Research on Data Prediction Methods of Structural Health Monitoring System

Trend prediction is virtually modeling process for dynamic data. The key to prediction is to establish a model in accordance with actual status, then use the model to predict the trend of object, and infer its behavior in future. Two prediction methods are researched to predict the trend on the observed points of the structure in this paper, which are regression prediction method and grey prediction method. The continuous time strain value of a measured point on Tianxingzhou Yangtze River Bridge is used as data sample for researching. The method of regression analysis is applied for predicting the trend of short-term data, and the method of grey model prediction for predicting long-term data. Regression prediction can assess the health status of the structure and obtain the alarm information effectively by comparing the actual monitoring data with the range of forecast interval. Grey prediction method has great advantages when dealing with poor information. By engineering example this study shows the pros and cons of these two methods, and proves that the method of grey model prediction is more suitable of predicting the trend of object in the structural health monitoring system.