Multi-device Collaborative Anomaly Diagnosis Method for Electric Mining Dump Trucks

Electric mining dump trucks (EMDTs) play an essential role in open pit mining. EMDTs are complex electromechanical equipment characterized by messy status information and numerous failure modes. For other devices to also be able to identify new faults, the new fault data should be shared among multiple devices in the poor network environment of the mine. This article proposes a collaborative multi-device fault diagnosis method for EMDTs. At the individual device's level, the powerful multivariate time-series data feature extraction capability of bidirectional long short-term memory (BLSTM) networks is utilized to identify the EMDT's overall fault state. At the multiple devices’ level, federal learning is employed to achieve collaborative optimization of numerous devices’ diagnostic models to identify new faults effectively. Collect actual operational data on EMDT and simulate faults based on this data to validate the strategy for fault diagnosis.