基于麻雀搜索算法优化的BiLSTM建筑能耗预测模型

BiLSTM energy consumption prediction model for buildings optimized by sparrow search algorithm

建筑能耗的准确预测对建筑能源的合理规划至关重要,为了在建筑能耗预测过程中选择重要的影响因子并提高建筑能耗预测模型的预测精度,提出基于熵加权K-means与随机森林相结合的特征选择方法和麻雀搜索算法,优化双向长短时记忆网络的预测模型(SSA-BiLSTM)进行建筑能耗预测。结果表明,能耗影响因子经过特征选择后,预测模型的计算精度显著提高,与单一的BiLSTM预测模型相比,SSA-BiLSTM预测模型在不同季节的能耗预测中均展现出良好的预测效果。

Accurate prediction of building energy consumption is crucial for the rational planning of building energy resources.To select important influencing factors and improve the prediction accuracy of the building energy consumption model during the prediction process,the paper proposes a feature selection method combining entropy-weighted K-means and random forest,and a prediction model optimized by the Sparrow Search Algorithm(SSA)for the Bidirectional Long Short-Term Memory Network(BiLSTM)for building energy consumption prediction.The results show that after feature selection,the computational accuracy of the prediction model is significantly improved.Compared with a single BiLSTM prediction model,the SSA-BiLSTM prediction model demonstrates good predictive performance in energy consumption prediction across different seasons.