Developing a robust model predictive control architecture through regional knowledge analysis of artificial neural networks

Tsai, P.H.; Chu, Jizheng; Jang, Shi-Shang; Shieh, Shyan-Shu

doi:10.1016/s0959-1524(02)00067-7

Cited by 27 publications

(10 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The main issue in implementation of nonlinear model predictive control is the computational complexity of control variables [32]. The rationale underlying model predictive control is to transform a control problem into an optimization one, so that at any sampling time, a sequence of future control values is computed by solving a finite horizon optimal problem [33][34][35]. For a nonlinear system, by setting the current values of the system as initial values, the data-based model is used to predict the future outputs over a length of a moving horizon N at each sampling instant, future control values are solved by a finite horizon M optimal problem.…”

Section: Model Predictive Control Of Data-based Modelmentioning

confidence: 99%

Data-based intelligent modeling and control for nonlinear systems

Sun

2011

Front. Electr. Electron. Eng. China

View full text Add to dashboard Cite

With the ever increasing complexity of industrial systems, model-based control has encountered difficulties and is facing problems, while the interest in data-based control has been booming. This paper gives an overview of data-based control, which divides it into two subfields, intelligent modeling and direct controller design. In the two subfields, some important methods concerning data-based control are intensively investigated. Within the framework of data-based modeling, main modeling technologies and control strategies are discussed, and then fundamental concepts and various algorithms are presented for the design of a data-based controller. Finally, some remaining challenges are suggested.

show abstract

Section: Model Predictive Control Of Data-based Modelmentioning

confidence: 99%

Data-based intelligent modeling and control for nonlinear systems

Sun

2011

Front. Electr. Electron. Eng. China

View full text Add to dashboard Cite

show abstract

“…Ho et al (2001) suggested that using neural networks with confidence bounds could provide more quality information on the performance of the deposition process for better decisionmaking and continuous improvement of a solder paste deposition process. Tsai et al (2002) developed a robust model predictive control architecture using artificial neural networks. The regional knowledge analysis method was proposed and incorporated in the analysis of dynamic artificial neural network models in process control.…”

Section: Machine Learning and Patterns Recognition Techniquesmentioning

confidence: 99%

Artificial intelligence for monitoring and supervisory control of process systems

Uraikul

Chan

Tontiwachwuthikul

2007

Engineering Applications of Artificial Intelligence

177

View full text Add to dashboard Cite

“…The pH process is a continuous stirred tank of reactions (CSTR) and has been studied by Palancar et al(1996Palancar et al( , 1998 and by the authors (Tsai et al, 2002). There are two inlet streams to the CSTR, the acid fl ow, an aqueous solution of acetic acid (AcH) and propionic acid (PrH) with fl owrate Q A Q A Q and concentrations C AcH C AcH C ,…”

Section: A Simulated Ph Process Under Cmpc and Mpcmentioning

confidence: 99%

A Comparative Study of Combined Feedforward/Feedback Model Predictive Control for Nonlinear Systems

2004

Self Cite

View full text Add to dashboard Cite

T ighter performance specifi cations from worldwide competition and ever increasing constraints from environmental and safety considerations give the practical driving force for the development of advanced control technology (Findeisen and Allgöwer, 2001). Large amounts of industrial practice and academic research have made model predictive control (MPC) the de facto standard algorithm for advanced control in the process industry (Nikolaou, 2001). MPC is a general and mathematically feasible scheme to integrate our knowledge about a target, process into controller design and operation, which allows fl exible and effi cient exploitation of our understanding of a target, and thus optimal performance of a system under various constraints. Three key factors responsible for the great success of MPC are the incorporation of a process model, an algorithm considering plant behaviour over a future horizon in time, and explicit treatment of constraints (Qin and Badgwell, 1999), with the model being foremost and fundamental.For an accurate and reliable prediction of a model, suffi cient information input is the most important factor, though algorithms in the model are equally important regarding the use of the input information. It is well known that combined feedforward plus feedback control can signifi cantly improve performance over simple feedback control whenever a major disturbance exists. It can be measured before it affects the process output, and methods for designing linear feedforward/feedback control systems are well documented in standard textbooks such as that of Stephanopoulos (1984). Such methods are also suitable for nonlinear systems where the effect of measured disturbances (DVs) can be separated from that of manipulated variables (MVs). However, for systems where DVs and MVs are closely coupled together, a unifi ed model correlating both kinds of variables is necessary and should be inverted online. At this time, MPC is a ready solution. Just as pointed out by Economou et al. (1986), MPC has the capability to combine the advantages of open-loop (feedforward) and feedback control. A future horizon in time considered in the MPC algorithm means that the effects of measured and unmeasured disturbances can be predicted and eliminated (Qin and Badgwell, 1999). Developing a valid model for process dynamics is often the major work in the implementation of an advanced control. More than 75% of the expenditure in an advanced control project normally goes to modelling. Artifi cial neural networks (ANNs) as a process model for control purpose are superior to other conventional modelling methods for reasons of complexity, accuracy, fl exibility, generality, execution speed and cost (Bhat and McAvoy, 1990;MacMurray and Himmelblau, 1995; Hussain, 1995). They have been widely studied in various modelbased control strategies. Various types of neural networks have been studied in the literature of process control, and the multilayered feedfor- * Author to whom correspondence may be addressed. E-mail address: ssjang@che.nt...

show abstract

Developing a robust model predictive control architecture through regional knowledge analysis of artificial neural networks

Cited by 27 publications

References 27 publications

Data-based intelligent modeling and control for nonlinear systems

Data-based intelligent modeling and control for nonlinear systems

Artificial intelligence for monitoring and supervisory control of process systems

A Comparative Study of Combined Feedforward/Feedback Model Predictive Control for Nonlinear Systems

Contact Info

Product

Resources

About