Fault diagnosis based on improved D-S evidential theory for hydroelectric generator units
Cheng Jiatan · Journal of Hydroelectric Engineering · 2014
An integrated fault diagnosis method based on improved D-S evidence theory is presented for vibration faults of hydroelectric generator units to improve conventional diagnosis of low reliability. It applies a hybrid algorithm of simulation annealing-particle swarm optimization to basic probability assignment of evidence theory according to frequency domain signals of vibration faults, realizes reasonable probability assignment by improved D-S theory, and finally implements fusion decision of multi-source information. Case studies show that this method is reliable in vibration fault diagnosis whether the original evidences are in high conflict or consistent and it achieves effective recognition of vibration failure modes of hydroelectric generator units.