基于PLC的液位控制系統(tǒng)的設(shè)計

基于PLC的液位控制系統(tǒng)的設(shè)計,基于PLC的液位控制系統(tǒng)的設(shè)計,基于,plc,控制系統(tǒng),設(shè)計 Control Engineering Practice embedded Vranc Ljubljana, b Nova Gorica Polytechnic, Nova Gorica, Slovenia identication steps to provide reliable operation. The controller monitors and evaluates the control performance of the closed-loop system. The controller was implemented on a programmable logic controller (PLC). The performance is illustrated on a eld test in industrial applications, as summarised below: ARTICLE IN PRESS C3 Corresponding author. Tel.: +38614773994; 1. Because of the diversity of real-life problems, a single nonlinear control method has a relatively narrow 0967-0661/$-see front matter r 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.conengprac.2005.05.006 fax: +38614257009. E-mail address: samo.gerksicijs.si (S. Gerksic). application for control of pressure on a hydraulic valve. r 2005 Elsevier Ltd. All rights reserved. Keywords: Control engineering; Fuzzy modelling; Industrial control; Model-based control; Nonlinear control; Programmable logic controllers; Self- tuning regulators 1. Introduction Modern control theory offers many control methods to achieve more efcient control of nonlinear processes than provided by conventional linear methods, taking advantage of more accurate process models (Bequette, 1991; Henson Murray-Smith Seborg, 1999) indicate that while there is a considerable and growing market for advanced con- trollers, relatively few vendors offer turn-key products. Excellent results of advanced control concepts, based on fuzzy parameter scheduling (Tan, Hang, Babuska, Oosterhoff, Oudshoorn, Gundala, Hoo, Ha gglund accepted 15 May 2005 Abstract This paper presents an innovative self-tuning nonlinear controller ASPECT (advanced control algorithms for programmable logic controllers). It is intended for the control of highly nonlinear processes whose properties change radically over its range of operation, and includes three advanced control algorithms. It is designed using the concepts of agent-based systems, applied with the aim of automating some of the conguration tasks. The process is represented by a set of low-order local linear models whose parameters are identied using an online learning procedure. This procedure combines model identication with pre- and post- c University of Ljubljana, Faculty of Electrical Engineering, Ljubljana, Slovenia Advanced control algorithms logic controller Samo Gerksic a,C3 , Gregor Dolanc a , Damir Saso Blazic c , Igor S krjanc c , Zoran Marins Robert King e , Mincho Hadjiski a Jozef Stefan Institute, in a programmable ic a , Jus Kocijan a,b , Stanko Strmcnik a , ek d , Miha Bozicek d , Anna Stathaki e , f , Kosta Boshnakov f Slovenia () friend matic indust from ling, proced controller monitors the resulting control performance ARTICLE IN PRESS a nonlinear process model. The model is obtained operating process signals by experimental model- using a novel online learning procedure. This The from for implementation on PLC or open controller rial hardware platforms. controller parameters are automatically tuned featu adapted ssioning of the controller is simplied by auto- experimentation and tuning. A distinguishing re of the controller is that the algorithms are mete commi The ASPECT controller is an efcient and user- ly engineering tool for implementation of para- r-scheduling control in the process industry. The used, the sensor readings, specic hardware platforms are etc. is demanded to eld of application. Therefore, more exible methods or a toolbox of methods are required in industry. 2. New methods are usually not available in a ready-to- use industrial form. Custom design requires consider- able effort, time and money. 3. The hardware requirements are relatively high, due to the complexity of implementation and computational demands. 4. The complexity of tuning (Babuska et al., 2002) and maintenance makes the methods unattractive to nonspecialised engineers. 5. The reliability of nonlinear modelling is often in question. 6. Many nonlinear processes can be controlled using the well-known and industrially proven PID controller. A considerable direct performance increase (nancial gain) is demanded when replacing a conventional control system with an advanced one. The main- tenance costs of an inadequate conventional control solution may be less obvious. The aim of this work is to design an advanced controller that addresses some of the aforementioned problems by using the concepts of agent-based systems (ABS) (Wooldridge Section 3 gives a brief description of the CT; and nally, Section 4 describes the application of the controller to a pilot plant where it is used for control of the pressure difference on a hydraulic valve in a valve test apparatus. 2. Run-Time Module The RTM of the ASPECT controller comprises a set of modules, linked in the form of a multi-agent system. Fig. 1 shows an overview of the RTM and its main modules: the signal pre-processing agent (SPA), the online learning agent (OLA), the model information agent (MIA), the control algorithm agent (CAA), the control performance monitor (CPM), and the operation supervisor (OS). 2.1. Multi-faceted model (MFM) The ASPECT controller is based on the multi-faceted model concept proposed by Stephanopoulus, Henning, and Leone (1990) and incorporates several model forms required by the CAA and the OLA. Specically, the MFM includes a set of local rst- and second-order local learning approach (Murray-Smith j ykb 1; j uk C0du j c 1; j vk C0dv j r j , (1) while the model equation of second order models is yk 1C0a 1; j ykC0a 2; j yk C01b 1; j uk C0du j b 2; j uk C01C0du j c 1; j vk C0dv j c 2; j vk C01C0dv j r j , 2 where k is the discrete time index, j is the number of the local model, y(k) is the process output signal, u(k) is the process input signal, v(k) is the optional measured disturbance signal (MD), du is the delay in the model branch from u to y,dv is the delay in the model branch from v to y, and a i,j , b i,j , c i,j and r j are the parameters of the jth local model. The set of local models can be interpreted as a TakagiSugeno fuzzy model, which in the case of a second order model can be expressed in the Fig. 1. Run-time module g Practice () 3 following form: yk 1C0 X m j1 b j ka 1; j ykC0 X m j1 b j ka 2; j yk C01 X m j1 b j kb 1; j uk C0du j X m j1 b j kb 2;j uk C01C0du j X m j1 b j kc 1; j nk C0dn j X m j1 b j kc 2; j nk C01C0dn j X m j1 b j kr j , 3 where b j ( k) is the value of the membership function of the jth local model at the current value of the scheduling variable s(k). Normalised triangular membership func- tions are used, as illustrated in Fig. 2. overview. ARTICLE IN PRESS The scheduling variable s(k) is calculated using coefcients k r , k y , k u ,andk v , using the weighted sum skk r rkk y ykk u uk C01k v vk. (4) The coefcients are congured by the engineer accord- ing to the nature of the process nonlinearity. 2.2. Online Learning Agent (OLA) The OLA scans the buffer of recent real-time signals, prepared by the SPA, and estimates the parameters of the local models that are excited by the signals. The most recently derived parameters are submitted to the MIA only when they pass the verication test and are proved to be better than the existing set. The OLA is invoked upon demand from the OS or autonomously, when an interval of the process signals with sufcient excitation is available for processing. It processes the signals batch-wise and using the local learning approach. An advantage of the batch-wise concept is that the decision on whether to adapt the model is not performed in real-time but following a delay that allows for inspection of the identication result before it is applied. Thus, better means for control over data selection is provided. A problem of distribution of the computation time required for identication appears with batch-wise processing of data (opposed to the online recursive processing that is typically used in adaptive controllers). This problem is resolved using a multi-tasking operation system. Since the OLA typically requires considerably Fig. 2. Fuzzy membership functions of local models in the MFM. S. Gerksic et al. / Control Engineerin4 more computation than the real-time control algorithm, it runs in the background as a low-priority task. The following procedure, illustrated in Fig. 3,is executed when the OLA is invoked. 2.2.1. Copy signal buffer The buffer of the real-time signals is maintained by the SPA. When the OLA is invoked, the relevant section of the buffer is copied for further processing. 2.2.2. Excitation check A quick excitation check is performed at the start, so that processing of the signals is performed only when they contain excitation. If the standard deviations of the signals r(k), y(k), u(k), and v(k) in the active buffer are below their thresholds, the execution is cancelled. 2.2.3. Copy active MFM from MIA The online learning procedure always compares the newly identied local models with the previous set of parameters. Therefore, the active MFM is copied from the MIA where it is stored. A default set of model parameters is used for the local models that have not yet been identied (see Section 2.3). 2.2.4. Select local models A local model is selected if the sum of its membership functions b j (k) over the active buffer normalised by the active buffer length exceeds a given threshold. Only the selected local models are included in further processing. 2.2.5. Identification The local model parameters are identied using the fuzzy instrumental variables (FIV) identication method developed by Blazic et al. (2003). It is an extension of the linear instrumental variables identication procedure (Ljung, 1987) for the specied MFM, based on the local learning approach (Murray-Smith MIA is copied from the active MFM, and the covariance matrix P j,MIA is initialised to 10 5 I (identity matrix). The FLS (fuzzy least-squares) estimates, h j;FLS and P j,FLS , are obtained using weighted least-squares identication, with b j (k) used for weighting. The calculation is performed recursively to avoid matrix inversion. The FIV (fuzzy instrumental variables) estimates, h j;FIV and P j,FIV , are calculated using weighted instrumental variables identi- cation. In order to prevent result degradation by noise, a g Practice () dead zone is used in each step of FIV and FLS recursive ARTICLE IN PRESS S. Gerksic et al. / Control Engineerin estimation. The vector of parameters and the covariance matrix are updated only if the absolute weighted difference between the process output and its prediction is above the congured noise threshold. Fig. 3. Online learning g Practice () 5 In case of lack of excitation in the branch from u to y or in the model branch from v to y (or when measured disturbance is not present at all), variants of the method with reduced parameter estimate vectors are used. procedure. C15 wi ARTICLE IN PRESS 2.2.6. Verification/validation This step is performed by comparing the simulation output of a selected local model with the actual process output in the proximity of the local model position. The normalised sum of mean square errors (MSE j )is calculated. The proximity is dened by the membership functions b j . For each of the selected local models, this step is carried out with three sets of model parameters: h j;MIA ; h j;FLS ; and h j;FIV : The set with the lowest MSE j is selected. Then, global verication is performed by comparing the simulation output of the fuzzy model including the selected set with the actual process output. The normal- ised sum of mean square errors (MSE G ) is calculated. If the global verication result is improved compared to the initial fuzzy model, the selected set is sent to the MIA as the result of online learning, otherwise the original set h j;MIA remains in use. For each processed local model, the MIA receives the MSE j , which serves as a condence index, and a ag indicating whether the model is new or not. 2.2.7. Model structure estimation Two model structure estimation units are also included in the OLA. The dead-time unit (DTU) estimates the process time delay. The membership function unit (MFU) suggests whether a new local model should be inserted. It estimates an additional local model in the middle of the interval between the two neighbouring local models that are the most excited. The model is submitted to the MIA if the global validation of the resulting fuzzy model is sufciently improved, compared to the original fuzzy model. 2.3. Model Information Agent (MIA) The task of the MIA is to maintain the active MFM and its status information. Its primary activity is processing the online learning results. When a new local model is received from the OLA, it is accepted if it passes the stability test and its condence index is sufcient. If it is accepted, a ready for tuning ag is set for the CAA. Another ag indicates whether the local model has been tuned since start-up or not. If the model condence index is very low, the automatic mode may be disabled. The MIA contains a mechanism for inserting addi- tional local models into the MFM. This may occur either by request or automatically, using the MFU of the OLA. The MIA may also store the active MFM to a local database or recall a previously stored one, which is useful for changing of modes. At initial conguration, the MIA is lled with default local models based on the initial estimation of the process dynamics. They are not exact but may provide S. Gerksic et al. / Control Engineerin6 reliable (although sluggish) control performance, similar procedure of the controller parameters from the MFM when the MIA reports that a new local model is generated if auto-tuning is enabled. The parameters of the control layer and the scheduling layer are replaced in such a manner that real-time control is not disturbed. Three CAAs have been developed and each has been proved effective in specic applications: the Fuzzy parameter-scheduling controller (FPSC), the dead-time compensation controller (DTCC), and the rule-based neural controller (RBNC). In this paper, only the concept of the FPSC is described briey in the following subsection. 2.4.1. Fuzzy parameter-scheduling controller 3. lers, so that in conjunction with the control layer a xed-parameter nonlinear controller is realised. The tuning layer executes the automatic tuning The scheduling layer performs real-time switching or scheduling (blending) of tuned local linear control- 2. 1. The control layer offers the functionality of a local linear controller (or several local linear controllers simultaneously), including everything required for industrial control, such as handling of constraints with anti-windup protection, bump-less mode switch- ing, etc. consisting Auto mode (or several auto modes with different tuning parameters): a nonlinear controller. The CAAs share a common interface of interaction th the OS and a common modular internal structure, of three layers: C15 Manual mode: open-loop operation (actuator con- straints are enforced). Safe mode: a xed PI controller with conservatively tuned parameters. C15 to the Safe mode. Using online learning through experiments or normal operating records (when the conditions are appropriate for closed-loop identica- tion), an accurate model of the plant is estimated gradually by receiving identied local models from the OLA. 2.4. Control Algorithm Agent (CAA) The CAA comprises an industrial nonlinear control algorithm and a procedure for automatic tuning of its parameters from the MFM. Several different CAAs may be used in the controller and may be interchanged in the initial conguration phase. The following modes of operation are supported: g Practice () An overview of the FPSC is shown in Fig. 4. ARTICLE IN PRESS S. Gerksic et al. / Control Engineerin The control layer of the FPSC includes a single PID controller in the form suitable for controller blending using velocity-based linearisation. It is equipped with anti-windup protection and bump-less transfer. The scheduling layer of the FPSC performs fuzzy blending of the controller parameters (in the case of Ti, its inverse value) according to the scheduling variable s(k) and the membership functions b j (k) of the local models. The instrument of velocity-based linearisation enables the dynamics of the blended global controller to be a linear combination of the local controller dynamics in the entire operating region, not just around equili- brium operating points. This provides the potential to improve performance with few local models and more transparent behaviour in off-equilibrium operating points (Leith Kocijan, Z unic, Strmcnik, if so, it terminates processing. Otherwise, it lters the signals y and v and performs a low-level analysis. The SC scans the pre-processed buffer for the last recognisable event that may be evaluated, or is otherwise important, e.g., a step change of the reference signal, a step change of the measured disturbance signal, an occurrence of an unmeasured disturbance, or the presence of oscillation. If an event that may be evaluated is detected and the conditions for feature estimation are fullled (there is no excessive oscillation, the signal-to- noise ratio is sufcient, the process response has settled after the event and there was a period of steady-state before the event), the corresponding buffer interval is sent to the PE, otherwise the execution is terminated. The PE may extract the following features of detected events: overshoot, settling time, rise time, oscillation decay rate, and tracking error measure or regulation error measure. Using a fuzzy evaluation procedure, an overall performance index (PI) is also calculated from the features. The CPM results are sent to the humanmachine interface. If poor performance is detected, the CPM triggers an automatic switchover to the Safe mode. Other automatic actions include, for example, blocking the OLA if oscillation is detected. Modications of the CAA parameters based on CPM results are not generally performed because such actions are highly process-specic, but they may be implemented in specic applications. 2.6. Operation supervisor (OS) The OS coordinates the control, modelling, and tuning activities of the agents and user interaction through the hierarchical set of dialogue windows of the human-machine interface (HMI). The OS and the HMI include the functionality required for automatic user- friendly experimentation, which is usually required for controller tuning. The controller commissioning procedure comprises the phases of basic settings, approximate estimation of the process dynamics for safe controller tuning, non- linear modelling and tuning of the scheduling controller, and conguring the regime for regular operation. The OS supports the control engineer by automatic execution of experiments for identication of local S. Gerksic et al. / Control Engineerin8 models. These experiments consist of a series of step which models have been tuned and their condence indices. While initiative and suggestions of the agents are helpful during system conguration, this may not be desired during regular operation. Therefore, at the end of the commissioning procedure, the system may be recongured to simplify the operation as much as possible. A range of operating regimes can be congured by enabling or disabling the agents and changing their conguration parameters. This results in a exible control system that covers the requirements of a wide range of applications and may help diagnose problems. Thus, although designed for control of nonlinear processes, the ASPECT controller may also be used for adaptive control using a single linear model or as a tool for PID controller tuning. Some specic operating regime options are listed below: C15 The OLA and/or the CPM may be invoked autono- mously (during regular operation) or upon OS demand (following scheduled experiments), or both. C15 The OLA may estimate the process dead-time c