Neural network modeling of temperature and humidity effects on residential air conditioner load

Seasonal variations are important factors affecting energy consumption patterns. Outdoor weather conditions are crucial in determining residential energy consumption for heating, ventilation and air-conditioning (HVAC) household appliances. In particular, in hot and humid summers, an important proportion of the overall electric power is dedicated to satisfy air conditioner loads. In this paper we address the modeling of outdoor weather conditions impact on predominantly residential air conditioner power usage. The main emphasis is on the temperature and humidity changes. Important field data has been collected for several years covering a wide range of ambient temperature and humidity. After recognizing that humidity can be divided into three different comfort levels, a multiple layer feed-forward neural network model is used to model the two dimensions impact of temperature and humidity. The results are compared to a numerically determined closed form mathematical relation for every humidity level. The conducted test results were satisfactory and the Neural Network modeling gave better representation of the nonlinear relationship between power pattern and outdoor weather conditions.