基于遗传算法和支持向量机的城镇土地定级方法研究

A study on the method for grading urban land based on genetic algorithm and support vector machine

基于支持向量机理论,采用遗传算法进行全局最优参数选择,构建城镇土地定级模型.以某市中心城区建成区及规划区为研究区,通过GIS空间分析技术构建定级因子属性数据训练样本和实验样本,实现基于遗传算法优化支持向量机的商服用地定级.在此基础上,将基于遗传算法优化支持向量机的城镇土地定级方法和传统的多因素综合评定定级方法结果进行对比分析.结果表明,基于遗传算法优化支持向量机的城镇土地定级方法效率较高,推广性较强,能够客观反映城镇内部土地使用价值差异.定级结果符合研究区实际情况,是实现城镇土地定级的有效手段.

Traditionally, multi-factor comprehensive evaluation and spatial clustering methods are employed in the evaluation of urban land gradation. Here, we build an urban land gradation model based on support vector machine and using genetic algorithm （GA-SVM model） to select global optimal parameters. Taking the built and planned area of a city in Guangdong province as a case study area, we selected the factors for grading the commercial land in the city on the basis of the ＇regulations for gradation and classification on urban land＇ and the characteristics of the study area. Then, we built training and testing samples of land gradation factors according to GIS spatial analysis techniques. The values of the nine factors for grading commercial land were inputted into the GA-SVM model to determine the grade of the commercial land. Finally, to confirm the fitness of the GA-SVM model we proposed, we compared the results of the GA-SVM model with the gradation gained from the traditional multi-factor comprehensive evaluation. This comparison showed that the GA-SVM model can accurately reflect the distribution of urban land quality with high efficiency and strong generalization performance. Furthermore, the gradation results gained from the GA-SVM model were in accordance with the actual situation of the study area. Therefore, we conclude that the GA-SVM model is as effective method for grading urban land.