基于IGWO-BP和超声衰减法的变压器油氧化安定性检测方法研究

Research on Transformer Oil Oxidation Stability Detection Based on IGWO-BP&Ultrasonic Attenuation Method

为实现对变压器油氧化安定性的快速准确检测,提出了基于改进灰狼算法(IGWO)优化BP神经网络和超声衰减法的变压器油氧化安定性检测方法。首先,利用超声波在变换器油中的衰减特性,检测得到超声波的衰减系数等特性参数;然后对灰狼算法的收敛因子进行改进,并利用改进后的灰狼算法优化BP神经网络,得到收敛速度快、全局寻优能力强和预测精度高的IGWO-BP算法;最后,利用IGWO-BP算法训练样本,得到变压器油氧化安定性的检测模型。试验结果验证了方法的可行性。

In order to realize quick accurate detection of the oxidation stability of transformer oil,a method was proposed,based on improved grey wolf algorithm optimized(IGWO)-BP neural network and ultrasonic attenuation method.Firstly,the attenuation coefficient of the ultrasonic wave was measured by using the attenuation characteristic of ultrasonic wave in transformer oil.Then,the convergence factor of the grey wolf algorithm was improved,and the improved grey wolf algorithm optimization-BP neural network was used to obtain IGWO-BP algorithm with fast convergence speed,strong global optimization capability and high prediction accuracy.Finally,the IGWO-BP algorithm was used to train samples to obtain a detection model for the oxidation stability of transformer oil.The feasibility of the proposed approach was verified through the experimental results.