Development Of Portable LVF Spectrometer Applied To Component Information Detection Of Crops

With the rapid development of computer technology and remote sensing technology, near infrared spectroscopy is widely used in the research of crops’ component information detection. At present, most of the spectrometers applied in agriculture are developed for the purpose of academic research. Although they have the advantages of high accuracy and wide range of detection, they have some disadvantages when they were applied in the field, such as complex operation, large volume, inconvenient to carry, long detection period, no online testing and so on. Those spectrometers are very high to the professional quality of the operators and not easy to popularize application. With the demand of fast, miniaturization and low cost, a portable spectrometer which could detect component information of crops was developed. This research was supported by 863 Project(2012AA10A503),the Natural Science Foundation of Beijing(4142019) and the Ministry of Agriculture Public Benefit Research Foundation(201003008-07).The main research contents include the followings:1. The importance of rapid detection in crops’ component information in precision agriculture was expounded. The research status of spectrometers at home and abroad was analysed. The application requirements of spectrometers in crops, vegetables and fruits were analysed. The requirement analysis combined with the actual was made. The technical analysis according to the requirement analysis was made. The last, the overall design of the system was proposed.2. The hardware system of the instrument mainly consisted of two parts: the optical system and the electronic hardware system. The optical module respectively used the optical fiber and Linear Variable Filter(LVF) to complete the functions of incident light’s transmission and splitting. In the electronic hardware system, CMOS photoelectric detector and C9001 driving circuit were used to complete the function of photoelectric conversion. The data acquisition board was used to complete the function of processing and analysis signals. GPRS wireless transmission module was used to transfer data to the server. Each device was compared to ensure that the choice had a good effect, high integration and small size. Finally, the whole hardware system was built and the instrument was encapsulated.3. The operating system of the software was Win CE 6.0. The development environment was Visual Studio 2008. The main functions of interactive interface included spectral data acquisition, spectral curves displaying, dark current calibration, whiteboard calibration, spectral data saving, integral time setting, scan times setting, nutritional components analysis and displaying, transmitting the component information to the server by GPRS module. For facilitating the operation, this study designed a simple interface.4. This paper used a tunable laser to correct the error caused by the machining and assembly when the prototype design had been completed. The maximum deviation and the minimum deviation of standard wavelength and the wavelength before calibration were 11.506 and 0.947. The maximum deviation and the minimum deviation of standard wavelength and the wavelength after calibration were 4.577 and 0.481. It was feasible that wavelength calibration was necessary and had a certain effect.5. The experiments verified the feasibility of the prototype in detecting component information of wheats, vegetables and apples.(1) In the outdoor natural light conditions, the canopy spectral data of winter wheats in 5 main physiological periods was collected. Partial least squares method was used to establish prediction models of the content of chlorophyll and water in winter wheat. The correlation coefficients were 0.893 and 0.918 respectively. The results showed that the prototype was feasible in the detection of component information in winter wheat leaves.(2) Tungsten halogen lamp was used as the light source. Experiments were conducted on Chinese cabbage, rape, spinach and lettuce leaves. The values of 600nm-1100 nm spectrum were gotten. Multiple regression method was used to establish the prediction models of the four kinds of vegetables. The correlation coefficients were 0.854, 0.93, 0.73 and 0.696 respectively. The results showed that the prototype was feasible in the detection of component information in vegetables.(3) In the indoor halogen tungsten light conditions, experiments were conducted on apples. Partial least squares method was used to establish prediction models of sugar degree in apples. The correlation coefficient was 0.89. The results showed that the prototype was feasible in the detection of sugar degree in apples.The spectrometer prototype named the Portable LVF Spectrometer in the paper. The instrument could not only solve the problems of complex operation and inconvenient to carry in ordinary spectrometer, but also had the function of analysis results displaying in real time. The instrument embedded inversion models of the crops’ component information, and could output component information in real-time, such as chlorophyll, water content and so on. It allowed the operator to obtain the data more intuitively.