CN112084711A - Ship electric propulsion system fault diagnosis method based on ABC-SVM (active-support vector machine) expert system - Google Patents

Ship electric propulsion system fault diagnosis method based on ABC-SVM (active-support vector machine) expert system Download PDF

Info

Publication number
CN112084711A
CN112084711A CN202010927759.XA CN202010927759A CN112084711A CN 112084711 A CN112084711 A CN 112084711A CN 202010927759 A CN202010927759 A CN 202010927759A CN 112084711 A CN112084711 A CN 112084711A
Authority
CN
China
Prior art keywords
support vector
vector machine
food source
electric propulsion
propulsion system
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.)
Pending
Application number
CN202010927759.XA
Other languages
Chinese (zh)
Inventor
丁宇
马生平
贲虹凯
随从标
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Harbin Engineering University
Original Assignee
Harbin Engineering University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Harbin Engineering University filed Critical Harbin Engineering University
Priority to CN202010927759.XA priority Critical patent/CN112084711A/en
Publication of CN112084711A publication Critical patent/CN112084711A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • 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
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/004—Artificial life, i.e. computing arrangements simulating life
    • G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N5/00—Computing arrangements using knowledge-based models
    • G06N5/04—Inference or reasoning models
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2111/00—Details relating to CAD techniques
    • G06F2111/06—Multi-objective optimisation, e.g. Pareto optimisation using simulated annealing [SA], ant colony algorithms or genetic algorithms [GA]

Abstract

The invention aims to provide a ship electric propulsion system fault diagnosis method based on an ABC-SVM (active-back support vector machine) expert system, which is characterized in that the knowledge and experience of an expert are stored in a knowledge base, a support vector machine optimized by using an artificial bee colony is used in an inference engine, the knowledge in the knowledge base is preprocessed and then used for carrying out classification training on the inference engine, and a trained model is stored; the real-time data collected from the ship electric propulsion system is stored in a dynamic database, characteristic values are extracted, the extracted characteristic values are input into a trained inference machine, whether the ship electric propulsion system has faults or not is judged, which equipment has faults is judged, and the reasons for the faults are explained. The invention greatly improves the practicability and the real-time performance of the expert system, overcomes the problem of local optimal solution of the support vector machine to a great extent, and improves the accuracy of fault diagnosis.

Description

Ship electric propulsion system fault diagnosis method based on ABC-SVM (active-support vector machine) expert system
Technical Field
The invention relates to a fault diagnosis method, in particular to a ship fault diagnosis method.
Background
At present, the ship structure is more and more complicated, the tonnage of the ship is also continuously increased, the power required to be provided is also more and more large, and the ship electric propulsion system is taken as a main provider of the ship power and has an important position. In practical application, the ship electric propulsion system is influenced by various factors, aging and the like, and the probability of occurrence of faults of the ship electric propulsion system is increasing day by day. After the current ship electric propulsion system breaks down, adverse effects are brought to normal work and navigation of a ship, and the change relation among all parts of the ship electric propulsion system is very complex, so that the fault diagnosis of the ship electric propulsion system faces huge challenges.
In the fault diagnosis, the selection of a fault diagnosis method is a key technology. The conventional fault diagnosis method uses an expert system method, and by inputting expert knowledge into a knowledge base, causes and positions of faults are sequentially searched from the knowledge base. However, the method has the following disadvantages that only a fixed special problem can be solved, the problem of exceeding the knowledge range cannot be solved, the problem of bottleneck in acquiring the expert knowledge is solved, and the method does not have self-learning capability. However, the electric propulsion system of the ship is complex, each system involves many fault problems, and it is very difficult to quickly and timely judge the type and cause of the fault in the process of operating and acquiring data. Therefore, the conventional fault diagnosis method is no longer applicable. In order to overcome the defects of knowledge acquisition of an expert system based on knowledge and improve the fault diagnosis capability, the fault diagnosis system combining the expert system and a neural network is provided. The neural network can obtain the fault diagnosis result of the ship electric propulsion system with high accuracy for diagnosing the fault problem of the ship electric propulsion system with a large number of samples. However, because the neural network has the defect of overfitting, the failure diagnosis effect is extremely poor for the problem of the failure of the ship electric propulsion system with a small number of samples. Therefore, the expert system based on the neural network has poor stability of the fault diagnosis result of the ship electric propulsion system, and sometimes a fault diagnosis phenomenon with a high error rate occurs.
A Support Vector Machine (SVM) is a general learning method based on a finite sample statistical learning theory, and the SVM is based on a structural risk minimization principle and has good generalization capability; the method can automatically acquire knowledge and break through the bottleneck of acquisition of expert system knowledge; has the self-adapting, self-learning, memory and induction capabilities. Compared with a neural network, the method effectively solves the problems of small sample, high dimension, nonlinearity and the like, overcomes the defect of low convergence speed of the artificial neural network, and improves the generalization capability of the learning method.
The parameters of the support vector machine are optimized by using the artificial bee colony algorithm, so that the problem of local optimal solution of the support vector machine can be solved to a great extent, and the precision of the support vector machine is greatly improved.
In summary, the method of combining the support vector machine for artificial bee colony optimization and the expert system can quickly and accurately judge the position and reason of the occurrence of the fault of the ship electric propulsion system, which is a brand new research and design strategy.
Disclosure of Invention
The invention aims to provide a fault diagnosis method of a ship electric propulsion system based on an ABC-SVM expert system, which is used for classifying and judging faults by applying the ABC-SVM to an inference machine.
The purpose of the invention is realized as follows:
the invention relates to a ship electric propulsion system fault diagnosis method based on an ABC-SVM expert system, which is characterized by comprising the following steps:
(1) storing expert knowledge and experience into a knowledge base through a knowledge acquisition module;
(2) the data management module provides an operation interface for the dynamic database, and stores real-time operation data of each device of the ship electric propulsion system in the dynamic database through the data acquisition module;
(3) preprocessing the knowledge in the knowledge base before the inference engine acquires the knowledge;
(4) selecting a kernel function;
(5) optimizing a punishment parameter C and a kernel function parameter sigma of the support vector machine by using an artificial bee colony algorithm;
(6) training and classifying the preprocessed data by the inference machine according to parameters C and sigma after the artificial bee colony algorithm is optimized, and generating a corresponding support vector machine model;
(7) extracting characteristic values of real-time data of each module of the ship electric propulsion system in the dynamic database;
(8) the inference machine carries out classification learning on real-time data of the ship electric propulsion system according to the trained support vector machine model, and judges the fault type;
(9) feeding back a fault conclusion obtained by a user through an interpreter;
(10) and outputting the diagnosis result.
The present invention may further comprise:
1. the inference engine adopts a support vector machine based on artificial bee colony optimization, and specifically comprises the following steps:
(a) initializing parameters including maximum iteration number M, number Np of food sources, optimization parameter D, search range of model parameters [ C, sigma ] and cycle number G;
(b) np food sources were randomly generated by the following formula:
xij＝xjmin+rand(0,1)(xjmax-xjmin)
in the formula, xijFor the ith bee the searched position, x, corresponding to the jth dimensionjmax、xjminRespectively an upper bound and a lower bound of a j-dimension variable;
fitness function Fit according toiEach food source is evaluated, and the optimal food source is found:
Figure BDA0002669035860000031
in the formula, yiIs an actual value, yi^ is a predicted value;
(c) the bee is adopted to do neighborhood search according to the following formula to generate a new solution,
Vij＝xij+rij(xij-xkj)
where i, k ∈ (1,2, …, Np), j ∈ (1,2) is a randomly chosen subscript, rij∈[-1,1]A random value between xijIs a food source xiComponent of the j-th dimension, xkjFor random selection of food sources xkJ is a component;
adaptive equation of
Figure BDA0002669035860000032
It is found that when the fitness before searching is higher than that of the new food source, the new food source replaces the previous optimal food source, otherwise, the position of the old food source is kept unchanged, and the position of the food source is x for the individual food sourceiRecording the times of non-updating;
(d) the probability P of selecting food source by observing bee is obtained byiThe new food source is represented by formula Vij＝xij+rij(xij-xkj) Searching nearby food sources, calculating the fitness of a new food source, and replacing the optimal food source by the new food source if the fitness value of the new food source is smaller:
Figure BDA0002669035860000033
(e) abandoning food sources which are not optimized after G cycles, changing the corresponding honey bees into detection bees, and obtaining new food sources by the formula xij＝xjmin+rand(0,1)(xjmax-xjmin) Generating;
(f) and (c) if the operation does not reach the iteration times M or the fitness preset precision, returning to the step (c), otherwise, ending the operation and outputting the optimal value of the parameter.
2. The preprocessing of the inference engine is as follows: the method comprises the steps of converting knowledge in a knowledge base into an input format required by a support vector machine, adopting a multi-class support vector machine by an inference machine, adopting a one-to-one classification method, constructing k classes of classification problems into two classes of SVM classifiers (support vector machine), wherein N is k (k-1)/2, a classification model used in the support vector machine is C-SVC, and classifying by adopting the idea of punishing parameters.
3. The kernel function is selected by using a Radial Basis Function (RBF) expressed as K (x, y) { (- σ | x-y { [ gamma ] } { (y) } y { (R {) y } n2)}。
The invention has the advantages that: on the basis of deep research on the fault diagnosis technology of the ship electric propulsion system, the invention introduces the support vector machine technology into the design of an expert system and optimizes the support vector machine by using an artificial bee colony algorithm. The support vector machine is based on the principle of minimizing the structural risk and has good generalization capability; knowledge can be automatically acquired; has the self-adapting, self-learning, memory and induction capabilities. The application of the support vector machine can overcome the problems of bottleneck of expert system knowledge acquisition, difficulty in knowledge maintenance, low reasoning speed and the like, thereby greatly improving the practicability and the real-time performance of the expert system. The punishment parameter C and the kernel function parameter sigma of the support vector machine can be optimized by using the artificial bee colony algorithm, the problem of local optimal solution of the support vector machine can be solved to a great extent, and the accuracy of fault diagnosis is improved.
Drawings
FIG. 1 is a structural web of the present invention;
FIG. 2 is an ABC-SVM flow diagram;
FIG. 3 is a two-dimensional plan view of a support vector machine.
Detailed Description
The invention will now be described in more detail by way of example with reference to the accompanying drawings in which:
with reference to fig. 1-3, the invention relates to a fault diagnosis method for a ship electric propulsion system based on an ABC-SVM expert system, which mainly uses an expert system as a model, and uses a support vector machine optimized by an artificial bee colony in an inference engine of the expert system to diagnose faults occurring in the ship electric propulsion system. The core of the expert system is an inference engine and a knowledge base, the knowledge and experience of experts are stored in the knowledge base, a support vector machine optimized by using an artificial bee colony is used in the inference engine, the knowledge in the knowledge base is preprocessed and then used for carrying out classification training on the inference engine, and a trained model is stored; the real-time data collected from the ship electric propulsion system is stored in a dynamic database, characteristic values are extracted, the extracted characteristic values are input into a trained inference machine, whether the ship electric propulsion system has faults or not is judged, which equipment has faults is judged, and the reasons for the faults are explained. The method specifically comprises the following steps:
(1) the method comprises the steps that the knowledge and experience of experts in the field of ship electric propulsion system faults are stored in a knowledge base through a knowledge acquisition module;
(2) the data management module provides an operation interface for the dynamic database, and stores real-time operation data of various devices of the ship electric propulsion system such as a diesel generator set, a distribution board, a frequency converter, a transformer, a propulsion motor and the like in the dynamic database through the data acquisition module;
(3) the inference engine preprocesses the knowledge in the knowledge base before acquiring the knowledge, and the preprocessing step is to convert the knowledge in the knowledge base into an input format required by the support vector machine. The inference engine adopts a multi-class support vector machine, adopts a one-to-one classification method, and constructs k classes of classification problems into two classes of SVM classifiers (N is k (k-1)/2). The classification model used in the support vector machine is C-SVC, and classification is carried out by adopting the idea of punishment parameters. Wherein C is a penalty parameter representing the penalty degree for errors, and the larger C is, the more heavy C is
(4) Selecting a kernel function: the kernel function is the most applied radial basis function (RBF function), and its expression is K (x, y) ═ exp { (- σ | x-y |2) }. Using the RBF function as the kernel function of the SVM, two parameters have to be considered: penalty coefficient C and kernel parameter σ. Parameter C makes a trade-off between misclassified samples and interface simplicity. Low C values smooth the interfaces, while high C values ensure that all samples are correctly classified by increasing the model freedom to select more support vectors. The parameter σ defines the magnitude of the impact of a single training sample, with smaller values having greater impact and larger values having less impact. The parameter σ can be seen as the inverse of the radius of influence of the sample selected by the model as the support vector.
(5) As shown in fig. 2, the method for optimizing the support vector machine by the artificial bee colony includes the following steps:
step 1: and initializing parameters. The maximum iteration number M is 50, the number Np of food sources is 30, the optimization parameter D is 2, the search range [0.1, 1000] of the model parameter [ C, sigma ], and the cycle number G is 30.
Step 2: randomly generating Np food sources by the formula (1) according to the fitness function Fit of the formula (2)iAnd evaluating each food source to find the optimal food source.
xij＝xjmin+rand(0,1)(xjmax-xjmin) (1)
In the formula, xijFor the ith bee the searched position, x, corresponding to the jth dimensionjmax、xjminRespectively, an upper bound and a lower bound of the variable of the j-th dimension.
Figure BDA0002669035860000051
In the formula, yiIs an actual value, yiAnd ^ is a predicted value.
Step 3: : the bee is adopted to do neighborhood search according to the following formula to generate a new solution,
Vij＝xij+rij(xij-xkj) (3)
where i, k ∈ (1,2, …, Np), j ∈ (1,2) is a randomly chosen subscript. r isij∈[-1,1]A random value in between. x is the number ofijIs a food source xiComponent of the j-th dimension, xkjThe j th component of the food source xk is selected randomly.
The fitness is calculated by the formula (2). When the fitness before searching is higher than that of a new food source, the new food source replaces the previous optimal food source, otherwise, the position of the old food source is kept unchanged, and the number of times that the individual xi of the food source is not updated is recorded.
Step 4: the probability Pi of selecting a food source by the observing bees is calculated by equation (4), a new food source can be searched near the food source by equation (3), and the fitness of the new food source is calculated, and if the fitness value thereof becomes small, the optimal food source is replaced by it.
Figure BDA0002669035860000061
Step 5: abandoning food sources which are not optimized after G cycles, and mutating honey bees corresponding to the food sources into scout bees, wherein a new food source is generated by the formula (1).
Step 6: if the operation does not reach the iteration times M or the preset accuracy of the fitness is not reached, returning to Step3, otherwise, ending the operation and outputting the optimal value of the parameter.
(6) Training and classifying the preprocessed data by the inference machine according to parameters C and sigma after the artificial bee colony algorithm is optimized, and generating a corresponding support vector machine model;
(7) extracting characteristic values of real-time data of each module of the ship electric propulsion system in the dynamic database;
(8) the inference machine carries out classification learning on real-time ship electric propulsion system data according to the trained support vector machine model, and judges the fault type;
(9) the interpreter tells the user what the reason the system concludes the fault is;
(10) and outputting the diagnosis result.
As shown in fig. 1, the core of the fault diagnosis structure diagram based on a support vector machine and an expert system is a support vector machine in an inference engine, the support vector machine is mainly used for classifying problems, and a ship electric propulsion system has many subsystems, so that a multi-class support vector machine is used, and a one-to-one method with good performance is used for the multi-class support vector machine: for the class k classification problem, all possible two classes of SVM classifiers are constructed, and N ═ k (k-1)/2 two classes of classifiers can be constructed in total. In constructing the two-class SVM classifier between the ith and jth classes, points within the ith and jth classes are labeled +1 and-1, respectively. In the test, the test data is respectively substituted into the two types of classifiers, namely, N ═ k (k-1)/2, and the test data is tested, the scores of the categories are accumulated, the score is selected, and the category corresponding to the highest one is the category to which the test data belongs.
The support vector machine is developed from the optimal classification plane in the linear separable case, and the basic idea is illustrated in the case of the two-dimensional plane of fig. 3.
In fig. 3, the dots and the squares represent two types of samples, the thick solid line in the middle is a classification line, the two broken lines in the vicinity are straight lines which pass through the sample closest to the classification line in each type and are parallel to the classification line, and the interval between the two broken lines is the classification interval. The optimal classification line is required to separate two classes correctly, i.e. the training error rate is 0, and to maximize the classification interval. For classification line < w.xiThe normalization process is performed so that for a linearly separable sample set S, the inequality is satisfied:
yi(＜w·xi＞+b)≥1,i＝1,2,… (5)
wherein w ∈ Rn is the normal of the hyperplane, b ∈ R is the intercept,<·>representing an inner vector product operation, xi∈Rn，yi∈[+1,-1]，i＝1,2,…N，xiIs the i-th characteristic quantity, yiIs output class, positive case when it equals + 1; the negative example is given when the value is-1.
At this time, the classification interval is equal to 2/| w |, making the interval maximally equivalent to making | w | | count non-woven cells2And minimum. Training samples can be correctly divided, and | | w | | | non-woven cells are made2The smallest classification surface is the optimal classification surface, and the training sample points located on the two dotted lines are called support vectors.
Maximizing the classification interval is actually the control of the generalization ability, which is one of the core ideas of SVM. The statistical learning theory indicates that in an N-dimensional space, if samples are distributed in a hypersphere with a radius of R, an indication function set f (x, w, b) formed by a regular hyperplane satisfying the condition | | | w | < a is sgn { < w · x {iThe VC dimension of > + b } satisfies the following bound.
h≤min([R2A2],N)+1 (6)
In the formula, min represents the shortest distance from a point to a hyperplane, and the radius of the hypersphere of R.
Thus, making | | w | | non-luminous2The minimum is to minimize the bounds on the VC dimension, thereby achieving the choice of functional complexity in the SRM criterion.
The problem then translates into a constrained nonlinear programming problem:
Figure BDA0002669035860000081
s.t.yi(＜w·xi＞+b)≥1,i＝1,2,…,l (8)
defining the Lagrange function:
Figure BDA0002669035860000082
wherein alpha is more than or equal to 0 and is constraint yi(w·xi+ b) Lagrange multiplier of 1 or more.
According to the KKT theorem
Figure BDA0002669035860000083
Figure BDA0002669035860000084
αi((＜w·xi＞+b)-1)＝0 (12)
By substituting equations (10) and (11) into equations (7) and (8), the optimal hyperplane problem is transformed into a dual quadratic programming problem.
Figure BDA0002669035860000085
In the formula, max represents the maximum distance of a point from the hyperplane.
Figure BDA0002669035860000086
By solving this optimal problem, the optimal hyperplane can be determined. And usually only a small fraction of ai is not 0, these non-zero solution corresponding samples are support vectors. The optimal classification function is obtained as:
Figure BDA0002669035860000087
in the formula, phi (x) is a nonlinear change, and an n-dimensional input and 1-dimensional output sample vector (xi, y) is mapped to a high-dimensional space K (·) from an original space as a kernel function.

Claims (4)

1. A ship electric propulsion system fault diagnosis method based on an ABC-SVM expert system is characterized by comprising the following steps:
(1) storing expert knowledge and experience into a knowledge base through a knowledge acquisition module;
(2) the data management module provides an operation interface for the dynamic database, and stores real-time operation data of each device of the ship electric propulsion system in the dynamic database through the data acquisition module;
(3) preprocessing the knowledge in the knowledge base before the inference engine acquires the knowledge;
(4) selecting a kernel function;
(5) optimizing a punishment parameter C and a kernel function parameter sigma of the support vector machine by using an artificial bee colony algorithm;
(6) training and classifying the preprocessed data by the inference machine according to parameters C and sigma after the artificial bee colony algorithm is optimized, and generating a corresponding support vector machine model;
(7) extracting characteristic values of real-time data of each module of the ship electric propulsion system in the dynamic database;
(8) the inference machine carries out classification learning on real-time data of the ship electric propulsion system according to the trained support vector machine model, and judges the fault type;
(9) feeding back a fault conclusion obtained by a user through an interpreter;
(10) and outputting the diagnosis result.
2. The method for diagnosing the fault of the ship electric propulsion system based on the expert system of the ABC-SVM as claimed in claim 1, wherein the method comprises the following steps: the inference engine adopts a support vector machine based on artificial bee colony optimization, and specifically comprises the following steps:
(a) initializing parameters including maximum iteration number M, number Np of food sources, optimization parameter D, search range of model parameters [ C, sigma ] and cycle number G;
(b) np food sources were randomly generated by the following formula:
xij＝xjmin+rand(0,1)(xjmax-xjmin)
in the formula, xijFor the ith bee the searched position, x, corresponding to the jth dimensionjmax、xjminRespectively an upper bound and a lower bound of a j-dimension variable;
fitness function Fit according toiEach food source is evaluated, and the optimal food source is found:
Figure FDA0002669035850000021
in the formula, yiIs an actual value, yi^ is a predicted value;
(c) the bee is adopted to do neighborhood search according to the following formula to generate a new solution,
Vij＝xij+rij(xij-xkj)
where i, k ∈ (1,2, …, Np), j ∈ (1,2) is a randomly chosen subscript, rij∈[-1,1]A random value between xijIs a food source xiComponent of the j-th dimension, xkjFor random selection of food sources xkJ is a component;
adaptive equation of
Figure FDA0002669035850000022
It is found that when the fitness before searching is higher than that of the new food source, the new food source replaces the previous optimal food source, otherwise, the position of the old food source is kept unchanged, and the position of the food source is x for the individual food sourceiRecording the times of non-updating;
(d) the probability P of selecting food source by observing bee is obtained byiThe new food source is represented by formula Vij＝xij+rij(xij-xkj) Searching nearby food sources, calculating the fitness of a new food source, and replacing the optimal food source by the new food source if the fitness value of the new food source is smaller:
Figure FDA0002669035850000023
(e) abandoning food sources which are not optimized after G cycles, changing the corresponding honey bees into detection bees, and obtaining new food sources by the formula xij＝xjmin+rand(0,1)(xjmax-xjmin) Generating;
(f) and (c) if the operation does not reach the iteration times M or the fitness preset precision, returning to the step (c), otherwise, ending the operation and outputting the optimal value of the parameter.
3. The method for diagnosing the fault of the ship electric propulsion system based on the expert system of the ABC-SVM as claimed in claim 1, wherein the method comprises the following steps: the preprocessing of the inference engine is as follows: the method comprises the steps of converting knowledge in a knowledge base into an input format required by a support vector machine, adopting a multi-class support vector machine by an inference machine, adopting a one-to-one classification method, constructing k classes of classification problems into two classes of SVM classifiers (support vector machine), wherein N is k (k-1)/2, a classification model used in the support vector machine is C-SVC, and classifying by adopting the idea of punishing parameters.
4. The method for diagnosing the fault of the ship electric propulsion system based on the expert system of the ABC-SVM as claimed in claim 1, wherein the method comprises the following steps: the kernel function is selected by using a Radial Basis Function (RBF) expressed as K (x, y) { (- σ | x-y { [ gamma ] } { (y) } y { (R {) y } n2)}。
CN202010927759.XA 2020-09-07 2020-09-07 Ship electric propulsion system fault diagnosis method based on ABC-SVM (active-support vector machine) expert system Pending CN112084711A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010927759.XA CN112084711A (en) 2020-09-07 2020-09-07 Ship electric propulsion system fault diagnosis method based on ABC-SVM (active-support vector machine) expert system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010927759.XA CN112084711A (en) 2020-09-07 2020-09-07 Ship electric propulsion system fault diagnosis method based on ABC-SVM (active-support vector machine) expert system

Publications (1)

Publication Number Publication Date
CN112084711A true CN112084711A (en) 2020-12-15

Family

ID=73732037

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010927759.XA Pending CN112084711A (en) 2020-09-07 2020-09-07 Ship electric propulsion system fault diagnosis method based on ABC-SVM (active-support vector machine) expert system

Country Status (1)

Country Link
CN (1) CN112084711A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113311280A (en) * 2021-07-30 2021-08-27 中国人民解放军海军工程大学 Health grading monitoring device for complex electromechanical system

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104036112A (en) * 2014-04-24 2014-09-10 河海大学 Fault diagnosis method based on support vector machine and expert system
CN105930864A (en) * 2016-04-15 2016-09-07 杭州电子科技大学 EEG (electroencephalogram) signal feature classification method based on ABC-SVM

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104036112A (en) * 2014-04-24 2014-09-10 河海大学 Fault diagnosis method based on support vector machine and expert system
CN105930864A (en) * 2016-04-15 2016-09-07 杭州电子科技大学 EEG (electroencephalogram) signal feature classification method based on ABC-SVM

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
季伟 等: "基于人工蜂群算法优化支持向量机的变压器故障诊断", 《科技创新与应用》 *
邓建球等: "ABC算法优化SVR的磨损故障预测模型", 《兵工自动化》 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113311280A (en) * 2021-07-30 2021-08-27 中国人民解放军海军工程大学 Health grading monitoring device for complex electromechanical system
CN113311280B (en) * 2021-07-30 2021-12-28 中国人民解放军海军工程大学 Health grading monitoring device for complex electromechanical system

Similar Documents

Publication Publication Date Title
CN112286751B (en) Intelligent diagnosis system and method for high-end equipment fault based on edge cloud cooperation
Liu et al. Supervised learning
CN101464964B (en) Pattern recognition method capable of holding vectorial machine for equipment fault diagnosis
CN108921285B (en) Bidirectional gate control cyclic neural network-based classification method for power quality disturbance
Yang et al. Combined wireless network intrusion detection model based on deep learning
Li et al. Rolling bearing fault diagnosis based on horizontal visibility graph and graph neural networks
CN109214460A (en) Method for diagnosing fault of power transformer based on Relative Transformation Yu nuclear entropy constituent analysis
CN108803554A (en) A kind of intelligent diagnosing method of generator signal
Yuan et al. Review of resampling techniques for the treatment of imbalanced industrial data classification in equipment condition monitoring
CN116842194A (en) Electric power semantic knowledge graph system and method
Igoe et al. How useful are gradients for ood detection really?
CN112084711A (en) Ship electric propulsion system fault diagnosis method based on ABC-SVM (active-support vector machine) expert system
Dhibi et al. Interval-valued reduced ensemble learning based fault detection and diagnosis techniques for uncertain grid-connected PV systems
Cao et al. No-delay multimodal process monitoring using Kullback-Leibler divergence-based statistics in probabilistic mixture models
Dhurandhar et al. Enhancing simple models by exploiting what they already know
CN109409434A (en) The method of liver diseases data classification Rule Extraction based on random forest
Djelloul et al. Fault isolation in manufacturing systems based on learning algorithm and fuzzy rule selection
CN116186633A (en) Power consumption abnormality diagnosis method and system based on small sample learning
US11741146B2 (en) Embedding multi-modal time series and text data
Nojima et al. A meta-fuzzy classifier for specifying appropriate fuzzy partitions by genetic fuzzy rule selection with data complexity measures
Zhang et al. Text classification with support vector machine and back propagation neural network
Sousa et al. Efficient selection of data samples for fault classification by the clustering of the SOM
CN116538127B (en) Axial flow fan and control system thereof
Dhiman et al. Fault Diagnosis of a Transformer using Fuzzy Model and PSO optimized SVM
Wenjiao et al. Data-Driven Fault Parameter Extraction and Fault Diagnosis Model for High-Voltage Circuit Breakers

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
RJ01 Rejection of invention patent application after publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20201215