The Analysis On The Occurrence Dynamics Prediction Of Dendrolimus Punctatus Walker In QianShan City

This project was originated from the national forest public welfare industry scientific research project(201404410).Dendrolimus punctatus was always one of the main pests in china’s forests and seriously restricted the development of forestry in CHINA and affected the safety of forest resources.Therefore,timely and accurate prediction of the occurrence dynamics of D.punctatus was an urgent problem to be solved.In this study,the relevant data of the occurrence dynamics of D.punctatus Walker was used as the research sample.The mathematical methods that Periodic distance method,regression analysis method,stationary time series method,Bayes discriminant analysis method,Markov chain method,artificial BP neural network method,multi-stage-factor contingency table analysis,GM(1,1)catastrophe prediction method and fuzzy comprehensive evaluation method were used to predict the occurrence dynamics of D.punctatus larvae and eggs in Qianshan City and the historical coincidence rate was used as the evaluation standard.The advantages and disadvantages of the prediction results of various methods were compared and analyzed in order to provide a scientific basis for the comprehensive management of D.punctatus Walker.Different methods and the selection of forecasted factors affected the accuracy of the forecasted model.The results of mathematical analysis of D.punctatus for the occurrence dynamics in Qianshan City,Anhui Province was shown as follows: In the prediction of the peak period of D.punctatus egg,the historical coincidence rate of BP neural network method,regression analysis method,Bayes discriminant analysis method and Markov chain method reached more than 95%,and the prediction result of stationary time series had the largest deviation from the actual values,and the prediction effect was relatively poor;In the prediction of the beginning period of D.punctatus larvae,five models including BP neural network method,regression analysis method,Bayes discriminant analysis method,fuzzy comprehensive evaluation method,multi-stage-factor contingency table analysis had excellent prediction performance in other methods,and the historical coincidence rate had reached more than 90%;In the prediction of the peak period of D.punctatus larvae,the historical coincidence rate of Bayes discriminant method for the second generation(100%)was higher than that of the first generation(91.07%).However,in the prediction of the occurrence of D.punctatus larvae,the historical coincidence rate of Bayes discriminant method for the first generation(96.97%)was higher than that of the second generation(81.82%),which indicated that the selection of predicted factors that were closely related to the forecasted quantity was the key to the accuracy of the method.The historical coincidence rate of predicting peak period of D.punctatus eggs based on the catastrophe method was over 90%.The combined prediction model of D.punctatus for the first generation egg was better than the single model.According to Comprehensive comparative analysis,regression analysis,BP neural network method and Bayes discriminant method had the highest accuracy.The key to improving the accuracy of model was the selection of predicted factors and scientific grading standards.The accuracy for the combined model was better than that of the single model.The forecast accuracy of the combined model depended on the combination of different mathematical methods.