论文部分内容阅读
人工神经网络技术已经在变压器的状态诊断得到应用,为了克服故障分析中BP神经网络存在的不足,提出了一种自适应混沌粒子群优化神经网络在变压器故障诊断的新方法。该算法通过进化速度因子和聚集因子调整惯性权重,并改进学习因子,引入混沌系统,构成混沌粒子群算法优化神经网络参数,有效地克服常规BP算法训练收敛速度慢、易陷入局部极小值等缺点。最后基于DGA对变压器故障实例分析仿真,对比常规变压器诊断方法结果表明,该算法能够提高诊断效率以及故障模式识别的准确性。
Artificial neural network technology has been applied to diagnose the status of transformers. In order to overcome the shortcomings of BP neural network in fault analysis, a new adaptive chaos particle swarm optimization neural network is proposed to diagnose faults of transformers. The algorithm adjusts the inertia weight by the evolutionary speed factor and the clustering factor, and improves the learning factor. The chaotic particle swarm optimization algorithm is introduced to optimize the neural network parameters. This algorithm can effectively overcome the slow convergence of traditional BP algorithm and easily fall into the local minima Disadvantages. Finally, based on DGA, the transformer fault analysis is simulated. Compared with the conventional transformer diagnosis method, the results show that this algorithm can improve the diagnostic efficiency and the accuracy of fault pattern recognition.