| DC motor plays an important role in the transmission system with precise speed control, it is necessary to establish the motor fault diagnosis system for those purposes such as the DC motor drive system's stable and economic operation, to reduce man-made accidents, to improve safety and reliability of equipment. Timely, continuous and objective of the unit's safety analysis, judge whether the equipment or process is normal, guide operation, management and maintenance to equipment, transition regular maintenance by the way to state maintenance.Now motor fault diagnosis method majority based on a single parameter, these parameters carry motor fault features, such as armature current, rotor speed, vibration frequency, surface temperature, etc. The result gets from the single parameter is easy to be wrong and has low reliability. In order to improve the correct rate on fault diagnosis, more theoretical techniques need to be applied to the process, we can introduce artificial intelligence expert system, artificial neural networks, and so on. So DC motor fault diagnosis based on BP neural network was proposed.The feature vector is composed by processed data, and is evidence information. Using various training algorithms to train the neural network, the trained neural network for fault diagnosis of DC motor, compare and analysis diagnosis results of different networks. Neural networks built up according to this idea have enough ability on identify nonlinear dynamic system, they can accurately map the input vectors an output vectors' relationship in the experiment, and give the extent permitted diagnosis results.The BP neural network used on fault diagnosis of DC motor is a new attempt, as the complexity of the DC motor, the research idea is firstly study neural network used in a simple normal circumstances to the motor state detection, then from simple to complex, easy to hard to explore the application of the network, study single and multiple fault, and compare the results.
|
PDF Full Text Request