Fault Detection Using an Artificial Neural Network and Variable Structure Based Approach

Abstract

The ultimate goal of fault detection and isolation is to maximize the life span of equipment and minimize the cost of maintenance. The development of intelligent diagnostic, prognostic, and health management technology has proven to be important for industrial and defense maintenance procedures in recent years. While diagnostic technology for aircraft have existed for more than 50 years, modern CPUs permit on-board intelligent and estimation-based fault detection methods. This theses discussed two strategies in particular: artificial neural networks and smooth variable structure filters. The purpose of this theses is to propose a method of health state awareness for a helicopter blade using an artificial neural network as well as develop a variable structure-based fault detection and diagnosis strategy for an electromechanical actuator.