Applications of Cerebellar Model Articulation Controller Neural Network on Fault Diagnosis of Mechanical and Electrical Equipments

• Main Authors: Yi-Shin Lin, 林宜生
• Other Authors: Chin-Pao Hung
• Format: Others
• Language: zh-TW
• Published: 2007
• Online Access: http://ndltd.ncl.edu.tw/handle/21878328631104500839

碩士 === 國立勤益科技大學 === 資訊與電能科技研究所 === 95 === The objective of this thesis is to study the applications of cerebellar model articulation controller (CMAC) neural network on the fault diagnosis of mechanical and electrical equipments. To demonstrate the feasibility of the proposed scheme, we take water circulation system and steam turbine generator sets as examples. Depending on the pstterns collection of each possible fault type, we built the diagnosis architecture based on the CAMC neural network firstly. Then a steepest descent learning rule is used to train the diagnosis system until the cost function smaller than a small positive value. Finally, the diagnosis system can be used to diagnose the fault types of water circulation system or turbine generator system. In the case study, the diagnosis results demonstrated the proposed scheme outperforms than traditional schemes on the correctness, noise rejection ability and the learning speed. In this thesis, the proposed diagnosis system is implemented on a personal computer and microcontroller system simultaneously. On the personal computer, the user can input the diagnosed data by a friendly interface and obtain the diagnosis results. Also, a remote fault diagnosis web site is built in our laboratory to benefit the data collection and diagnosis test for remote users. On the microcontroller system, a portable diagnosis apparatus is implemented to benefit the fault diagnosis on the work field. All the necessary technology described above will be discussed in the thesis.