基于人工神经网络的建筑多目标预测模型

Building multi-objective predicting model based on artificial neural network

为得出一种能快够速且准确预测建筑能耗和室内热舒适状况的方法,提出应用人工神经网络来预测建筑能耗和室内热舒适状况的方法,并通过遗传优化算法对神经网络的连接权进行优化;其次,对影响建筑能耗和室内热舒适状况的主要因素进行分析,并针对这些主要因素建立基于GA-BP网络的建筑能耗和室内热舒适状况的预测模型;结合EnergyPlus模型计算所得出的144组样本数据,训练和测试所建立的住宅建筑能耗和室内热舒适状况的GA-BP网络模型,测试结果表明该模型有较高的预测精度。该预测方法的建立使建筑师在设计阶段能够简单且准确地获得设计建筑的能耗和室内舒适状况,从而使设计向着有利于建筑节能和改善室内热环境的方向发展。

In order to predict energy consumption and indoor thermal comfort quickly and accurately, a method applying artificial neuron network （ANN） was proposed. Meanwhile, the connect weight of ANN network was optimized using genetic algorithm. Furthermore, the main influence factors effecting the energy consumption and indoor thermal comfort was analyzed, it was put forward in terms of the main factors that the predicting model was based on the GA-BP network to predict energy consumption and indoor thermal comfort. Eventually, the GA-BP network model was trained and tested with 144 samples which was calculated by the EnergyPlus software, and the result proves that the model predicts energy consumption and indoor thermal comfort with high accuracy.