【作者】 杨芳；

【导师】 罗广祥；

【作者基本信息】 长安大学 ， 地图学与地理信息系统， 2014， 硕士

【摘要】 基准地价是政府宏观调控土地市场的重要手段，地价的确定对一个城市的经济发展乃至民生具有十分重要的意义。由于地价影响因素众多，每个影响因素权重的确定以及基准地价评价模型的选择具有很大的主观性、经验性与不确定性，传统的地价评估方法往往费时费力、智能化以及功能化程度低。因此另辟蹊径，研究新的评估模型对解决现阶段基准地价更新工作中的问题具有一定借鉴与参考价值。B-P神经网络作为人工网络的核心部分，由于具有独特的学习能力和信息并行分布处理的优势，被广泛应用于各行各业的研究领域中。神经网络非常适合解决主观性强同时智能化低的多影响因素下的不精确、信息模糊性问题，这就决定了B-P神经网络应用于基准地价更新的可行性与合理性。由于B-P神经网络本身也具有一定缺陷，如何对其进行改进，并能成功地解决实际应用问题也是相关领域的研究热点。本文在深入分析基准地价影响因素、评价模型的基础上，从遗传算法和Levenberg-Marquardt算法的特点和计算过程出发，尝试性地研究了一种基于遗传算法和Levenberg-Marquardt算法相结合改进B-P神经网络的基准地价评估方法，分析论证了所提方法的合理性与可行性，并将其成功应用到延安市城区基准地价评估的实际工作中。更多还原

【Abstract】 Benchmark price is an important means of government regulation of the land market， and it has a great significance for the economic development of a city or even livelihood to determine the premium.There are so many factors influencing of the premium，in the weightsdetermining of every factors and the selection of benchmark price evaluation model there area lot of subjectivityempirical and uncertainty. The traditional land evaluation methods arealways time-consuming、low degree of intelligence and functionality by the mean time. So ithas certain reference value to solve the updating problems of the benchmark price in another new assessment model.For the unique learning ability and the advantages of parallel distributed processinginformation，the B-P neural network as a core part of the artificial network is widely used inall walks of life research. Neural network is suited to solve the inaccurate informationambiguity problem of multi-factors that have imprecise subjectivity and low intelligent，andthis determines the feasibility and rationality of B-P neural network benchmark premiumupdating. Since the B-P neural network has some flaws，and how to improve it andsuccessfully solve the practical problem is the hotspot in some related research areas.Based on the analysis of the influence factor and evaluation modelabout benchmark landprice，start from the characteristic and counting process of GA and LM arithmetic，study theassessment method of benchmark land price based on the B-P neural network combined GAwith LM arithmeticin tentative，analyze and demonstrate the rationality and feasibility of thismethod，apply to the update of datum land pricein Yan’an city successfully.更多还原