基于PSO的改进BP算法在变压器故障诊断上的研究

Fault Diagnosis for Transformer Based on PSO-BP Algorithm

变压器是电力系统的重要组成部分,其运行状态对电力系统的稳定、安全运行有着重要的意义。研究变压器故障诊断方法,加强变压器的运行维护,可以有效减少故障隐患所带来的安全事故。BP神经网络具有并行分布式计算、自适应、记忆及聚类等诸多优点,能准确表达变压器油中溶解气体与变压器内部故障之间存在的映射关系;但是BP算法存在收敛速度慢、易陷入局部极小点的缺陷,而PSO算法具有全局寻优的能力,可有效地改善BP神经网络收敛速度,提高故障诊断准确率。将BP神经网络模型与PSO算法改进的BP神经网络模型应用于变压器故障诊断,结果表明,故障诊断的可靠性和准确性都得到了明显提高。

Transformer is an important part of power system. Its operation state of power system stability and safe op- eration is of great significance. The transformer fault diagnosis methods can strengthen the operation of the transformer ma- intenance, and can effectively reduce the problems brought about by the safety accident. The BP neural network with adap- tive, parallel and distributed computation, memory and clustering, and many other advantages can accurately express the dissolved gas in transformer oil and the mapping relationship between the internal fault of transformer. But BP algorithm is with slow convergence speed and easy to fall into local minimum point defects. However, the PSO algorithm has the ability of global optimization, which can effectively improve the convergence speed of BP neural network, enhance the accuracy of fault diagnosis. In this paper, the BP network model and the improved PSO algorithm of BP network model is applied to transformer fault diagnosis. The results show that the reliability of the diagnosis accuracy of PSO-BP algorithm is improved obviously.