| Grain yield has spatial variability in the farmland, and it can reflect the growth and management situation of crop. Therefore, the accurate information of yield space distribution is an important foundation for implementing precision agriculture. Many domestic and foreign researchers have put forward a variety of grain yield monitoring methods and models, however, the accuracy and generality of the models are not satisfactory. Therefore, a set of grain yield intelligent monitoring network was build. Meanwhile, on the basis of a dual plate and impact-type grain flow sensor, a grain yield monitoring system was also developed for Chinese harvester. Consequently, a certain foundation could be laid to achieve the commercialization of Chinese yield monitoring system. The researches mainly included:(1) Key technology research on vehicle-mounted yield monitoring systemTo weaken the vibration interference, a kind of shock absorber was used for the vibration reduction of grain flow sensor. Based on the research of elevator speed variation, the kinematics and physical mechanics of grain movement toward the impact-type sensor was analyzed, and then a mass model of the grain flow sensor named "voltage/elevater speed" model was put forward. The relationship between grain mass flow and impact force was described deeply by the model. Hence, it could eliminate the influence of elevater speed variation on grain flow signal and improve the estimation accuracy of grain yield.(2) Grain flow signal processing and yield modelingIn order to eliminate the vibration signal, a regression difference method was used to reduce the vibration interference, moreover, a dual threshold de-noising and two kinds of interpolation method were proposed to eliminate the singular value of the difference voltages. Furthermore, the effects of the wavelet de-noising method and adaptive noise cancellation filter were also studied.The verification results showed that the adaptive noise cancellation filter was better than wavelet de-noising method; however, the root mean square error (RMSE) of yield estimation was both greater than 5.00%; while the RMSE of yield estimation by regression difference method was minimum, which was 3.15%. It meant that the regression difference method was the best. On the other hand, fitting on the normalization of "voltage/elevator speed" parameters by least square method, the regressive "voltage/elevator speed" model of total yield was obtained, whose RMSE was only 2.03%, and it had a higher accuracy and robustness than the voltage model of total yield.(3) Yield mapping technology and error analysisThe error sources of yield map were given, especially, the errors caused by transmission delay and speed variation were studied, and the solutions were also presented. In addition, three types of interpolation methods were analyzed and adopted to generate the yield map; and then, the effectiveness was evaluated.(4) Development of remote grain yield monitoring and management systemThree platforms were designed and developed in the remote grain yield monitoring and management system including database management platform, grain yield remote monitoring and decision-making platform, as well as Web service release platform. The system could respectively realize a lot of functions, such as real-time yield information monitoring, yield information storage and query, yield point map drawing, track and playback, decision-making and information release, etc. In addition, a simple variable rate fertilization model based on yield variation was proposed, and the prescription map could be generated.(5) Development of grain yield monitoring system based on intelligent mobileA yield monitoring system based on Android mobile was developed through the intelligent mobile phone. It was possible to acquire the real-time yield data from remote grain yield monitoring and management system at anytime and anywhere. The system provided a very convenient method to obtain information by Web Service technology.(6) Optimization and field experiment of grain yield monitoring systemA bipolar acquisition circuit was designed, and the dual polarity signal could reflect the forward vibration of grain flow sensor. Consequently, it could improve the accuracy of difference signal and yield predicted. Secondly, the sampling frequency of yield monitoring system was increased and carried on optimization. The test results showed that the yield monitoring system with 50Hz sampling frequency had the highest cost performance, whose accuracy was higher and the cost was lower.Through the optimization of vehicle-mounted system, this study has improved the accuracy and practicability of the grain yield monitor, and using the intelligent grain yield monitoring system, it is convenient for data processing and query, which can promote the implementation of precision management on cereal planting. |
PDF Full Text Request