Optimal spatial search using genetic algorithms and GIS

Abstract

This study demonstrates that genetic algorithms are capable of producing satisfying results for optimal spatial search under complex situations. We successfully solve a spatial search problem using the proposed method to allocate the facility according to the population constraint from GIS. The search algorithm is very simple using the mechanics of natural selection in biology. The proposed method can be used as a planning tool that can help urban planners to improve development efficiency in site selection. The method is developed by a common computer language which can directly use the full functions of a commercial GA package through the DLL and can import the spatial data from GIS. This integration is useful for solving realistic problems by using large spatial data sets. The programming can be easily adapted to other applications by just modifying the fitness functions instead of changing the model itself. The proposed method has been tested in the city of Hong Kong, a densely populated region. The population data are obtained from the census department and the population density is prepared in GIS as the main inputs to the GA programming.

資源的有效利用和管理往往涉及到空間的優化配置問題。例如需要在空間上確定n個設施的最佳位置。當選址問題涉及多個目標和不同的約束性條件時,就會變得十分復雜。利用一般的brute-force搜索方法無法對涉及高維數據的問題進行求解。利用遺傳算法和GIS結合來解決復雜的空間優化配置問題,具有智能的搜索方法可以大大提高空間的搜索能力。在基于進化的優化過程中,根據GIS的空間數據來計算不同解決方案(染色體) 的適應度。針對不同的應用目的,GIS可以給出不同的適應度函數。實驗表明,所提出的方法比簡單的搜索方法和退火算法有更大的優越性。該方法在處理復雜的空間優化問題有更好的表現。