The Application And Research On The Reuse Of The Expressway Network Toll Data Disposal With Genetic Programming Method

Genetic programming (GP) has had many applications and research on the fields of prediction and classification now. Since GP fulfills the aims of prediction and classification by creating concrete prediction and classification functions, GP is a very proper method to the problem which are concerning of both the accuracy and the relationship between the input(prediction, classification) variables and the output variables. This article introduces the prediction method and improves an multi-classification method based on GP method. Moreover, with the help of the symbolic regression ability of GP, the paper also bring forward a kind of integer linear programming method based on GP method. To validate the validity of the algorithm above, the paper analyzed the result of the algorithms by taking the expressway network toll data as study object. According to the reuse of the expressway network toll data, the toll amount prediction by car type amount, the classification to fraud toll collectors and solve the integer linear programming problem of toll collector job scheduling are the three application solution examples. That is, the toll amount prediction and toll collector fraud classification ask that the knowledge expression ability should exist in the relationship between the input and output variables on the premise of the accuracy rates satisfy the demands. Only under this circumstances, useful information and decision making could be offer to the sides, such as road operating and management, public services, government and sciences. But the current counting and artificial method can not satisfy the two conditions simultaneously due to the the algorithm itself or can not raise the accuracy rates by reducing the knowledge expression ability when the accuracy rate are not high. In addition, the job scheduling of toll collector is an integer programming problem, and integer programming is a kind of typical NP hard problem which is solved usually by genetic algorithm (GA). This paper solves the integer programming problem by GP method and because GP encodes the problem with layer structure, the structure and scale of the problem has dynamic changeability and few artificial intervene factors. The experiments show that the accuracy rates of prediction and classification are roughly have an inverse ratio with the knowledge expression ability. If the knowledge expression ability is not so high, the prediction and classification accuracy rates could be very high. The solution result of the toll collector job scheduling problem shows that the integer linear programming method based on GP is effective and the solution result of the example in this paper is better than GA method.