CN107168292B - Submarine navigation device circuit failure diagnosis method based on ELM algorithm - Google Patents
Submarine navigation device circuit failure diagnosis method based on ELM algorithm Download PDFInfo
- Publication number
- CN107168292B CN107168292B CN201710475045.8A CN201710475045A CN107168292B CN 107168292 B CN107168292 B CN 107168292B CN 201710475045 A CN201710475045 A CN 201710475045A CN 107168292 B CN107168292 B CN 107168292B
- Authority
- CN
- China
- Prior art keywords
- navigation device
- device circuit
- sample data
- submarine navigation
- failure
- 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.)
- Active
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
- G05B23/0243—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 model based detection method, e.g. first-principles knowledge model
-
- 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
- G05B2219/00—Program-control systems
- G05B2219/20—Pc systems
- G05B2219/24—Pc safety
- G05B2219/24065—Real time diagnostics
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Automation & Control Theory (AREA)
- Testing And Monitoring For Control Systems (AREA)
- Medicines Containing Antibodies Or Antigens For Use As Internal Diagnostic Agents (AREA)
Abstract
The invention belongs to fault diagnosis field, it is specifically related to the submarine navigation device circuit failure diagnosis method based on ELM algorithm, comprising: the phenomenon of the failure and failure cause for collecting submarine navigation device circuit establish sample data set;According to sample data set, learning network is trained using extreme learning machine ELM algorithm, obtains learning model;New data are inputted to learning model;Using nearest neighbor matching algorithm, the output result of learning model is matched with known fault type;Submarine navigation device circuit is diagnosed according to matching result.The present invention can be in the case where sample be less, to faster and accurately diagnosing to submarine navigation device circuit.
Description
Technical field
The invention belongs to fault diagnosis field, it is specifically related to the submarine navigation device circuit fault diagnosis based on ELM algorithm
Method.
Background technique
In recent years, fault diagnosis field is quickly grown, and various method for diagnosing faults emerge one after another, the accuracy of fault diagnosis
Also it steps up.Fault diagnosis, new-energy automobile fault diagnosis, unmanned vehicle fault diagnosis and expansible
Method for diagnosing faults etc..
In the prior art, traditional diagnostic method is using traditional artificial neural network, and the hidden node of network is joined
Number is to carry out repeatedly optimizing and finally determining by certain iterative algorithm.These iterative steps often make the training of parameter
Process occupies a large amount of time.Also, existing algorithm is easy to generate locally optimal solution, to make the efficiency of network training process
It cannot be guaranteed.
Current is less for submarine navigation device circuit failure diagnosis method, and in the diagnostic method of existing minority,
Processing speed is excessively slow or fault sample demand is excessive or cost is too high, and goes to reduce without suitable method
The influence power of interference factor.
Summary of the invention
The technical problem to be solved in the present invention is that overcoming the deficiencies of the prior art and provide the underwater boat based on ELM algorithm
Row device circuit failure diagnosis method, can be in the case where sample be less, to faster and accurately to submarine navigation device electricity
Road is diagnosed.
Submarine navigation device circuit failure diagnosis method according to the present invention based on ELM algorithm, comprising the following steps:
The phenomenon of the failure and failure cause for collecting submarine navigation device circuit, establish sample data set;
According to sample data set, learning network is trained using ELM algorithm, obtains learning model;
New data are inputted to learning model;
Using nearest neighbor matching algorithm, the output result of learning model is matched with known fault type;
Submarine navigation device circuit is diagnosed according to matching result.
Further, described that learning network is trained using ELM algorithm, further include later,
Using nearest neighbor matching algorithm, by the output result of learning network in training and the progress of known fault type
Match, obtains the error amount of learning model;
Whether within the allowable range to judge the error amount of learning model:
If the error amount of learning model is within the allowable range, enters and input new data to learning model;
If the error amount of learning model within the allowable range, is not returned and is trained using ELM algorithm to learning network.
Further, the phenomenon of the failure and failure cause for collecting submarine navigation device circuit, establishes sample data set,
It specifically includes:
Collect the phenomenon of the failure and failure cause of submarine navigation device circuit;
The phenomenon of the failure and failure cause are equivalent to phenomenon sample data and reason sample data respectively;
Standard normalized is carried out to the phenomenon sample data and center of circle sample data, respectively obtains phenomenon of the failure number
According to collection and failure cause data set;
Phenomenon of the failure data set and failure cause data set are constituted into sample data set.
Still further, it is described according to sample data set, learning network is trained using ELM algorithm, is learnt
Model, comprising:
If sample data set is (X, J), the X representing fault phenotype data collection, J representing fault reason data collection;
It is (X, J) according to sample data set, calculates the output valve of learning network;
The output equation of the learning network is constructed, the output equation is learning model.
Also further, the formula of standard normalized is carried out to the phenomenon sample data and center of circle sample data
Are as follows:
In formula (1), μ is the mean value of sample data, and σ is the standard deviation of sample data;That is:
Further, the output valve for calculating learning network, formula are as follows:
In formula (4), βiFor the weight of output valve;G(ai, bi, X) indicate i-th of hidden node output;A hereini,
biCenter and impact factor for i-th radial basis function node.
In the above-mentioned technical solutions, the G (ai, bi, X) and it is acquired by following equation:
G(ai, bi, X) and=g (bi||X-ai||) (5)。
Preferably, the output equation of the learning network are as follows:
H β=J+E (6)
In formula (6),
Wherein, ai, biValue given at random in [- 1,1] section, β is in a given at randomi, biOn the basis of it is optimal
Export weight.
Preferably,
The formula of the nearest neighbor matching algorithm are as follows:
s0(k)-si(k)/range (s (k)) indicates k-th of the index and known fault type of current learning network output
The dissimilar degree of k-th of index, i.e. matching result;ωkIndicate the current criteria specific gravity shared when evaluating fault type.
Preferably, described that submarine navigation device circuit is diagnosed according to matching result, it specifically includes:
If E (s0, si, W) >=0.9, decide that current failure type is the known fault type of current matching.
In the present invention, firstly, arithmetic speed is fast using ELM algorithm, the number of nodes of demand is less, solves tradition
Algorithm problem at high cost;Secondly, carrying out reasonable to the output and fault type of ELM using nearest neighbor matching algorithm
Match, reduce the influence of interference factor, improves matched accuracy.
ELM algorithm is grown up on the basis of neural networks with single hidden layer, and neural networks with single hidden layer is greatly improved
Data processing speed.ELM algorithm is one of neural network research algorithm, is a kind of extensive single hidden layer Feedforward Neural Networks
Network.
The hidden node parameter of ELM algorithm randomly selects, in the training process without adjusting, it is only necessary to which setting is implicit
The number of layer neuron, can obtain unique optimal solution;And the outer power (i.e. output weight) of network is flat by minimizing
The least square solution that square loss function obtains.It is not necessarily to any iterative step during the determination of network parameter in this way, thus significantly
Reduce the regulating time of network parameter.Compared with traditional training method, this method is fast with pace of learning, Generalization Capability is good
The advantages that.
Detailed description of the invention
Present invention will be further explained below with reference to the attached drawings and examples, in attached drawing:
Fig. 1 is the frame diagram of present invention method;
Fig. 2 is the method flow diagram of the embodiment of the present invention.
Specific embodiment
For a clearer understanding of the technical characteristics, objects and effects of the present invention, now control attached drawing is described in detail
A specific embodiment of the invention.
As depicted in figs. 1 and 2, the submarine navigation device circuit event of the present invention based on ELM (extreme learning machine) algorithm
Hinder diagnostic method, which comprises the following steps:
101, the phenomenon of the failure and failure cause for collecting submarine navigation device circuit, establish sample data set;
1011, the phenomenon of the failure and failure cause of submarine navigation device circuit are collected;
1012, the phenomenon of the failure and failure cause are equivalent to phenomenon sample data and reason sample data respectively;
1013, standard normalized is carried out to the phenomenon sample data and center of circle sample data, respectively obtains failure
Phenotype data collection and failure cause data set;
1014, phenomenon of the failure data set and failure cause data set are constituted into sample data set.
102, according to sample data set, learning network is trained using ELM algorithm, obtains learning model;
201, using nearest neighbor matching algorithm, the output result of learning network in training and known fault type are carried out
Matching, obtains the error amount of learning model;
202, whether within the allowable range to judge the error amount of learning model:
If the error amount of learning model is within the allowable range, enters and input new data to learning model;
If the error amount of learning model within the allowable range, is not returned and is trained using ELM algorithm to learning network.
103, new data are inputted to learning model;
104, using nearest neighbor matching algorithm, the output result of learning model is matched with known fault type;
105, submarine navigation device circuit is diagnosed according to matching result.
It is described according to sample data set, learning network is trained using ELM algorithm, obtains learning model, comprising:
If sample data set is (X, J), the X representing fault phenotype data collection, J representing fault reason data collection;
It is (X, J) according to sample data set, calculates the output valve of learning network;
The output equation of the learning network is constructed, the output equation is learning model.
The formula of standard normalized is carried out to the phenomenon sample data and center of circle sample data are as follows:
In formula (1), μ is the mean value of sample data, and σ is the standard deviation of sample data;That is:
The output valve for calculating learning network, formula are as follows:
In formula (4), βiFor the weight of output valve;G(ai, bi, X) indicate i-th of hidden node output;A hereini,
biCenter and impact factor for i-th radial basis function node.
G (ai, bi, X) and it is acquired by following equation:
G(ai, bi, X) and=g (bi||X-ai||) (5)。
Because the neural networks with single hidden layer of building cannot approach data sample with zero error, therefore corresponding to the above output and in fact
It also can be with error between the output of border.Therefore the output equation of the learning network are as follows:
H β=J+E (6)
In formula (6),
Wherein, ai, biValue given at random in [- 1,1] section, β is in a given at randomi, biOn the basis of it is optimal
Export weight.
The formula of the nearest neighbor matching algorithm are as follows:
s0(k)-si(k)/range (s (k)) indicates k-th of the index and known fault type of current learning network output
The dissimilar degree of k-th of index, i.e. matching result;ωkIndicate the current criteria specific gravity shared when evaluating fault type.
If E (s0, si, W) >=0.9, decide that current failure type is the known fault type of current matching.
The present invention can be used for the fault diagnosis for AC distribution plate in submarine navigation device circuit.Beneficial effect is: 1.
Limit of utilization learning machine (ELM) algorithm forms diagnostic network, and required sample size is few, and calculating speed is fast, avoids similar nerve net
A large amount of iterative calculation of network and parameter setting;2. utilizing nearest neighbor matching algorithm, successful match rate is higher;3. by visual
Breakdown judge has carried out quantification treatment, to infer its fault type by machine algorithm.
The embodiment of the present invention is described with above attached drawing, but the invention is not limited to above-mentioned specific
Embodiment, the above mentioned embodiment is only schematical, rather than restrictive, those skilled in the art
Under the inspiration of the present invention, without breaking away from the scope protected by the purposes and claims of the present invention, it can also make very much
Form, all of these belong to the protection of the present invention.
Claims (8)
1. a kind of submarine navigation device circuit failure diagnosis method based on ELM algorithm, which comprises the following steps:
The phenomenon of the failure and failure cause for collecting submarine navigation device circuit, establish sample data set;
According to sample data set, learning network is trained using extreme learning machine ELM algorithm, obtains learning model;
New data are inputted to learning model;
Using nearest neighbor matching algorithm, the output result of learning model is matched with known fault type;
Submarine navigation device circuit is diagnosed according to matching result;
It is described that learning network is trained using ELM algorithm, further include later,
Using nearest neighbor matching algorithm, the output result of learning network in training is matched with known fault type, is obtained
To the error amount of learning model;
Whether within the allowable range to judge the error amount of learning model:
If the error amount of learning model is within the allowable range, enters and input new data to learning model;
If the error amount of learning model within the allowable range, is not returned and is trained using ELM algorithm to learning network;
The phenomenon of the failure and failure cause for collecting submarine navigation device circuit, establishes sample data set, specifically includes:
Collect the phenomenon of the failure and failure cause of submarine navigation device circuit;
The phenomenon of the failure and failure cause are equivalent to phenomenon sample data and reason sample data respectively;
Standard normalized is carried out to the phenomenon sample data and center of circle sample data, respectively obtains phenomenon of the failure data set
With failure cause data set;
Phenomenon of the failure data set and failure cause data set are constituted into sample data set.
2. the submarine navigation device circuit failure diagnosis method according to claim 1 based on ELM algorithm, which is characterized in that
It is described according to sample data set, learning network is trained using ELM algorithm, obtains learning model, comprising:
If sample data set is (X, J), the X representing fault phenotype data collection, J representing fault reason data collection;
It is (X, J) according to sample data set, calculates the output valve of learning network;
The output equation of the learning network is constructed, the output equation is learning model.
3. the submarine navigation device circuit failure diagnosis method according to claim 2 based on ELM algorithm, which is characterized in that
The formula of standard normalized is carried out to the phenomenon sample data and center of circle sample data are as follows:
(1)
In formula (1),For the mean value of sample data,For the standard deviation of sample data;That is:
(2)
(3).
4. the submarine navigation device circuit failure diagnosis method according to claim 3 based on ELM algorithm, which is characterized in that
The output valve for calculating learning network, formula are as follows:
(4)
In formula (4),For the weight of output valve;Indicate the output of i-th of hidden node;
HereinCenter and impact factor for i-th radial basis function node.
5. the submarine navigation device circuit failure diagnosis method according to claim 4 based on ELM algorithm, which is characterized in that
It is describedIt is acquired by following equation:
(5).
6. the submarine navigation device circuit failure diagnosis method according to claim 5 based on ELM algorithm, which is characterized in that
The output equation of the learning network are as follows:
(6)
In formula (6),
(7)
(8)
(9)
(10)
Wherein,Value given at random in [- 1,1] section, β be at random giveBasis
On optimal output weight.
7. the submarine navigation device circuit failure diagnosis method according to claim 6 based on ELM algorithm, which is characterized in that
The formula of the nearest neighbor matching algorithm are as follows:
(11)
Indicate k-th of index of current learning network output with
Know the dissimilar degree of k-th of index of fault type, i.e. matching result;Indicate current criteria in evaluation fault type when institute
The specific gravity accounted for.
8. the submarine navigation device circuit failure diagnosis method according to claim 7 based on ELM algorithm, which is characterized in that
It is described that submarine navigation device circuit is diagnosed according to matching result, it specifically includes:
If, decide that current failure type is the known fault type of current matching.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710475045.8A CN107168292B (en) | 2017-06-21 | 2017-06-21 | Submarine navigation device circuit failure diagnosis method based on ELM algorithm |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710475045.8A CN107168292B (en) | 2017-06-21 | 2017-06-21 | Submarine navigation device circuit failure diagnosis method based on ELM algorithm |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107168292A CN107168292A (en) | 2017-09-15 |
CN107168292B true CN107168292B (en) | 2019-10-25 |
Family
ID=59819083
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710475045.8A Active CN107168292B (en) | 2017-06-21 | 2017-06-21 | Submarine navigation device circuit failure diagnosis method based on ELM algorithm |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107168292B (en) |
Families Citing this family (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108594788B (en) * | 2018-03-27 | 2020-09-22 | 西北工业大学 | Airplane actuator fault detection and diagnosis method based on deep random forest algorithm |
CN109001073A (en) * | 2018-06-01 | 2018-12-14 | 东北大学 | A kind of Novel pulp densimeter based on ELM network and a kind of measurement method |
CN111078456B (en) * | 2019-12-26 | 2023-05-16 | 新奥数能科技有限公司 | Device fault diagnosis method, device, computer readable storage medium and electronic device |
CN111796519B (en) * | 2020-06-14 | 2022-05-06 | 武汉理工大学 | Automatic control method of multi-input multi-output system based on extreme learning machine |
CN114811443B (en) * | 2021-01-22 | 2024-04-16 | 北京科益虹源光电技术有限公司 | Waterway state monitoring system |
CN113283292B (en) * | 2021-04-13 | 2023-08-08 | 江苏大学镇江流体工程装备技术研究院 | Method and device for diagnosing faults of underwater micro-propeller |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1851555A (en) * | 2006-05-19 | 2006-10-25 | 李鹏 | Method for realizing two-dimensional panoramic true imaging |
CN101246509A (en) * | 2008-02-28 | 2008-08-20 | 上海交通大学 | Entire car personalized configuration system based on case illation technology |
CN102221363A (en) * | 2011-04-12 | 2011-10-19 | 东南大学 | Fault-tolerant combined method of strapdown inertial integrated navigation system for underwater vehicles |
CN103983453A (en) * | 2014-05-08 | 2014-08-13 | 南京航空航天大学 | Differentiating method of fault diagnosis of executing mechanism of aeroengine and sensor |
CN104503420A (en) * | 2014-09-19 | 2015-04-08 | 北京化工大学 | Non-linear process industry fault prediction method based on novel FDE-ELM and EFSM |
CN105242205A (en) * | 2015-09-06 | 2016-01-13 | 南京航空航天大学 | Aviation three-level AC power generator rotary rectifier online fault diagnosis method |
CN105606914A (en) * | 2015-09-06 | 2016-05-25 | 南京航空航天大学 | IWO-ELM-based Aviation power converter fault diagnosis method |
CN106597260A (en) * | 2016-12-29 | 2017-04-26 | 合肥工业大学 | Simulation circuit fault diagnosis method based on continuous wavelet analysis and ELM network |
-
2017
- 2017-06-21 CN CN201710475045.8A patent/CN107168292B/en active Active
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1851555A (en) * | 2006-05-19 | 2006-10-25 | 李鹏 | Method for realizing two-dimensional panoramic true imaging |
CN101246509A (en) * | 2008-02-28 | 2008-08-20 | 上海交通大学 | Entire car personalized configuration system based on case illation technology |
CN102221363A (en) * | 2011-04-12 | 2011-10-19 | 东南大学 | Fault-tolerant combined method of strapdown inertial integrated navigation system for underwater vehicles |
CN103983453A (en) * | 2014-05-08 | 2014-08-13 | 南京航空航天大学 | Differentiating method of fault diagnosis of executing mechanism of aeroengine and sensor |
CN104503420A (en) * | 2014-09-19 | 2015-04-08 | 北京化工大学 | Non-linear process industry fault prediction method based on novel FDE-ELM and EFSM |
CN105242205A (en) * | 2015-09-06 | 2016-01-13 | 南京航空航天大学 | Aviation three-level AC power generator rotary rectifier online fault diagnosis method |
CN105606914A (en) * | 2015-09-06 | 2016-05-25 | 南京航空航天大学 | IWO-ELM-based Aviation power converter fault diagnosis method |
CN106597260A (en) * | 2016-12-29 | 2017-04-26 | 合肥工业大学 | Simulation circuit fault diagnosis method based on continuous wavelet analysis and ELM network |
Also Published As
Publication number | Publication date |
---|---|
CN107168292A (en) | 2017-09-15 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107168292B (en) | Submarine navigation device circuit failure diagnosis method based on ELM algorithm | |
CN108038300B (en) | Optical fiber state evaluation method based on improved membership function combined with neural network | |
CN110929918B (en) | 10kV feeder fault prediction method based on CNN and LightGBM | |
CN108197648A (en) | A kind of Fault Diagnosis Method of Hydro-generating Unit and system based on LSTM deep learning models | |
CN103971160B (en) | particle swarm optimization method based on complex network | |
CN104809722B (en) | A kind of fault diagnosis method based on infrared thermal imagery | |
CN104376389B (en) | Master-slave mode microgrid power load prediction system and method based on load balancing | |
CN107169610A (en) | Aquaculture dissolved oxygen prediction method and device | |
CN107316099A (en) | Ammunition Storage Reliability Forecasting Methodology based on particle group optimizing BP neural network | |
CN108665112A (en) | Photovoltaic fault detection method based on Modified particle swarm optimization Elman networks | |
CN105956702A (en) | Aquaculture water quality dissolved oxygen prediction method based on data fusion | |
CN111445010B (en) | Distribution network voltage trend early warning method based on evidence theory fusion quantum network | |
CN105427241B (en) | Distortion correction method for large-view-field display equipment | |
CN102567782A (en) | Neural-network-based automobile engine torque estimation method | |
CN112130086B (en) | Method and system for predicting remaining life of power battery | |
EP3871938A1 (en) | Method and device for determining pavement rating, storage medium and automobile | |
CN106647272A (en) | Robot route planning method by employing improved convolutional neural network based on K mean value | |
CN109858611A (en) | Neural network compression method and relevant device based on channel attention mechanism | |
CN109934422A (en) | Neural network wind speed prediction method based on time series data analysis | |
CN109239603A (en) | A kind of extreme learning machine under manifold regularization frame predicts power battery SOC method | |
CN108985313A (en) | AGC system discrimination method based on big data and Bayesian neural network | |
CN105568732A (en) | Disc mill control method | |
CN105389614B (en) | A kind of implementation method of neutral net self refresh process | |
CN106599541B (en) | A kind of structure and parameter on-line identification method of dynamic power load model | |
CN106529816B (en) | The method of adjustment and system of power line channel |
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 |