Develop Of The Device Of Distribution Network Fault Line Detection Based On Embedded Linux System

Form the early1990s, It has been launched a variety of fault line-selection devices in small-current grounding system. However, as a result of the limitation on line-selection method and hardware, it was not ideal affect after these devices had been applied to the field. Nowdays, these issues has been resolved depend on the constant development of the fault line-selection theory on small-current grounding system and advantment of hardware.This paper, which aims at6KV distribution network system on the power plant of Dexing copper mine, would have researched a set of intelligent fault line-selection devices based on data fusion line-selection method of BP neural network. The device would been adopted S3C2440chip on a series of ARM9on Samsung as the processor and the embedded Linux system as software platform. The device has many advanced advantages such as smaller volume, stronger function, higher reliability, better portability, higher accuracy in line-selection and so on.The device would have adopted data fusion line-selection method based on the BP neural network. The device will be to the active power, the five harmonic component and the wavelet energy of zero-sequence current on line as the characteristic parameter, to BP neural as data fusion comprehensive line-selection model. This paper will research some main contents:(1) It will be researched to the data fusion line-selection method on BP neural network. Through analyzing steady state information, transient state information and harmonic information on fault zero-sequence of single-phese grounding system, it is identified that the small-current grounding fault line-selection should choose the active power, the five harmonic component dan the wavelet energy of zero-sequence current as the fault characteristic. According to the feature of the fault characteristic define their fault measures respectively. To their fault measures as the input of BP neural network model, it eventually abtains the result of fault diagnosis.(2) The implementation of the driver of CAN bus on the underlying embedded Linux system. The host computer, which communicate with the front sampling machine via the CAN bus, obtain the zero-sequence current data for fault diagnosis.(3) The software design of fault diagnosis algorithm. Including the calculation of active power and reactive power of zero-sequence current, the extraction of the harmonic component of zero-sequence current, the disintegration of zero-sequence by the wavelet, the calculation of fault measure of each characteristic, and the establishment of BP neural network model.(4) making the QT user interface. The QT user interface make user very convenient to direct check the real-time data of fault diagnosis, that will bring a big convenience to the worker in the field.