CN114970681B - Induction motor fault diagnosis method and system - Google Patents
Induction motor fault diagnosis method and system Download PDFInfo
- Publication number
- CN114970681B CN114970681B CN202210468372.1A CN202210468372A CN114970681B CN 114970681 B CN114970681 B CN 114970681B CN 202210468372 A CN202210468372 A CN 202210468372A CN 114970681 B CN114970681 B CN 114970681B
- Authority
- CN
- China
- Prior art keywords
- fault
- optimal
- model
- vector machine
- support vector
- 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
- 238000003745 diagnosis Methods 0.000 title claims abstract description 266
- 238000000034 method Methods 0.000 title claims abstract description 82
- 230000006698 induction Effects 0.000 title claims abstract description 36
- 238000012706 support-vector machine Methods 0.000 claims abstract description 150
- 239000002245 particle Substances 0.000 claims abstract description 48
- 230000009466 transformation Effects 0.000 claims abstract description 30
- 230000002068 genetic effect Effects 0.000 claims abstract description 17
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 12
- 230000036541 health Effects 0.000 claims abstract description 8
- 238000012549 training Methods 0.000 claims description 71
- 238000012360 testing method Methods 0.000 claims description 35
- 238000012795 verification Methods 0.000 claims description 34
- 238000005457 optimization Methods 0.000 claims description 13
- 239000013598 vector Substances 0.000 claims description 6
- 238000010276 construction Methods 0.000 claims description 5
- 238000000605 extraction Methods 0.000 claims description 3
- 238000006243 chemical reaction Methods 0.000 description 6
- 238000007796 conventional method Methods 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 238000010606 normalization Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 230000035945 sensitivity Effects 0.000 description 3
- 230000007547 defect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 108010074864 Factor XI Proteins 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000004568 cement Substances 0.000 description 1
- 239000003245 coal Substances 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010845 search algorithm Methods 0.000 description 1
- 238000000844 transformation Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
- Y04S10/52—Outage or fault management, e.g. fault detection or location
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Software Systems (AREA)
- Medical Informatics (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Control Of Electric Motors In General (AREA)
Abstract
The invention provides a fault diagnosis method and system of an induction motor, belonging to the field of fault diagnosis of induction motors, wherein the method comprises the following steps: collecting stator current of a motor, performing Hilbert series transformation on the stator current, performing Fourier decomposition on a transformation result, and extracting a fault characteristic frequency component to be diagnosed; inputting the characteristic frequency components of faults to be diagnosed into an optimal support vector machine diagnosis model to perform fault diagnosis on the motor, evaluating the health degree of the motor, and outputting fault types if the motor has faults; the fault characteristic frequency component comprises frequency components of fundamental frequencies of multiples of two and multiples of three corresponding to the asymmetry of the motor structure caused by faults; optimizing core parameters in the support vector machine diagnosis model by adopting a grid search method or a genetic algorithm or a particle swarm algorithm to obtain an optimal support vector machine diagnosis model; the invention effectively solves the problem that the diagnosis accuracy of the traditional method is reduced when the induction motor has slight faults.
Description
Technical Field
The invention belongs to the field of fault diagnosis of induction motors, and particularly relates to a fault diagnosis method and system of an induction motor.
Background
The induction motor has simple structure and high cost performance, and has quite wide application in various industries, but various faults can occur in the running process; if the fault can be diagnosed in early stage and the shutdown maintenance is timely arranged, a great deal of unnecessary loss of manpower, material resources and financial resources can be avoided.
The existing induction motor fault diagnosis method focuses on detection of the stator current characteristic frequency component corresponding to the fault, and ignores the stator current characteristic frequency component corresponding to the motor structure asymmetry caused by the fault. When the fault characteristics of the motor are relatively obvious, the traditional method can obtain a relatively good diagnosis effect; when the failure of the motor is slight, the failure diagnosis accuracy of the conventional method is reduced to some extent, and the situation is in need of improvement.
Disclosure of Invention
Aiming at the defects of the prior art, the invention aims to provide an induction motor fault diagnosis method and system, and aims to solve the problems that the existing induction motor fault diagnosis method ignores stator current characteristic frequency components corresponding to motor structure asymmetry caused by faults, and when the faults of a motor are very slight, the fault diagnosis accuracy is low.
In order to achieve the above object, in one aspect, the present invention provides a fault diagnosis method for an induction motor, comprising the steps of:
Collecting stator current of a motor, performing Hilbert series transformation on the stator current, performing Fourier decomposition on a transformation result, and extracting a fault characteristic frequency component to be diagnosed;
Inputting the fault characteristic frequency component to be diagnosed into an optimal support vector machine diagnosis model to perform fault diagnosis on the motor, evaluating the health degree of the motor, and outputting fault types if the motor has faults;
the method for constructing the optimal support vector machine diagnosis model comprises the following steps:
respectively collecting stator currents of a motor which normally operates and different fault motors, carrying out Hilbert series transformation on the stator currents, carrying out Fourier decomposition on transformation results, and extracting fault characteristic frequency components for constructing a model;
The fault characteristic frequency component used for constructing the model comprises frequency components of fundamental frequencies of multiples of two and multiples of three corresponding to motor structure asymmetry caused by faults;
constructing a support vector machine diagnosis model based on the fault characteristic frequency components for constructing the model;
Optimizing core parameters in the support vector machine diagnosis model by adopting a grid search method or a genetic algorithm or a particle swarm algorithm to obtain an optimal support vector machine diagnosis model;
the core parameters are parameters affecting the performance of the optimal objective function of the support vector machine diagnosis model.
Further preferably, the method of hilbert series transformation is:
squaring the stator current;
Performing Hilbert transform on the stator current, and squaring a transform result;
and summing the squares of the square sum conversion results of the stator currents, and outputting the conversion results of the Hilbert series.
Further preferably, the fault signature frequency components also include stator inter-turn short circuits, rotor cage bar and end ring breaks, air gap eccentricities, and bearing faults.
Further preferably, the training method for supporting the vector machine diagnostic model includes the following steps:
Taking an actual fault class used for training a model and a corresponding class of fault characteristic frequency components used for training the model as one sample, projecting all the samples into a high-dimensional characteristic space, and converting nonlinear fault class identification into linear fault class identification;
Replacing dot product operation in the support vector machine diagnosis model by using a kernel function, wherein the object for dot product operation is fault characteristic frequency for training the model;
and multiplying the actual fault category used for training the model by the corresponding Lagrangian coefficient, and then combining the kernel function, inputting the multiplied actual fault category into a symbol function of the support vector machine diagnosis model, and outputting the predicted fault category to complete the construction of the support vector machine diagnosis model.
Further preferably, the support vector machine diagnostic model is a C-SVC (C-Support Vector Classification) model, and the kernel function is an RBF kernel function (radial basis function: radial Basis Function); the core parameter of the support vector machine diagnosis model is a penalty factor of the C-SVC model;
the core parameters of the RBF kernel function are the weight factors of the kernel function.
Further preferably, the method for optimizing the support vector machine diagnostic model by using the grid search method comprises the following steps:
s1: the fault characteristic frequency used for training the model is divided into a plurality of groups of training sets and test sets after being normalized; each training set and test set is used as a verification set;
S2: initializing optimal fault diagnosis accuracy, optimal penalty factors and optimal weight factors, and setting a support vector machine diagnosis model according to the optimal fault diagnosis accuracy, the optimal penalty factors and the optimal weight factors;
S3: inputting each verification set into a support vector machine diagnosis model under the current iteration, and outputting a predicted fault class;
S4: calculating the fault accuracy corresponding to each verification set by adopting the predicted fault category and the actual fault category for training the model;
S5: selecting the maximum fault accuracy as the fault accuracy under the current iteration;
S6: judging whether the fault accuracy under the current iteration is greater than the optimal fault diagnosis accuracy, if so, respectively updating the optimal fault diagnosis accuracy, the optimal punishment factor and the optimal weight factor into the fault accuracy, the punishment factor and the weight factor under the current iteration, and turning to S7; otherwise, turning to S7;
s7: updating the penalty factors and the weight factors in the support vector machine diagnosis model according to the grid, judging whether the updated penalty factors and the weight factors exceed a preset range, if not, executing S3, otherwise, setting the support vector machine diagnosis model by adopting the optimal penalty factors and the optimal weight factors, and completing the optimization of the support vector machine diagnosis model; wherein the grid is divided into grids based on a preset range of penalty factors and weight factors.
Further preferably, the method for optimizing the support vector machine diagnostic model by adopting a genetic algorithm comprises the following steps:
D1: the fault characteristic frequency used for training the model is divided into a plurality of groups of training sets and test sets after being normalized; each training set and test set is used as a verification set; initializing the population;
D2: initializing optimal fault diagnosis accuracy, optimal penalty factors and optimal weight factors, and setting a support vector machine diagnosis model according to the optimal fault diagnosis accuracy, the optimal penalty factors and the optimal weight factors;
D3: inputting each verification set into a current support vector machine diagnosis model, and obtaining fault accuracy under current iteration;
d4: judging whether the fault accuracy under the current iteration is greater than the optimal fault diagnosis accuracy, if so, turning to D5; otherwise, turning to D6;
D5: updating the optimal fault diagnosis accuracy acc, the optimal penalty factor C and the optimal weight factor gamma coefficient;
D6: calculating the fitness of the population;
D7: judging whether the current iteration number exceeds an evolution algebra, if so, stopping iteration, and outputting an optimal fault diagnosis accuracy acc, an optimal penalty factor C and an optimal weight factor gamma coefficient; otherwise, executing D8;
d8: updating the iteration step length and the iteration direction of the core parameters under the current iteration according to the fitness, further updating the penalty factor C and the weight factor gamma coefficient, and setting the current support vector machine diagnosis model according to the penalty factor C and the weight factor gamma coefficient;
d9: selecting, crossing and mutating the population, updating the population, and adding 1 to the iteration times; go to D3.
Further preferably, the method for optimizing the support vector machine diagnosis model by adopting the particle swarm algorithm comprises the following steps:
K1: the fault characteristic frequency used for training the model is divided into a plurality of groups of training sets and test sets after being normalized; each training set and test set is used as a verification set; initializing particles, particle positions and particle speeds;
K2: initializing optimal fault diagnosis accuracy, optimal penalty factors and optimal weight factors, and setting a support vector machine diagnosis model according to the optimal fault diagnosis accuracy, the optimal penalty factors and the optimal weight factors;
k3: inputting each verification set into a current support vector machine diagnosis model, and calculating the fault accuracy under the current iteration;
and K4: judging whether the fault accuracy under the current iteration is greater than the optimal fault diagnosis accuracy, if so, switching to K5; otherwise, turning to K6;
And K5: updating the optimal fault diagnosis accuracy acc, the optimal penalty factor C and the optimal weight factor gamma coefficient;
k6: calculating the adaptability of the particle swarm;
K7: judging whether the current iteration number exceeds an evolution algebra, if so, stopping iteration, and outputting an optimal fault diagnosis accuracy acc, an optimal penalty factor C and an optimal weight factor gamma coefficient; otherwise, executing K8;
And K8: updating the iteration step length and the iteration direction of the core parameters under the current iteration according to the fitness, further updating the penalty factor C and the weight factor gamma coefficient, and setting the current support vector machine diagnosis model according to the penalty factor C and the weight factor gamma coefficient;
k9: searching individual and group extremum, updating the position and speed of particles, and adding 1 to the iteration times; go to K3.
In another aspect, the present invention provides an induction motor fault diagnosis system, comprising:
The characteristic extraction module is used for collecting stator current of the motor, carrying out Hilbert series transformation on the stator current, carrying out Fourier decomposition on a transformation result, and extracting a fault characteristic frequency component to be diagnosed;
The method comprises the steps of respectively collecting stator currents of a motor which normally operates and different fault motors, carrying out Hilbert series transformation on the stator currents, carrying out Fourier decomposition on transformation results, and extracting fault characteristic frequency components for constructing a model;
the characteristic frequency components used for constructing the model comprise frequency components of fundamental frequencies of multiples of two and multiples of three corresponding to motor structure asymmetry caused by faults;
the support vector machine diagnosis module is internally provided with a support vector machine diagnosis model and is used for inputting the characteristic frequency component of the fault to be diagnosed into the optimal support vector machine diagnosis model to perform fault diagnosis on the motor, evaluating the health degree of the motor, and outputting fault types if the motor has faults; the building module of the support vector machine diagnosis model is used for building an optimal support vector machine diagnosis model and setting the optimal support vector machine diagnosis model in the support vector machine diagnosis module, and comprises the following components:
the model building unit is used for building a support vector machine diagnosis model based on the fault characteristic frequency components used for building the model;
The model optimization unit is used for optimizing core parameters in the support vector machine diagnosis model by adopting a grid search method or a genetic algorithm or a particle swarm algorithm to obtain an optimal support vector machine diagnosis model;
the core parameters are parameters affecting the performance of the optimal objective function of the support vector machine diagnosis model.
Further preferably, the method of hilbert series transformation is:
squaring the stator current;
Performing Hilbert transform on the stator current, and squaring a transform result;
and summing the squares of the square sum conversion results of the stator currents, and outputting the conversion results of the Hilbert series.
Further preferably, the fault signature frequency components also include stator inter-turn short circuits, rotor cage bar and end ring breaks, air gap eccentricities, and bearing faults.
Further preferably, the specific execution method of the model building unit includes the steps of:
Taking an actual fault class used for training a model and a corresponding class of fault characteristic frequency components used for training the model as one sample, projecting all the samples into a high-dimensional characteristic space, and converting nonlinear fault class identification into linear fault class identification;
In the training process of the support vector machine diagnosis model, kernel functions are utilized to replace dot product operation in the support vector machine diagnosis model, wherein the object for dot product operation is the fault characteristic frequency for training the model.
And multiplying the actual fault category used for training the model by the corresponding Lagrangian coefficient, and then combining the kernel function, inputting the multiplied actual fault category into a symbol function of the support vector machine diagnosis model, and outputting the predicted fault category to complete the construction of the support vector machine diagnosis model.
Further preferably, the support vector machine diagnostic model is a C-SVC model and the kernel function is an RBF kernel function; the core parameter of the support vector machine diagnosis model is a penalty factor of the C-SVC model; the core parameters of the RBF kernel function are the weight factors of the kernel function.
Further preferably, the model optimizing unit optimizes the method for supporting the vector machine diagnosis model by using a grid search method, comprising the steps of:
s1: the fault characteristic frequency used for training the model is divided into a plurality of groups of training sets and test sets after being normalized; each training set and test set is used as a verification set;
S2: initializing optimal fault diagnosis accuracy, optimal penalty factors and optimal weight factors, and setting a support vector machine diagnosis model according to the optimal fault diagnosis accuracy, the optimal penalty factors and the optimal weight factors;
S3: inputting each verification set into a support vector machine diagnosis model under the current iteration, and outputting a predicted fault class;
S4: calculating the fault accuracy corresponding to each verification set by adopting the predicted fault category and the actual fault category for training the model;
S5: selecting the maximum fault accuracy as the fault accuracy under the current iteration;
S6: judging whether the fault accuracy under the current iteration is greater than the optimal fault diagnosis accuracy, if so, respectively updating the optimal fault diagnosis accuracy, the optimal punishment factor and the optimal weight factor into the fault accuracy, the punishment factor and the weight factor under the current iteration, and turning to S7; otherwise, turning to S7;
s7: updating the penalty factors and the weight factors in the support vector machine diagnosis model according to the grid, judging whether the updated penalty factors and the weight factors exceed a preset range, if not, executing S3, otherwise, setting the support vector machine diagnosis model by adopting the optimal penalty factors and the optimal weight factors, and completing the optimization of the support vector machine diagnosis model; wherein the grid is divided into grids based on a preset range of penalty factors and weight factors.
Further preferably, the method for optimizing the support vector machine diagnosis model by the model optimizing unit by adopting a genetic algorithm comprises the following steps:
D1: the fault characteristic frequency used for training the model is divided into a plurality of groups of training sets and test sets after being normalized; each training set and test set is used as a verification set; initializing the population;
D2: initializing optimal fault diagnosis accuracy, optimal penalty factors and optimal weight factors, and setting a support vector machine diagnosis model according to the optimal fault diagnosis accuracy, the optimal penalty factors and the optimal weight factors;
D3: inputting each verification set into a current support vector machine diagnosis model, and obtaining fault accuracy under current iteration;
d4: judging whether the fault accuracy under the current iteration is greater than the optimal fault diagnosis accuracy, if so, turning to D5; otherwise, turning to D6;
D5: updating the optimal fault diagnosis accuracy acc, the optimal penalty factor C and the optimal weight factor gamma coefficient;
D6: calculating the fitness of the population;
D7: judging whether the current iteration number exceeds an evolution algebra, if so, stopping iteration, and outputting an optimal fault diagnosis accuracy acc, an optimal penalty factor C and an optimal weight factor gamma coefficient; otherwise, executing D8;
d8: updating the iteration step length and the iteration direction of the core parameters under the current iteration according to the fitness, further updating the penalty factor C and the weight factor gamma coefficient, and setting the current support vector machine diagnosis model according to the penalty factor C and the weight factor gamma coefficient;
d9: selecting, crossing and mutating the population, updating the population, and adding 1 to the iteration times; go to D3.
The method for optimizing the support vector machine diagnosis model by the model optimizing unit by adopting a particle swarm algorithm comprises the following steps:
K1: the fault characteristic frequency used for training the model is divided into a plurality of groups of training sets and test sets after being normalized; each training set and test set is used as a verification set; initializing particles, particle positions and particle speeds;
K2: initializing optimal fault diagnosis accuracy, optimal penalty factors and optimal weight factors, and setting a support vector machine diagnosis model according to the optimal fault diagnosis accuracy, the optimal penalty factors and the optimal weight factors;
k3: inputting each verification set into a current support vector machine diagnosis model, and calculating the fault accuracy under the current iteration;
and K4: judging whether the fault accuracy under the current iteration is greater than the optimal fault diagnosis accuracy, if so, switching to K5; otherwise, turning to K6;
And K5: updating the optimal fault diagnosis accuracy acc, the optimal penalty factor C and the optimal weight factor gamma coefficient;
k6: calculating the adaptability of the particle swarm;
K7: judging whether the current iteration number exceeds an evolution algebra, if so, stopping iteration, and outputting an optimal fault diagnosis accuracy acc, an optimal penalty factor C and an optimal weight factor gamma coefficient; otherwise, executing K8;
And K8: updating the iteration step length and the iteration direction of the core parameters under the current iteration according to the fitness, further updating the penalty factor C and the weight factor gamma coefficient, and setting the current support vector machine diagnosis model according to the penalty factor C and the weight factor gamma coefficient;
k9: searching individual and group extremum, updating the position and speed of particles, and adding 1 to the iteration times; go to K3.
In general, the above technical solutions conceived by the present invention have the following beneficial effects compared with the prior art:
Based on the existing stator current fault characteristic frequency, the invention additionally considers the stator current characteristic frequency component which is caused by faults and corresponds to the motor structure asymmetry, namely frequency components of the fundamental frequency of multiples of two and multiples of three; amplifying the fault frequency component and separating the fault frequency component from the fundamental frequency component using a hilbert transform; the method comprises the steps of adopting a support vector machine method to improve diagnosis sensitivity and optimizing core parameters of a support vector machine diagnosis model through a grid search algorithm, a particle swarm algorithm and a genetic algorithm; the problem that the diagnosis accuracy of the traditional method is reduced when the induction motor has slight faults is effectively solved.
The invention adopts genetic algorithm and particle swarm algorithm to iterate the variable step length of the core parameters of the support vector machine diagnosis model, overcomes the defects of fixed iteration step length and limited iteration range of the grid search method, shortens the diagnosis time and improves the diagnosis accuracy.
Drawings
FIG. 1 is a flow chart of a method for diagnosing a slight fault of an induction motor according to an embodiment of the present invention;
FIG. 2 is a diagnostic flow of a support vector machine diagnostic model provided by an embodiment of the present invention;
FIG. 3 is a schematic diagram of the iteration range of core parameters of a support vector machine diagnostic model according to an embodiment of the present invention;
FIG. 4 is an iterative flow chart of a grid search method provided by an embodiment of the present invention;
FIG. 5 is an iterative flow chart of a particle swarm algorithm provided by an embodiment of the invention;
FIG. 6 is a flowchart of an iteration of a genetic algorithm provided by an embodiment of the present invention;
FIG. 7 is a schematic diagram of a fault diagnosis result of a conventional method according to an embodiment of the present invention;
fig. 8 is a schematic diagram of fault diagnosis results of the improved method according to the embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
Aiming at the problem of low accuracy of light fault diagnosis of the existing induction motor, the invention provides a light fault diagnosis method of the induction motor, which has the following overall thought: based on the original stator current fault characteristic frequency, additionally considering the stator current characteristic frequency component corresponding to the motor structure asymmetry caused by the fault; amplifying the fault frequency component and separating the fault frequency component from the fundamental frequency component using a hilbert transform; the sensitivity of diagnosis is improved by adopting a support vector machine method; and finally, the diagnosis accuracy of the induction motor when slight faults occur is effectively improved.
Example 1
As shown in fig. 1, a method for diagnosing a slight fault of an induction motor includes the steps of:
(1) Respectively collecting stator currents of a normal motor and different fault motors, carrying out Hilbert series transformation on the stator currents, carrying out Fourier decomposition on a transformation result, and extracting fault characteristic frequency components;
The fault characteristic frequency component also comprises frequency components of frequency of fundamental waves of multiples of two and multiples of three which are asymmetrically corresponding to the motor structure caused by faults;
In the embodiment, a Hall current sensor is adopted to collect the stator current of the motor;
in an alternative embodiment, the performing a hilbert series of transformations on the stator current in step (1) includes:
Squaring the stator current;
performing Hilbert transform on the stator current, and squaring a transform result;
Summing the two square results, and outputting a Hilbert series transformation result;
before fault diagnosis, determining fault types and characteristic frequency components corresponding to faults; for example, in the present embodiment, the main classes of faults considered are stator inter-turn short circuit, rotor cage bar and end ring breaks, air gap eccentricity, and bearing faults; each fault corresponds to a different characteristic frequency component, and besides, the fault characteristic frequency component also comprises frequency components of frequency of multiple times of two and frequency of multiple times of three, which are asymmetrically corresponding to the motor structure caused by the fault; wherein the fundamental wave frequency is 50Hz;
(2) Constructing a support vector machine diagnosis model according to the characteristic frequency components, and optimizing core parameters of the model to obtain an optimal diagnosis model;
in step (2) of the present embodiment, as shown in fig. 2, the support vector machine diagnosis model is:
Projecting a sample (x i,yi) into a high-dimensional feature space, replacing dot product operation of a support vector machine diagnosis model by using a kernel function K (x i, x), and then carrying alpha i *yi into a symbol function f (x) to output a fault diagnosis class; wherein f (x) is an output function of the support vector machine diagnosis model, and the result is a fault type; x is the fault class of the sample to be predicted;
The specific expression form of f (x) is as follows:
Wherein x i is a fault-characteristic frequency component; y i is the fault class; since there are at least three fault categories, the sample (x i,yi) needs to be projected to the high-dimensional feature space; alpha i * represents the Lagrangian coefficient; b * denotes offset; i=1, 2,3, …, n, n is an integer, representing the number of samples;
It is to be noted that: one fault category corresponds to a series of fault frequencies, not to one fault frequency;
In an alternative embodiment, in step (2), a method for constructing a support vector machine diagnostic model includes the steps of:
The support vector machine diagnosis model selects a C-Support Vector Classification (C-SVC for short) model, the model has a punishment factor C and a relaxation factor xi, and the C-SVC model can give consideration to the diagnosis accuracy and the control of the error-divided sample proportion; the support vector machine model can form an efficient diagnosis model by matching with a kernel function; the kernel function selects RBF kernel function, the kernel parameter of the function is weight factor gamma coefficient, the use of the kernel function can avoid complex dot product operation in high-dimensional characteristic space, the diagnosis time is greatly reduced, and the rapidity of diagnosis is improved;
The RBF kernel function specifically comprises the following steps:
Wherein σ is the standard deviation of the kernel function; the RBF kernel function is expressed as a weight factor gamma coefficient:
K(x,xi)＝exp(-γ||x-xi||2)γ>0
Thus, γ=1/2σ 2;
obtaining optimal core parameters by an iterative optimization method, and constructing an optimal support vector machine diagnosis model based on the optimal core parameters;
In this embodiment, the core parameters are penalty factor C and weighting factor γ coefficients; the iterative optimization method is a grid searching method, a genetic algorithm and a particle swarm algorithm, and the grid searching method is simple and easy to implement and is suitable for occasions with fewer motors and fewer fault types; the genetic algorithm and the particle swarm algorithm are relatively complex, and are suitable for occasions with more motors and more fault types; the genetic algorithm and the particle swarm algorithm are relatively complex, and are suitable for occasions with more motors and more fault types; under the same condition, a core parameter combination with smaller punishment factor C in the optimization result is preferably selected, so that a large number of support vectors can be avoided to be constructed, diagnosis time is saved, over-learning is prevented, and generalization capability of a diagnosis model is improved;
in step (2) of the present embodiment, as shown in fig. 3, 4, 5 and 6, the iterative optimization method further includes:
As shown in fig. 4, before iteration, the grid search method initializes the optimal fault diagnosis accuracy acc, the optimal penalty factor C and the optimal weight factor gamma coefficient, and performs normalization processing on a characteristic sample formed by collecting stator current data, and then divides the characteristic sample into a training set and a testing set; cross checking is carried out according to the initialization data and the training set and the testing set, and fault diagnosis accuracy is output; after the updating data is completed by any iteration, performing cross verification; judging whether the current acc is larger than the optimal acc according to the cross verification result, if so, updating the optimal fault diagnosis accuracy acc, the optimal penalty factor C and the optimal weight factor gamma coefficient, and then selecting the next parameter combination according to FIG. 3 for cross verification; if not, directly selecting the next parameter combination for cross test until the iteration is finished beyond the parameter combination range, and outputting the optimal fault diagnosis accuracy acc, the optimal penalty factor C and the optimal weight factor gamma coefficient; the parameters here are C and γ shown in FIG. 3;
as shown in fig. 5, a specific implementation method of the genetic algorithm includes the following steps:
(1) Dividing the fault characteristic frequency into a plurality of groups of training sets and test sets after normalization processing; each training set and test set is used as a verification set; initializing the population;
(2) Initializing optimal fault diagnosis accuracy, optimal penalty factors and optimal weight factors, and setting a support vector machine diagnosis model according to the optimal fault diagnosis accuracy, the optimal penalty factors and the optimal weight factors;
(3) Inputting each verification set into a current support vector machine diagnosis model, and calculating the fault accuracy under the current iteration;
(4) Judging whether the fault accuracy under the current iteration is greater than the optimal fault diagnosis accuracy, if so, turning to the step (5); otherwise go to step (6);
(5) Updating the optimal fault diagnosis accuracy acc, the optimal penalty factor C and the optimal weight factor gamma coefficient;
(6) Calculating the fitness of the population;
(7) Judging whether the current iteration number exceeds an evolution algebra, if so, stopping iteration, and outputting an optimal fault diagnosis accuracy acc, an optimal penalty factor C and an optimal weight factor gamma coefficient; otherwise, executing the step (8);
(8) Updating the iteration step length and the iteration direction of the core parameters under the current iteration according to the fitness, further updating the penalty factor C and the weight factor gamma coefficient, and setting the current support vector machine diagnosis model according to the penalty factor C and the weight factor gamma coefficient;
(9) Selecting, crossing and mutating the population, updating the population, and adding 1 to the iteration times; turning to step (3).
The iteration range of the core parameters is larger than that of the grid searching method;
As shown in fig. 6, a specific implementation method of the particle swarm algorithm includes the following steps:
(1) Dividing the fault characteristic frequency into a plurality of groups of training sets and test sets after normalization processing; each training set and test set is used as a verification set; initializing particles, particle positions and particle speeds;
(2) Initializing optimal fault diagnosis accuracy, optimal penalty factors and optimal weight factors, and setting a support vector machine diagnosis model according to the optimal fault diagnosis accuracy, the optimal penalty factors and the optimal weight factors;
(3) Inputting each verification set into a current support vector machine diagnosis model, and obtaining fault accuracy under current iteration;
(4) Judging whether the fault accuracy under the current iteration is greater than the optimal fault diagnosis accuracy, if so, turning to the step (5); otherwise go to step (6);
(5) Updating the optimal fault diagnosis accuracy acc, the optimal penalty factor C and the optimal weight factor gamma coefficient;
(6) Calculating the adaptability of the particle swarm;
(7) Judging whether the current iteration number exceeds an evolution algebra, if so, stopping iteration, and outputting an optimal fault diagnosis accuracy acc, an optimal penalty factor C and an optimal weight factor gamma coefficient; otherwise, executing the step (8);
(8) Updating the iteration step length and the iteration direction of the core parameters under the current iteration according to the fitness, further updating the penalty factor C and the weight factor gamma coefficient, and setting the current support vector machine diagnosis model according to the penalty factor C and the weight factor gamma coefficient;
(9) Searching individual and group extremum, updating the position and speed of particles, and adding 1 to the iteration times; turning to step (3). The method for optimizing the support vector machine diagnosis model by adopting the particle swarm optimization comprises the following steps:
K1: the fault characteristic frequency used for training the model is divided into a plurality of groups of training sets and test sets after being normalized; each training set and test set is used as a verification set; initializing particles, particle positions and particle speeds;
K2: initializing optimal fault diagnosis accuracy, optimal penalty factors and optimal weight factors, and setting a support vector machine diagnosis model according to the optimal fault diagnosis accuracy, the optimal penalty factors and the optimal weight factors;
k3: inputting each verification set into a current support vector machine diagnosis model, and calculating the fault accuracy under the current iteration;
and K4: judging whether the fault accuracy under the current iteration is greater than the optimal fault diagnosis accuracy, if so, switching to K5; otherwise, turning to K6;
And K5: updating the optimal fault diagnosis accuracy acc, the optimal penalty factor C and the optimal weight factor gamma coefficient;
k6: calculating the adaptability of the particle swarm;
K7: judging whether the current iteration number exceeds an evolution algebra, if so, stopping iteration, and outputting an optimal fault diagnosis accuracy acc, an optimal penalty factor C and an optimal weight factor gamma coefficient; otherwise, executing K8;
And K8: updating the iteration step length and the iteration direction of the core parameters under the current iteration according to the fitness, further updating the penalty factor C and the weight factor gamma coefficient, and setting the current support vector machine diagnosis model according to the penalty factor C and the weight factor gamma coefficient;
K9: searching individual and group extremum, updating the position and speed of particles, and adding 1 to the iteration times; turning to K3; the meaning of the cross test is as follows: dividing a fault sample into a plurality of training sets and a test set, wherein the training sets are used for constructing a diagnosis model; the test set is used for evaluating the performance of the diagnostic model;
(3) Performing online real-time fault diagnosis by adopting an optimal diagnosis model, and evaluating the health degree of the motor;
The diagnosis method corresponding to the embodiment can perform online diagnosis in the process of putting the motor into use; and estimating the overall health level of the induction motor according to the fault type and the fault component size obtained through diagnosis.
According to the embodiment, on the basis of the original stator current fault characteristic frequency, the stator current characteristic frequency component corresponding to the motor structure asymmetry caused by the fault is additionally considered, and the Hilbert transformation and support vector machine method are combined to improve the diagnosis sensitivity and effectively improve the diagnosis accuracy when the induction motor has slight faults. The method has high reliability and high diagnosis accuracy, and is suitable for any system based on induction motor driving, such as a coal mine power system based on induction motor driving, a cement plant power system and the like.
The above embodiment is utilized to diagnose the slight fault of the induction motor, and the diagnosis results provided based on the conventional method and the embodiment are shown in fig. 7 and 8 respectively; wherein the fault categories "1", "2", "3" and "4" represent no fault, stator inter-turn short circuit fault, rotor cage bar and end ring fracture fault, air gap eccentricity and bearing fault, respectively. According to the test results shown in fig. 7 and 8, based on the above embodiment, under the condition of 100 data samples, the diagnosis accuracy is improved from 87% to 96%, and the effect is remarkable, so that the invention provides a method for diagnosing the slight fault of the induction motor, which can effectively solve the technical problem of low accuracy of the conventional method for diagnosing the slight fault of the induction motor.
Example 2
The invention provides an induction motor fault diagnosis system, comprising:
The characteristic extraction module is used for collecting stator current of the motor, carrying out Hilbert series transformation on the stator current, carrying out Fourier decomposition on a transformation result, and extracting a fault characteristic frequency component to be diagnosed;
The method comprises the steps of respectively collecting stator currents of a motor which normally operates and different fault motors, carrying out Hilbert series transformation on the stator currents, carrying out Fourier decomposition on transformation results, and extracting fault characteristic frequency components for constructing a model;
the characteristic frequency components used for constructing the model comprise frequency components of fundamental frequencies of multiples of two and multiples of three corresponding to motor structure asymmetry caused by faults;
the support vector machine diagnosis module is internally provided with a support vector machine diagnosis model and is used for inputting the characteristic frequency component of the fault to be diagnosed into the optimal support vector machine diagnosis model to perform fault diagnosis on the motor, evaluating the health degree of the motor, and outputting fault types if the motor has faults; the building module of the support vector machine diagnosis model is used for building an optimal support vector machine diagnosis model and setting the optimal support vector machine diagnosis model in the support vector machine diagnosis module, and comprises the following components:
the model building unit is used for building a support vector machine diagnosis model based on the fault characteristic frequency components used for building the model;
The model optimization unit is used for optimizing core parameters in the support vector machine diagnosis model by adopting a grid search method or a genetic algorithm or a particle swarm algorithm to obtain an optimal support vector machine diagnosis model;
the core parameters are parameters affecting the performance of the optimal objective function of the support vector machine diagnosis model.
Further preferably, the method of hilbert series transformation is:
squaring the stator current;
Performing Hilbert transform on the stator current, and squaring a transform result;
and summing the squares of the square sum conversion results of the stator currents, and outputting the conversion results of the Hilbert series.
Further preferably, the fault signature frequency components also include stator inter-turn short circuits, rotor cage bar and end ring breaks, air gap eccentricities, and bearing faults.
Further preferably, the specific execution method of the model building unit includes the steps of:
Taking an actual fault class used for training a model and a corresponding class of fault characteristic frequency components used for training the model as one sample, projecting all the samples into a high-dimensional characteristic space, and converting nonlinear fault class identification into linear fault class identification;
In the training process of the support vector machine diagnosis model, kernel functions are utilized to replace dot product operation in the support vector machine diagnosis model, wherein the object for dot product operation is the fault characteristic frequency for training the model.
And multiplying the actual fault category used for training the model by the corresponding Lagrangian coefficient, and then combining the kernel function, inputting the multiplied actual fault category into a symbol function of the support vector machine diagnosis model, and outputting the predicted fault category to complete the construction of the support vector machine diagnosis model.
Further preferably, the support vector machine diagnostic model is a C-SVC model and the kernel function is an RBF kernel function; the core parameter of the support vector machine diagnosis model is a penalty factor of the C-SVC model; the core parameters of the RBF kernel function are the weight factors of the kernel function.
The optimization method of the support vector machine diagnosis model is consistent with the optimization method in the induction motor fault diagnosis method.
It will be readily appreciated by those skilled in the art that the foregoing description is merely a preferred embodiment of the invention and is not intended to limit the invention, but any modifications, equivalents, improvements or alternatives falling within the spirit and principles of the invention are intended to be included within the scope of the invention.
Claims (10)
1.A method for diagnosing faults of an induction motor, comprising the steps of:
Collecting stator current of a motor, performing Hilbert series transformation on the stator current, performing Fourier decomposition on a transformation result, and extracting a fault characteristic frequency component to be diagnosed;
Inputting the fault characteristic frequency component to be diagnosed into an optimal support vector machine diagnosis model to perform fault diagnosis on the motor, evaluating the health degree of the motor, and outputting fault types if the motor has faults;
the method for constructing the optimal support vector machine diagnosis model comprises the following steps:
respectively collecting stator currents of a motor which normally operates and different fault motors, carrying out Hilbert series transformation on the stator currents, carrying out Fourier decomposition on transformation results, and extracting fault characteristic frequency components for constructing a model;
The fault characteristic frequency component used for constructing the model comprises frequency components of fundamental frequencies of multiples of two and multiples of three corresponding to motor structure asymmetry caused by faults;
constructing a support vector machine diagnosis model based on the fault characteristic frequency components for constructing the model;
Optimizing core parameters in the support vector machine diagnosis model by adopting a grid search method or a genetic algorithm or a particle swarm algorithm to obtain an optimal support vector machine diagnosis model;
the core parameters are parameters affecting the performance of the optimal objective function of the support vector machine diagnosis model.
2. The method for diagnosing faults in an induction motor according to claim 1, wherein the training method for supporting a diagnosis model of a vector machine comprises the steps of:
Taking an actual fault class used for training a model and a corresponding class of fault characteristic frequency components used for training the model as one sample, projecting all the samples into a high-dimensional characteristic space, and converting nonlinear fault class identification into linear fault class identification;
Replacing dot product operation in the support vector machine diagnosis model by using a kernel function, wherein the object for dot product operation is fault characteristic frequency for training the model;
and multiplying the actual fault category used for training the model by the corresponding Lagrangian coefficient, and then combining the kernel function, inputting the multiplied actual fault category into a symbol function of the support vector machine diagnosis model, and outputting the predicted fault category to complete the construction of the support vector machine diagnosis model.
3. The induction motor fault diagnosis method according to claim 1 or 2, characterized in that the support vector machine diagnosis model is a C-SVC model, and the kernel function is an RBF kernel function; the core parameter of the support vector machine diagnosis model is a penalty factor of the C-SVC model; the core parameters of the RBF kernel function are the weight factors of the kernel function.
4. A method for diagnosing an induction motor fault as claimed in claim 3, wherein the method for optimizing the support vector machine diagnostic model using a grid search method comprises the steps of:
s1: the fault characteristic frequency used for training the model is divided into a plurality of groups of training sets and test sets after being normalized; each training set and test set is used as a verification set;
s2: initializing optimal fault diagnosis accuracy, optimal penalty factors and optimal weight factors, and building a support vector machine diagnosis model according to the optimal fault diagnosis accuracy, the optimal penalty factors and the optimal weight factors;
S3: inputting each verification set into a support vector machine diagnosis model under the current iteration, and outputting a predicted fault class;
S4: calculating the fault accuracy corresponding to each verification set according to the predicted fault category and the actual fault category used for training the model;
S5: selecting the maximum fault accuracy as the fault accuracy under the current iteration;
S6: judging whether the fault accuracy under the current iteration is greater than the optimal fault diagnosis accuracy, if so, respectively updating the optimal fault diagnosis accuracy, the optimal punishment factor and the optimal weight factor into the fault accuracy, the punishment factor and the weight factor under the current iteration, and turning to S7; otherwise, turning to S7;
s7: updating the penalty factors and the weight factors in the support vector machine diagnosis model according to the grid, judging whether the updated penalty factors and the weight factors exceed a preset range, if not, executing S3, otherwise, setting the support vector machine diagnosis model by adopting the optimal penalty factors and the optimal weight factors, and completing the optimization of the support vector machine diagnosis model; wherein the grid is divided into grids based on a preset range of penalty factors and weight factors.
5. The method of claim 4, wherein the method of optimizing the support vector machine diagnostic model using a genetic algorithm comprises the steps of:
D1: the fault characteristic frequency used for training the model is divided into a plurality of groups of training sets and test sets after being normalized; each training set and test set is used as a verification set; initializing the population;
d2: initializing optimal fault diagnosis accuracy, optimal penalty factors and optimal weight factors, and building a support vector machine diagnosis model according to the optimal fault diagnosis accuracy, the optimal penalty factors and the optimal weight factors;
d3: inputting each verification set into a current support vector machine diagnosis model, and calculating the fault accuracy under the current iteration;
d4: judging whether the fault accuracy under the current iteration is greater than the optimal fault diagnosis accuracy, if so, turning to D5; otherwise, turning to D6;
D5: updating the optimal fault diagnosis accuracy acc, the optimal penalty factor C and the optimal weight factor gamma coefficient;
D6: calculating the fitness of the population;
D7: judging whether the current iteration number exceeds an evolution algebra, if so, stopping iteration, and outputting an optimal fault diagnosis accuracy acc, an optimal penalty factor C and an optimal weight factor gamma coefficient; otherwise, executing D8;
d8: updating the iteration step length and the iteration direction of the core parameters under the current iteration according to the fitness, further updating the penalty factor C and the weight factor gamma coefficient, and setting the current support vector machine diagnosis model according to the penalty factor C and the weight factor gamma coefficient;
d9: selecting, crossing and mutating the population, updating the population, and adding 1 to the iteration times; go to D3.
6. The method of claim 4, wherein the method for optimizing the support vector machine diagnostic model using the particle swarm algorithm comprises the steps of:
K1: the fault characteristic frequency used for training the model is divided into a plurality of groups of training sets and test sets after being normalized; each training set and test set is used as a verification set; initializing particles, particle positions and particle speeds;
K2: initializing optimal fault diagnosis accuracy, optimal penalty factors and optimal weight factors, and setting a support vector machine diagnosis model according to the optimal fault diagnosis accuracy, the optimal penalty factors and the optimal weight factors;
k3: inputting each verification set into a current support vector machine diagnosis model, and calculating the fault accuracy under the current iteration;
and K4: judging whether the fault accuracy under the current iteration is greater than the optimal fault diagnosis accuracy, if so, switching to K5; otherwise, turning to K6;
And K5: updating the optimal fault diagnosis accuracy acc, the optimal penalty factor C and the optimal weight factor gamma coefficient;
k6: calculating the adaptability of the particle swarm;
K7: judging whether the current iteration number exceeds an evolution algebra, if so, stopping iteration, and outputting an optimal fault diagnosis accuracy acc, an optimal penalty factor C and an optimal weight factor gamma coefficient; otherwise, executing K8;
And K8: updating the iteration step length and the iteration direction of the core parameters under the current iteration according to the fitness, further updating the penalty factor C and the weight factor gamma coefficient, and setting the current support vector machine diagnosis model according to the penalty factor C and the weight factor gamma coefficient;
k9: searching individual and group extremum, updating the position and speed of particles, and adding 1 to the iteration times; go to K3.
7. An induction motor fault diagnosis system, comprising:
The characteristic extraction module is used for collecting stator current of the motor, carrying out Hilbert series transformation on the stator current, carrying out Fourier decomposition on a transformation result, and extracting a fault characteristic frequency component to be diagnosed;
The method comprises the steps of respectively collecting stator currents of a motor which normally operates and different fault motors, carrying out Hilbert series transformation on the stator currents, carrying out Fourier decomposition on transformation results, and extracting fault characteristic frequency components for constructing a model;
The fault characteristic frequency component used for constructing the model comprises frequency components of fundamental frequencies of multiples of two and multiples of three corresponding to motor structure asymmetry caused by faults;
the support vector machine diagnosis module is internally provided with a support vector machine diagnosis model and is used for inputting the characteristic frequency component of the fault to be diagnosed into the optimal support vector machine diagnosis model to perform fault diagnosis on the motor, evaluating the health degree of the motor, and outputting fault types if the motor has faults; the building module of the support vector machine diagnosis model is used for building an optimal support vector machine diagnosis model and setting the optimal support vector machine diagnosis model in the support vector machine diagnosis module, and comprises the following components:
the model building unit is used for building a support vector machine diagnosis model based on the fault characteristic frequency components used for building the model;
The model optimization unit is used for optimizing core parameters in the support vector machine diagnosis model by adopting a grid search method or a genetic algorithm or a particle swarm algorithm to obtain an optimal support vector machine diagnosis model;
the core parameters are parameters affecting the performance of the optimal objective function of the support vector machine diagnosis model.
8. The induction motor fault diagnosis system according to claim 7, characterized in that the specific execution method of the model building unit comprises the steps of:
Taking an actual fault class used for training a model and a corresponding class of fault characteristic frequency components used for training the model as one sample, projecting all the samples into a high-dimensional characteristic space, and converting nonlinear fault class identification into linear fault class identification;
Replacing dot product operation in the support vector machine diagnosis model by using a kernel function, wherein the object for dot product operation is fault characteristic frequency for training the model;
and multiplying the actual fault category used for training the model by the corresponding Lagrangian coefficient, and then combining the kernel function, inputting the multiplied actual fault category into a symbol function of the support vector machine diagnosis model, and outputting the predicted fault category to complete the construction of the support vector machine diagnosis model.
9. The induction motor fault diagnosis system according to claim 8, wherein the support vector machine diagnosis model is a C-SVC model and the kernel function is an RBF kernel function; the core parameter of the support vector machine diagnosis model is a penalty factor of the C-SVC model; the core parameters of the RBF kernel function are the weight factors of the kernel function.
10. The induction motor fault diagnosis system according to claim 9, wherein the method for optimizing the support vector machine diagnosis model by the model optimizing unit using a genetic algorithm comprises the steps of:
D1: the fault characteristic frequency used for training the model is divided into a plurality of groups of training sets and test sets after being normalized; each training set and test set is used as a verification set; initializing the population;
D2: initializing optimal fault diagnosis accuracy, optimal penalty factors and optimal weight factors, and setting a support vector machine diagnosis model according to the optimal fault diagnosis accuracy, the optimal penalty factors and the optimal weight factors;
D3: inputting each verification set into a current support vector machine diagnosis model, and obtaining fault accuracy under current iteration;
d4: judging whether the fault accuracy under the current iteration is greater than the optimal fault diagnosis accuracy, if so, turning to D5; otherwise, turning to D6;
D5: updating the optimal fault diagnosis accuracy acc, the optimal penalty factor C and the optimal weight factor gamma coefficient;
D6: calculating the fitness of the population;
D7: judging whether the current iteration number exceeds an evolution algebra, if so, stopping iteration, and outputting an optimal fault diagnosis accuracy acc, an optimal penalty factor C and an optimal weight factor gamma coefficient; otherwise, executing D8;
d8: updating the iteration step length and the iteration direction of the core parameters under the current iteration according to the fitness, further updating the penalty factor C and the weight factor gamma coefficient, and setting the current support vector machine diagnosis model according to the penalty factor C and the weight factor gamma coefficient;
d9: selecting, crossing and mutating the population, updating the population, and adding 1 to the iteration times; turning to D3;
The method for optimizing the support vector machine diagnosis model by the model optimizing unit by adopting a particle swarm algorithm comprises the following steps:
K1: the fault characteristic frequency used for training the model is divided into a plurality of groups of training sets and test sets after being normalized; each training set and test set is used as a verification set; initializing particles, particle positions and particle speeds;
K2: initializing optimal fault diagnosis accuracy, optimal penalty factors and optimal weight factors, and setting a support vector machine diagnosis model according to the optimal fault diagnosis accuracy, the optimal penalty factors and the optimal weight factors;
k3: inputting each verification set into a current support vector machine diagnosis model, and calculating the fault accuracy under the current iteration;
and K4: judging whether the fault accuracy under the current iteration is greater than the optimal fault diagnosis accuracy, if so, switching to K5; otherwise, turning to K6;
And K5: updating the optimal fault diagnosis accuracy acc, the optimal penalty factor C and the optimal weight factor gamma coefficient;
k6: calculating the adaptability of the particle swarm;
K7: judging whether the current iteration number exceeds an evolution algebra, if so, stopping iteration, and outputting an optimal fault diagnosis accuracy acc, an optimal penalty factor C and an optimal weight factor gamma coefficient; otherwise, executing K8;
And K8: updating the iteration step length and the iteration direction of the core parameters under the current iteration according to the fitness, further updating the penalty factor C and the weight factor gamma coefficient, and setting the current support vector machine diagnosis model according to the penalty factor C and the weight factor gamma coefficient;
k9: searching individual and group extremum, updating the position and speed of particles, and adding 1 to the iteration times; go to K3.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210468372.1A CN114970681B (en) | 2022-04-29 | 2022-04-29 | Induction motor fault diagnosis method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210468372.1A CN114970681B (en) | 2022-04-29 | 2022-04-29 | Induction motor fault diagnosis method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114970681A CN114970681A (en) | 2022-08-30 |
| CN114970681B true CN114970681B (en) | 2024-08-20 |
Family
ID=82979184
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210468372.1A Active CN114970681B (en) | 2022-04-29 | 2022-04-29 | Induction motor fault diagnosis method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114970681B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101916359A (en) * | 2005-01-27 | 2010-12-15 | 剑桥研究和仪器设备股份有限公司 | Methods and apparatus for classifying different parts of a sample into respective classes |
| CN106650815A (en) * | 2016-12-28 | 2017-05-10 | 北京航天测控技术有限公司 | Dynamic particle swarm optimization-based multi-core support vector machine pod fault diagnosis method and apparatus |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB201404226D0 (en) * | 2014-03-11 | 2014-04-23 | Rolls Royce Plc | Fault detection in induction machines |
| CN107132450B (en) * | 2017-04-27 | 2019-06-11 | 上海电力学院 | A kind of sea double feedback electric engine stator winding inter-turn short circuit initial failure discrimination method |
-
2022
- 2022-04-29 CN CN202210468372.1A patent/CN114970681B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101916359A (en) * | 2005-01-27 | 2010-12-15 | 剑桥研究和仪器设备股份有限公司 | Methods and apparatus for classifying different parts of a sample into respective classes |
| CN106650815A (en) * | 2016-12-28 | 2017-05-10 | 北京航天测控技术有限公司 | Dynamic particle swarm optimization-based multi-core support vector machine pod fault diagnosis method and apparatus |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114970681A (en) | 2022-08-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110361176B (en) | Intelligent fault diagnosis method based on multitask feature sharing neural network | |
| CN110703057B (en) | Power equipment partial discharge diagnosis method based on data enhancement and neural network | |
| CN110376522B (en) | Motor fault diagnosis method of data fusion deep learning network | |
| CN109376801B (en) | Wind driven generator blade icing diagnosis method based on integrated deep neural network | |
| CN108664010A (en) | Generating set fault data prediction technique, device and computer equipment | |
| CN112396109A (en) | Motor bearing fault diagnosis method based on recursion graph and multilayer convolution neural network | |
| CN110082136B (en) | Rotary machine fault diagnosis method based on cloud genetic algorithm optimization support vector machine | |
| WO2024065777A1 (en) | Method, apparatus, electronic device, and storage medium for diagnosing industrial fault | |
| CN111207067A (en) | Air compressor fault diagnosis method based on fuzzy support vector machine | |
| CN112541524B (en) | BP-Adaboost multisource information motor fault diagnosis method based on attention mechanism improvement | |
| Júnior et al. | The design of multiple linear regression models using a genetic algorithm to diagnose initial short-circuit faults in 3-phase induction motors | |
| CN114048787B (en) | Method and system for intelligently diagnosing bearing fault in real time based on Attention CNN model | |
| CN112541510A (en) | Intelligent fault diagnosis method based on multi-channel time series data | |
| CN114638318A (en) | Power grid fault diagnosis method and system based on SE-inclusion network | |
| CN111766513A (en) | Capsule network-based variable-working-condition multi-fault diagnosis method for three-phase induction motor | |
| CN117556360A (en) | Photovoltaic inverter abnormality monitoring method and device | |
| CN114970681B (en) | Induction motor fault diagnosis method and system | |
| CN117992777A (en) | Motor electrical fault diagnosis method based on image fusion | |
| CN117218602A (en) | Structural health monitoring data abnormality diagnosis method and system | |
| CN116861165A (en) | Rotor fault diagnosis method based on variational modal decomposition complex network modeling | |
| CN116106016A (en) | Rolling bearing fault diagnosis method based on joint learning convolutional neural network | |
| CN116224950A (en) | Intelligent fault diagnosis method and system for self-organizing reconstruction of unmanned production line | |
| CN115146822A (en) | Photovoltaic power generation prediction method and device and terminal equipment | |
| CN110874088B (en) | Monitoring method of ship key equipment system based on multi-dimensional vector model | |
| CN116861260A (en) | Blower motor fault diagnosis method based on genetic algorithm and LightGBM |
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 |