Rolling Bearing Fault Diagnosis Method Based on Parallel QPSO-BPNN Under Spark-GPU Platform

Lanjun Wan,Hongyang Li,Mansheng Xiao,Gen Zhang,Changyun Li,Junfeng Man

doi:10.1109/access.2021.3072596

Lanjun Wan, Hongyang Li + Show 4 more

Open Access

https://doi.org/10.1109/access.2021.3072596

Copy DOI

Abstract

Facing the massive rolling bearing vibration data, how to improve the training efficiency, diagnosis efficiency, and diagnosis accuracy of the rolling bearing fault diagnosis model is a challenge. Considering that the Spark-GPU platform provides powerful distributed parallel computing capabilities and back propagation neural network (BPNN) optimized by quantum particle swarm optimization (QPSO) algorithm has the characteristics of low computational complexity and high diagnosis accuracy, a rolling bearing fault diagnosis method based on parallel QPSO-BPNN under Spark-GPU platform is proposed. First, the distributed parallelization of QPSO-BPNN model based on Spark-GPU platform is realized, which can improve the training efficiency and diagnosis efficiency of rolling bearing fault diagnosis model in the big data environment. Second, in order to improve the convergence speed of fault diagnosis model, a parameter update strategy suitable for the distributed parallel training of QPSO-BPNN model is designed. At each iteration during training, the local parameters of each worker node are collected to the master node, and the global parameters are updated according to the weights and synchronized to each worker node. Third, a combination strategy of multiple QPSO-BPNN models based on ensemble learning is proposed. The weighted voting method is adopted to combine the output results of different QPSO-BPNN models to obtain the best fault diagnosis result of a sample, which can improve the fault diagnosis accuracy to a certain extent. Experimental results show that the proposed method can quickly perform model training and fault diagnosis for large-scale rolling bearing vibration data, and the fault diagnosis accuracy reaches 98.73%.

Highlights

Rolling bearing is one of the key components of mechanical equipment, and fault diagnosis of rolling bearing is essential to ensure long-term efficient and stable operation of mechanical equipment [1]
A rolling bearing fault diagnosis method based on parallel quantum particle swarm optimization (QPSO)-back propagation neural network (BPNN) under Spark-GPU platform is proposed, which aims to fully exploit the powerful distributed parallel computing capabilities provided by the Spark-GPU platform and take advantage of QPSO-BPNN with low computational complexity and high diagnosis accuracy to achieve more efficient and accurate fault diagnosis of rolling bearing in the big data environment
To perform fast and accurate rolling bearing fault diagnosis in the big data environment, a rolling bearing fault diagnosis method based on parallel QPSO-BPNN under Spark-GPU platform is proposed

Summary

Introduction

Rolling bearing is one of the key components of mechanical equipment, and fault diagnosis of rolling bearing is essential to ensure long-term efficient and stable operation of mechanical equipment [1]. The traditional rolling bearing fault diagnosis methods based on signal processing technology have been widely used, such as enhanced singular spectrum decomposition [2], frequency phase space empirical wavelet transform [3], adaptive generalized demodulation [4], high-order synchrosqueezing transform [5], recycling variational mode decomposition [6], and resonance-based. The above methods can effectively diagnose rolling bearing faults when the time-frequency domain features of vibration signals are obvious. With the rapid development of machine learning and deep learning, there are more and more data-driven rolling bearing fault diagnosis methods based on machine learning and deep learning, such as naive bayes algorithm [8], least square support vector machine [9], iterative random forest [10], BP neural network [11], one-dimensional convolutional neural network [12], two-dimensional convolutional neural network [13], LSTM recurrent neural.

Methods

Results

Conclusion