Fault Diagnosis System Based On Virtual Instrument Of Power Electronic Rectifier Study

Power electronic equipments are usually used as critical component in power supply and power controller in many Production fields. With the increasingly widespread use of power electronic circuits, there are highly requirements on their reliability and maintainability, and these need reliable and accurate fault diagnosis. When the system goes wrong, protection device will cut off power source of main circuit quickly( a few to several ten milliseconds) for preventing the fault enlarge, and this will also make the fault information disappeared to bring a lot of difficulties in equipment maintenance. Therefore, it is important to have the power electronic circuits dynamically monitored for on-line fault diagnosis .In-depth analysis of current research status of online monitoring and fault diagnosis system for power electronic rectifier circuit, this thesis presents a research on the implementation of online monitoring and fault diagnosis system of power electronic rectifier circuit based on virtual instrument. Take the power electronic rectifier circuit as an example, the failure feature signals are extracted by frequency spectrum analysis and neural network is used in fault diagnosis system. This thesis focuses on the network structure and training algorithm of multilayer feedforward network and back-propagation learning algorithm. The method is simulated by matlab. The results of the simulation show that the method could improve the network's convergence speed and give excellent diagnosis effect.For the implementation of system, virtual instrument technology is used to simplify hardware, data acquisition modules and signal adjusting module are configured on PC, data analyzing and processing is done with software to realize fault diagnosis of power electronic circuits. Compared with traditional instrument, virtual instrument has the following significance features: powerful data processing speed, storage, call, display as well as object-oriented programming method of Computer technology, which evidently enhances the flexibility of fault diagnosis analysis and has engineering values.