Parameter Estimation with Colored Noise Effect for Grey Models and Its Application to Hydrological Analysis

• Main Authors: Lee, Ru-huang, 李如晃
• Other Authors: Ru-yih,Wang
• Format: Others
• Language: zh-TW
• Published: 1998
• Online Access: http://ndltd.ncl.edu.tw/handle/63734833898675550675

博士 === 國立臺灣大學 === 農業工程學系研究所 === 86 === The hydrological system is regarded as a grey single input/single output (SISO ) system and a grey multiple input/single output (MISO) system in this study, respectively. The main consideration of colored noise effect in parameter esti mation for a linear model is to make the estimators tend to the true values as close as possible. The differential hydrological grey model with dual series (DHGM(2,2)) and the multivariable grey model (MGM(1,N)) are adopted to investi gate the influence of colored noise effect fTo verify the applicability of the proposed procedures, the upstream watershed of Tseng-wen Reservoir and the do wnstream plain zone of Tseng-wen River Basin in southern Taiwan are chosen as the project areas. By comparing the simulated results and the model''s residual errors, both models of DHGM(2,2) and MGM(1,N) employed the colored noise effe ct in parameter estimation show better results than that without colored noise effect. It is also verified that colored noise effect plays an important role not only in the parameter estimation, but in the simulation of rainfall-runof f process as well.