Recognition algorithm for light intensity variation of LED lamps using neural network with statistics characteristics


Abstract: The primary causes of light intensity variation of LED lamps are input alternating current power supply, and harmonic emission and control mode of LED chips drive circuit. The light intensity variation of different-type LED lamps is different. The light intensity variation recognition of LED lamps is the basis of fault diagnosis, aging evaluation and anti-counterfeiting identification. In this paper, the recognition algorithm for light intensity variation of LED lamps is proposed using BP neural network with statistics characteristics, which is applied to identify LED lamps. Statistics characteristics of light intensity variation of LED lamps are extracted, including skewness, kurtosis, and coefficient of variation. Taking the statistical characteristics as the input node and the type parameter of LED lamp as the output node, the BP neural network model is established for identify LED lamps. To validate the proposed recognition algorithm, the light intensity variation of 10 LED lamps are collected. The experimental results demonstrate the proposed recognition algorithm can effectively identify the type of LED lamps.

Keywords: LED lamp; Light intensity variation; Pattern recognition; Neural network; Statistics characteristics