| Artificial Intelligence is a dream when human being develops into computer age. It provided the probality for nuifying automatiation of knowledge learning and acquireing, common adapting of knowledge expressing way, high efficiency and entirely of searching solution and intelligence body activation into envirnment. To meet the need of deeply research and model establishing for complexity system, aiming at water-soil project special complexity, this paper has applied Artificial intelligence into the research of this project. There have been explorely research about apply of ANN and GA in ET0 and CRWS. The main conclusions are as follows: (1) ANN has been applied for the research of ET0. Models of 4 factors (radiation, temperature, humidity, wind speed) and 3 factors (radiation, temperature, humidity) for ET0 single station have been established. At the same time, the applying conditions of models have been put forward. In crop growing stages, 4 factors model can show the effect of weather factors to ET0 and has higher precision. When wind speed is lower, the effect of wind speed to ET0 can be neglected and can meet the need of producing. The researches have been making for ET0's space-time variety and relativity in big dimension area (arid and cold area of china north) and ET0's linearity models for different second district have been developed. On base of these, ANN models of ET0 for different second districts have been set up. At the same time, the model has been evaluated. Research showed that the ANN model of ET0 has higher precision because of characters of ANN, such as nonlinear, self-adaptive. This research is a supplement to tradition computer models of ET0. (2) On base of water-saving irrigation experiment in salty soil, ANN has been applied in the study of crop response to water salt and with aiming at the complexity of CRWS and sufficiency using the ANN merits of many parameters and nonlinear. this paper has established 10-factors and 6-factors ANN models for CRWS. By testing, these two models have higher precision. Considering the complexity of models, this research recommenced using 6-factors model. Crop sensitive to water and salt press has been analyzed with the BP-CRWS established in this research. The result showed in light-moderate salty soil, the sunflower is most sensitive to water in squaring stage and the following is flowing stage, seedling stage. But in gravity salty soil, the order is seedling stage, squaring stage, flowing stage. This simulative result is same to the experiment result and breaks a new road for CRWS and relative research.(3) At present, usually parameters and systems have been optimized by least square techniques and dynamic programming in the research of crop water model and crop response to water and salt. These tradition optimum ways can trap in part optimum points. GA put a new way for full optimum. This paper has applied GA into relative research of CRWS. The sensitive indexes of crop water model for spring wheat has been get with RAGA. Results showed sensitive parameters obtained by GA have same adaptability with that by least square techniques. Because GA can control the scope of parameters, the result is more reasonable than result by least square techniques. Soil's optimum water content for crop (sunflower) stages in different salty soil has been get by GA-ANN coupling model. Analysis and research showed that GA can avoid the shortcoming of trapping into part optimize point of tradition optimize algorithmic and provide a short-cut for optimum of many parameters and nonlinear problem in water soil engineering research. Now, most applying of artificial intelligence in water-soil engineering belongs to attempt research. This research applied ANN and GA into model's establishing and parameters optimizing and showed the feasibility of their applying in water-soil engineering research. The purpose of this paper is taking a sample for similar research and making some complexity problem that can't be solved with tradition way can solved. |
PDF Full Text Request