【摘要】 电力三产用电量预测模型,基于自适应神经网络模糊推理系统(ANNFRS)原理及电力负荷的历史数据建立。该系统由 7 个隐含层数,5 个隶属度函数组数据模型组成。网络使用 Levenberg-Marquardt 算法训练,每个局部映射参数采用线性最小二乘估计算法。仿真结果表明拟合误差的数量级为 10-5,能满足负荷预测要求。更多还原

【Abstract】 In adaptive NN-based fuzzy reasoning system (ANNFRS), the predicting model of power consumption for electric three-industry was structured according to principle of the system and history data of electric load. It consists of 7 pieces of hidden layer and 5 pieces of data model with membership grade function array. The network learning was trained with Levenberg-Marquardt algorithm, every local mapping parameter was estimated with linear least squares solution algorithm. The simulation results show that the quantity level of error is 10-5, which can satisfy the demands of load forecast.更多还原