Abstract: Aiming at the problem of low accuracy of traditional asphalt pavement serviceability prediction, the asphalt pavement serviceability prediction model based on rough set theory (RS) and PCA-(principal component analysis) APSO-(adaptive particle swarm optimization) SVM (support vector machine) was established. Considering the time-series indexes and influencing factor indexes of asphalt pavement, 11 initial prediction indexes (including the pavement serviceability, the equivalent axle times, the road age, the maintenance nature, the pothole rate, the repair rate, the annual precipitation, the average temperature and sunshine hours in previous 3 years) were established. Nine core indexes were screened out by rough RS attribute reduction, four principal components were extracted by PCA and a dataset based on the four principal components was obtained. The adaptive particle swarm algorithm (APSO) was introduced into the SVM, the dataset was trained and the parameters of the SVM model were also optimized. The PCA-APSO-SVM prediction model of the pavement serviceability was established, and the prediction of the pavement serviceability was carried out by taking a certain section of the G6 Beijing-Tibet Expressway in Gansu as an example. The research results show that the prediction accuracy of PCA-APSO-SVM model has been significantly improved compared to PCA-PSO-SVM, APSO-SVM and PSO-SVM, and the prediction results are more in line with the actual situation, which can provide relevant references for the decision-making of pavement maintenance.

摘要： 针对传统沥青路面使用性能预测精度较低的问题，建立了基于粗糙集理论（rough set, RS）与主成分分析法(principal compoent analysis, PCA)-自适应粒子群算法(adaptive particle swarm optimization, APSO)-支持向量机(support vector machine, SVM)的沥青路面使用性能预测模型。基于沥青路面的时序指标与影响因素指标，建立了11个初始预测指标(包括前3年的路面使用性能、当量轴次、路龄、养护性质、坑槽率、修补率、年降水量、平均气温、日照时数)；通过RS属性约减筛选出9个核心指标；利用PCA提取4个主成分，得到了基于4个主成分的数据集；将APSO引入到SVM中，对数据集进行训练，并优化了SVM模型参数；建立了路面使用性能的PCA-APSO-SVM预测模型，并以G6京藏高速甘肃境内某段道路为例，对路面使用性能进行预测。研究结果表明：PCA-APSO-SVM模型预测精度较PCA-PSO-SVM、APSO-SVM、PSO-SVM有较大提高，预测结果与实际情况更加符合，能为路面养护决策提供相关参考。