Switched Unfalsified Multicontroller Nonparametric Model Based Design
In this paper, we present a controller design strategy for the implementation of a multicontroller structure for single-input single-output (SISO) plants. The overall control system can be viewed as a feedback interconnection of a SISO plant, a set of can
- PDF / 409,414 Bytes
- 10 Pages / 439.37 x 666.142 pts Page_size
- 44 Downloads / 160 Views
Switched Unfalsified Multicontroller Nonparametric Model Based Design Fernando Coito, Luı´s Brito Palma, and Fernando Costa
Abstract In this paper, we present a controller design strategy for the implementation of a multicontroller structure for single-input single-output (SISO) plants. The overall control system can be viewed as a feedback interconnection of a SISO plant, a set of candidate controllers and a switched selection scheme that supervises the switching process among the candidate controllers. The switching scheme is designed without explicit assumptions on the plant model, based on the unfalsified control concept. A switched multicontroller structure is implemented and experimental results are presented.
5.1
Introduction
Dealing with nonlinear systems is an inherently difficult problem. As a consequence models and analysis of nonlinear systems will be less precise than for the simpler linear case. Thus, one should look for model representations and tools that utilize less precise system knowledge than the traditional approaches. This is indeed the trend in the area of intelligent control where a range of approaches, such as Fuzzy Logic, Neural Networks and Probabilistic Reasoning are being explored [1, 2]. The current paper uses operating regime decomposition for the partitioning of the operating range of the system in order to solve modeling and control problems. The operating regime approach [3–6] leads to multiple-model or multiple controller synthesis, where different local models or controllers are used under different operating conditions (see Fig. 5.1). A supervisory mechanism selects the controller according to the current local region in which the process is operating.
F. Coito (*) • L.B. Palma • F. Costa Departamento de Engenharia Electrote´cnica, Faculdade de Cieˆncias e Tecnologia, UNL, 2829-516 Caparica, Portugal e-mail: [email protected]; [email protected]; [email protected] A. Madureira et al. (eds.), Computational Intelligence and Decision Making: Trends and Applications, Intelligent Systems, Control and Automation: Science and Engineering 61, DOI 10.1007/978-94-007-4722-7_5, # Springer Science+Business Media Dordrecht 2013
47
48
F. Coito et al.
Fig. 5.1 Switching control. The switching decision between the controllers is performed trough the switching signal s
In performance-based supervision the supervisor attempts to directly assess the potential performance of the candidate controllers, without estimating the process model [6–9]. Under the unfalsification concept the best controller among a set of candidate is selected straight from input/output data. The candidate controllers’ performance is evaluated without actually applying them to the feedback loop. Controllers that do not yield the desired closed loop dynamics are entitled falsified. Only unfalsified controllers are candidate to actually control the process. The process is controlled by the controller presenting the best performance.
5.2
The Unfalsified Multicontroller
Control specifications are mapped into a
Data Loading...
