APPLICATION OF WEIGHTED BLOCK RECURSIVE PARTIAL LEAST SQUARES REGRESSION FOR DAM SAFETY MONITORING

As a new modeling method, partial least squares regression has been widely used in dun safety monitoring data analysis, but with the accumulation of data, If off-line analysis still is used, calculated rate will be affected, and predicting accuracy will be reduced. In this paper, weighted block recursive partial least squares regression(WBRPLSR) is designed, in this method, weights of sample data are allocated by time sequence, and then by means of recursive time blocks, prediction model for dam safety monitoring is established by partial least squares regression. The result of an example shows that efficiency and prediction performance of WBRPLSR model are greatly improved, which has some popularized value.