CN109116833A - Based on improvement drosophila-bat algorithm mechanical failure diagnostic method - Google Patents
Based on improvement drosophila-bat algorithm mechanical failure diagnostic method Download PDFInfo
- Publication number
- CN109116833A CN109116833A CN201811007052.6A CN201811007052A CN109116833A CN 109116833 A CN109116833 A CN 109116833A CN 201811007052 A CN201811007052 A CN 201811007052A CN 109116833 A CN109116833 A CN 109116833A
- Authority
- CN
- China
- Prior art keywords
- parameter
- drosophila
- support vector
- algorithm
- vector machines
- 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
- 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
- G05B23/0259—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterized by the response to fault detection
- G05B23/0262—Confirmation of fault detection, e.g. extra checks to confirm that a failure has indeed occurred
-
- 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
Abstract
It is specially a kind of based on drosophila-bat algorithm mechanical failure diagnostic method is improved the invention belongs to mechanical fault diagnosis field, include the following steps and extracts Time-domain Statistics feature and frequency domain statistical nature from the Mechanical Running Condition signal of acquisition；Using mechanical fault diagnosis training sample set come Training Support Vector Machines；Using drosophila algorithm as frame, the echolocation thought of bat algorithm is incorporated, designs and improves drosophila-bat parameter optimization method, and finds global optimum's parameter of support vector machines using this method；Obtained global optimum's parameter is substituted into support vector machines, completes the building of the fault diagnosis model based on support vector machines.The present invention can obtain optimal support vector machines parameter in a relatively short period of time, effectively increase the building efficiency and failure modes accuracy rate of the fault diagnosis model based on support vector machines, have good practical application effect.
Description
Technical field
The present invention relates to the fields such as mechanical fault diagnosis, mode identification technology, are related to a kind of Swarm Intelligent Algorithm pair
Supporting vector relevant parameter optimizing, it is specially a kind of to construct the machine fault diagnosis model based on the optimal supporting vector of parameter
Based on improvement drosophila-bat algorithm mechanical failure diagnostic method.
Background technique
Modern machinery and equipment has become the complication system for collecting the multipotencys domain such as mechanical, electrical, liquid, gas, magnetic, multi-scale coupling, and
Quite a few mechanical equipment works under the rugged environments such as varying load, variable speed for a long time, the service life of mechanical equipment
And safety all receives strong influence, the especially load-supporting parts such as mechanical equipment main shaft, gear-box, bearing and event easily occurs
Barrier.Also, mechanical equipment enlargement and complication increasingly, the caused loss of mechanical breakdown also become very huge, e.g., 1992
One 600MW thermal power generation unit of Kansai Electric Power Co. is in hypervelocity experiment, due to not predicting in unit in time
The failure of bearing causes unit that high vibration occurs and causes the damage of unit unrepairable, economic loss caused by the accident
Up to 5,000,000,000 yen；The bearing catastrophic failure of vertical mill, leads to the axis in Wuhan steel plant high-speed wire production line in 2003
On gear there is fracture defect so that whole production line stop production 68 hours, cause huge economic loss.Therefore,
The reliable mechanical failure diagnostic method of research and development realizes that the Precise Diagnosis of mechanical equipment fault just seems most important.
Mechanical fault diagnosis is exactly the process of trouble-shooting reason, and its essence is understand and grasp equipment running process
In state, assessment, pre- measurement equipment reliability, early detection potential faults, and to its reason, position, degree of danger etc. into
Row identification, predicts the development trend of failure, finally realizes Predictive Maintenance.In essence, mechanical fault diagnosis is a mode
Identification problem generally includes Signal Pretreatment, feature extraction and fault identification, and the superiority and inferiority of fault identification algorithm will be straight
Connect the precision for influencing fault diagnosis.Common fault identification algorithm includes K arest neighbors sorting algorithm (KNNC), artificial neural network
(ANN), k-means etc..But KNNC algorithm is the algorithm for pattern recognition based on statistical theory, is usually needed in training pattern
A large amount of training sample is wanted, this is generally difficult to meet in practical applications；The recognition effect of k-means algorithm depends critically upon k
The selection of value, and in many cases the estimation of k value be it is very difficult, this also causes k-means algorithm in practical application
Middle less effective；ANN algorithm is the algorithm for pattern recognition based on empirical risk minimization principle, is tended in training pattern
The problem of falling into overfitting causes the generalization ability of fault diagnosis model insufficient.Support vector machines (SVM) is then a kind of based on knot
The algorithm for pattern recognition of structure risk minimization, by establishing an optimizing decision hyperplane, so that this is flat for two lateral extent of plane
The distance between two nearest class samples of face maximize, to provide good generalization ability to classification problem.SVM also has meter
High-efficient advantage, while can simply and effectively realize the extensive transplanting of algorithm, therefore the intelligent trouble diagnosis based on SVM
It is widely applied.But SVM itself has certain limitation again, and mainly SVM needs to be arranged punishment parameter C and core ginseng
Number, the setting of the two parameters has a great impact to the overall performance of SVM, however SVM optimal parameters selection method is still
Urgent need to solve the problem on engineer application at present.Currently, existing method all carries out tune ginseng using intelligent algorithm mostly, to obtain one
A ideal result.
In recent years, it proposes and has developed a large amount of intelligent algorithm to solve parameters optimization problem, Swarm Intelligence Algorithm is exactly
The global optimization method that the intelligent phenomenon of simulation biocenose develops.Drosophila algorithm and bat algorithm be not all in 2011 by
Same scholar proposes that both methods simulates the biological phenomenon that drosophila and bat are looked for food respectively, belongs to heuristic swarm intelligence
Algorithm.Compared to other intelligent algorithms, the maximum advantage of drosophila algorithm is that algorithm is easy to accomplish, convenient for turning its theoretical thought
Change program code and it can be readily appreciated that also advantageous in computational efficiency into.And bat algorithm possesses excellent local search and complete
Office's search capability, it is possible to prevente effectively from algorithm falls into locally optimal solution.But for the parameter optimization process of SVM, due to parameter
Value range it is very big, to intelligent algorithm, more stringent requirements are proposed for this, this both required algorithm search go out parameter energy maximum limit
Degree ground promotes the performance of SVM, and requires the computational efficiency of algorithm high.Therefore, suitable intelligent algorithm is chosen and improves to obtain
Optimal application is still the direction of worth research and discovery.
Summary of the invention
Mechanical fault diagnosis is substantially a pattern recognition problem, generally includes Signal Pretreatment, feature extraction and event
Barrier identification and etc., and the superiority and inferiority of fault identification algorithm will directly affect the precision of fault diagnosis.Support vector machines is a kind of normal
Algorithm for pattern recognition is widely applied in mechanical fault diagnosis field, but the performance heavy dependence of support vector machines
In the selection of parameter.
Based on problem of the existing technology, the present invention is in order to further enhance the fault diagnosis mould based on support vector machines
The overall performance of type, proposes a kind of improvement drosophila-bat parameter optimization algorithm to realize the arameter optimization of support vector machines, from
And build the optimal support vector machines fault diagnosis model of parameter.The present invention is mainly using improvement drosophila-bat algorithm to branch
The arameter optimization of vector machine is held, to play support vector machines optimum performance, the failure modes in elevating mechanism fault diagnosis are accurate
Degree.This method drosophila algorithm calculating process it is simple, be easily programmed on the basis of, further merge the echolocation of bat algorithm
Thought, to obtain more preferably parameter value under faster calculating speed.
The concrete scheme that the present invention uses is as follows:
It is a kind of based on improve drosophila-bat algorithm mechanical failure diagnostic method, comprising the following steps:
S1, Time-domain Statistics feature and frequency domain statistical nature are extracted from the Mechanical Running Condition signal of acquisition；
S2, using mechanical fault diagnosis training sample set come Training Support Vector Machines, support vector machines mathematical model is such as
Under:
s.t.yi(wTφ(xi)+b)-1+εi≥0；
Wherein, | | | |2Indicate two norms；W indicates the normal vector of hyperplane, wTIndicate the transposition of w vector；C indicates punishment
Parameter；εiIndicate slack variable；yiIndicate the label to i-th of Hyperplane classification；B indicate displacement, that is, arrive hyperplane away from
From；φ(xi) indicate by kernel function treated sample point xi；The dimension of N expression support vector machines.
In training sample data, data are usually Nonlinear separability, need to penetrate data print using kernel function at this time
To reach linear separability in order to classify on to higher dimension.And use most kernel functions for gaussian kernel function, table
It is as follows up to formula:
Wherein, g indicates nuclear parameter, xi、xjIndicate sample point；j∈(1,2,...,N).Therefore the model needs to be related to
Two big parameters are punishment parameter C and kernel functional parameter g；
S3, using drosophila algorithm as frame, incorporate the echolocation thought of bat algorithm, design and improve drosophila-bat ginseng
Optimization method is counted, and finds global optimum's parameter of support vector machines using this method；
S4, the global optimum's parameter for obtaining step S3 substitute into support vector machines, complete the failure based on support vector machines
The building of diagnostic model.
Further, the step S3 is specifically included:
S301, to the parameter assignment of support vector machines obtained by step S2, the punishment parameter of support vector machines is expressed as C, core
Parameter is expressed as g, and (C, g) is denoted as to the initial position of drosophila population；
S302, random direction and distance, C are enclosed to each drosophilai=C+rand, gi=g+rand；Wherein, CiIt indicates more
Punishment parameter value after new；giIndicate updated nuclear parameter value；Rand indicates equally distributed random number；
If S303, random number rand are less than the first parameter r set according to bat algorithmi, then drosophila population is carried out
Local search: Ci=Ci+ 0.01 × randn, gi=gi+ 0.01 × randn, rand indicate to obey equally distributed random number；It is no
It then jumps in next step；
S304, resulting drosophila population position is substituted into evaluation function, calculates its fitness value；Evaluation function indicates
Are as follows: Fitness=function (Ci,gi)；function(Ci,gi) indicate (Ci,gi) the corresponding wrong classification rate of parameter；SeFor
For the training samples number of misclassification class, S is training sample total quantity；
S305, Calculation Estimation functional minimum value, i.e., current optimal solution: Fitnessbest=min (function (Ci,
gi))；
S306, global search is carried out to the result of the current optimal solution of gained: if the current optimal solution FitnessbestIt is excellent
In global optimum best, and the second parameter A set according to bat algorithmiGreater than the random number rand of generation, then receive to work as
Preceding optimal value, and current optimal value is assigned to global optimum, i.e. best=Fitnessbest, by the initial position of step S302
It is updated to the corresponding position of optimal drosophila individual, i.e. (C, g) → (Cbest,gbest)；Otherwise the current optimal value is abandoned, and is maintained
The initial position of step S302；
S307, by initial position determined by step S306, return step S302, until meet stopping criterion for iteration, this
When record optimal parameter value.
Further, after step S306 updates the initial position in step S302, if updated initial position is super
The value interval of parameter out then needs to carry out mapping processing to the initial position of each drosophila, where mapping that in section:
Ci=CLB+ U (0,1) × (CUB-CLB)；
gi=gLB+ U (0,1) × (gUB-gLB)；
Wherein, CLBFor the corresponding lower boundary of C, gLBFor the corresponding lower boundary of g；CUBFor the corresponding coboundary C, gUBIt is corresponding for g
Coboundary；U (0,1) is the uniform random number on section [0,1].
Further, the first parameter r set according to bat algorithmiFor the impulse ejection frequency in bat algorithm,
Parameter riIt is incremented by with the propulsion of operation time:
Wherein,Indicate the first parameter when t circulation；T is the time,For initial transmissions frequency,'s
Value range is [0,1]；γ is constant.
Further, the second parameter A set according to bat algorithmiFor the loudness in bat algorithm, the parameter are as follows:
Wherein, α is constant, and value range is [0.85,0.95]；Indicate the second parameter at the t+1 moment；Ai tIt indicates
For in the second parameter of t moment.
Beneficial effects of the present invention:
The present invention proposes one kind in order to further enhance the overall performance of the fault diagnosis model based on support vector machines
Drosophila-bat parameter optimization algorithm is improved to realize the arameter optimization of support vector machines, to build the optimal support of parameter
Vector machine fault diagnosis model.Improve drosophila-bat parameter optimization algorithm can obtain within a short period of time optimal support to
Amount machine parameter effectively improves the building efficiency and failure modes accuracy rate of the fault diagnosis model based on support vector machines,
With good practical application effect.
Detailed description of the invention
Fig. 1 is the method flow diagram that the present invention uses；
Fig. 2 is the classification process figure for the vector machine that the present invention uses；
Fig. 3 is the specific flow chart of drosophila algorithm and bat algorithm that the present invention uses；
Fig. 4 is the mechanical fault diagnosis classification accuracy curve graph using the method for the present invention；
Fig. 5 is the mechanical fault diagnosis classification accuracy curve graph based on particle swarm algorithm；
Fig. 6 is the mechanical fault diagnosis classification accuracy curve graph based on genetic algorithm.
Specific embodiment
In order to make the objectives, technical solutions, and advantages of the present invention clearer, below in conjunction with attached drawing to of the invention real
The technical solution applied in example is clearly and completely described, it is clear that described embodiment is only that present invention a part is implemented
Example, instead of all the embodiments.
To make the objectives, technical solutions, and advantages of the present invention clearer, below in conjunction with attached drawing to the present invention make into
The detailed description of one step:
As shown in Figure 1, the system flow chart that the present invention uses includes:
S1, Time-domain Statistics feature and frequency domain statistical nature are extracted from the Mechanical Running Condition signal of acquisition；
S2, using mechanical fault diagnosis training sample set come Training Support Vector Machines；
S3, using drosophila algorithm as frame, incorporate the echolocation thought of bat algorithm, design and improve drosophila-bat ginseng
Optimization method is counted, and finds global optimum's parameter of support vector machines using this method；
S4, the global optimum's parameter for obtaining step S3 substitute into support vector machines, complete the failure based on support vector machines
The building of diagnostic model.
Wherein, the present invention is to the support vector cassification flow chart of use as shown in Fig. 2, carrying out parameter to drosophila algorithm
Initialization generates drosophila population, on the other hand, inputs mechanical fault diagnosis sample data, after pre-processing to data, generates
Training set carries out classification based training to the SVM parameter that drosophila population generates using training set, calculates fitness value, judge whether
Meet stopping criterion for iteration, continues Population Regeneration if not, otherwise export result
Further, the step S3 is specifically included, as shown in Figure 3:
S301, to the parameter assignment of support vector machines obtained by step S2, the punishment parameter of support vector machines is expressed as C, core
Parameter is expressed as g, and (C, g) is denoted as to the initial position of drosophila population；
S302, random direction and distance, C are enclosed to each drosophilai=C+rand, gi=g+rand；Wherein, CiIt indicates more
Punishment parameter value after new；giIndicate updated nuclear parameter value；Rand indicates equally distributed random number；
If S303, random number rand are less than the first parameter r set according to bat algorithmi, then drosophila population is carried out
Local search: Ci=Ci+ 0.01 × randn, gi=gi+ 0.01 × randn, rand indicate to obey equally distributed random number；It is no
It then jumps in next step；
S304, resulting drosophila population position is substituted into evaluation function, calculates its fitness value；Evaluation function indicates
Are as follows: Fitness=function (Ci,gi)；function(Ci,gi) indicate (Ci,gi) the corresponding wrong classification rate of parameter；SeFor
For the training samples number of misclassification class, S is training sample total quantity；
S305, Calculation Estimation functional minimum value, i.e., current optimal solution: Fitnessbest=min (function (Ci,
gi))；
S306, global search is carried out to the result of the current optimal solution of gained: if the current optimal solution FitnessbestIt is excellent
In global optimum best, and the second parameter A set according to bat algorithmiGreater than the random number rand of generation, then receive to work as
Preceding optimal value, and current optimal value is assigned to global optimum, i.e. best=Fitnessbest, by the initial position of step S302
It is updated to the corresponding position of optimal drosophila individual, i.e. (C, g) → (Cbest,gbest)；Otherwise the current optimal value is abandoned, and is maintained
The initial position of step S302；
S307, by initial position determined by step S306, return step S302, until meet stopping criterion for iteration, this
When record optimal parameter value.
Further, after step S306 updates the initial position in step S302, if updated initial position is super
The value interval of parameter out then needs to carry out mapping processing to the initial position of each drosophila, where mapping that in section:
Ci=CLB+ U (0,1) × (CUB-CLB)；
gi=gLB+ U (0,1) × (gUB-gLB)；
Wherein, CLBFor the corresponding lower boundary of C, gLBFor the corresponding lower boundary of g；CUBFor the corresponding coboundary C, gUBIt is corresponding for g
Coboundary；U (0,1) is the uniform random number on section [0,1].
Further, the first parameter r set according to bat algorithmiFor the impulse ejection frequency in bat algorithm,
Parameter riIt is incremented by with the propulsion of operation time:
Wherein,Indicate the first parameter when t circulation；T is the time,For initial transmissions frequency,Value range
For [0,1]；γ is constant.
Further, the second parameter A set according to bat algorithmiFor the loudness in bat algorithm, the parameter are as follows:
Wherein, α is constant,It is expressed as the second parameter in t moment；α value is [0.85,0.95].The present embodiment takes
Value is α=0.9.
The present invention is described in detail below with reference to specific embodiment and attached drawing.
First of all for verifying, the present invention has good mode identificating ability, selects UCI (University of
California at Irvin) data in machine learning knowledge base are tested as embodiment, specific data such as 1 institute of table
Show.
The description of 1 data of table
It after obtaining data, is carried out according to process shown in Figure of abstract, in order to compare (FFBA-SVM) of the invention
Advantage, at the same using grid search SVM tune ginseng and genetic algorithm SVM tune ginseng (GA-SVM) to identical data carry out operation, every kind
Each operation of method ten times, the average classification accuracy that each algorithm is obtained is recorded in table 2.
Table 2 and grid search, performance of genetic algorithms comparing result
Last in table 2 is classified as the average classification accuracy of the method for the present invention, compares other two methods, and the present invention is bright
It shows more for advantage.
In order to further compare calculating time and accuracy rate of the invention, we are carried out using method similar with the present invention
Comparison, such as particle swarm algorithm tune ginseng (PSO-SVM) and basic bat algorithm (BA-SVM), by three kinds of methods population quantity with
Maximum number of iterations is all respectively set to 20 and 100, comparing calculation time that in this way can be more fair.Comparing result is recorded in table
In 3.
Table 3 and particle swarm algorithm, original bat algorithm performance comparing result
Table 3 the result shows that, select particle swarm algorithm tuning, accuracy rate be significantly lower than bat algorithm and side of the invention
Method, this is because population lack echolocation operation, it is weaker so as to cause part and ability of searching optimum, but also because for this purpose,
It is very fast that particle swarm algorithm calculates the time, and present invention incorporates the calculating of the global optimizing ability of bat algorithm and drosophila algorithm effects
The high feature of rate, therefore average classification accuracy of the invention is higher than particle swarm optimization, and average calculation times lead over bat
Bat algorithm.
In order to verify application effect of the present invention in mechanical fault diagnosis, pass through the bearing fault number of Case Western Reserve University
According to come the validity that illustrates this method.The data are that the single-point event of bearing different parts is simulated using electrical discharge machining grooving
Barrier, mainly by the way that different notch width: 0.007,0.014,0.021 (1 inch=2.54 centimetres) is arranged, to simulate bearing
Minor failure, moderate failure and catastrophe failure.Speed of mainshaft 1772rpm, load 1HP, sample frequency are selected in this experiment
9 kinds of fault vibration signals for the driving end vibration bearing of 12000Hz are analyzed, comprising: outer ring minor failure, outer ring moderate
Failure, outer ring catastrophe failure, inner ring minor failure, inner ring moderate failure, inner ring catastrophe failure and rolling element minor failure, rolling
Kinetoplast moderate failure, rolling element catastrophe failure.50 groups of samples, each specimen sample points are measured under every kind of malfunction respectively
Be 2048 points, will wherein 20 groups of samples are as training sample, remaining 30 groups of sample is as test sample.Respectively from each sample
24 time-frequency domain statistical natures are extracted in vibration signal, and fault diagnosis is then carried out using the SVM of parameter optimization again.Fig. 4 exhibition
Show that the present invention carries out the fitness curve convergence figure of fault diagnosis to the data, it can be seen from the figure that the present invention obtained
Optimal classification accuracy rate is 98.89%, and from figs. 5 and 6, it can be seen that the accuracy rate based on particle swarm algorithm and genetic algorithm
The bright nicety of grading of we is all not achieved.Thus the present invention is to have a good application prospect in mechanical fault diagnosis.
The above description is only a preferred embodiment of the present invention, is not intended to restrict the invention, it is clear that those skilled in the art
Various changes and modifications can be made to the invention by member without departing from the spirit and scope of the present invention.
Those of ordinary skill in the art will appreciate that all or part of the steps in the various methods of above-described embodiment is can
It is completed with instructing relevant hardware by program, which can be stored in a computer readable storage medium, storage
Medium may include: ROM, RAM, disk or CD etc..
Embodiment provided above has carried out further detailed description, institute to the object, technical solutions and advantages of the present invention
It should be understood that embodiment provided above is only the preferred embodiment of the present invention, be not intended to limit the invention, it is all
Any modification, equivalent substitution, improvement and etc. made for the present invention, should be included in the present invention within the spirit and principles in the present invention
Protection scope within.
Claims (7)
1. a kind of based on improvement drosophila-bat algorithm mechanical failure diagnostic method, which comprises the following steps:
S1, Time-domain Statistics feature and frequency domain statistical nature are extracted from the Mechanical Running Condition signal of acquisition；
S2, using mechanical fault diagnosis training sample set Training Support Vector Machines model；
S3, using drosophila algorithm as frame, incorporate bat algorithm echolocation thought, design improve drosophila-bat parameter it is excellent
Change method, and find using this method global optimum's parameter of support vector machines；
S4, the global optimum's parameter for obtaining step S3 substitute into support vector machines, complete the fault diagnosis based on support vector machines
The building of model.
2. according to claim 1 a kind of based on drosophila-bat algorithm mechanical failure diagnostic method is improved, feature exists
In supporting vector machine model described in step S2 includes:
s.t.yi(wTφ(xi)+b)-1+εi≥0；
Wherein, | | | |2Indicate two norms；W indicates the normal vector of hyperplane, wTIndicate the transposition of w vector；C indicates punishment parameter；
εiIndicate slack variable；yiIndicate the label to i-th of Hyperplane classification；B indicates displacement, that is, arrives the distance of hyperplane；φ
(xi) indicate by kernel function treated sample point xi；The dimension of N expression support vector machines.
3. according to claim 2 a kind of based on drosophila-bat algorithm mechanical failure diagnostic method is improved, feature exists
In the kernel function is gaussian kernel function, expression formula are as follows:
Wherein, g indicates nuclear parameter, xi、xjIndicate sample point；j∈(1,2,...,N).
4. according to claim 1 a kind of based on drosophila-bat algorithm mechanical failure diagnostic method is improved, feature exists
In the step S3 is specifically included:
S301, assignment is carried out to the parameter of support vector machines obtained by step S2, the punishment parameter of support vector machines is expressed as C, core
Parameter is expressed as g, and (C, g) is denoted as to the initial position of drosophila population；
S302, random direction and distance, C are enclosed to each drosophilai=C+rand, gi=g+rand；Wherein, CiIt indicates after updating
Punishment parameter value；giIndicate updated nuclear parameter value；Rand indicates to obey equally distributed random number；
If S303, random number rand are less than the first parameter r set according to bat algorithmi, then part is carried out to drosophila population and searched
Rope: Ci=Ci+ 0.01 × randn, gi=gi+ 0.01 × randn, randn indicate to obey the random number of standardized normal distribution；It is no
It then jumps in next step；
S304, resulting drosophila population position is substituted into evaluation function, calculates its fitness value；Evaluation function indicates are as follows:function(Ci,gi) indicate (Ci,gi) the corresponding mistake classification of parameter
Rate；SeFor the training samples number of misclassification class, S is training sample total quantity；
S305, Calculation Estimation functional minimum value, i.e., current optimal solution: Fitnessbest=min (function (Ci,gi))；
S306, global search is carried out to the result of the current optimal solution of gained: if the current optimal solution FitnessbestBetter than the overall situation
Optimal value best, and the second parameter A set according to bat algorithmiGreater than the random number rand of generation, then receive current optimal
Value, and current optimal value is assigned to global optimum, i.e. best=Fitnessbest, the initial position of step S302 is updated to
The individual corresponding position of optimal drosophila, i.e. (C, g) → (Cbest,gbest)；Otherwise the current optimal value is abandoned, and maintains step
The initial position of S302；
S307, by initial position determined by step S306, return step S302 remembers at this time until meeting stopping criterion for iteration
Optimal parameter value under record.
5. according to claim 4 a kind of based on drosophila-bat algorithm mechanical failure diagnostic method is improved, feature exists
In after step S306 updates the initial position in step S302, if updated initial position exceeds the value area of parameter
Between, then it needs to carry out mapping processing to the initial position of each drosophila, where mapping that in section:
Ci=CLB+ U (0,1) × (CUB-CLB)；
gi=gLB+ U (0,1) × (gUB-gLB)；
Wherein, CLBFor the corresponding lower boundary of C, gLBFor the corresponding lower boundary of g；CUBFor the corresponding coboundary C, gUBFor g it is corresponding on
Boundary；U (0,1) is the uniform random number on section [0,1].
6. according to claim 4 a kind of based on drosophila-bat algorithm mechanical failure diagnostic method is improved, feature exists
In the first parameter r set according to bat algorithmiFor the impulse ejection frequency in bat algorithm, parameter riWith operation
The propulsion of time and be incremented by:
ri t+1=ri 0[1-exp(-γt)]
Wherein, ri t+1Indicate the first parameter when t circulation；T is time, ri 0For initial transmissions frequency, ri 0Value range be
[0,1]；γ is constant.
7. according to claim 4 a kind of based on drosophila-bat algorithm mechanical failure diagnostic method is improved, feature exists
In the second parameter A set according to bat algorithmiFor the loudness in bat algorithm, the parameter are as follows:
Wherein, α is constant, and value range is [0.85,0.95]；Indicate the second parameter at the t+1 moment；It is expressed as in t
Second parameter at moment.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811007052.6A CN109116833B (en) | 2018-08-31 | 2018-08-31 | Mechanical fault diagnosis method based on improved fruit fly-bat algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811007052.6A CN109116833B (en) | 2018-08-31 | 2018-08-31 | Mechanical fault diagnosis method based on improved fruit fly-bat algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109116833A true CN109116833A (en) | 2019-01-01 |
| CN109116833B CN109116833B (en) | 2021-04-16 |
Family
ID=64861497
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811007052.6A Active CN109116833B (en) | 2018-08-31 | 2018-08-31 | Mechanical fault diagnosis method based on improved fruit fly-bat algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109116833B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111159903A (en) * | 2019-12-31 | 2020-05-15 | 重庆邮电大学 | Design and manufacturing method of compact type multi-channel multi-fluid heat exchange device |
| CN111639695A (en) * | 2020-05-26 | 2020-09-08 | 温州大学 | Method and system for classifying data based on improved drosophila optimization algorithm |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104794368A (en) * | 2015-05-15 | 2015-07-22 | 哈尔滨理工大学 | Rolling bearing fault classifying method based on FOA-MKSVM (fruit fly optimization algorithm-multiple kernel support vector machine) |
| CN106802563A (en) * | 2016-10-13 | 2017-06-06 | 湖南工业大学 | A kind of sewage procedure optimization control method based on fruit bat optimization and LSSVM |
| CN107330781A (en) * | 2017-06-19 | 2017-11-07 | 南京信息工程大学 | A kind of individual credit risk appraisal procedure based on IFOA SVM |
| WO2018071896A1 (en) * | 2016-10-14 | 2018-04-19 | Celmatix Inc. | Methods and systems for assessing fertility based on subclinical genetic factors |
| CN108121999A (en) * | 2017-12-10 | 2018-06-05 | 北京工业大学 | Support vector machines parameter selection method based on mixing bat algorithm |
| CN108416366A (en) * | 2018-02-06 | 2018-08-17 | 武汉大学 | A kind of power-system short-term load forecasting method of the weighting LS-SVM based on Meteorological Index |
-
2018
- 2018-08-31 CN CN201811007052.6A patent/CN109116833B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104794368A (en) * | 2015-05-15 | 2015-07-22 | 哈尔滨理工大学 | Rolling bearing fault classifying method based on FOA-MKSVM (fruit fly optimization algorithm-multiple kernel support vector machine) |
| CN106802563A (en) * | 2016-10-13 | 2017-06-06 | 湖南工业大学 | A kind of sewage procedure optimization control method based on fruit bat optimization and LSSVM |
| WO2018071896A1 (en) * | 2016-10-14 | 2018-04-19 | Celmatix Inc. | Methods and systems for assessing fertility based on subclinical genetic factors |
| CN107330781A (en) * | 2017-06-19 | 2017-11-07 | 南京信息工程大学 | A kind of individual credit risk appraisal procedure based on IFOA SVM |
| CN108121999A (en) * | 2017-12-10 | 2018-06-05 | 北京工业大学 | Support vector machines parameter selection method based on mixing bat algorithm |
| CN108416366A (en) * | 2018-02-06 | 2018-08-17 | 武汉大学 | A kind of power-system short-term load forecasting method of the weighting LS-SVM based on Meteorological Index |
Non-Patent Citations (2)
| Title |
|---|
| 吴易轩: "果蝇优化算法改进及应用研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
| 李素等: "群智能算法优化支持向量机参数综述", 《智能系统学报》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111159903A (en) * | 2019-12-31 | 2020-05-15 | 重庆邮电大学 | Design and manufacturing method of compact type multi-channel multi-fluid heat exchange device |
| CN111639695A (en) * | 2020-05-26 | 2020-09-08 | 温州大学 | Method and system for classifying data based on improved drosophila optimization algorithm |
| CN111639695B (en) * | 2020-05-26 | 2024-02-20 | 温州大学 | Method and system for classifying data based on improved drosophila optimization algorithm |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109116833B (en) | 2021-04-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Amraee et al. | Transient instability prediction using decision tree technique | |
| CN102221655B (en) | Random-forest-model-based power transformer fault diagnosis method | |
| CN105759201B (en) | High-voltage circuitbreaker self-diagnosing method based on exceptional sample identification | |
| US8588984B2 (en) | Rotor angle stability prediction using post disturbance voltage trajectories | |
| CN108051660A (en) | A kind of transformer fault combined diagnosis method for establishing model and diagnostic method | |
| CN109740859A (en) | Transformer condition evaluation and system based on Principal Component Analysis and support vector machines | |
| CN102944769A (en) | Fault diagnosis method of power transformer based on extreme learning machine | |
| CN109659933A (en) | A kind of prediction technique of power quality containing distributed power distribution network based on deep learning model | |
| CN106482967B (en) | A kind of Cost Sensitive Support Vector Machines locomotive wheel detection system and method | |
| CN106980877A (en) | A kind of Prediction of Blasting Vibration method based on particle cluster algorithm Support Vector Machines Optimized | |
| CN110362048A (en) | Blower critical component state monitoring method and device, storage medium and terminal | |
| CN109492790A (en) | Wind turbines health control method based on neural network and data mining | |
| CN104123678A (en) | Electricity relay protection status overhaul method based on status grade evaluation model | |
| CN109921415A (en) | A kind of pernicious online defence method of Data Injection Attacks of power grid towards hybrid measurement | |
| CN108595884A (en) | Power system transient stability appraisal procedure and device | |
| CN109116833A (en) | Based on improvement drosophila-bat algorithm mechanical failure diagnostic method | |
| CN105204438A (en) | Machining quality control method for cylinder cover components | |
| CN106408016A (en) | Distribution network power outage time automatic identification model construction method | |
| CN104218571A (en) | Running state evaluation method for wind power generation equipment | |
| CN103440527A (en) | Method for improving ant colony algorithm optimization support vector machine parameters | |
| CN110705831A (en) | Power angle instability mode pre-judgment model construction method after power system fault and application thereof | |
| CN111062569A (en) | Low-current fault discrimination method based on BP neural network | |
| CN109799454A (en) | Generator stator insulation residual breakdown strength prediction technique based on particle group optimizing | |
| CN109325553A (en) | A kind of wind turbine gearbox fault detection method, system, equipment and medium | |
| Halilcevic et al. | Prediction of power system security levels |
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 |