Supervised-Learning-Based Intelligent Fault Diagnosis for Mechanical Equipment

Geonkyo Hong,Dongjun Suh

doi:10.1109/access.2021.3104189

Abstract

Recently, anomaly detection for improving the productivity of machinery in industrial environments has drawn considerable attention. As large-scale data collection and processing are becoming easier owing to technological developments, data-based deep-learning technology is being developed to detect anomalies in mechanical equipment operation. This study proposes an ensemble model that combines stacked two-dimensional and one-dimensional convolutional neural networks (CNNs), residual long short-term memory (LSTM), and LSTM based on supervised learning. The model, which is called the SCRLSTM model, can detect abnormal data generated by mechanical equipment. The proposed model can extract the spatial features of data using a CNN model and detect anomalous states in the time-series-based vibration datasets of machinery under various environments through residual LSTM. To verify this model, data augmentation was applied to the original time-series-based mechanical vibration dataset, which had unbalanced samples that lowered the performance of the abnormal anomaly detection model. In addition, an image-based analysis was performed by converting time-series-based raw-signal data to Mel-spectrogram images, thereby achieving better performance in the fault diagnosis system to which data augmentation was applied. The proposed SCRLSTM model shows better performance than other supervised-learning-based models on datasets having different lengths under various conditions. This indicates that the proposed anomaly detection model can be expected to improve the productivity of mechanical equipment in industrial settings.

Highlights

As the technology used with mechanical equipment advances, the complexity of industrial environments and uncertainties pertaining to productivity increase
SCRLSTM MODEL FOR FAULT DIAGNOSIS This study proposes a new model that combines 1D and 2D convolutional neural networks (CNNs) and residual long short-term memory (LSTM) and LSTM to detect anomalies in a time-series-based industrial machine vibration dataset
The main contribution of this study is that it overcomes the data imbalance problem in the time-series-based data for bearings and industrial machines through data augmentation

Summary

INTRODUCTION

As the technology used with mechanical equipment advances, the complexity of industrial environments and uncertainties pertaining to productivity increase. Machine learning has been used to study and develop time-series-based methods for anomaly detection of industrial machinery and bearing faults using data from mechanical equipment. Liu et al proposed a domain adaptive approach by developing a deep-learning-based JDDA model for fault diagnosis of an electromechanical drive system [13] These studies established deep learning for mechanical equipment anomaly detection using the time and frequency domains. We propose fault diagnosis model that operates on deep learning and can detect time-series based anomalies in machine vibration data, thereby overcoming much limitation posed by existing intelligent-data-based defects. 6) The proposed SCRLSTM model using various load, noise, and time series-based mechanical equipment datasets demonstrates superior generalization ability as it shows better performance in average accuracy and confusion plot compared to the other five network models This generalization capability effectively handles the task of diagnosing machine faults.

BACKGROUND

EXPERIMENTAL ANALYSIS

Findings

CONCLUSION