Techno-economic analysis and thermal-electrical demand optimization of a sustainable residential building using machine learning approach

The progression of data-driven methods has been expediting within the past few years, and the application of machine learning methods in relation to load forecasting, energy consumption optimization, and predictive control models has been expanding as well. In that matter, several investigations have attempted to promote the energy-saving potential of building energy consumption due to the optimization of influential parameters. Various machine learning methods have been utilized using different training datasets such as environmental parameters, geometrical characteristics or occupancy profiles, etc. However, the impact of heating, ventilation and air conditioning (HVAC) setpoint temperature in combination with other parameters has been infrequently investigated. In the present study, a neural network approach has been utilized to achieve thermal comfort-oriented optimized load to implement as input data for sizing renewable energy hybrid systems due to a sustainability point of view. In that matter, the required training data set was acquired due to building energy model's (BEM) results from Energyplus Software. The outdoor dry-bulb temperature, the relative humidity, thermal comfort, and HVAC equipment setpoint temperature were considered as the training set's key features. The results revealed that the energy-saving potential can be reached at 4.7% and 64% for hot and cold seasons, respectively, which indicates a significant opportunity for HVAC load optimization. Moreover, the economic and environmental analysis was performed on the optimized hybrid renewable energy system.

Automated summary

The progression of data-driven methods has been accelerating in recent years, with machine learning methods being increasingly applied to load forecasting, energy consumption optimization, and predictive control models. However, the impact of HVAC setpoint temperature in combination with other parameters has been rarely investigated. This study utilized a neural network approach to optimize load based on thermal comfort, and the results showed that significant energy-saving potential exists in HVAC load optimization. The study also conducted economic and environmental analysis on the optimized hybrid renewable energy system.