UNIFORM LUMINANCE DESIGN OF THE DIRECT-TYPE BACKLIGHT UNIT OF LIQUID CRYSTAL DISPLAYS USING NEURAL NETWORKS

Light efficiency optimization of a direct-type backlight unit (BLU) of the liquid crystal display (LCD) is considered in this paper. The purpose of this research is to obtain the greatest uniformity in a direct-type BLU, while the brightness is maintained in a satisfactory level. It is difficult to find a numerical optimization algorithm that is readily applicable to this problem because of the existence of discrete variables, implicit objective functions and constraints, or even binary constraints. The sequential neural-network approximation (SNA) method is designed specifically for this type of engineering design optimization problem. In this paper, a two variable LCD BLU design example is presented first to illustrate the design process using the SNA method. Then a four variable LCD BLU design optimization problem is solved.