| At present,our rice fertilization is still mainly based on traditional even fertilization.Using this way to fertilize not only consumes a large amount of fertilizer and low fertilizer utilization,but also wastes resources,increases production costs and causes extremely serious pollution to environment.Therefore,it is of great significance to reduce environmental pollution and improve fertilizer utilization rate by applying rice fertilizer accurately and in real time.Rice variable fertilizer technology can reduce the waste and loss of fertilizer and save the cost,which is the inevitable direction of the future development of rice fertilizer technology.Based on the analysis of the research status and characteristics of rice fertilization strategy models,rice fertilization machines,ultrasonic sensors,conductivity sensors and temperature sensors at home and abroad,a rice lateral depth variable fertilization control system integrated with a variety of sensors was designed to improve fertilizer utilization and reduce operation costs.The main research contents and conclusions are as follows:(1)In order to realize real-time data acquisition of rice soil parameters and real-time fertilization operation of control fertilizer drainer,the hardware part of this device was selected and assembled.The hardware part of this model was mainly composed of data acquisition module,data processing module,satellite positioning module and fertilizer discharge module.The data acquisition module includes ultrasonic sensor,soil conductivity sensor and temperature sensor.The ultrasonic sensor can detect the surface depth of rice soil;the conductivity sensor can detect the conductivity of rice soil in the area through which the machine passes;the temperature sensor can detect the internal temperature of rice soil and compensate the detection results of the ultrasonic and conductivity sensors.The satellite positioning module detects the information of position and operation speed of machinery and tools in real time,selects STM32 series single chip microcomputer as the data processing module,and sends the target fertilizer amount to the fertilizer controller of the fertilizer discharge module after calculating by STM32 single chip microcomputer.The fertilizer controller realizes lateral depth variable fertilizer operation by driving and controlling the rotation speed of fertilizer discharge motor.The test results show that the correlation coefficients(R2)of ultrasonic and temperature sensors were 0.998 and0.996,respectively.The maximum error of the conductivity sensor was 11μs/cm-1,and the detection error of the three sensors was less than 3%.(2)By studying the current situation and demand of rice variable fertilization strategy model and soil nutrient content prediction model,the neural network control model was adopted in the software of this control system.Firstly,the correlation between soil nitrogen content,surface depth and electrical conductivity was verified.Then,using Matlab software,soil electrical conductivity and surface depth parameters in soil sample test data were used as input neurons of the neural network model,and soil nitrogen content was used as output neurons to establish the neural network prediction model and verify the prediction accuracy.The results show that the root mean square error RMSE was 40.172,and the coefficient of determination R was 0.90231.In order to increase the prediction accuracy,the grey Wolf algorithm was used to optimize the traditional model.The RMSE of the optimized model was 7.29,and the average R was0.94341.Linear fitting of soil nitrogen content and nitrogen application amount in soil measurement formula data of Qixing Farm was carried out to obtain regression equation.Keilu Vision5 software was used for algorithm implementation of the optimized model and burned into STM32 microcontroller,so as to realize calculation and control of data.(3)In order to verify the performance of this model and fertilization module,the precision test of fertilizer discharge and field production verification test were carried out in the rice variable fertilization demonstration field of Qixing Farm,Jiamusi City,Heilongjiang Province from May 2021 to May 2022.The stability test results of fertilizer discharge system showed that the maximum inter-row consistent coefficient of variation of fertilizer discharge apparatus under the three paths was 2.23%.The maximum variation coefficient of stability of single row fertilizer drainer was 2.20%,and the maximum error of fertilizer system was 3.72%.Field production experiments showed that the variable fertilization model could save 10%~15%of fertilizer and the crops grew more evenly with this model than with traditional side deep fertilization. |