配电网运维成本研究与预测

Research and Prediction of Distribution Network Operation and Maintenance Cost

可靠性价值分析是配电系统规划和运行的重要工具,中断成本预测模型直接影响可靠性价值评估的准确性。基于此,利用径向基函数(RBF)神经网络和正交最小二乘(OLS)学习方法提出了2种中断成本预测模型,包括平均或聚合模型(AAM)和概率分布模型(PDM)。利用所提神经网络技术对AAM和PDM中的居民和工业中断成本进行集成,采用蒙特卡洛时间序列模拟技术进行价值评估预测,并通过评估台电系统安装断开开关、横向熔断器、变压器及备用电源的可靠性价值,对该技术进行测试。结果表明,2种成本模型产生了不同的中断成本,PDM对系统的建模更符合实际。

Reliability value analysis is an important tool for planning and operating distribution systems,and the interruption cost prediction model directly affects the accuracy of reliability value assessment.Based on this,this paper uses radial basis function(RBF)neural network combined with orthogonal least squares(OLS)learning method to construct two kinds of interruption cost prediction models,namely average or aggregate model(AAM)and probability distribution model(PDM).The proposed neural network technology was used to integrate the costs of residential and industrial interruptions in AAM and PDM,and Monte Carlo time series simulation technology was used for value evaluation and prediction.The reliability value of the installation of disconnect switches,transverse fuses,transformers,and backup power sources in the Taiwan Power System was evaluated,and the technology was tested.The results indicate that the two cost models generated different interruption costs,and PDM′s modeling of the system is more in line with reality.