| As China’s highway network begins to transition from the large-scale construction phase to the high-quality development phase.Developing green and low-carbon road construction,improving the service life of pavement structures and improving pavement maintenance management are central to this transition.Based on this,asphalt pavement performance evaluation is the key to gaining an in-depth understanding of new pavement structures and the current state of pavement quality,as well as guiding pavement maintenance plans.However,the development of the pavement evaluation system in China is relatively late,resulting in serious shortcomings for the current road network.In particular,there are obvious shortcomings for the subjective biases in the current evaluation systems,for understanding and evaluating the complex and variable pavement performance evolution processes,and for the comprehensive evaluation of multi-indicator performance.In response to these challenges,this paper focuses on the evaluation model of asphalt pavement performance,proposes several evaluation methods based on the complex network approach,and establishes the data-driven pavement performance evaluation model,the main contents of which are as follows:1.A pavement performance network model is proposed to address the issue of subjective biases in the current pavement evaluation systems.Specifically,the pavement performance network is constructed by treating each pavement structure as a node in the network and edges among nodes are computed through the performance similarities.Then,by using the community detection method,the pavement structures are divided into several groups based on their performance.Pavement structures with similar performance are grouped together and those with differentiated performance are separated.Further,these groups are compared in the statistical level,yield significant performance differences.The model is validated in the Research Institute of Highway Ministry track(RIOHTrack),demonstrated the effectiveness of analyzing the pavement structures.2.A temporal pavement performance network model is proposed to address the issue of neglecting the evolutionary process of pavement performance in the current evaluation systems.Specifically,a temporal network with regard to the pavement performance is constructed using multiplicative temporal derivatives(MTD).MTD derives performance similarities among each pavement structures at each performance collection time point.This establishes the evolutionary process of pavement structures.Then,the performance evolution process is measured by dynamic community detection method,which provides groups of pavement structures with similar performance at each time point.The characteristics of the evolution process are further computed by the number of times each pavement structure changes their group affiliation during their evaluation period.This provides an essential insight into the stability of pavement performance.The proposed model is tested in the RIOHTrack,demonstrated the significance in considering the evolution process of pavement performance in the evaluation system.3.A multidimensional model is proposed to address the issue of biased in selecting multiple performance indicators.The key to this model is to capture not only the performance conditions per se,but also the nonlinear interactions among these attributes.This provides an in-depth higher-order property of pavements’service condition.Specifically,pavement performance attributes are modeled into multilayer network with each layer representing an aspect of pavement condition.Subsequently,this multilayer pavement condition network is mapped into low-dimensional space through the network representation learning for systematic evaluation.Finally,unsupervised cluster analysis derives groups of pavements which share similar overall condition for future decision-making process.The proposed method is validated with in the RIOHTrack and experimental results demonstrate the effectiveness in categorizing pavement condition based on multi-attributes.4.The validity of the main pavement evaluation model,which contains the network construction and community detection algorithms,proposed in this paper is further verified in the widely discussed correlation-based network.First,a social network is constructed using a public available dataset in the field of emotion recognition.The aim is to search for the potential diversity groups in emotion experience which proposed in sociology.Second,through construction and analysis of brain network based on EEG data,the importance of minimum spanning tree characteristic of brain network is demonstrated and validated as a key to understand human emotion.Finally,minimum spanning tree characteristics show statistical significant differences between detected groups from social network,thus further demonstrate the validity of complex network approach. |
