基于改进粒子群小波神经网络的智能变电站网络流量预测

Smart substation network traffic prediction based on improved particle swarm wavelet neural network

智能变电站过程层网络流量一旦发生异常,将直接影响继电保护动作的可靠性、快速性和灵敏性,然而目前缺乏针对智能变电站网络流量异常预警的方法.基于此,提出一种基于改进粒子群小波神经网络的网络流量预测模型,为智能变电站网络性能分析预测、网络故障和病毒入侵预警提供决策依据.分析智能变电站网络流量的特点,对流量数据进行归一化处理,建立小波神经网络预测模型,利用粒子群优化算法对传统的小波神经网络模型的网络结构和参数进行优化.在实际智能变电站运行环境中的实验表明,所提模型预测精度高,收敛速度快,提高了智能变电站网络流量预测的准确性和快速性,保障电网安全运行.

Once an exception occurs in the network traffic of the smart substation process layer,the reliability,rapidity and agility of relay protection action will be affected instantly.However,there is no effective method for network traffic warning of smart substation,currently.According to the characteristics of network traffic of smart substations,a network traffic prediction model,which is based on improved particle swarm wavelet neural network,is proposed to assist decision-making for the network performance analysis and prediction,network failures and virus invasion warning of smart substations.The network traffic data is first normalized to feed to the proposed model,which utilizes particle swarm algorithm to optimize the structure and parameters of the conventional wavelet neural network.Experiments conducted in the real smart substation environment validate the high accuracy and fast convergence of the prediction model,so as to improve the accuracy and rapidity of smart substation network traffic prediction and ensure the safe operation of grid.