For access to this article, please select a purchase option:
IET members benefit from discounts to all IET publications and free access to E&T Magazine. If you are an IET member, log in to your account and the discounts will automatically be applied.
Your recommendation has been sent to your librarian.
There are common faults in planetary gearbox that it is not suitable for shutdown detection at the initial stage or there are many kinds of faults which are not easy to classify accurately. Based on the above reasons, this paper proposes a method combining multi population genetic algorithm (MPGA) and BP neural network. Traditional BP neural network uses a variety of genetic algorithms to optimize the initial weights between layers and the initial threshold corresponding to the single layer network. The traditional method greatly increases the global optimization ability of BP neural network when gradient drops, so we avoid the problem that the local optimal of selecting initial weight and initial threshold. This paper uses the tradition BP neural network and optimized MPGA BP neural network to classify the common faults of planetary gearbox. Then we compare the results of traditional BP neural network and MPGA BP neural network in planetary gearbox fault classification. The results show that: MPGA BP neural network has higher prediction accuracy than traditional BP neural network, so this method can be used for fault classification of planetary gearboxes.
Inspec keywords: fault diagnosis; backpropagation; mechanical engineering computing; gears; genetic algorithms; neural nets
Subjects: Mechanical engineering applications of IT; Optimisation techniques; Mechanical drives and transmissions; Civil and mechanical engineering computing; Optimisation; Neural nets