Multi-feature Extraction for Bearing Fault Diagnosis Using Binary-tree Mahalanobis-Taguchi System

Rolling bearings are widely used in rotary machinery systems, their condition directly determines the normal operation of machinery, and effective recognition of bearing fault can greatly reduce maintenance time and cost. This paper proposes an integrated model which combines Binary-tree and Mahalanobis-Taguchi system (BT-MTS) to detect faults of rolling bearing. The BT-MTS consists of three stages. Firstly, the time domain, frequency domain, and time- frequency domain features are extracted from the vibration signals by feature extraction technology, such as complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN). Then, the Fisher ratio method is utilized at each node of the binary tree to design the architecture of the BT-MTS modeling. Ultimately, the trained BT-MTS is used to predict the fault types. The experimental results demonstrate that the proposed method is effective in recognizing the different categories of rolling bearings faults.