Research On Pavement State Recognition And Flatness Evaluation Method

With the current substantial increase in the number of motor vehicles and the continuous extension of highway mileage,the requirements for the detection and evaluation of pavement flatness and maintenance efficiency are getting higher and higher.Integrating advanced sensing technology and artificial intelligence technology,short detection cycle,intelligent,low-cost pavement flatness detection and evaluation methods and technologies are urgent needs of road management departments.This paper uses a large number of vehicle video capture equipment such as driving recorders to collect cheap driving videos,extract the jitter data and characteristics of the driving videos,combine the driving speed,and use an improved machine learning algorithm to carry out pavement state recognition and pavement flatness statistical evaluation.Every vehicle that can provide driving video can be used as a detection vehicle,which can meet the real needs of short detection cycle and low cost.The main research work of this paper is as follows:(1)Vehicle jitter vector detection based on improved gray projection algorithm.Use image processing to enhance gray-scale contrast and median filter to eliminate noise interference;perform block gray-scale projection processing on the image to remove large-deviation jitter vectors and moving target interference,and improve the accuracy of global jitter vectors;use particle swarm optimization(PSO)algorithm to improve the search mode of the correlation of the gray projection curve in the traditional gray projection algorithm,and improve the efficiency of jitter detection.(2)Pavement state recognition based on driving video.The particle swarm optimization(PSO)algorithm is used to determine the optimal parameters of the support vector machine(SVM)and random forest(RF)models,and a pavement state combined recognition model based on SVM and RF is constructed,and cross-validation is used to train the model.The results show that the effect of the combined recognition model is significantly better than the traditional single-application recognition model.The example results show that the combined recognition model has a high recognition accuracy for road conditions such as flat pavements,diseased pavements,manhole cover pavements,and bridge-head jumps.(3)Graded evaluation of pavement flatness based on driving video.The genetic algorithm(GA)is used to improve the K-means algorithm;on the basis of the pavement flatness classification,the improved K-means algorithm is used to obtain the optimal clustering center of the jitter data of different vehicle speed ranges,so as to realize the pavement flatness graded evaluation.(4)Pavement state recognition and flatness evaluation system.Based on the pavement state recognition model and the flatness evaluation method,a pavement state recognition and flatness evaluation system is constructed,which can analyze the driving video to obtain the recognition and classification results of the pavement state and the flatness graded evaluation results,the feasibility of the real application of the evaluation algorithm is verified.