基于遗传算法优化神经网络的建筑物电力负荷预测

建立BP神经网络模型,解决了建筑物电力负荷预测由于强耦合性、滞后性和非线性而难于建立模型的问题。利用遗传算法的全局搜索能力对网络模型进行权值优化,解决了传统BP神经网络易陷入局部最优的困扰,使预测更为精准。通过MATLAB软件进行仿真试验,验证了此方法的可行性。

Building Power Load Forecasting Based on Genetic Algorithm Optimizing Neural Network

Through the establishment of tile BP neural network model,this paper solved the building power load forecasting as a result of strong coupling, hysteresis and nonlinear problem which was difficult to establish model. The global search ability of genetic algorithm was used to optimize the weight value. It solved the trouble of traditional BP neural network which was easy to fall into the local optimum,to make the fi）recasting more accurate. Finally,through tile MATLAB simulation experiments,the feasibility of this method is verfied.