Prediction of Cooling Load Demand of Target Building Based on Bidirectional Recurrent Neural Network

Prediction of Cooling Load Demand of Target Building Based on Bidirectional Recurrent Neural Network

연구 분야: Artificial Intelligence

학회: 2023 International Conference on Applied Physics and Computing (ICAPC)

In the building, to determine which refrigeration units and operating parameters to turn on or off, the facility personnel need to make assumptions about the cooling load demand and outdoor environmental factors in the future time, but all assumptions cannot reflect the actual situation, such as assuming that the building cooling load demand is too high will cause energy waste, or when the assumed cooling load demand is too low. The indoor temperature is too high, affecting the comfort level. This paper comprehensively considers the influence of temperature, humidity, ultraviolet intensity, and rainfall on the cooling load demand of the target building, and proposes a method based on a bidirectional recurrent neural network to forecast the cooling load demand of a building. The results show that the method proposed in this paper has lower prediction error and better prediction effect than other neural network prediction models.