CN109116833B - Mechanical fault diagnosis method based on improved fruit fly-bat algorithm - Google Patents
Mechanical fault diagnosis method based on improved fruit fly-bat algorithm Download PDFInfo
- Publication number
- CN109116833B CN109116833B CN201811007052.6A CN201811007052A CN109116833B CN 109116833 B CN109116833 B CN 109116833B CN 201811007052 A CN201811007052 A CN 201811007052A CN 109116833 B CN109116833 B CN 109116833B
- Authority
- CN
- China
- Prior art keywords
- parameter
- support vector
- vector machine
- algorithm
- fault diagnosis
- 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
- 238000004422 calculation algorithm Methods 0.000 title claims abstract description 87
- 238000003745 diagnosis Methods 0.000 title claims abstract description 43
- 238000000034 method Methods 0.000 title claims abstract description 42
- 235000013399 edible fruits Nutrition 0.000 title claims description 9
- 238000012706 support-vector machine Methods 0.000 claims abstract description 55
- 238000012549 training Methods 0.000 claims abstract description 21
- 238000005457 optimization Methods 0.000 claims abstract description 19
- 241000255581 Drosophila <fruit fly, genus> Species 0.000 claims abstract description 9
- 238000010276 construction Methods 0.000 claims abstract description 6
- 241000255588 Tephritidae Species 0.000 claims description 24
- 230000005540 biological transmission Effects 0.000 claims description 6
- 238000011156 evaluation Methods 0.000 claims description 6
- 230000009191 jumping Effects 0.000 claims description 3
- 238000009827 uniform distribution Methods 0.000 claims description 3
- 238000006073 displacement reaction Methods 0.000 claims description 2
- 238000012545 processing Methods 0.000 claims description 2
- 238000009826 distribution Methods 0.000 claims 1
- 230000000694 effects Effects 0.000 abstract description 5
- 230000006870 function Effects 0.000 description 16
- 238000004364 calculation method Methods 0.000 description 9
- 239000002245 particle Substances 0.000 description 8
- 238000003909 pattern recognition Methods 0.000 description 8
- 230000008901 benefit Effects 0.000 description 7
- 230000002068 genetic effect Effects 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000005096 rolling process Methods 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 3
- 238000007781 pre-processing Methods 0.000 description 3
- XEEYBQQBJWHFJM-UHFFFAOYSA-N Iron Chemical compound [Fe] XEEYBQQBJWHFJM-UHFFFAOYSA-N 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 230000002431 foraging effect Effects 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000003860 storage Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 241001391944 Commicarpus scandens Species 0.000 description 1
- 229910000831 Steel Inorganic materials 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000007635 classification algorithm Methods 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000009760 electrical discharge machining Methods 0.000 description 1
- 229910052742 iron Inorganic materials 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 230000007935 neutral effect Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000010187 selection method Methods 0.000 description 1
- 239000010959 steel Substances 0.000 description 1
- 238000002054 transplantation Methods 0.000 description 1
Images
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
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Automation & Control Theory (AREA)
- Testing And Monitoring For Control Systems (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention belongs to the field of mechanical fault diagnosis, and particularly relates to a mechanical fault diagnosis method based on an improved drosophila-bat algorithm, which comprises the following steps of extracting time domain statistical characteristics and frequency domain statistical characteristics from collected mechanical running state signals; training a support vector machine by adopting a mechanical fault diagnosis training sample set; the drosophila algorithm is taken as a frame, the echo positioning idea of the bat algorithm is integrated, an improved drosophila-bat parameter optimization method is designed, and the method is adopted to search the global optimal parameters of the support vector machine; and substituting the obtained global optimal parameters into the support vector machine to complete the construction of the fault diagnosis model based on the support vector machine. The method can obtain the optimal support vector machine parameters in a short time, effectively improves the construction efficiency and the fault classification accuracy of the fault diagnosis model based on the support vector machine, and has good practical application effect.
Description
Technical Field
The invention relates to the technical field of mechanical fault diagnosis, pattern recognition and the like, in particular to a mechanical fault diagnosis method based on an improved drosophila-bat algorithm, which is used for optimizing relevant parameters of a support vector by a swarm intelligence optimization algorithm to construct a support vector machine fault diagnosis model based on parameter optimization.
Background
Modern mechanical equipment becomes a complex system integrating multiple energy domains such as mechanical, electric, hydraulic, pneumatic, magnetic and the like and coupling in multiple scales, and a part of mechanical equipment works in severe environments such as variable load, variable rotating speed and the like for a long time, so that the service life and the safety of the mechanical equipment are greatly influenced, and particularly, bearing parts such as a main shaft, a gear box, a bearing and the like of the mechanical equipment are easy to break down. Moreover, mechanical equipment is increasingly large-sized and complicated, and loss caused by mechanical failure is also very large, for example, in an overspeed experiment of a 600MW thermal power generating set of a Japan Guanxi electric power company in 1992, the failure of a bearing in the set is not predicted in time, so that the set is damaged irreparably due to severe vibration, and the economic loss caused by the accident is up to 50 billion yen; in 2003, the bearing of the neutral rolling mill in the high-speed wire production line of the Wuhan iron and Steel works suddenly fails, so that the gear on the shaft has a fracture fault, the whole production line stops producing for 68 hours, and huge economic loss is caused. Therefore, it is very important to research and develop a reliable mechanical fault diagnosis method to realize accurate diagnosis of mechanical equipment faults.
The fault diagnosis of the mechanical equipment is a process of searching fault causes, and the essence of the fault diagnosis is to know and master the state of the equipment in the operation process, evaluate and predict the reliability of the equipment, find fault hidden dangers at early stage, identify the causes, parts, danger degree and the like, predict the development trend of the fault and finally realize the predicted maintenance. In essence, mechanical fault diagnosis is a pattern recognition problem, which generally includes signal preprocessing, feature extraction, fault recognition and other steps, and the quality of a fault recognition algorithm directly affects the precision of fault diagnosis. Commonly used fault identification algorithms include K nearest neighbor classification algorithm (KNNC), Artificial Neural Network (ANN), K-means, and the like. However, the KNNC algorithm is a pattern recognition algorithm based on a statistical theory, and a large number of training samples are usually required in training a model, which is usually difficult to satisfy in practical application; the identification effect of the k-means algorithm depends heavily on the selection of the k value, and the estimation of the k value is very difficult in many cases, which also results in the poor effect of the k-means algorithm in practical application; the ANN algorithm is a pattern recognition algorithm based on an empirical risk minimization principle, and is prone to being caught in the problem of over-learning when a model is trained, so that the generalization capability of a fault diagnosis model is insufficient. The Support Vector Machine (SVM) is a pattern recognition algorithm based on structure risk minimization, and the distance between two types of samples closest to a plane on two sides of the plane is maximized by establishing an optimal decision hyperplane, so that good generalization capability is provided for classification problems. The SVM also has the advantage of high calculation efficiency, and meanwhile, the generalization transplantation of the algorithm can be simply and effectively realized, so that the intelligent fault diagnosis based on the SVM is widely applied. However, the SVM has certain limitations, mainly the SVM needs to set a penalty parameter C and a kernel parameter, and the setting of the two parameters has a great influence on the overall performance of the SVM, however, the SVM optimal parameter selection method is still a difficult problem which needs to be solved urgently in the current engineering application. At present, most of the existing methods adopt an intelligent algorithm to adjust parameters so as to obtain a more ideal result.
In recent years, a large number of intelligent algorithms are proposed and developed to solve the problem of parameter optimization, and the group intelligent algorithm is a global optimization method developed by simulating the intelligent phenomenon of biological groups. The fruit fly algorithm and the bat algorithm are proposed by different scholars in 2011, and the two methods respectively simulate the biological phenomena of fruit fly foraging and bat foraging and belong to a heuristic group intelligent algorithm. Compared with other intelligent algorithms, the fruit fly algorithm has the greatest advantages that the algorithm is easy to implement, the theoretical idea of the fruit fly algorithm is convenient to convert into program codes and easy to understand, and the fruit fly algorithm also has advantages in computational efficiency. The bat algorithm has excellent local search and global search capabilities, and can effectively avoid the algorithm from falling into a local optimal solution. However, in the parameter optimization process of the SVM, the value range of the parameter is extremely large, so that higher requirements are provided for an intelligent algorithm, the performance of the SVM can be improved to the maximum extent by the parameter searched by the algorithm, and the calculation efficiency of the algorithm is high. Therefore, it is still worth studying and exploring to select and improve the appropriate intelligent algorithm for optimal application.
Disclosure of Invention
The mechanical fault diagnosis is essentially a pattern recognition problem, and generally comprises the steps of signal preprocessing, feature extraction, fault recognition and the like, and the quality of a fault recognition algorithm directly influences the precision of fault diagnosis. The support vector machine is a common pattern recognition algorithm and widely applied to the field of mechanical fault diagnosis, but the performance of the support vector machine is seriously dependent on the selection of parameters.
Based on the problems in the prior art, the invention provides an improved drosophila-bat parameter optimization algorithm to realize parameter tuning of the support vector machine in order to further improve the overall performance of the fault diagnosis model based on the support vector machine, so that the fault diagnosis model of the support vector machine with optimal parameters is constructed. The invention mainly adopts an improved drosophila-bat algorithm to optimize the parameters of the support vector machine so as to exert the best performance of the support vector machine and improve the fault classification accuracy in the mechanical fault diagnosis. On the basis of simple fruit fly algorithm calculation process and easy programming, the method further fuses the echo positioning idea of the bat algorithm, thereby obtaining a better parameter value at a higher calculation speed.
The invention adopts the following specific scheme:
a mechanical fault diagnosis method based on an improved drosophila-bat algorithm comprises the following steps:
s1, extracting time domain statistical characteristics and frequency domain statistical characteristics from the collected mechanical operation state signals;
s2, training a support vector machine by adopting the mechanical fault diagnosis training sample set, wherein the support vector machine mathematical model is as follows:
s.t.yi(wTφ(xi)+b)-1+εi≥0；
wherein | | | purple hair2Represents a two-norm; w denotes the normal vector of the hyperplane, wTRepresents a transpose of the w vector; c represents a penalty parameter; epsiloniRepresents a relaxation variable; y isiA label representing a classification of the ith hyperplane; b represents the displacement, i.e. the distance to the hyperplane; phi (x)i) Representing the sample point x after kernel processingi(ii) a N denotes the dimension of the support vector machine.
When training sample data, the data is usually non-linearly separable, and then a kernel function is needed to print the data to a higher dimension so as to achieve linear separability for classification. The most used kernel function is a gaussian kernel function, and the expression is as follows:
wherein g denotes the nuclear parameter, xi、xjAll represent sample points; j ∈ (1, 2.. multidot.n). Therefore, two parameters needed to be involved in the model are a penalty parameter C and a kernel function parameter g;
s3, taking the fruit fly algorithm as a frame, integrating the echo positioning idea of the bat algorithm, designing an improved fruit fly-bat parameter optimization method, and searching the global optimal parameter of the support vector machine by adopting the method;
and S4, substituting the global optimal parameters obtained in the step S3 into the support vector machine to complete the construction of the fault diagnosis model based on the support vector machine.
Further, the step S3 specifically includes:
s301, assigning a value to the parameter of the support vector machine obtained in the step S2, wherein the punishment parameter of the support vector machine is represented as C, the kernel parameter is represented as g, and the (C, g) is recorded as the initial position of the fruit fly population;
s302, attaching random direction and distance C to each fruit flyi＝C+rand，giG + rand; wherein, CiRepresenting the updated penalty parameter value; giRepresenting the updated core parameter values; rand denotes a uniformly distributed random number;
s303, if the random number rand is less than the first parameter r set according to the bat algorithmiAnd then, carrying out local search on the fruit fly population: ci＝Ci+0.01×randn，gi＝gi+0.01 × randn, rand representing a random number subject to uniform distribution; otherwise, jumping to the next step;
s304, substituting the obtained fruit fly population position into an evaluation function, and calculating the fitness value of the fruit fly population position; the merit function is expressed as: function (C)i,gi)；function(Ci,gi) Is represented by (C)i,gi) The error classification rate corresponding to the parameters; seThe number of training samples is classified as error, and S is the total number of training samples;
s305, calculating the minimum value of the evaluation function, namely the current optimal solution: fitnessbest＝min(function(Ci,gi))；
S306, carrying out global search on the result of the obtained current optimal solution: if the current isOptimal solution FitnessbestA second parameter A which is superior to the global optimum best value best and is set according to the bat algorithmiIf the random number is greater than the generated random number rand, the current optimal value is received, and the current optimal value is assigned to the global optimal value, namely best is FitnessbestUpdating the initial position of the step S302 to the position corresponding to the optimal individual drosophila, namely (C, g) → (C)best,gbest) (ii) a Otherwise, abandoning the current optimum value and maintaining the initial position of step S302;
and S307, returning the initial position determined in the step S306 to the step S302 until an iteration termination condition is met, and recording an optimal parameter value at the moment.
Further, after the initial position in step S302 is updated in step S306, if the updated initial position exceeds the value interval of the parameter, the initial position of each drosophila needs to be mapped into the interval where the initial position is located:
Ci＝CLB+U(0，1)×(CUB-CLB)；
gi＝gLB+U(0，1)×(gUB-gLB)；
wherein, CLBIs the lower boundary, g, for CLBIs the lower boundary corresponding to g; cUBIs the upper boundary corresponding to C, gUBIs the upper boundary corresponding to g; u (0, 1) is the interval [0, 1]]Uniformly distributed random numbers.
Further, the first parameter r set according to the bat algorithmiThe parameter r is the pulse transmission frequency in the bat algorithmiIncrement as the operation time advances:
wherein,
representing a first parameter at t cycles; t is the time of day and t is,
in order to be the initial transmission frequency,
is/are as follows
The value range is [0, 1 ]; gamma is a constant.
Further, the second parameter A set according to the bat algorithmiThe loudness in the bat algorithm is given by:
wherein alpha is a constant and has a value range of [0.85,0.95 ]]；
Represents a second parameter at time t + 1; a. thei tIndicated as the second parameter at time t.
The invention has the beneficial effects that:
in order to further improve the overall performance of the fault diagnosis model based on the support vector machine, the invention provides an improved drosophila-bat parameter optimization algorithm to realize parameter tuning and optimization of the support vector machine, so that the fault diagnosis model of the support vector machine with optimal parameters is constructed. The improved drosophila-bat parameter optimization algorithm can obtain the optimal support vector machine parameters in a short time, effectively improves the construction efficiency and the fault classification accuracy of the fault diagnosis model based on the support vector machine, and has good practical application effect.
Drawings
FIG. 1 is a flow chart of a method employed in the present invention;
FIG. 2 is a flow chart of the classification of the vector machine employed in the present invention;
FIG. 3 is a detailed flow chart of the Drosophila algorithm and bat algorithm employed by the present invention;
FIG. 4 is a graph of mechanical fault diagnosis classification accuracy using the method of the present invention;
FIG. 5 is a graph of mechanical failure diagnosis classification accuracy based on particle swarm optimization;
FIG. 6 is a graph of classification accuracy for mechanical fault diagnosis based on genetic algorithm.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more clearly and completely apparent, the technical solutions in the embodiments of the present invention are described below with reference to the accompanying drawings, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments.
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be further described in detail with reference to the accompanying drawings:
as shown in fig. 1, the system flow chart adopted by the present invention includes:
s1, extracting time domain statistical characteristics and frequency domain statistical characteristics from the collected mechanical operation state signals;
s2, training a support vector machine by adopting a mechanical fault diagnosis training sample set;
s3, taking the fruit fly algorithm as a frame, integrating the echo positioning idea of the bat algorithm, designing an improved fruit fly-bat parameter optimization method, and searching the global optimal parameter of the support vector machine by adopting the method;
and S4, substituting the global optimal parameters obtained in the step S3 into the support vector machine to complete the construction of the fault diagnosis model based on the support vector machine.
The method comprises the steps of initializing parameters of a fruit fly algorithm to generate a fruit fly population, inputting mechanical fault diagnosis sample data, preprocessing the data to generate a training set, performing classification training on SVM parameters generated by the fruit fly population by using the training set, calculating a fitness value, judging whether an iteration termination condition is met, if not, continuing to update the population, otherwise, outputting a result
Further, the step S3 specifically includes, as shown in fig. 3:
s301, assigning a value to the parameter of the support vector machine obtained in the step S2, wherein the punishment parameter of the support vector machine is represented as C, the kernel parameter is represented as g, and the (C, g) is recorded as the initial position of the fruit fly population;
s302, attaching random direction and distance C to each fruit flyi＝C+rand，giG + rand; wherein, CiRepresenting the updated penalty parameter value; giRepresenting the updated core parameter values; rand denotes a uniformly distributed random number;
s303, if the random number rand is less than the first parameter r set according to the bat algorithmiAnd then, carrying out local search on the fruit fly population: ci＝Ci+0.01×randn，gi＝gi+0.01 × randn, rand representing a random number subject to uniform distribution; otherwise, jumping to the next step;
s304, substituting the obtained fruit fly population position into an evaluation function, and calculating the fitness value of the fruit fly population position; the merit function is expressed as: function (C)i,gi)；function(Ci,gi) Is represented by (C)i,gi) The error classification rate corresponding to the parameters; seThe number of training samples is classified as error, and S is the total number of training samples;
s305, calculating the minimum value of the evaluation function, namely the current optimal solution: fitnessbest＝min(function(Ci,gi))；
S306, carrying out global search on the result of the obtained current optimal solution: if the current optimal solution FitnessbestA second parameter A which is superior to the global optimum best value best and is set according to the bat algorithmiIf the random number is greater than the generated random number rand, the current optimal value is received, and the current optimal value is assigned to the global optimal value, namely best is FitnessbestUpdating the initial position of the step S302 to the position corresponding to the optimal individual drosophila, namely (C, g) → (C)best,gbest) (ii) a Otherwise, abandoning the current optimum value and maintaining the initial position of step S302;
and S307, returning the initial position determined in the step S306 to the step S302 until an iteration termination condition is met, and recording an optimal parameter value at the moment.
Further, after the initial position in step S302 is updated in step S306, if the updated initial position exceeds the value interval of the parameter, the initial position of each drosophila needs to be mapped into the interval where the initial position is located:
Ci＝CLB+U(0，1)×(CUB-CLB)；
gi＝gLB+U(0，1)×(gUB-gLB)；
wherein, CLBIs the lower boundary, g, for CLBIs the lower boundary corresponding to g; cUBIs the upper boundary corresponding to C, gUBIs the upper boundary corresponding to g; u (0, 1) is the interval [0, 1]]Uniformly distributed random numbers.
Further, the first parameter r set according to the bat algorithmiThe parameter r is the pulse transmission frequency in the bat algorithmiIncrement as the operation time advances:
wherein,
representing a first parameter at t cycles; t is the time of day and t is,
in order to be the initial transmission frequency,
has a value range of [0, 1]](ii) a Gamma is a constant.
Further, the second parameter A set according to the bat algorithmiThe loudness in the bat algorithm is given by:
wherein, alpha is a constant, alpha is a linear,
expressed as a second parameter at time t;alpha is 0.85,0.95]. The value α in this embodiment is 0.9.
The present invention will be described in detail below with reference to specific embodiments and the accompanying drawings.
First, in order to verify that the present invention has a good pattern recognition capability, data in a uci (university of California at irvin) machine learning knowledge base is selected as an embodiment for testing, and specific data is shown in table 1.
TABLE 1 data description
After the data are obtained, the operation is carried out according to the flow shown in the abstract drawing, in order to compare the advantages of the invention (FFBA-SVM), grid search SVM parameter and genetic algorithm SVM parameter (GA-SVM) are simultaneously adopted to operate the same data, each operation is carried out for ten times, and the average classification accuracy obtained by each algorithm is recorded in table 2.
Table 2 results of comparing performance of grid search and genetic algorithm
The last column in table 2 is the average classification accuracy of the method of the present invention, which is significantly more advantageous than the other two methods.
In order to further compare the calculation time and the accuracy of the method, the method similar to the method is adopted for comparison, such as particle swarm optimization (PSO-SVM) and basic bat optimization (BA-SVM), the population number and the maximum iteration number in the three methods are respectively set to be 20 and 100, and therefore the calculation time can be compared fairly. The comparative results are reported in table 3.
Table 3 compares the results with the particle swarm algorithm and the original bat algorithm
The results in table 3 show that the particle swarm optimization is adopted, the accuracy is obviously lower than that of the bat algorithm and the method of the invention, because the particle swarm lacks echo positioning operation, the local and global searching capability is weaker, but because of the fact, the particle swarm optimization calculation time is faster, and the invention combines the characteristics of the global optimizing capability of the bat algorithm and the high calculation efficiency of the drosophila algorithm, so the average classification accuracy of the invention is higher than that of the particle swarm optimization, and the average calculation time is ahead of that of the bat algorithm.
In order to verify the application effect of the method in mechanical fault diagnosis, the effectiveness of the method is illustrated by bearing fault data of the university of Keyssierra. The data is obtained by simulating single-point faults at different parts of the bearing by utilizing spark machining grooving, and mainly by setting different grooving widths: 0.007, 0.014, 0.021(1 inch ═ 2.54 cm) to simulate light, medium and severe bearing failure. This experiment selects and uses the main shaft rotational speed 1772rpm, the load is 1HP, sample frequency is 12000Hz the drive end vibration bearing's 9 kinds of trouble vibration signals carry out the analysis, include: slight fault of the outer ring, moderate fault of the outer ring, serious fault of the outer ring, slight fault of the inner ring, moderate fault of the inner ring, serious fault of the inner ring, slight fault of the rolling body, moderate fault of the rolling body and serious fault of the rolling body. Respectively measuring 50 groups of samples under each fault state, wherein the sampling point number of each sample is 2048 points, 20 groups of samples are used as training samples, and the rest 30 groups of samples are used as test samples. 24 time-frequency domain statistical characteristics are extracted from the vibration signals of each sample respectively, and then the SVM with optimized parameters is adopted to carry out fault diagnosis. Fig. 4 shows a fitness curve convergence diagram of the fault diagnosis of the data, and it can be seen from the diagram that the optimal classification accuracy obtained by the invention is 98.89%, while it can be seen from fig. 5 and 6 that the accuracy based on the particle swarm algorithm and the genetic algorithm does not reach the classification accuracy of the invention. Therefore, the invention has good application prospect in mechanical fault diagnosis.
The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, and it is apparent that those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention.
Those skilled in the art will appreciate that all or part of the steps in the methods of the above embodiments may be implemented by associated hardware instructed by a program, which may be stored in a computer-readable storage medium, and the storage medium may include: ROM, RAM, magnetic or optical disks, and the like.
The above-mentioned embodiments, which further illustrate the objects, technical solutions and advantages of the present invention, should be understood that the above-mentioned embodiments are only preferred embodiments of the present invention, and should not be construed as limiting the present invention, and any modifications, equivalents, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (6)
1. A mechanical fault diagnosis method based on an improved fruit fly-bat algorithm is characterized by comprising the following steps:
s1, extracting time domain statistical characteristics and frequency domain statistical characteristics from the collected mechanical operation state signals;
s2, training a support vector machine model by adopting a mechanical fault diagnosis training sample set;
s3, taking the fruit fly algorithm as a frame, integrating the echo positioning idea of the bat algorithm, designing an improved fruit fly-bat parameter optimization method, and searching the global optimal parameter of the support vector machine by adopting the method;
s301, assigning the parameters of the support vector machine obtained in the step S2, wherein the punishment parameters of the support vector machine are represented as C, the kernel parameters are represented as g, and the (C, g) is recorded as the initial position of the fruit fly population;
s302, attaching random direction and distance C to each fruit flyi＝C+rand，giG + rand; it is composed ofIn, CiRepresenting the updated penalty parameter value; giRepresenting the updated core parameter values; rand denotes a random number subject to uniform distribution;
s303, if the random number rand is less than the first parameter r set according to the bat algorithmiAnd then, carrying out local search on the fruit fly population: ci＝Ci+0.01×randn，gi＝gi+0.01 × randn, randn representing a random number following a standard normal distribution; otherwise, jumping to the next step;
s304, substituting the obtained fruit fly population position into an evaluation function, and calculating the fitness value of the fruit fly population position; the merit function is expressed as:
function(Ci,gi) Is represented by (C)i,gi) The error classification rate corresponding to the parameters; seThe number of the training samples classified as errors is obtained, and S is the total number of the training samples;
s305, calculating the minimum value of the evaluation function, namely the current optimal solution: fitnessbest＝min(function(Ci,gi))；
S306, carrying out global search on the result of the obtained current optimal solution: if the current optimal solution FitnessbestA second parameter A which is superior to the global optimum best value best and is set according to the bat algorithmiIf the random number is greater than the generated random number rand, the current optimal value is received, and the current optimal value is assigned to the global optimal value, namely best is FitnessbestUpdating the initial position of the step S302 to the position corresponding to the optimal individual drosophila, namely (C, g) → (C)best,gbest) (ii) a Otherwise, abandoning the current optimum value and maintaining the initial position of step S302;
s307, returning the initial position determined in the step S306 to the step S302 until an iteration termination condition is met, and recording an optimal parameter value at the moment;
and S4, substituting the global optimal parameters obtained in the step S3 into the support vector machine to complete the construction of the fault diagnosis model based on the support vector machine.
2. The method for diagnosing mechanical failure based on improved drosophila-bat algorithm as claimed in claim 1, wherein said support vector machine model in step S2 comprises:
s.t.yi(wTφ(xi)+b)-1+εi≥0；
wherein | | | purple hair2Represents a two-norm; w denotes the normal vector of the hyperplane, wTRepresents a transpose of the w vector; c represents a penalty parameter; epsiloniRepresents a relaxation variable; y isiA label representing a classification of the ith hyperplane; b represents the displacement, i.e. the distance to the hyperplane; phi (x)i) Representing the sample point x after kernel processingi(ii) a N denotes the dimension of the support vector machine.
3. The method for diagnosing mechanical failure based on improved drosophila-bat algorithm as claimed in claim 2, wherein said kernel function is a gaussian kernel function whose expression is:
wherein g denotes the nuclear parameter, xi、xjAll represent sample points; j ∈ (1, 2.. multidot.n).
4. The method as claimed in claim 1, wherein after the step S306 updates the initial position in the step S302, if the updated initial position exceeds the value range of the parameter, the initial position of each drosophila needs to be mapped into the range:
Ci＝CLB+U(0，1)×(CUB-CLB)；
gi＝gLB+U(0，1)×(gUB-gLB)；
wherein, CLBIs the lower boundary, g, for CLBIs the lower boundary corresponding to g; cUBIs the upper boundary corresponding to C, gUBIs the upper boundary corresponding to g; u (0, 1) is in the interval [0, 1]]Uniformly distributed random numbers.
5. The method for diagnosing mechanical failure based on improved drosophila-bat algorithm as claimed in claim 1, wherein the first parameter r set according to the bat algorithmiThe parameter r is the pulse transmission frequency in the bat algorithmiIncrement as the operation time advances:
ri t+1＝ri 0[1-exp(-γt)]
wherein r isi t+1Representing a first parameter at t cycles; t is time, ri 0Is the initial transmission frequency, ri 0Has a value range of [0, 1]](ii) a Gamma is a constant.
6. The method for diagnosing mechanical failure based on improved drosophila-bat algorithm as claimed in claim 1, wherein the second parameter A set according to the bat algorithmiThe loudness in the bat algorithm is given by:
wherein alpha is a constant and has a value range of [0.85,0.95 ]]；
Represents a second parameter at time t + 1;
indicated as the second parameter at time t.
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 CN109116833A (en) | 2019-01-01 |
| CN109116833B true 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) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111159903B (en) * | 2019-12-31 | 2023-07-21 | 重庆邮电大学 | Design and manufacturing method of compact multi-channel multi-fluid heat exchange device |
| CN111639695B (en) * | 2020-05-26 | 2024-02-20 | 温州大学 | Method and system for classifying data based on improved drosophila optimization algorithm |
| CN117407817B (en) * | 2023-10-25 | 2024-07-09 | 国网四川省电力公司绵阳供电公司 | Abnormality monitoring system of power distribution automation machine room |
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 |
|---|
| 果蝇优化算法改进及应用研究;吴易轩;《中国优秀硕士学位论文全文数据库 信息科技辑》;20180215;全文 * |
| 群智能算法优化支持向量机参数综述;李素等;《智能系统学报》;20180130;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109116833A (en) | 2019-01-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111272429B (en) | Bearing fault diagnosis method | |
| CN109116833B (en) | Mechanical fault diagnosis method based on improved fruit fly-bat algorithm | |
| CN110006649B (en) | Bearing fault diagnosis method based on improved ant lion algorithm and support vector machine | |
| CN108073158A (en) | Based on PCA and KNN density algorithm Wind turbines Method for Bearing Fault Diagnosis | |
| CN106777606A (en) | A kind of gearbox of wind turbine failure predication diagnosis algorithm | |
| CN105678343B (en) | Hydropower Unit noise abnormality diagnostic method based on adaptive weighted group of sparse expression | |
| CN106770939A (en) | A kind of Diagnosis Method of Transformer Faults based on support vector description and the nearly barycenter neighbours of K | |
| CN110321940A (en) | The feature extraction of aircraft telemetry and classification method and device | |
| CN110263834B (en) | Method for detecting abnormal value of new energy power quality | |
| CN110987436B (en) | Bearing fault diagnosis method based on excitation mechanism | |
| CN115099296A (en) | Sea wave height prediction method based on deep learning algorithm | |
| CN112729826A (en) | Bearing fault diagnosis method for artificial shoal-frog leaping optimization extreme learning machine | |
| CN112564107A (en) | Transient stability assessment method for power system | |
| CN115600088A (en) | Distribution transformer fault diagnosis method based on vibration signals | |
| CN113538037A (en) | Method, system, equipment and storage medium for monitoring charging event of battery car | |
| CN116010884A (en) | Fault diagnosis method of SSA-LightGBM oil-immersed transformer based on principal component analysis | |
| CN107195297B (en) | Data normalization fused self-adaptive variation bird group voice recognition system | |
| CN115275990A (en) | Evaluation method and system for broadband oscillation risk of regional power grid | |
| CN117131425B (en) | Numerical control machine tool processing state monitoring method and system based on feedback data | |
| CN112362292B (en) | Method for anomaly detection of wind tunnel test data | |
| CN113869451A (en) | Rolling bearing fault diagnosis method under variable working conditions based on improved JGSA algorithm | |
| CN108760813B (en) | Gas turbine blade health monitoring system and method based on temperature signals | |
| CN105116323B (en) | A kind of electrical fault detection method based on RBF | |
| CN116561692A (en) | Dynamic update real-time measurement data detection method | |
| Xiao et al. | Health assessment for piston pump using LSTM neural network |
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 |