Research On Prediction And Dispersion Of Burglary Risk Based On Diffusion Model

With the development of urbanization,industrialization and informationization in our country,crime is still the main problem faced by urban residents and the public security department.How to overcome it is a matter of concern to researchers.Crime does not happen randomly in the city,but related to environmental factors such as population and economy in study area.Under specific geographical environment,it is characterized by a specific space-time distribution.Understanding the space-time distribution of burglary by considering environmental characteristic of geographical space which contains the burglary is of great significance for crime prevention for the public security department,especially in the face of police shortage,increasing burglary prevention and hard detection.This paper focuses on crime prediction based on the diffusion model,combining with the characteristic of the space-time distance of the transmission of crime risk,the crime data with time and space,space constraints,as well as the social environment of geographic units in the area,to estimates the crime risk in a microgeographic space.The results of this research improve the accuracy and spatial resolution of the crime prediction.It is of great academic value and application value in saving the public security department and preventing crime.The specific research work and results can be summarized as the following:(1)In view of the traditional diffusion model seldom considering geographical environment characteristics,this paper applies a crime prediction method that takes geographical environment characteristics into account.The crime has different time-space distribution forms in different geographical environment.Traditional diffusion model describes the space and time of crime risk dynamic,but lack of the information of geographical background.So,it is difficult to further improve the prediction of the spatial resolution and accuracy.In order to solve this problem,this paper puts forward a crime prediction method from the perspectives of both space-time and environmental similarity,based on the anisotropic diffusion model,combining the characteristic of the space-time distance of the transmission of crime risk,establishing the environmental features of the geographic unit in which the case is located,and using the diffusion coefficient function to introduce environmental features for traditional diffusion model.Finally,the effectiveness of our proposed method is verified by experiments.(2)Considering that traditional crime prediction method having low prediction accuracy in micro scale geographic space,this paper puts forward a crime prediction method based on non-homogeneous diffusion model.Different from the crime prediction at the macro scale,the traditional space-time process analysis and regression model at the micro scale are difficult to achieve a good effect of crime prediction.This paper summarizes the theories related to the daily activities of criminals in criminology.And the paper proposes crime risk estimation model based on the non-homogeneous diffusion equation.This model uses homogeneous part to describe crime risk of local transmission,and at the same time uses non-homogeneous to model the overall transfer of crime risk which always to be overlooked,in order to deal with variation in spatial distribution in the micro space,trying to fully understand space-time distribution of the crime.The experimental results show that compared with the crime prediction model based on local transmission of crime risk,this method can better deal the variation of spatial distribution of crimes in micro scale geographic space.(3)Considering the absence of police deployment planning in the micro-scale at the present stage,this paper puts forward a dynamic police deployment multi-objective optimization taking into account the police workload.Based on the existing mathematical model of police deployment optimal,this paper uses the crime risk map with higher spatial resolution in order to directly calibrate the geographic units that need to be used for crime prevention.It increases constraints on police workload and simulates the influence of police force on the spatial distribution of crime risk.A multi-objective optimization problem of police force deployment at the micro scale geographic space is proposed to balance the contradiction between the efficiency of crime control and the workload of police.Police deployment optimization model is base on the amount of police resources,the police work load,crime risk,and the allocation of police resources,to optimie the deployments the police in micro scale space.Experiments show that this optimization method can improve the efficiency of the police law enforcement and provide operable solution for burglary prevention.