Application Software Development Of Coupling The Model Of Mean Generating Function And BPNeural Network For Forecasting Rainfall And Temperature In Horqin Sandy Land Simulation

Hydrology of climate factors changes profoundly affects the ecological environment and the update change, the development and utilization of resources, especially in arid regions, because of the relatively fragile ecological environment, it is sensitive to hydrology and climate factor changes. In recent decades, the effect of global changes of hydrology and climate factors on hydrology, water resources and ecological environment has become the important research content. The simulation prediction research of hydrology and climate factors can further improve the simulation prediction theory, which has certain reference significance to the future climate change research, ecological environment improvement, reasonable development and utilization of water resources, etc.Based on the analysis and summary various kinds of relevant existing simulation prediction theories, this paper selected mean generating function and BP neural network as essential methods. Through analysis mean generating function and BP neural network, this paper found them and their existing coupled manners have make full use of their advantages in simulation prediction and remain to be improved. Through improving the roughing factor sets, featured crude anthology combination and precision control condition, further playing the respective advantages of mean generating function and BP neural network. This paper established a new coupling simulation prediction model with mean generating function and BP neural network model(MGF-BP-I of simulation and forecast model).Using MGF-OSR, MGF-OSR-BP, MGF-BP-I of three ways and five modes to simulate predirect the average annual precipitation in Horqin sandy area. The results as follows: Firstly, during the modeling stage, MGF-OSR-BP, MGF-BP-I both the overall best, MGF-BP-I modeling stage optimal mode 98% number of years relative error distributed in 0 to 0.5%, from 0 to 1.25%,0 to 0.1%, fitting results are good, MGF-BP-I model modeling stage optimal accuracy is better than MGF-OSR-BP nearly 3 times. Secondly, In the test phase, the three models relative error of MGF-OSR, MGF-OSR-BP, MGF-BP-I optimal modeling phase distributed between 0 to 70%, have large span and high error. The 4 years relative error of MGF-BP-I testing phase mode concentrated in 0-1% and 0-2.3%. The overall optimal model of MGF-BP-I relative error is 6.45% and has higher accuracy. There is one year relative error of MGF-BP-I testing phase model was 2.66%, has high precision and the error range is intensive. The accuracy of MGF-BP-I test stage optimal increased 8 to 27 times faster than other modes. Thirdly, Prediction found that the precipitation of 2015 and 2017 is relatively rich, the other for the past decade years. Since the data have been found out in 2013 and 2014, compared to them and found the relative error was 10.98% and 8.65%, the precision of prediction is higher. Fourthly, Overall, MGF-BP-I model takes into account more fully, the accuracy is much higher than the other two methods. MGF-BP-I while the best overall model is more realistic applications, the results are satisfactory and can be used for simulation and prediction of the hydro-climatic factors.Based on the principle of MGF-BP-I model, this paper adopts modular design and uses Visual C++6.0 development tools to develop a MGF-BP-I application software. Software Key features include the achievment of three methods and five modes, the viewing features such as the simulation predicted data, connectivity, input and output features and help functions. The application software used in simulation and prediction of Horqin sandy area average annual temperature, the result show that:first, in the modeling stage, MGF-OSR, the the 93% number of years relative error of MGF-BP-I test stage optimal mode focused on 1% to 15% and 0% to 0.3%.MGF-BP-I while the best overall, MGF-BP-I and modeling stage optimal MGF-OSR-BP mode 95% number of years relative error distribution at 0 to 0.03%, 0.01% and 0 to 0 to 0.1% and had high precision. Secondly, in the test phase, the relative error of MGF-OSR, MGF-OSR-BP, MGF-BP-I model the optimal phase distributed from 0 to 40%. The 4 years relative error of while overall the best modes of MGF-BP-I focused on 0 to 5% and had higher precision. The accuracy of MGF-BP-I test stage optimal mode compared to other modes was higher 1 to 3 times. Third, forecast found that temperatures next largest growth in 2015, the highest in 2016, and the temperature of 2017 has dropped. Comparison with the measured data of 2013,2014, the relative error was 7.75% and 11.98%, the accuracy of simulation and prediction can be well. Four, Overall, MGF-BP-I analog prediction model has high precision and the software available can be used to predict the temperature simulation. Other, by using the Yellow River basin of Inner Mongolia section of precipitation. Meanwhile, the application software module has division clear, clear structure, simple interface design, functional, and general applicability can be more widely applied in hydro-climatic factors.