A Dynamic Model for Predicting Indoor Temperatures based on CBLSTM and Auto Encoder Approach

There are several options for controlling the temperature and humidity within buildings like schools, offices, hospitals, houses, and even grocery stores. Many different theories of control have been suggested and investigated during the past few decades. The most effective method of controlling the temperature in a wide range of building types remains a matter of debate. A straightforward method of regulation is needed to create a comfortable interior environment while reducing energy consumption. The proposed method includes preprocessing, feature selection, and model training. Normalization and standardization procedures are employed in the preprocessing stage. When data is normalized, the maximum and minimum values are swapped out. In the realm of machine learning, standardization is a common practice for preprocessing data. Correlation based feature selection is utilized in the process of feature selection. CBLS TM-AE is used to train the models after the features have been collected. When compared to widely used algorithms like LSTM and AE, the proposed technique comes out on top. The probability of success when employing this method is 95.24%.