摘要
列举了实际桥梁健康监测系统中数据缺失的几种形式,根据桥梁健康监测系统中监测数据是时间序列集的特点,以及神经网络强大的映射能力,利用神经网络及时间序列混合模型的方法来填补缺失数据,并将该方法与时间序列法的填补结果进行对比,结果表明该方法处理缺失数据的误差较低。
Because of shutting down of the power supply, disturbance or noise in transmission, replacement of sensors, and etc. , the missing data is occurred frequently in bridge health monitoring system. This will affect the bridge health monitoring system to evaluate the state of bridge. So, it is very important to impute these missing data. In this paper, some missing data examples occurred in an actual bridge health monitoring system are listed. By means of time series＇ feature of bridge health monitoring system＇ s data and neural network＇ s powerful mapping capability, the paper proposed a hybrid model method of time series and a neural network. It could be utilized to forecast the missing data. In the paper, the hybrid model＇ s imputation results and time series＇ imputation results were compared. It shows that the hybrid model method can treat the missing data with a lower error.
出处
《重庆理工大学学报（自然科学）》
CAS
2011年第4期79-85,共7页
Journal of Chongqing University of Technology：Natural Science
基金
重庆市自然科学基金资助项目(CSTC2009BB6037)
研究生科研创新基金个人资助项目(CDJXS11120015)
关键词
缺失数据
填补
神经网络及时间序列混合模型
桥梁健康监测系统
missing data
imputation
hybrid modal of neural network and time series
bridge health monitoring system
