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.
Dreadful hyperparmeters limit the performance of deep learning model. In this paper, we present a novel Kernel Principal Analysis(KPCA)-Deep Belief Network(DBN) hyperparameter optimization approach to diagnose conditions of rolling bearings. The method involves the construction of KPCA-DBN from fuzzy relations, where KPCA is used to reduce the dimension of bearing vibration data. Considering a multi-object optimization problem, Particle Swarm Optimization(PSO) is further performed on the hyperparameter optimization of KPCA-DBN. Experiment results demonstrate that the proposed method could effectively find the optimal hyperparameters.
Inspec keywords: condition monitoring; fault diagnosis; vibrational signal processing; vibrations; deep learning (artificial intelligence); fuzzy set theory; particle swarm optimisation; principal component analysis; rolling bearings; belief networks
Subjects: Vibrations and shock waves (mechanical engineering); Inspection and quality control; Optimisation techniques; Principal component analysis; Mechanical components; Mechanical engineering applications of IT; Combinatorial mathematics; Neural nets; Optimisation; Combinatorial mathematics; Optimisation techniques; Civil and mechanical engineering computing; Principal component analysis; Signal processing and detection; Inspection and quality control; Statistics; Combinatorial mathematics; Maintenance and reliability