CN110837851A - Fault diagnosis method for hydraulic pump of electro-hydrostatic actuator - Google Patents

Fault diagnosis method for hydraulic pump of electro-hydrostatic actuator Download PDF

Info

Publication number
CN110837851A
CN110837851A CN201911024933.3A CN201911024933A CN110837851A CN 110837851 A CN110837851 A CN 110837851A CN 201911024933 A CN201911024933 A CN 201911024933A CN 110837851 A CN110837851 A CN 110837851A
Authority
CN
China
Prior art keywords
hydraulic pump
electro
quantum
particle
hydrostatic actuator
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
CN201911024933.3A
Other languages
Chinese (zh)
Inventor
丁建军
贺梓洲
仙丹
刘阳鹏
陈鹏
李涛
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.)
Xian Jiaotong University
Original Assignee
Xian Jiaotong 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 Xian Jiaotong University filed Critical Xian Jiaotong University
Priority to CN201911024933.3A priority Critical patent/CN110837851A/en
Publication of CN110837851A publication Critical patent/CN110837851A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F04—POSITIVE - DISPLACEMENT MACHINES FOR LIQUIDS; PUMPS FOR LIQUIDS OR ELASTIC FLUIDS
    • F04B—POSITIVE-DISPLACEMENT MACHINES FOR LIQUIDS; PUMPS
    • F04B51/00—Testing machines, pumps, or pumping installations
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N10/00—Quantum computing, i.e. information processing based on quantum-mechanical phenomena
    • 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]

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Software Systems (AREA)
  • Computing Systems (AREA)
  • Evolutionary Biology (AREA)
  • Mathematical Physics (AREA)
  • Computational Linguistics (AREA)
  • Molecular Biology (AREA)
  • Mechanical Engineering (AREA)
  • Computational Mathematics (AREA)
  • Condensed Matter Physics & Semiconductors (AREA)
  • Mathematical Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Pure & Applied Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Health & Medical Sciences (AREA)
  • Fluid-Pressure Circuits (AREA)

Abstract

The invention discloses a method for diagnosing a hydraulic pump fault of an electro-hydrostatic actuator, which comprises the steps of firstly utilizing a vibration acceleration sensor or a pressure sensor to collect vibration signals on a hydraulic pump shell of the electro-hydrostatic actuator or pressure signals at an outlet of a hydraulic pump, constructing a characteristic vector based on a wavelet packet decomposition technology, and coding the fault type of the hydraulic pump of the electro-hydrostatic actuator to be used as an output sample of training. Then, on a vibration signal data set or a pressure signal data set of a hydraulic pump of the electro-hydrostatic actuator, automatically optimizing model hyperparameters C and delta of a support vector machine by using a quantum particle group algorithm to obtain values of the optimal hyperparameters C and delta, bringing the values into the support vector machine model, and training the support vector machine model of the optimal hyperparameters by using feature vectors extracted from a training data set of the hydraulic pump of the electro-hydrostatic actuator and fault codes corresponding to the feature vectors; and finally, performing hydraulic pump fault diagnosis on the electro-hydrostatic actuator on the hydraulic pump test data set.

Description

Fault diagnosis method for hydraulic pump of electro-hydrostatic actuator
Technical Field
The invention relates to a method for monitoring and diagnosing an early fault state of a hydraulic pump of an electro-hydrostatic actuator, in particular to a method for diagnosing a fault of a hydraulic pump of an electro-hydrostatic actuator.
Background
An Electro-Hydrostatic Actuator (EHA) is a novel airborne hydraulic Actuator widely applied to the field of aerospace. The hydraulic pump is the core drive device of the electro-hydrostatic actuator hydraulic system, and the working environment is extremely severe. In a service period, a hydraulic pump of the electro-hydrostatic actuator is affected by abrasion, erosion, impact, vibration, fatigue and the like, and gradually loses efficacy to generate faults, and the normal operation of a space flight vehicle, a civil aircraft and the like can be directly affected by the fault of the hydraulic pump, so that serious life and property loss and even disastrous consequences are caused. Therefore, the state of a hydraulic pump, which is a core driving device of a hydraulic system of the electro-hydrostatic actuator, must be monitored, and early fault states must be diagnosed accurately and quickly. So that the staff can take relevant measures to avoid causing great loss of life and property.
The previous research on the hydraulic pump of the electro-hydrostatic actuator mainly focuses on the aspects of design, manufacture and power analysis, and the research on the aspects of monitoring, diagnosing and maintaining the early fault state is less. The method comprises the steps of performing fault diagnosis on a hydraulic pump of the electro-hydrostatic actuator, and on one hand, performing analysis diagnosis based on fault characteristic frequency appearing in a vibration signal; on the other hand, the diagnostic evaluation can also be carried out on the basis of the fault-frequency pressure pulsations occurring in the pressure signal. At present, the method for monitoring and diagnosing the state of a hydraulic pump of a hydraulic system of an electro-hydrostatic actuator mainly comprises the following steps: artificial neural networks, support vector machines, and the like. The existing hydraulic pump state monitoring method has the problems that the learning model of the self-adaptive regulator cannot be subjected to over-parameter, the diagnosis accuracy rate needs to be improved and the like. Therefore, a series of group intelligent optimization algorithms are applied to the acquisition of the optimal hyper-parameters of the machine learning model, such as a particle swarm algorithm and a genetic algorithm. The particle swarm algorithm is simple in concept and easy to implement, so that the particle swarm algorithm is widely concerned by the field of computers, but the standard particle swarm algorithm is easy to fall into local optimization, and the global convergence can not be guaranteed.
Disclosure of Invention
The invention aims to solve the problems of the existing hydraulic pump early fault state monitoring and diagnosis technology, and provides a hydraulic pump fault diagnosis method for an electro-hydrostatic actuator.
The invention is realized by adopting the following technical scheme:
a method for diagnosing hydraulic pump faults of an electro-hydrostatic actuator comprises the steps of collecting vibration signals on a hydraulic pump shell of the electro-hydrostatic actuator or pressure signals at an outlet of a hydraulic pump, constructing a characteristic vector based on a wavelet packet decomposition technology, and coding the fault type of the hydraulic pump of the electro-hydrostatic actuator to serve as an output sample of training; optimizing model hyperparameters C and delta of a support vector machine by utilizing a quantum particle group algorithm on a vibration signal data set or a pressure signal data set of a hydraulic pump of the electro-hydrostatic actuator to obtain values of the optimal hyperparameters C and delta, bringing the values into the support vector machine model, and training the support vector machine model of the optimal hyperparameters by utilizing feature vectors extracted from a training data set of the hydraulic pump of the electro-hydrostatic actuator and fault codes corresponding to the feature vectors; and finally, performing hydraulic pump fault diagnosis on the electro-hydrostatic actuator on the hydraulic pump test data set.
A further improvement of the invention is that the method comprises in particular the following steps:
step 1, collecting vibration signals on a shell or pressure signal data of an outlet of a hydraulic pump of an electro-hydrostatic actuator in a normal state or an abnormal state;
step 2, constructing a frequency band energy characteristic vector for a hydraulic pump data set of the electro-hydrostatic actuator based on a wavelet packet decomposition technology, and coding the fault type of the hydraulic pump of the electro-hydrostatic actuator to be used as an output sample of training;
3, searching model hyperparameter C and delta optimal values of a support vector machine by using quantum particle groups on a hydraulic pump data set of the electro-hydrostatic actuator;
step 4, training the quantum particle swarm optimization support vector machine by utilizing a hydraulic pump training sample data set of the electro-hydrostatic actuator to obtain a hydraulic pump fault diagnosis model of the electro-hydrostatic actuator;
and 5, performing fault diagnosis on the hydraulic pump of the electro-hydrostatic actuator on the hydraulic pump test set of the electro-hydrostatic actuator by using the trained quantum particle swarm optimization support vector machine model.
The invention is further improved in that in step 1, the vibration signal of the hydraulic pump is detected by using 2 acceleration sensors ax and ay in the direction vertical to the outlet of the hydraulic pump of the electro-hydrostatic actuator, or the pressure signal of the hydraulic pump is detected by using a pressure sensor P at the outlet of the hydraulic pump of the electro-hydrostatic actuator.
The further improvement of the invention is that in the step 2, before the characteristic vector is constructed, the vibration signal or the pressure signal of the hydraulic pump data set of the electro-hydrostatic actuator is subjected to proportion 4: 1, randomly extracting to obtain a training sample data set and a test data set;
the fault types of the hydraulic pump of the electro-hydrostatic actuator comprise normal hydraulic pump, loose ball head, misalignment of shaft, abrasion of a valve plate, abrasion of a bearing, fatigue of the bearing, fracture of the bearing and deformation of the bearing.
The further improvement of the invention is that in the step 3, the quantum particle swarm algorithm is adopted to optimize the hyperparameters C and delta of the support vector machine, and the specific optimizing process is as follows:
step 3.1, initializing the two-dimensional position P of the particle mapped by the population scale of the quantum particle swarm, the maximum iteration times, the support vector machine hyper-parameter penalty factor C and the nuclear parameter delta, wherein the iteration times t is 0;
step 3.2, the position of the particle of the quantum probability amplitude code is subjected to solution space transformation
Performing solution space transformation before calculating the objective function value of the model, wherein the solution space transformation result is the current search result of the solved model hyperparameter;
3.3, training a comprehensive evaluation formula Q for fault classification of a support vector machine based on a training data sample set characteristic vector of a hydraulic pump vibration signal or a pressure signal of the electro-hydrostatic actuator to serve as a target function, and calculating a target function value of each particle according to the target function;
step 3.4, updating the new local optimal position P of each particle according to the objective function valueib；
Step 3.5, updating the global optimal position P according to the objective function valueg；
3.6, updating the position of each particle in the population based on the quantum revolving gate;
and 3.7, repeating the steps 3.2-3.7 until the iteration is finished, and obtaining the global optimal position as an optimization result, namely the penalty factor C and the kernel parameter delta of the optimal support vector machine.
The further improvement of the invention is that in step 3.1, the model hyper-parameters C and delta are quantum encoded through quantum probability argument, and the specific method is as follows:
carrying out initialization coding on particle positions P by using quantum probability amplitude:
particle position:
Figure BDA0002248361240000041
wherein, thetaij2 pi × rand, rand being a random number between (0, 1); i is the particle population size; j e 1,2 represents the particle dimension, i.e. the number of parameter classes to be searched.
The further improvement of the invention is that in step 3.2, the solution space mapping of the quantum particle swarm positions:
marking the solution space transformation result of the particle position as X, distinguishing different dimensionality solution space mapping results by an X upper mark, and distinguishing sine and cosine solution space mapping results by an X lower mark;
then the result of the quantum probability amplitude cosine solution space mapping in the first qubit is:
Figure BDA0002248361240000042
the first qubit quantum probability amplitude sinusoidal solution space mapping result is
Figure BDA0002248361240000043
Wherein, CmaxAnd CminSolving the upper limit and the lower limit of the range for the optimal value of the penalty factor C, wherein subscripts C and s are used for distinguishing the mapping results of cosine and sine, C represents cosine, and s represents sine;
then the result of the quantum probability amplitude cosine solution space mapping in the second qubit is:
Figure BDA0002248361240000044
then the result of the sinusoidal solution space mapping of the quantum probability amplitude in the second qubit is:
Figure BDA0002248361240000045
wherein, deltamaxAnd deltaminSolving the upper limit and the lower limit of the range for the optimal value of the kernel parameter delta, wherein subscripts c and s are used for distinguishing mapping results of cosine and sine, c represents cosine, and s represents sine;
two solutions of each quantum bit corresponding to the optimization problem are obtained according to the solution space mapping result, and therefore the search efficiency is improved.
The further improvement of the invention is that in step 3.3, the diagnosis precision and recall rate of the fault diagnosis model are comprehensively considered, and a comprehensive evaluation formula Q is provided as a fault diagnosis target function of the hydraulic pump of the electro-hydrostatic actuator of the quantum particle swarm optimization support vector machine, wherein the comprehensive evaluation formula is as follows:
Figure BDA0002248361240000051
wherein, TP is the number of correctly classified fault classes, FP is the number of incorrectly classified fault classes, and FN is the number of incorrectly classified normal classes.
The further improvement of the invention is that in steps 3.4, 3.5 and 3.6, the quantum probability argument is updated through the quantum revolving gate based on the local optimum position and the global optimum position of the quantum particle swarm, and when the local optimum position and the global optimum position are both cosine positions, the updating formula is as follows:
the local optimal position traversed by the quantum particle at present is a cosine position, that is:
Pib＝(cos(θib1),cos(θib2))
the global optimal position traversed by the quantum population is also a cosine position:
Pg＝(cos(θg1),cos(θg2))
quantum amplitude change of particle:
Δθij(t+1)＝ω×Δθij(t)+c1×Δθbc+c2×Δθgc
where ω is the inertia factor, c1Is a self factor, c2Is a global factor, both are constants;
Figure BDA0002248361240000053
then the quantum revolving door:
Figure BDA0002248361240000054
wherein i is the particle population size; j is belonged to {1,2} and represents a particle dimension, namely the number of parameter types needing to be searched; b and g are used for distinguishing a local optimal value and a global optimal value of the particle, wherein b represents a local optimal value, and g represents a global optimal value; subscript c represents a cosine;
updating the particle probability amplitude coding position based on the quantum revolving gate:
Figure BDA0002248361240000061
wherein i is the particle population size; j ∈ {1,2} represents the particle dimension;
after the update, the particle PiTwo new positions of (c):
Pic＝(cos(θi1(t)+Δθi1(t+1)),cos(θi2(t)+Δθi2(t+1)))
Pis＝(sin(θi1(t)+Δθi1(t+1)),sin(θi2(t)+Δθi2(t+1)))。
the further improvement of the invention is that in steps 3.4, 3.5 and 3.6, the quantum probability argument is updated by the quantum revolving gate based on the local optimum position and the global optimum position of the quantum particle swarm, and when the local optimum position and the global optimum position are sinusoidal positions, the updating formula is as follows:
the local optimal position currently traversed by the quantum particle is a sinusoidal position, namely:
Pib＝(sin(θib1),sin(θib2))
the global optimal position traversed by the quantum population is also a sinusoidal position:
Pg＝(sin(θg1),sin(θg2))
quantum amplitude change of particle:
Δθij(t+1)＝ω×Δθij(t)+c1×Δθbs+c2×Δθgs
where ω is the inertia factor, c1Is a self factor, c2Is a global factor, both are constants;
Figure BDA0002248361240000063
then the quantum revolving door:
wherein i is the particle population size; j is belonged to {1,2} and represents a particle dimension, namely the number of parameter types needing to be searched; b and g are used for distinguishing a local optimal value and a global optimal value of the particle, wherein b represents a local optimal value, and g represents a global optimal value; subscript s represents sine;
updating the particle probability amplitude coding position based on the quantum revolving gate:
Figure BDA0002248361240000072
after the update, the particle PiTwo new positions of (c):
Pic＝(cos(θi1(t)+Δθi1(t+1)),cos(θi2(t)+Δθi2(t+1)))
Pis＝(sin(θi1(t)+Δθi1(t+1)),sin(θi2(t)+Δθi2(t+1)))。
the invention has at least the following beneficial technical effects:
the invention provides a method for diagnosing the fault of a hydraulic pump of an electro-hydrostatic actuator, which comprises the steps of collecting vibration signals or pressure pulsation signals of the hydraulic pump through a signal collecting system, wherein the vibration signals or the pressure pulsation signals comprise normal state signals and fault state signals of the hydraulic pump of the electro-hydrostatic actuator, randomly extracting the signals to construct a training data sample set and a testing data sample set, and calculating a frequency band energy characteristic vector based on wavelet packet decomposition; the positions of the particles are coded by utilizing the quantum probability argument, and then the quantum revolving door is introduced to enable the quantum argument to rotate by a certain angle, so that the positions of the particles are updated, and the searching efficiency is greatly improved. And then, the quantum particle swarm and the support vector machine model are combined to construct a hydraulic pump fault diagnosis model of the electro-hydraulic actuator based on quantum particle swarm optimization support vector, so that the super-parameter of the support vector machine model which realizes the highest hydraulic pump fault diagnosis accuracy on a hydraulic pump fault data set of the electro-hydraulic actuator is searched, and the accurate and rapid diagnosis of the hydraulic pump fault of the electro-hydraulic actuator is realized.
In conclusion, the self-adaptive model hyperparameter adjustment of the hydraulic pump fault diagnosis model of the electro-hydrostatic actuator is realized, and compared with the conventional electro-hydrostatic actuator diagnosis method, the method can improve the fault diagnosis accuracy of the hydraulic pump of the electro-hydrostatic actuator, is beneficial to reducing life and property losses caused by faults of aerospace vehicles, and improves the operation benefits of the aerospace vehicles.
Drawings
FIG. 1 is a schematic diagram of a hydraulic pump monitoring and diagnosing of an electro-hydrostatic actuator.
FIG. 2 is a route diagram of a fault diagnosis method of an electro-hydrostatic actuator based on a quantum particle swarm optimization support vector machine.
FIG. 3 is a flow chart of the method for searching the optimal hyper-parameters C and delta of the support vector machine based on the quantum particle swarm.
Detailed Description
The invention is described in further detail below with reference to the accompanying drawings.
In order to improve the accuracy of the fault diagnosis of the hydraulic pump of the electro-hydrostatic actuator, the invention provides a method for diagnosing the fault of the hydraulic pump of the electro-hydrostatic actuator. And then optimizing the model hyperparameters C and delta of the support vector machine by utilizing a quantum particle group algorithm on a vibration signal data set or a pressure signal data set of the hydraulic pump of the electro-hydrostatic actuator to obtain values of the optimal hyperparameters C and delta, bringing the values into the support vector machine model, and training the support vector machine model of the optimal hyperparameters by utilizing the characteristic vectors extracted from the training data set of the hydraulic pump of the electro-hydrostatic actuator and fault codes corresponding to the characteristic vectors. And finally, performing hydraulic pump fault diagnosis on the electro-hydrostatic actuator on the hydraulic pump test data set. Compared with a support vector machine and a particle swarm algorithm, the method for diagnosing the hydraulic pump fault of the electro-hydrostatic actuator has higher diagnosis precision and shorter training time.
As shown in figure 1, the method for monitoring the state of the hydraulic pump and diagnosing the fault of the electro-hydrostatic actuator mainly comprises a motor, the hydraulic pump, a servo system and a pressurizing oil tank. A vibration acceleration sensor is arranged on an oil outlet shell of the hydraulic pump or a pressure sensor is arranged in an oil outlet oil loop to monitor the running state of the electro-hydrostatic actuator, so that early fault intelligent diagnosis is carried out.
The invention discloses a method for diagnosing the fault of a hydraulic pump of an electro-hydrostatic actuator, which has the main training route shown in figure 2 and comprises the following steps:
step 1, collecting data of a hydraulic pump normal state and an abnormal state of an electro-hydrostatic actuator
Detecting the vibration of a hydraulic pump by using 2 acceleration sensors ax and ay in the vertical direction of an outlet of a hydraulic pump of an electro-hydrostatic actuator, or detecting the pressure pulsation of the hydraulic pump by using a pressure sensor P at the outlet of the hydraulic pump of the electro-hydrostatic actuator;
and 2, constructing a frequency band energy characteristic vector for the hydraulic pump data set of the electro-hydrostatic actuator based on a wavelet packet decomposition technology, and encoding the fault type of the hydraulic pump of the electro-hydrostatic actuator to be used as an output sample of training.
Before constructing the characteristic vector, the vibration signal or the pressure signal of the hydraulic pump data set of the electro-hydrostatic actuator is proportioned by 4: 1, randomly extracting to obtain a training sample data set and a test data set;
the fault types of the hydraulic pump of the electro-hydrostatic actuator comprise normal hydraulic pump, loose ball head, misalignment of shaft, abrasion of a valve plate, abrasion of a bearing, fatigue of the bearing, fracture of the bearing and deformation of the bearing. During the experiment, the fault codes are 1 if the fault occurs, and the fault which does not occur is 0; for example, ball head loosening fault coding: [0, 1, 0, 0, 0, 0, 0, 0], port plate wear failure coding: [0,0,0,1,0,0,0,0].
And 3, searching model hyperparameter C and delta optimal values of the support vector machine by using the quantum particle group on a hydraulic pump data set of the electro-hydrostatic actuator. The method adopts a quantum particle swarm algorithm to optimize the hyperparameters C and delta of the support vector machine, the optimization flow is shown in figure 3, and the specific optimization process is as follows:
step 3.1, initializing the population size of the quantum particle swarm, the maximum iteration times, and mapping a support vector machine hyperparameter penalty factor C and a kernel parameter delta into a two-dimensional position P of the particle, wherein the iteration times t is 0;
carrying out initialization coding on particle positions P by using quantum probability amplitude:
particle position:
Figure BDA0002248361240000091
wherein, thetaij2 pi × rand, rand being a random number between (0, 1); i is the particle population size; j is belonged to {1,2} and represents a particle dimension, namely the number of parameter types needing to be searched;
step 3.2, the position of the particle of the quantum probability amplitude code is subjected to solution space transformation
Performing solution space transformation on the positions of the quantum probability amplitude encoded particles:
and marking the solution space transformation result of the particle position as X, distinguishing different dimensionality solution space mapping results by an X upper mark, and distinguishing sine and cosine solution space mapping results by an X lower mark.
Then the result of the quantum probability amplitude cosine solution space mapping in the first qubit is:
Figure BDA0002248361240000101
the first qubit quantum probability amplitude sinusoidal solution space mapping result is
Figure BDA0002248361240000102
Wherein, CmaxAnd CminSolving the upper limit and the lower limit of the range for the optimal value of the penalty factor C, wherein subscripts C and s are used for distinguishing the mapping results of cosine and sine, C represents cosine, and s represents sine。
Then the result of the quantum probability amplitude cosine solution space mapping in the second qubit is:
Figure BDA0002248361240000103
then the result of the sinusoidal solution space mapping of the quantum probability amplitude in the second qubit is:
wherein, deltamaxAnd deltamin is the upper and lower limits of the optimal value solving range of the nuclear parameter delta, subscripts c and s are used for distinguishing mapping results of cosine and sine, c represents cosine, and s represents sine.
It can be seen that each qubit corresponds to two solutions to the optimization problem, which can speed up the search efficiency.
3.3, training a comprehensive evaluation formula Q for fault classification of a support vector machine based on a training data sample set characteristic vector of a hydraulic pump vibration signal or a pressure signal of the electro-hydrostatic actuator to serve as a target function, and calculating a target function value of each particle according to the target function;
the comprehensive evaluation formula is as follows:
Figure BDA0002248361240000105
wherein, TP is the number of correctly classified fault classes, FP is the number of incorrectly classified fault classes, and FN is the number of incorrectly classified normal classes.
Step 3.4, updating the new local optimal position P of each particle according to the objective function valueib；
Step 3.5, updating the global optimal position P according to the objective function valueg；
3.6, updating the position of each particle in the population based on the quantum revolving gate;
the local optimal position traversed by the quantum particle at present is a cosine position, that is:
Pib＝(cos(θib1),cos(θib2))
the global optimal position traversed by the quantum population is also a cosine position:
Pg＝(cos(θg1),cos(θg2))
quantum amplitude change of particle:
Δθij(t+1)＝ω×Δθij(t)+c1×Δθbc+c2×Δθgc
where ω is the inertia factor, c1Is a self factor, c2Is a global factor, both are constants, usually taken to be 1.
Figure BDA0002248361240000111
Figure BDA0002248361240000112
Then the quantum revolving door:
Figure BDA0002248361240000113
wherein i is the particle population size; j is belonged to {1,2} and represents a particle dimension, namely the number of parameter types needing to be searched; b and g are used for distinguishing a local optimal value and a global optimal value of the particle, wherein b represents a local optimal value, and g represents a global optimal value; subscript c represents a cosine;
updating the particle probability amplitude coding position based on the quantum revolving gate:
Figure BDA0002248361240000114
wherein i is the particle population size; j ∈ {1,2} represents the particle dimension
After the update, the particle PiTwo new positions of (c):
Pic＝(cos(θi1(t)+Δθi1(t+1)),cos(θi2(t)+Δθi2(t+1)))
Pis＝(sin(θi1(t)+Δθi1(t+1)),sin(θi2(t)+Δθi2(t+1)))
or
The local optimal position currently traversed by the quantum particle is a sinusoidal position, namely:
Pib＝(sin(θib1),sin(θib2))
the global optimal position traversed by the quantum population is also a sinusoidal position:
Pg＝(sin(θg1),sin(θg2))
quantum amplitude change of particle:
Δθij(t+1)＝ω×Δθij(t)+c1×Δθls+c2×Δθgs
where ω is the inertia factor, c1Is a self factor, c2Is a global factor, both are constants, usually taken to be 1.
Figure BDA0002248361240000121
Figure BDA0002248361240000122
Then the quantum revolving door:
Figure BDA0002248361240000123
wherein i is the particle population size; j is belonged to {1,2} and represents a particle dimension, namely the number of parameter types needing to be searched; b and g are used for distinguishing a local optimal value and a global optimal value of the particle, wherein b represents a local optimal value, and g represents a global optimal value; subscript s represents sine;
updating the particle probability amplitude coding position based on the quantum revolving gate:
Figure BDA0002248361240000124
then it is moreAfter all, particle PiTwo new positions of (c):
Pic＝(cos(θi1(t)+Δθi1(t+1)),cos(θi2(t)+Δθi2(t+1)))
Pis＝(sin(θi1(t)+Δθi1(t+1)),sin(θi2(t)+Δθi2(t+1)))
3.7, repeating the steps 3.2-3.7 until the iteration is finished, and obtaining a global optimal position as an optimization result, namely a penalty factor C and a kernel parameter delta of the optimal support vector machine;
and 4, training the quantum particle swarm optimization support vector machine by utilizing the electro-hydrostatic actuator hydraulic pump training sample data set to obtain the electro-hydrostatic actuator hydraulic pump fault diagnosis model.
And 5, performing fault diagnosis on the hydraulic pump of the electro-hydrostatic actuator on the hydraulic pump test set of the electro-hydrostatic actuator by using the trained quantum particle swarm optimization support vector machine model.
Inputting the feature vector data of the electro-hydrostatic actuator hydraulic pump vibration signal or pressure signal test data set into a trained support vector machine model for fault diagnosis, and outputting a fault diagnosis classification result, namely a fault code, of the electro-hydrostatic actuator hydraulic pump. And (4) comparing the fault codes in the step (2) to obtain the fault type of the hydraulic pump of the electro-hydrostatic actuator.

Claims (10)

1. A method for diagnosing the fault of a hydraulic pump of an electro-hydrostatic actuator is characterized by comprising the steps of collecting a vibration signal on a hydraulic pump shell of the electro-hydrostatic actuator or a pressure signal at an outlet of the hydraulic pump, constructing a characteristic vector based on a wavelet packet decomposition technology, and coding the fault type of the hydraulic pump of the electro-hydrostatic actuator to be used as an output sample of training; optimizing model hyperparameters C and delta of a support vector machine by utilizing a quantum particle group algorithm on a vibration signal data set or a pressure signal data set of a hydraulic pump of the electro-hydrostatic actuator to obtain values of the optimal hyperparameters C and delta, bringing the values into the support vector machine model, and training the support vector machine model of the optimal hyperparameters by utilizing feature vectors extracted from a training data set of the hydraulic pump of the electro-hydrostatic actuator and fault codes corresponding to the feature vectors; and finally, performing hydraulic pump fault diagnosis on the electro-hydrostatic actuator on the hydraulic pump test data set.
2. The method for diagnosing the fault of the hydraulic pump of the electro-hydrostatic actuator as claimed in claim 1, wherein the method comprises the steps of:
step 1, collecting vibration signals on a shell or pressure signal data of an outlet of a hydraulic pump of an electro-hydrostatic actuator in a normal state or an abnormal state;
step 2, constructing a frequency band energy characteristic vector for a hydraulic pump data set of the electro-hydrostatic actuator based on a wavelet packet decomposition technology, and coding the fault type of the hydraulic pump of the electro-hydrostatic actuator to be used as an output sample of training;
3, searching model hyperparameter C and delta optimal values of a support vector machine by using quantum particle groups on a hydraulic pump data set of the electro-hydrostatic actuator;
step 4, training the quantum particle swarm optimization support vector machine by utilizing a hydraulic pump training sample data set of the electro-hydrostatic actuator to obtain a hydraulic pump fault diagnosis model of the electro-hydrostatic actuator;
and 5, performing fault diagnosis on the hydraulic pump of the electro-hydrostatic actuator on the hydraulic pump test set of the electro-hydrostatic actuator by using the trained quantum particle swarm optimization support vector machine model.
3. The method for diagnosing the fault of the hydraulic pump of the electro-hydrostatic actuator as claimed in claim 2, wherein in step 1, the vibration signal of the hydraulic pump is detected by using 2 acceleration sensors ax, ay in a direction perpendicular to the outlet of the hydraulic pump of the electro-hydrostatic actuator, or the pressure signal of the hydraulic pump is detected by using the pressure sensor P at the outlet of the hydraulic pump of the electro-hydrostatic actuator.
4. The method for diagnosing the fault of the hydraulic pump of the electro-hydrostatic actuator as claimed in claim 2, wherein in the step 2, before the feature vector is constructed, the vibration signal or the pressure signal of the hydraulic pump data set of the electro-hydrostatic actuator is subjected to proportion 4: 1, randomly extracting to obtain a training sample data set and a test data set;
the fault types of the hydraulic pump of the electro-hydrostatic actuator comprise normal hydraulic pump, loose ball head, misalignment of shaft, abrasion of a valve plate, abrasion of a bearing, fatigue of the bearing, fracture of the bearing and deformation of the bearing.
5. The method for diagnosing the fault of the hydraulic pump of the electro-hydrostatic actuator as claimed in claim 2, wherein in the step 3, the quantum particle swarm algorithm is adopted to optimize the hyperparameters C and delta of the support vector machine, and the specific optimization process is as follows:
step 3.1, initializing the two-dimensional position P of the particle mapped by the population scale of the quantum particle swarm, the maximum iteration times, the support vector machine hyper-parameter penalty factor C and the nuclear parameter delta, wherein the iteration times t is 0;
step 3.2, the position of the particle of the quantum probability amplitude code is subjected to solution space transformation
Performing solution space transformation before calculating the objective function value of the model, wherein the solution space transformation result is the current search result of the solved model hyperparameter;
3.3, training a comprehensive evaluation formula Q for fault classification of a support vector machine based on a training data sample set characteristic vector of a hydraulic pump vibration signal or a pressure signal of the electro-hydrostatic actuator to serve as a target function, and calculating a target function value of each particle according to the target function;
step 3.4, updating the new local optimal position P of each particle according to the objective function valueib；
Step 3.5, updating the global optimal position P according to the objective function valueg；
3.6, updating the position of each particle in the population based on the quantum revolving gate;
and 3.7, repeating the steps 3.2-3.7 until the iteration is finished, and obtaining the global optimal position as an optimization result, namely the penalty factor C and the kernel parameter delta of the optimal support vector machine.
6. The method for diagnosing the hydraulic pump fault of the electro-hydrostatic actuator as claimed in claim 5, wherein in step 3.1, the model hyper-parameters C and δ are quantum-encoded by quantum probability argument, and the method comprises the following steps:
carrying out initialization coding on particle positions P by using quantum probability amplitude:
particle position:
Figure FDA0002248361230000031
wherein, thetaij2 pi × rand, rand being a random number between (0, 1); i is the particle population size; j e 1,2 represents the particle dimension, i.e. the number of parameter classes to be searched.
7. The method for diagnosing the fault of the hydraulic pump of the electro-hydrostatic actuator as claimed in claim 6, wherein in step 3.2, the solution space mapping of the positions of the quantum particle swarm is as follows:
marking the solution space transformation result of the particle position as X, distinguishing different dimensionality solution space mapping results by an X upper mark, and distinguishing sine and cosine solution space mapping results by an X lower mark;
then the result of the quantum probability amplitude cosine solution space mapping in the first qubit is:
Figure FDA0002248361230000032
the first qubit quantum probability amplitude sinusoidal solution space mapping result is
Figure FDA0002248361230000033
Wherein, CmaxAnd CminSolving the upper limit and the lower limit of the range for the optimal value of the penalty factor C, wherein subscripts C and s are used for distinguishing the mapping results of cosine and sine, C represents cosine, and s represents sine;
then the result of the quantum probability amplitude cosine solution space mapping in the second qubit is:
then the result of the sinusoidal solution space mapping of the quantum probability amplitude in the second qubit is:
Figure FDA0002248361230000035
wherein, deltamaxAnd deltaminSolving the upper limit and the lower limit of the range for the optimal value of the kernel parameter delta, wherein subscripts c and s are used for distinguishing mapping results of cosine and sine, c represents cosine, and s represents sine;
two solutions of each quantum bit corresponding to the optimization problem are obtained according to the solution space mapping result, and therefore the search efficiency is improved.
8. The method for diagnosing the fault of the hydraulic pump of the electro-hydrostatic actuator according to claim 5, wherein in step 3.3, a comprehensive evaluation formula Q is provided as a target function for diagnosing the fault of the hydraulic pump of the electro-hydrostatic actuator of the quantum particle swarm optimization support vector machine by comprehensively considering the diagnosis precision and the recall rate of a fault diagnosis model, and the comprehensive evaluation formula is specifically as follows:
Figure FDA0002248361230000041
wherein, TP is the number of correctly classified fault classes, FP is the number of incorrectly classified fault classes, and FN is the number of incorrectly classified normal classes.
9. The method for diagnosing the hydraulic pump fault of the electro-hydrostatic actuator as claimed in claim 5, wherein in steps 3.4, 3.5 and 3.6, the quantum probability argument is updated through the quantum revolving gate based on the local optimal position and the global optimal position of the quantum particle swarm, and when the local optimal position and the global optimal position are both cosine positions, the updating formula is as follows:
the local optimal position traversed by the quantum particle at present is a cosine position, that is:
Pib＝(cos(θib1),cos(θib2))
the global optimal position traversed by the quantum population is also a cosine position:
Pg＝(cos(θg1),cos(θg2))
quantum amplitude change of particle:
Δθij(t+1)＝ω×Δθij(t)+c1×Δθbc+c2×Δθgc
where ω is the inertia factor, c1Is a self factor, c2Is a global factor, both are constants;
Figure FDA0002248361230000042
then the quantum revolving door:
Figure FDA0002248361230000044
wherein i is the particle population size; j is belonged to {1,2} and represents a particle dimension, namely the number of parameter types needing to be searched; b and g are used for distinguishing a local optimal value and a global optimal value of the particle, wherein b represents a local optimal value, and g represents a global optimal value; subscript c represents a cosine;
updating the particle probability amplitude coding position based on the quantum revolving gate:
wherein i is the particle population size; j ∈ {1,2} represents the particle dimension;
after the update, the particle PiTwo new positions of (c):
Pic＝(cos(θi1(t)+Δθi1(t+1)),cos(θi2(t)+Δθi2(t+1)))
Pis＝(sin(θi1(t)+Δθi1(t+1)),sin(θi2(t)+Δθi2(t+1)))。
10. the method for diagnosing the hydraulic pump fault of the electro-hydrostatic actuator as claimed in claim 5, wherein in steps 3.4, 3.5 and 3.6, the quantum probability argument is updated through the quantum rotary gate based on the local optimal position and the global optimal position of the quantum particle swarm, and when the local optimal position and the global optimal position are sinusoidal positions, the formula is updated as follows:
the local optimal position currently traversed by the quantum particle is a sinusoidal position, namely:
Pib＝(sin(θib1),sin(θib2))
the global optimal position traversed by the quantum population is also a sinusoidal position:
Pg＝(sin(θg1),sin(θg2))
quantum amplitude change of particle:
Δθij(t+1)＝ω×Δθij(t)+c1×Δθbs+c2×Δθgs
where ω is the inertia factor, c1Is a self factor, c2Is a global factor, both are constants;
Figure FDA0002248361230000052
then the quantum revolving door:
Figure FDA0002248361230000054
wherein i is the particle population size; j is belonged to {1,2} and represents a particle dimension, namely the number of parameter types needing to be searched; b and g are used for distinguishing a local optimal value and a global optimal value of the particle, wherein b represents a local optimal value, and g represents a global optimal value; subscript s represents sine;
updating the particle probability amplitude coding position based on the quantum revolving gate:
Figure FDA0002248361230000061
after the update, the particle PiTwo new positions of (c):
Pic＝(cos(θi1(t)+Δθi1(t+1)),cos(θi2(t)+Δθi2(t+1)))
Pis＝(sin(θi1(t)+Δθi1(t+1)),sin(θi2(t)+Δθi2(t+1)))。
CN201911024933.3A 2019-10-25 2019-10-25 Fault diagnosis method for hydraulic pump of electro-hydrostatic actuator Pending CN110837851A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911024933.3A CN110837851A (en) 2019-10-25 2019-10-25 Fault diagnosis method for hydraulic pump of electro-hydrostatic actuator

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911024933.3A CN110837851A (en) 2019-10-25 2019-10-25 Fault diagnosis method for hydraulic pump of electro-hydrostatic actuator

Publications (1)

Publication Number Publication Date
CN110837851A true CN110837851A (en) 2020-02-25

Family

ID=69575647

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911024933.3A Pending CN110837851A (en) 2019-10-25 2019-10-25 Fault diagnosis method for hydraulic pump of electro-hydrostatic actuator

Country Status (1)

Country Link
CN (1) CN110837851A (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111472827A (en) * 2020-04-10 2020-07-31 太原理工大学 Intelligent decision-making method for hydraulic support group following propulsion behavior
CN111722174A (en) * 2020-05-31 2020-09-29 宁夏隆基宁光仪表股份有限公司 System and method for realizing electric energy meter abnormity diagnosis by applying quantum particle group algorithm
CN113191232A (en) * 2021-04-21 2021-07-30 西安交通大学 Electro-hydrostatic actuator fault identification method based on multi-mode homologous features and XGboost model
CN113807524A (en) * 2021-08-12 2021-12-17 上海工程技术大学 Rolling bearing fault diagnosis method for optimizing SVM (support vector machine) by quantum differential evolution algorithm
CN114098757A (en) * 2021-11-12 2022-03-01 南京海量物联科技有限公司 ECG signal monitoring method based on quantum particle swarm optimization

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107621782A (en) * 2017-05-08 2018-01-23 安徽理工大学 A kind of method for diagnosing faults of grid bipolar transistor (IGBT) module
CN107884706A (en) * 2017-11-09 2018-04-06 合肥工业大学 The analog-circuit fault diagnosis method approached based on vector value canonical kernel function
CN108414923A (en) * 2018-02-05 2018-08-17 武汉大学 A kind of analog-circuit fault diagnosis method based on the extraction of depth confidence network characterization

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107621782A (en) * 2017-05-08 2018-01-23 安徽理工大学 A kind of method for diagnosing faults of grid bipolar transistor (IGBT) module
CN107884706A (en) * 2017-11-09 2018-04-06 合肥工业大学 The analog-circuit fault diagnosis method approached based on vector value canonical kernel function
WO2019090878A1 (en) * 2017-11-09 2019-05-16 合肥工业大学 Analog circuit fault diagnosis method based on vector-valued regularized kernel function approximation
CN108414923A (en) * 2018-02-05 2018-08-17 武汉大学 A kind of analog-circuit fault diagnosis method based on the extraction of depth confidence network characterization

Non-Patent Citations (5)

* Cited by examiner, † Cited by third party
Title
XIUZHEN GUO等: ""A Novel Feature Extraction Approach Using Window Function Capturing and QPSO-SVM for Enhancing Electronic Nose Performance"", 《SENSORS》 *
李士勇等: ""求解连续空间优化问题的量子粒子群算法"", 《量子电子学报》 *
杨光春等: ""基于QPSO-SVM的轴承故障诊断方法"", 《机械传动》 *
贾茂力等: ""基于一种QPSO算法的阵列天线方向图综合"", 《电子设计工程》 *
金剑等: ""基于小波包和PSO-SVM的柴油机故障诊断"", 《组合机床与自动化加工技术》 *

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111472827A (en) * 2020-04-10 2020-07-31 太原理工大学 Intelligent decision-making method for hydraulic support group following propulsion behavior
CN111722174A (en) * 2020-05-31 2020-09-29 宁夏隆基宁光仪表股份有限公司 System and method for realizing electric energy meter abnormity diagnosis by applying quantum particle group algorithm
CN113191232A (en) * 2021-04-21 2021-07-30 西安交通大学 Electro-hydrostatic actuator fault identification method based on multi-mode homologous features and XGboost model
CN113191232B (en) * 2021-04-21 2023-04-18 西安交通大学 Electro-hydrostatic actuator fault identification method based on multi-mode homologous features and XGboost model
CN113807524A (en) * 2021-08-12 2021-12-17 上海工程技术大学 Rolling bearing fault diagnosis method for optimizing SVM (support vector machine) by quantum differential evolution algorithm
CN113807524B (en) * 2021-08-12 2023-07-25 上海工程技术大学 Rolling bearing fault diagnosis method for optimizing SVM by quantum differential evolution algorithm
CN114098757A (en) * 2021-11-12 2022-03-01 南京海量物联科技有限公司 ECG signal monitoring method based on quantum particle swarm optimization
CN114098757B (en) * 2021-11-12 2024-02-09 南京海量物联科技有限公司 ECG signal monitoring method based on quantum particle swarm optimization

Similar Documents

Publication Publication Date Title
CN110837851A (en) Fault diagnosis method for hydraulic pump of electro-hydrostatic actuator
CN106980822B (en) A kind of rotary machinery fault diagnosis method based on selective ensemble study
Wang et al. Modified multiscale weighted permutation entropy and optimized support vector machine method for rolling bearing fault diagnosis with complex signals
CN111651937A (en) Method for diagnosing similar self-adaptive bearing fault under variable working conditions
CN113188794B (en) Gearbox fault diagnosis method and device based on improved PSO-BP neural network
CN111459144A (en) Airplane flight control system fault prediction method based on deep cycle neural network
CN107247231A (en) A kind of aerogenerator fault signature extracting method based on OBLGWO DBN models
CN110046409B (en) ResNet-based steam turbine component health state evaluation method
Zhang et al. A novel data-driven method based on sample reliability assessment and improved CNN for machinery fault diagnosis with non-ideal data
CN109298633A (en) Chemical production process fault monitoring method based on adaptive piecemeal Non-negative Matrix Factorization
CN115438714A (en) Clustering federal learning driven mechanical fault diagnosis method, device and medium
Liu et al. Automatically designing network-based deep transfer learning architectures based on genetic algorithm for in-situ tool condition monitoring
CN104677629A (en) Fault detection method for vehicle transmission
Liu et al. Prediction study of the heavy vehicle driving state based on digital twin model
Zheng et al. An unsupervised transfer learning method based on SOCNN and FBNN and its application on bearing fault diagnosis
Yafei et al. Fault Diagnosis of Axial Piston Pump Based on Extreme‐Point Symmetric Mode Decomposition and Random Forests
Xu et al. A new diagnosis method with few-shot learning based on a class-rebalance strategy for scarce faults in industrial processes
CN113743010A (en) Rolling bearing running state evaluation method based on EEMD energy entropy
Zhao et al. A capsnet-based fault diagnosis method for a digital twin of a wind turbine gearbox
Wang et al. Gearbox fault diagnosis based on a novel hybrid feature reduction method
Zhang et al. Review of remaining useful life prediction using support vector machine for engineering assets
Li et al. Class imbalanced fault diagnosis via combining K-means clustering algorithm with generative adversarial networks
CN110762081A (en) Large hydraulic machine fault diagnosis method and device based on HSMM-SVM model
CN115730255A (en) Motor fault diagnosis and analysis method based on transfer learning and multi-source information fusion
He et al. Probabilistic model based algorithms for prognostics

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

Application publication date: 20200225

RJ01 Rejection of invention patent application after publication