Analysis And Application Of Abnormal Behavior For Expressway Toll Data

The construction of expressway network is an indispensable booster for regional economic development.At the same time,various forms of toll evasion also emerge in an endless stream,causing a huge loss of national financial revenue.With the wide popularization of the expressway network toll system,every passing and payment behavior of vehicles can be tracted.By analyzing the relevant fields in the toll history,we can find the abnormalities.However,the existing manual inspection method is inefficient.How to detect the abnormal data in the expressway network toll collection system with the help of data mining technology,and then analyze the abnormal behavior characteristics of the vehicles reflected by the abnormal data,are the key problems to be solved by the current traffic supervision departments.Therefore,this dissertation combines the data mining technology with the expressway abnormal behavior analysis business,and designs a specific algorithm model,which is applied to the abnormal detection of expressway network toll data and the analysis of vehicle abnormal behavior characteristics.The main work of this dissertation is as follows:(1)For the problem of high dimension of dataset,a feature selection algorithm based on peak density is proposed.Taking the maximum information coefficient of features as the measurement,and based on the idea of density peak clustering,the importance index of features is defined,and the feature subset after dimension reduction is obtained.Experiments show that the feature subset can represent the features of all the data.Finally,the dimension of Expressway data set is reduced.(2)For the problem that the effect of DBSCAN clustering algorithm is greatly affected by the input parameters,K-means algorithm is proposed to adaptively determine the value of DBSCAN parameters.For the problem of high time complexity of DBSCAN algorithm,the neighborhood query method is improved.Finally,experiments show that the improved DBSCAN algorithm can improve the clustering effect and execution efficiency.(3)In order to solve the problem that a large number of candidate itemsets are generated by the operation of frequent itemsets in Eclat association rules algorithm,the pruning optimization of frequent itemsets is carried out based on the pruning idea of Apriori algorithm.For the problem of redundant rules generated by Eclat algorithm,the interest model is introduced.The experimental results show that the improved algorithm can improve the operation efficiency and produce more valuable association rules.(4)This dissertation designs and implements the expressway abnormal behavior analysis system,which integrates three main function modules: data management,data analysis and result display.Users can choose the algorithm model according to the business requirements and analyze the business data,which has high practical application value.