Diagnosis of Rotating Machines Faults Using Artificial Intelligence Based on Preprocessing for Input Data

Abstract

machines are extensively utilized in industrial life, since it represents a vital element in industrial processes. Therefore, early detection of faults of rotating machines is necessary to avoid the forced stopping for frequent maintenance in industrial processes. Various condition monitoring and detecting procedures are used to diagnose the rotating machinery faults based on vibration signature analysis, temperature monitoring, noise signature analysis, lubricant signature analysis, Artificial Intelligence (AI) techniques. many AI methods are in use for bearing defects diagnosis. For instance, Fuzzy Inference System (FIS); Artificial Neural Networks (ANNs) and Adaptive Neuro-Fuzzy Inference System (ANFIS). AI techniques are advanced technologically for classifying and detection various rotating machinery faults. This research work describes an inclusive method to some extent for detecting and classification of bearing faults using two methods of artificial intelligence which are ANN and ANFIS. The study is implemented offline in MATLAB environment. The proposed data were taken via using accelerometer which is mounted on the bearing housing in Machinery Fault Simulator (MFS) by M.Samy [1]. The obtained data of rolling element bearing were classified into four main conditions: Healthy, Outer Faulty, Inner Faulty, And Ball Faulty. The four conditions were imported to ANNs and ANFIS models. This work presents a comparison between the diagnosing based on Fast Fourier Transform (FFT), ANNs and ANFIS. The input data were preprocessed before entering to ANNs and ANFIS models by using three techniques as the following: the normalized data in range (0-1), the time domain features, and finally the Auto Regressive (AR) model. The accomplished outcomes of ANN and ANFIS models in case of AR model give high accuracy results in classification issue. The achieved outcomes are encouraging and promising in the field of diagnosis of mechanical machinery faults.

Keywords

• fault diagnosis
• bearings
• artificial neural networks
• adaptive neuro-fuzzy inference system
• mechanical faults
• fault detection
• artificial techniques
• normalized data
• time domain features
• auto regressive (ar) model