Fault Analysis And Relay Protection Of Transmission Line Based On Wavelet Transform And Neural Network

With the rapid development of China, the power industry has entered a UHV, high intelligence stage and the reliability of the power supply system has become increasingly important. The power system mainly includes transformers, transmission lines, generators and loads. The power grid is made up of the first two. Stable and secure operation of the power system depends largely on the operating status of transmission lines, so it needs to enhance protection for the safe operation of the transmission line. Especially for the single-phase grounded fault in lines. It should be given in a timely excision in order to avoid damaging the circuit normal run, which has been the top priority in safe and reliable operation of the power system.Firstly, the paper analyzes transient characteristics of the electrical quantity when the small current grounding system occurs grounding fault. The model of small current grounding system is built on the MATLAB, then taking zero sequence current and voltage of different fault types as the research object. Through the analysis and comparison between the waveform simulated and the fault theory studied by domestic and foreign, the power system model in this paper can more accurately simulate the fault state.According to the data characteristics of the transient zero sequence current, this paper uses the wavelet packet transform which can be used to analyze the fault signal in the low frequency and high frequency. In the fault signal process, the wavelet packet has a better time-frequency characteristics than traditional wavelet analysis. Db4 basis function of wavelet is chosen to decompose the fault zero sequence current obtained by simulation in three layer of wavelet packet. To analyze and extract the energy value of every continuous spectrum, then the fault line of transmission line is analyzed by combining the characteristic energy and the intelligent algorithm. Through the simulation of wavelet toolbox in MATLAB, it confirms that the algorithm can extract the fault energy more accurately and effectively, and shows that there is a certain difference between the energy values of different faults.Finally, the energy value of the fault feature extraction is used as the input of BP neural network to train the neural network. In view of the problem that the BP neural network is easy to fall into local minimum point and the slow iterative training process, the paper uses Elman neural network to improve the structure of the neural network. Finally, the neural network is tested by 50 sets of test sample. Experimental results show that when the neural network determines the correct fault line, the system will enter the fault interrupt handling process, issued a fault trip command and the results are transmitted to the output circuit of switch signal. Then the relay switch on the fault line cuts off the fault line and finishes the task of relay protection.