Research On Prison Risk Forecast And Security Patrol Planning Method

As a place where criminals are held,the security of prison has a great impact on social stability.At present,although the security of prison has been greatly improved compared with before,there are still many problems,showing that the security of prison is still facing a severe test.In recent years,with the continuous maturity of artificial intelligence technologies such as big data and machine learning,people begin to gradually apply some new technologies to multiple fields of social development,among which smart prison is an innovative application in the field of artificial intelligence in the current era.At present,the risk forecast and security patrol of prison are still dominated by traditional manual disposal,which is characterized by strong subjectivity and poor real-time performance.There are many problems such as delayed detection of hidden risks and low efficiency of security patrol.Therefore,aiming at the above problems,this thesis proposes a risk spatio-temporal forecast method and an optimal route security patrol planning method for prison.Firstly,the risk scenario area is forecasted in time according to the risk forecast algorithm,and then the optimal route security patrol of the risk scenario area is conducted through the patrol planning method.The whole process is oriented by science and technology.It has changed the traditional working mode of manual disposal,which can effectively decrease the occurrence of risk accidents in prisons and improve the level of intelligent management in prisons.The main contributions of this thesis are as follows:(1)A risk forecast model for spatio-temporal graph convolution based on multi-feature attention mechanism is proposed.Aiming at the problems of late risk discovery and low risk forecast accuracy in prison,this thesis proposes a spatio-temporal graph convolution risk forecast model based on multi-feature attention mechanism.The model is composed of different feature components independently,and each feature component contains four different modules:spatio-temporal data graph construction,spatio-temporal attention mechanism,spatio-temporal graph convolution network,and feature fusion output.Through the spatio-temporal attention mechanism module,the model can be more focused on the input of spatio-temporal information in the prison.Through the spatio-temporal graph convolution network module,the model can be correlated to extract spatio-temporal features,and the dependence of features in spatio-temporal dimensions can be deeply explored.Through the feature fusion output module,the forecast results of risk features can be fused and output according to different weight learning parameters.Thus,the final risk forecast result is obtained,and the accuracy of the risk forecast result is effectively improved.(2)A security patrol planning model of optimal route for prison based on improved ant colony algorithm is proposed.Aiming at the low efficiency of security patrol in prison,this thesis proposes a security patrol planning model of optimal route based on improved ant colony algorithm.According to the risk forecast model,the risk scenario area can be forecasted.For the risk scenario area,security patrol can be conducted through the patrol planning model.Through this model,an optimal patrol route with less time,short distance and less personnel input can be planned for the police in the prison,which greatly improves the work efficiency of the police and reduces the waste of police resources in the prison.(3)A risk forecast and security patrol planning system is constructed.Based on the risk forecast model and the security patrol planning model,the visualization system of risk forecast and security patrol planning is constructed,and the system is applied to the smart prison big data visualization platform system,and the risk forecast analysis of the prison and the patrol route planning of the risk area are realized.