基于半监督学习和稀疏表示的质谱仪运行灵敏度故障检测算法

A fault detection algorithm for running sensitivity of mass Spectrometer based on semi-supervised learning and sparse representation

为了解决质谱仪运行灵敏度故障对环境监测和化学污染监测产生的影响,提高监测数据的准确性和可靠性,提出了一种基于半监督学习和稀疏表示研究质谱仪运行灵敏度故障检测算法。根据半监督学习的原理对质谱仪的运行数据进行划分,在混合数据中标记运行灵敏度的故障特征。通过对应特征处理原始数据,在去噪和归一化特征的基础上求取灵敏度系数。采用稀疏表示理论对系数转换,以约束范数关系建立故障监测目标函数,检测质谱仪运行灵敏度故障。结果表明:以3种类型运行灵敏度故障作为测试对象,新算法可以实现较高精度的故障检测,且不受样本数量的限制,具有一定的应用价值。

In order to solve the impact of mass spectrometer operating sensitivity failure on environmental monitoring and chemical pollution monitoring,and improve the accuracy and reliability of monitoring data,a fault detection algorithm for mass spectrometer operating sensitivity based on semi supervised learning and sparse representation is proposed.According to the principle of semi supervised learning,the operation data of mass spectrometer was divided,and the fault characteristics of operation sensitivity were marked in the mixed data.By processing the raw data with corresponding features,the sensitivity coefficient was calculated based on denoising and normalization features.Using sparse representation theory to transform coefficients,a fault monitoring objective function was established based on constrained norm relationships to detect sensitivity faults in mass spectrometer operation.The experimental results showed that the new algorithm could achieve high-precision fault detection with three types of sensitivity faults as test objects,and was not limited by the number of samples,which had certain application value.