Power Grid Fault Diagnosis Based On Fuzzy Neural Network

With the power grid scale expanding now, the number of power grid components affected by all kinds of failure is also increasing. Currently, the fault diagnosis of power grid is mainly using artificial intelligence techniques, movement protection and circuit breaker information. However, due to the power grid becomes more and more complex now, the causes have become more and more complex and diverse, leading protection and circuit breakers can not diagnosis fault components due to signal distortion. Based on this, in order to diagnosis fault power grid components quickly and accurately, the main work is as follows:Build a simple power grid structure model through the analysis to the action of the fault components, first.Then build the action relationship between the fault components and protection, and sorting out the fault test sample set of power grid. Then using fuzzy scale of power grid failure frequency to fuzzy process training samples, and representing in the form of vectors to obtain the input vector set of fault power grid components and output vector set of fault power grid components. Finally, constructed fuzzy neural network through hybrid fuzzy neural network classifier, used MATLAB to simulate, and analyzing the results of simulation.Simulated faulty power components through the standard BP neural network, and obtaining the error curve of BP neural network. It can be seen from the convergence curve, the convergence of BP neural network is poor, and there is a big error shock phenomenon. Simulating fault power gird components through hybrid fuzzy neural network classifier, we can obtain the error curve of fuzzy neural network. It can be seen from the convergence curve that the convergence rate of fuzzy neural network is significantly quickly than standard BP network. In addition to the network to achieve the desired goal of precision error, the error curve is shocking very weak, in another word is that the convergence curve is more smoothly. Indicated the training of the network model had begun to mature.Through the simulated power grid structure, and compaired with the traditional artificial intelligence technology, fuzzy neural network. Convergence is quickly, and actual output of simulation is very close to theoretical output. So, fuzzy neural network has more advantages in diagnosing of complex and uncertain fault than the traditional artificial intelligence technology.The main innovation of this paper:The integrated use of two traditional artificial intelligence techniques, artificial neural network, characteristics of fuzzy logic theory to construct fuzzy neural network combined with movement protection theory and identify complex fuzzy fault information.