Evaluation Of Traffic Operation State Based On Driving Noise

At present,the problem caused by traffic congestion is becoming more and more serious,and it has become the main problem that people all over the world are facing and needs to be solved urgently.The timely and effective detection of traffic operation state is of great practical significance to the macro-control of government departments,the implementation of traffic managers’ plans,and the decision-making of residents’ travel.Based on the analysis of driving noise data,the evaluation of traffic operation state is studied in this paper.Aiming at the shortcomings of traditional characteristics of driving noise and traditional recognition algorithms in the evaluation of traffic operation state,three improved methods are proposed in this paper.In order to verify the effectiveness of the three improved evaluation methods,the sound acquisition equipment is used to collect the driving noise data from a same four-lane urban road section,and the different evaluation methods are compared and analyzed through experiments.The specific research are as follows:(1)The traditional Mel Frequency Cepstrum Coefficient(MFCC),the traditional Particle Swarm Optimization(PSO),and the Support Vector Machine(SVM)classifier are combined for identifying the traffic operation state.The limitations of traditional MFCC features and traditional PSO algorithm to optimize the parameters of SVM classifier in traffic operation state evaluation are analyzed through the experimental results under complex noise environment.(2)Aiming at the problems that the traditional MFCC features are weak in representing the traffic operation state under complex noise environment,the Teager Energy Operator(TEO)is introduced into the feature extraction of driving noise.A new driving noise feature T-MFCC is formed by combining TEO and traditional MFCC features,and a traffic operation state evaluation method based on T-MFCC and PSO-SVM is proposed.The experimental results show that the evaluation method based on T-MFCC and PSO-SVM is better than the traditional evaluation method based on MFCC and PSO-SVM,and the accuracy increased by 3.685%.(3)In order to solve the problem of premature convergence and local optimization in traditional PSO algorithm,some new parameters are defined,then the Genetic Algorithm(GA)and K-Means Clustering(KMC)are introduced to improve the PSO algorithm.The improved algorithm is named GK-PSO and it is used to optimize SVM parameters.The experimental results show that the evaluation accuracy of traffic operation state evaluation method based on MFCC and GK-PSO-SVM is 1.340%higher than that of traditional evaluation method,and 2.345% lower than that of the improved evaluation method in(2).(4)Combining the T-MFCC features with the SVM classifier optimized by GK-PSO algorithm,a traffic operation state evaluation method based on T-MFCC and GK-PSO-SVM is proposed in this paper.Four different evaluation methods(including the traditional method)are compared through experiments.The experimental results show that the evaluation accuracy of T-MFCC and GK-PSO-SVM evaluation method is4.690% higher than that of the traditional method,1.005% higher than that of the improved evaluation method in(2),and 3.350% higher than that of the improved evaluation method in(3).