CN109507989A - A kind of method of unmanned plane sensor fault diagnosis - Google Patents
A kind of method of unmanned plane sensor fault diagnosis Download PDFInfo
- Publication number
- CN109507989A CN109507989A CN201811582140.9A CN201811582140A CN109507989A CN 109507989 A CN109507989 A CN 109507989A CN 201811582140 A CN201811582140 A CN 201811582140A CN 109507989 A CN109507989 A CN 109507989A
- Authority
- CN
- China
- Prior art keywords
- unmanned plane
- time delay
- fault diagnosis
- sample sequence
- delay sample
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B23/00—Testing or monitoring of control systems or parts thereof
- G05B23/02—Electric testing or monitoring
- G05B23/0205—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
- G05B23/0218—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults
- G05B23/0221—Preprocessing measurements, e.g. data collection rate adjustment; Standardization of measurements; Time series or signal analysis, e.g. frequency analysis or wavelets; Trustworthiness of measurements; Indexes therefor; Measurements using easily measured parameters to estimate parameters difficult to measure; Virtual sensor creation; De-noising; Sensor fusion; Unconventional preprocessing inherently present in specific fault detection methods like PCA-based methods
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Automation & Control Theory (AREA)
- Gyroscopes (AREA)
Abstract
The invention discloses a kind of methods of unmanned plane sensor fault diagnosis, comprising: the real boat data for obtaining unmanned plane medium-rate gyro, using real boat data as the first time delay sample sequence；Gray model carries out modeling and forecasting to the first time delay sample sequence, obtains prediction model using minimum variance criteria；The weight of the hidden layers of Leman neural network regulating networks, input layer and output layer；The method that the method combination Leman neural network of gray model modeling and forecasting is adjusted obtains combination latency prediction model；Combination latency prediction model is applied in unmanned plane sensor fault diagnosis, the failure of the unmanned plane sensor is diagnosed.The method of the unmanned plane sensor fault diagnosis can carry out failure predication and diagnosis to unmanned plane sensor.
Description
Technical field
The invention belongs to unmanned plane fault diagnosis field more particularly to a kind of methods of unmanned plane sensor fault diagnosis.
Background technique
With the raising and micromation of computer capacity, the progress of artificial intelligence and nanotechnolgy, the machine of unmanned plane
Load electronic equipment is more and more, and system also becomes increasingly complex, this proposes reliability, the maintainability of unmanned plane higher and higher
It is required that thus unmanned plane online system failure diagnosis is born.And UAV system is the comprehensive system of a complicated electromechanical integration
System, fault characteristic are that complexity is high, non-linear strong and phenomenon of the failure is varied, it is difficult to establish accurate mathematical model.
Since the time delay in UAV system has time-varying and random characteristic, time delay is that one of influence system performance is important
Factor.In order to improve the control performance and quality of system, accurate measurement, analysis and prediction time delay are to studying and control this kind of system
It is very important.
Summary of the invention
The technical problem to be solved in the present invention is to provide a kind of methods of unmanned plane sensor fault diagnosis, can be to nobody
Machine sensor carries out failure predication and diagnosis.
To solve the above problems, the technical solution of the present invention is as follows:
A kind of method of unmanned plane sensor fault diagnosis, the method is based on gray model and Leman neural network is pre-
Survey model, comprising the following steps:
S1: obtaining the real boat data of unmanned plane medium-rate gyro, using the real boat data as the first time delay sample sequence；
S2: the gray model carries out modeling and forecasting to the first time delay sample sequence, is obtained using minimum variance criteria
To prediction model；
S3: the weight of the hidden layers of the Leman neural network regulating networks, input layer and output layer；
S4: in conjunction with the step S2 and the step S3, combination latency prediction model is obtained；
S5: the combination latency prediction model is applied in the unmanned plane sensor fault diagnosis, the nothing is diagnosed
The failure of man-machine sensor.
The method for the unmanned plane sensor fault diagnosis that one embodiment of the invention provides, by the unmanned plane medium-rate top
The reality of spiral shell navigates data as the first time delay sample sequence and includes:
The real boat data of rate gyroscope described in choosing 500 groups, wherein when data of navigating in fact described in 400 groups are as described first
Prolong sample sequence, data of navigating in fact described in remaining 100 groups are used for the verifying of the prediction model, set the first time delay sample sequence
It is classified as X(0):
X(0)=(x(0)(1),x(0)(2),x(0)(3),…x(0)(n)) wherein,
x(0)(i), i=1,2,3 ... n are the first time delay sample sequence at i moment, and n is the time delay sample data
Number；
Accumulation process is carried out to the first time delay sample sequence and obtains the second time delay sample sequence X(1),
X(1)=(x(1)(1),x(1)(2),x(1)(3),…x(1)(n)) wherein,
x(1)(i), i=1,2,3 ... n is the second time delay sample sequence at i moment after carrying out accumulation process.
The method for the unmanned plane sensor fault diagnosis that one embodiment of the invention provides, the gray model is to described the
One time delay sample sequence carries out modeling and forecasting
The time delay queue length of the gray model is taken as 20, the gray model first time delay sample described in 400 groups
Sequence carries out modeling and forecasting, obtains coefficient a and coefficient b；
Wherein, coefficient a is development coefficient, and coefficient b is that grey acts on coefficient of discharge；
Parameter a ' and parameter b ' are found out by minimum variance criteria.
The method for the unmanned plane sensor fault diagnosis that one embodiment of the invention provides, the Leman neural network are adjusted
The weight of the hidden layer of network, input layer and output layer includes:
The number of plies of the input layer is 20, the number of plies of the hidden layer is 10 and the number of plies of the output layer is 1；
The hidden layer, the input layer and the output layer is adjusted in the Leman neural network.
The method for the unmanned plane sensor fault diagnosis that one embodiment of the invention provides, in conjunction with the step S2 and described
Step S3, obtaining combination latency prediction model includes:
In conjunction with the step S2 and the step S3, weight coefficient γ is obtained, finally obtain the combination latency prediction mould
Type.
The present invention due to using the technology described above, makes it have the following advantages that and actively imitate compared with prior art
Fruit:
1) one embodiment of the invention provide unmanned plane sensor fault diagnosis method in gray model to acquisition
The real boat data of unmanned plane medium-rate gyro carry out modeling and forecasting, and obtain prediction model according to minimum variance criteria, then strangle
The weight of the hidden layers of graceful neural network regulating networks, input layer and output layer, then by the side of gray model modeling and forecasting
The method that method combination Leman neural network is adjusted obtains combination latency prediction model, will finally combine the application of latency prediction model
Into unmanned plane sensor fault diagnosis, the failure of the unmanned plane sensor is diagnosed.This method can be to unmanned plane sensor
Carry out failure predication and diagnosis.
2) gray model and Leman neural network that one embodiment of the invention provides, wherein gray model is for steady
Time-delay series precision of prediction it is higher, Leman neural network is good for Nonlinear Delay sequence prediction effect；Gray model is weak
Change the randomness of original time-delay series and the regularity of cumulative sequence, non-linear with Leman neural network combines, and can make
Prediction effect is more accurate.
3) one embodiment of the invention provide Leman neural network using dynamic back propagation algorithm to network into
Row training, is adjusted the weight of each layer of network, and the mean square error that sample exports and identification exports can be made to reach minimum.
Detailed description of the invention
Fig. 1 is a kind of flow chart of the method for unmanned plane sensor fault diagnosis of the invention；
Fig. 2 is a kind of schematic diagram of the method for unmanned plane sensor fault diagnosis of the invention；
Fig. 3 is the comparison of the prediction time-delay series and practical time-delay series of gray model of the invention；
Fig. 4 is the comparison of the prediction time-delay series and practical time-delay series of Leman neural network of the invention；
Fig. 5 is the comparison of the prediction time-delay series and practical time-delay series of combination forecasting of the invention；
Fig. 6 is the comparison of the prediction time-delay series and practical time-delay series of LS-SVM of the invention；
Fig. 7 is the comparison of the prediction time-delay series and practical time-delay series of AR algorithm of the invention；
Fig. 8 is fault diagnosis at stuck 0 ° of one embodiment of the invention；
Fig. 9 is 3 ° of permanent deviation fault diagnosis of another embodiment of the present invention.
Specific embodiment
Below in conjunction with the drawings and specific embodiments to a kind of method of unmanned plane sensor fault diagnosis proposed by the present invention
It is described in further detail.According to following explanation and claims, advantages and features of the invention will be become apparent from.
Embodiment 1
Referring to Fig. 1, Fig. 2, Fig. 1 are the flow chart of the method for unmanned plane sensor fault diagnosis provided by the invention, are based on
Gray model and Leman neural network prediction model, comprising the following steps:
S1: the real boat data of unmanned plane medium-rate gyro are obtained, using real boat data as the first time delay sample sequence；
S2: gray model carries out modeling and forecasting to the first time delay sample sequence, obtains prediction mould using minimum variance criteria
Type；
S3: the weight of the hidden layers of Leman neural network regulating networks, input layer and output layer；
S4: in conjunction with step S2 and step S3, combination latency prediction model is obtained；
S5: combination latency prediction model is applied in unmanned plane sensor fault diagnosis, diagnosis unmanned plane sensor
Failure.
It is appreciated that nothing of the gray model to acquisition in the method for unmanned plane sensor fault diagnosis provided in this embodiment
The real boat data of man-machine medium-rate gyro carry out modeling and forecasting, and obtain prediction model according to minimum variance criteria, then Leman
The weight of the hidden layers of neural network regulating networks, input layer and output layer, then by the method for gray model modeling and forecasting
In conjunction with the method that Leman neural network is adjusted, combination latency prediction model is obtained, is finally applied to combination latency prediction model
In unmanned plane sensor fault diagnosis, the failure of the unmanned plane sensor is diagnosed.This method can to unmanned plane sensor into
Row failure predication and diagnosis.
Above-mentioned steps specifically:
The real boat data of rate gyroscope described in choosing 500 groups, wherein when data of navigating in fact described in 400 groups are as described first
Prolong sample sequence, data of navigating in fact described in remaining 100 groups are used for the verifying of the prediction model, set the first time delay sample sequence
It is classified as X(0):
X(0)=(x(0)(1),x(0)(2),x(0)(3),…x(0)(n)) wherein,
x(0)(i), i=1,2,3 ... n are the first time delay sample sequence at i moment, and n is the number of time delay sample data；
Accumulation process is carried out to the first time delay sample sequence and obtains the second time delay sample sequence X(1),
X(1)=(x(1)(1),x(1)(2),x(1)(3),…x(1)(n)) wherein,
x(1)(i), i=1,2,3 ... n is the second time delay sample sequence at i moment after carrying out accumulation process.
Obtain single argument single order gray model are as follows:
Wherein, coefficient a is development coefficient, and coefficient b is that grey acts on coefficient of discharge；
The time delay queue length of gray model is taken as 20,
Order matrixWhen according to described first
Prolong sample sequence and the second time delay sample sequence, obtains coefficient a and coefficient b.
Coefficient is acquired finally by least square methodWherein T indicates transposition.
Then prediction model is obtained:
Leman neural network is by input layer, hidden layer, undertaking layer and output layer.The number of plies of input layer is taken as 20, implies
The number of plies of layer be taken as 10 and the number of plies of output layer be taken as 1；Leman neural network is using dynamic back propagation algorithm to net
Network is trained, and the weight of each layer of network is adjusted, and the mean square error that sample exports and identification exports is made to reach minimum.
If network-external list entries is u (t), accepting layer output is yc(t), the output of network is y (t), w1To accept layer
Connection weight matrix, w to hidden layer2For the connection weight matrix and w of input layer to hidden layer3For the company of hidden layer to output layer
Connect weight matrix；
Wherein,
Wherein i=1,2 ... m；J=1,2 ..., n；Q=1,2 ... .r；η1,η2,η3It is to accept layer to hidden layer respectively
Connection weight matrix w1Learning Step, input layer to hidden layer connection weight matrix w2Learning Step, hidden layer to output layer
Connection weight matrix w3Learning Step.
Wherein, referring to Fig. 3, Fig. 4, wherein horizontal axis is series number, and vertical pivot is time delay value.By select 400 groups of real data of navigating
Verifying is compared with remaining 100 groups real boat data, error is made to reach smaller.
Further, gray model weakens the randomness of original time-delay series and the regularity combination Leman mind of cumulative sequence
Through the non-linear of network, combination latency prediction model is obtained；
If d (1), d (2) ..., d (n) be the first time delay sample sequence, using the first time delay sample sequence to d (n+1) into
Row prediction, ifWithThe respectively latency prediction value of gray model and Leman neural network,For prediction model
Latency prediction value establishes following combination latency prediction model:
Wherein, γ is the weight coefficient of prediction model；
The error sequence of gray model are as follows:
The error sequence of Leman neural network are as follows:
Seek the variance of error sequence:
D (e)=γ2D(e1)+(1-γ)2D(e2)+2γ(1-γ)Vc(e1,e2)
Pass through solutionIt can then make variance D (e) minimum, then have:
Wherein, γbestFor the optimal weights coefficient of prediction model；
Due to gray model and Leman neural network Independent modeling, it can thus be assumed that error sequence e1,e2Between mutually solely
It is vertical, therefore its covariance V can be enabledc(e1,e2)=0, therefore have following formula
It can finally obtain combination latency prediction model are as follows:
According to the first time delay sample sequence, and then according to formulaIt can be calculated
γbest, and then obtain combination latency prediction model.
Fig. 5 is the comparison of the prediction time-delay series and practical time-delay series of the combination forecasting in the present embodiment, wherein
Horizontal axis is series number, and vertical pivot is time delay value.Gray model is combined with Leman neural network, finds out γbest, make combined prediction
The error of model is smaller.
Finally referring to Fig. 6 and Fig. 7, using LS-SVM, (Least Squares Support Vector Machines is minimum
Two multiply support vector machines) prediction time-delay series in weight coefficient γ and practical time-delay series comparison and AR algorithm it is pre-
The order p and practical time-delay series surveyed in time-delay series is compared.It is found that the combination forecasting in the present embodiment is compared with other methods
With higher precision of prediction.
Gray model is higher for stable time-delay series precision of prediction, the gray model when time delay train wave moves bigger
Precision of prediction can reduce；Leman neural network is good for Nonlinear Delay sequence prediction effect, but is easy to fall into local optimum.
Therefore gray model is weakened to the randomness of original time-delay series and the regularity of cumulative sequence, it is non-thread with Leman neural network
Property combine, combination latency prediction model can be obtained, prediction effect is more accurate.
It is appreciated that gray model and Leman neural network in the present embodiment, wherein gray model is for stable
Time-delay series precision of prediction is higher, and Leman neural network is good for Nonlinear Delay sequence prediction effect；Gray model is weakened
The regularity of the randomness of original time-delay series and cumulative sequence, non-linear with Leman neural network combine, and can make pre-
It is more accurate to survey effect.
Finally combination latency prediction model is applied in unmanned plane sensor fault diagnosis, diagnoses the unmanned plane sensing
The failure of device.
In the sensor fault diagnosis research of unmanned plane height-lock control, as shown in fig. 7, being added in 200S at stuck 0 °
Failure, and combination latency prediction model is applied in the sensor fault diagnosis of unmanned plane, it can be diagnosed to be sensing in time
The stuck failure of device.
Embodiment 2
Based on the same inventive concept, combination latency prediction model is applied to nothing on the basis of embodiment 1 by the present embodiment
In the sensor fault diagnosis research of man-machine horizontal lateral movement, rest part is same as Example 1 or similar, no longer superfluous herein
It states.
Referring to Fig. 9, generally, the state of flight of the horizontal lateral movement of unmanned plane has orientation to fly nonstop to, turn left and turns right.
Now by taking the following roll angle of right-hand bend 10KM gives 30 ° of situations as an example, 3 ° of permanent deviations are added in 250s, to rate gyroscope output
Roll angle carries out real-time monitoring, will will finally combine latency prediction model and be applied in the sensor fault diagnosis of unmanned plane,
In the sensor failure of unmanned plane, the failure of sensor can be detected in time.
Embodiments of the present invention are explained in detail above in conjunction with attached drawing, but the present invention is not limited to above-mentioned implementations
Mode.Even if to the present invention, various changes can be made, if these variations belong to the model of the claims in the present invention and its equivalent technologies
Within enclosing, then still fall within the protection scope of the present invention.
Claims (5)
1. a kind of method of unmanned plane sensor fault diagnosis, which is characterized in that the method is based on gray model and Leman
Neural network prediction model, comprising the following steps:
S1: obtaining the real boat data of unmanned plane medium-rate gyro, using the real boat data as the first time delay sample sequence；
S2: the gray model carries out modeling and forecasting to the first time delay sample sequence, is obtained using minimum variance criteria pre-
Survey model；
S3: the weight of the hidden layers of the Leman neural network regulating networks, input layer and output layer；
S4: in conjunction with the step S2 and the step S3, combination latency prediction model is obtained；
S5: the combination latency prediction model is applied in the unmanned plane sensor fault diagnosis, the unmanned plane is diagnosed
The failure of sensor.
2. the method for unmanned plane sensor fault diagnosis as described in claim 1, which is characterized in that by the unmanned plane middling speed
The reality of rate gyro navigates data as the first time delay sample sequence and includes:
The real boat data of rate gyroscope described in choosing 500 groups, wherein data of navigating in fact described in 400 groups are as the first time delay sample
This sequence, data of navigating in fact described in remaining 100 groups be used for the prediction model verifying, set the first time delay sample sequence as
X(0):
X(0)=(x(0)(1),x(0)(2),x(0)(3),…x(0)(n)) wherein,
x(0)(i), i=1,2,3 ... n are the first time delay sample sequence at i moment, and n is of the time delay sample data
Number；
Accumulation process is carried out to the first time delay sample sequence and obtains the second time delay sample sequence X(1),
X(1)=(x(1)(1),x(1)(2),x(1)(3),…x(1)(n)) wherein,
x(1)(i), i=1,2,3 ... n is the second time delay sample sequence at i moment after carrying out accumulation process.
3. the method for unmanned plane sensor fault diagnosis as claimed in claim 2, which is characterized in that the gray model is to institute
Stating the first time delay sample sequence progress modeling and forecasting includes:
The time delay queue length of the gray model is taken as 20, the gray model first time delay sample sequence described in 400 groups
Modeling and forecasting is carried out, coefficient a and coefficient b are obtained；
Wherein, coefficient a is development coefficient, and coefficient b is that grey acts on coefficient of discharge；
Coefficient a ' and coefficient b ' are found out by minimum variance criteria.
4. the method for unmanned plane sensor fault diagnosis as described in claim 1, which is characterized in that the Leman neural network
The weight of the hidden layers of regulating networks, input layer and output layer includes:
The number of plies of the input layer is 20, the number of plies of the hidden layer is 10 and the number of plies of the output layer is 1；
The hidden layer, the input layer and the output layer is adjusted in the Leman neural network.
5. the method for unmanned plane sensor fault diagnosis as described in claim 3 or 4, which is characterized in that in conjunction with the step
The S2 and step S3, obtaining combination latency prediction model includes:
In conjunction with the step S2 and the step S3, weight coefficient γ is obtained, finally obtain the combination latency prediction model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811582140.9A CN109507989A (en) | 2018-12-24 | 2018-12-24 | A kind of method of unmanned plane sensor fault diagnosis |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811582140.9A CN109507989A (en) | 2018-12-24 | 2018-12-24 | A kind of method of unmanned plane sensor fault diagnosis |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109507989A true CN109507989A (en) | 2019-03-22 |
Family
ID=65754390
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811582140.9A Pending CN109507989A (en) | 2018-12-24 | 2018-12-24 | A kind of method of unmanned plane sensor fault diagnosis |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109507989A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110705133A (en) * | 2019-11-06 | 2020-01-17 | 中国联合网络通信集团有限公司 | Predictive maintenance method and predictive maintenance equipment |
| CN111625762A (en) * | 2020-04-29 | 2020-09-04 | 粤电集团贵州有限公司 | Fan fault diagnosis method |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106054617A (en) * | 2016-08-15 | 2016-10-26 | 江南大学 | Online prediction method for random time delay of forward channel of network control system |
| CN106789214A (en) * | 2016-12-12 | 2017-05-31 | 广东工业大学 | It is a kind of based on the just remaining pair network situation awareness method and device of string algorithm |
| CN108508863A (en) * | 2017-10-12 | 2018-09-07 | 上海智容睿盛智能科技有限公司 | A kind of electromechanical equipment method for diagnosing faults based on gray model |
-
2018
- 2018-12-24 CN CN201811582140.9A patent/CN109507989A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106054617A (en) * | 2016-08-15 | 2016-10-26 | 江南大学 | Online prediction method for random time delay of forward channel of network control system |
| CN106789214A (en) * | 2016-12-12 | 2017-05-31 | 广东工业大学 | It is a kind of based on the just remaining pair network situation awareness method and device of string algorithm |
| CN108508863A (en) * | 2017-10-12 | 2018-09-07 | 上海智容睿盛智能科技有限公司 | A kind of electromechanical equipment method for diagnosing faults based on gray model |
Non-Patent Citations (2)
| Title |
|---|
| 王余伟，曹东，李勇: "基于在线模型的飞控系统传感器的故障诊断", 《电光与控制》 * |
| 王洋、华容: "组合时延预测的无人机传感器故障诊断研究", 《电光与控制》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110705133A (en) * | 2019-11-06 | 2020-01-17 | 中国联合网络通信集团有限公司 | Predictive maintenance method and predictive maintenance equipment |
| CN110705133B (en) * | 2019-11-06 | 2022-09-23 | 中国联合网络通信集团有限公司 | Predictive maintenance method and predictive maintenance equipment |
| CN111625762A (en) * | 2020-04-29 | 2020-09-04 | 粤电集团贵州有限公司 | Fan fault diagnosis method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20230330848A1 (en) | Reinforcement and imitation learning for a task | |
| US10692018B2 (en) | Machine learning device and machine learning method for learning optimal object grasp route | |
| CN110806759B (en) | Aircraft route tracking method based on deep reinforcement learning | |
| US20210064970A1 (en) | Distributional reinforcement learning | |
| US20210201156A1 (en) | Sample-efficient reinforcement learning | |
| US20220366245A1 (en) | Training action selection neural networks using hindsight modelling | |
| US11113605B2 (en) | Reinforcement learning using agent curricula | |
| CN111989696A (en) | Neural network for scalable continuous learning in domains with sequential learning tasks | |
| EP3793786A1 (en) | Self-supervised robotic object interaction | |
| WO2021058583A1 (en) | Training action selection neural networks using q-learning combined with look ahead search | |
| Peng et al. | An improved particle swarm optimization algorithm applied to long short-term memory neural network for ship motion attitude prediction | |
| Ma et al. | Deep auto-encoder observer multiple-model fast aircraft actuator fault diagnosis algorithm | |
| CN109507989A (en) | A kind of method of unmanned plane sensor fault diagnosis | |
| WO2020260067A1 (en) | Error compensation in analog neural networks | |
| CN114677339A (en) | Power transmission line bolt out-of-stock defect detection method introducing attention mechanism | |
| CN106687871A (en) | System and method for controller adaptation | |
| Shi et al. | Wasserstein distance based multi-scale adversarial domain adaptation method for remaining useful life prediction | |
| CN107038450A (en) | Unmanned plane policing system based on deep learning | |
| Zhou et al. | Multihorizons transfer strategy for continuous online prediction of time‐series data in complex systems | |
| Zhang et al. | Network security situation prediction model based on multi-swarm chaotic particle optimization and optimized grey neural network | |
| EP4143745A1 (en) | Training an action selection system using relative entropy q-learning | |
| Zheng et al. | Application of Principal Component Analysis‐Assisted Neural Networks for the Rotor Blade Load Prediction | |
| CN108960406B (en) | MEMS gyroscope random error prediction method based on BFO wavelet neural network | |
| CN113821012B (en) | Fault diagnosis method for variable-working-condition satellite attitude control system | |
| WO2024068789A1 (en) | Learning tasks using skill sequencing for temporally-extended exploration |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190322 |
|
| WD01 | Invention patent application deemed withdrawn after publication |