利用人工神经网络进行国有土地价格评估的探讨

Research of Appraising State Owned Land Price with Artificial Neural Network

在研究国内城市土地评价方法和分析了山西省自然、经济和社会状况的基础上,提出利用人工神经网络、结合特尔菲法建立城市国有土地资产总量评估指标体系,以城市的综合平均地价为评价目标的国有土地资产总量评估方法。本文在模型建立过程中,以山西省22个城市中的18城市的资料为学习样本,以20个城市评估指标为输入,以城市综合平均地价为输出,建立了三层BP人工神经网络,并以其余4个城市的资料作为验证,取得了较好的模拟效果。有效地解决了利用有限的城市(镇)地价资料,结合城市(镇)统计资料,进行国有土地价格的评估,进而为国有土地资产总量的评估提出新的探讨途径。

Based on analysis of the national urban land evaluation methods and natural, economic and social states of Shanxi Province, the paper introduced artificial neural network and Delphi approach to establish the evaluation index system of land assets considering complex average land price of urban as its evaluation goal. The model of 3 layers of BP network was established based on the data sets of 18 of 22 cities of Shanxi province, inputing value of 20 evaluation index of cities and outputing value of complex average urban land price. The simulation results were well validated by the data sets of another 4 cities. It showed that the model can effectively solve the evaluation of national land assets and complex average land price of cities using the limited urban land price and statistical data of cities.