基于机器学习算法的农田挥发氨多传感器阵列检测技术研究

Rapid detection of volatile ammonia in farmland using mulitsensors with machine learning algorithms

【目的】设计能够快速、低成本、便捷检测农田挥发氨装置。【方法】构建基于二氧化锡(SnO_(2))半导体气体传感器阵列检测系统,并在新鲜空气(氨气质量浓度为0 mg·m^(-3))和氨气质量浓度分别为75.9、151.8、303.6 mg·m^(-3)条件下,以及混合有乙醇的空气、纯乙醇气体(质量浓度为151.8 mg·m^(-3))、混合有氨气的空气和纯氨气气体(质量浓度为151.8 mg·m^(-3))样品下,通过主成分分析法(principal component analysis,PCA)、K-最近邻算法(K-nearest neighbors,KNN)和支持向量机算法(support vector machine,SVM)对多传感器阵列响应稳态阶段和暂态阶段的数据进行分类处理,分析该系统对不同质量浓度氨气和混合气体环境下氨气的区分效果。【结果】该装置能够明显区分不同质量浓度氨气,稳态阶段的主成分1值超过90%。KNN与SVM算法稳态阶段平均准确率超过97%,暂态阶段平均准确率68%,KNN与SVM平均分类准确率为68%。【结论】该多传感器阵列检测系统不需要等待传感器进入稳态阶段便可以读取数据,有助于农田环境中氨气快速和连续检测。

【Objective】Design a device for the rapid,low-cost,and convenient detection of ammonia in farmland.【Method】An electronic nose device based on a tin dioxide(SnO2)semiconductor gas sensor array was constructed.Under conditions of fresh air(ammonia mass concentration of 0 mg·m^(-3))and ammonia mass concentrations of 75.9,151.8 and 303.6 mg·m^(-3) respectively,as well as air mixed with ethanol,pure ethanol gas(mass concentration of 151.8 mg·m^(-3)),air mixed with ammonia,and pure ammonia gas(mass concentration of 151.8 mg·m^(-3)),the data responded in the steady-state stage and transient stage of multisensory array were classified by using principal component analysis(PCA),K-nearest neighbors(KNN),and support vector machine(SVM)algorithms.Furthermore,the system’s ability to differentiate between different mass concentrations of ammonia and mixed gas environments was evaluated.【Results】The device can clearly differentiate between different mass concentrations of ammonia.In the steady-state phase,the PC1 proportion exceeds 90%,and both KNN and SVM algorithms achieve accuracy rates exceeding 97%.While in the transient phase,the average accuracy rate is 68%,and the average classification accuracy for KNN and SVM is 68%.【Conclusion】The multisensor array detection system can read the data without waiting a steady-state phase,which can facilitate the rapid and continuous detection of volatile ammonia in farmland.