rotation matrix can be calculated from the measurement data of navigation sensors such as star sensors and gyroscopes. is Relative position expressed in a coordinate system.

[0058] Step 1-3: The thruster distribution is based on the three-axis desired thrust vector of the satellite body obtained by the control algorithm. (3×1) and the three-axis desired torque vector (3×1), and the layout of each thruster, to obtain the output thrust of each thruster (12×1) (variable thrust is achieved through time modulation, 12 is the number of thrusters involved in the allocation).

[0059] The control allocation matrix of the thruster is defined as , the desired thrust vector and torque vector obtained by control calculation are , then the thrust output of each thruster is The following constraints must be met:

[0060]

[0061]

[0062] The optimization goal is , that is, the fuel consumption is optimal. The above problem is described as a quadratic programming problem, and the interior point convex optimization algorithm is used to solve the quadratic programming problem. When a thruster failure is found, the row data where it is located can be removed accordingly, and the 12-dimensional data can be processed after dimensionality reduction.

[0063] Control algorithms such as CW guidance and attitude control and convex optimization algorithms are well known and will not be described in detail here.

[0064]

[0065] in

[0066]

[0067] Step 2: Create a fault data set, where the fault data set includes the output of the control system under various fault states during the rendezvous and docking process of the aircraft.

[0068] Specifically, various fault conditions are set, including the fault conditions of each thruster, which are divided into thrust valve failure to open, thrust valve normally open and unable to be closed, thrust reduction and thrust leakage according to the fault type. In this embodiment, the thrust reduction is 30%, 50%, and thrust leakage is 30% and 50% respectively. For the 6 fault modes of each thruster, 1000 sets of data are generated for each mode, and a total of 12x6x1000=72000 sets of data are generated for parameter training. In the process of rendezvous and docking, the outputs of the guidance, navigation and control system under various fault states are collected, including three-axis relative attitude measurement data, three-axis relative position measurement data and three-axis relative attitude controller output and three-axis relative position controller. Figure 3 This is a schematic diagram of the system closed loop. According to the connection relationship, data under fault conditions is generated for subsequent training. It should be noted that all modules in the block diagram are mathematical simulation modules, not physical modules.

[0069] The guidance and control controller in the figure performs guidance and control calculations for the tracker, carries out rendezvous and manipulation tasks, receives the measurement values ​​of its own attitude from the navigation sensor and the actual relative position and relative attitude measurement values ​​of the two aircraft, and then calculates the control algorithm. After the calculation is completed, the desired control force and control torque are output to the control distribution module.

[0070] The control distribution module distributes the desired control force and control torque as the command u of each thruster The output is sent to the thruster module. The control distribution module also receives instructions from the liquid neural network fault detection module regarding the faulty thruster, such as shutting down a faulty thruster. The control distribution module will isolate the faulty thruster during distribution.

[0071] The thruster module sends the commands u to each thruster. Add delayed generation The data is output to the tracker dynamics module, where thruster faults can be set. The fault types are described in step 1-1.

[0072] The tracker dynamics and target dynamics are simulated in a unified time scenario, with the initial positions and UTC times of their respective orbits preset, and the simulation is performed based on the inertia and mass characteristics of the aircraft, which can receive the power-on time of the thrusters. The data changes the position and attitude of the tracker, outputs the quaternion and relative position of the two aircraft systems relative to the orbital coordinate system and the real-time angular velocity data of the two aircraft to the navigation sensor.

[0073] The navigation sensor calculates the input data to obtain the true relative position and relative attitude of the two aircraft, adds the corresponding noise data according to the sensor measurement error model, and outputs it to the liquid neural network fault detection module.

[0074] The liquid neural network fault detection module only records data including When the training data is completed, the system performs real-time fault detection according to the trained parameters when in closed-loop use, and gives the command to close the faulty thruster.

[0075]

[0076] is the relative position measurement value, controlling the model , for The relative attitude measurement value after conversion to Euler angle, the controller command output , is the relative attitude control instruction; It is the control instruction of relative position.

[0077] Step 3: Build and train the liquid neural network, such as Figure 4 As shown, the liquid neural network includes an input layer, a liquid layer and a classifier. The input layer is used to integrate the original time series data of the aircraft control system into a unified input vector; the liquid layer includes N neurons, which are used to dynamically model the input sequence and extract time series features; the classifier is used to map to the fault category probability through the fully connected layer.

[0078] In this embodiment, thruster fault diagnosis is achieved using raw data consisting of three-axis relative attitude measurement data, three-axis relative position measurement data, three-axis relative attitude controller output data, and three-axis relative position controller output data.

[0079] Specifically, the input layer integrates the original time series data from the aircraft control system into a unified input vector, and the state variables are three-axis relative attitude measurement data, three-axis relative position measurement data, three-axis relative attitude controller output (torque), and three-axis relative position controller output (thrust).

[0080]

[0081] The liquid layer is the core layer of the system, responsible for dynamic modeling of input sequences and time series feature extraction. Through recursive updates, the internal state at each moment memorizes the information of the previous moment while integrating the current input, thereby capturing the subtle dynamic changes during flight and providing sufficient dynamic information for fault detection. Hidden state of liquid neurons Evolves over time and is described by the differential equation:

[0082]

[0083] Among them, the neuron state is represented by: Assume that there are N neurons in the liquid layer, let represents the hidden state at time t.

[0084] Input processing: through the weight matrix The input vector is calculated to obtain the dynamic time constant, , is the constant of neuron i at time t, is the learnable bias value, Sigmoid activation function.

[0085] Synaptic coupling: , is the synaptic weight matrix, defining the connection strength between neurons, is the hidden state, The activation function generated for the synaptic weights uses the Tanh or LeakyReLU function. , Concatenate the hidden state and input into a vector, and are the learnable weight matrices and biases.

[0086] Hide status updates: is the sampling time interval (which can be determined by the data sampling frequency), is the dynamic time constant used to adjust the decay speed of the state, is the nonlinear activation function tanh.

[0087] Classifier settings:

[0088] The hidden state at the final time step T Mapped to fault category probability through a fully connected layer:

[0089]

[0090] is the output layer weight matrix;

[0091] is the bias for the output layer.

[0092] , K is the number of fault categories.

[0093] Furthermore, in this embodiment, the training method for the constructed liquid neural network is:

[0094] For each of the six fault modes of each thruster, 1000 sets of data were generated for each mode, and a total of 12*6 types*1000 sets of working conditions, 72000 sets of data were generated for the training of liquid neural network parameters. After the test set accuracy rate was obtained above 98%, the test set accuracy rate test was carried out. If the accuracy rate reached above 97%, it was considered qualified and the parameters were determined, such as Figure 5 shown.

[0095] Backpropagation parameter update: cross entropy loss function ), and update the parameters by solving the gradient of each parameter through back propagation. are the true labels provided in the training data.

[0096] Update parameters using gradient descent (Adam optimizer)

[0097]

[0098] , is the learning rate.

[0099] Step 4: Obtain real-time data output by the aircraft's control system and apply the pre-trained liquid neural network for fault diagnosis.

[0100] In this embodiment, neural network parameters are used in the aircraft guidance, navigation and control software for real-time fault diagnosis, such as Figure 3 As shown, after a thruster fault is found in the liquid neural network fault detection module, the faulty thruster is isolated when it is determined to be a certain fault for N consecutive beats. In this embodiment, N is set to 5. When the liquid neural network detects a thruster fault, if it is determined to be a certain fault for 5 consecutive beats, the faulty thruster is isolated and the thruster is reallocated for control.

[0101] Example 2

[0102] The difference between this embodiment and embodiment 1 is that, on the basis of embodiment 1, safety corridor identification is further introduced.

[0103] It is understandable that since the probability of fault identification cannot reach 100%, there is still a probability that the fault diagnosis will not be detected when issuing a command to shut down a faulty thruster. Therefore, a safety corridor design is adopted in the algorithm of the guidance and control controller. The safety corridor is symmetrical with the target docking axis and the half angle is =8° cone. The apex of the cone in the docking corridor takes the origin of the target spacecraft docking surface coordinate system as the reference point. If the relative position exceeds the corridor area or the relative attitude is greater than 8 degrees, it is considered abnormal and evacuation control is required to abandon the rendezvous and docking or space control mission.

[0104] It should be noted that safety corridor identification is a conventional technical means. The safety corridor can be geometrically defined as a dynamically adjusted conical space area, whose axis is aligned with the orbit tangent of the target spacecraft, and whose boundary is determined by the safety tolerance of relative position, velocity and attitude, to ensure that the tracking spacecraft can approach safely and avoid collision. The lateral distance, velocity vector angle, and relative attitude angle deviation must not exceed the threshold.

[0105] It is understandable that the introduction of corridor identification on the basis of the intelligent fault diagnosis method of the on-orbit service vehicle thruster can effectively improve the reliability and safety of rendezvous and docking and space control missions.

[0106] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and descriptions are only preferred examples of the present invention and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. An intelligent fault diagnosis method for thrusters of an on-orbit servicing aircraft, characterized in that: The steps include: Step 1: Establish a mathematical model related to the control of the aircraft and configure the propeller of the aircraft, wherein the mathematical model includes the dynamic equation of the aircraft and the dynamic method of the flight environment, and the configuration includes the nominal thrust, thrust direction and installation position; Step 2: Create a fault data set, where the fault data set includes the output of the control system under various fault states during the rendezvous and docking process of the aircraft; Step 3: construct and train a liquid neural network, wherein the liquid neural network includes an input layer, a liquid layer, and a classifier, wherein the input layer is used to integrate the original time series data of the control system of the aircraft into a unified input vector; the liquid layer includes N neurons, and the parameters of the neurons change in real time with the input and state to adaptively capture the time series features; A classifier, used to map to fault class probabilities through an output layer; Step 4: Obtain real-time data output by the aircraft's control system and apply the pre-trained liquid neural network for fault diagnosis.

2. The method for intelligent fault diagnosis of thrusters of an on-orbit servicing aircraft according to claim 1, characterized in that: In step 2, when making the fault data set, it is necessary to set the fault conditions, which include the thruster's propulsion valve being unable to open, the thrust valve being normally open and unable to be closed, thrust reduction, and thrust leakage.

3. The method for intelligent fault diagnosis of thrusters of an on-orbit servicing aircraft according to claim 1, characterized in that: In step 2, when the fault condition is set, the thrust decreases by 30% and 50% respectively, and the thrust leakage leaks by 30% and 50% respectively.

4. The method for intelligent fault diagnosis of an on-orbit servicing aircraft thruster according to claim 1, characterized in that: In step 2, the output of the control system under various fault conditions during the rendezvous and docking process of the aircraft includes three-axis relative attitude measurement data, three-axis relative position measurement data, three-axis relative attitude controller output and three-axis relative position controller output data.

5. The method for intelligent fault diagnosis of thrusters of an on-orbit servicing aircraft according to claim 1, characterized in that: The method for extracting the time series characteristics of the liquid layer is: Represents the hidden state at time t. The input vector is projected into the hidden state space through the weight matrix, and the state at the previous moment is projected into the hidden state space through the hidden state weight matrix. Introducing updates to the current moment to capture the dynamic continuity between states; As the formula As shown, is the state at the previous moment Based on this, add a time interval and time constant The relevant adjustments and , integrating the current input and historical information to obtain a new state change, where Contains the temporal characteristics of the input sequence.

6. The method for intelligent fault diagnosis of thrusters of an on-orbit servicing aircraft according to claim 5, characterized in that: When the classifier predicts the probability of fault category, the hidden state of the final time step T obtained by the liquid layer is Mapped to fault class probability through the output layer.

7. The method for intelligent fault diagnosis of thrusters of an on-orbit servicing aircraft according to any one of claims 1 to 6, characterized in that: In step 3, when training the liquid neural network, a cross entropy loss function and a weight decay function are used, and the parameters are updated by back propagation and gradient descent.

8. The method for intelligent fault diagnosis of thrusters of an on-orbit servicing aircraft according to claim 7, characterized in that: In step 3, a threshold is set, and when the failure probability predicted by the liquid neural network reaches the set threshold, the parameter is fixed.

9. The method for intelligent fault diagnosis of thrusters of an on-orbit servicing aircraft according to claim 1, characterized in that: In step 4, the method of using the pre-trained liquid neural network for real-time fault diagnosis is as follows: when the liquid neural network detects a thruster fault and N consecutive beats determine that they are all a certain fault, the faulty thruster is isolated and the thruster is reallocated for control.

10. The method for intelligent fault diagnosis of thrusters of an on-orbit servicing aircraft according to claim 9, characterized in that: In step 4, after the fault diagnosis is completed, there is still a probability that the fault diagnosis has not been detected, so the safety corridor judgment is introduced, and the evacuation is carried out after exceeding the safety corridor, and the rendezvous and docking mission is abandoned.

11. The method for intelligent fault diagnosis of thrusters of an on-orbit servicing aircraft according to claim 10, characterized in that: The safety corridor is symmetrical with the docking axis of the target aircraft for rendezvous and docking with the aircraft, and the half angle is =8° cone, the cone vertex of the docking corridor of the safety corridor takes the origin of the target spacecraft docking surface coordinate system as the reference point. If the relative position exceeds the corridor area or the relative attitude is greater than 8 degrees, it is considered abnormal and evacuation control is required to abandon the rendezvous and docking or space control mission.

Images