摘要
针对短期供热负荷控制预测的问题,提出了一种基于改进黏菌算法优化BiLSTM的预测模型。利用猫映射、T分布变异和随机反向学习等改进策略对黏菌算法进行改进,改进后的黏菌算法优化BiLSTM网络参数,构建ISMA-BiLSTM模型,对换热站热负荷进行预测。实验结果表明,ISMA-BiLSTM模型与SMA-BiLSTM、BiLSTM和LSTM模型相比,预测结果更加合理且预测精度有所提高,在短期供热负荷预测中能满足实际工程控制需要。
Aiming at the problem of short-term heating load control prediction,a prediction model based on improved slime mould algorithm(ISMA)was proposed to optimize bidirectional long short term memory(BiLSTM).The slime mould algorithm was improved by using improved strategies such as cat mapping,T-distribution variation and stochastic reverse learning,and the improved slime mould algorithm optimized the parameters of BiLSTM network.The ISMA-BiLSTM model was constructed to predict the heating load of heat exchange station.Experimental results show that the prediction results of the as-proposed model are more reasonable and the prediction accuracy is improved to some extent,compared with SMA-BiLSTM,BiLSTM and LSTM models,so the ISMA-BiLSTM model can meet the needs of actual engineering control in short-term heating load prediction.
作者
薛贵军
赵广昊
史彩娟
XUE Guijun;ZHAO Guanghao;SHI Caijuan(School of Electrical Engineering,North China University of Science and Technology,Tangshan 063210,Hebei,China)
出处
《沈阳工业大学学报》
CAS
北大核心
2024年第4期434-441,共8页
Journal of Shenyang University of Technology
基金
河北省自然科学基金项目(E2020209121)。
关键词
集中供热系统
热负荷
短期供热负荷控制预测
黏菌算法
双向长短期记忆网络
猫映射
T分布变异
随机反向学习
central heating system
heat load
short-term heating load control prediction
slime mould algorithm
bidirectional long short term memory
cat mapping
T-distribution variation
stochastic reverse learning