都市計畫草圖替選方案分析模式

Abstract

當規劃者無助地面對一張地圖，不知從何下手配置實質項目時，他最大的願望應該是手邊能有一個可以有效地協助規劃者進行草稿配置作業的分析工具；如此，規劃者就不需要只是窩在辦公室裡，依據自己的好惡，勉強湊出一、二個替選方案應景，而是更有系統地面對並處理市民的感受，藉由先進科技的協助，產生很多相匹敵的替選方案，這樣繼續進行下一步評選與細部設計工作方有意義。 本研究即在構建上述規劃者所期望擁有的分析工具，它是由一個多目標數學規劃模式與一套基因演算法所組成；前者將配置問題提綱挈領地規範成數學規劃模式，後者則負責為這個模式找出近似非劣解集作為替選方案。模式採漸進複雜的方法構建，演算法經過多次測試而確認其求解能力；所需輸入的參數資料，必需經由調查方能獲得，以確實掌握民眾的感受；分析結果為接近柏拉圖最適境的替選方案；整體而言，這個分析工具應該符合規劃者的需要。 為瞭解所構建分析工具在實際問題上的表現，本研究針對「社子島地區」與「淡海新市鎮」二個都市計畫區進行實例研究，並對後者進行參數敏感度分析；除驗證並展示其實用價值外，並發現以下重要結論：一是配置內容愈單純的都市計畫案，愈有可能只找到單一方案；二是分析範圍應以規劃範圍為中心予以適度擴大；三是土地使用與交通運輸兩種手段搭配使用，在提高發展效率的目標上，會較各自單獨運用來得好；四是分區尺度愈小，配置結果將愈細緻明確；五是運算時間、基因字串與分區數之間存在函數關係，可據以由問題規模估計作業時間；六是配置結果確能反映輸入參數值所隱含之規劃理念與偏好期望，且演算結果之穩定度相當高。

When a planner do not know how to start to allocate the physical items and look at the planned area map helplessly, it is wished that an analysis tool can be generated to assist planner to proceed the sketch layout task. With this tool, the planner can sincerely face to and systematically deal with the preference of citizen to generate nondominated alternatives, rather than decide the layout environment pretentiously in the office and generate few alternatives in a forced manner. Thus the next tasks, evaluation and precise design will be more meaningful. The objective of this study is to construct an analysis tool, which is needed by the planners as mentioned above. The tool consists of two elements: a mathematical programming model and a genetic algorithm. The model is constituted from the key factors of layout problem in an increasingly complex process. The algorithm is used to find the approximating nondominated solutions for the model, and its searching ability is identified through many tests. The input data for this tool must be investigated in order to follow the preference of citizen. The alternatives generated by this tool are approximating to the Pareto optimum. We think that this tool is nearly matched with the planners' need. Two case studies, Sheh-Tzyy-Dao and Tan-Hai new town, are described to illustrate the application process and outputs. There are some findings in the case studies. First, the fewer types of land uses the case is, the more possible to generate only one solution. Second, the analysis area should be the appropriate enlargement of planned area. Third, the integrated strategies of land use and transportation have better performance on increasing development efficiency than individual strategies of land use or transportation. Fourth, the smaller the cell is, the more detail and clear the layout will be. Fifth, the function relationships among calculating time, string length and number of cells can be used to estimate operation time from the problem scale. Sixth, the outputs of model do reflect the planning principles and expectation, which are implied in the input parameters, and the algorithm is very stable on the solutions seeking.