基于集成智能方法的电力短期负荷预测

将自组织(SOM)和反向传播(BP)两种神经网络结合起来, 并使用模糊理论, 建立了一种基于集成智能方法的日负荷预测智能模型, 该模型首先利用SOM网络的竞争学习能力将历史数据分成若干类别从而找出与预测日同类型的预测类别. 然后, 把温度、日类型等不确定性扰动因素分离出去, 利用BP算法的非线性函数逼近功能, 完成电力负荷的基本分量部分的预测工作. 在处理温度、天气情况、日类型等不确定因素对负荷的影响时, 采用模糊逻辑理论对负荷基本分量进行修正. 提出了一种基于进化树的自组织神经网络算法(SOETA), 该算法是一种无监督基于二叉树的自组织特征映射网络模型, 采用进化思想进行无监督学习, 具有灵活的拓扑结构和精确的模式识别. 本文以2007年厦门市的电力负荷数据为例, 试验结果表明, SOETA+BP+模糊理论的预测精度最优, 有效提高了电力短期负荷预测精度.

A model for short-term load forecasting in power system based on multi-AI methods

A model for short-term load forecasting in power system is developed based on multi-AI (artificial intelligence) methods. Combination algorithms of BP plus fuzzy, SOM plus BP and fuzzy, SOETA plus BP and fuzzy, are applied to forecast the day load of Xiamen. Firstly self-organizing neural network is used to the classification of related data. The historical load data are divided into several kinds, and the certain kind of data is found out, which is as same as that of the forecasting day. Then BP algorithm is used to get basic load forecasting value which takes use of the powerful learning ability and nonlinear reflecting functions. The fuzzy logic is used to construct the function of membership, which can modify the basic load value. Experiment results show that the load forecasting model can acquire satisfactory effect and the combination of SOETA plus BP and fuzzy is the best way, which can improve load forecast accuracy.