摘要 :
In this paper we present a novel method for short term forecast of time series based on Knot-Optimizing Spline Networks (KOSNETS). The time series is first approximated by a nonlinear recurrent system. The resulting recurrent syst...
展开
In this paper we present a novel method for short term forecast of time series based on Knot-Optimizing Spline Networks (KOSNETS). The time series is first approximated by a nonlinear recurrent system. The resulting recurrent system is then approximated by feedforward B-spline networks, yielding a nonlinear optimization problem. In this optimization problem, both the knot points and the coefficients of the S-splines are decision variables so that the solution to the problem has both optimal coefficients and partition points. To demonstrate the usefulness and accuracy of the method, numerical simulations and tests using various model and real time series are performed. The numerical simulation results are compared with those from a well-known regression method, MARS. The comparison shows that our method outperforms MARS for nonlinear problems.
收起
摘要 :
Climatic variability and human activities are the two primary factors that affect basin hydrology, and thus quantification of their effects is of great importance for water resources management and sustainable development at a cat...
展开
Climatic variability and human activities are the two primary factors that affect basin hydrology, and thus quantification of their effects is of great importance for water resources management and sustainable development at a catchment scale. In this study, the writers investigated the long-term trends and abrupt changes in hydroclimatic variables, including precipitation, potential evapotranspiration (PET), and runoff, from 1957-2000 in the Hutuo River Basin by the nonparametric Mann-Kendall test and the precipitation-runoff double cumulative curve method. A two-parameter hydrological model and linear regression method were employed to separate and quantify the effects of climatic variability and human activities on runoff. The results are the following: (1) significant downward trends for annual precipitation and annual runoff were detected by the Mann-Kendall test at a 99% confidence level, (2) a change in the gradient of precipitation-runoff double cumulative curves and an abrupt change in runoff series can both be found in 1979, indicating that the relationship between precipitation and runoff has changed; as a result, the annual runoff from 1957-2000 can be divided into two periods termed the baseline (1957-1979) and human-induced (1980-2000) periods, and (3) the climate variability was the primary cause for the decrease in annual runoff from the baseline to the human-induced period, despite certain effects of human activities on the change with respect to annual runoff.
收起