Data Acquisition Using TMS320F2812 And Power Cable Fault Recognition

Power cable is used to transmit and allocate the power energy and it is very important component of power network. It will cause power disruption in local or whole areas once power cable has fault, and then it will bring enormously economic loss for government and the life of people. Therefore, ensuring the normal running of power cable is the precondition to guarantee the normal running of power network. So, they are very important that studying the data acquisition and identifying the fault of power cable.Based on the hardware platform of DSP system & computer and the theory of wavelet analysis & artificial neural network, this paper designs the data acquisition module by using TMS320F2812. It gives a profound study in fault recognition technology. It achieves fault recognition and classification of the typical power cable fault. The contribution of this paper consists of the following parts.In order to obtain the signal of fault voltage or current when power cable has fault, the data acquisition module is firstly designed and realizied by using TMS320F2812. This module includes the schematic circuit diagram, the PCB of the smallest system of DSP, the peripheral extended circuit and the interface circuit of I/O communication. All the working principles of each part are discussed. Secondly, the main programs of the data acquisition module, the DSP initialization, the interrupt timing disposal, the A/D acknowledge interrupt and the UART transmission are accomplished with environment of CCS2.21. The working processes of all programs are analyzed in detail and the flow charts are given. Thirdly, the hardware and software function of data acquisition transmission module are testes. The experiment proves that data acquisition module works very well and the design is correct.To the problem of power cable fault recognition, for one thing, this paper studies the relative theories of artificial neural network and wavelet analysis, constructs two typical simulative models for producing fault signal. For another, this paper builds the feedback artificial neural network module of Elman, writes the program of fault recognition with Matlab, successfully fulfill the recognition of cable open circuit fault and short circuit fault. Finally, the loose module of wavelet neural network is built by using the fault feature signal based on wavelet package decomposing---energy method. The neural networks model can correctly recognize and classify the cable fault patterns among the 1 phase short circuit and grounding fault, 2 phase short circuit, 3 phase short circuit and grounding fault. Two examples show that the simulative model is valid and two classifiers have good performances.