A Fault Diagnosis Framework for Centrifugal Pumps by Scalogram-Based Imaging and Deep Learning

Abstract

Centrifugal pumps are the most vital part of any process industry. A fault in centrifugal pump can affect imperative industrial processes. To ensure reliable operation of the centrifugal pump, this paper proposes a novel automated health state diagnosis framework for centrifugal pump that combines a signal to time-frequency imaging technique and an Adaptive Deep Convolution Neural Network model (ADCNN). First, the vibration signals corresponding to different health conditions of the centrifugal pump are acquired. Vibration signals obtained from the centrifugal pump carry a great deal of information and generally, statistical features are extracted from the vibration signals to retain meaningful fault information. However, these features are either insensitive to weak incipient faults or unsuitable for tracking severe faults, thus, decreasing the fault classification accuracy. To tackle this problem, a signal to time-frequency imaging technique is applied to the pump vibration signals. For this purpose, Continuous Wavelet Transform (CWT) is applied to decompose the vibration signals over different time-frequency scales and extract the pump fault information in both the time and frequency domains. The CWT scales form two-dimensional timefrequency images commonly referred to as scalograms. The CWT scalograms are then converted into grayscale images (SGI). Over the past few decades, CNN models have been established as an effective practice to process images for classification and pattern recognition. Consequently, the extracted CWTSGIs are finally provided as inputs to the proposed ADCNN architecture to achieve feature extraction and classification for centrifugal pump faults. The performance of the proposed diagnostic framework (CWTSGI + ADCNN) is validated with a vibration dataset collected from a testbed specifically designed for centrifugal pump diagnosis. The experimental results suggest that the proposed technique based on CWTSGI and ADCNN outperformed existing methods with an average performance improvement of 4.7 -15.6%.

Monitoring and Predictive Maintenance of Centrifugal Pumps Based on Smart Sensors

“…These failure forms will cause changes in the pump vibration. Therefore, a sensor and IoT system [16][17][18][19][20] can be designed to form a digital unit and an industrial internet platform for smart pumps, which can effectively perform signal acquisition, analysis, feature extraction, and fault diagnosis, and realize the digital upgrading of pump equipment and the intelligent operation and maintenance of pump units [21,22]. The bearing is an important part of the centrifugal pump that supports the rotor.…”

A Method for Pipeline Leak Detection Based on Acoustic Imaging and Deep Learning

“…A CWT transforms a time-domain complex signal into the time-frequency domain with the help of the source wavelet function. The source or mother wavelet is usually a short time-based signal such as a vibrating signal having cycles [20]. The mother wavelet is the basis for decomposition, and a complex signal is decomposed into coefficients local- Step 1: AE signals are collected from the pipeline testbed using AE sensors;…”

“…Deep learning (DL) methods can solve the problem of leak-related information extraction and classification [20]. DL algorithms enable feature extraction from the image without human intervention [21].…”

Efficacy Study of Fault Trending Algorithm to Prevent Fault Occurrence on Automatic Trampoline Webbing Machine

“…Furthermore, a CWT transform can be utilised as a feature extraction method to integrate with deep learning algorithms and build a system that can autonomously identify the fault at its incipient stage. Hasan et al [28] applied CWT as a tool to convert the scalogram to grey-scale images and then apply an adaptive CNN (convolutional neutral network) to identify the pump condition.…”