GIS母线接头过热已成为典型的过热性故障,为实现GIS母线接头温度的准确预测,本文在研究最小二乘支持向量机(Least squares support vector machines, LSSVM)算法的基础上,引入混沌理论改进的人工蜂群算法对LSSVM的参数进行优化,建立了一种基于参数优化LSSVM的GIS母线接头温度预测模型。通过GIS母线温度物理模拟实验,将实验所获得的负荷电流、GIS母线筒外壳测点温度及环境温度作为输入量,GIS母线接头温度作为输出量,对该模型进行了训练。结果表明,该模型的预测误差仅为0.193%,优于ABC-LSSVM、LSSVM和RBF神经网络。本文提出的温度预测模型可实现母线接头温度的精确预测,为防止GIS母线接头过热性故障的研究奠定了基础。
英文摘要:
Overheating of the GIS bus junction has become a typical overheating fault. In order to accu-rately predict the realization of GIS bus connector temperature, On the basis of Least squares support vector machines (LSSVM) algorithm, this paper introduces chaos theory and improved artificial bee colony algorithm to optimize the parameters of LSSVM, and a temperature predic-tion model of GIS bus joint based on parameter optimization LSSVM is established. Through the GIS bus temperature rise physical simulation experiment, the model is trained by the load current, the temperature of the GIS bus shell, the temperature of the measuring point and the ambient temperature as the input, and the temperature of the GIS bus junction as the output. The results show that the prediction error of this model is 0.193%, better than that of ABC-LSSVM, LSSVM and RBF neural networks. The temperature prediction model proposed in this paper can accurately predict the temperature of bus bar connector, thus laying the foundation for avoiding overheating faults of GIS bus connector.
