| Rice is one of the main grain crops in China,and the planting areas are relatively extensive,mainly distributed in the northern producing areas represented by Heilongjiang and the southern producing areas represented by Jiangsu and Anhui.With the deepening of scientific research and the promotion of technology,rice production has continued to rise in recent years,and the fundamentals of supply and demand are relatively loose.However,in the drying process after the harvest of rice,there is currently no mature drying equipment integrated with the domestic selfdeveloped online monitoring system for water collection that has been widely promoted.In order to improve the intelligent level of agricultural equipment and ensure the quality and yield of rice,the research of online detection technology of rice moisture content is of great significance.Online detection technology of rice moisture content refers to the technology of simultaneous moisture content detection in the actual harvesting,drying and other working processes of rice,and detection accuracy is an important factor restricting the development of this technology.In order to improve the accuracy of online detection of rice moisture during the drying process,this paper analyzes the principle of capacitive moisture content detection and its influencing factors,conducts electrostatic field simulation and optimization tests on the electrode plate,designs a dynamic acquisition device and detection system,establishes a prediction model of genetic algorithm to optimize BP neural network,predicts and corrects the moisture value of the upper computer,and carries out indoor static test and indoor dynamic test to verify the accuracy,stability and reliability of the device.The main research contents of this paper are:(1)By reviewing the literature and considering the test conditions of the device,the flat plate structure was used as the capacitor structure of the capacitive sensor,and the equivalent model of the three-pole plate capacitor was established,and the relative permittivity and pore ratio of dry matter,water and air of rice were obtained as the main factors affecting the moisture content detection results of rice,and the edge effect of flat plate capacitors was analyzed.(2)The electrostatic field simulation analysis of the triode plate capacitor was carried out by COMSOL software,and the relationship between the plate thickness,plate spacing,relative area and edge effect of the capacitor was obtained.Through the three-factor and five-level Central Composite experimental design of the plate thickness,plate spacing and relative area of the triode plate capacitor,the optimal plate structure parameters of the triode plate capacitor are obtained.(3)This device is mainly composed of rice dynamic collection device and real-time detection system.The dynamic acquisition device was modeled in 3D using UG software,the overall structure and working principle of the dynamic acquisition device were determined,and the design of each component of the dynamic acquisition device was introduced in detail.Altium Designer software was used to design the rice moisture content detection circuit with STM32F407ZGT6 as the core control chip,and the capacitive sensor,temperature sensor,power module and communication module were selected as the main modules of the detection circuit.(4)The information fusion of capacitive sensor and temperature sensor is realized by genetic algorithm optimization BP neural network,the process of genetic algorithm optimization BP neural network based on MATLAB software is described in detail,the regression model of rice moisture content is established through calibration test and the model is verified,the data obtained by calibration test is used for neural network training,and the GA-BP neural network prediction model is established,and the prediction result is better,which can provide reference for the moisture correction of the upper computer.(5)Indoor static test is carried out,and the actual detection range of the device is 9%～28%,and the accuracy error,repeatability and sensitivity are in line with the expected design requirements.The feasibility of the device was verified by indoor dynamic test,and the average relative error between the online detection result and the actual moisture content(measured by the constant weight method at 105°C)was 2.103%,which met the expected online detection accuracy requirements of bench test. |
