The Development Of The Water Saving Agricultural Expert System Of Loess Plateau

The WSA-LP (the Water Saving Agriculture Expert System of Loess Plateau), by using the artificial intelligence technology, is developed for improving the agriculture water resource utilization efficiency and water use efficiency, and developing water saving agriculture. The WSA-LP is composed of wheat mulching sub-system (WMS),fertilizer recommending sub-system(FRS),farming system recommending sub-system(FSRS),chemicaI water saving agent recommending sub-system(CWSARS) and wheat variety choosing sub-system(WVCS), including 16 databanks(the total records beyond 500) and 9 rules banks(the product rules beyond 1000). The WSA-LP is established both based on the PAID expert system developing platform and under network environment, and the modularized design idea is applied.The rule system and system flow chart is compiled under the conclusions of wheat mulching cultivation knowledge on Loess Plateau. The WMS is composed of specialist recommending module, instance module, the amount of soil film estimating module, frequent ask question & machine module. The weight-based rule, the product rule and the model-based rule is established to present the knowledge. The wheat mulching experiment bank of part Loess Plateau ,the spatial resolving rate is town(Xiang in China), is set up. The model for calculating the amount of soil film, which adopts to irregulars field, is also given.The wheat fertilizer recommendation technology system for the agriculture production on Loess Plateau is advanced by the fertilizer experiment data. The regression equation between base moisture of soil and yield is set up, in which the soil base moisture of previous year is used as a limited factor on the yield of later year, which can forecast a upper limit of the yield with 95% statistical significance, so the opportunity which appears when the yield determined by user disagrees to the reality, and the degree that fertilizer recommend result tallies reality is improved. The databank of model between yield under fertilizer effect on Loess Plateau are established, so the system estimate ability is enhanced. Two fertilizers distribution regime is adopted, which is favorite to different user choosing.The FSRS is designed by the object-oriented method in which the condition that determines a farming system feasible or unfeasible are accumulate temperature (growth period) and precipitation. Four object classes, crop object class, region object class, control bar class, work window field object class, is created. The object is managed by itsattribute, and the data is delivered by massage, which achieves the system modularized.The CWSARS is set up on the summary of the research result of and application experience of chemical water saving agent. CWSARS, by product rules, can recommend the proper water saving agent according to the crop variety, growth stage and soil conditions. CWSARS integrates the experimental result, so it is conveniences for user's selecting.The WVCS is designed after the collection and analysis of the wheat variety description issued by the authoritative department. The knowledge source which have relative stable form and glossary is analyzed, and the data mining step for this kind of knowledge is designed, viz. filter by key word-intercept-field specialist confirm. The standard glossary table and the "value" of each words is established, which enhance the expert system ability to knowledge acquisition and light the rough task of knowledge banks maintenance.