| Summary: | 碩士 === 國立高雄第一科技大學 === 光電工程研究所 === 96 === This research proposes a newly developed optimization method for light guide plate (LGP) of extra thin back light system. The design goal will be focused at 13 to 15 inches monitor, which might be specific for notebook computer. Traditionally, LGP is optimized either by curve fitting or time-consuming try errors, which might have disadvantages in case uniformity and brightness has to perfectly be balanced. Besides, extra-thin back light system might complicate optical design.
In this paper, the Back-Propagation Neural Network (BPNN) is employed to optimize the distribution of LGP micro-structure in order to achieve the well-balanced for uniformity for backlight systems. In this paper, distribution spacing of each region of the LGP is employed as the input values of the input layer in BPNN. After calculation, luminance values of each regions are treated as the target output values of the output layer of the BPNN; then have both the weighting values and the bias values well and repeatedly trained in the hidden layer of the BPNN in order to get most optimal results.
The experiment shows successful results in system uniformity after finding out the input values with the best inferential output values in BPNN and Genetic Algorithms (GA). Well-balanced uniformity for extra-thin backlight system might be achieved by this newly developed optimization method.
|
|---|
