Prediction Model Of Fuel Moisture Content Based On Partial Linear Model

The forest fuel moisture content is a dominant factor in the study of forecasting forest fire, and it is also an important factor in the control spread of forest fire. It is correlation between forest fuel moisture content and meteorological factors, and the influence of meteorological factors on the forest fuel moisture content is the integrated result of meteorological factors. The study on relationship between forest fuel moisture content and meteorological factors provides theoretical basis for forecasting forest fire.In this paper, stepwise regression analysis and partial linear model are used to analysis relationship between forest fuel moisture content and meteorological factors. Firstly, we establish stepwise regression model about forest fuel moisture content and meteorological factors, and predict for forest fuel moisture content. The results showed that the error of model is less than 3%, indicating that the method is feasible. But in the model, the explanatory variables selected only temperature 1h before and moisture content 1h before, and did not take into account other meteorological factors, for example,wind speed etc.Secondly, in this paper establish partial linear model with forest fuel moisture content and meteorological factors. Kernel estimation, local polynomial estimation and smoothing spline estimation are used to fit relationship between fuel moisture content and meteorological factors. Fitting results showed that error of forest fuel moisture content prediction model on partial linear model is less than 5%, and prediction result is almost consistent with actual value, indicating that the model is applicability.Finally, we discuss differences of stepwise regression model and partial linear model. Stepwise regression model requires independent variable and response variable is correlation. Partial linear model is consist of linear and nonlinear section. In the process of model fit, partial linear model take into account all the meteorological factors to avoid some of question that correlation is weak between forest fuel moisture content and meteorological factors is missed.