本文探討建構LED路燈透鏡與陣列結合為照明燈具的程序，在符合道路環境需要及照明法規的基礎之下，予以最佳化照明的性能。根據二次光學的原理建構單顆LED透鏡之外型曲面，再以工程製圖軟體建立實體模型。為了配合路燈照射在長方形路面的範圍，將計算所得的多顆透鏡排列成矩型，其排列方法是依據照射面積的長與寬的比例，設定為3.5:1。 LED路燈透鏡的建構及分析時，因子與性能間之數學運算及相關程序，並非顯性且為連續的數學，欲進行最佳化程序時構成困難，所以本文採用實驗計畫法的因子分析方法。依據我國法規CNS，照明的均勻度需超過法規所定的最小值以上；照度需達到法規的標準定值，效率需越大越好；光度分布需達到法規要求。於最佳化之性能分析時，可轉換為最小化LED透鏡的顆數及最大化燈具的均勻度。由實驗計畫法建構燈具的均勻度及透鏡顆數的兩個反應曲面式，形成兩個設計目標，再執行雙目標最佳化。由最佳設計結果的性能分析及比較，可得到優於其他相似燈具的性能，也驗證本文的LED路燈照明之實驗設計最佳化程序的適用性。 A completed process of constructing street fixture including LED lens and array in the lamp is introduced and developed. This process is based on public road lighting regulations and specifications to satisfy real road environmental needs. Further optimization can be utilized for improving several primary lighting performances. The beginning process contains the use of secondary optical theory for building the surface-contour of a single LED lens; then the engineering analysis model can be built by computerized graphical software. To mach the rectangular lighting area on road, a rectangle array of 3.5:1 ratio of a group lens is introduced herein that synthesize the lighting fixture. During the analysis of single lens, there exist several implicit and non-continuous mathematical expressions describing the whole computation process. It is impossible of utilizing these expressions for further applying in the work of optimization. In this thesis, the formal experimental plan and analysis are adopted to obtain the explicit expressions between performances of lighting fixture and five factors. According to lighting regulations of Taiwan, the degree of uniformity of lighting fixture has to be greater than a specific value and as larger as possible. The illuminance must be equal to a specific value in written regulation. The photometric distribution must achieve to specific value. The efficiency of lighting fixture required to be as high as possible. The above three main objectives can be transformed to the minimum number of LED lens and maximum value of uniformity of lighting fixture. Those two objectives can be constructed as two response functions by DOE method and software. Then the global criterion strategy is used for solving this two-objectives optimization problem. From the comparison and discussion shows that the final result is better than the similar fixtures in published pagers. It also shows that the presented process is suitable for building a road lamp and fixture.