Influence of the equivalent electric load strategy on energy demand forecasting

In recent years, interest in artificial intelligence has been increasing in the energy sector, with the aim of improving efficiency and decreasing overall consumption. Innovative methods such as pattern recognition, artificial neural networks or other machine learning methods are being investigated to improve the accuracy of forecasting. Moreover, a suitable operation strategy may improve the efficiency of energy systems. The electric equivalent load is an operation strategy proposed for systems where electricity and heat (or cooling) are requested simultaneously, and there is equipment such as a heat pump to cover the heating (or cooling) load. In this paper, the electric equivalent load is defined to cover both the electricity request of the user and the electricity consumption of the heat pump. Thus, the energy system follows the electric equivalent load to cover both electricity and heating/cooling requests. This study aims to analyse how the accuracy varies when forecasting energy requests (e.g. electricity and heat) separately with respect to forecasting directly the electric equivalent load. The results show that electric equivalent load forecasting averages the accuracy of electricity and heat, improving the operation system by decreasing the electricity demand from the grid and increasing energy production during peak hours.