A Daylight Adaptive Energy Optimization Technique for Indoor Lighting System Using Machine Learning

The power consumption through lighting system in all over the world is around 40 % considering buildings, offices, schools and colleges or even other public institutions. This study proposes to design an effective daylight utilization and enhanced energy saving for indoor lighting environment through applying machine learning strategies like Neural Net Fitting (NNFT), Regression Learner (RL). In addition, the artificial LED lighting has been adjusted based on the diffused indoor light intensity with respect to presence of daylight. The next goal is to reduce overall power consumption by optimizing light intensity for the user in a selected specific room. In the process of developing a model simulation software is utilized that is MATLAB platform. The lux meter is used to collect data of diffused daylight as well as indoor light intensity data. With the change of outer daylight intensity diffused indoor light also changes inside the room. A neural net fitting model has been developed to maintain a constant illuminance in the indoor irrespective of variable daylight intensity. The diffuse indoor light data has been trained with NNFT technique to predict the contribution of illuminance from the artificial LED light. The simulation result shows that a reduction of 26 % energy usage can be achieved by implementing the proposed model in comparison with conventional constant LED lighting system.