Plenary lecture 1: fault detection and isolation using neuro-fuzzy systems

Intelligent systems have been widely used in many industrial applications such as: Control systems, Identification, Pattern recognition and fault detection and diagnosis; being fault detection one of the most developed area cause of it direct incidence on productivity and security. Artificial Neural Networks and Fuzzy systems have been some of the Artificial Intelligence techniques that have been used for these activities. Some of the reasons for using Artificial Neural Networks are: Can "Learn" from historical data, so they can be used as associative memories. they have great generalization capabilities, so they can give accurate outputs for input patterns different that the used in the training phase. They can be used with corrupt or incomplete data, because the "knowledge" is spread over the networks interconnection weights. Can give input/output maps from data without apparent relation. They are easy for computer implantation. There exist a great number of learning algorithms that can be used for specific problems. On the other hand, fuzzy logic emulates the human classification capabilities using multivaluated criteria instead of the classical binary logic used in computational environments. Fuzzy logic creates some fuzzy sets which are described using linguistic labels and a membership level with values between [0,1] according to the real partial belonging to each of the created fuzzy sets. In this plenary speech it will be presented fault detection schemes based on diverse Neuro-Fuzzy configurations. It will be also presented some industrial examples.