Zone Based Control Methodology of Smart Indoor Lighting Systems Using Feedforward Neural Networks

A smart, accurate, and energy efficient control strategy to adjust dimming level of luminaires in an indoor environment is proposed in this paper. The control block in lighting system is nonlinear and time variant, since multiple reflections of objects and daylight variation are related to daytime and they can directly affect the system. According to the complexity of equations which model the lighting system, a control system based on Neural Network (NN) and learning machine is developed. By considering each zone as an independent structure, occupancy in each zone is added. In addition, photodetectors are placed at the work zones and hence increasing the accuracy. The occupancy condition for other zones in the environment are considered as bias to the inputs of the system. Therefore, multiple reflections in the environment are considered in the design of the proposed control method. Accuracy and system performance is improved by separation of control block for each zone as an autonomous control unit, whereas complexity of the system is reduced. The proposed design is evaluated in test beds developed using DIALux and MATLAB. The mean error varies according to the effect of zones on each other. The method is suitable for indoor environment that zones does not have common luminaires. The mean error in the case study that is not proper for the method does not exceed 20%. Although, the error seems to be high but compared to the methods that have ceiling mount sensors is accurate and power and power efficient. Besides, the case with zones that has separated luminaires the mean error is less than 5%.