ABSTRACT
Condition monitoring and faults classification are a multi-disciplinary and integrated technology. It consists of the knowledge of intelligent system and the other parts such as analysis of vibration signal, current, temperature or even all of acquired signal from data acquisition. So it depends on how and which parts the intelligent knowledge will be dealt with. Developing intelligent system, combining modern signal analysis and soft computing theory, data mining, realizing the real-time, online dynamic monitoring, are a developmental direction to reach relatively best performance in condition monitoring and faults detection of machine. Recently, the research area in machine learning has been applied to perform condition monitoring and faults classification. State of the art for developing and improving performance of system is still on going research. One of state of the art is kernel trick; it is one of the crucial tricks for machine learning. Its basic idea is to project the input data into a high-dimensional implicit feature space with a nonlinear mapping, and then the data is analyzed so that nonlinear relations of the input data can be described.
