Research Of Multi-sensor Data Fusion In The Application Of Greenhouse Monitoring System

Multi-sensor data fusion technology is a national promoting scientific research projects recently.Many countries in the world have invested a lot of manpower and material resources to study it.Although data fusion technology is born in the military field,it has been extended widely in the civilian areas,especially in the intelligent greenhouse monitoring field.This thesis focuses on the key technologies of multi-sensor data fusion technology applied in the greenhouse monitoring.Firstly,the basic concepts of multi-sensor data fusion are introduced.The contradictions between the growing scale of facility agriculture and the backward of its monitoring means are analyzed.According to the greenhouse environment characteristics,the application of multi-sensor data fusion technology in the greenhouse monitoring system is proposed to improve the accuracy of monitoring which could guarantee the yield and quality of the greenhouse.Then,the characteristics of multi-sensor data fusion,the functional model,the structural model and the mathematic model are discussed.The existing data fusion algorithms are analyzed and compared and their advantages and disadvantages as well as the scope of application are obtained.For greenhouse monitoring application background,the thesis focuses on three algorithms of weighted average mean,Kalman filtering and D-S evidence theory.D-S evidence theory is chosen as the fusion algorithm after analysis.The traditional D-S evidence is improved in order to overcome the shortcomings of "fusion rules failure with big evidence conflicts","bad robustness","one-vote veto" and "element fuzzy".The improved algorithm works from both the consistency of evidence and the focus of element.In the consistency of evidence,distance function is used to represent the difference between evidences.The consistency parameters concluded from distance function are used as weights to reassign the basic probability assignment function.In the focus of element,the focusing factor and proportionality coefficient are used to adjust the degree of focus from large subset to small subset.The improved algorithm overcomes the drawbacks of the traditional algorithm from simulation analysis.Finally,the structure of a greenhouse monitoring system model is built on the hardware platform,using two-stage fusion and two-stage control.In the partial fusion center,a method of compatibility matrix is used to remove invalid data and optimize valid data.In the global fusion center,the improved D-S evidential reasoning algorithm is applied to multi-sensor data fusion.The experiment results show that the monitoring system can make an accurate environment state judgment of greenhouse and control actuator to adjust the greenhouse environment.