CN110097143A - A kind of Fault Diagnosis of Gear Case method based on fish-swarm algorithm Optimizing BP Network - Google Patents
A kind of Fault Diagnosis of Gear Case method based on fish-swarm algorithm Optimizing BP Network Download PDFInfo
- Publication number
- CN110097143A CN110097143A CN201910455439.6A CN201910455439A CN110097143A CN 110097143 A CN110097143 A CN 110097143A CN 201910455439 A CN201910455439 A CN 201910455439A CN 110097143 A CN110097143 A CN 110097143A
- Authority
- CN
- China
- Prior art keywords
- fish
- artificial fish
- network
- visual
- neural network
- 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.)
- Withdrawn
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- 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
- G06N3/084—Backpropagation, e.g. using gradient descent
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Maintenance And Management Of Digital Transmission (AREA)
Abstract
The invention discloses a kind of Fault Diagnosis of Gear Case methods based on fish-swarm algorithm Optimizing BP Network.For the mechanism of ammonium fixation in two the major parameter visuals field and step-length in standard intraocular's fish-swarm algorithm, the present invention first improves original fish-swarm algorithm, propose a kind of dynamic self-adapting artificial fish-swarm algorithm (DLAFSA), recycle the initial weight and threshold value of dynamic self-adapting artificial fish school algorithm BP neural network, DLAFSA-BP network model is constructed, and is applied to Fault Diagnosis of Gear Case.The present invention has given full play to the global optimizing ability of algorithm operation artificial fish-swarm algorithm early period and the local search ability that later period BP neural network is fine, can more acurrate, reliably identify the failure of gear-box.Those skilled in the art only needs for this method to be used for collected gear-box vibration data, can judge fault type, while solving the problems, such as that the precision of conventional gearbox fault diagnosis technology is low, poor reliability.
Description
Technical field
The present invention relates to fault diagnosis technology field, especially a kind of gear-box event based on fish-swarm algorithm Optimizing BP Network
Hinder diagnostic method.
Background technique
Gear-box as a kind of actuating unit, be one of with widest mechanical equipment in real life, such as
Wind-driven generator, high speed motor car and space equipment etc., gear-box are wherein play extremely critical role.Due in gear-box
Structure is complicated in portion, and working environment is severe, and the processing technologys such as components therein such as gear, bearing are complicated, assembly precision requirement
Height, and the often continuous work under high-speed overload are the main reason for inducing mechanical disorder, in some instances it may even be possible to will lead to entire machinery
The paralysis of system causes the loss that can not be retrieved.Therefore, the operating status of Accurate Diagnosis gear-box whether health and fault bit
It sets significant.For the troubleshooting issue of gear-box, currently, domestic and foreign scholars propose the method for many fault diagnosises,
From the time domain of early stage, frequency domain analysis Time-frequency Analysis finally and present machine learning field.North China Electric Power University
Dragon's fountain etc. benefit is proposed in paper " the gearbox of wind turbine method for diagnosing faults based on particle group optimizing BP neural network "
With the weight and threshold value of particle swarm algorithm Optimized BP Neural Network, and by the network after optimization be used for gearbox of wind turbine therefore
Barrier diagnosis, this method lack the dynamic regulation of speed, fall into locally optimal solution easily to reduce diagnostic accuracy.Institutes Of Technology Of He'nan
Liu Jingyan proposed in paper " application of the genetic neural network in Fault Diagnosis of Gear Case " it is a kind of based on genetic algorithm
The Fault Diagnosis of Gear Case method of Optimized BP Neural Network, this method have dependence, and operation for the selection of initial population
Amount is big.
These method for diagnosing faults are all inevitably present respective defect, limit fault diagnosis technology in engineering
With the application in actual production.It can be seen that the problem of precision of existing Gear Box Fault Diagnosis Technology is low, poor reliability is urgently
It solves.
Summary of the invention
For the disadvantages described above of the prior art, higher, good reliability that the main purpose of the present invention is to provide a kind of precision
Fault Diagnosis of Gear Case method.
To achieve the above object, the present invention provides a kind of Fault Diagnosis of Gear Case based on fish-swarm algorithm Optimizing BP Network
Method, comprising the following steps:
1) using the vibration signal of acceleration transducer acquisition gear-box speed change system, normal, retainer fracture event is obtained
Vibration under five kinds of barrier, broken teeth failure, bearing outer ring failure, combined fault (existing simultaneously gear tooth breakage and outer ring crackle) states
Data;
2) Fault characteristic parameters are extracted, feature ginseng is extracted according to sensibility of each index to five kinds of gear-box different operating conditions
Number, and be normalized;
3) using said extracted and make the characteristic parameter of normalized as dynamic self-adapting artificial fish school algorithm
Five kinds of states of the input of BP neural network (DLAFSA-BP), gear-box are exported as network;
4) Fault Diagnosis of Gear Case result is obtained using DLAFSA-BP network model.
The improvement content of standard intraocular's fish-swarm algorithm is specific as follows:
1) mechanism of ammonium fixation in two major parameter visuals field and step-length in standard intraocular's fish-swarm algorithm is adjusted to a kind of dynamic
State mechanism is adjusted parameter visual using the dynamic changeability of Logistic model, if the variation model of visual field visual
It encloses for [visualmin, visualmax], the variation of parameter visual is set as formula:
The then calculation formula of visual are as follows:
Wherein, t is the number of iterations of algorithm;α is initial attenuation speed, and α value is bigger, and visual decrease speed is faster.
Step-length Step is set with the dynamic self-adapting changing rule such as following formula of visual field visual:
Step=β * Visual (3)
Wherein, β is view step coefficient 0 < β < 1.
2) increase bulletin board in improved fish-swarm algorithm, ignore original crowding factor.It is recorded in bulletin board each
Artificial Fish executes the highest Artificial Fish state of food concentration after a behavior, is always ensured that Artificial Fish in bulletin board in iterative process
State be optimal.
Shoal of fish behavior representation formula of bunching in conjunction with bulletin board is as follows:
Wherein, rand indicates a random number between [0,1];XcWhen indicating the t times iteration, n (t) neighbours in neighborhood
The center of fish;Ggbx indicates the state of Artificial Fish in bulletin board;Xnext(i) next state of i-th Artificial Fish is indicated.
X in formula (4)cCalculation formula it is as follows:
Wherein, XmiIndicate Artificial Fish XiM (0 < m < n (t)) article neighbours fish.
Shoal of fish foraging behavior expression in conjunction with bulletin board is as follows:
Wherein, XmaxIndicate Artificial Fish XiThat highest neighbours fish of food concentration in neighbour structure.
Using the adaptive dynamic shoal of fish algorithm optimization BP neural network, the specific steps of which are as follows:
1) BP neural network topological structure is determined, the neuron number including the network number of plies and each layer network;
2) according to N Artificial Fish of network structure random initializtion, the initial shoal of fish is formed;
3) parameter of dynamic self-adapting artificial fish-swarm algorithm (DLAFSA) is set, including repeats to explore number try-
Number, maximum number of iterations T, field range [visualmin, visualmax], rate of decay α, depending on walking factor beta;
4) food concentration of every Artificial Fish in the initial shoal of fish is calculated.The state of Artificial Fish is set as the initial of neural network
Weight and threshold value, every Artificial Fish all correspond to a neural network, food of the inverse of neural metwork training error as Artificial Fish
Object concentration;
5) DLAFSA algorithm is run, after algorithm, extracts the state ggbx of Artificial Fish in bulletin board;
6) the state ggbx of Artificial Fish in bulletin board is assigned to BP neural network as initial weight and threshold value;
7) training BP neural network, and fault diagnosis is carried out with trained network.
The state of Artificial Fish is set as to the initial weight and threshold value of neural network, the conduct reciprocal of neural metwork training error
The food concentration of Artificial Fish, then network weight and the searching process of threshold value are as follows:
The state X of Artificial Fish is expressed as an I*J+J+J*K+K dimension by the neural network for being I-J-K to a topological structure
Vector:
X=(w11...wi1,b1,...,w1j...wij,bj,v11...vj1,a1,...,v1k...vjk,ak) (7)
Wherein, w11...wi1Indicate weight of the input layer to first neuron of hidden layer, b1Indicate hidden layer the
The threshold value of one neuron, w1j...wijIndicate the weight between input layer and hidden layer j-th neuron, bjIt indicates
The threshold value of hidden layer j-th neuron, v11...vj1Indicate the power between first neuron of hidden layer neuron and output layer
Value, a1Indicate the threshold value of first neuron of hidden layer, v1k...vjkIndicate hidden layer neuron and k-th of neuron of output layer
Between weight, akIndicate the threshold value of k-th of neuron of hidden layer.BP neural network instruction is set by the food concentration of Artificial Fish
Practice the inverse of error, i.e. Y (X)=1/e (X), then any two Artificial Fish XpAnd XqBetween Euclidean distance calculation formula it is as follows:
Wherein Xp、XqIn element subtract each other in strict accordance with dimension one-to-one correspondence, an Artificial Fish basic act of every executions is refreshing
It is just adjusted through network weight and threshold value primary.
In general, the gear-box event of the present invention based on dynamic self-adapting artificial fish school algorithm BP neural network
Hinder diagnostic method compared with prior art, can achieve the following beneficial effects:
The present invention is proposed for the mechanism of ammonium fixation in two the major parameter visuals field and step-length in the artificial fish-swarm algorithm of standard
A kind of dynamic self-adapting artificial fish-swarm algorithm, improves the speed and precision of algorithm global optimizing;It is examined in the failure of gear-box
Faulted-stage section utilizes the initial weight and threshold value of dynamic self-adapting artificial fish school algorithm BP neural network, the DLAFSA- of building
BP network model can accurately and reliably identify fault type.Those skilled in the art only needs to be used to collect by this method
Gear-box vibration data, can judge fault type.
Detailed description of the invention
Invention is described further with reference to the accompanying drawing:
Fig. 1 is that the Fault Diagnosis of Gear Case of the present invention based on dynamic self-adapting fish-swarm algorithm Optimized BP Neural Network is total
Body flow chart;
Fig. 2 is dynamic self-adapting artificial fish-swarm algorithm flow chart of the present invention;
Fig. 3 is the flow chart of dynamic self-adapting artificial fish school algorithm BP neural network of the present invention;
Fig. 4 is the gear-box event based on dynamic self-adapting artificial fish school algorithm BP neural network of the embodiment of the present invention
Hinder diagnostic result figure.
Specific embodiment
It is clear to be more clear the objectives, technical solutions, and advantages of the present invention, below in conjunction with the drawings and specific embodiments
The present invention will be described in further detail.
Fig. 1 is that the gearbox fault of the present invention based on dynamic self-adapting artificial fish school algorithm BP neural network is examined
Disconnected overview flow chart.Believe as shown in Figure 1, acquiring the vibration under five kinds of operating conditions of gear-box speed change system with acceleration transducer first
Number;Extract input of the Fault characteristic parameters as dynamic self-adapting artificial fish school algorithm BP neural network model;Iteration
After the completion of training, output gear case fault diagnosis result.
Fig. 2 is dynamic self-adapting artificial fish-swarm algorithm flow chart of the present invention.As shown in Fig. 2, the shoal of fish is initialized first,
The food concentration of every Artificial Fish is calculated, and assigns the highest Artificial Fish of food concentration to bulletin board, then executes fish school behavior
Rule updates the value in the state and bulletin board of Artificial Fish, judges whether to meet termination condition, exports bulletin board if meeting
Value, carries out next iteration if being unsatisfactory for, until meeting termination condition.
The mechanism of ammonium fixation in two major parameter visuals field and step-length in standard intraocular's fish-swarm algorithm is adjusted to a kind of dynamic
Mechanism is adjusted parameter visual using the dynamic changeability of Logistic model, if the variation range of visual field visual
For [visualmin, visualmax], the variation of parameter visual is set as formula:
The then calculation formula of visual are as follows:
Wherein, t is the number of iterations of algorithm;α is initial attenuation speed, and α value is bigger, and visual decrease speed is faster.
Step-length Step with visual field visual dynamic self-adapting changing rule such as following formula:
Step=β * Visual (3)
Wherein, β is view step coefficient (0 < β < 1).
Shoal of fish behavior representation formula of bunching in conjunction with bulletin board is as follows:
Wherein, rand indicates a random number between [0,1];XcWhen indicating the t times iteration, n (t) neighbours in neighborhood
The center of fish;Ggbx indicates the state of Artificial Fish in bulletin board;Xnext(i) next state of i-th Artificial Fish is indicated.
X in formula (4)cCalculation formula it is as follows:
Wherein, XmiIndicate Artificial Fish XiM (0 < m < n (t)) article neighbours fish.
Shoal of fish foraging behavior expression in conjunction with bulletin board is as follows:
Wherein, XmaxIndicate Artificial Fish XiThat highest neighbours fish of food concentration in neighbour structure.
Fig. 3 is the flow chart of dynamic self-adapting fish-swarm algorithm Optimized BP Neural Network of the present invention.As shown in figure 3,
DLAFSA algorithm optimization BP neural network the following steps are included:
1) BP neural network topological structure is determined, the neuron number including the network number of plies and each layer network;
2) according to N Artificial Fish of network structure random initializtion, the initial shoal of fish is formed;
3) parameter of improved artificial fish-swarm algorithm (DLAFSA) is set, including repeats to explore number try-number,
Maximum number of iterations T, field range [visualmin, visualmax], rate of decay α, depending on walking factor beta;
4) food concentration of every Artificial Fish in the initial shoal of fish is calculated.The state of Artificial Fish is set as the initial of neural network
Weight and threshold value, every Artificial Fish all correspond to a neural network, food of the inverse of neural metwork training error as Artificial Fish
Object concentration;
5) DLAFSA algorithm is run, after algorithm, extracts the Artificial Fish state ggbx in bulletin board;
6) the state ggbx of Artificial Fish in bulletin board is assigned to BP neural network as initial weight and threshold value;
7) training BP neural network, and fault diagnosis is carried out with the network after training.
In step 4), the state of Artificial Fish is set as to the initial weight and threshold value of neural network, neural metwork training error
Food concentration of the inverse as Artificial Fish, then network weight and the searching process of threshold value are as follows:
The state X of Artificial Fish is expressed as an I*J+J+J*K+K dimension by the neural network for being I-J-K to a topological structure
Vector:
X=(w11...wi1,b1,...,w1j...wij,bj,v11...vj1,a1,...,v1k...vjk,ak) (7)
Wherein, w11...wi1Indicate weight of the input layer to first neuron of hidden layer, b1Indicate hidden layer the
The threshold value of one neuron, w1j...wijIndicate the weight between input layer and hidden layer j-th neuron, bjIt indicates
The threshold value of hidden layer j-th neuron, v11...vj1Indicate the power between first neuron of hidden layer neuron and output layer
Value, a1Indicate the threshold value of first neuron of hidden layer, v1k...vjkIndicate hidden layer neuron and k-th of neuron of output layer
Between weight, akIndicate the threshold value of k-th of neuron of hidden layer.BP neural network instruction is set by the food concentration of Artificial Fish
Practice the inverse of error, i.e. Y (X)=1/e (X), then any two Artificial Fish XpAnd XqBetween Euclidean distance calculation formula it is as follows:
Wherein Xp、XqIn element subtract each other in strict accordance with dimension one-to-one correspondence, an Artificial Fish basic act of every executions is refreshing
It is just adjusted through network weight and threshold value primary.
Using acceleration transducer acquisition gear-box in normal, retainer fracture defect, broken teeth failure, bearing outer ring event
Vibration under five kinds of barrier, retainer fracture defect and combined fault (existing simultaneously gear tooth breakage failure and outer ring crack fault) operating conditions
Dynamic 50 groups of signal, every kind 10 groups of operating condition, wherein 8 groups are used as training sample, 2 groups are used as test sample, extract nargin, kurtosis, partially
Eight coefficient, pulse index, barycenter of frequency spectrum, Spectral variance, harmonic factor and correlation factor characteristic parameters are spent as DLAFSA-BP
The input of network model, output use binary coding representation are as follows: normal (1 000 0), retainer fracture (0 100
0), broken teeth (0 010 0), bearing outer ring crackle (0 001 0), combined fault (0 000 1).
The parameter setting of ADAFSA algorithm: artificial fish-swarm scale N is 50, regard step factor beta as 0.6, and rate of decay α is 100,
Field range [1,8], maximum number of iterations T are 100, repeat to explore number try-number to be 30.
The parameter setting of BP neural network: network topology structure 8-8-5.Hidden layer is adopted respectively with output layer transmission function
With tansig function and logsig function, train function sets for trainlm, frequency of training 1000, training objective 0.0001,
Learning rate 0.05.
Fig. 4 is that the gearbox fault based on dynamic self-adapting artificial fish school algorithm BP neural network is examined in the present embodiment
Disconnected result figure.Can be seen that from Fig. 4 test result can using the BP neural network model after the optimization of dynamic self-adapting fish-swarm algorithm
Accurately to identify gearbox fault type, high reliablity.
In conclusion the above is merely preferred embodiments of the present invention, the protection scope being not intended to limit the invention.
All within the spirits and principles of the present invention, made any modification, equivalent improvement, replacement etc., should all include of the invention
Within protection scope.
Claims (2)
1. a kind of Fault Diagnosis of Gear Case method based on fish-swarm algorithm Optimizing BP Network, which comprises the following steps:
Step 1: obtaining normal, retainer fracture event using the vibration signal of acceleration transducer acquisition gear-box speed change system
Vibration under five kinds of barrier, broken teeth failure, bearing outer ring failure, combined fault (existing simultaneously gear tooth breakage and outer ring crackle) states
Data;
Step 2: extracting Fault characteristic parameters, feature ginseng is extracted according to sensibility of each index to five kinds of gear-box different operating conditions
Number, and be normalized;
Step 3: using said extracted and make normalized characteristic parameter as dynamic self-adapting artificial fish school algorithm
Five kinds of states of the input of BP neural network (DLAFSA-BP), gear-box are exported as network;
Step 4: obtaining Fault Diagnosis of Gear Case result using DLAFSA-BP network model.
2. the Fault Diagnosis of Gear Case method according to claim 1 based on fish-swarm algorithm Optimizing BP Network, feature exist
In in step 3, the mechanism of ammonium fixation in two major parameter visuals field and step-length in standard intraocular's fish-swarm algorithm is adjusted to a kind of
Dynamic mechanism is adjusted parameter visual using the dynamic changeability of Logistic model, if the variation of visual field visual
Range is [visualmin, visualmax], the variation of parameter visual is set as formula:
The then calculation formula of visual are as follows:
Wherein, t is the number of iterations of algorithm;α is initial attenuation speed, and α value is bigger, and visual decrease speed is faster;
Step-length Step is set with the dynamic self-adapting changing rule such as following formula of visual field visual:
Step=β * Visual (3)
Wherein, β is view step coefficient 0 < β < 1;
2) in improved fish-swarm algorithm, ignore original crowding factor.Each Artificial Fish is recorded in bulletin board executes one
The highest Artificial Fish state of food concentration after behavior is always ensured that the state of Artificial Fish in bulletin board is optimal in iterative process
's;
Shoal of fish Behavior changes of bunching in conjunction with bulletin board are as follows:
Wherein, rand indicates a random number between [0,1];XcWhen indicating the t times iteration, n (t) neighbours fish in neighborhood
Center;Ggbx indicates the state of Artificial Fish in bulletin board;Xnext(i) next state of i-th Artificial Fish is indicated;
X in formula (4)cCalculation formula it is as follows:
Wherein, XmiIndicate Artificial Fish XiM (0 < m < n (t)) article neighbours fish;
Shoal of fish foraging behavior expression in conjunction with bulletin board is as follows:
Wherein, XmaxIndicate Artificial Fish XiThat highest neighbours fish of food concentration in neighbour structure;
Using the adaptive dynamic shoal of fish algorithm optimization BP neural network, the specific steps of which are as follows:
1) BP neural network topological structure is determined, the neuron number including the network number of plies and each layer network;
2) according to N Artificial Fish of network structure random initializtion, the initial shoal of fish is formed;
3) parameter of dynamic self-adapting artificial fish-swarm algorithm (DLAFSA) is set, including repeats to explore number try-number, most
Big the number of iterations T, field range [visualmin, visualmax], rate of decay α, depending on walking factor beta;
4) food concentration of every Artificial Fish in the initial shoal of fish is calculated.The state of Artificial Fish is set as to the initial weight of neural network
And threshold value, every Artificial Fish all correspond to a neural network, the inverse of neural metwork training error is dense as the food of Artificial Fish
Degree;
5) DLAFSA algorithm is run, after algorithm, extracts the state ggbx of Artificial Fish in bulletin board;
6) the state ggbx of Artificial Fish in bulletin board is assigned to BP neural network as initial weight and threshold value;
7) training BP neural network, and fault diagnosis is carried out with trained network;
The state of Artificial Fish is set as to the initial weight and threshold value of neural network, the reciprocal of neural metwork training error is used as manually
The food concentration of fish, then network weight and the searching process of threshold value are as follows:
The neural network for being I-J-K to a topological structure, is expressed as an I*J+J+J*K+K dimensional vector for the state X of Artificial Fish:
X=(w11...wi1,b1,...,w1j...wij,bj,v11...vj1,a1,...,v1k...vjk,ak) (7)
Wherein, w11...wi1Indicate weight of the input layer to first neuron of hidden layer, b1Indicate hidden layer first
The threshold value of neuron, w1j...wijIndicate the weight between input layer and hidden layer j-th neuron, bjIndicate implicit
The threshold value of layer j-th neuron, v11...vj1Indicate the weight between first neuron of hidden layer neuron and output layer, a1
Indicate the threshold value of first neuron of hidden layer, v1k...vjkIt indicates between k-th of neuron of hidden layer neuron and output layer
Weight, akIndicate the threshold value of k-th of neuron of hidden layer.BP neural network training is set by the food concentration of Artificial Fish to miss
The inverse of difference, i.e. Y (X)=1/e (X), then any two Artificial Fish XpAnd XqBetween Euclidean distance calculation formula it is as follows:
Wherein Xp、XqIn element in strict accordance with dimension one-to-one correspondence subtract each other, an Artificial Fish basic act of every execution, nerve net
Network weight and threshold value just adjust once.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910455439.6A CN110097143A (en) | 2019-05-29 | 2019-05-29 | A kind of Fault Diagnosis of Gear Case method based on fish-swarm algorithm Optimizing BP Network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910455439.6A CN110097143A (en) | 2019-05-29 | 2019-05-29 | A kind of Fault Diagnosis of Gear Case method based on fish-swarm algorithm Optimizing BP Network |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110097143A true CN110097143A (en) | 2019-08-06 |
Family
ID=67449489
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910455439.6A Withdrawn CN110097143A (en) | 2019-05-29 | 2019-05-29 | A kind of Fault Diagnosis of Gear Case method based on fish-swarm algorithm Optimizing BP Network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110097143A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108459406A (en) * | 2018-03-15 | 2018-08-28 | 上海理工大学 | Microscope auto-focusing window selection method based on artificial fish-swarm algorithm |
CN112446457A (en) * | 2020-12-02 | 2021-03-05 | 中国计量大学 | Gas leakage source positioning method based on improved artificial fish swarm algorithm |
-
2019
- 2019-05-29 CN CN201910455439.6A patent/CN110097143A/en not_active Withdrawn
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108459406A (en) * | 2018-03-15 | 2018-08-28 | 上海理工大学 | Microscope auto-focusing window selection method based on artificial fish-swarm algorithm |
CN108459406B (en) * | 2018-03-15 | 2020-06-23 | 上海理工大学 | Microscope automatic focusing window selection method based on artificial fish swarm algorithm |
CN112446457A (en) * | 2020-12-02 | 2021-03-05 | 中国计量大学 | Gas leakage source positioning method based on improved artificial fish swarm algorithm |
CN112446457B (en) * | 2020-12-02 | 2023-07-18 | 中国计量大学 | Gas leakage source positioning method based on improved artificial fish swarm algorithm |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106769052B (en) | A kind of mechanical system rolling bearing intelligent failure diagnosis method based on clustering | |
CN110410282B (en) | SOM-MQE and SFCM (Small form-factor pluggable) based wind turbine generator health state online monitoring and fault diagnosis method | |
CN111046945B (en) | Fault type and damage degree diagnosis method based on combined convolutional neural network | |
CN100485342C (en) | Integrated supporting vector machine mixed intelligent diagnosing method for mechanical fault | |
CN106682814A (en) | Method for intelligently diagnosing wind turbine unit faults based on fault knowledge base | |
CN109376620A (en) | A kind of migration diagnostic method of gearbox of wind turbine failure | |
CN109782603A (en) | The detection method and monitoring system of rotating machinery coupling fault | |
CN112257530B (en) | Rolling bearing fault diagnosis method based on blind signal separation and support vector machine | |
Lin et al. | Gear fault diagnosis based on CS-improved variational mode decomposition and probabilistic neural network | |
CN103808509A (en) | Fan gear box fault diagnosis method based on artificial intelligence algorithm | |
CN114124038A (en) | Deconvolution enhancement-based rolling bearing acoustic signal weak fault diagnosis method | |
CN111337256A (en) | Method for diagnosing fault depth local migration of rolling bearing weighted by domain asymmetry factor | |
CN110097143A (en) | A kind of Fault Diagnosis of Gear Case method based on fish-swarm algorithm Optimizing BP Network | |
CN113375941A (en) | Open set fault diagnosis method for high-speed motor train unit bearing | |
CN111766512A (en) | Generator fault maintenance system and method | |
CN112329520B (en) | Truck bearing fault identification method based on generation countermeasure learning | |
Shang et al. | Fault diagnosis method of rolling bearing based on deep belief network | |
CN110348468A (en) | A kind of bearing inferior health recognition methods of the strong reconstruct edge noise reduction autocoder of Method Using Relevance Vector Machine optimization | |
CN115545070A (en) | Intelligent diagnosis method for unbalance-like bearing based on comprehensive balance network | |
CN115163424A (en) | Wind turbine generator gearbox oil temperature fault detection method and system based on neural network | |
CN111539381B (en) | Construction method of wind turbine bearing fault classification diagnosis model | |
CN113469252A (en) | Extra-high voltage converter valve operation state evaluation method considering unbalanced samples | |
CN111783941A (en) | Mechanical equipment diagnosis and classification method based on probability confidence degree convolutional neural network | |
CN116625686A (en) | On-line diagnosis method for bearing faults of aero-engine | |
CN114964783B (en) | Gearbox fault detection model based on VMD-SSA-LSSVM |
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 | ||
WW01 | Invention patent application withdrawn after publication | ||
WW01 | Invention patent application withdrawn after publication |
Application publication date: 20190806 |