基于支持向量机的信息融合在筑路质量中的预测

筑路质量受多种因素影响难以在线监测，通过其他易测指标信息融合预测筑路质量可以确保施工质量．选择影响筑路质量的关键因素，对20组施工现场实测数据进行预处理，利用神经网络进行局部时间序列预测，最后基于支持向量信息融合，对筑路质量进行在线学习和预测．3组预留数据预测误差为21％、25％和17％．在数据量较少情况下，基于支持向量机的信息融合方法适用于筑路质量的预测．

Road Construction Quality Prediction with SVM-Based Information Fusion

Construction quality, influenced by many factors, is not easy to be detected. It is effective to ensure mad quality and solve the problem duly using information fusion by other measurable indexes. After choosing key influential factors, twenty records of constructing yard data were pretreated. And then local time series was predicted by artificial neural network. Using support vector machine（SVM）, construction quality was on-line learned and estimated. Prediction errors of three groups of reserved data were 21% , 25% and 17% respectively. Information fusion based on SVM was suitable for construction quality prediction, especially in case of lack of data.