Fault Diagnosis of Harmonic Drive With Imbalanced Data Using Generative Adversarial Network
37
Citation
30
Reference
10
Related Paper
Citation Trend
Abstract:
Harmonic drive is the core component of the industrial robot, and its fault diagnosis is crucial to the reliability and performance of the equipment. Most machine learning methods achieve good results based on the assumption of data balance. However, the scarce fault data of harmonic drive is difficult to collect, resulting in various imbalanced health status samples, which has an adverse effect on fault diagnosis. In this article, we propose a data generation method based on generative adversarial networks (GANs) to solve the problem of data imbalance and utilize the multiscale convolutional neural network (MSCNN) to realize the fault diagnosis of the harmonic drive. First, the data collected from three vibration acceleration sensors are preprocessed by fast Fourier transform (FFT) to obtain the frequency spectrum of the vibration signal. Second, multiple GANs were adopted to generate various fault spectrum data and the data selection module (DSM) is elaborately designed to filter and purify these data. Third, the filtered generated data will be combined with the real data to form a balanced dataset, and then the MSCNN is used to achieve multiclassification of the health status of the harmonic drive. Finally, the experiments have been implemented on an industrial robot vibration test bench to validate the effectiveness of our approach. The results have shown the fault multiclassification accuracy as 98.49% under imbalanced fault data conditions, which outperforms that of the other compared methods.Keywords:
Harmonic
The efficient Fourier transform (EFT) and FFT algorithms are described and their computational efficiencies with respect to the direct method are discussed. An efficient procedure is proposed for the reordering of data set; the use of EFT algorithm for the initial Fourier transforms and restricting the size of final subsets to not less than 4 is also suggested for saving computation time in the FFT. It is found that on average the FFT with the proposed modifications is more than twice as fast as the original FFT. The amount of overhead operations involved in computer routine, based on the modified FFT is estimated.
Split-radix FFT algorithm
Prime-factor FFT algorithm
Rader's FFT algorithm
Twiddle factor
Cooley–Tukey FFT algorithm
Cite
Citations (0)
Split-radix FFT algorithm
Prime-factor FFT algorithm
Rader's FFT algorithm
Twiddle factor
Cite
Citations (0)
The fast Fourier Transform (FFT), a reduced-complexity formulation of the Discrete Fourier Transform (DFT), is an important tool in many areas of science and engineering. FFTW is a well-known package that follows this approach and is currently one of the fastest available implementations of the FFT. NVIDIA introduced its version of FFTW called cuFFT that achieves high performance on the GPUs. In this work we present a novel way to map the FFT algorithm on the newly introduced Tensor Cores by adapting the the Cooley-Tukey recursive FFT algorithm. We present four major types of optimizations that enhance the performance of our approach for varying FFT sizes and show that the approach consistently outperforms cuFFT with a speedup of about 15% to 250% on average.
Split-radix FFT algorithm
Speedup
Prime-factor FFT algorithm
Rader's FFT algorithm
Twiddle factor
Cooley–Tukey FFT algorithm
Implementation
Cite
Citations (9)
Vibration spectrum analysis is now performed using FFT computing devices almost exclusively. The digital nature of the FFT instrument has led to the incorporation of additional computational capability to enhance processing and interpretation. Much greater capabilities are becoming available through the marriage of computers and FFT analyzers, the most advanced of which are for modal analysis. The power and potential of a general or special purpose computer in combination with an FFT instrument-type analyzer is discussed with respect to typical vibration and acoustic problems that can be solved.
Cite
Citations (0)
In this letter, we propose new trainable Fast Fourier Transform (FFT) structures to increase the accuracy of beamspace transformation in the multi-user (MU) mode of Massive Multiple Input Multiple Output (MIMO) receiver. The FFT is a widely employed beamspace transformation technique, thanks to its low transforming complexity. Unfortunately, there is a significant performance loss of MU signal detection in the FFT beamspace, as default FFT beams are not co-directed with angles of arrivals. We address this issue with trainable FFT that outperforms not only fixed FFT approaches, but also other trainable FFT techniques.
Split-radix FFT algorithm
Prime-factor FFT algorithm
Cite
Citations (0)
Real time object detection in traffic surveillance is one of the latest topics in today's world using Region based Convolutional Neural Networks algorithm in comparison with Convolutional Neural Networks. Real-Time Object Detection is performed using Regional Convolutional Neural Networks (N=78) over Convolutional Neural Networks (N=78) with the split size of training and testing dataset 70% and 30% respectively. Regional Convolutional Neural Networks had significantly better accuracy (75.6%) compared to Convolutional Neural Networks (47.7%) and attained significance value of p=0.041. Regional Convolutional Neural Networks achieved significantly better object detection than Convolutional Neural Networks for identifying real-time objects in traffic surveillance.
Cite
Citations (8)
Split-radix FFT algorithm
Prime-factor FFT algorithm
Cite
Citations (9)
Split-radix FFT algorithm
Prime-factor FFT algorithm
Rader's FFT algorithm
Twiddle factor
Cite
Citations (1)
The efficient Fourier transform (EFT) and FFT algorithms are described and their computational efficiencies with respect to the direct method are discussed. An efficient procedure is proposed for the reordering of data set; the use of EFT algorithm for the initial Fourier transforms and restricting the size of final subsets to not less than 4 is also suggested for saving computation time in the FFT. It is found that on average the FFT with the proposed modifications is more than twice as fast as the original FFT. The amount of overhead operations involved in computer routine, based on the modified FFT is estimated.
Split-radix FFT algorithm
Prime-factor FFT algorithm
Rader's FFT algorithm
Twiddle factor
Cooley–Tukey FFT algorithm
Cite
Citations (0)
This work proposes an original multidimensional fast Fourier transform (FFT) algorithm where the computation is first organized into multiplier-free butterflies and then completed by 1-D FFTs. The properties of well-known 1-D FFT algorithms blend in quite nicely with those of the proposed multidimensional FFT scheme, extending their computational and structural characteristics to it. Strong points of the proposed method are that its total computational cost decreases as the signal space dimensions increase and that its efficiency is superior to that of any other multidimensional FFT algorithm.
Split-radix FFT algorithm
Prime-factor FFT algorithm
Rader's FFT algorithm
Twiddle factor
Cite
Citations (4)