Research On Transformer Microcomputer Protection Based On Artificial Neural Network

With the development of country economy,there has been more and more requirements of electrical power energy,and power system capacity is enlarged constantly.Because large-scale power transformers are adopted widely,many requirements for the transformer protection are put forward,for example fast action ability and dependability. Differential protection has long been the key protection for power transformer. How to improve its rapidity,sensitivity and reliability is a popular research field.So far, the core problem of transformer differential protection is still how to precisely identify the magnetizing inrush current.Inrush current identification methods currently being used, can't very well meet the requirements of modern large power transformer protection. In order to adapt to the requirements of large transformer protection, it is important to search a new method which can discriminate the inrush current and short current faster and more reliably.Radial basis functuon neural network is a new and effective neural network.It has the best classification ability and universal approximation property,simple structure and fast training speed.So it has particular advantages when applied in system indentification.The choice of quantity and position of hidden layer radial basis functions is very important and directly affects the goodness of fit of overall network classification ability.In this paper,a new optimization algorithm based on particle swarm optimization for radial basis functuon neural network is presented after traditional algorithms are thoroughly researched.For a long while,the application of radial basis functuon neural network to ptotection is based on classification ability.In this paper, radial basis functuon neural network is trained by sample data.The various faults and inrush current of power transformer can be distinguished and judged by the trained radial basis functuon neural network.Applying this model, power transformer fault and inrush current were analyed and compared in two different algorithms of radial basis functuon neural network,and two simulation results were acquired.The whole results from the experiment shows that the radial basis functuon neural network model designed in this thesis can discriminate the magnetizing inrush current and fault current successfully and quickly. Finally, a complete differential key peotection scheme based on radial basis functuon neural network is presented, and the fault treatment's process of the main differential protection scheme is designed.