Anomaly detection for health assessment and prediction of diesel generator set

Abstract

Diesel generator set is widely used in a variety of fields including industry, agriculture and daily life. In order to obtain the operation status information of generator set in time to facilitate health management and fault prediction, a wireless sensor network based data collection system is developed. Most of the sensor data is streaming time series data, in which anomalies provide important information in critical situations. Hierarchical temporal memory (HTM) is a technology of cone neuron model based on the interaction between neuroscience and physiology of pyramidal neurons in the cerebral cortex of the human brain. HTM learns time-based patterns in unlabeled data on a continuous basis and it is very robust to noise and high capacity. HTM is used for anomaly detection for diesel generator set health assessment and prediction, and the primary results show that HTM base anomaly detection method is superior to other anomaly detection methods and has the potential to be used in the health assessment and prediction of diesel generator set.