基于ISSA-LSTM的热舒适短期预测模型

Short-term prediction model for thermal comfort based on ISSA-LSTM

摘要:

为解决在测试日内的短期预测过程中，农村城镇人体热舒适中建筑惰性及人员等随机因素使人体感受变化的样本对预测结果影响大而导致预测精准度低的问题，提出基于改进麻雀搜索算法（Improvement Sparrow Search Algorithm, ISSA）优化长短期记忆神经网络（Long Short-Term Memory Neural Network, LSTM）的方法建立新型户用空调热舒适短期预测模型。首先，对测试日气象数据进行动态性分析，对数据进行有效性验证并构建多种热舒适预测模型；随后选用新型户用热舒适短期预测模型（ISSA-LSTM）对热舒适进行预测。结果表明，模型的最高预测均方误差（Mean Squared Error，MSE）比麻雀搜索算法（Sparrow Search Algorithm，SSA）和蜣螂优化算法（Dung beetle optimizer，DBO）优化LSTM分别提高了0.02296和0.10827，采用ISSA-LSTM方法后改善了短期热舒适预测的精度问题，并提高了分体式空调通过热舒适来控制温度的性能。

Abstract:

In order to solve the problem of low prediction accuracy due to the influence of random factors such as building inertia and personnel in human thermal comfort in rural towns on the prediction results during short-term prediction within the test day, we propose an optimized Long Short-Term Memory Neural Network (LSTM) based on the Improved Sparrow Search Algorithm (ISSA). A new short-term prediction model for thermal comfort of residential air conditioners is developed based on the Long Short-Term Memory Neural Network (LSTM) method. Firstly, we analyzed the dynamics of the weather data on the test days, verified the validity of the data and constructed various thermal comfort prediction models; then, we selected the new household thermal comfort short-term prediction model (ISSA-LSTM) to predict thermal comfort. The results showed that the highest prediction mean squared error (MSE) of the model was 0.02296 and 0.10827 higher than that of the Sparrow Search Algorithm (SSA) and Dung beetle optimizer (DBO) optimized LSTM, respectively. The ISSA-LSTM method improves the accuracy problem of short-term thermal comfort prediction and improves the performance of split air conditioners to control temperature through thermal comfort.