一种基于GA-BP自优化的建筑能耗预测方法研究

A Self-optimizing Algorithm Based on GA-BP for Solving the Prediction of Building Energy

针对传统BP神经网络的建筑能耗预测中不变的预测影响因素难以保证预测的准确性和人工确定网络结构耗时长的问题,本文提出一种基于GA-BP的自优化的建筑能耗预测方法。该方法利用遗传算法对建筑能耗BP神经网络预测模型的输入因素和网络结构进行自动寻优确定,有效地减少了最佳预测模型的设计时间,节省了人工实验成本。利用该方法建设的建筑能耗预测系统已应用在某建筑群的能耗预测中,有效地减少了建筑能源浪费。

In traditional prediction of building energy based on BP neural network, the influent factors and the network structure were manually determined. And it always took a lot of time and lead to inaccurate results. In this paper, we proposed a new self-optimizing algorithm based on genetic algorithm （GA） and BP neural network algorithm. The algorithm used GA to determine input influent factors and neural network structure, and saved a lot of time and cost. The algorithm was used to develop the prediction system of building energy, which has been implemented for some buildings and help to reduce energy waste.