Research On Data Fusion In Seed Moisture Content Measurement

With the development of modern agriculture, the method of improving seeds'quality and increasing burgeon rate by selecting, drying and coating series of processing steps becomes more and more important to heighten efficiency and quality of agricultural products. During drying course, the core technique is the online test and control technology, which is the essential part of the processing system. Due to property of seeds, there is few methods for measuring moisture content directly. However, standard method for seed moisture content couldn't reach the requirements of online measurement. It limits development and application of seed drying machine.In this paper, we go deep into analyzing the method of testing and controlling in theory,and explain the way of system designing from aspects of mechanics and circuit, introduce the design of temperature sensor and weight sensor. Capacitive sensor structure is optimized by using AD7746 single-ended input mode. The data from temperature, weight and capacitive sensor are merged using data fusion method by multi-dimension regressions, parameter estimation and artificial neural networks method respectively. Result indicates that it works well in measurement precision,real time and repetition of moisture instrument. In this dissertation, attention is put on three key technologies: structure of seed drying instrument, design of sensors circuit and data fusion in seed moisture content measurement.Firstly, the background and significance of research is analyzed, principle and common methods of modern moisture measurement technology, especially the realistic meaning of celerity measurement is introduced.Secondly, design of structure of mechanical and sensors circuit,including the structure of seed drying instrument, temperature sensor and weight sensor circuit. The performance after amelioration is compared and analyzed. Insufficiency of former capacitive sensor circuit is discussed. Design capacitive circuit, including CMOS chip choosing and periphery circuit design.Thirdly, preparation of sample of seed moisture measurement, design and implement of project, analysis of data which affect moisture measurement. The relationship of moisture, capacitance, temperature and weight has been pointed out.Fourthly, purpose, application, foreground and methods of data fusion technology have been discussed. Analysis of data measured with different methods, and optimized mathematics model of seed moisture measurement is brought forward based on contrast of several parameters. The particular operating of data fusion methods based on multi-dimension regressions, artificial neural networks and parameter estimation are used. The validation is proved by increasing the measurement precision and reducing the uncertain factors by using data fusion technology.