The Study Of Toll Audit In Networked Expressway Based On Neural Network

With the expansion of expressway network, the amount of one-time payment for vehicles increased. This has increased the behavior of toll evasion. Toll auditing has became a necessary means to reduce property losses, but it mainly depended on manual inspection, lack of necessary technical support.Artificial neural network (ANN) can be used for Massively Parallel Processing of information, has a strong fault tolerance, good at associative memory, summary, analogy and reasoning, and also has very strong self-learning ability. ANN has excellent performance for pattern recognition and prediction. So the neural network can be used to aid highway toll audit, it can improve the efficiency and accuracy of toll audit.This paper analyzed the connotation, function, contents and methods of toll audit under expressway networked toll collection mode in detail. This paper conducted a comprehensive analysis and research for the toll evasion phenomenon, also gave a classification and detailed description for the phenomenon. It found that the toll record and toll evasion phenomenon are associated.This paper researched the principle, structure, transfer function, algorithm, and deficiency of BP neural network. It also choosed improved method for the deficiency of BP neural network. Make use of the association between toll record and toll evasion phenomenon, combined the actual toll audit, this paper analyzed and designed a BP neural network model for the toll audit. After the selected sample data pre-processing, it implemented network training and testing through the programming by MATLAB.Through the analysis for simulation results, the BP network model of this paper has faster convergence and higher prediction accuracy. For the changing category, changing card, license does not match, U route and other behavior of toll evasion, the model has a good audit performance. It proved that the use of BP neural network aid in toll audit is feasible and practical.