遗传算法在多目标土地利用规划中的应用

Multi-objective Programming Method for Land Use Based on Genetic Algorithm

在建立和求解多目标土地利用规划模型过程中,采用遗传算法并结合多目标模糊优选理论处理多目标土地利用结构优化问题。其详细步骤为:在分析区域特征基础上构建时间序列;通过现状分析设置土地利用类型决策变量、确定优化目标和约束条件,构建土地利用结构多目标优化模型;用灰色GM(1,1)预测模型对土地利用结构优化多目标模型参数进行预测;用遗传算法多目标优化方法对土地利用结构优化模型求解得pareto解集;最后运用多目标模糊动态优选方法对规划期内各时间段的pareto解集进行模糊动态优选,以最大优属度判定原则确定出最优决策序列。自贡实例表明,该方法有助于解决多目标协调问题,通过遗传算法避免了人为偏好权重的主观性,并通过提供多方案选择性,从而使规划更具灵活性,同时又便于充分吸纳公众参与,减少规划中的不确定性,增加规划的科学性、现实性和可操作性。

A method of multi-objective programming for land use is suggested based on genetic algorithm combined with multi-objective fuzzy optimization. The logical approaches comprise ① building time series based on carefully analyzing land utilization system, ② setting up land use pattern variables and determining optimizing goal and the restriction condition proceeding from the present situation, ③ applying grey GM （1,1） to predict parameters needed by the model, ④studying obtaining pareto non-inferior desegregations by genetic algorithm, and ⑤ seeking optimum planning decision array from pareto disaggregates by means of the principle of maximum dominance. An instance of land use planning in Zigong region through time slicing 2010, 2015 and extending into 2020 shows the genetic algorithm integrated in multi-objective fuzzy optimum programming is conducive to solving the multi-objective coordination by providing more choice schemes that in the meantime also help make more flexibility and expedience to public participants in the process of planning. All procedures make the planning take on the characteristics of scientificity, feasibility and maneuverability.