Jialing Zhang

The demand for personalized lighting environments based on households is steadily increasing among users. This article proposes a novel intelligent control system for personalized lighting in home environments, aiming to automatically capture user information, such as homecoming time and light switching behavior, in order to train a model that intelligently regulates the lights for users. Facial recognition technology is employed by this system to identify users and record their lighting data. Subsequently, nine commonly used machine learning models were evaluated, revealing that the error back-propagation neural network algorithm exhibits excellent performance in time-series analysis. The BPNN weights were optimized using genetic algorithms, resulting in an improved coefficient of determination (R²) of 0.99 for turn-on time and 0.82 for turn-off time test sets. Furthermore, testing was conducted on data collection duration which demonstrated that even with only 20 time-series data collected from new users, the model still displayed exceptional performance in training prediction tasks. Overall, this system effectively identifies users and automatically adjusts the lighting environment according to their preferences, providing comfortable and convenient lighting conditions tailored to individual needs. Consequently, a broader goal of energy conservation and environmental sustainability can be achieved.