在高压调门油动机的运行监控中，由于正常状态的样本远多于故障样本，故障数据稀缺且采集相对困难，此外还存在故障发生的不确定性，传统的监测方法难以有效应用。对此，本文提出了一种基于支持向量数据描述（Support Vector Data Description，SVDD）异常检测和支持向量机（Support Vector Machine，SVM）故障诊断的高压调门油动机状态监测系统。首先，从原始数据中提取时域（Time Domain，T）、频域（Frequency Domain，F）和时频域小波包子带能量（Wavelet Packet Subband Energy，W）特征，并通过特征融合及归一化的方式形成新的多维融合特征向量TFW。随后，采用卷积神经网络（Convolutional Neural Network，CNN）对TFW进行深层次挖掘，生成更具表现力的特征TFWCNN，以此作为SVDD和SVM模型的输入。搭建了高压调门油动机故障模拟试验台，用以采集数据并验证所提方法的有效性。研究结果表明，在三个具有不同阀位开度的高压调门油动机动态数据集上，SVDD异常检测的F1分数分别达到0.9991、0.9978和0.9760，SVM故障诊断的F1分数分别为0.9988、0.9950和0.9867，不仅说明该方法在高压调门油动机的状态监测中表现出的优异性能，同时也说明深度TFWCNN特征在高压调门油动机状态监测中的有效性和准确性，还为类似的汽轮机状态监测诊断系统提供了一种有效的技术方案。
Abstract
In the monitoring of high-pressure servo motor, traditional methods face significant challenges due to the scarcity of fault data, which is difficult to acquire, and the inherent uncertainty of fault occurrences, with normal samples vastly outnumbering fault samples. To address these issues, this paper proposes a state monitoring system for high-pressure servo motor based on Support Vector Data Description (SVDD) for anomaly detection and Support Vector Machine (SVM) for fault diagnosis. Initially, time-domain (T), frequency-domain (F), and wavelet packet subband energy (W) features are extracted from raw data. These features are then fused and normalized to form a new multidimensional feature vector, TFW. Subsequently, a Convolutional Neural Network (CNN) is employed to deeply mine the TFW, generating more expressive features, TFWCNN, which serve as inputs to the SVDD and SVM models. An experimental platform for simulating high-pressure servo motor faults was constructed to collect data and validate the proposed method. The results indicate that on three dynamic datasets with different valve opening positions, the F1 scores for SVDD anomaly detection are 0.9991, 0.9978, and 0.9760, and for SVM fault diagnosis, the F1 scores are 0.9988, 0.9950, and 0.9867, respectively. These findings not only demonstrate the superior performance of the proposed method in the state monitoring of high-pressure servo motor but also highlight the efficacy and accuracy of deep TFWCNN features. Furthermore, this study provides an effective technical solution for similar turbine state monitoring and diagnostic systems.
关键词
高压调门油动机 /
SVDD异常检测 /
SVM故障诊断 /
状态监测系统
{{custom_keyword}} /
Key words
high-pressure servo motor /
SVDD anomaly detection /
SVM fault diagnosis /
state monitoring system
{{custom_keyword}} /
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] 周伟杰, 胡义宏, 唐军. 基于支持向量数据描述的油动机故障预警研究 [J]. 热力透平, 2022, 51(4): 290-294.
ZHOU Weijie, HU Yihong, TANG Jun. Research on Oil Engine Fault Warning Based on Support Vector Data Description [J]. Thermal Turbine, 2022, 51(4): 290-294.
[2] CHEN K, TANG R, LI C, et al. Fractional order P I λ controller synthesis for steam turbine speed governing systems [J]. ISA Transactions, 2018, 77: 49-57.
[3] HUANG B, PENG Y H, HU L S, et al. Incipient fault detection approach based on piecewise linear shape-based global embedding for steam turbine plants [J]. Applied Energy, 2024, 370: 123563.
[4] BUCHAIAH S, SHAKYA P. Bearing fault diagnosis and prognosis using data fusion based feature extraction and feature selection [J]. Measurement, 2022, 188: 110506.
[5] HAN L, LI C W, GUO S L, et al. Feature extraction method of bearing AE signal based on improved FAST-ICA and wavelet packet energy [J]. Mechanical Systems and Signal Processing, 2015, 62-63: 91-99.
[6] BIN G F, GAO J J, LI X J, et al. Early fault diagnosis of rotating machinery based on wavelet packets—Empirical mode decomposition feature extraction and neural network [J]. Mechanical Systems and Signal Processing, 2012, 27: 696-711.
[7] 熊青松, 熊海贝, 袁程, 等. 基于多通道一维卷积神经网络的建筑结构损伤识别 [J]. 振动与冲击, 2024, 43(24): 216-224.
XIONG Qingsong, XIONG Haibei, YUAN Cheng, et al. Damage identification of building structures based on multi-channel one-dimensional convolutional neural networks [J]. Journal of Vibration and Shock, 2024, 43(24): 216-224.
[8] 魏晓良, 潮群, 陶建峰, 等. 基于LSTM和CNN的高速柱塞泵故障诊断 [J]. 航空学报, 2021, 42(3): 435-445.
WEI Xiaoliang, CHAO Qun, TAO Jianfeng, et al. Fault Diagnosis of High-Speed Plunger Pump Based on LSTM and CNN [J]. Acta Aeronautica, 2021, 42(3): 435-445.
[9] JIANG W, MA L, ZHANG P, et al. Anomaly Detection of Axial Piston Pump Based on the DTW-RCK-IF Composite Method Using Pressure Signals [J]. Applied Sciences, 2023, 13(24): 13133.
[10] 林桐, 陈果, 滕春禹, 等. 基于超球优化支持向量数据描述的滚动轴承故障检测 [J]. 振动与冲击, 2019, 38(02): 204-210+225.
LIN Tong, CHEN Guo, TENG Chunyu, et al. Fault detection of rolling bearings based on hypersphere optimization support vector data description [J]. Journal of Vibration and Shock, 2019, 38(02): 204-210+225.
[11] LIN K, PAN J, XI Y, et al. Vibration anomaly detection of wind turbine based on temporal convolutional network and support vector data description [J]. Engineering Structures, 2024, 306: 117848.
[12] LI D, GAO H, YANG K, et al. Abnormal identification of oil monitoring based on LSTM and SVDD [J]. Wear, 2023, 526-527: 204793.
[13] 聂浩淼, 车驰东. 基于振动信号PSD-SVM方法的不定负荷下柴油机气阀间隙异常故障诊断 [J]. 振动与冲击, 2024, 43(02): 299-305.
NIE Haomiao, CHE Chidong. Fault diagnosis of diesel valve clearance under variable load based on vibration signal PSD-SVM method [J]. Journal of Vibration and Shock, 2024, 43(02): 299-305.
[14] WANG B, QIU W, HU X, et al. A rolling bearing fault diagnosis technique based on recurrence quantification analysis and Bayesian optimization SVM [J]. Applied Soft Computing, 2024, 156: 111506.
[15] SALAHSHOOR K, KORDESTANI M, KHOSHRO M S. Fault detection and diagnosis of an industrial steam turbine using fusion of SVM (support vector machine) and ANFIS (adaptive neuro-fuzzy inference system) classifiers [J]. Energy, 2010, 35(12): 5472-5482.
[16] ZHU Z, ZHANG L, WU H, et al. Wavelet packet energy proportion-based early warning for the failure of lithium-ion batteries [J]. IEEE Transactions on Transportation Electrification, 2024: 1-1.
[17] HE W, MAO J, LI Z, et al. Fault Identification of Rotating Machinery Based on Dynamic Feature Reconstruction Signal Graph [J]. IEEE/ASME Transactions on Mechatronics, 2024, 29(3): 2056-2066.
[18] 姜万录, 李振宝, 雷亚飞, 等. 基于深度学习的滚动轴承故障诊断与性能退化程度识别方法 [J]. 燕山大学学报, 2020, 44(6): 526-536.
JIANG Wanlu, LI Zhenbao, LEI Yafei, et al. Fault Diagnosis and Performance Degradation Degree Recognition Method of Rolling Bearing Based on Deep Learning [J]. Journal of Yanshan University, 2020, 44(6): 526-536.
[19] TAX D M J, DUIN R P W. Support Vector Data Description [J]. Machine Learning, 2004, 54(1): 45-66.
[20] CORTES C, VAPNIK V. Support-vector networks [J]. Machine Learning, 1995, 20: 273–297.
{{custom_fnGroup.title_cn}}
脚注
{{custom_fn.content}}
PDF(3288 KB)
