No abstract

Searching the predictors in each level of the NCEP data by use the long time series data of NCEP and short time sequence data of wind observation. And filtering the information and extraction the components for these primary predictors using the method of partial least-squares regression (PLS), then takes the new comprehensive variables (names components) as predictors and using the neural network with the features including adaptive and learning and the logical reasoning ability of fuzzy system to establish the wind field calculation model with fuzzy neural network(FNN) through combining fuzzy neural network system and adjustment the system parameters using BP algorithm. Comparing the calculation result shows that the errors of combining model with partial least-square regression(PLS) and fuzzy neural network(FNN) is smaller than that the multiple linear regression model. The length time sequence data of wind could be calculated according to the short time sequence data of observation wind and the long time series data of NCEP by the combining model with PLS and FNN in practical, therefore this model is better practicability and popularize value for it provide the basis to research the exploitation wind resources.

No abstract

With the practical problem in application of the neural network in weather forecast, the effect of the learning matrix in neural network forecast model with the multi-collinearity on the generalization capability is researched. The results show, in the context of the same input knot number, whatever the smaller network or the network getting larger, there is few changes in simulation error both the neural network models without or with multi-collinearity, and the mean simulation errors for both of the two types model are very close, but the generalization capability of the neural network with multi-collinearity is obvious superior than that without multi-collinearity. Further more, it is to analyses the generalization capability for the two types of models in different training times from 5000 to 20000, the results also indicate that the multi-collinearity have the remarkable effect on decrease the forecast precision to the neural network forecast model.

No abstract