Search

Abstract: Detecting fault before it deteriorates the system performance is crucial for the reliability and safety of many engineering systems. This paper develops an intelligent technique based on fuzzy-genetic algorithm (FGA) for automatically detecting faults on HVAC system. Many researchers have proposed only using fuzzy systems to effect fault detection and diagnosis. Other applications of the FGA are mainly focused on the synthesis of fuzzy control rules. The proposed automatic fault detection system (AFD) monitors the HVAC system states continuously by fuzzy system. The optimization capability of genetic algorithms allows the generation of optimal fuzzy rules. Faults are represented as different fault levels in the AFD system and are distinguished by fuzzy system after tuning its rule table. Simulation studies are conducted to verify the proposed AFD system for the single zone air handler system. [Copyright &y& Elsevier]

A fuzzy controller with online learning capability is reported in this paper. The controller learns from a standard proportional plus derivative (PD) controller. It is implicitly assumed that the tuning parameters of the PD controller are already known. The learning is realized via Wang's table lookup scheme. The controllers are applied successfully to control an open-loop unstable system, i.e., the ball and plate system. Experimental studies have demonstrated the performance of the proposed controller. [ABSTRACT FROM AUTHOR]

This paper proposes an antecedent validity adaptation (AVA) principle for fuzzy systems tuning. It is suggested that the fuzzy rules may be updated with respect to the validity of antecedents. This adaptation principle agrees with the human intuition and fuzzy logic reasoning. The principle is first applied to table look-up (TL) scheme to model recorded data; an on-line fuzzy identification algorithm is also proposed. Next, the principle is applied to fuzzy clustering algorithm. These methods are successfully applied to model nonlinear systems. [Copyright &y& Elsevier]

Examines the problem of optimization of fused fuzzy systems (FFS) via genetic algorithms. Use of rule-table transformation to separate a large fuzzy rule optimization problem; Application of the cart-pole systems; Characteristics of FFS.

Conventional hill-climbing type adaptive algorithms suffer from large transient error in abrupt changing environments. In this paper, we suggest a novel idea to address this problem. A data re-initialization (DR) scheme is proposed using multiple models approach with special initialization technique to smooth transition between different environments. The design is based on adaptive fuzzy nearest neighbor clustering algorithm with on-line enhancements. The fast adaptation is realized by the data re-initialization and switching between different models; learning is realized by recording and retrieving the trained up models. The algorithm is conceptually simple and feasible for real-time applications. The performance of the algorithm is tested with simulation studies. [ABSTRACT FROM AUTHOR]