Automated Excavator Activity Diagnosis Via a Topology and Statistical Information Based Classifier

For automated diagnosis of excavator activities, a Topology and Statistical Information based Classifier (TSIC) is put forward, and it employs topology and statistical information of excavator activity samples. Specifically, a small sensor network is built on the excavator for its activity data acquisition. Distance metric learning is improved to explore sample features, making the same-class samples closer and different-class ones further. An improved mountain function is employed to construct covers for topology feature extraction, and then a weighted linear classifier is designed to diagnose (or classify) excavator activities. Generalization performance of TSIC is discussed. Experiments on excavator construction datasets and public datasets demonstrate competitive performance of TSIC.