Study On Self-Memory Characteristics Of Hydrology Dynamic System And Its Application

The nonlinearity and high complexityof hydrological phenomena demands for applying new theories, developing new theories and putting forward new theories in higher level and more comprehensive point, in order to solve the problems that can't explain and unceasingly appears until now. Hydrological dynamic system was the research objects in this study. Aiming at the problems existing in hydrology analyze, gray theory, multiple theories, and modern optimization algorithms were used to study on self-memory simulation models and prediction models of hydrological time series. This study, being of great theoretical significance and application value to develop potential and raise predicting level, provided a new way to research nonlinear simulating prediction and enriched the research contents of hydrology.The main achievements were as follows:(1) Long memory characteristics of hydrological time series was researched. The reasons of long memory characteristics of hydrological time series were discussed in virtue of structure transformation. Modified R/S analysis was used to checking long memory characteristics and the statistic was analyzed to determine the length of memory. The study showed that the hydrological time series had long memory characteristics, but the causes of long memory characteristic needed further investigation.(2) Lag errors phenomena of gray self-memory model often existed in real application. The key factors that named gray system core background value and resulted in lag error were found, according to the mechanism of modeling. New gray self-memory model was set up by modifying the background value, and the practical example showed the affectivity of the model.(3) Based on multivariable inversion theory, hydrological multivariable time series inversion self-memory mode was put forward, combined with self-memory theory, and hydrological multivariable time series inversion self-memory model was set up as well. The practical example showed that this model had good adaptability. (4) In view of the deficiency of self-memory model that adopted least squares method to estimate the memory coefficient, self-memory modeling theory and process that based on parameter optimization were put forward, combined with modern optimization algorithms. The example showed that self-memory model, based on parameter optimization, advanced the adaptability and had a better fitting effect and prediction effect.