Research On Highway Abnormal Data Detection Methods

In recent years,the mileage of highway network in our country has been rapidly increasing,with consequent abnormal behaviors such as escaping charges.These abnormal behavior of escaping fees will not only affect the normal operation of the expressway,but also bring about huge property losses to the country.Therefore,in order to detect fleeing vehicles accurately and quickly,this paper investigates the common means of fleeing vehicles and summarizes the important features of the abnormal behavior of fleeing vehicles.In this paper,two solutions are proposed based on the characteristics of the data extracted from charging data:(1)The IGA-IBP algorithm is proposed to analyze the characteristic attributes of vehicles in expressways,and the multiple golden parts method is used to improve the mutation operator in genetic algorithm to ensure the diversity of the population.Secondly,step by step,Between the learning rate and taking into account the impact factors of the previous learning rate,as well as dynamically adjust the vehicle travel time to improve the accuracy of the algorithm for highway escape fraud detection accuracy and stability.The experimental results show that the IGA-IBP algorithm improves the detection speed and detection accuracy,and is superior to the traditional algorithms in RMSE,MAPE and R2 evaluation indexes,which is of great significance to the decision of the traffic department.(2)Proposed improved firefly algorithm optimized weighted K-Means algorithm model for the highway escape behavior analysis and detection,the Rosenbrock search algorithm and firefly algorithm,the use of Rosenbrock algorithm powerful search capabilities,constantly updated search direction,making the initial clustering Center selection is more accurate.In view of the main features shared by escape routes,taking into account the influence of abnormal points in high-speed data and the impact of each sample data on the clustering results are different,we introduce weights into the objective function and eliminate the initial The selection of cluster centers is affected by abnormal data.Through the different brightness of fireflies,the fireflies with large brightness attract the fireflies with small brightness,and the positions are updated continuously to obtain the final clustering result.Experimental results show that the improved algorithm in this paper can effectively detect abnormal data in expressway data,which is more stable,accurate and efficient than traditional algorithms.