CN112504682A - Chassis engine fault diagnosis method and system based on particle swarm optimization algorithm - Google Patents

Chassis engine fault diagnosis method and system based on particle swarm optimization algorithm Download PDF

Info

Publication number
CN112504682A
CN112504682A CN202011520723.6A CN202011520723A CN112504682A CN 112504682 A CN112504682 A CN 112504682A CN 202011520723 A CN202011520723 A CN 202011520723A CN 112504682 A CN112504682 A CN 112504682A
Authority
CN
China
Prior art keywords
data
principal component
kernel
neural network
particle swarm
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
CN202011520723.6A
Other languages
Chinese (zh)
Inventor
陈悦峰
孙希明
于深
邹永显
王嫒娜
麻雄
陈文英
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.)
63963 TROOPS PLA
Dalian University of Technology
Original Assignee
63963 TROOPS PLA
Dalian University of Technology
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 63963 TROOPS PLA, Dalian University of Technology filed Critical 63963 TROOPS PLA
Priority to CN202011520723.6A priority Critical patent/CN112504682A/en
Publication of CN112504682A publication Critical patent/CN112504682A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
    • G01M15/00—Testing of engines
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/004—Artificial life, i.e. computing arrangements simulating life
    • G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/084—Backpropagation, e.g. using gradient descent

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Molecular Biology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Testing And Monitoring For Control Systems (AREA)

Abstract

The invention discloses a chassis engine fault diagnosis method and system based on a particle swarm optimization algorithm, wherein the diagnosis method comprises the following steps: determining parameters of a kernel function of a kernel principal component analysis algorithm by adopting a particle swarm optimization algorithm of dynamic inertia factors; carrying out optimization processing on the initial weight and the threshold of the BP neural network model by utilizing a particle swarm optimization algorithm, and training the initialized BP neural network model by utilizing a training sample set to obtain a trained BP neural network model; determining whether the monitoring data is fault data or not by adopting a kernel principal component analysis algorithm with determined parameters; and when the monitoring data is fault data, inputting the monitoring data into the trained BP neural network model, and determining the fault type corresponding to the monitoring data. The invention combines the particle swarm optimization algorithm, the kernel principal component analysis algorithm and the BP neural network model, and realizes the efficient and accurate diagnosis of the chassis engine fault by utilizing the abundant fault information contained in the lubricating oil of the chassis engine.

Description

Chassis engine fault diagnosis method and system based on particle swarm optimization algorithm
Technical Field
The invention relates to the technical field of fault diagnosis, in particular to a chassis engine fault diagnosis method and system based on a particle swarm optimization algorithm.
Background
The chassis engine is the main power structure of the vehicle and is the core part of the vehicle, and once a fault occurs, the running condition of the whole vehicle is affected. With the development of the technology, the structure of the engine is more and more precise and more complex, the traditional fault diagnosis method inevitably consumes a large amount of manpower and material resources, and the diagnosis accuracy is not ideal. In recent years, the development of artificial intelligence provides a new scheme for fault diagnosis, and fault diagnosis methods based on data driving are more and more applied to practice and achieve good effects.
The existing intelligent fault diagnosis methods are many, and a neural network, a support vector machine, a rough set theory, Kalman filtering, a Bayesian network and the like are applied in many cases. The most mature neural network model is the BP neural network model, but the BP neural network model also has some defects, such as easy falling into local minimum, slow convergence speed and the like, and influences the precision of fault diagnosis.
Disclosure of Invention
The invention aims to provide a chassis engine fault diagnosis method and system based on a particle swarm optimization algorithm so as to efficiently and accurately diagnose the chassis engine fault.
In order to achieve the purpose, the invention provides the following scheme:
a chassis engine fault diagnosis method based on a particle swarm optimization algorithm comprises the following steps:
acquiring historical lubricating oil data of a chassis engine, extracting characteristic parameters, and establishing a training sample set;
determining parameters of a kernel function of a kernel principal component analysis algorithm for chassis engine fault diagnosis by using the training sample set and taking the accuracy of fault monitoring as a fitness function and adopting a particle swarm optimization algorithm of dynamic inertia factors to obtain the kernel principal component analysis algorithm with well-determined parameters;
optimizing the initial weight and the threshold of the BP neural network model by utilizing a particle swarm optimization algorithm to obtain an initialized BP neural network model, and training the initialized BP neural network model by utilizing the training sample set to obtain a trained BP neural network model;
obtaining lubricating oil data of a monitoring state of a chassis engine, and extracting characteristic parameters to obtain monitoring data;
determining whether the monitoring data is fault data or not by adopting a kernel principal component analysis algorithm with determined parameters;
and when the monitoring data are fault data, inputting the monitoring data into a trained BP neural network model, and determining the fault type corresponding to the monitoring data.
Optionally, the speed updating formula of the particle swarm optimization algorithm of the dynamic inertia factor is as follows:
Figure BDA0002848813140000021
the position updating formula of the particle swarm optimization algorithm of the dynamic inertia factors is as follows:
Figure BDA0002848813140000022
the dynamic inertia factor updating formula of the particle swarm optimization algorithm of the dynamic inertia factors is as follows:
w＝wmax-t*(wmax-wmin)/Tmax；
wherein the content of the first and second substances,
Figure BDA0002848813140000023
and
Figure BDA0002848813140000024
respectively representing the velocity vector and the position vector of the ith particle before the updating of the t iteration,
Figure BDA0002848813140000025
and
Figure BDA0002848813140000026
respectively representing the velocity vector and the position vector of the ith particle after the ith iteration update, w is an inertia factor of the particle swarm optimization algorithm, c1And c2Respectively a first learning factor and a second learning factor, lambda1And λ2A first random number and a second random number between 0 and 1, respectively; w is amaxAnd wminRespectively representing the maximum inertia factor and the minimum inertia factor, T, set in the iterative processmaxThe maximum number of iterations is indicated,
Figure BDA0002848813140000027
representing the individual optimal positions during the t-th iteration,
Figure BDA0002848813140000028
representing the global optimum position during the t-th iteration.
Optionally, the lubricating oil data of the monitoring state of the chassis engine is acquired, and the characteristic parameters are extracted to obtain the monitoring data, which specifically includes:
standardizing the lubricating oil data in a monitoring state to obtain the standardized lubricating oil data;
calculating a kernel matrix of the lubricating oil data after the standardization treatment;
using formulas
Figure BDA0002848813140000029
Performing centralization processing on the kernel matrix to obtain a core matrix subjected to centralization processing; wherein K represents the kernel matrix prior to centralization,
Figure BDA00028488131400000210
representing the kernel matrix after centralization, 1NRepresenting an N-dimensional square matrix, 1NEach element in the lubricating oil is 1/N, and N represents the number of lubricating oil data;
and solving the eigenvalue and the eigenvector of the core matrix after the centralization treatment, and sequencing the eigenvalue and the eigenvector from large to small to obtain the sequenced eigenvalue and eigenvector as monitoring data.
Optionally, determining whether the monitoring data is fault data by using a kernel principal component analysis algorithm with well-determined parameters specifically includes:
calculating nonlinear principal components of the monitoring data by adopting a kernel principal component analysis algorithm with well-determined parameters;
using formula T based on nonlinear principal components of the monitored data2＝(t1，t2，...，tq)Λ-1(t1，t2，...，tq)TCalculating T of the monitoring data2Statistics; wherein, t1，t2，...，tqIs a nonlinear principal component, Lambda, of the kernel principal component analysis extracted monitoring data-1Representing a diagonal inverse matrix formed by the characteristic values corresponding to each nonlinear principal component, wherein q is the number of the nonlinear principal component reserved in the kernel principal component analysis;
using a formula based on the nonlinear principal component of the monitored data
Figure BDA0002848813140000031
Calculating SPE statistic of the monitoring data; wherein, tSRepresenting the nonlinear principal component corresponding to the s-th non-zero eigenvalue, k representing each non-zero eigenvalueThe linear principal component corresponds to the number of non-zero eigenvalues, t, in the eigenvaluesjRepresenting the jth nonlinear principal component of the kernel principal component analysis extraction monitoring data;
judging T of the monitoring data2Whether the statistic and the SPE statistic exceed T respectively2Obtaining a judgment result by the control limit and the SPE control limit;
if the judgment result shows that the data is the fault data, determining the monitoring data as the fault data;
and if the judgment result shows that the data is not normal, determining the monitoring data as normal data.
Optionally, the determining, by using the kernel principal component analysis algorithm with the determined parameters, whether the monitored data is fault data, before further including:
using formulas
Figure BDA0002848813140000032
Determination of T2A control limit; wherein N represents the number of training samples in normal operation state, q represents the number of nonlinear principal component components retained in kernel principal component analysis in each training sample in normal operation state, α represents the level of examination, and Fq，N-q，αRepresenting the distribution critical value when the degree of freedom is q, the detection level is alpha and the condition is N-q in the F distribution;
using formulas
Figure BDA0002848813140000041
Determining SPE control limit; wherein, h represents an intermediate variable,
Figure BDA0002848813140000042
θ1、θ2and theta3Respectively representing a first feature value accumulated value, a second feature accumulated value and a third feature accumulated value, CαRepresenting the critical value of a normal distribution at the test level α.
Optionally, the optimizing processing is performed on the initial weight and the threshold of the BP neural network model by using the particle swarm optimization algorithm to obtain the initialized BP neural network model, and the training sample set is used to train the initialized BP neural network model to obtain the trained BP neural network model, which specifically includes:
determining the optimal initial weight and threshold of the BP neural network model by using the training sample set and the mean square error output by the BP neural network model as a fitness function and adopting a particle swarm optimization algorithm to obtain the initialized BP neural network model;
and training the initialized BP neural network model by using the training sample set to obtain the trained BP neural network model.
A chassis engine fault diagnosis system based on particle swarm optimization algorithm, the diagnosis system comprising:
the training sample set establishing module is used for acquiring historical lubricating oil data of the chassis engine, extracting characteristic parameters and establishing a training sample set;
the kernel principal component analysis parameter determination module is used for determining parameters of a kernel function of a kernel principal component analysis algorithm for chassis engine fault diagnosis by using the training sample set, taking the accuracy of fault monitoring as a fitness function and adopting a particle swarm optimization algorithm of dynamic inertia factors to obtain the kernel principal component analysis algorithm with well-determined parameters;
the BP neural network model training module is used for carrying out optimization processing on the initial weight and the threshold of the BP neural network model by utilizing a particle swarm optimization algorithm to obtain an initialized BP neural network model, and training the initialized BP neural network model by utilizing the training sample set to obtain a trained BP neural network model;
the characteristic parameter extraction module is used for acquiring lubricating oil data of a chassis engine in a monitoring state and extracting characteristic parameters to obtain monitoring data;
the fault diagnosis module is used for determining whether the monitoring data is fault data or not by adopting a kernel principal component analysis algorithm with determined parameters;
and the fault type determining module is used for inputting the monitoring data into a trained BP neural network model when the monitoring data is fault data, and determining the fault type corresponding to the monitoring data.
Optionally, the speed updating formula of the particle swarm optimization algorithm of the dynamic inertia factor is as follows:
Figure BDA0002848813140000051
the position updating formula of the particle swarm optimization algorithm of the dynamic inertia factors is as follows:
Figure BDA0002848813140000052
the dynamic inertia factor updating formula of the particle swarm optimization algorithm of the dynamic inertia factors is as follows:
w＝wmax-t*(wmax-wmin)/Tmax；
wherein the content of the first and second substances,
Figure BDA0002848813140000053
and
Figure BDA0002848813140000054
respectively representing the velocity vector and the position vector of the ith particle before the updating of the t iteration,
Figure BDA0002848813140000055
and
Figure BDA0002848813140000056
respectively representing the velocity vector and the position vector of the ith particle after the ith iteration update, w is an inertia factor of the particle swarm optimization algorithm, c1And c2Respectively a first learning factor and a second learning factor, lambda1And λ2A first random number and a second random number between 0 and 1, respectively; w is amaxAnd wminRespectively representing the maximum inertia factor and the minimum inertia factor, T, set in the iterative processmaxThe maximum number of iterations is indicated,
Figure BDA0002848813140000057
representing the individual optimal positions during the t-th iteration,
Figure BDA0002848813140000058
representing the global optimum position during the t-th iteration.
Optionally, the feature parameter extraction module specifically includes:
the standardized processing submodule is used for carrying out standardized processing on the lubricating oil data in the monitoring state to obtain the standardized lubricating oil data;
the kernel matrix calculation submodule is used for calculating a kernel matrix of the lubricating oil data after the standardization processing;
a centralized processing submodule for utilizing the formula
Figure BDA0002848813140000059
Performing centralization processing on the kernel matrix to obtain a core matrix subjected to centralization processing; wherein K represents the kernel matrix prior to centralization,
Figure BDA00028488131400000510
representing the kernel matrix after centralization, 1NRepresenting an N-dimensional square matrix, 1NEach element in the lubricating oil is 1/N, and N represents the number of lubricating oil data;
and the characteristic extraction submodule is used for solving the characteristic value and the characteristic vector of the core matrix after the centralization processing, sequencing the characteristic value and the characteristic vector from large to small, and obtaining the sequenced characteristic value and characteristic vector as monitoring data.
Optionally, the fault diagnosis module specifically includes:
the principal component analysis submodule is used for calculating the nonlinear principal component of the monitoring data by adopting a kernel principal component analysis algorithm with determined parameters;
T2a statistic calculation submodule for using the formula T according to the nonlinear principal component of the monitored data2＝(t1，t2，...，tq)Λ-1(t1，t2，...，tq)TCalculating T of the monitoring data2Statistics; wherein, t1，t2，...，tqIs a nonlinear principal component, Lambda, of the kernel principal component analysis extracted monitoring data-1Representing a diagonal inverse matrix formed by the characteristic values corresponding to each nonlinear principal component, wherein q is the number of the nonlinear principal component reserved in the kernel principal component analysis;
SPE statistic calculation submodule for calculating nonlinear principal component of monitored data by using formula
Figure BDA0002848813140000061
Calculating SPE statistic of the monitoring data; wherein, tsRepresenting the non-linear pivot component corresponding to the s-th non-zero eigenvalue, k representing the number of non-zero eigenvalues in the eigenvalue corresponding to each non-linear pivot component, tjRepresenting the jth nonlinear principal component of the kernel principal component analysis extraction monitoring data;
a judgment result submodule for judging T of the monitoring data2Whether the statistic and the SPE statistic exceed T respectively2Obtaining a judgment result by the control limit and the SPE control limit;
the fault diagnosis submodule is used for determining the monitoring data as fault data if the judgment result shows that the monitoring data is positive; and if the judgment result shows that the data is not normal, determining the monitoring data as normal data.
Compared with the prior art, the invention has the beneficial effects that:
the invention provides a chassis engine fault diagnosis method and system based on a particle swarm optimization algorithm, wherein the diagnosis method comprises the following steps: acquiring historical lubricating oil data of a chassis engine, extracting characteristic parameters, and establishing a training sample set; determining parameters of a kernel function of a kernel principal component analysis algorithm for chassis engine fault diagnosis by using the training sample set and taking the accuracy of fault monitoring as a fitness function and adopting a particle swarm optimization algorithm of dynamic inertia factors to obtain the kernel principal component analysis algorithm with well-determined parameters; optimizing the initial weight and the threshold of the BP neural network model by utilizing a particle swarm optimization algorithm to obtain an initialized BP neural network model, and training the initialized BP neural network model by utilizing the training sample set to obtain a trained BP neural network model; obtaining lubricating oil data of a monitoring state of a chassis engine, and extracting characteristic parameters to obtain monitoring data; determining whether the monitoring data is fault data or not by adopting a kernel principal component analysis algorithm with determined parameters; and when the monitoring data are fault data, inputting the monitoring data into a trained BP neural network model, and determining the fault type corresponding to the monitoring data. The invention combines the particle swarm optimization algorithm, the kernel principal component analysis algorithm and the BP neural network model, and realizes the efficient and accurate diagnosis of the chassis engine fault by utilizing the abundant fault information contained in the lubricating oil of the chassis engine.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without inventive exercise.
FIG. 1 is a flow chart of a chassis engine fault diagnosis method based on a particle swarm optimization algorithm provided by the invention;
FIG. 2 is a flow chart of determining parameters of a kernel function of a kernel principal component analysis algorithm provided by the present invention;
FIG. 3 is a block diagram of a BP neural network model provided by the present invention;
fig. 4 is a flowchart of a training process of the BP neural network model provided in the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The invention aims to provide a chassis engine fault diagnosis method and system based on a particle swarm optimization algorithm so as to efficiently and accurately diagnose the chassis engine fault.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.
As shown in FIG. 1, the invention provides a chassis engine fault diagnosis method based on a particle swarm optimization algorithm, which comprises the following steps:
step 101, obtaining historical lubricating oil data of a chassis engine, extracting characteristic parameters, and establishing a training sample set.
The lubricating oil of the chassis engine contains abundant fault information, and the diagnosis of the common abrasion fault of the chassis engine can be realized by monitoring abrasive particles and other physical and chemical property parameters in the lubricating oil.
After the invention adopts the sensor to obtain the lubricating oil data information of the chassis engine, the original characteristic parameters of the data information are extracted:
(1) the method comprises the steps of carrying out standardization processing on collected oil sample data characteristic information in a normal operation state, wherein the average value is 0 and the variance is 1 after the standardization processing, and eliminating the influence of different amplitudes of each characteristic on a model.
(2) And calculating the normalized sample data set kernel matrix.
(3) Performing centralization processing on the core matrix, wherein a centralization formula of the core matrix is as follows:
Figure BDA0002848813140000081
where K represents the kernel matrix prior to centralization,
Figure BDA0002848813140000082
representing the kernel matrix after centralization, 1NRepresents an N-dimensional square matrix with each element being 1/N, and N represents the number of samples.
(4) And (4) solving the eigenvalue and the eigenvector of the kernel matrix, and arranging the eigenvalue and the eigenvector in the descending order.
And 102, determining parameters of a kernel function of a kernel principal component analysis algorithm for chassis engine fault diagnosis by using the training sample set and a particle swarm optimization algorithm of dynamic inertia factors with the accuracy of fault monitoring as a fitness function to obtain the kernel principal component analysis algorithm with the determined parameters.
And (4) performing dimensionality reduction and fault monitoring on the original characteristic parameters extracted in the step 101 by adopting a kernel principal component analysis method. The kernel function of the kernel principal component analysis method adopts a Gaussian kernel function:
exp(||x-y||2/w)
where w is equal to or greater than 0 and x and y are known vectors. Different values of the Gaussian kernel function w parameter have great influence on the final dimensionality reduction result and the fault monitoring result, and the Gaussian kernel function w parameter is optimized by adopting a particle swarm optimization algorithm. The principle of the particle swarm optimization algorithm is as follows:
assuming that m particles are provided, each particle represents a potential solution, the variable of the optimization problem is d, the search space of the particle swarm optimization algorithm is d-dimensional, and the particles update the speed and the position of the particles by continuously tracking the optimal value of the individual particles and the optimal value of the group of particles. Wherein:
Figure BDA0002848813140000083
in the above formula
Figure BDA0002848813140000084
Represents the velocity value of the ith particle at time t, where VminAnd VmaxA minimum and a maximum value of the artificially set particle velocity.
Figure BDA0002848813140000091
In the above formula
Figure BDA0002848813140000092
Represents the position of the ith particle at time t, where XminAnd XmaxA minimum and a maximum of the artificially set particle positions.
The optima for individual particles are represented by pbest and for population particles by gbest. The particle continuously updates the speed and the position of the particle by tracking the individual optimal value and the group optimal value, and the updating formula is as follows:
Figure BDA0002848813140000093
Figure BDA0002848813140000094
in the above updating formula
Figure BDA0002848813140000095
And
Figure BDA0002848813140000096
representing the speed and position after updating, and w is an inertia factor of the particle swarm optimization algorithm, and generally takes a value between 0.4 and 0.9. c. C1And c2For learning factor, λ1And λ2Is a random number between 0 and 1.
When the value w is larger, the global search capability of the particle swarm optimization algorithm is stronger, but the local search capability is weaker. when the w value is smaller, the local searching capability of the particle swarm optimization algorithm is stronger, and the global searching capability is weaker. The best search effect can be achieved by adopting a method of dynamic inertia factors, and the formula is as follows:
w＝wmax-T*(wmax-wmin)/Tmax
in the above formula wmaxFor maximum inertia factor, w, set in the iterative processminFor minimum inertia factor, T, set during iterationmaxAnd T is the iteration times of the current algorithm and is the same as T.
103, carrying out optimization processing on the initial weight and the threshold of the BP neural network model by using a particle swarm optimization algorithm to obtain an initialized BP neural network model, and training the initialized BP neural network model by using the training sample set to obtain a trained BP neural network model;
wherein the standard BP neural network model is shown in fig. 3.
Wherein X represents input, and in practical application, the number of input of the characteristic parameters is taken as the number of input of the fault characteristic parameters after dimension reduction. Y represents an output, and is a fault phenomenon in actual application, and how many fault phenomena have output. The model comprises an input layer, a hidden layer and an output layer, wherein the hidden layer is generally one layer, all layers are connected, and none layer is connected. The number of neurons in the hidden layer is generally obtained by an empirical formula or repeated experiments, and the empirical formula is as follows:
Figure BDA0002848813140000101
wherein n ishFor the number of hidden layer neurons, n0Is the number of neurons in the input layer, n1L is an integer between 1 and 10 for the number of neurons in the output layer.
The traditional BP neural network model has the defects of low convergence speed and easy trapping in local minimum points, and the performance of the BP neural network model can be obviously improved by optimizing the initial weight and the threshold of the BP neural network model by adopting the particle swarm optimization algorithm. As shown in fig. 4, the specific implementation steps are as follows:
and determining the neuron numbers of an input layer, a hidden layer and an output layer of the BP neural network model according to the actual situation, and determining the total number of the initial weight and the threshold.
And initializing each parameter of the particle swarm optimization algorithm by taking the mean square error output by the neural network model as a fitness function of the particle swarm optimization algorithm.
And (4) iterative optimization of the particle swarm optimization algorithm is carried out, and the optimal initial weight and threshold of the BP neural network model are found.
And taking the optimization result of the particle swarm optimization algorithm as the initial weight and the threshold of the BP neural network model, and training the neural network by using training data to obtain the specific BP neural network model.
And 104, acquiring lubricating oil data of the monitoring state of the chassis engine, and extracting characteristic parameters to obtain monitoring data.
Step 104, acquiring lubricating oil data of a monitoring state of the chassis engine, and extracting characteristic parameters to obtain monitoring data, wherein the method specifically comprises the following steps: standardizing the lubricating oil data in a monitoring state to obtain the standardized lubricating oil data; calculating a kernel matrix of the lubricating oil data after the standardization treatment; using formulas
Figure BDA0002848813140000102
Performing centralization processing on the kernel matrix to obtain a core matrix subjected to centralization processing; wherein K represents the kernel matrix prior to centralization,
Figure BDA0002848813140000103
representing the kernel matrix after centralization, 1NRepresenting an N-dimensional square matrix, 1NEach element in the lubricating oil is 1/N, and N represents the number of lubricating oil data; and solving the eigenvalue and the eigenvector of the core matrix after the centralization treatment, and sequencing the eigenvalue and the eigenvector from large to small to obtain the sequenced eigenvalue and eigenvector as monitoring data. The method for extracting the characteristics is the same as the method for processing historical lubricating oil data in the process of establishing the training sample set, and details are not repeated here.
And 105, determining whether the monitoring data is fault data by adopting a kernel principal component analysis algorithm with determined parameters.
Step 105, determining whether the monitored data is fault data by using the kernel principal component analysis algorithm determined by the parameters, specifically including: calculating nonlinear principal components of the monitoring data by adopting a kernel principal component analysis algorithm with well-determined parameters; using formula T based on nonlinear principal components of the monitored data2＝(t1，t2，...，tq)Λ-1(t1，t2，...，tq)TCalculating T of the monitoring data2Statistics; wherein, t1，t2，...，tqIs a nonlinear principal component, Lambda, of the kernel principal component analysis extracted monitoring data-1Representing a diagonal inverse matrix formed by the characteristic values corresponding to each nonlinear principal component, wherein q is the number of the nonlinear principal component reserved in the kernel principal component analysis; using a formula based on the nonlinear principal component of the monitored data
Figure BDA0002848813140000111
Calculating SPE statistic of the monitoring data; wherein, tsRepresenting the non-linear pivot component corresponding to the s-th non-zero eigenvalue, k representing the number of non-zero eigenvalues in the eigenvalue corresponding to each non-linear pivot component, tjRepresenting the jth nonlinear principal component of the kernel principal component analysis extraction monitoring data; judging T of the monitoring data2Whether the statistic and the SPE statistic exceed T respectively2Obtaining a judgment result by the control limit and the SPE control limit; if the judgment result shows that the data is the fault data, determining the monitoring data as the fault data; and if the judgment result shows that the data is not normal, determining the monitoring data as normal data.
Wherein, T2The control limit and the SPE control limit need to be determined in advance, and the specific determination mode is as follows: using formulas
Figure BDA0002848813140000112
Determination of T2A control limit; wherein N represents the number of training samples in normal operation state, and q represents the number of non-training samples reserved in kernel principal component analysis in each training sample in normal operation stateNumber of linear principal component, alpha representing the level of examination, Fq，N-q，αRepresenting the distribution critical value when the degree of freedom is q, the detection level is alpha and the condition is N-q in the F distribution; using formulas
Figure BDA0002848813140000113
Determining SPE control limit; wherein, h represents an intermediate variable,
Figure BDA0002848813140000114
θ1、θ2and theta3Respectively representing a first feature value accumulated value, a second feature accumulated value and a third feature accumulated value, CαRepresenting the critical value of a normal distribution at the test level α.
As shown in fig. 2, step 105 specifically includes:
1) the method comprises the steps of carrying out standardization processing on collected oil sample data characteristic information in a normal operation state, wherein the average value is 0 and the variance is 1 after the standardization processing, and eliminating the influence of different amplitudes of each characteristic on a model.
2) A suitable kernel function is selected, typically a gaussian kernel function.
3) And optimizing the w parameter of the Gaussian kernel function by adopting a particle swarm optimization algorithm, wherein the optimization target is that the fault monitoring accuracy is highest.
4) And calculating the normalized sample data set kernel matrix.
5) Performing centralization processing on the core matrix, wherein a centralization formula of the core matrix is as follows:
Figure BDA0002848813140000121
where K represents the kernel matrix prior to centralization,
Figure BDA0002848813140000122
representing the kernel matrix after centralization, 1NRepresents an N-dimensional square matrix with each element being 1/N, and N represents the number of samples.
6) And (4) solving the eigenvalue and the eigenvector of the kernel matrix, and arranging the eigenvalue and the eigenvector in the descending order.
7) And calculating the nonlinear principal components of the sample data which normally runs.
8) Calculating T of normal operation sample data2And SPE statistics, and calculating the control limit of the statistics under certain confidence. Wherein T is2The statistic reflects the state of a variable by representing the fluctuation of a nonlinear principal component vector mode in a sample data kernel principal component model, and the SPE statistic refers to the error of each sampling of the statistical model on the change trend. Their calculation formula is as follows:
T2＝(t1，t2，...，tq)Λ-1(t1，t2，...，tq)T
wherein t is1，t2，...，tqIs a nonlinear principal component, Lambda, of the sample data extracted by kernel principal component analysis-1Is a diagonal inverse matrix, T, formed by eigenvalues corresponding to each principal element2The control limit for the statistic is calculated as follows:
Figure BDA0002848813140000123
wherein N represents the number of samples, q is the number of kernel principal elements retained in kernel principal component analysis, α is the test level, and the distribution critical value in F distribution when the degree of freedom is q, the test level is α, and the condition is N-q is Fq，N-q，α。
Figure BDA0002848813140000124
Where k represents the number of non-zero eigenvalues of the N eigenvalues and q is the number of reserved kernel principal elements.
The control limit for the SPE statistic is calculated as follows:
Figure BDA0002848813140000125
wherein theta isiFor the eigenvalue accumulated values from q +1 to k,
Figure BDA0002848813140000126
Cαrepresents the critical value of a normal distribution at the test level a.
9) Processing the collected test sample data according to the steps to calculate the T of the test sample2And SPE statistics, if the control limit is exceeded at the same time, indicating that a fault occurs.
And 106, when the monitoring data are fault data, inputting the monitoring data into a trained BP neural network model, and determining the fault type corresponding to the monitoring data.
In step 105, the dimension reduction and fault monitoring of sample data are realized, and the data without faults are stored in the database without alarming. When fault data are monitored, the fault data are stored in a database and an alarm prompt is given.
And the monitored fault data is used as the input of the BP neural network model to realize classification, namely fault diagnosis.
The invention takes the fault data after dimension reduction obtained in the step 105 as the input of the BP neural network model, and realizes fault classification, namely fault diagnosis.
The invention also provides a chassis engine fault diagnosis system based on the particle swarm optimization algorithm, which comprises the following steps:
and the training sample set establishing module is used for acquiring historical lubricating oil data of the chassis engine, extracting characteristic parameters and establishing a training sample set.
And the kernel principal component analysis parameter determining module is used for determining the parameters of the kernel function of the kernel principal component analysis algorithm for chassis engine fault diagnosis by using the training sample set, taking the accuracy of fault monitoring as a fitness function and adopting a particle swarm optimization algorithm of dynamic inertia factors to obtain the kernel principal component analysis algorithm with well-determined parameters.
The speed updating formula of the particle swarm optimization algorithm of the dynamic inertia factors is as follows:
Figure BDA0002848813140000131
the position updating formula of the particle swarm optimization algorithm of the dynamic inertia factors is as follows:
Figure BDA0002848813140000132
the dynamic inertia factor updating formula of the particle swarm optimization algorithm of the dynamic inertia factors is as follows:
w＝wmax-t*(wmax-wmin)/Tmax；
wherein the content of the first and second substances,
Figure BDA0002848813140000133
and
Figure BDA0002848813140000134
respectively representing the velocity vector and the position vector of the ith particle before the updating of the t iteration,
Figure BDA0002848813140000135
and
Figure BDA0002848813140000136
respectively representing the velocity vector and the position vector of the ith particle after the ith iteration update, w is an inertia factor of the particle swarm optimization algorithm, c1And c2Respectively a first learning factor and a second learning factor, lambda1And λ2A first random number and a second random number between 0 and 1, respectively; w is amaxAnd wminRespectively representing the maximum inertia factor and the minimum inertia factor, T, set in the iterative processmaxThe maximum number of iterations is indicated,
Figure BDA0002848813140000141
representing the individual optimal positions during the t-th iteration,
Figure BDA0002848813140000142
representing the global optimum position during the t-th iteration.
The BP neural network model training module is used for carrying out optimization processing on the initial weight and the threshold of the BP neural network model by utilizing a particle swarm optimization algorithm to obtain an initialized BP neural network model, and training the initialized BP neural network model by utilizing the training sample set to obtain a trained BP neural network model;
and the characteristic parameter extraction module is used for acquiring lubricating oil data of the monitoring state of the chassis engine and extracting characteristic parameters to obtain monitoring data.
The characteristic parameter extraction module specifically comprises: the standardized processing submodule is used for carrying out standardized processing on the lubricating oil data in the monitoring state to obtain the standardized lubricating oil data; the kernel matrix calculation submodule is used for calculating a kernel matrix of the lubricating oil data after the standardization processing; a centralized processing submodule for utilizing the formula
Figure BDA0002848813140000143
Performing centralization processing on the kernel matrix to obtain a core matrix subjected to centralization processing; wherein K represents the kernel matrix prior to centralization,
Figure BDA0002848813140000144
representing the kernel matrix after centralization, 1NRepresenting an N-dimensional square matrix, 1NEach element in the lubricating oil is 1/N, and N represents the number of lubricating oil data; and the characteristic extraction submodule is used for solving the characteristic value and the characteristic vector of the core matrix after the centralization processing, sequencing the characteristic value and the characteristic vector from large to small, and obtaining the sequenced characteristic value and characteristic vector as monitoring data.
And the fault diagnosis module is used for determining whether the monitoring data is fault data by adopting a kernel principal component analysis algorithm with determined parameters.
The fault diagnosis module specifically comprises: the principal component analysis submodule is used for calculating the nonlinear principal component of the monitoring data by adopting a kernel principal component analysis algorithm with determined parameters; t is2A statistic calculation submodule for using the formula T according to the nonlinear principal component of the monitored data2＝(t1，t2，...，tq)Λ-1(t1，t2，...，tq)TCalculating T of the monitoring data2Statistics; wherein, t1，t2，...，tqIs a nonlinear principal component, Lambda, of the kernel principal component analysis extracted monitoring data-1Representing a diagonal inverse matrix formed by the characteristic values corresponding to each nonlinear principal component, wherein q is the number of the nonlinear principal component reserved in the kernel principal component analysis; SPE statistic calculation submodule for calculating nonlinear principal component of monitored data by using formula
Figure BDA0002848813140000145
Calculating SPE statistic of the monitoring data; wherein, tsRepresenting the non-linear pivot component corresponding to the s-th non-zero eigenvalue, k representing the number of non-zero eigenvalues in the eigenvalue corresponding to each non-linear pivot component, tjRepresenting the jth nonlinear principal component of the kernel principal component analysis extraction monitoring data; a judgment result submodule for judging T of the monitoring data2Whether the statistic and the SPE statistic exceed T respectively2Obtaining a judgment result by the control limit and the SPE control limit; the fault diagnosis submodule is used for determining the monitoring data as fault data if the judgment result shows that the monitoring data is positive; and if the judgment result shows that the data is not normal, determining the monitoring data as normal data.
And the fault type determining module is used for inputting the monitoring data into a trained BP neural network model when the monitoring data is fault data, and determining the fault type corresponding to the monitoring data.
Compared with the prior art, the invention has the beneficial effects that:
the existing fault diagnosis technology generally has no fault monitoring link or has low fault monitoring accuracy. The invention realizes the dimension reduction and fault monitoring of the sample data by utilizing the kernel principal component analysis method improved by the particle swarm optimization algorithm, improves the fault monitoring accuracy, reduces the workload of subsequent fault diagnosis tasks, and improves the identification accuracy of the subsequent fault diagnosis method.
The BP neural network model method in the existing fault diagnosis method has certain defects, such as easy falling into local minimum, slow convergence rate and the like. The invention utilizes the particle swarm optimization algorithm to carry out optimization processing on the initial weight and the threshold of the BP neural network model, improves the global search capability of the BP neural network model, accelerates the convergence speed of the neural network, and improves the efficiency and the accuracy of fault diagnosis in practical application.
The equivalent embodiments in the present specification are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts between the equivalent embodiments can be referred to each other.
The principles and embodiments of the present invention have been described herein using specific examples, which are provided only to help understand the method and the core concept of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In summary, this summary should not be construed to limit the present invention.

Claims (10)

1. A chassis engine fault diagnosis method based on a particle swarm optimization algorithm is characterized by comprising the following steps:
acquiring historical lubricating oil data of a chassis engine, extracting characteristic parameters, and establishing a training sample set;
determining parameters of a kernel function of a kernel principal component analysis algorithm for chassis engine fault diagnosis by using the training sample set and taking the accuracy of fault monitoring as a fitness function and adopting a particle swarm optimization algorithm of dynamic inertia factors to obtain the kernel principal component analysis algorithm with well-determined parameters;
optimizing the initial weight and the threshold of the BP neural network model by utilizing a particle swarm optimization algorithm to obtain an initialized BP neural network model, and training the initialized BP neural network model by utilizing the training sample set to obtain a trained BP neural network model;
obtaining lubricating oil data of a monitoring state of a chassis engine, and extracting characteristic parameters to obtain monitoring data;
determining whether the monitoring data is fault data or not by adopting a kernel principal component analysis algorithm with determined parameters;
and when the monitoring data are fault data, inputting the monitoring data into a trained BP neural network model, and determining the fault type corresponding to the monitoring data.
2. The particle swarm optimization algorithm-based chassis engine fault diagnosis method according to claim 1, wherein the speed updating formula of the particle swarm optimization algorithm of the dynamic inertia factors is as follows:
Figure FDA0002848813130000011
the position updating formula of the particle swarm optimization algorithm of the dynamic inertia factors is as follows:
Figure FDA0002848813130000012
the dynamic inertia factor updating formula of the particle swarm optimization algorithm of the dynamic inertia factors is as follows:
w＝wmax-t*(wmax-wmin)/Tmax；
wherein the content of the first and second substances,
Figure FDA0002848813130000013
and
Figure FDA0002848813130000014
respectively representing the velocity vector and the position vector of the ith particle before the updating of the t iteration,
Figure FDA0002848813130000015
and
Figure FDA0002848813130000016
respectively representing the velocity vector and the position vector of the ith particle after the ith iteration update, w is an inertia factor of the particle swarm optimization algorithm, c1And c2Respectively a first learning factor and a second learning factor, lambda1And λ2A first random number and a second random number between 0 and 1, respectively; w is amaxAnd wminRespectively representing the maximum inertia factor and the minimum inertia factor, T, set in the iterative processmaxThe maximum number of iterations is indicated,
Figure FDA0002848813130000017
representing the individual optimal positions during the t-th iteration,
Figure FDA0002848813130000018
representing the global optimum position during the t-th iteration.
3. The chassis engine fault diagnosis method based on the particle swarm optimization algorithm according to claim 1, wherein the obtaining of the lubricating oil data of the chassis engine in the monitoring state and the feature parameter extraction are performed to obtain the monitoring data specifically comprises:
standardizing the lubricating oil data in a monitoring state to obtain the standardized lubricating oil data;
calculating a kernel matrix of the lubricating oil data after the standardization treatment;
using formulas
Figure FDA0002848813130000022
Centralizing the kernel matrixProcessing to obtain a core matrix after centralized processing; wherein K represents the kernel matrix prior to centralization,
Figure FDA0002848813130000023
representing the kernel matrix after centralization, 1NRepresenting an N-dimensional square matrix, 1NEach element in the lubricating oil is 1/N, and N represents the number of lubricating oil data;
and solving the eigenvalue and the eigenvector of the core matrix after the centralization treatment, and sequencing the eigenvalue and the eigenvector from large to small to obtain the sequenced eigenvalue and eigenvector as monitoring data.
4. The chassis engine fault diagnosis method based on the particle swarm optimization algorithm according to claim 1, wherein the determining whether the monitored data is fault data by using the kernel principal component analysis algorithm with determined parameters specifically comprises:
calculating nonlinear principal components of the monitoring data by adopting a kernel principal component analysis algorithm with well-determined parameters;
using formula T based on nonlinear principal components of the monitored data2＝(t1，t2，...，tq)Λ-1(t1，t2，...，tq)TCalculating T of the monitoring data2Statistics; wherein, t1，t2，...，tqIs a nonlinear principal component, Lambda, of the kernel principal component analysis extracted monitoring data-1Representing a diagonal inverse matrix formed by the characteristic values corresponding to each nonlinear principal component, wherein q is the number of the nonlinear principal component reserved in the kernel principal component analysis;
using a formula based on the nonlinear principal component of the monitored data
Figure FDA0002848813130000021
Calculating SPE statistic of the monitoring data; wherein, tsRepresenting the non-linear pivot component corresponding to the s-th non-zero eigenvalue, and k represents the eigenvalue corresponding to each non-linear pivot componentNumber of medium non-zero eigenvalues, tjRepresenting the jth nonlinear principal component of the kernel principal component analysis extraction monitoring data;
judging T of the monitoring data2Whether the statistic and the SPE statistic exceed T respectively2Obtaining a judgment result by the control limit and the SPE control limit;
if the judgment result shows that the data is the fault data, determining the monitoring data as the fault data;
and if the judgment result shows that the data is not normal, determining the monitoring data as normal data.
5. The particle swarm optimization algorithm-based chassis engine fault diagnosis method according to claim 1, wherein the parameter-determined kernel principal component analysis algorithm is used for determining whether the monitored data is fault data, and the method further comprises the following steps:
using formulas
Figure FDA0002848813130000031
Determination of T2A control limit; wherein N represents the number of training samples in normal operation state, q represents the number of nonlinear principal component components retained in kernel principal component analysis in each training sample in normal operation state, α represents the level of examination, and Fq，N-q，αRepresenting the distribution critical value when the degree of freedom is q, the detection level is alpha and the condition is N-q in the F distribution;
using formulas
Figure FDA0002848813130000032
Determining SPE control limit; wherein, h represents an intermediate variable,
Figure FDA0002848813130000033
θ1、θ2and theta3Respectively representing a first feature value accumulated value, a second feature accumulated value and a third feature accumulated value, CαRepresenting the critical value of a normal distribution at the test level α.
6. The chassis engine fault diagnosis method based on the particle swarm optimization algorithm according to claim 1, wherein the optimizing processing is performed on the initial weight and the threshold of the BP neural network model by using the particle swarm optimization algorithm to obtain the initialized BP neural network model, and the training sample set is used to train the initialized BP neural network model to obtain the trained BP neural network model, which specifically comprises:
determining the optimal initial weight and threshold of the BP neural network model by using the training sample set and the mean square error output by the BP neural network model as a fitness function and adopting a particle swarm optimization algorithm to obtain the initialized BP neural network model;
and training the initialized BP neural network model by using the training sample set to obtain the trained BP neural network model.
7. A chassis engine fault diagnosis system based on particle swarm optimization algorithm, which is characterized by comprising:
the training sample set establishing module is used for acquiring historical lubricating oil data of the chassis engine, extracting characteristic parameters and establishing a training sample set;
the kernel principal component analysis parameter determination module is used for determining parameters of a kernel function of a kernel principal component analysis algorithm for chassis engine fault diagnosis by using the training sample set, taking the accuracy of fault monitoring as a fitness function and adopting a particle swarm optimization algorithm of dynamic inertia factors to obtain the kernel principal component analysis algorithm with well-determined parameters;
the BP neural network model training module is used for carrying out optimization processing on the initial weight and the threshold of the BP neural network model by utilizing a particle swarm optimization algorithm to obtain an initialized BP neural network model, and training the initialized BP neural network model by utilizing the training sample set to obtain a trained BP neural network model;
the characteristic parameter extraction module is used for acquiring lubricating oil data of a chassis engine in a monitoring state and extracting characteristic parameters to obtain monitoring data;
the fault diagnosis module is used for determining whether the monitoring data is fault data or not by adopting a kernel principal component analysis algorithm with determined parameters;
and the fault type determining module is used for inputting the monitoring data into a trained BP neural network model when the monitoring data is fault data, and determining the fault type corresponding to the monitoring data.
8. The particle swarm optimization algorithm-based chassis engine fault diagnosis system according to claim 7, wherein the speed updating formula of the particle swarm optimization algorithm of the dynamic inertia factors is as follows:
Figure FDA0002848813130000041
the position updating formula of the particle swarm optimization algorithm of the dynamic inertia factors is as follows:
Figure FDA0002848813130000042
the dynamic inertia factor updating formula of the particle swarm optimization algorithm of the dynamic inertia factors is as follows:
w＝wmax-t*(wmax-wmin)/Tmax；
wherein the content of the first and second substances,
Figure FDA0002848813130000043
and
Figure FDA0002848813130000044
respectively representing the velocity vector and the position vector of the ith particle before the updating of the t iteration,
Figure FDA0002848813130000045
and
Figure FDA0002848813130000046
respectively representing the velocity vector and the position vector of the ith particle after the ith iteration update, w is an inertia factor of the particle swarm optimization algorithm, c1And c2Respectively a first learning factor and a second learning factor, lambda1And λ2A first random number and a second random number between 0 and 1, respectively; w is amaxAnd wminRespectively representing the maximum inertia factor and the minimum inertia factor, T, set in the iterative processmaxThe maximum number of iterations is indicated,
Figure FDA0002848813130000051
representing the individual optimal positions during the t-th iteration,
Figure FDA0002848813130000052
representing the global optimum position during the t-th iteration.
9. The system for diagnosing the fault of the chassis engine based on the particle swarm optimization algorithm according to claim 7, wherein the characteristic parameter extraction module specifically comprises:
the standardized processing submodule is used for carrying out standardized processing on the lubricating oil data in the monitoring state to obtain the standardized lubricating oil data;
the kernel matrix calculation submodule is used for calculating a kernel matrix of the lubricating oil data after the standardization processing;
a centralized processing submodule for utilizing the formula
Figure FDA0002848813130000053
Performing centralization processing on the kernel matrix to obtain a core matrix subjected to centralization processing; wherein K represents the kernel matrix prior to centralization,
Figure FDA0002848813130000054
representing the kernel matrix after centralization, 1NRepresenting an N-dimensional square matrix, 1NEach element in the lubricating oil is 1/N, and N represents the number of lubricating oil data;
and the characteristic extraction submodule is used for solving the characteristic value and the characteristic vector of the core matrix after the centralization processing, sequencing the characteristic value and the characteristic vector from large to small, and obtaining the sequenced characteristic value and characteristic vector as monitoring data.
10. The particle swarm optimization algorithm-based chassis engine fault diagnosis system according to claim 7, wherein the fault diagnosis module specifically comprises:
the principal component analysis submodule is used for calculating the nonlinear principal component of the monitoring data by adopting a kernel principal component analysis algorithm with determined parameters;
T2a statistic calculation submodule for using the formula T according to the nonlinear principal component of the monitored data2＝(t1，t2，...，tq)Λ-1(t1，t2，...，tq)TCalculating T of the monitoring data2Statistics; wherein, t1，t2，...，tqIs a nonlinear principal component, Lambda, of the kernel principal component analysis extracted monitoring data-1Representing a diagonal inverse matrix formed by the characteristic values corresponding to each nonlinear principal component, wherein q is the number of the nonlinear principal component reserved in the kernel principal component analysis;
SPE statistic calculation submodule for calculating nonlinear principal component of monitored data by using formula
Figure FDA0002848813130000055
Calculating SPE statistic of the monitoring data; wherein, tsRepresenting the non-linear pivot component corresponding to the s-th non-zero eigenvalue, k representing the number of non-zero eigenvalues in the eigenvalue corresponding to each non-linear pivot component, tjRepresenting the jth nonlinear principal component of the kernel principal component analysis extraction monitoring data;
a judgment result submodule for judging T of the monitoring data2The statistic and the SPE statistic areWhether or not exceed T respectively2Obtaining a judgment result by the control limit and the SPE control limit;
the fault diagnosis submodule is used for determining the monitoring data as fault data if the judgment result shows that the monitoring data is positive; and if the judgment result shows that the data is not normal, determining the monitoring data as normal data.
CN202011520723.6A 2020-12-21 2020-12-21 Chassis engine fault diagnosis method and system based on particle swarm optimization algorithm Pending CN112504682A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011520723.6A CN112504682A (en) 2020-12-21 2020-12-21 Chassis engine fault diagnosis method and system based on particle swarm optimization algorithm

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011520723.6A CN112504682A (en) 2020-12-21 2020-12-21 Chassis engine fault diagnosis method and system based on particle swarm optimization algorithm

Publications (1)

Publication Number Publication Date
CN112504682A true CN112504682A (en) 2021-03-16

Family

ID=74921880

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011520723.6A Pending CN112504682A (en) 2020-12-21 2020-12-21 Chassis engine fault diagnosis method and system based on particle swarm optimization algorithm

Country Status (1)

Country Link
CN (1) CN112504682A (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113139606A (en) * 2021-04-27 2021-07-20 上海联创设计集团股份有限公司 Heating and ventilation fan fault detection method and system based on artificial intelligence algorithm
CN113188794A (en) * 2021-04-29 2021-07-30 东北电力大学 Gearbox fault diagnosis method and device based on improved PSO-BP neural network
CN116150676A (en) * 2023-04-19 2023-05-23 山东能源数智云科技有限公司 Equipment fault diagnosis and identification method and device based on artificial intelligence
CN116937820A (en) * 2023-09-19 2023-10-24 深圳凯升联合科技有限公司 High-voltage circuit state monitoring method based on deep learning algorithm

Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107065842A (en) * 2017-05-26 2017-08-18 宁波大学 A kind of fault detection method based on particle group optimizing core independent component analysis model
CN108062565A (en) * 2017-12-12 2018-05-22 重庆科技学院 Double pivots-dynamic kernel principal component analysis method for diagnosing faults based on chemical industry TE processes
CN108334059A (en) * 2018-02-26 2018-07-27 上海理工大学 Based on particle cluster algorithm Optimized BP Neural Network Symbolic fault diagnosis method
CN108734202A (en) * 2018-04-27 2018-11-02 西安工程大学 A kind of Fault Diagnosis for HV Circuit Breakers method based on improved BP
CN108921230A (en) * 2018-07-25 2018-11-30 浙江浙能嘉华发电有限公司 Method for diagnosing faults based on class mean value core pivot element analysis and BP neural network
CN109426889A (en) * 2017-09-01 2019-03-05 南京理工大学 Short-term load forecasting method based on KPCA in conjunction with improvement neural network
CN109447236A (en) * 2018-09-28 2019-03-08 重庆邮电大学 A kind of method for diagnosing faults of hybrid vehicle heat management system
CN111351668A (en) * 2020-01-14 2020-06-30 江苏科技大学 Diesel engine fault diagnosis method based on optimized particle swarm algorithm and neural network
US20200241487A1 (en) * 2019-01-24 2020-07-30 Qingdao university of technology Networked control system time-delay compensation method based on predictive control
CN111521398A (en) * 2020-03-18 2020-08-11 武汉工程大学 Gear box fault diagnosis method and system based on BP neural network and principal component analysis method

Patent Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107065842A (en) * 2017-05-26 2017-08-18 宁波大学 A kind of fault detection method based on particle group optimizing core independent component analysis model
CN109426889A (en) * 2017-09-01 2019-03-05 南京理工大学 Short-term load forecasting method based on KPCA in conjunction with improvement neural network
CN108062565A (en) * 2017-12-12 2018-05-22 重庆科技学院 Double pivots-dynamic kernel principal component analysis method for diagnosing faults based on chemical industry TE processes
CN108334059A (en) * 2018-02-26 2018-07-27 上海理工大学 Based on particle cluster algorithm Optimized BP Neural Network Symbolic fault diagnosis method
CN108734202A (en) * 2018-04-27 2018-11-02 西安工程大学 A kind of Fault Diagnosis for HV Circuit Breakers method based on improved BP
CN108921230A (en) * 2018-07-25 2018-11-30 浙江浙能嘉华发电有限公司 Method for diagnosing faults based on class mean value core pivot element analysis and BP neural network
CN109447236A (en) * 2018-09-28 2019-03-08 重庆邮电大学 A kind of method for diagnosing faults of hybrid vehicle heat management system
US20200241487A1 (en) * 2019-01-24 2020-07-30 Qingdao university of technology Networked control system time-delay compensation method based on predictive control
CN111351668A (en) * 2020-01-14 2020-06-30 江苏科技大学 Diesel engine fault diagnosis method based on optimized particle swarm algorithm and neural network
CN111521398A (en) * 2020-03-18 2020-08-11 武汉工程大学 Gear box fault diagnosis method and system based on BP neural network and principal component analysis method

Non-Patent Citations (6)

* Cited by examiner, † Cited by third party
Title
FIROOZ SADJADI 等: "《自动目标识别的物理学原理》", 30 April 2014, 国防工业出版社 *
岳海豹: "基于核主元分析与信息熵的柴油机故障诊断", 《CNKI中国优秀硕士学位论文全文数据库工程科技Ⅱ辑》 *
曹建芳: "《大规模场景图像的情感语义分析若干关键技术研究》", 31 January 2018 *
李柱国: "《油液分析诊断技术》", 31 January 1997 *
王秋滢: "《船用调制型惯性导航及其组合导航技术》", 31 March 2017 *
魏秀业: "基于粒子群优化的齿轮箱智能故障诊断研究", 《CNKI中国博士学位论文全文数据库工程科技Ⅱ辑》 *

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113139606A (en) * 2021-04-27 2021-07-20 上海联创设计集团股份有限公司 Heating and ventilation fan fault detection method and system based on artificial intelligence algorithm
CN113188794A (en) * 2021-04-29 2021-07-30 东北电力大学 Gearbox fault diagnosis method and device based on improved PSO-BP neural network
CN116150676A (en) * 2023-04-19 2023-05-23 山东能源数智云科技有限公司 Equipment fault diagnosis and identification method and device based on artificial intelligence
CN116150676B (en) * 2023-04-19 2023-09-26 山东能源数智云科技有限公司 Equipment fault diagnosis and identification method and device based on artificial intelligence
CN116937820A (en) * 2023-09-19 2023-10-24 深圳凯升联合科技有限公司 High-voltage circuit state monitoring method based on deep learning algorithm
CN116937820B (en) * 2023-09-19 2024-01-05 深圳凯升联合科技有限公司 High-voltage circuit state monitoring method based on deep learning algorithm

Similar Documents

Publication Publication Date Title
CN112504682A (en) Chassis engine fault diagnosis method and system based on particle swarm optimization algorithm
CN110516305B (en) Intelligent fault diagnosis method under small sample based on attention mechanism meta-learning model
CN112254964A (en) Rolling bearing fault diagnosis method based on rapid multi-scale convolution neural network
CN109800875A (en) Chemical industry fault detection method based on particle group optimizing and noise reduction sparse coding machine
CN109145516B (en) Analog circuit fault identification method based on improved extreme learning machine
CN111562108A (en) Rolling bearing intelligent fault diagnosis method based on CNN and FCMC
CN109389171B (en) Medical image classification method based on multi-granularity convolution noise reduction automatic encoder technology
CN112000084B (en) Intelligent BIT design method of controller module based on 1D-CNN and GRU-SVM
CN111753891B (en) Rolling bearing fault diagnosis method based on unsupervised feature learning
CN113642754A (en) Complex industrial process fault prediction method based on RF noise reduction self-coding information reconstruction and time convolution network
CN110880369A (en) Gas marker detection method based on radial basis function neural network and application
CN114048688A (en) Method for predicting service life of bearing of wind power generator
CN114169377A (en) G-MSCNN-based fault diagnosis method for rolling bearing in noisy environment
CN111079926A (en) Equipment fault diagnosis method with self-adaptive learning rate based on deep learning
CN114298134A (en) Wind power prediction method and device and electronic equipment
CN114819054A (en) Power electronic system state monitoring method based on physical information neural network
CN114429152A (en) Rolling bearing fault diagnosis method based on dynamic index antagonism self-adaption
CN114091525A (en) Rolling bearing degradation trend prediction method
CN115859077A (en) Multi-feature fusion motor small sample fault diagnosis method under variable working conditions
CN117056678B (en) Machine pump equipment operation fault diagnosis method and device based on small sample
CN112001436A (en) Water quality classification method based on improved extreme learning machine
CN113126489A (en) CNN-GRU-BINN-based intelligent BIT design method for heavy-duty gas turbine control system
CN113159088A (en) Fault monitoring and diagnosis method based on multi-feature fusion and width learning
CN112365093A (en) GRU deep learning-based multi-feature factor red tide prediction model
CN116400168A (en) Power grid fault diagnosis method and system based on depth feature clustering

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: 20210316