奶牛个体瘤胃碳排放量在线检测装备设计与试验

Design and Testing of Online Monitoring Equipment for Individual Cow Rumen Carbon Emissions

摘要:

发展奶牛个体瘤胃碳排放在线检测技术对于推动畜牧业减排降碳、实现持续高质量发展具有重要意义。针对我国奶牛养殖个体瘤胃碳排放检测/监测流程繁琐、效率低、适用装备缺乏等问题，设计了一种奶牛个体瘤胃碳排放在线检测装备，并提出了基于支持向量机（SVM）的奶牛个体日碳排放量预测模型。以STM32微处理器为主控制核心，集成RFID/甲烷/二氧化碳等传感器采集数据，采用LoRa/4G双模通信技术传输数据，基于Qt Creator软件/云平台分析与显示数据，实现奶牛个体碳排放数据本地/远程一体化监测、传输、分析与可视化显示。采用背景浓度扣除法与时间点对齐重构方式，构建了基于SVM的奶牛个体碳排放预测模型，实现浓度数据到奶牛个体碳排放量准确测算。在北京某牧场对25头荷斯坦奶牛进行了实地测试，与二氧化碳平衡法相比，基于SVM回归预测模型的奶牛平均日二氧化碳排放量、甲烷排放量、碳排放量MAE平均值分别为511.13、58.16、2 202.10 g/d;MRE平均值分别为3.78%、14.34%、8.55%;在牛舍打开风机情况下，SVM预测奶牛个体平均二氧化碳排放量、甲烷排放量、碳排放量波动幅度分别为10.85、8.15、255.35 g/d，均显著低于二氧化碳平衡法对应差值。研究结果能够实现奶牛个体瘤胃甲烷和二氧化碳排放量实时、快速、准确测定，可为农牧业碳排放量精准核算提供技术参考。

Abstract:

The online detection technology for rumen carbon emissions of individual dairy cows is of great significance for promoting emissions reduction and carbon mitigation in animal husbandry, and achieving sustainable high-quality development. In order to solve the problems of cumbersome, inefficient and a lack of suitable equipment for detecting individual rumen carbon emissions in China's dairy farming, an online detecting device was designed. Additionally, an SVM-based model for predicting daily methane emissions from individual cows was proposed. The equipment was mainly controlled by STM32 microprocessor, integrating RFID/methane/carbon dioxide sensors to collect data, and using LoRa/4G dual-mode communication technology to transmit data. Based on Qt Creator software/cloud platform, data was analyzed and displayed to achieve local/remote integrated monitoring, transmission, analysis and visualization of individual carbon emissions data of cows. To obtain the daily carbon emissions of individual cows, an SVM-based prediction model was constructed by using the background concentration deduction and time point alignment reconstruction methods. Field tests were conducted on 25 Holstein cows at a certain ranch in Beijing. Compared with the CO2 balance method, the SVM regression prediction model showed MAE averages of 511.13 g/d, 58.16 g/d, and 2 202.10 g/d, and MRE averages of 3.78%, 14.34%, and 8.55% for the average cattle emissions of carbon dioxide, methane, and carbon, respectively. Moreover, when the fan was turned on, the SVM model predicted that the average carbon dioxide emissions, methane emissions, and carbon emission fluctuations of cattle individuals were 10.85 g/d, 8.15 g/d, and 255.35 g/d, respectively. These fluctuations were significantly smaller than those derived from the carbon dioxide balance method. This study can achieve real-time, rapid, and accurate measurement of methane and carbon dioxide emissions from individual cow rumens. It can provide technical support for the precise accounting of carbon emissions in the agricultural and livestock sectors.