An Improved Signal Pre-Processing Method for Gearbox Fault Features Extraction

Abstract

Time synchronous averaging (TSA) has been extensively applied for gearbox fault analysis. But it cannot unveil the vibration features precisely in conditions of different speeds and load excitation. This paper presents an improved method which is a combination of Time synchronous averaging (TSA) and Empirical mode decomposition (EMD) to extract the intermittent vibration feature from the oscillated vibration response for the fault diagnosis of the gearbox. In this technique, time-synchronous averaging and empirical mode decomposition is performed on the gearbox vibration response to detect fault features. First, TSA is performed on the gearbox raw vibration response. By applying TSA on raw vibration response, a noiseless vibration response with the power spectrum is produced. Finally, decomposition technique is applied on noiseless vibration response and a sequence of IMFs are produced with the power spectrum to obtain more enhanced and improved mono-component of noiseless regular vibration response. Simulation and Practical investigation of the proposed method show the advantages and effectiveness in extracting faulty vibration feature with different speed and load excitation. An experimental investigation has been carried out on the Machinery Fault Simulator with one healthy and two faulty gearbox conditions i.e., healthy tooth, fragmented tooth and missing tooth under three different speed and load excitation.