Dynamic Model Identification of High Power LED Lighting System Based on BP-Network

In this paper, a system identification method based on BP-network is proposed to realize the study of dynamic characteristics of high power LED luminaire. LED input-output characteristic relations are obtained according to the lighting system structure, and response curves are received using the method of input isolation. A Back Propagation network model is established to simulate input and output relationships of actual system using the approximation ability to nonlinear mapping which multi-layer network has. The neural network is trained to make the input and output of network infinitely close to that of the practical system, and make the network characteristics approximately same to the actual lamps. The research provides the theory basis to establish feedback control system of the LED lighting luminaire.