Study on the Road Condition Monitoring Based on Vehicular 3D Acceleration Sensor
|School||Dalian University of Technology|
|Course||Electronics and Communication Engineering|
|Keywords||Intelligent Transportation System 3D Acceleration Sensors Mixed Gaussian Background Model Support Vector Machine (SVM)|
In recent years, as the development of cities in our country is faster and faster, population and vehicles are more and more.At the same time,it has also brought great pressure for all aspects of urban development,for example traffic management, the road maintenance, citizens travel, energy conservation and emission reduction and so on.Therefore,in order to alleviate this pressure,people begin to research intelligent transportation system.Although our country has made good progress on traffic and road display system based on the vehicular (taxi) GPS technology,mass data information implicited in the GPS data is not deeply excavated, and modern sensing technology and 3G communication have not also been made full use of.This makes the government lack advanced technology in transportation policy formulation and the management of the operation of taxi.So,combined with the vehicular sensor technology, wireless communication technology and mass data mining technology, the research of the taxi intelligence regulation and the analysis and application to existing GPS data are of great theoretical significance and application value.This paper studies systematically a specific application of the vehicular sensor technolo-gy in public transportation, which is the road conditions monitoring based on vehicular 3D acceleration sensor, further analyses and researches of the data information about the road conditions collected from 3D acceleration sensors, and gives the detective results of current road situation. The main work of this paper includes:Firstly, the paper introduces briefly the background and significance of the research about vehicular sensor network in intelligent transportation system and the research status quo and the development at home and abroad, and introduces the system structure and system function of vehicular sensor networks.Secondly, this paper does a theoretical analysis for common noise and denoising algorit-m in a communication system, and provides the experiment results of each algorithm. The res-earching methods include singular value decomposition (SVD) method, empirical mode deco-mposition (EMD) method and adaptive filter method. At the same time, the paper also analyzes the performance of signal and gives the shortcomings of traditional filter.For the last, this paper does a analysis of theory and result for these algorithms, and finds a suitable algorithm for vehicular 3D acceleration sensor signal,namely empirical mode decomposition method. Thirdly, this paper studies a method on pavement abnormal events extraction based on vehicular 3D acceleration sensor, which is mixed gaussian background model method. It can make up for the shortcomings of the median filtering method to establish the background model which requires a lot of storage space, kalman filtering method to establish the background model whose calculation time is very long and single gaussian background model method whi-ch can’t be successfully applied to the reality unitl now. And this method shows good perform-ance under the different events.Fourthly, on the basis of extracted abnormal events, this paper provides a new method of pavement abnormal events classification, which is the classification method of the support vector machine (SVM) based on energy characteristic extracted by EMD. And compared with k neighbor classification, the support vector machine classifier is superior to the k neighbor classifier on effect and performance, which makes up for the shortcomings of selecting k value difficultly of the k neighbor classifier.At last,this paper summarizes the research work of road conditions monitoring based on vehicullar 3D acceleration sensor, provides two new methods which are a new method of pavement abnormal events extraction and a new method of pavement abnormal events classification, and provides a prospect for the following job.