The Research On Intelligent Methods Of Power Transformer Fault Diagnosis

The development of the power transformer fault diagnosis technology has been closely linked to the economic efficiency and safety situation of the entire power system. With the continuous improvement of scientific and technological level, power transformer trends toward high-capacity, high voltage, large power grid. Fault occurred more and more, and the fault is more complexly. At present, the gas analysis is the most widely in the field of transformer fault diagnosis. These years, researchers at home and abroad are working on varieties of intelligent algorithm in the transformer fault diagnosis and have also made a lot of breakthrough results. On the basis of the extensive literature, the main work of this paper is as follows:First of all, on the basis of elementary particle swarm optimization, in this paper, because it has some shortcomings of falling into local minimum easily and slow convergence rate. I have improved the inertia weight and acceleration factor, retains its advantages of powerful search capabilities, setting up less parameters and easy algorithms. Through experimental simulation, with improved particle swarm algorithm instead of BP algorithm to train the neural network, this method can access to the network the smallest error and identify the type of transformer failure. Simulation results show that the improved particle swarm algorithms applied to transformer fault diagnosis. Compared with the improved BP algorithm, the iterations are fewer and network convergence performance has been greatly improved.Second, this paper considered to use the method of basing on AHP and the fuzzy optimization model. Fuzzy optimization model is an improved method of the fuzzy comprehensive evaluation method. This method has some advantages of a clear physical concept and easy calculation. The previous fuzzy optimization model for the weight of the evaluation factors to use the mean, this paper attempts to use AHP method to strike the weight, and then substituted into the fuzzy optimization model to identify the transformer fault diagnosis. The test result proved that this method can be applied to transformer fault diagnosis.In this paper, these two intelligent diagnosis methods for the power transformer fault diagnosis have some practical meaning and application value. Now, there is no integral method to find out all the fault diagnosis accurately. So the field of intelligent fault diagnosis also needs more in-depth research and development.