Real-time monitoring and diagnosis scheme for IoT-enabled devices using multivariate SPC techniques
提出一种多元统计过程控制方案，利用空间秩和改进的自适应弹性网算法检测物联网设备高维数据流中的稀疏均值偏移，并定位故障变量，适用于非正态分布和参考样本不足的情况，通过风电机组案例验证了实时监控与诊断效果。
This article is aimed at condition monitoring and fault identification for Internet of Things (IoT) devices, and proposes a multivariate statistical process control scheme. The new method aims to detect sparse mean shifts using spatial rank and an improved adaptive elastic net algorithm, which can monitor the high-dimension data stream collected by IoT devices and pinpoint faulty variables. The new method is also applicable in the presence of a non-normal distribution and insufficient reference samples. Numerical simulations verify that the proposed method has clear advantages over existing methods. The case of wind turbines shows that the method can be applied to real-time monitoring and diagnosis of real IoT devices, which could provide valuable diagnosis of root cause and optimize subsequent maintenance strategies.