Early Classification of Industrial Alarm Floods Based on Semisupervised Learning
Haniyeh Seyed Alinezhad, Jun Shang, Tongwen Chen
Abstract
Early classification of ongoing alarm floods in industrial monitoring systems is crucial to provide a safe and efficient operation. It can provide online decision support for plant operators to take timely action, without waiting for the end of an alarm flood. In this article, a data-driven approach is proposed to address the early classification problem with unlabeled historical data. To prioritize earlier activated alarms and take advantage of the triggering time information of alarms, a vector representation called exponentially attenuated component (EAC) is used to represent alarm floods. This makes alarm sequences fit for different powerful machine learning algorithms, which can be easily implemented online with acceptable computational complexities. A method based on the time information of unlabeled historical alarm floods is formulated to determine the attenuation coefficient for EAC representation. With the Gaussian mixture model, an efficient semisupervised approach is proposed to provide an early classification of alarm floods using unlabeled historical data. It includes two phases: offline clustering and online classification, where the clustering step is automated in terms of choosing the optimal number of clusters by applying an efficient cluster validity index. The efficiency of the proposed method is validated by the Tennessee Eastman process benchmark and a real industrial dataset.