Spatial Analysis Of Urban Transportation And Land Use Integration Based On Neural Network Models

The spatial differentiation of land use induces traffic demand and guides the construction of traffic supply.Urban traffic condition is an important factor that determines land use.A close interaction has been found between urban traffic and land use.Recently,urban big data such as points of interest has gradually accumulated,making it possible to interpret the nonlinear relationship between grid-level traffic and land use indicators,but big data has also brought challenges to system modeling.Previous studies have neglected important phenomena such as the lag of urban construction,the potential transference,the jointy effect of multiple types of land to traffic demand,and the correlation of transportation supply facilities along roads and subway stations.This has led to limited accuracy and affected the analysis of the rule of urban development.Aiming at the problem of ignoring this phenomenon in the existing research,this research explores the improvement of the methods and models.The multi-dimensional long and short memory network(MDLSTM)is used in the analysis of the impact of urban traffic on land use;the improved model-convolutional neural network-multidimensional long and short memory network(CNN-MDLSTM)is purposed in the analysis of the impact of land use on traffic demand;In the analysis of the impact of land use on traffic supply,an improved model.Convolutional Neural Network-Cross Long Short Memory Network(CNN-CLSTM)is applied.Based on the results of the model,the interaction mechanism between traffic and land use is revealed,and the model was verified with the data of three cities as cases,and policy recommendations are also explored.About the impact of urban traffic on land use,this research considers the hysteresis of land use change,introduces the concepts of vitality and potential,and establishs an MDLSTM model that explains the rule between traffic and land use.Results show that the MDLSTM model that takes into account the hysteresis and potential transfer has higher accuracy than the traditional models;the larger the city,the greater the land use potential of the city's leading industry and tertiary industry,and the more significant the impact of traffic.About the impact of land use on traffic demand,this research considers the simultaneous impact of multiple land use types on traffic demand.A CNN-MDLSTM model that explains the rules of land use on traffic demand is established.Results show that the accuracy of the CNN-MDLSTM model that takes into account the simultaneous effects of multiple land use is higher than that of the sub-models(CNN and MDLSTM)and other networks;high vilality grids has greater impact on traffic demand;the mixed land use grids has closer relationship between land use and traffic demand.About the impact of land use on traffic supply,this research considers the characteristic that the high value of the number of transportation facilities is concentrated in the adjacent grids of roads and the grids where the upstream and downstream stations of the rail line are located.A CNN-CLSTM model is established to explain the rule of the impact of land use on traffic supply.The results show that the CNN-CLSTM model considering the above characteristics has higher accuracy than the traditional model;the larger the scale of the urban intensive transportation mode,the greater the development potential.In general,this research uses neural network-based methods at the urban grid level to depict representative phenomena of urban development,analyze the interactive relationship between traffic and land use.The results reflect the model's effective simulation of urban rules.The case study reveals the differences in the rules of different cities to guide a healthy development of cities and transportation.