CAJ | 학술논문

为改进以往神经网络对建筑能耗预测的不足，提出应用遗传算法结合 Levenberg-Marquardt 算法(GALM)改进神经网络对建筑能耗进行预测。首先，利用遗传算法优化神经网络的权值和阈值；其次，利用 Levenberg-Marquardt 算法优化神经网络训练，针对影响建筑能耗的主要因素建立 GALM 神经网络的建筑能耗预测模型。通过建立建筑能耗监测平台采集某公共建筑1个月的能耗数据，对该模型进行训练和测试。实验结果表明，该模型可以准确且高效地对建筑能耗进行短期预测。
위개진이왕신경망락대건축능모예측적불족，제출응용유전산법결합 Levenberg-Marquardt 산법(GALM)개진신경망락대건축능모진행예측。수선，이용유전산법우화신경망락적권치화역치；기차，이용 Levenberg-Marquardt 산법우화신경망락훈련，침대영향건축능모적주요인소건립 GALM 신경망락적건축능모예측모형。통과건립건축능모감측평태채집모공공건축1개월적능모수거，대해모형진행훈련화측시。실험결과표명，해모형가이준학차고효지대건축능모진행단기예측。
In order to improve the conventional method of predicting building energy consumption using ANN, it proposed that the improved neural network optimized by genetic algorithm and Levenberg-Marquardt algorithm, which was applied to predict building energy consumption. First, the genetic algorithm was used to optimize the weight and threshold of ANN, and Levenberg-Marquardt algorithm was adopted to optimize the neural network training. Then, the predicting model based on GALM was set up in terms of the main factors affecting the energy consumption. Furthermore, a public building power consumption data for one month is collected by establishing a monitoring platform to train and test the model. Eventually, the result proves that the model is qualified to predict short-term energy consumption accurately and efficiently.