基于贝叶斯优化BiLSTM模型的输电塔损伤识别

PDF(2909 KB)

振动与冲击 ›› 2023, Vol. 42 ›› Issue (1) : 238-248.

论文

魏佳恒，郭惠勇

作者信息 +

Damage identification of transmission tower based on BO-BiLSTM model

WEI Jiaheng, GUO Huiyong

Author information +

文章历史 +

摘要

结构的加速度响应可以反映结构的状态信息，蕴含结构的损伤特征。针对目前输电塔健康监测系统产生大量数据而无法有效分析和诊断输电塔损伤的问题，本文利用结构输出加速度响应数据的时序关系，提出了基于双向长短时记忆网络（Bi-directional Long and Short-Term Memory ,BiLSTM）的损伤识别方法，并采用概率寻优方法贝叶斯优化（Bayesian optimization,BO）确定网络模型超参数。首先描述了BiLSTM的基本原理，给出基于贝叶斯优化的超参数选取策略，从而提出了基于BO-BiLSTM模型的损伤识别方法。然后使用该方法对输电塔有限元模型进行了损伤定位与模式识别，测试集的整体识别准确率达到94.2%。为了验证该方法对实际结构的损伤识别效果，提出基于异源数据的损伤识别方式:将输电塔有限元模型数据作为模型训练的样本训练BO-BiLSTM模型，使用实验数据用作验证集检验损伤识别效果。识别结果表明BO-BiLSTM可以较为准确的识别真实结构的损伤情况，识别效果较BiLSTM以及BO-LSTM更稳定。

Abstract

The structure’s acceleration response sequence reflects the original structural state information and contains many damage characteristics. In response to the problem that the current transmission tower health monitoring system generates a large amount of data and cannot effectively analyze and diagnose the damage of the transmission tower, based on the time series relationship of the the structure’s output acceleration response data, this paper proposes a damage identification method based on a Bi-directional Long and Short-Term Memory (BiLSTM) network, and adopts a probability optimization method Bayesian optimization(BO) to determine the neural network hyperparameters. Firstly, this article describes the related theories of BiLSTM.Then it proposes a hyperparameter selection strategy based on Bayesian optimization and a damage identification method based on the BO-BiLSTM model. Then the method is applied to the finite element model of the transmission tower for damage localization and pattern identification, and the overall identification accuracy rate of test set reaches 94.2%. Finally, in order to verify method on the actual structure, a damage identification method based on heterogeneous data is proposed: the transmission tower finite element model data is used as the model training sample to train the BO-BiLSTM model, and the experimental data is used as the test set to test the damage recognition effect . The results show that the BO-BiLSTM can identify the real structure damage accurately, and the recognition effect is more stable than BiLSTM and BO-LSTM.

导出引用

魏佳恒，郭惠勇. 基于贝叶斯优化BiLSTM模型的输电塔损伤识别[J]. 振动与冲击, 2023, 42(1): 238-248

WEI Jiaheng, GUO Huiyong. Damage identification of transmission tower based on BO-BiLSTM model[J]. Journal of Vibration and Shock, 2023, 42(1): 238-248

参考文献

[1] 张武能,汪毅,张留斌,吴克华,王晓晨.输电铁塔螺栓连接结构健康监测技术进展与展望[J].中国电力,2021,54(02):18-26.
Zhang Wuneng, Wang yi, Zhang Liubin, Wu Kehua, Wang Xiaochen. Development and Prospect of Structure HealthMonitoring Technologies for Bolted Joints of Transmission Towers[J]. Electric Power, 2021, 54(02):18-26. (in Chinese)
[2] 李惠, 鲍跃全, 李顺龙, 等. 结构健康监测数据科学与工程[J]. 工程力学, 2015, 32(8): 1−7.
Li Hui, Bao Yuequan, Li Shunlong, et al. Data science and engineering for structural health monitoring [J].Engineering Mechanics, 2015, 32(8):1−7.(in Chinese)
[3] Yuequan Bao, Zhicheng Chen,Shiyin Wei, Yang Xu, Zhiyi Tang, Hui Li. The State of the Art of Data Science and Engineering in Structural Health Monitoring[J]. Engineering,2019,5(2): 234-242.
[4] Hinton G E, Salakhutdinov R R. Reducing the Dimensionality of Data with Neural Networks[J].Science, 2006, 313(5786): 504-507.
[5] 王慧, 郭晨林, 王乐, 张敏照. 基于内积矩阵及深度学习的结构健康监测研究[J]. 工程力学. doi: 10.6052/j.issn.1000-4750.2020.12.0935
Wang Hui, Guo Chenlin, Wang Le, Zhang Minzhao. Structural health monitoring based inner product matrix and deep learing[J]. Engineering Mechanics. doi: 10.6052/j.issn.1000-4750.2020.12.0935 (in Chinese)
[6] Talebi N, Sadrnia M A, Darabi A. Fault detection of wind energy conversion systems using recurrent neural networks[J].International Journal of Sustainable Energy, 2014, 34(1):52-70.
[7] 赵志宏,赵敬娇,魏子洋.基于BiLSTM的滚动轴承故障诊断研究[J].振动与冲击,2021,40(01):95-101.
Zhao Hongzhi, Zhao Jingjiao, Wei Ziyang. Rolling bearing fault diagnosis based on BiLSM network[J]. Journal of Vibration and Shock,2021,40(01):95-101. (in Chinese)
[8] 杜小磊,陈志刚,王衍学,张楠.基于IEWT和IFractalNet的滚动轴承故障诊断[J].振动与冲击,2020,39(24):134-142.
Du Xiaolei, Chen Zhigang, Wang Yanxue, Zhang Nan. Li Bin, Liu Xiangqing. Fault diagnosis of rolling bearings based on improved empirical wavelet transform and IfractalNet[J]. Journal of Vibration and Shock, 2020,39(24):134-142. (in Chinese)
[9] Cabrera D., Guamán A. and Zhang S. et al., Bayesian approach and time series dimensionality reduction to LSTM-based model-building for fault diagnosis of a reciprocating compressor[J]. Neurocomputing, 2020,380(7): 51-66.
[10] 江敏,陈一民.贝叶斯优化算法的发展综述[J].计算机工程与设计,2010,31(14):3254-3259.
Jiang Ming, Chen Yiming. Survey on Bayesian optimization algorithm [J].Computer Engineering and design,2010,31(14):3254-3259. (in Chinese)
[11] Zhong Jingjing, Peter W. Tse, Wang,Dong. Novel Bayesian inference on optimal parameters of support vector machines and its application to industrial survey data classification[J]. Neurocomputing,2016,211(26): 159-171..
[12] Wang Dong,Miao Qiang. Some Improvements on a General Particle Filter Based Bayesian Approach for Extracting Bearing Fault Features[J]. Journal of Vibration and Acoustics,2015,137(4): 041016-1-9.
[13] Gers F., Schmidhuber J., Cummins F. Learning to Forget: Continual Prediction with LSTM[J]. Neural Computation,2000,12(10): 51-71.
[14] 郑秋怡, 周广东, 刘定坤. 基于长短时记忆神经网络的大跨拱桥温度-位移相关模型建立方法[J]. 工程力学,2021,38(4):68-79.doi:10.6052/j.issn.1000-4750.2020.05.0323
Zheng Qiuyi, Zhou Guangdong, Liu Dingkun. Method of modeling temperature-displacement correlation for Long-span ARCH bridges based on long short-term memory neural networks[J]. Engineering Mechanics, 2021, 38(4): 68-79. doi: 10.6052/j.issn.1000-4750.2020.05.0323 (in Chinese)
[15] 牟含笑,郑建飞,胡昌华,赵瑞星,董青.基于CDBN与BiLSTM的多元退化设备剩余寿命预测[J]. 航空学报, doi: 10.7527/S1000-6893.2021.25403.
Mou Hanxiao, Zheng Jianfei, Hu Changhua, Zhao Ruixing, Dong Qing. Remaining useful life prediction of multivariate degradation equipment based on CDBN and BiLSTM[J/OL]. Acta Aeronautica ET Astronautica Sinica, doi: 10.7527/S1000-6893.2021.25403.
[16] 崔佳旭,杨博.贝叶斯优化方法和应用综述[J].软件学报,2018,29(10):3068-3090.
Cui Jiaxu, Yang Bo. Survey on Bayesian Optimization Methodology and Applications [J]. Journal of Software, 2018,29(10):3068-3090. (in Chinese)
[17] 高治鑫,包腾飞,李扬涛,王一兵.基于贝叶斯优化LightGBM的大坝变形预测模型[J/OL].长江科学院院报:1-6[2021-06-10].http://kns.cnki.net/kcms/detail/42.1171.TV.20210107.1228.024.html.
Gao Zhixin, Bao Pengfei, Li Yangtao, Wang Yibin. Dam Deformation Prediction Model Based on Bayesian Optimization and Light GBM[J/OL]. Journal of Yangtze River Scientific Research Institute: 1-6[2021-06-10].http://kns.cnki.net/kcms/detail/42.1171.TV.20210107.1228.024.html. (in Chinese)
[18] Rasmussen Carl Edward,Williams Christopher K. I.. Gaussian Processes for Machine Learning[M].The MIT Press:2005-11-23.