Volume 3,Issue 7
基于深度学习的反应器过程故障诊断方法研究
在工业反应器运行过程中，高维过程信号的特征提取效果与噪声抑制水平，会直接影响故障检测与诊断的整体性能。为解决这一问题，本文设计一种新型多通道时序双向长短期记忆- 门控循环单元神经网络（MTBiLSTM-GRU），用于高维过程信号的深度特征学习。首先采用小波变换对高维过程信号进行多频带特征提取，分离不同频率分量信息；其次通过快速傅里叶变换实现时域信号到频域信号的转换，简化信号特征呈现形式；最后利用 MTBiLSTM-GRU 网络从多尺度过程信号中学习具有区分度的时频联合特征。在 Tennessee Eastman 数据集上开展多组故障分类方法对比实验，结果表明：小波变换降噪条件下模型分类准确率达到 90.71%，当选用 haar 小波基且降噪等级设置为 3 时，模型分类准确率可提升至 96.1%。此外，在行业内公认难以分类的第 3 类、第 9 类与第 15 类故障数据上，本文模型分别实现 99.50%、98.63%、97.38% 的高分类精度。
[1]Faradisa, I. S., et al. (2016). Identification of phonocardiogram signal based on STFT and Marquart Lavenberg Backpropagation. 2016 International Seminar on Intelligent Technology and Its Applications (ISITIA).
[2]Daubechies, I. (1990). "The wavelet transform, time-frequency localization and signal analysis." IEEE Transactions on Information Theory 36(5): 961-1005.
[3]Krizhevsky A, Sutskever I, Hinton GE (2017) ImageNet classifcation with deep convolutional neural networks. Commun ACM 60(6):84–90
[4]Shahnazari, H. (2020). "Fault diagnosis of nonlinear systems using recurrent neural networks." Chemical Engineering Research and Design 153: 233-245.
[5]Sun, W., et al. (2020). "Fault detection and identification using Bayesian recurrent neural networks." Computers & Chemical Engineering 141: 106991.
[6]Kang, J.-L. (2020). "Visualization analysis for fault diagnosis in chemical processes using recurrent neural networks." Journal of the Taiwan Institute of Chemical Engineers 112: 137-151.
[7]Han, Y., et al. (2020). "An optimized long short-term memory network based fault diagnosis model for chemical processes." Journal of Process Control 92: 161-168.
[8]Zhang, R. and Z. Xiong (2019). Recurrent Neural Network Model with Self-Attention Mechanism for Fault Detection and Diagnosis. 2019 Chinese Automation Congress (CAC).
[9]Mirzaei, S., et al. (2022). "A comparative study on long short-term memory and gated recurrent unit neural networks in fault diagnosis for chemical processes using visualization." Journal of the Taiwan Institute of Chemical Engineers 130.
[10]Yu, J., et al. (2020). "Multichannel one-dimensional convolutional neural network-based feature learning for fault diagnosis of industrial processes." Neural Computing and Applications 33(8): 3085-3104.
[11]Pöppelbaum, J., et al. (2022). "Contrastive learning based self-supervised time-series analysis." Applied Soft Computing 117.