基于长短期记忆网络的工业空调启动时间预测

Start-up Time Prediction of Industrial Air Conditioner Based on Long Short-Term Memory Network

为了降低企业生产车间空调能耗,基于长短期记忆(LSTM)网络构建了一种工业空调启动时间预测模型。使用该模型对车间空调提前启动时间进行预测,并将预测结果应用于车间空调系统的启动控制,以达到节能目的;采用平均绝对百分误差(MAPE)对预测模型进行整体误差评估,实验结果表明:LSTM较好地解决了生产车间空调系统启动时间预测问题,相较于传统预测方法有着更小的MAPE。优化控制后的空调系统能够在保证车间生产环境达标的同时,降低空调系统约27.9%的能耗。

In order to reduce the energy consumption of air conditioners in production workshops,an industrial air conditioning energy consumption prediction model based on long short-term memory(LSTM)network is constructed.The model is used to predict the early start-up time of the workshop air conditioner,and the prediction result is applied to the startup control of the workshop air-condition⁃ing system to achieve the purpose of energy saving.The overall error evaluation of the forecast model is carried out by mean absolute percentage error(MAPE).The experimental results show that LSTM better solves the problem of starting time prediction of air-condi⁃tioning system in production workshop,and has a smaller MAPE than the traditional prediction method.The optimized air conditioning system can reduce the energy consumption of the air conditioning system by approximately 27.9%while ensuring the production envi⁃ronment of the workshop.