Font Size: a A A

Research On Substructure Quality Evaluation Algorithm For Highway Pavement Based On Evolutionary Fuzzy Neural Network

Posted on:2013-01-30Degree:MasterType:Thesis
Country:ChinaCandidate:Y L JiangFull Text:PDF
GTID:2218330374463963Subject:Signal and Information Processing
Abstract/Summary:PDF Full Text Request
With the constantly increase of highway mileage, the expansion of highway network scale and the increase of service life of the highway in China, the task of highway protection and rehabitation is becoming heavier and more important. Ground Penetrating Radar (GPR) is the first choice for the highway daily detection with its advantage of fast, nondestructive. high accuracy and resolution at present. However, currently, the GPR data interpretation is typically done by manually, which the analysis result is depending on the experience of experts, resulting in the problems such as the varying conclusions made by different experts and long interpretation period and so on. It is urgent to develop an automatic high-precision algorithm for evaluation of substructure.The changes of the substructure's quality generate the variations of permittivity and conductivity of substructure materials. So. the parameters acquired from GPR record such as the thickness of layer, time delay and the reflection amplitude will change accordingly. Combining the image processing, signal detection, digital processing and pattern recognition. the author presents a novel approach for substructure fuzzy evaluation basing on Evolutionary Fuzzy Neural Network to solve the challenges above.This paper mainly analyzes the GPR data which obtains from the Nanchang section Changjiu highway in Jiangxi province, and the presented evaluation of substructure includes the following three modules:1. The preprocessing module of GPR:At the first, it is the clutter suppression processing of GPR original data. And at the second to detect the substructure of highway pavement based on signal processing and image processing methods respectively. And then to extract Regions of Interest (ROI) of the substructure of highway pavement which has been detected. At the last, to extract vectors including six features from the time domain and the wavelet domain.2. The rating module of substructure quality:Combining the expert experience with the drilling core sample build the ground truth database, using method of cross-validation to divide the whole sample sets into training set and testing set. At the last. Evolutionary Fuzzy Neural Network will be used to classify the feature.3. The post-processing module:To obtain the results of the substructure quality using evaluation algorithm for highway pavement after the processing of the output results based on Evolutionary Fuzzy Neural Network.The agreement of the final processing results has reached92.6%with the ground truth database. It meets the needs of engineering applications, also guides the development of appropriate conservation strategies and provides a scientific basis for allocation of conservation funds.
Keywords/Search Tags:Ground Penetrating Radar (GPR), Evolutionary clustering, Featureextraction, Evolutionary Fuzzy Neural Network (EFUNN), Automaticscoring system
PDF Full Text Request
Related items