基于介电特性的稻田水体稻瘟灵监测装置研制与试验研究

Development and Experimental Research on the Monitoring Device of Isoprothiolane in Paddy-field Water Based on Dielectric Properties

【目的】稻田农药使用为稻田作业中至关重要的一环,精准施药对于农业生产、环境及水资源保护等方面具有重要意义。通过以稻田常用农药稻瘟灵作为研究对象,设计农药含量实时监测装置以实现稻田水体稻瘟灵含量的远程实时监测,为水体相关农药快速检测提供科学理论依据。【方法】采用介电特性与LoRa技术开发了一种稻田水体稻瘟灵含量的实时在线监测装置。装置的硬件部分包括硬件电路与结构组件。该装置的硬件电路以STM32单片机作为主控,相关模块包括激励信号源模块、检测与信号采集模块、输入输出模块;结构组件由太阳能面板、主控电路盒、支撑部件和检测探头盒4个部分组成,采用监测站式安装工作。软件部分以MDK 5.0为开发环境,采用C语言开发设计。程序主要由初始化程序、AD9859扫频程序、数据采集程序、LibSVM程序、串口发送程序、LoRa配置程序和LCD显示程序组成。试验对幅值比、相位差、幅值比原始电压、相位差原始电压4种数据经CARS(competitive adaptive reweighted sampling)特征提取后建立了SVR回归预测模型。【结果】对该装置进行了试验性能验证,以稻田水体作为空白水样,依次添加稻瘟灵原液后得到被测样本。在25℃的试验温度下,针对0~44 mg/L质量浓度的23个低浓度稻瘟灵溶液样本进行检测。其中以幅值比建立的预测模型决定系数R^(2)高达0.9959,RMSE仅为2.433 mg/L,且4种预测模型R2都在0.99以上。通过对频率点特征提取,使得单次检测仅需30 min,可实现快速、简便、准确的测量稻瘟灵含量。【结论】该装置对于稻田水体稻瘟灵含量具有较高精度的检测,单次检测速度较快,稳定性高。同时设备具有全天候工作能力,配合LoRa物联网可实现远程在线实时监测。

[Objective]The use of pesticides in paddy fields is a crucial part of paddy field operations,and precise pesticide application is of great significance to agricultural production,environment and water resources protection.By taking isoprothiolane,a commonly-used pesticide in paddy fields,as the research object,a realtime monitoring device for pesticide content was designed to realize the remote real-time monitoring of the content of Isoprothiolane in paddy field water,thus providing a scientific theoretical basis for rapid detection of water-related pesticides[.Method]A real-time online monitoring device for isoprothiolane in paddy field water was developed by using dielectric properties and LoRa technology.The hardware part of the device includes hardware circuits and structural components.STM32 single-chip microcomputer was used as the main control of the hardware circuit,which included excitation signal source module,detection and signal acquisition module,and input and output module.The structural components are composed of four parts:solar panel,main control circuit box,support component and detection probe box.This device was installed in the way of monitoring station.The software part used MDK 5.0 as the development environment,and C language was used to develop the software.The program was mainly composed of initialization program,AD9859 frequency sweep program,data acquisition program,LibSVM program,serial port transmission program,LoRa configuration program and LCD display program.In the experiment,the SVR regression prediction models were established after the four kinds of data of amplitude ratio,phase difference,amplitude ratio raw voltage and phase difference raw voltage were extracted by Competitive Adaptive Reweighted Sampling(CARS)feature.[Result]The experimental performance verification of the device was conducted.The paddy field water was used as a blank water sample,and the isoprothiolane stock solution was added to obtain the tested sample.At the experimental temperature of 25℃,23 low-concentration isoprothiolane solution samples in the concentration range of 0-44 mg/L were tested.The coefficient of determination R^(2) of the prediction model established by the amplitude ratio was up to 0.9959,the RMSE was only 2.433 mg/L,and the R^(2) of the four prediction models were all above 0.99.It only takes half a minute for a single detection by extracting the features of the frequency points,which could realize fast,simple and accurate measurement of isoprothiolane content.[Conclusion]The device can detect the content of Isoprothiolane in paddy-field water with high precision.The single detection speed is fast and the stability is high.At the same time,the device had all-weather working ability,and it could realize remote online real-time monitoring with the LoRa internet of things.