Projection Pursuit Model And Its Applications In Water And Soil Resources

Practical research of projection pursuit method has been improved in this paper.As a result of multidimension,nonlinearity and lack of samples were most characteristics in water and soil resources,the new projection technique has been applied in water and soil resources research including integrated evaluation and prediction in this paper,the multiple complexity in water and soil resources has been analyzed and rational results has been achieved. There are three new ideas sparkling in this paper.1.The real coded accelerating genetic algorithm has been used to optimize the projection direction in projection pursuit method.The difficult problem of projection pursuit technique in computer has been resolved with the new projection pursuit algorithm,which improves the practice of the method into a new stage.2.Classification and integrated evaluation in water and soil resources,on the basis of structural characteristic of evaluation sample or some evaluation criterion,projection pursuit cluster models based on different functions of projection indexes has been presented in this paper.When standard deviation and partial density are used to structure the function of projection indexes,it deduces and acquires the empirical formula of calculation of exclusive parameter-the density window breadth of projection pursuit cluster model. Contraposing the problems of multicollinearity among indexes of evaluation sample,projection pursuit principal components analysis has been presented in this paper.3.Prediction in water and soil resources,on the basis of the correlations among prediction sample and prediction ingredients,projection pursuit regression model,projection pursuit threshold regression model,coupling error back propagation and projection pursuit model,projection pursuit based on partial least-squares regression model and coupling error back propagation and projection pursuit based on partial least-squares regression model have been presented in this paper.On the basis of structural characteristic of projection sample,projection pursuit auto regression model and projection pursuit auto regressive based on error back propagation model has been presented in this paper. The multiple complexity in water resources prediction has been analyzed and rational results has been achieved by prediction models.The projection pursuit method that has been successfully in this paper embodies the thoughts of stressing principle contradiction.By using the method,the paper deals with water and soil resources multiple complexity,offers new technology for resolving some present pratical problem and suggests some new ideas for water and soil resources research.