Development of immune intelligent computing techniques for real-time condition monitoring and fault diagnosis

Maintenance operation is important for the process of industrial product and efficiency. Predictive maintenance is more effective than reactive one for reducing of costs in unexpected fault condition. Condition monitoring based predictive maintenance, which is based on measuring of some system parameters and data acquisition, is used to fault detection, diagnosis, and prognosis by extracting certain features from acquired data.In this thesis, the aim is to develop intelligent computing techniques for real-time and online condition monitoring and fault diagnosis. In accordance with this aim, theoretical methods are developed and theirs practical applications are implemented. Electrical machines, which are rigorously used in industrial systems and important components, are selected as an application problem in thesis scope. Intelligent computing techniques based real time methods are developed for fault detection, diagnosis, and prognosis in these machines.In method development stage covered the theoretical side of the thesis; artificial immune systems, support vector machines, artificial neural networks, and fuzzy logic are proposed for fault detection, diagnosis and prognosis. The parameters of the algorithms are adjusted by using particle swarm optimization, clonal selection, and genetic algorithm to improve the convergence speed and diagnosis performance of them. Signal processing based methods are developed for extracting distinctive features. Developed methods detect faults with a little data necessity and under different operation speeds.In the application stage of the thesis, signals are acquired from system by founding experimental setup. The proposed methods are performed as online and they are implemented on Altera Cyclone III FPGA card as real time. The developed methods are verified by experimental studies.As a result, developed techniques are supported by FUBAP and TUBITAK 1001 projects conducted in the thesis process. Studies are concluded by national and international level of academic publications.