Senior ScientistContact Madhav
An essential part of the Prognostic and Health Management of rotating machines is dedicated to diagnosis operations, where the reliable early diagnosis of single and multiple faults is an important requirement for many practical scenarios of machine operations
In this presentation, we focus on well-known challenges of multiple faults, and discuss the solution based on a new blended ensemble framework utilising Convolutional Neural Network with Support Vector Machine (BECNN-SVM) model for multiple and single faults diagnosis of usual suspects in rotating machinery: gears, bearings and shafts. The proposed approach benefits also from pre-processing of the acquired signals using complementary signal processing techniques, forming thus inputs to 2D Convolutional Neural Network base learners. The outputs from these networks then are fused using a meta-learner for fault detection purposes. Discriminative properties of the complementary features ensure the high capabilities of the approach and provide good results under different load, speed, and fault conditions of the rotating machinery components. The experimental results show that the proposed method can detect faults with good accuracy and when compared to other state-of-the-art methods, it offers improved overall effectiveness in the fault diagnosis process.
Free of charge
Dr. Ivan Petrunin is a Senior Lecturer in Signal Processing for Autonomous Systems at the School of Aerospace, Transport and Manufacturing of the Cranfield University, UK.
He has extensive knowledge and experience in applied signal processing for various industrial and civil engineering applications related to rotating machinery and, more recently, related to autonomous system applications for detection and classification of weak non-stationary signatures in RF & acoustic.