Research On The Key Technology Of Multi-component Detection Of Nutrient Solution

In recent years, problems such as soil pollution, water shortage and agricultural land reduction have become more and more serious. Soilless culture technique has been drawn wide attentions due to its advantages of high utilization rate of water resources and nutrient components and less space needs. Automatic detection system of nutrient solution provides data to the control system. Thus it is an important part of soilless culture precise control system and of great research significance. Ion selective electrode (ISE) techniques make possible automatic detection of nutrient solution components. However, the cross-sensitivity phenomenon of ISE limits its application in the nutrient solution detection. Meanwhile, there are no mature electrodes available for the detection of some elements such as SO42-,H2PO4-. Researches on calibration of ISE cross-sensitivity and the soft sensors of SO42- and H2PO4- are key technology for automatic detection system of nutrient solution.The main work of this dissertation is as follows:1. Cross-sensitivity inhibition of ISE based on least squares support vector machine regression (LS-SVR)Firstly, the cross-sensitivity experiments of K and Ca ISE is conducted. Then LS-SVR was used to build cross-sensitivity inhibition model for ISE, and the regularization parameter is determined by Fuzzy reasoning. At last field soilless culture data is used to validate the effect and the result shows this model can reduce cross-sensitivity interference.2. Soft sensor of the nutrient solution componentsThe change regularities of the nutrient solution composition are studied. LS-SVR is used to model soft sensors for SO42-,H2PO4- according to the measurable parameters of the nutrient solution. The experimental results show the soft sensors are effective and feasible.3. Intelligent on-line detection system of nutrient solution is developed.