Improving fault diagnosis precision in flexible manufacturing systems through PSO-BFA hybrid optimisation and multimodal data fusion

View at publisher Summarise with VIRAT
Authors: Mingwei Li, Pengbo Shi, GuangHua Yang, Na Li, Xin Li
Cited by: 0
References: 30

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

To overcome the problems of low iteration efficiency and a propensity to drift towards local optimal solutions when choosing optimal parameters in the standard CNC machine tool fault data training model, an intelligent optimisation technique based on PSO-BFA is proposed. By simulating the bacteria’s local foraging activity, the BFA algorithm is introduced, giving the particle swarm the qualities of local convergence, replicability, and migration. This enhancement improves the model’s fitness value, facilitates precise local parameter optimisation, and accelerates convergence towards the optimal solution. The fault data training model determines the average fitness value of the particles in the swarm, selects a deep confidence network model with customisable scale, and uses the PSO method to get the ideal value in the global range. When the training process rapidly converges to the optimal solution, the ability to identify problematic spots and diagnose problems is improved. Simulation results show that the PSO-BFA intelligent optimisation method outperforms the traditional swarm intelligence strategy in multi-fault diagnosis and classification, achieving the peak fitting value in less iterations.

Fields of Study & Subjects

Metaheuristic Optimization Algorithms Research Machine Fault Diagnosis Techniques Engineering Technology and Methodologies

Publication details

Journal / source
Australian Journal of Mechanical Engineering
Pages
1–12
Publication date
25 March 2026
Language
EN