Identification Of Low-Utility Land In Urban Centres Using Entropy Weight-BP Neural Network

At present,China’s economic development has entered a period of rapid development.The rapid urban development under the new normal must be accompanied by the continuous expansion of urban scale,which directly leads to the inability of urban land supply to keep pace with the demand for urban development land,and the contradiction between land supply and demand has become one of the important factors impeding the operation and development of urban economy.The contradiction between land supply and demand has become one of the most important factors hindering the economic operation and development of cities.Therefore,how to quickly identify and evaluate the low-utility land in the city,further explore the redevelopment potential of urban land,improve the efficiency of land conservation and intensive use,and promote the green,healthy and sustainable development of urban land is of great significance.This paper starts from the basic concept and connotation of urban low-utility land,analyses in depth the spatial and temporal characteristics and evolution process of lowutility land in urban stock land,and constructs a multi-factor coupled evaluation index system that takes into account the internal and external characteristics of the land for the two main types of land,namely residential and industrial,to make up for the shortcomings of the current evaluation object and index selection from a single perspective,and the evaluation system only considers the overall nature of the land parcel and ignores the impact of local neighborhood on low-utility land.To address the problems of high human intervention,inability to reflect the internal linkage of indicators and separation of the weight determination method from the indicator system,this paper proposes a low-utility land identification and evaluation weight determination model based on BP neural network algorithm,which is applicable to internal-external characteristics and integrates the indicator system and indicator weights into a unified whole,so as to improve the objectivity and accuracy of low-utility land identification results.Finally,the study is validated by using the Binhu District of Wuxi City as the research object.The main research contents and results are as follows.(1)Construction of an evaluation index system for the identification of low-utility land in the central area of cityThis paper firstly summarizes and discusses the connotation and definition of lowutility land in the central area of city,analyses the spatial characteristics,attribute characteristics and evolutionary characteristics of low-utility land in the central area of city using the idea of six elements of geographic scenes,summarizes the factors of lowutility land formation in terms of natural factors,economic factors and policy planning factors.From the summarized characteristics and formation factors of low-utility land,the variable factors that can be used to identify and evaluate low-utility land are summarized,the interaction forces between land parcels that are often neglected are considered,the distance-constrained neighborhood influence index factor is added,and the R-type cluster analysis method is used to filter the variable factors and eliminate redundant indicator factors.The system is based on the coupling of multiple elements of the system to construct an index system that can reflect the internal nature and external characteristics of the site.(2)Research on the determination of weights based on entropy weighting method and BP neural networkThe initial weights of internal and external indicators are calculated using the entropy weight method,and on this basis,the initial weights are trained and learned based on BP neural network learning,and the internal relationship between internal and external indices and indicators is established,and the compensation value of feedback weights between indicators is calculated,so as to determine the optimal weight value of indicators.(3)Construction of evaluation model for low utility land identification based on internal-external characteristicsBased on the constructed index system of internal nature and external characteristics,the index evaluation hierarchy was established with reference to relevant national standards and existing research results,and the quantitative model of internal and external indicators was constructed.(4)Case study on the identification and evaluation of low-utility land in Wuxi city centerThis paper used the current land use data,OSM road data,river and water system data,service-oriented POI data and statistical yearbook data,etc.as the basic experimental data,and uses the method proposed in this paper to identify and evaluate low-utility land in residential and industrial land in Binhu District,Wuxi,and validates the identification and evaluation results through Baidu Street View Map.The results show that: the overall utilization efficiency of residential land in Binhu District,Wuxi City is good,mainly with lighter grade and moderate grade low-utility land,among which lowutility land above moderate grade is mainly concentrated in the south and northeast of the study area;the overall situation of low-utility industrial land is poor,among which there are more low-utility industrial land above moderate grade,mainly concentrated in the old industrial park in the northeast corner.