基于粗糙集和BP网络的微网短期负荷预测

Short-Term Load Forecasting of Micro Grid Based on Rough Sets and BP Neural Network

结合粗糙集和BP神经网络两种智能控制算法提出了微网短期负荷预测模型。首先将影响微网负荷的气象和日类型等因素利用粗糙集建立历史数据属性决策表,通过属性约简算法对其进行属性约简,找到影响微网负荷的核心因素,然后将该核心因素作为BP神经网络的输入量对微网负荷进行预测。BP网络具有收敛速度慢和易陷入局部最优等缺陷,据此提出一种基于模拟退火遗传算法优化的BP神经网络新模型。实验表明,采用粗糙集和改进BP神经网络的新模型对微网负荷进行预测取得了良好的效果,证明了该方法的有效性。

A short-term load forecasting model based on two integrated intelligent algorithms, i.e. attribute reduction of rough sets and BP artificial neural network, is proposed. At first, according to the historical data such as weather and type of day influencing the micro-grid loads, an attribute decision table is built up and the date mining is performed by means of attribute reduction algorithm, thus the kernel factors influencing the loads are determined and using them as the input vectors of the BP artificial neural network the load forecasting is conducted. In order to solve the defects of BP network which has slow convergence and is easy to trap in local optimum, a BP network which is optimized by genetic simulated annealing algorithm is proposed. Forecasting results of calculation examples show that this method has achieved good results in the prediction speed and precision and is suitable for short-term load forecasting.