The Research On Wheat Scab Prediction Model

This paper describes the research of wheat scab, analyzed the influence factors of wheat scab, common wheat disease forecasting methods are reviewed, and introduces the advantages and disadvantages of various forecasting methods. On this basis, the paper presents a new predicting method - Projection pursuit cluster based on BP network forecast combination methods, the first application of projection pursuit cluster model of wheat scab in China on the impact of meteorological factors analyzed, screening, filtering out the most influential meteorological factors as BP network input node, and then modified BP learning algorithm, greatly reducing the dimension of the network input layer,the advantage of projection pursuit can exclude independent data structure and feature or variable interference with little relations, make high-dimensional data projected into one-dimensional space, then ,analysis one-dimensional data after projecting,and then comparison of one-dimensional result,find out the best projection.Therefore,Projection Pursuit is excellent method whichanalysis to identify good projection, is the analysis of high dimensional, nonlinear data and other traditional problem .This paper selected mid-April, late April and early May day average temperature, precipitation in late, late hours of sunshine, relative humidity, mid-average maximum temperature, mean minimum temperature late as the meteorological factors affecting wheat scab, respectively, stepwise regression model and Projection pursuit cluster based on combination of BP network prediction model in Anhui Province occurred Tongcheng degree of Scab of wheat were predicted by stepwise regression analysis to establish the extent of occurrence of wheat scab prediction model not only full, objective conditions and the use of meteorological data, and qualitative and quantitative prediction of combined forecasting to improve the quality of the forecasting methods and forecasting. Based on Projection Pursuit Clustering in BP neural network's forecasting method using the projection pursuit method of processing high dimensional data analysis capabilities, according to input factors the importance of variables for output to solve the BP neural network input dimension is difficult to determine the problem and improve the neural network of high-dimensional data processing capabilities, but also to speed up network convergence. Two kinds of forecasting methods on the results of comparative analysis show that the combined use of projection pursuit model of BP neural network prediction performance is much better than stepwise regression prediction model, can serve as a new forecasting method for prediction of wheat diseases and insect pests.In order to better achieve a variety of forecast model prediction in the application of wheat, by the current technology level and realistic background research and put forward a real-time information collection agricultural forecast system, this system of computer technology, communication technology, network technology and microelectronics technology, and through some network equipment to establish communication links with a high degree of integration, integrating software, hardware as a whole, on a variety of agricultural information automatically correct the data collection, transmission, statistics and comprehensive analysis and forecast. With the acquisition speed, high accuracy and good real-time acquisition outstanding advantages. Automatic data acquisition system can reduce the labor intensity collected; predictive control mechanism to timely forecast and warning data to guide agricultural production, increase farm production and further promote agricultural technology development.