Research On Concentration Prediction Of Gas Dissolved In Oil And Fault Diagnosis For Power Transformer

In the process of building a smart grid with UHV grid as the backbone grid,the scale of the power system is gradually expanding,and the stability of power supply becomes more and more important.The power transformer is the core part of the power system,and its operating state is related to the entire power system stability.When a fault occurs,it will immediately cause power grid to paralyze in the entire area.In the event of a failure,it will immediately lead to a power grid of the entire area.Therefo re,fault diagnosis of power transformers is a very important task.The purpose of analyzing the dissolved gas in transformer oil is to determine the oil-immersed transformer’s operating status and is an important basis for fault diagnosis.Concentration of oil dissolved gas in the transformer is an important basis for transformer insulation fault diagnosis.The concentration prediction of the transformer oil dissolved gas is helpful for timely prediction of failure.In order to solve the disadvantages of the traditional least squares vector machine algorithm applied in transformer gas prediction,such as hard to select parameter,low prediction accuracy and existing concentration prediction models for dissolved characteristic gases lack considerations of the influence of oil temper ature,loads and interaction among gases.Grey relational analysis and least squares vector machine optimized by flower pollination algorithm based prediction of dissolved gas content in transformer oil is adopted Firstly,the thesis makes use of grey relational degree analyses recent dissolved gases analysis data to extract strong-correlated factors.Secondly,a flower pollination algorithm is applied to optimize the parameters of the least squares vector machine.The example analysis results show that the parameters of the least squares vector machine can affect the accuracy of gas prediction,and the flower pollination algorithm can effectively optimize the parameters.Compared with support vector machine and least squares vector machine,the model adopted in this paper has better prediction accuracy.Wavelet neural network is a powerful tool for solving the problem with nonlinearity and high dimension.In order to solve the disadvantages of the traditional wavelet neural network algorithm applied in transformer fault diagnosis,such as uneven sample distribution of training samples,hard to select parameter,long training time.Fuzzy clustering and wavelet neural network optimized by flower pollination algorithm based transformer fault diagnosis is adopted.Firstly,fuzzy clustering is applied to deal with transformer fault sample data so as to remove the bad data;Secondly,a flower pollination algorithm is applied to optimize the parameters of the wavelet neural network.Based on the determination of the input and output the transfer function and the number of hidden neurons of wavelet neural network network.Finally the network model of wavelet neural network is established.The example analysis results show that the parameters of wavelet neural network can affect the accuracy of fault diagnosis,and the flower pollination algorithm can effectively optimize the parameters.Wavelet neural network based on flower pollination algorithm has better convergence,lower diagnosis error rate and shorter training time compared with wavelet neural network based on particle swarm algorithm and it is more suitable for transformer fault diagnosis.