Study On Self-Memory Model Of Hydrologic Nonlinear Time Series Analysis

The study object of many problems comes down to hydrologic time series in the field of hydrology and water resources. Time series analysis method is playing an important role in hydrologic regular analysis and hydrologic analogy as well as hydrologic forecasting and so forth. Traditional hydrologic time series analysis method is mainly based on stochastic theory is an uncertainty method, which has a linear characteristic. Time series analysis self-memory model based on self-memorization principle of dynamic system is a kind of nonlinear method, which combining stochastic with dynamic. The self-memory model proceeds from dynamic model of system and introduces into the memory function which contains multi-time values, and combines dynamic calculating method with model parameter estimated using historical data, thus the nonlinear feature of hydrologic system is revealed and the forecasting accuracy is raised markedly. For hydrologic system, generally only the observed values of hydrologic time series are known and the dynamic equation is unknown. In this paper, firstly, themethods of using hydrologic time series to retrieve hydrology dynamic model are studied, and the problems to establish self-memory prediction model on the basis of retrieving hydrology dynamic model are investigated by real examples. Secondly, according to the theory of retrieving model using time series and the self-memorization principle, the grey self-memory model based on grey system theory and phase space self-memory model established using chaotic theory are put forward, and the two kinds of models are applied to the analogy and forecasting of hydrologic time series, and the application field of the self-memory model is further expanded from extent and profundity. The grey self-memory model is that the grey differential equation of system is retrieved by means of grey system theory, and the self-memory model is further established through taking the equation as dynamic model. The phase space self-memory model is that the phase space is reconstructed for chaotic time series, and the dynamic model of phase space is deduced according to the retrieving method of multi-variable time series, thus the phase space self-memory forecasting model is established. The study and application by way of the self-memory model and the grey self-memory model as well as the phase space self-memory model show that the self-memorization principle and its forecasting model have good suitability, and that the grey self-memory model is especially suitable for the hydrologic time series with considerable fluctuation and amplitude, moreover, the forecasting method that the phase space self-memory model lead stochastic- dynamic or statistic- dynamic into the phase space is feasible for the forecasting of chaotic time series. In addition, the concrete methods of applying self-memory model to forecasting for hydrologic time series with periodical variation andseasonality fluctuation are studied deep. To sum up, the current study makes the self-memory model develop, and carries through an active exploration in order to further enrich hydrologic forecasting means by applying new theories and methods and technologies.