CN114611377A - Nuclear extreme learning machine-based gliding missile reentry trajectory real-time reconstruction method - Google Patents
Nuclear extreme learning machine-based gliding missile reentry trajectory real-time reconstruction method Download PDFInfo
- Publication number
- CN114611377A CN114611377A CN202210088426.1A CN202210088426A CN114611377A CN 114611377 A CN114611377 A CN 114611377A CN 202210088426 A CN202210088426 A CN 202210088426A CN 114611377 A CN114611377 A CN 114611377A
- Authority
- CN
- China
- Prior art keywords
- fault
- learning machine
- kernel
- aircraft
- gliding
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 40
- 238000003745 diagnosis Methods 0.000 claims abstract description 18
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 9
- 230000008569 process Effects 0.000 claims abstract description 8
- 238000012549 training Methods 0.000 claims description 26
- 230000006870 function Effects 0.000 claims description 24
- 239000011159 matrix material Substances 0.000 claims description 21
- 210000002569 neuron Anatomy 0.000 claims description 16
- 238000012360 testing method Methods 0.000 claims description 16
- 239000013598 vector Substances 0.000 claims description 8
- 238000013507 mapping Methods 0.000 claims description 5
- 238000005457 optimization Methods 0.000 claims description 5
- 230000008859 change Effects 0.000 claims description 4
- 238000004088 simulation Methods 0.000 claims description 4
- 238000012544 monitoring process Methods 0.000 claims description 3
- 238000013461 design Methods 0.000 claims description 2
- 238000010586 diagram Methods 0.000 description 7
- 230000009467 reduction Effects 0.000 description 3
- 238000000605 extraction Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 210000002364 input neuron Anatomy 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 210000004205 output neuron Anatomy 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/15—Vehicle, aircraft or watercraft design
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Computer Hardware Design (AREA)
- Computational Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Automation & Control Theory (AREA)
- Aviation & Aerospace Engineering (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Monitoring And Testing Of Nuclear Reactors (AREA)
Abstract
The invention discloses a method for reconstructing a gliding reentry track in real time based on a nuclear extreme learning machine, which is an online track reconstruction scheme of a hypersonic-speed boosting gliding aircraft after a fault occurs in the reentry process of the hypersonic-speed boosting gliding aircraft. Firstly, establishing a fault model of damaged pneumatic surfaces of a gliding reentry section of an aircraft and extracting fault characteristics; then, a fault diagnosis method based on a kernel-limit learning machine algorithm is designed, and finally, a fault-tolerant guidance and track reconstruction scheme of the aircraft under the condition that the fault occurs in the glide section is designed based on a fault diagnosis result. The fault diagnosis method provided by the invention can overcome the dependence on an aircraft model and realize rapid and accurate diagnosis on the fault in a data-driven mode; the fault-tolerant guidance and track reconstruction strategy can adaptively adjust the control instruction and the task target according to the fault diagnosis result, and realize online fault active fault tolerance.
Description
Technical Field
The invention relates to the field of spacecraft guidance, in particular to a real-time track reconstruction method considering faults in the boosting, gliding and reentry process of a hypersonic aircraft.
Background
The hypersonic-velocity boosting gliding aircraft is a very complex multivariable system, has the characteristics of strong nonlinearity, strong coupling, fast time-varying property and the like, has obvious viscous effect and large change of working environment conditions, and is easy to generate faults of damage to actuators, sensors, structures and the like. In recent years, with the development and utilization of space resources, the demand for a world round trip system capable of completing tasks quickly and smoothly is urgent. Research shows that the most severe phase in the world round trip process is the reentry phase, so the atmosphere reentry field is regarded as the key for widely developing space application, and the reentry guidance and control problem becomes a research hotspot in the aerospace field. Considering the reentry process of the hypersonic aircraft, when an emergency occurs, the offline track no longer meets the requirement of a flight task, the track needs to be adjusted online, and a new track is designed to be used as a target for tracking guidance, namely an online track reconstruction guidance method.
Disclosure of Invention
Aiming at the problems, the invention provides a method for reconstructing the gliding reentry track in real time based on a nuclear extreme learning machine, which is an online track reconstruction scheme after a hypersonic-speed-assisted gliding aircraft breaks down in the reentry process.
The invention relates to a method for reconstructing a gliding reentry track in real time based on a nuclear extreme learning machine, which comprises the following specific steps:
step 1: and establishing a fault model of the damaged pneumatic surface of the gliding reentry section of the aircraft.
Step 2: and extracting fault characteristics of fault signals under various fault types.
And step 3: and acquiring the state information of the aircraft in real time in the flight process, inputting the state information into the nuclear limit learning machine subjected to offline training, and monitoring the fault condition according to the output result of the nuclear limit learning machine.
And 4, step 4: and accurately estimating the aerodynamic coefficient change degree caused by the fault, and transmitting the estimated fault information to a guidance algorithm.
The invention has the advantages that:
1. according to the method for reconstructing the gliding reentry trajectory in real time based on the kernel limit learning machine, the proposed fault diagnosis method can overcome the dependence on an aircraft model, and can realize rapid and accurate diagnosis on the fault in a data driving mode.
2. In the method for reconstructing the gliding reentry trajectory in real time based on the kernel-limit learning machine, the fault-tolerant guidance and trajectory reconstruction strategy can adaptively adjust the control instruction and the task target according to the fault diagnosis result, and realize the online fault active fault tolerance.
Drawings
FIG. 1 is a flow chart of a method for reconstructing a gliding reentry trajectory in real time based on a kernel-based extreme learning machine according to the present invention.
FIG. 2 shows the application of k at the 5 th s of glideiFault trajectory diagram without any fault tolerance after a fault of 0.5.
Figure 3 is a ballistic inclination curve for different failure levels.
Figure 4 is a ballistic inclination curve at different moments of failure.
Fig. 5 is a schematic diagram of fault feature vectors under different fault degrees.
Fig. 6 is a diagram showing the construction of the extreme learning machine.
Fig. 7 is a diagram illustrating a result of the fault diagnosis.
FIG. 8 is a diagram illustrating reachable region estimation.
FIG. 9 is a schematic diagram of a reconstructed trajectory without modifying the target.
FIG. 10 is a schematic diagram of a reconstructed trajectory of an altered object.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings.
The invention relates to a method for reconstructing a gliding reentry trajectory in real time based on a nuclear extreme learning machine, which is specifically designed as shown in figure 1:
step 1: and establishing a fault model of the damaged aerodynamic surface of the gliding reentry section of the aircraft.
Consider a failure in which the aerodynamic surfaces of an aircraft are damaged during flight in the glide phase, which failure can lead to a reduction in the lift coefficient of the aircraft. Therefore, the fault diagnosis method used by the invention is driven by the aircraft ballistic data, and in order to acquire fault data, different degrees of lift coefficient changes are applied at different times on the normal trajectory through simulation so as to simulate the fault trajectory under different fault conditions, namely: twenty times after the start of the glide period, namely 0.5s, 1s, 1.5s … … 9s, 9.5s, and 10sThe lift coefficient is changed to obtain 200 different sets of fault lift coefficients CL' (failure condition):
CL'＝kiCL,ki＝0.1,0.2,0.3～1
in the formula, CL' is the failure lift coefficient, kiIs the coefficient of aerodynamic surface damage, CLNormal lift coefficient; as shown in fig. 2, k is applied at the 5 th s of the glide sectioniAfter a fault of 0.5, no fault tolerance measures are taken.
In order to classify and diagnose pneumatic faults of different degrees, ballistic calculations are carried out according to nominal control commands under the 200 groups of fault conditions, and the simulation time is (t)f+10)s，tfIs the time of occurrence of the fault. As shown in FIG. 3 as tfTake 1s, kiTaking a trajectory inclination angle curve within 10s after a fault occurs under the condition of 0.1-1; as shown in FIG. 4 as tfTake 1-10 s, kiThe ballistic dip curve was taken within 10s after the failure occurred at 0.1. It can be seen that the ballistic inclination angle data under different fault degrees are different at the same moment, and the ballistic inclination angle data under the same fault degree have certain similarity at different moments, which meets the requirement of fault classification by using supervised learning.
Step 2: and extracting fault characteristics.
In order to facilitate subsequent fault diagnosis, feature extraction needs to be performed on fault signals under various fault types, that is, dimension reduction is performed on original signals output by the sensors. In consideration of measurement noise, the method carries out fault diagnosis based on trajectory inclination angle data within 10s after the fault occurs, and the aircraft state quantity within 10s after the fault occurs is obtained through simulation
Data, where r is the geocentric distance; theta is longitude;
is dimension; v is the speed; gamma is a ballistic inclination angle; psi is the ballistic declination; a curve fit is then made to each state quantity,its first order coefficients are taken and combined into a feature vector for the set of data. Shown in FIG. 5 as tfTake 1s, kiAnd taking each fault feature vector under the condition of 0.1-1.
And step 3: and fault diagnosis based on the core limit learning machine.
The method comprises the steps of acquiring aircraft state information in real time in the flight process, inputting the aircraft state information into a nuclear limit learning machine subjected to offline training, and monitoring the fault condition according to the output result of the nuclear limit learning machine; the method specifically comprises the following steps:
A. nuclear extreme learning machine (KELM) design
Given N arbitrarily different training samples
Wherein xiIs an input vector of samples i, tiThe desired output vector for sample i. As shown in FIG. 6, the standard extreme learning machine has N input neurons, L hidden layer neurons and M output neurons, E in the figure1～E2jRespectively are input layer data; l1～lMRespectively are output layer data; omega and b are weight and bias of the hidden layer neuron node respectively; v is the output layer weight.
The activation function g (x) is expressed mathematically as follows:
Hβ＝T
h is a random feature mapping matrix, beta represents a weight matrix between an output layer and a hidden layer, and T represents an expected output matrix of a training sample. In the invention, the input layer of the extreme learning machine is a fault feature vector, the output layer is a data label, the number of the neurons is 10, and the neurons respectively correspond to kiAnd taking a pneumatic fault type of 0.1-1.
After the weights ω and biases b of the hidden layer neuron nodes are randomly generated from arbitrary continuous sample distribution probabilities and given training samples, the hidden layer output matrix H is actually known and remains unchanged. The aim of extreme learning machine algorithm training is to solve the minimum norm least square solution beta of a linear system H beta T:
β＝H+T
wherein H+A Moore-Penrose generalized inverse matrix representing the hidden layer output matrix H. Under the condition that the specific form of the feature mapping h (x) of the hidden layer is unknown, a kernel function needs to be introduced to measure the similarity between samples, and a kernel matrix of the KELM can be defined according to the Mercer condition, and is expressed as follows:
ΩELM＝HH+:ΩELMi,j＝h(xi)h(xj)＝K(xi,xj)
in the formula, i is the number of input layer neuron, j is the number of hidden layer neuron, xiRepresenting input layer data, xjRepresenting hidden layer data, ΩELMi,jAnd K (x)i,xj) A kernel matrix and a kernel function between the ith input layer neuron and the jth hidden layer neuron, respectively. Thus, the model output of the KELM may be expressed as:
from the above formula, where C is the regularization coefficient and I is the identity matrix, it can be seen that the kernel matrix ΩELM＝HH+Only sum input data xiAnd the number of training samples. In KELM, the kernel function K (x)i,xj) Inputting data (x) in a low-dimensional spacei,xj) Conversion to inner product h (x) in high dimensional feature spacei)·h(xj) Therefore, the dimension of the feature space cannot be influenced, and dimension disaster is successfully avoided. The KELM only needs to select a proper kernel function and does not need to set the number of hidden layer neurons, so that the time for optimizing the number of hidden layer neurons is saved. Compared with the traditional ELM algorithm, the KELM can effectively solve the problem of generalization and stability reduction caused by random assignment of hidden layer neurons.
B. Nuclear limit learning machine parameter optimization
The kernel function used by the invention is a radial basis function, is a global kernel function, can effectively resist noise and interference, is suitable for linear inseparable data, and has the following mathematical form:
where s is a width parameter of the radial basis kernel function, and the result obtained by classification is greatly influenced by the parameter s. The extreme learning machine with the radial basis function as the kernel function comprises two parameters to be optimized, a regularization parameter C and a kernel function parameter s. The regularization parameter C affects the training accuracy and stability, and can be considered as a parameter for balancing the training accuracy and the testing accuracy. The larger C is, the higher training precision of the kernel limit learning machine is, but the stability is poor, and the value range is usually selected to be 10-4,104]. The other kernel parameter s is used for controlling the deviation and variance of the kernel function, the larger s is, the larger the deviation is, the smaller the variance is, and the value range is usually selected to be [0.1,10 ]]. The training result of the extreme learning machine is embodied in four aspects: training accuracy AtrainAnd test accuracy AtestTraining time and testing time. The selection of the parameters C and s influences A thereintrainAnd AtestThe optimized objective function is thus set to:
minF＝W·[Atrain,Atest]'
wherein W ═ W1,w2]Is a constant coefficient matrix, w1For training the coefficients, w2Are test coefficients. Two parameters of the kernel-extreme learning machine are optimized using a differential biophysical optimization algorithm.
C. Nuclear extreme learning machine training and testing
After the feature extraction is carried out on the two hundred groups of fault data in the step 1, the two hundred groups of fault data are divided into two parts at random: half of the data are used as training samples, and the fault occurrence time is t f1, 2, 3,. 10 s; the other half of the data is used as a test sample, and the fault occurrence time is tf0.5, 1.5, 2.5,. 9.5 s. For the kernel extreme learning machine, the feature vector and the kernel function are utilized to directly obtain the weight matrix beta of the output layer, the matrix is a mapping matrix obtained after training, and the corresponding relation from the hidden layer to the fault type is reflected. Then, training of the extreme learning machine is realized by using the training samples, two parameters of the extreme learning machine are optimized through a differential biophysical optimization algorithm, and the optimization result is [ C, s [ ]]＝[3317.4,1.3107]. The test sample is input into the trained extreme learning machine, and the valley station and the diagnosis result are obtained according to the output value of the test sample, as shown in fig. 7, the diagnosis precision is extremely high, and the feasibility of the fault diagnosis method is verified.
And 4, step 4: fault-tolerant guidance and trajectory reconstruction scheme
When the aircraft generates aerodynamic surface damage faults in the gliding section, the aerodynamic coefficients are changed due to the faults, and new aerodynamic coefficients are generated. And accurately estimating the aerodynamic coefficient change degree caused by the fault by adopting a fault diagnosis method based on the kernel limit learning machine, and transmitting the estimated fault information to a guidance algorithm. And estimating the reachable area of the aircraft in the current state according to the new aerodynamic coefficient and making a decision, as shown in FIG. 8: if the original target point is in the reachable area, redesigning the control command to enable the aircraft to hit the target, as shown in fig. 9; if the original target point is not in the reachable area, the terminal point is reselected in the reachable area and a control command is designed to make the aircraft drive to a new target, as shown in fig. 10.
Claims (5)
1. A method for reconstructing a gliding reentry trajectory in real time based on a kernel-limit learning machine is characterized in that: the method comprises the following specific steps:
step 1: establishing a fault model of damaged pneumatic surfaces of a gliding reentry section of the aircraft;
step 2: extracting fault characteristics of fault signals under various fault types;
and step 3: acquiring the state information of the aircraft in real time in the flight process, inputting the state information into the core limit learning machine subjected to offline training, and monitoring the fault condition according to the output result of the core limit learning machine;
and 4, step 4: and accurately estimating the aerodynamic coefficient change degree caused by the fault, and transmitting the estimated fault information to a guidance algorithm.
2. The method as claimed in claim 1, wherein the method comprises the following steps: in step 1, in order to acquire fault data, lift coefficient changes of different degrees are applied at different times on a normal trajectory through simulation so as to simulate fault trajectories under different fault conditions.
3. The method as claimed in claim 1, wherein the method comprises the following steps: in step 2, fault diagnosis is carried out based on trajectory inclination angle data within 10s after the fault occurs, and the state quantity (r, theta,
v, γ, ψ) data, where r is the geocentric distance; theta is longitude;
is dimension; v is the speed; gamma is a ballistic inclination angle; psi is the ballistic declination; and then, carrying out primary curve fitting on each state quantity, taking a primary term coefficient of each state quantity and combining the coefficients into a characteristic vector of the group of data.
4. The method as claimed in claim 1, wherein the method comprises the following steps: the design method of the kernel limit learning machine in the step 3 comprises the following steps:
under the condition that the feature mapping h (x) of the hidden layer of the standard extreme learning machine is unknown in specific form, a kernel function is introduced to measure the similarity between training samples, and a kernel matrix of the kernel extreme learning machine is defined according to Mercer conditions and is expressed as follows:
ΩELM＝HH+:ΩELMi,j＝h(xi)h(xj)＝K(xi,xj)
where H is a random feature mapping matrix, i is the number of input layer neurons, j represents the number of hidden layer neurons, xiRepresenting input layer data, xjRepresenting hidden layer data, ΩELMi,jAnd K (x)i,xj) A kernel matrix and a kernel function between the ith input layer neuron and the jth hidden layer neuron, respectively; thus, the model output of the KELM may be expressed as:
in the above formula, C is a regularization coefficient; i is an identity matrix; t represents an expected output matrix of the training sample; beta represents a weight matrix between the output layer and the hidden layer; n total number of training samples.
5. The method as claimed in claim 4, wherein the method comprises the following steps: the method for optimizing the parameters of the kernel limit learning machine comprises the following steps:
the kernel function of the kernel limit learning machine is a radial basis function, and the mathematical form is as follows:
wherein s is a width parameter of the radial basis kernel function;
the extreme learning machine with the radial basis function as the kernel function comprises two parameters to be optimized, a regularization parameter C and a kernel function parameter s;
the training result of the extreme learning machine is embodied in four aspects: training accuracy AtrainAnd test accuracy AtestTraining time and testing time; the selection of the parameters C and s influences A thereintrainAnd AtestThe optimized objective function is thus set to:
min F＝W·[Atrain,Atest]'
wherein W is [ W ═ W1,w2]Is a constant coefficient matrix; two parameters of the kernel-extreme learning machine are optimized using a differential biophysical optimization algorithm.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210088426.1A CN114611377B (en) | 2022-01-25 | 2022-01-25 | Method for reconstructing reentry track of gliding missile in real time based on kernel extreme learning machine |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210088426.1A CN114611377B (en) | 2022-01-25 | 2022-01-25 | Method for reconstructing reentry track of gliding missile in real time based on kernel extreme learning machine |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114611377A true CN114611377A (en) | 2022-06-10 |
| CN114611377B CN114611377B (en) | 2024-06-07 |
Family
ID=81857641
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210088426.1A Active CN114611377B (en) | 2022-01-25 | 2022-01-25 | Method for reconstructing reentry track of gliding missile in real time based on kernel extreme learning machine |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114611377B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170341779A1 (en) * | 2015-04-15 | 2017-11-30 | The Government Of The United States Of America, As Represented By The Secretary Of The Navy | Control System and Method for a Plane Change for Satellite Operations |
| CN107436604A (en) * | 2017-07-31 | 2017-12-05 | 中南大学 | Controlling planning method is intelligently decomposed in carrying robot path under a kind of intelligent environment |
| CN112116101A (en) * | 2020-09-17 | 2020-12-22 | 南京航空航天大学 | Aero-engine fault diagnosis method based on group reduction kernel limit learning machine |
-
2022
- 2022-01-25 CN CN202210088426.1A patent/CN114611377B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170341779A1 (en) * | 2015-04-15 | 2017-11-30 | The Government Of The United States Of America, As Represented By The Secretary Of The Navy | Control System and Method for a Plane Change for Satellite Operations |
| CN107436604A (en) * | 2017-07-31 | 2017-12-05 | 中南大学 | Controlling planning method is intelligently decomposed in carrying robot path under a kind of intelligent environment |
| CN112116101A (en) * | 2020-09-17 | 2020-12-22 | 南京航空航天大学 | Aero-engine fault diagnosis method based on group reduction kernel limit learning machine |
Non-Patent Citations (2)
| Title |
|---|
| LI ZHAOYING 等: "Trajectory tracking of tail-sitter aircraft by L1 adaptive fault tolerant contro", 《JOURNAL OF SYSTEMS ENGINEERING AND ELECTRONICS》, 31 December 2021 (2021-12-31), pages 1477 - 1489 * |
| 寇英信 等: "基于改进核极限学习机和集成学习理论的目标机动轨迹预测", 《国防科技大学学报》, 31 December 2021 (2021-12-31), pages 23 - 35 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114611377B (en) | 2024-06-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108594788B (en) | Airplane actuator fault detection and diagnosis method based on deep random forest algorithm | |
| Abbaspour et al. | Neural adaptive observer-based sensor and actuator fault detection in nonlinear systems: Application in UAV | |
| CN106156401B (en) | Multi-combination classifier based data driving system state model online identification method | |
| CN111190429B (en) | Unmanned aerial vehicle active fault-tolerant control method based on reinforcement learning | |
| CN114564000B (en) | Active fault tolerance method and system based on intelligent aircraft actuator fault diagnosis | |
| CN104850889A (en) | Airplane rotation actuator drive unit adaptive fault detection, isolation and confidences assessment method | |
| Ma et al. | Deep auto-encoder observer multiple-model fast aircraft actuator fault diagnosis algorithm | |
| CN111638034B (en) | Strain balance temperature gradient error compensation method and system based on deep learning | |
| CN112393934A (en) | Wind turbine generator fault diagnosis method based on sparse self-coding and extreme learning machine | |
| CN111176263A (en) | Online aircraft thrust fault identification method based on BP neural network | |
| Cao et al. | System identification method based on interpretable machine learning for unknown aircraft dynamics | |
| Qin et al. | Sensor fault diagnosis of autonomous underwater vehicle based on LSTM | |
| Han et al. | Quadratic-Kalman-filter-based sensor fault detection approach for unmanned aerial vehicles | |
| Emami et al. | Online identification of aircraft dynamics in the presence of actuator faults | |
| CN110412997B (en) | Spacecraft attitude control spray pipe fault intelligent diagnosis system and method based on neural network | |
| Kumar et al. | Data-driven method based aerodynamic parameter estimation from flight data | |
| Park et al. | Data-driven fault detection and isolation of system with only state measurements and control inputs using neural networks | |
| Yan et al. | Active fault-tolerant control integrated with reinforcement learning application to robotic manipulator | |
| Tong et al. | Machine learning for uav propeller fault detection based on a hybrid data generation model | |
| CN108959787A (en) | Consider the thermal deformation prediction technique and system of the macro dual drive system of actual condition | |
| CN112597706A (en) | Carrier rocket executing mechanism fault online diagnosis method | |
| CN114611377B (en) | Method for reconstructing reentry track of gliding missile in real time based on kernel extreme learning machine | |
| CN114462149B (en) | Aircraft pneumatic parameter identification method based on pre-training and incremental learning | |
| Yao et al. | A data-driven fault detection approach for Modular Reconfigurable Flying Array based on the Improved Deep Forest | |
| Zhao et al. | Active fault-tolerant strategy for flight vehicles: Transfer learning-based fault diagnosis and fixed-time fault-tolerant control |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant |