智能电网超短期电力负荷预测ELM方法,效果超级好.docVIP

智能电网超短期电力负荷预测S/C-WMM-ELM方法* 摘要：随着对电能质量要求的提高以及各项技术的发展，智能电网应运而生。它具有不同于传统电网的清洁性、安全性以及交互性，在一定程度上改变了电力系统的运行方式和供需关系。特别是，由于实时交互性的要求，需要信息在供电侧与消费侧实时、高速、双向的传输，从而赋予了智能电网下电力负荷预测新的特点。本文针对智能电网中要求负荷预测的实时性，提出基于更新准则的加权多模型极限学习算法。该算法建立多个前馈神经网络模型，基于更新准则将模型集分成三部分：即时更新模型、批次更新模型和挂起模型。任一时刻对于该时刻输出误差较小的即时更新模型，无需采用随机方式进行在线更新；对于每一时刻都出现较大误差的批次更新模型，则利用新来的真实数据，用随机方式重新在线计算；对于输出误差较大的挂起模型，采用随机方式进行在线更新。因其在训练过程中无需迭代学习即可确定网络参数，达到了节省训练时间，提高预测快速性的目的。最后分别采用平日、节日、假日等多时间段数据进行应用研究，结果表明无论在拟合精度，还是计算时间上都满足要求。 关键字：智能电网，负荷预测，超短期，加权多模型，极限学习机 S/C-WMM-ELM Method for Ultra-short Term Load forecasting in Smart Grid Abstract：With the increase of power quality and the development of technologies，the smart grid’s different from the traditional power grid with cleanliness，safety and interaction . the supply and demand. As the requirements of real-time interaction, information is needed in a real-time, high-speed and two-way interaction way between the supply side and consumer side. So, it gives the new features to the smart grid. In this paper, Weighted Suspend Multiple Models Extreme Learning Machine is proposed for its requirement of real-time. It establishes Multiple Single Layer Feedforward Neural Networks (SLFNN). Then a generalization suspending criterion is designed to separate the whole models into three parts: the sequential single models, the sequential chunk model and the updating models. For the sequential single models with minor error in this moment, it needn’t adopt the random selection method to update online, which is used to decrease the overflowed output. For the sequential chunk models with major error in every moment, the latest real data can be used to re- calculate by random method. For the updating models with major error, it must utilize the random selection method to update the input weighting matrix to ensure the adaptive characteristics of the multiple models. Due to the network parameters is determined without iteration, this algorithm save much time and ensure the fast forecasting. The forec