CN106503288B - Interconnection hydraulic cylinder mechanical property prediction method based on Support Vector Machines for Regression - Google Patents
Interconnection hydraulic cylinder mechanical property prediction method based on Support Vector Machines for Regression Download PDFInfo
- Publication number
- CN106503288B CN106503288B CN201610828286.1A CN201610828286A CN106503288B CN 106503288 B CN106503288 B CN 106503288B CN 201610828286 A CN201610828286 A CN 201610828286A CN 106503288 B CN106503288 B CN 106503288B
- Authority
- CN
- China
- Prior art keywords
- hydraulic cylinder
- regression
- interconnection
- support vector
- vector machines
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/06—Power analysis or power optimisation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Hardware Design (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Investigating Strength Of Materials By Application Of Mechanical Stress (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The present invention relates to a kind of interconnection hydraulic cylinder mechanical properties prediction algorithm based on Support Vector Machines for Regression includes the following steps: that (1) develops interconnection hydraulic cylinder device, and carries out corresponding force performance test, and the response obtained under different operating conditions exports；(2) theory analysis is carried out to interconnection hydraulic cylinder mechanics output experimental data, establishes the interconnection hydraulic cylinder mechanical properties prediction model based on Support Vector Machines for Regression；(3) training set test set of the experimental data as prediction model is chosen；(4) corresponding kernel function and initialization model parameter are selected；(5) using cross validation method to Model Parameter Optimization；(6) interconnection hydraulic cylinder mechanics being carried out using prediction model and predicting and carry out performance evaluation, the present invention can effectively predict the mechanical property of similar Hydraulic coupling element.
Description
Technical field
The present invention relates to a kind of mechanical property prediction methods for interconnecting hydraulic cylinder, refer in particular to a kind of based on regression supporting vector
The interconnection hydraulic cylinder mechanical properties prediction algorithm of machine.
Background technique
In recent years, traffic accident declaration of an accident in the whole world is more frequent, and the data that NHTSA in 2004 is provided show nearly three points of the U.S.
One of traffic accident accident origin turn on one's side in vehicle body, be especially embodied on lorry, SUV and car grade height mass center vehicle, and be suspended in
Guarantee to play key effect in terms of vehicle safety.For vehicle rollover problem, a large amount of scholars have carried out corresponding research, and are directed to and set
Different suspension frame structures are counted, wherein hydraulic interconnection suspension (Hydraulic interconnected suspension, HIS) system
It unites due to its brilliant control stability and good ride performance energy, receives domestic and foreign scholars' extensive concern.
Foreign scholar Hawley proposes oil pipe interconnection damper earliest, and elaborates interconnection that may be present between take turns more
Mode.Moulton is based on hydraulic interconnection suspension comparative test, establishes the simple two degrees of freedom vehicle differential equation.Liu is proposed
A kind of anti-side is inclined hydraulic interconnection suspension, and has carried out theory analysis for fluid compression in turbulent valve damage and cylinder, but lack
To compression, the research of fluid inertia and hydroelasticity wave effect in transfer route loss, fluid hose, while lacking test and testing
Card.Network synthesis theory is applied to the passive interconnection suspension system of tradition by Mace, proposes a kind of mechanical admittance matrix, and be directed to
Three kinds of damping structures carry out theory analysis.Smith is then based on network synthesis theory and proposes rigidity and damping decouplingization, but right
HIS dynamics research is insufficient, therefore network synthesis theory rationality in application can not determine.2005, Smith and Walker were comprehensive
Existing research achievement has carried out theoretical definition to interconnection suspension, it was confirmed that interconnection hydraulic cylinder can effectively promote the anti-inclination of suspension
Can, and theoretical research and the derivation of equation have been carried out to interconnection hydraulic cylinder mechanical model on this basis.
However the non-linear research model of interconnection hydraulic cylinder is relatively simple at present, can not react completely under actual condition mutually
Join hydraulic cylinder mechanical characteristic.By interconnecting the bench test of hydraulic cylinder mechanical property, further investigation non-linear factor is hydraulic to interconnecting
Cylinder Effect on Mechanical Properties mechanism, and support vector machines (support vector machine, SVM) prediction model is introduced to interconnection
Hydraulic cylinder mechanical property is predicted.Simultaneously in order to solve traditional SVM to data regression fit precision problem, it is insensitive to introduce ε
Loss function, building interconnection hydraulic cylinder Support Vector Machines for Regression prediction model (support vector machine for
Regression, SVR), and model built is verified using bench test.
Summary of the invention
It is an object of the invention to be directed to existing interconnection hydraulic cylinder nonlinear model Construct question, propose a kind of based on recurrence
The interconnection hydraulic cylinder mechanical properties prediction algorithm of type support vector machines, to realize the hydraulic rigid good mechanical properties prediction of interconnection.
For achieving the above object, the technical scheme adopted by the invention is as follows: the interconnection based on Support Vector Machines for Regression
Hydraulic cylinder mechanical properties prediction algorithm, which comprises the steps of: (1) select interconnection hydraulic cylinder, and carry out interconnection liquid
Cylinder pressure mechanical property test obtains the mechanical response under different operating conditions；(2) to interconnection hydraulic cylinder mechanical response experimental data into
Row theory analysis determines and uses the interconnection hydraulic cylinder mechanical properties prediction model based on Support Vector Machines for Regression, and selects back
Return the kernel function and initialization model parameter of type support vector machines；(3) experimental data is chosen as interconnection hydraulic cylinder mechanical property
The training set of prediction model；(4) interconnection hydraulic cylinder mechanical properties prediction model parameter is optimized using cross validation method；
(5) interconnection hydraulic cylinder mechanics is carried out using interconnection hydraulic cylinder mechanical properties prediction model to predict and carry out performance evaluation.
Further, in step (1), hydraulic cylinder mechanical property test is interconnected as single cylinder test, i.e., will interconnect hydraulic cylinder
One of hydraulic cylinder carries out mechanical test on actuating vibration table, another is unloaded, and the exciting input to interconnection hydraulic cylinder is difference
The sinusoidal input of frequency various amplitude.
Further, in step (2), the interconnection hydraulic cylinder mechanical properties prediction model based on Support Vector Machines for Regression is adopted
It is predicted with small sample, includes input layer, hidden layer and output layer, the input layer is interconnection hydraulic cylinder under multiple transient times
Displacement, velocity and acceleration, the output layer be interconnection hydraulic cylinder power output, the hidden layer be based on support vector machines
Linear regression function, including linear insensitive loss function of ε and Radial basis kernel function.
Further, in step (3), the selection of experimental data uses uniform sampling, and each sampling period is 20Hz, each
120 groups of group of data points of working condition acquiring are at training set；Training set includes training sample and test sample, wherein randomly selecting is more than instruction
Practice the data group of collection 2/3 as training sample, remaining data group is that test sample carries out precision of prediction judgement.
Further, in step (4), the process that interconnection hydraulic cylinder mechanical properties prediction model parameter optimizes includes: A
Determine Support Vector Machines for Regression parameter, the Support Vector Machines for Regression parameter includes penalty factor, kernel function variance g and
Linear insensitive loss function of ε；B averagely chooses penalty factor and kernel function variance as initial data, and is divided into k group (k is
Natural number)；C determines the prediction error of cross validation algorithm, calculates the output valve of each training sample, obtain each output valve with
Error between desired value obtains optimal penalty factor and kernel function variance.
Further, in step (5), using the mean square error E and coefficient of determination R of test set2It supports based on regression
The precision of prediction of vector machine forecast model is evaluated.
Beneficial effects of the present invention: it since multivariate features are presented by non-linear effects in interconnection hydraulic cylinder mechanical characteristic, utilizes
The mechanical properties prediction model of Support Vector Machines for Regression building is trained test data, can be to interconnection hydraulic cylinder output
Power carries out Accurate Prediction, and relatively supportive vector machine precision of prediction is promoted obviously, while to similar Hydraulic coupling element
Mechanical properties prediction provides a kind of Research Thinking.
Detailed description of the invention
Fig. 1 is the interconnection hydraulic cylinder mechanical properties prediction algorithm flow chart based on Support Vector Machines for Regression.
Fig. 2 is interconnection Hydraulic Cylinder Model figure.
Hydraulic cylinder mechanical properties prediction figure when Fig. 3 is 0.5Hz.
Hydraulic cylinder mechanical properties prediction figure when Fig. 4 is 5Hz.
Hydraulic cylinder mechanical properties prediction figure when Fig. 5 is 9Hz.
In figure: 1 is interconnection hydraulic cylinder two-end-point, and 2 be piston rod, and 3 be hydraulic cylinder, and 4 be interconnecting pipes, and 5 be gasbag-type
Accumulator.
Specific embodiment
The invention patent is described further with reference to the accompanying drawings and examples.
A kind of interconnection hydraulic cylinder mechanical properties prediction algorithm based on Support Vector Machines for Regression of the invention, flow chart
As shown in Figure 1, the specific steps are as follows:
(1) the interconnection hydraulic cylinder device developed is selected, as shown in Fig. 2, 1 being wherein interconnection hydraulic cylinder two-end-point, 2 be work
Stopper rod, 3 be hydraulic cylinder, and 4 be interconnecting pipes, and 5 be bladder accumulator.And interconnection hydraulic cylinder mechanical property test is carried out, it obtains
Take the mechanical response under different operating conditions；
In implementation, in the test of hydraulic interconnection suspension, using sinusoidal excitation signal as inputting, cylinder force is taken to make
For output, test frequency takes 0.1,0.5,1,3,5,7,9,11,13,15Hz, wherein and 0.1H~5Hz takes amplitude for 10mm, 6~
It is 5mm that 15Hz, which takes amplitude,.In test, force signal can be acquired in real time by the force sensor signals that excitation head carries and is stored to control
Platform processed.
(2) to interconnection hydraulic cylinder mechanical response experimental data carry out theory analysis, determine using based on regression support to
The interconnection hydraulic cylinder mechanical properties prediction model of amount machine, and select the kernel function and initialization model ginseng of Support Vector Machines for Regression
Number；Firstly, establishing the training sample set { (xi, yi), i=1,2 ... n } containing n training sample, wherein xi (xi ∈ Rd) be
The input column vector of i-th of training sample,Yi is the output valve of corresponding xi, wherein yi ∈ R.Its
In, RdRefer to the power operation collection of real number.
In order to simplify statistical model calculation amount, introduces the higher Radial basis kernel function of model prediction accuracy and replace tradition system
Meter learns model and clicks operation, kernel function are as follows:
Secondly it introduces linear insensitive loss function of ε and carries out error judgement:
Wherein regression function f (x) expression formula are as follows:
Slack variable and penalty factor are introduced into regression function, then w and b may be expressed as: in former regression function
For the magnitude differences problem of variate-value each in sample, data are normalized.Consider penalty factor
And kernel function variance g is affected to regression model performance and training samples number is few, therefore utilizes cross validation method pair
Penalty factor and kernel function variance g carry out optimizing, are solved, and the value for obtaining penalty factor is 0.5, the side in kernel function
Poor g value is 0.6.
After regression model foundation, training sample set is inputted, hydraulic cylinder certain period in time series is interconnected with left end
Displacement input, speed input be used as input sample, using left end interconnection hydraulic cylinder two-end-point force signal as export sample,
The sample input of middle interconnection hydraulic cylinder is sinusoidal mechanism excitation input, to improve precision of prediction, 120 groups of data of each working condition acquiring
Point carries out prediction output to the mechanical property of interconnection hydraulic cylinder at different frequencies.
(3) training set of the experimental data as prediction model is chosen；
(4) using cross validation method to Model Parameter Optimization；
For the magnitude differences problem of variate-value each in sample, data are normalized.Consider penalty factor
And kernel function variance g is affected to regression model performance and training samples number is few, therefore utilizes cross validation method pair
Penalty factor and kernel function variance g carry out optimizing, Optimizing Flow are as follows: 1) determine (punishment of Support Vector Machines for Regression parameter
The factor and kernel function variance) threshold value；2) penalty factor and kernel function are averagely chosen as initial data, and is divided into k group；3) really
The prediction error for determining cross validation algorithm calculates the output valve of each training sample, obtains between each output valve and desired value
Error obtains optimal penalty factor and kernel function variance.It is solved, the value for obtaining penalty factor is 0.5, in kernel function
Variance g value be 0.6.
(5) interconnection hydraulic cylinder mechanics is carried out using prediction model to predict and carry out performance evaluation
It introduces mean square error E and coefficient of determination R2 to judge the SVR forecast of regression model effect established, express
Formula are as follows:
Optimized Support Vector Machines for Regression interconnects hydraulic cylinder mechanical properties prediction result as shown in Fig. 3 Fig. 4 and Fig. 5,
The relatively traditional supporting vector machine model precision of prediction of Support Vector Machines for Regression prediction model is promoted obvious as the result is shown.Wherein
When excited frequency is 0.5Hz, the mean square error E and coefficient of determination R2 of Support Vector Machines for Regression are respectively 0.00146 He
99.38%, opposite support vector machines 0.004714 and 96.29%, precision of prediction is promoted obviously, and wherein mean square error compares the range of decrease
It is 69.03%；When excited frequency is 5Hz, the mean square error of SVR is reduced to 0.000873 from the 0.003673 of SVM, the range of decrease
76.23%, and the coefficient of determination is promoted from 96.89% to 99.48%；When excited frequency is 9Hz, the mean square error of SVR is compared
The range of decrease 63.18%, the coefficient of determination are promoted to 98.93% from 96.56%.
Multivariate features are presented by non-linear effects in interconnection hydraulic cylinder mechanical characteristic in summary, utilize regression supporting vector
The mechanical properties prediction model that mechanism is built is trained test data, can carry out to interconnection hydraulic cylinder power output accurate pre-
It surveys, and relatively supportive vector machine precision of prediction is promoted obviously, while to the mechanical properties prediction to similar Hydraulic coupling element
Provide a kind of Research Thinking.
Claims (5)
1. the interconnection hydraulic cylinder mechanical property prediction method based on Support Vector Machines for Regression, which is characterized in that including walking as follows
It is rapid:
(1) interconnection hydraulic cylinder is selected, and carries out interconnection hydraulic cylinder mechanical property test, obtains the mechanical response under different operating conditions；
(2) theory analysis is carried out to the experimental data of interconnection hydraulic cylinder mechanical response, determined using based on regression supporting vector
The interconnection hydraulic cylinder mechanical properties prediction model of machine, and select the kernel function and initialization model ginseng of Support Vector Machines for Regression
Number；
(3) training set of the experimental data as interconnection hydraulic cylinder mechanical properties prediction model is chosen；
(4) interconnection hydraulic cylinder mechanical properties prediction model parameter is optimized using cross validation method；Process includes:
A determines Support Vector Machines for Regression parameter, and the Support Vector Machines for Regression parameter includes penalty factor, kernel function side
Poor g and linear insensitive loss function of ε；
B averagely chooses penalty factor and kernel function variance as initial data, and is divided into k group, and k is natural number；
C determines the prediction error of cross validation algorithm, calculates the output valve of each training sample, obtains each output valve and expectation
Error between value obtains optimal penalty factor and kernel function variance；
(5) interconnection hydraulic cylinder mechanics is carried out using interconnection hydraulic cylinder mechanical properties prediction model to predict and carry out performance evaluation.
2. the interconnection hydraulic cylinder mechanical property prediction method based on Support Vector Machines for Regression according to claim 1, special
Sign is, in step (1), interconnects hydraulic cylinder mechanical property test as single cylinder test, i.e., will interconnect one of liquid of hydraulic cylinder
Cylinder pressure carries out mechanical test on actuating vibration table, another is unloaded, and the exciting input to interconnection hydraulic cylinder is that different frequency difference is shaken
The sinusoidal input of width.
3. the interconnection hydraulic cylinder mechanical property prediction method according to claim 2 based on Support Vector Machines for Regression,
It is characterized in that, in step (2), the interconnection hydraulic cylinder mechanical properties prediction model based on Support Vector Machines for Regression uses small sample
Prediction includes input layer, hidden layer and output layer, and the input layer is the displacement for interconnecting hydraulic cylinder under multiple transient times,
Velocity and acceleration, the output layer are interconnection hydraulic cylinder power output, and the hidden layer is based on Support Vector Machines for Regression
Linear regression function, including linear insensitive loss function of ε and Radial basis kernel function.
4. the interconnection hydraulic cylinder mechanical property prediction method according to claim 3 based on Support Vector Machines for Regression,
It is characterized in that, in step (3), the selection of experimental data uses uniform sampling, and each sampling period is 20Hz, each working condition acquiring
120 groups of group of data points are at training set；Training set includes training sample and test sample, wherein randomly selecting more than training set 2/3
Data group as training sample, remaining data group is that test sample carries out precision of prediction judgement.
5. the interconnection hydraulic cylinder mechanical property prediction method according to claim 1 based on Support Vector Machines for Regression,
It is characterized in that, in step (5), using the mean square error E and coefficient of determination R of test set2To pre- based on Support Vector Machines for Regression
The precision of prediction for surveying model is evaluated.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610828286.1A CN106503288B (en) | 2016-09-18 | 2016-09-18 | Interconnection hydraulic cylinder mechanical property prediction method based on Support Vector Machines for Regression |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610828286.1A CN106503288B (en) | 2016-09-18 | 2016-09-18 | Interconnection hydraulic cylinder mechanical property prediction method based on Support Vector Machines for Regression |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106503288A CN106503288A (en) | 2017-03-15 |
| CN106503288B true CN106503288B (en) | 2019-06-28 |
Family
ID=58291430
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610828286.1A Active CN106503288B (en) | 2016-09-18 | 2016-09-18 | Interconnection hydraulic cylinder mechanical property prediction method based on Support Vector Machines for Regression |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106503288B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107273615B (en) * | 2017-06-15 | 2021-06-04 | 东华大学 | Ultra-wideband microwave humidity detection method based on machine learning |
| CN109774399B (en) * | 2019-01-15 | 2021-12-14 | 南昌大学 | Semi-active control method for hydraulic interconnected suspension for identifying pavement excitation frequency range |
| CN110110423B (en) * | 2019-04-29 | 2022-11-01 | 西安理工大学 | Model selection method of rubber shock absorber |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102608285A (en) * | 2012-02-21 | 2012-07-25 | 南京工业大学 | Prediction method for explosion characteristics of organic mixture based on support vector machine |
| CN103258234A (en) * | 2013-05-02 | 2013-08-21 | 江苏大学 | Method for predicting mechanical properties of inertial container based on particle swarm optimization BP neural network |
| CN104392119A (en) * | 2014-11-13 | 2015-03-04 | 成都新核中创信息科技有限公司 | Multiphase support vector regression-based seismic wave crest and trough modeling method |
-
2016
- 2016-09-18 CN CN201610828286.1A patent/CN106503288B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102608285A (en) * | 2012-02-21 | 2012-07-25 | 南京工业大学 | Prediction method for explosion characteristics of organic mixture based on support vector machine |
| CN103258234A (en) * | 2013-05-02 | 2013-08-21 | 江苏大学 | Method for predicting mechanical properties of inertial container based on particle swarm optimization BP neural network |
| CN104392119A (en) * | 2014-11-13 | 2015-03-04 | 成都新核中创信息科技有限公司 | Multiphase support vector regression-based seismic wave crest and trough modeling method |
Non-Patent Citations (1)
| Title |
|---|
| QGA-SVM 在弓网滑动电接触下的最优载荷研究;任志玲等;《高压电路》;20150416;全文 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106503288A (en) | 2017-03-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109855823B (en) | Method for identifying damage of bridge structure by using test vehicle | |
| Law et al. | Moving force identification: optimal state estimation approach | |
| CN106503288B (en) | Interconnection hydraulic cylinder mechanical property prediction method based on Support Vector Machines for Regression | |
| CN102519651B (en) | Method for determining basic frequency of stay cable when testing cable tension of cable stayed bridge by using vibration method | |
| CN103852269B (en) | Bullet train runs kinetic parameter detection method | |
| CN104457681B (en) | Girder structure dynamic deflection monitoring method based on strain mode | |
| CN102425148B (en) | Rapid sub-grade settlement predicting method based on static sounding and BP (Back Propagation) neural network | |
| CN106323451B (en) | The method and device thereof of displacement signal are obtained by acceleration signals | |
| Zhang et al. | A flutter prediction method with low cost and low risk from test data | |
| CN106980744A (en) | Control stability and ride comfort cooperative optimization method in vehicle dynamics simulation analysis | |
| CN109398020B (en) | Nonlinear model-based prediction control method for vehicle hydro-electric coupling type ISD suspension | |
| CN108844702A (en) | The measuring method of Bridge Influence Line when vehicle at the uniform velocity passes through | |
| CN103557980B (en) | Method for precisely testing external prestressing tendon tensioning force | |
| Ni et al. | Traffic-induced bridge displacement reconstruction using a physics-informed convolutional neural network | |
| He et al. | FEM free damage detection of beam structures using the deflections estimated by modal flexibility matrix | |
| CN105351424A (en) | Optimizing and designing system for dynamic property of fluidic-resistance suspension of automobile powertrain | |
| Yong et al. | Hydraulic–mechanical coupling modeling by bond graph for impact system of a high frequency rock drill drifter with sleeve distributor | |
| CN114065373A (en) | Automobile control arm rack endurance test method, device and equipment | |
| CN104455157A (en) | Obtaining method of car seat suspension hydraulic buffer nonlinear speed characteristic parameter | |
| CN117272815A (en) | Multi-target performance calculation method for vehicle suspension vibration reduction system with knowledge and data fusion | |
| CN104807661A (en) | High-level and high-rise structure dynamic detection bearing capacity evaluating method | |
| CN107144384A (en) | A kind of tire produces the measuring and calculation method of impulsive force when passing through faulting of slab ends | |
| CN104182592B (en) | The discrimination method of equivalent three mass of human body of seat human body vibrating model | |
| CN115524086A (en) | Statistical moment-curvature beam type bridge damage identification method based on axle coupling vibration | |
| CN110321593A (en) | Bridge dynamic deflection vibration shape Matrix Construction Method based on accumulation modal mass participation rate |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |