Fault Tolerant Control Based on Adaptive LQG and Fuzzy Controllers
This paper describes a fault tolerant switching loop control study case for a three water tank laboratory setup with sensors and actuators redundancies. Fault diagnosis and tolerance are two main characteristics of intelligent setups when closed-loop cont
- PDF / 343,099 Bytes
- 12 Pages / 439.37 x 666.142 pts Page_size
- 69 Downloads / 184 Views
Fault Tolerant Control Based on Adaptive LQG and Fuzzy Controllers Carla Viveiros, Luis Brito Palma, and Jose´ Manuel Igreja
Abstract This paper describes a fault tolerant switching loop control study case for a three water tank laboratory setup with sensors and actuators redundancies. Fault diagnosis and tolerance are two main characteristics of intelligent setups when closed-loop control is used for large complex systems. Discrete adaptive LQG (linear quadratic) and Fuzzy fault tolerant controllers, for nominal situation, are presented and tested in the laboratory setup. For abrupt fault situation (structural fault), switching from one controller to another based on controller performance evaluation using PCA (principal component analysis) technique is implemented. The main contributions are: (a) proposed approach based on PCA to detect structural faults; (b) performance assessment based on performance index through Mahalanobis distance using T2 statistics and (c) control switching between LQG and Fuzzy controllers based on the performance index. Experimental results applied to a three-tank laboratory setup in the presence of a structural fault are shown to evaluate the performance of the proposed approach. Keywords Structural faults • Principal component analysis • Controller performance monitoring • Switching control
C. Viveiros (*) • J.M. Igreja Automation & Power Systems Engineering, Instituto Superior de Engenharia de Lisboa, 1959-007 Lisbon, Portugal e-mail: [email protected]; [email protected] L.B. Palma Electrical Engineering, Universidade Nova de Lisboa, 2829-516 Caparica, Portugal e-mail: [email protected] A. Madureira et al. (eds.), Computational Intelligence and Decision Making: Trends and 143 Applications, Intelligent Systems, Control and Automation: Science and Engineering 61, DOI 10.1007/978-94-007-4722-7_14, # Springer Science+Business Media Dordrecht 2013
144
14.1
C. Viveiros et al.
Introduction
Our present-day society is strongly dependent on the availability and correct functioning of control systems. One of the many benefits of Fault Tolerant Control (FTC) approach is to maintain overall system stability and acceptable performance in the presence of faults. FTC approaches [1] can be classified in two categories: passive (e.g. robust control) and active approach (e.g. adaptative control). To implement active fault tolerant control approaches two tasks are needed: fault detection and isolation (FDI) and controller reconfiguration or accommodation. In the area of controller performance there are several techniques proposed, e.g. [2–4], in this work the principal component analysis (PCA) technique will be used. PCA is a multivariate statistical method, within the class of linear methods, which uses linear correlations while reducing data dimension and is suitable for control loop performance assessment and fault detection. In [5], PCA and non linear PCA were presented and the performance index is based on Q statistics. In [6], a benchmark to monitor model predictive control wa
Data Loading...