CN103197668A - Data drive type control performance detection device and method - Google Patents
Data drive type control performance detection device and method Download PDFInfo
- Publication number
- CN103197668A CN103197668A CN2013101163656A CN201310116365A CN103197668A CN 103197668 A CN103197668 A CN 103197668A CN 2013101163656 A CN2013101163656 A CN 2013101163656A CN 201310116365 A CN201310116365 A CN 201310116365A CN 103197668 A CN103197668 A CN 103197668A
- Authority
- CN
- China
- Prior art keywords
- performance
- response
- analyser
- frequency response
- frequency
- 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.)
- Granted
Links
Images
Landscapes
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
- Feedback Control In General (AREA)
Abstract
The invention relates to the technical field of control system performance detection, in particular to a data drive type control performance detection device and a data drive type control performance detection method. The device comprises a data preprocessor, a frequency demultiplication processor, a frequency classifier and a performance evaluation device. According to the data drive type control performance detection device and the data drive type control performance detection method, modeling to a system is not needed, calculation is simple and efficiency of an algorithm is high, daily operation numbers in a control system are directly used for calculating, an experiment is not needed, any influence to normal operation of the system cannot be produced, non-invasive and system-friendly effects are achieved, and convenient application in an actual control system is achieved. In addition, detection results of the device are normalized, and the results are easy to understand.
Description
Technical field
The invention belongs to the automatic control technology field, relate in particular to a kind of control performance pick-up unit and method of data driven type.
Background technology
The performance of control system has direct influence to security and the economy of industrial processes.But in present industrial processes, the shared ratio of the bad control loop of performance is very high, has 66%~80% controller can not reach the performance that it should reach approximately.Therefore, the benchmark problem of research industrial control system has urgency, and relevant achievement will have broad application prospects.Current for control system at random performance evaluation mainly be the performance evaluation of model-driven, but this class methods complexity is difficult to use in practice.Method of evaluating performance based on data need not too much procedural knowledge owing to it, so its easier application in the performance evaluation of actual production process.Simultaneously, modern industrial processes all establish perfect production information collection and management system, therefrom can obtain a large amount of production run real time datas, realize having practical significance for the evaluation of control loop performance based on these data.
Summary of the invention
Purpose of the present invention is subject to deficiencies such as interference at the system control performance detection of complex of mentioning in the above-mentioned background technology based on model, has proposed a kind of control system device for detecting performance and method of data-driven.
A kind of control system device for detecting performance of data-driven, this device comprise data pre-processor, frequency division processor, frequency categorization device and performance evaluation device, wherein,
Described data pre-processor, frequency division processor, frequency categorization device and performance evaluation device are connected successively;
Described performance evaluation device is made up of high frequency response performance analyser, intermediate frequency response performance analyser, LF-response performance analyser and performance weighter, wherein, high frequency response performance analyser, intermediate frequency response performance analyser and LF-response performance analyser link to each other with the performance weighter with the frequency categorization device respectively.
A kind of control system method for testing performance of data-driven, this method may further comprise the steps:
Step 1: data pre-processor utilizes wavelet packet to decompose to system's output sampling of control system, obtains one group of narrowband response;
Step 2: the frequency division processor utilizes empirical mode decomposition method that each narrowband response is decomposed into one group of natural mode of vibration component;
Step 3: the frequency categorization device utilizes clustering method that the natural mode of vibration component is divided three classes: high frequency response, intermediate frequency response, LF-response;
Step 4: described high frequency response performance analyser utilizes high frequency response to pass through the randomness performance of minimum variance performance Index Calculation system; Described intermediate frequency response performance analyser utilizes intermediate frequency response by the determinacy performance of nondimensional determinacy performance Index Calculation system; Described LF-response performance analyser utilizes LF-response to pass through the steady-state error of its mean value computation system;
The high frequency response performance analyser is formed the randomness response output of system with the high fdrequency component in the cluster analysis, utilizes least-squares algorithm to come the minimum variance performance index of computing system, is used for the randomness performance of evaluation system;
The intermediate frequency response performance analyser at first, is chosen cluster and is obtained intermediate frequency component, estimates system transter G with the input/output relation between setting value and the intermediate frequency response again
Cl(s), and with its equivalence delay estimated value as system delay d; Secondly, by to giving G
Cl(s) do step disturbance and calculate the adjusting time T; At last, calculate nondimensional determinacy performance index
LF-response performance analyser, low frequency performance mainly corresponding to the steady-state error of control system, by the low frequency component that cluster obtains, ask the average of calculating low frequency component as steady-state error ε to it, and what steady-state error ε reacted is the control accuracy of system;
Step 5: randomness performance, determinacy performance and steady-state error performance are carried out comprehensively by FUZZY WEIGHTED
For:
η=ω
1η
1+ω
2η
2+ω
3η
3,
Wherein, η is the Comprehensive Control performance index; η
1Be normalization randomness performance index; η
2Be normalization determinacy performance index; η
3Be normalization steady-state error index; ω
iBe i weighting coefficient, i=1,2,3.
Beneficial effect of the present invention: the present invention need not system is carried out modeling, directly use the regular job number of control system to calculate, need not to test, more can not produce any influence to the normal operation of system, be the method and apparatus of the friendly type of no invasive system, can in actual control system, use very easily.In addition, this Device Testing result is normalized, and it is easy to understand as a result.
Description of drawings
Fig. 1 is that the one-piece construction of this device detects synoptic diagram.
Embodiment
The present invention is described further below in conjunction with drawings and Examples:
Preferred embodiment is elaborated.Should be emphasized that following explanation only is exemplary, rather than in order to limit the scope of the invention and to use.
The invention provides a kind of succinct, effectively, need priori is few, be easy to calculate, actual control system is normally moved the data driven type that does not have influence control system method for testing performance and device.
As shown in Figure 1, the present invention mainly contains four parts composition: data pre-processor, frequency division processor, frequency categorization device and performance evaluation device.The each several part major function is as follows: data pre-processor realizes the predecomposition of data, is about to the close loop maneuver number and is decomposed into one group of arrowband component; The frequency division processor becomes one group of natural mode of vibration component with each narrowband response; The frequency categorization device is divided into the high frequency group with all natural mode of vibration components, the intermediate frequency group, and low frequency group three classes, and constitute three groups of responses of senior middle school's low frequency; The performance evaluation device mainly is to finish the performance evaluation of different frequency range with comprehensive.Being implemented as follows of various piece:
1. data pre-processor
Have outstanding orthogonality, completeness and locality just because of wavelet package transforms, so just be used for signal is carried out pre-service.Utilize wavelet packet to decompose, the closed loop output Y of system is decomposed into trend component (low frequency component) A1 and details component (high fdrequency component) D1 with filtered device.And then trend component A1 is decomposed into new trend component and details component.Decompose successively till reaching the pretreated requirement of data.Decompose process for a n level, closed loop output Y will be broken down into 2
nIndividual arrowband component
2. frequency division processor
Because system performance is the contained system performance information difference method of evaluating performance difference of system then on different frequency ranges, handles so need carry out frequency division to system.Empirical modal decomposes can be decomposed into narrow band signal one group of intrinsic mode function.It is the method for data driven type that empirical modal decomposes, and divides measuring decomposition side as follows for j arrowband:
(1) approximating method adopts cubic spline to realize usually, the minimal value envelope x of match output number
Min(t) and maximum value envelope x
Max(t); Cubic spline is a smooth curve by a series of shape value points, on the mathematics by finding the solution the process that the three moments euqation group draws the curvilinear function group.
(2) the mean value curve of calculating minimal value envelope and maximum value envelope
Calculate intrinsic mode function L (t)=Y-me (t); Wherein, x
Min(t) be the minimal value envelope; x
Max(t) be the maximum value envelope; Me (t) is the mean value curve of minimal value envelope and maximum value envelope; Y is sampled data; L (t) is intrinsic mode function;
J arrowband component decomposites m by above step
jIndividual intrinsic mode function L
i(t) (i=1~m
j) and a remaining component, n arrowband component is divided into and solves
Individual intrinsic mode function.
3. frequency categorization device
The instantaneous energy that the frequency categorization device passes through to calculate each intrinsic mode function is divided into three groups by the K mean cluster with intrinsic mode function: high fdrequency component group, intermediate frequency component group and low frequency component group as clustering information.
At first by following formula to each intrinsic mode function c
i(t) carry out as down conversion:
Wherein: j is imaginary unit, and real part is got in Re () expression, and ω (t) is instantaneous frequency, and a (t) is amplitude,
Phase angle, t are the time, and τ is variable.
By following transformation calculations instantaneous energy:
Instantaneous energy vector with each intrinsic mode function is that clustering information finds three cluster centres as follows:
A) optional intrinsic mode function is as first cluster centre F1.
B) calculate each sample apart from the distance of F1, select from F1 distance sample farthest as second cluster centre.
C) calculate distance between each sample and fixed all cluster centres one by one, and select small distance wherein.In all minor increments, select a ultimate range, and that sample that will produce ultimate range is defined as newly-increased cluster centre.
After finding initial cluster centre, finish the K-mean cluster by following steps:
1) select 3 initial cluster centers, each object represents the center of a class;
2) and for remaining other object, then according to them and these distances of clustering centers, respectively they are distributed to the cluster of the cluster centre representative the most similar to it;
3) and then calculate the cluster centre of each new cluster that obtains, the i.e. average of all objects in this cluster;
4) calculate the distance at instantaneous energy vector to 3 center of each intrinsic mode function again, reclassify, revise new central point.Constantly repeating this process finishes when new distance center equals the central point of last time.
4. performance evaluation device
1) high frequency performance evaluator
Because the noise in the system mostly is high-frequency signal, performance evaluation is at random normally adopted in the evaluation of system's performance under noise disturbance.The randomness response output that the high frequency performance evaluator is mainly formed the high fdrequency component in the cluster analysis system utilizes least-squares algorithm to come the minimum variance performance index of computing system to come the randomness performance of evaluation system.Evaluation procedure is as follows:
A) matrix by at first being constructed as follows
Wherein y is system's output battle array, and X is structural matrix, and d is system delay, and m is that to get 10~20, n usually be hits to the number of coefficient.
B) the matrix of coefficients α by the least square method computing system:
α=(X
TX)
-1X
Ty
C) the randomness performance by the following formula computing system:
Wherein: η
1Be the randomness performance index.
2) frequency performance evaluator
Utilize cluster to obtain intermediate frequency component, the estimating step of its determinacy performance index is as follows:
A) choose cluster and obtain intermediate frequency component, estimate system transter G with the input/output relation between setting value and the intermediate frequency component again
Cl(s), and with its equivalence delay estimated value as d;
B) pass through giving G
Cl(s) do step disturbance and calculate the adjusting time T;
C) calculate nondimensional determinacy performance index
3) low frequency performance evaluator
Low frequency performance is mainly corresponding to the steady-state error of control system, the low frequency component that obtains by cluster, to its average of calculating low frequency component as steady-state error ε.What this error was reacted is the control accuracy of system.
1) performance weighter
After having calculated the randomness performance of system, determinacy performance, steady-state error, realize that by FUZZY WEIGHTED system performance is comprehensive.At first wanting three class performance index to carry out expert's scoring before comprehensively all is normalized to three indexs: η
1(randomness performance index), η
2(determinacy performance index), η
3(steady-state error index).After desired value normalization, need to calculate its weighting coefficient, adopt FUZZY WEIGHTED at this, its step is as follows:
A) by relatively setting up following comparator matrix D in twos;
B) calculate weighting matrix V by the D battle array:
C) it is as follows to calculate weighing vector by the V battle array:
ω wherein
iBe i weighting coefficient, after obtaining weighing vector, the control performance of system comprehensively is:
η=ω
1η
1+ω
2η
2+ω
3η
3
The method for testing performance of the data driven type that the present invention proposes can the mode by software programming or configuration be realized in the overwhelming majority's industrial control system (device).Also can realize by hardware simultaneously.Implementation is flexible.
The above; only for the preferable embodiment of the present invention, but protection scope of the present invention is not limited thereto, and anyly is familiar with those skilled in the art in the technical scope that the present invention discloses; the variation that can expect easily or replacement all should be encompassed within protection scope of the present invention.Therefore, protection scope of the present invention should be as the criterion with the protection domain of claim.
Claims (2)
1. the control system device for detecting performance of a data-driven is characterized in that this device comprises data pre-processor, frequency division processor, frequency categorization device and performance evaluation device, wherein,
Described data pre-processor, frequency division processor, frequency categorization device and performance evaluation device are contacted successively;
Described performance evaluation device is made up of high frequency response performance analyser, intermediate frequency response performance analyser, LF-response performance analyser and performance weighter, wherein, high frequency response performance analyser, intermediate frequency response performance analyser and LF-response performance analyser link to each other with the performance weighter with the frequency categorization device respectively.
2. the control performance detection method of a data-driven is characterized in that this method may further comprise the steps:
Step 1: data pre-processor utilizes wavelet packet to decompose to system's output sampling of control system, obtains one group of narrowband response;
Step 2: the frequency division processor utilizes empirical mode decomposition method that each narrowband response is decomposed into one group of natural mode of vibration component;
Step 3: the frequency categorization device utilizes clustering method that the natural mode of vibration component is divided three classes: high frequency response, intermediate frequency response, LF-response;
Step 4: described high frequency response performance analyser utilizes high frequency response to pass through the randomness performance of minimum variance performance Index Calculation system; Described intermediate frequency response performance analyser utilizes intermediate frequency response by the determinacy performance of nondimensional determinacy performance Index Calculation system; Described LF-response performance analyser utilizes LF-response to pass through the steady-state error of its mean value computation system;
The high frequency response performance analyser is formed the randomness response output of system with the high fdrequency component in the cluster analysis, utilizes least-squares algorithm to come the minimum variance performance index of computing system, is used for the randomness performance of evaluation system;
The intermediate frequency response performance analyser at first, is chosen cluster and is obtained intermediate frequency component, estimates system transter G with the input/output relation between setting value and the intermediate frequency response again
Cl(s), and with its equivalence delay estimated value as system delay d; Secondly, by to giving G
Cl(s) do step disturbance and calculate the adjusting time T; At last, calculate nondimensional determinacy performance index
Mainly corresponding to the steady-state error of control system, by the low frequency component that cluster obtains, as steady-state error ε, what steady-state error ε reacted is the control accuracy of system to its average of calculating low frequency component for LF-response performance analyser, low frequency performance;
Step 5: with randomness performance, determinacy performance and steady-state error performance comprehensively be by FUZZY WEIGHTED:
η=ω
1η
1+ω
2η
2+ω
3η
3,
Wherein, η Comprehensive Control performance index; η
1Be normalization randomness performance index; η
2Be normalization determinacy performance index; η
3Be normalization steady-state error index; ω
iBe i weighting coefficient, i=1,2,3.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310116365.6A CN103197668B (en) | 2013-04-03 | 2013-04-03 | Data drive type control performance detection device and method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310116365.6A CN103197668B (en) | 2013-04-03 | 2013-04-03 | Data drive type control performance detection device and method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN103197668A true CN103197668A (en) | 2013-07-10 |
CN103197668B CN103197668B (en) | 2015-02-25 |
Family
ID=48720331
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201310116365.6A Expired - Fee Related CN103197668B (en) | 2013-04-03 | 2013-04-03 | Data drive type control performance detection device and method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN103197668B (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112436531A (en) * | 2020-11-13 | 2021-03-02 | 新疆大学 | Series-parallel control wind power hybrid energy storage coordination optimization control method |
CN112990773A (en) * | 2021-04-23 | 2021-06-18 | 浙江浙能技术研究院有限公司 | Control loop performance evaluation method based on multi-index fusion |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6597958B1 (en) * | 2001-03-22 | 2003-07-22 | Abb Automation Inc. | Method for measuring the control performance provided by an industrial process control system |
CN102103376A (en) * | 2010-12-16 | 2011-06-22 | 浙江大学 | Method for evaluating control performance of PID (Proportion Integration Differentiation) loop based on subspace identification |
CN102768528A (en) * | 2012-07-27 | 2012-11-07 | 华北电力大学 | Detecting device and detecting method for control performance of multiple-input multiple-output control system |
-
2013
- 2013-04-03 CN CN201310116365.6A patent/CN103197668B/en not_active Expired - Fee Related
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6597958B1 (en) * | 2001-03-22 | 2003-07-22 | Abb Automation Inc. | Method for measuring the control performance provided by an industrial process control system |
CN102103376A (en) * | 2010-12-16 | 2011-06-22 | 浙江大学 | Method for evaluating control performance of PID (Proportion Integration Differentiation) loop based on subspace identification |
CN102768528A (en) * | 2012-07-27 | 2012-11-07 | 华北电力大学 | Detecting device and detecting method for control performance of multiple-input multiple-output control system |
Non-Patent Citations (4)
Title |
---|
刘吉臻 等: "基于数据驱动的电站燃烧稳定度综合评价", 《中国电机工程学报》, no. 35, 15 December 2007 (2007-12-15), pages 1 - 6 * |
孟庆伟 等: "一种热工控制系统综合性能的评价方法", 《中国电机工程学报》, no. 23, 15 August 2011 (2011-08-15), pages 101 - 107 * |
李庆芝 等: "多类扰动下系统综合性能的评价", 《电力科学与工程》, no. 12, 28 December 2011 (2011-12-28) * |
杨政厚 等: "串级控制回路的性能评价方法", 《动力工程》, no. 5, 15 May 2009 (2009-05-15), pages 426 - 431 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112436531A (en) * | 2020-11-13 | 2021-03-02 | 新疆大学 | Series-parallel control wind power hybrid energy storage coordination optimization control method |
CN112990773A (en) * | 2021-04-23 | 2021-06-18 | 浙江浙能技术研究院有限公司 | Control loop performance evaluation method based on multi-index fusion |
Also Published As
Publication number | Publication date |
---|---|
CN103197668B (en) | 2015-02-25 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104698837B (en) | A kind of time-varying linear structure operational modal parameter recognition methods, device and application | |
CN102183749B (en) | Sea target detecting system of adaptive radar and method thereof | |
CN110849627B (en) | Width migration learning network and rolling bearing fault diagnosis method based on same | |
Zhang et al. | Intelligent fault diagnosis of rotating machinery using support vector machine with ant colony algorithm for synchronous feature selection and parameter optimization | |
Yu et al. | A bearing fault diagnosis technique based on singular values of EEMD spatial condition matrix and Gath-Geva clustering | |
Wang et al. | Intelligent fault diagnosis of planetary gearbox based on adaptive normalized CNN under complex variable working conditions and data imbalance | |
CN108256488A (en) | A kind of radar target identification method based on micro-Doppler feature extraction and deep learning | |
CN105719023A (en) | Real-time wind power prediction and error analysis method based on mixture Gaussian distribution | |
CN102147465A (en) | System and method for detecting sea target by chaos optimizing radar | |
CN111368892B (en) | Electric energy quality disturbance efficient identification method for generalized S transformation and SVM | |
CN104714925A (en) | Gear drive noise analysis method based on fractional order Fourier transform and support vector machine | |
CN103678949B (en) | Density based is analyzed and many Extended target of spectral clustering follow the tracks of measurement collection division methods | |
CN101984360B (en) | Normalized leakage LMS self-adaptive mobile target detector based on FRFT | |
CN104237853B (en) | A kind of for the particle filter method of trace point mark sequence before multi frame detection | |
CN101799888A (en) | Industrial soft measurement method based on bionic intelligent ant colony algorithm | |
CN112564093A (en) | Low-frequency oscillation online control strategy based on pattern matching | |
CN103413134A (en) | Ground moving target micro-tremor signal characteristic extraction based on sparse decomposition | |
CN111709350B (en) | Low-frequency oscillation modal parameter identification method and system based on FCM clustering | |
CN104865075A (en) | Marine diesel engine vibration signal analysis system and method thereof | |
CN114662386A (en) | Bearing fault diagnosis method and system | |
CN103197668B (en) | Data drive type control performance detection device and method | |
CN102697491B (en) | Identification method and system of characteristic waveform of electrocardiogram | |
Pau et al. | Microcontroller architectures for battery state of charge prediction with tiny neural networks | |
CN104239689B (en) | Short-term wind speed forecasting method based on optimization Method Using Relevance Vector Machine | |
CN102183746B (en) | Radar marine target detection system and method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20150225 Termination date: 20160403 |