Based On Radial Basis Function (rbf) And The Ds Evidence Theory Circulating Nutrient Solution Of Multi-component Online Testing System

Greenhouse agriculture, the factory, intensive and technological agriculture, is a future tendency of modern agriculture. Hydroponics is an important plant method under greenhouse environment. The on-line detection of nutrient-liquid is a precondition of the intelligent adjustment and control for nutrient-liquid. The appropriate prescription of nutrient-liquid ensures crop to grow healthily under optimal environment. Therefore, it is necessary to establish the system of the multi-component detection on-line for circulation nutrient-liquid. On-line detection of nutrient liquid is used to adjust components concentrations of nutrient liquid in term of variable growth requirement. This research is helpful to provide an efficient approach from greenhouse agriculture to precision agriculture and concentrated agriculture .Through study of correlative contents of advanced computer cybernetics, artificial intelligence , the domain knowledge and special crop growth mechanism in greenhouse, we present the system of Multi-sensor Data Fusion (MSDF) based on Radial Basis Function Network (RBF) to implement on line detection for nutrient-liquid , which may realize multiple components detection on-line, for example NO3-, Cl-, Ca2+, PH, EC, NH4+, K+ and so on.The soft sensor's mechanism is introduced to overcome the limitations of sensor's manufacturing process. To improve the believe-degree of soft sensor's result, we analyze soft sensor's result by uncertain inferential capacity and combination rule of evidential theory. At the same time , we establish the experiment system of component detection on-line of circulation nutrient-liquid according to above theoretical frame .Experiment results show that MSDF's results approach to practice with less than ?% error in effect. That proves validity and credibility of theoretic frame as above. In summarization, Multi-sensor Data Fusion (MSDF), which realizes complex components detection on-line, is valid and can provide warrant of intelligent dynamic regulation and control for nutrient-liquidIn addition, this paper gives an introduction to the design and realization of MSDF system in detail, which include Single Chip Microcomputer System of sampling and preprocessing, management software system by Delphi, Matlab and communication between Micro-computer and Single Chip Microcomputer.