CN110287594A - A kind of aero-engine method for diagnosing status based on neural network algorithm - Google Patents
A kind of aero-engine method for diagnosing status based on neural network algorithm Download PDFInfo
- Publication number
- CN110287594A CN110287594A CN201910556416.4A CN201910556416A CN110287594A CN 110287594 A CN110287594 A CN 110287594A CN 201910556416 A CN201910556416 A CN 201910556416A CN 110287594 A CN110287594 A CN 110287594A
- Authority
- CN
- China
- Prior art keywords
- neural network
- layer
- network model
- aero
- engine
- 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
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
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation
Abstract
The present invention relates to a kind of aero-engine method for diagnosing status based on neural network algorithm includes the following steps: to obtain object sensing data；The object sensing data input neural network model that will acquire；Neural network model exports the Health Category and probability corresponding with Health Category of object according to the sensing data of input；The object includes but is not limited to any one of whole aero-engine, the anti-asthma control system of aero-engine and gas path component.Diagnostic method provided by the invention can simulate the mathematical model between the health status and each detection parameters of aero-engine, thus Health Category and probability locating for Precise Diagnosis current aerospace engine by complicated deep neural network.
Description
Technical field
The invention belongs to aero-engine fault diagnosis technique field more particularly to a kind of boats based on neural network algorithm
Empty engine condition diagnostic method.
Background technique
Aero-engine is that an extremely complex system or even its internal subsystems also have higher complexity
Degree.The mathematical model of current techniques, our standard difficult to use describes the whole service process of aero-engine, including
The operation health status of aero-engine subsystems.
Summary of the invention
(1) technical problems to be solved
For existing technical problem, the present invention provides a kind of aero-engine state based on neural network algorithm
Diagnostic method can be simulated between the health status and each detection parameters of aero-engine by complicated deep neural network
Mathematical model, thus Health Category and probability locating for Precise Diagnosis current aerospace engine.
(2) technical solution
In order to achieve the above object, the main technical schemes that the present invention uses include:
A kind of aero-engine method for diagnosing status based on neural network algorithm, includes the following steps:
Obtain object sensing data；
The object sensing data input neural network model that will acquire；
Neural network model exports the Health Category of object and corresponding with Health Category according to the sensing data of input
Probability；
The object includes but is not limited to whole aero-engine, the anti-asthma control system of aero-engine and gas circuit portion
Any one of part.
Preferably, the neural network model includes: input layer, hidden layer and output layer；
The input layer and object sensor Data Matching；
The hidden layer quantity is determined by sensing data dimension；
The hidden layer quantity is equal with the relationship of sensing dimension；
The output layer is each health status grade and corresponding probability.
It preferably, further include establishing nerve before the object sensing data that will acquire inputs neural network model
Network model, and the neural network model of foundation is trained；
Object sensing data is Z, and Health Category and probability are S, then training set is combined into { (Z1, S1), (Z2, S2) ... ...,
(Zi, Si) ... ..., (ZT, ST)}。
Preferably, the loss function of each neuron of the neural network model makes using simoid function；
The expression formula of the simoid function are as follows:
Preferably, the output layer of the neural network uses softmax function；
The expression formula of the softmax function are as follows:
Wherein SiIndicate i-th of health status, aiRefer to the output of i-th of neuron of a hidden layer on output layer.
Preferably, the hidden layer of the neural network model includes at least two layers, i.e. first layer and the last layer；
The Health Category number phase of the neuron number with output layer of the last layer in the hidden layer of the neural network model
Together.
Preferably, in the hidden layer of the neural network model in addition to the last layer, every layer of neuronal quantity and input layer Z
Quantity it is equal.
Preferably, the neural network model is the model established using deep neural network algorithm.
(3) beneficial effect
The beneficial effects of the present invention are: a kind of aero-engine state based on neural network algorithm provided by the invention is examined
Disconnected method has the advantages that
The present invention health status of aero-engine can be simulated by complicated deep neural network and each detecting is joined
Mathematical model between number, thus the probability of Health Category locating for Precise Diagnosis current aerospace engine, this probability can either
For judging current aerospace engine health status, Hidden Markov Model or Kalman filtering scheduling algorithm can be used for, into
Row more accurately calculates.
Detailed description of the invention
Fig. 1 is nerve net in a kind of aero-engine method for diagnosing status based on neural network algorithm provided by the invention
The structural schematic diagram of network model；
Fig. 2 is nerve net in a kind of aero-engine method for diagnosing status based on neural network algorithm provided by the invention
The computation model of neuron in network model.
Specific embodiment
In order to preferably explain the present invention, in order to understand, with reference to the accompanying drawing, by specific embodiment, to this hair
It is bright to be described in detail.
The present invention provides a kind of aero-engine method for diagnosing status based on neural network algorithm, including walk as follows
It is rapid:
Obtain object sensing data；
The object sensing data input neural network model that will acquire；
Neural network model exports the Health Category of object and corresponding with Health Category according to the sensing data of input
Probability；
The object includes but is not limited to whole aero-engine, the anti-asthma control system of aero-engine and gas circuit portion
Any one of part.
In detail, the method in the present embodiment can be the failure and health status of entire aero-engine for target,
It is also possible to the failure and health status, such as anti-asthma control system, gas path component etc. of the subsystem of aero-engine；It is described to examine
Disconnected method uses deep neural network algorithm；The input layer of the neural network algorithm is that each sensor of aero-engine is adopted
The data of collection；The depth of the neural network algorithm is hidden number of layers and is determined according to number of sensors；The diagnostic method will
The health status for diagnosing target divides multiple grades；The output layer of the neural network algorithm is each health status grade and phase
The probability answered.
It should be understood that the neural network model includes: input layer, hidden layer and output layer；The input layer and mesh
Mark the matching of object sensing data；The hidden layer quantity is determined by sensing data dimension；The pass of the hidden layer quantity and sensing dimension
System is equal；The output layer is each health status grade and corresponding probability.
Neural network algorithm input layer is that follow-up is stopped navigation the associated sensor data of sky engine target system.
In addition, further including before the object sensing data that will acquire inputs neural network model in the present embodiment
Neural network model is established, and the neural network model of foundation is trained；
Object sensing data is Z, and Health Category and probability are S, then training set is combined into { (Z1, S1), (Z2, S2) ... ...,
(Zi, Si) ... ..., (ZT, ST)}。
The loss function of each neuron of neural network model described in the present embodiment makes using simoid function；
The expression formula of the simoid function are as follows:
Secondly, the output layer of neural network described in the present embodiment uses softmax function；
The expression formula of the softmax function are as follows:
Wherein Si indicates i-th of health status, aiRefer to the output of i-th of neuron of a hidden layer on output layer.
It is noted that it is each strong that the output layer of the neural network is obtained after the output of softmax function
Health shape probability of state, i.e., current sensor data perception to engine state parameters under the aero-engine target system that is judged
The probability of the health status of system.
The hidden layer of neural network model described in the present embodiment includes at least two layers, i.e. first layer and the last layer；
The Health Category number phase of the neuron number with output layer of the last layer in the hidden layer of the neural network model
Together.
In the hidden layer of neural network model described in the present embodiment in addition to the last layer, every layer of neuronal quantity and input
The quantity of layer Z is equal.
Finally, it should be understood that neural network model described in the present embodiment is using deep neural network algorithm
The model of foundation.
Next, what input layer Z was indicated is each sensing data it is noted that in Fig. 1, and Fig. 2 illustrate it is each
The computation model of neuron.A large amount of refreshing hidden layers by network of neuron composition, the neuron number of each hidden layer (removing the last layer)
Amount is equal to the quantity of input layer Z, and the neuron number of the last one hidden layer and the number of levels of health status are identical.By hidden layer
Calculating after, the output of the last one hidden layer is classified using softmax, obtain output result.
As a kind of neural network algorithm, a large amount of training data is also needed for establishing sensing data and health status
Between mathematical model, if sensing data is indicated using Z, health status is indicated using S, then training set be combined into (Z1, S1),
(Z2,S2),…,(Zi,Si),…,(ZT,ST)}.The hidden layer quantity of neural network is determined by sensing data dimension, general to be arranged
To be equal, if relationship is complex, hidden layer quantity can also be increased.The loss function of each neuron of neural network makes
With simoid function:
The output layer of neural network is classified using softmax function:
Wherein SiIndicate i-th of health status, aiRefer to the output of i-th of neuron of a hidden layer on output layer.But nerve net
The classification results of network not instead of output state, the probability of each health status, i.e., current sensor data perception to start
The probability of the health status of the aero-engine goal systems judged under machine state parameter.
The technical principle of the invention is described above in combination with a specific embodiment, these descriptions are intended merely to explain of the invention
Principle shall not be construed in any way as a limitation of the scope of protection of the invention.Based on explaining herein, those skilled in the art
It can associate with other specific embodiments of the invention without creative labor, these modes fall within this hair
Within bright protection scope.
Claims (8)
1. a kind of aero-engine method for diagnosing status based on neural network algorithm, which comprises the steps of:
Obtain object sensing data；
The object sensing data input neural network model that will acquire；
Neural network model exports the Health Category of object and corresponding with Health Category general according to the sensing data of input
Rate；
The object includes but is not limited to that aero-engine is whole, in anti-asthma control system and gas path component of aero-engine
It is any.
2. diagnostic method according to claim 1, which is characterized in that
The neural network model includes: input layer, hidden layer and output layer；
The input layer and object sensor Data Matching；
The hidden layer quantity is determined by sensing data dimension；
The hidden layer quantity is equal with the relationship of sensing dimension；
The output layer is each health status grade and corresponding probability.
3. diagnostic method according to claim 1, which is characterized in that in the object sensing data input mind that will acquire
It further include establishing neural network model, and be trained to the neural network model of foundation before network model；
Object sensing data is Z, and Health Category and probability are S, then training set is combined into { (Z1, S1), (Z2, S2) ... ..., (Zi,
Si) ... ..., (ZT, ST)}。
4. diagnostic method according to claim 2 or 3, which is characterized in that
The loss function of each neuron of the neural network model makes using simoid function；
The expression formula of the simoid function are as follows:
5. diagnostic method according to claim 4, which is characterized in that the output layer of the neural network uses softmax
Function；
The expression formula of the softmax function are as follows:
Wherein SiIndicate i-th of health status, aiRefer to the output of i-th of neuron of a hidden layer on output layer.
6. diagnostic method according to claim 5, which is characterized in that
The hidden layer of the neural network model includes at least two layers, i.e. first layer and the last layer；
The Health Category number of neuron number with output layer of the last layer is identical in the hidden layer of the neural network model.
7. diagnostic method according to claim 6, which is characterized in that
In the hidden layer of the neural network model in addition to the last layer, every layer of neuronal quantity is equal with the quantity of input layer Z.
8. diagnostic method according to claim 7, which is characterized in that the neural network model is using depth nerve net
The model that network algorithm is established.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910556416.4A CN110287594B (en) | 2019-06-25 | 2019-06-25 | Aero-engine state diagnosis method based on neural network algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910556416.4A CN110287594B (en) | 2019-06-25 | 2019-06-25 | Aero-engine state diagnosis method based on neural network algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110287594A true CN110287594A (en) | 2019-09-27 |
| CN110287594B CN110287594B (en) | 2023-06-27 |
Family
ID=68005729
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910556416.4A Active CN110287594B (en) | 2019-06-25 | 2019-06-25 | Aero-engine state diagnosis method based on neural network algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110287594B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111259993A (en) * | 2020-03-05 | 2020-06-09 | 沈阳工程学院 | Fault diagnosis method and device based on neural network |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107886168A (en) * | 2017-11-07 | 2018-04-06 | 歌拉瑞电梯股份有限公司 | One kind carries out elevator faults using multilayer perceptron neutral net and knows method for distinguishing |
-
2019
- 2019-06-25 CN CN201910556416.4A patent/CN110287594B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107886168A (en) * | 2017-11-07 | 2018-04-06 | 歌拉瑞电梯股份有限公司 | One kind carries out elevator faults using multilayer perceptron neutral net and knows method for distinguishing |
Non-Patent Citations (1)
| Title |
|---|
| 蒋勇: "基于一致性融合和神经网络的航空发动机气路系统故障诊断方法", 《航空维修与工程》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111259993A (en) * | 2020-03-05 | 2020-06-09 | 沈阳工程学院 | Fault diagnosis method and device based on neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110287594B (en) | 2023-06-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US6240343B1 (en) | Apparatus and method for diagnosing an engine using computer based models in combination with a neural network | |
| CN112131760A (en) | CBAM model-based prediction method for residual life of aircraft engine | |
| CN108801387B (en) | System and method for measuring remaining oil quantity of airplane fuel tank based on learning model | |
| CN109709934B (en) | Fault diagnosis redundancy design method for flight control system | |
| CN110348752B (en) | Large industrial system structure safety assessment method considering environmental interference | |
| CN108256173A (en) | A kind of Gas path fault diagnosis method and system of aero-engine dynamic process | |
| CN105572572A (en) | WKNN-LSSVM-based analog circuit fault diagnosis method | |
| CN109635318A (en) | A kind of aero-engine sensor intelligent analytic redundancy design method based on KEOS-ELM algorithm | |
| US20200066062A1 (en) | Method and system for vehicle analysis | |
| CN109323754A (en) | A kind of train wheel polygon fault diagnosis detection method | |
| CN106339720B (en) | A kind of abatement detecting method of automobile engine | |
| CN109813542A (en) | The method for diagnosing faults of air-treatment unit based on production confrontation network | |
| CN102567782A (en) | Neural-network-based automobile engine torque estimation method | |
| CN108427400A (en) | A kind of aircraft airspeed pipe method for diagnosing faults based on neural network Analysis design | |
| US6292738B1 (en) | Method for adaptive detection of engine misfire | |
| CN106939840A (en) | Method and apparatus for determining the gas mass flow in internal combustion engine | |
| CN116625686A (en) | On-line diagnosis method for bearing faults of aero-engine | |
| CN108982096A (en) | Industrial robot crank axle wear detecting method based on heuristic rule system | |
| CN114357372A (en) | Aircraft fault diagnosis model generation method based on multi-sensor data driving | |
| CN110287594A (en) | A kind of aero-engine method for diagnosing status based on neural network algorithm | |
| CN105373112A (en) | A steering engine fault detection and diagnosis method based on multi-model parameter estimation | |
| CN111079348B (en) | Method and device for detecting slowly-varying signal | |
| CN109581194B (en) | Dynamic generation method for electronic system fault test strategy | |
| CN109855878A (en) | Computer-readable medium, engine failure detection device and ship | |
| CN115563868A (en) | Fault diagnosis method and device for oil circuit system of aviation alternating-current generator |
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 | ||
| GR01 | Patent grant |