Intelligent Anomaly Detection Method of Pump Set Based on Convolve-gated Self-attention Multi-source Data Fusion
-
摘要: 针对核动力泵组在运行过程中多变工况下难以利用多源检测信号进行诊断的问题，本文提出一种利用深度学习网络融合多源数据的泵组智能异常检测方法。该方法利用卷积神经网络（CNN）对多源数据进行融合，能够有效地对多源数据之间的关系进行分析。采用自注意力机制提取具有注意力权值的输入数据融合特征，使所构建的智能异常检测模型具有自主适应不同类型输入数据的能力，保证了所提方法在多源数据场景下的核动力泵组智能异常状态检测的准确度，同时加入残差块提升模型训练效果。通过搭建泵组故障模拟试验台来验证该方法的可靠性和准确性，结果表明，本文所提检测方法能够有效融合多源数据之间的信息特征，在此基础上能够充分完成泵组在运行过程中多变工况下故障诊断的任务，且具有较高的诊断精度。Abstract: To address the challenge of diagnosing nuclear power pump set under varying operating conditions using multi-source detection signals, this paper proposes an intelligent anomaly detection method based on deep learning for pump set by fusing multi-source data. The method employs Convolutional Neural Networks (CNN) to fuse multi-source data, effectively analyzing the relationships among diverse data sources. A self-attention mechanism is adopted to extract fusion features of input data with attention weights, enabling the constructed intelligent anomaly detection model to autonomously adapt to different types of input data. This ensures the high accuracy of the proposed method in detecting abnormal states of nuclear power pump set under multi-source data scenarios. Additionally, residual blocks are incorporated to enhance model training performance. The reliability and accuracy of the method are validated through a pump set fault simulation test bench. The results demonstrate that the proposed detection method effectively integrates the informational features of multi-source data, enabling the reliable diagnosis of faults in pump set under variable operating conditions with high diagnostic precision.
-
Key words:
- Pump set /
- Fault diagnosis /
- Feature fusion /
- Deep learning /
- Multi-source data
-
图 1 核动力泵组智能异常检测方法流程图
Figure 1. Flow Chart of Intelligent Anomaly Detection Method for Nuclear Pump Set
图 3 泵组故障模拟实验台上测点布置
Figure 3. Measuring Point Layout on Pump Set Fault Simulation Test Bench
表 1 泵组故障模拟实验工况设置
Table 1. Setting of Test Conditions for Pump Set Fault Simulation
工况编号 转速/（r · min−1） 水位/cm 1 1800 30 2 1800 16c 3 1800 23 4 2400 30 5 2400 16 6 2400 23 7 3000 30 8 3000 16 9 3000 23 
下载: 导出CSV
表 2 监测量与板卡、通道对应表
Table 2. Monitoring Parameters and Corresponding Boards and Channels
监测量名称 测点编号 传感器灵敏度 叶轮端X向振动 A-1-FX 22 mV/g 叶轮端Y向振动 A-2-FY 22 mV/g 叶轮端Z向振动 A-3-FZ 22 mV/g 电机U相电流 C-9-U 1 V/10 A 电机V相电流 C-10-V 1 V/10 A 电机W相电流 C-11-W 1 V/10 A 电机端X向振动 B-5-MX 22 mV/g 电机端Y向振动 B-6-MY 22 mV/g 电机端Z向振动 B-7-MZ 22 mV/g 水泵出水侧压力 A-4-出口压力 量程0~1 MPa 水泵进水侧流量 B-8-出口流量 量程2~40 m3/h
下载: 导出CSV
表 3 异常检测模型超参数表
Table 3. Hyperparameters of Anomaly Detection Model
网络组成 内核 步长 参数量 卷积层
（1Conv1d_1）18/10 2 19036 最大池化层1
（Maxpool_1）2 2 残差块
（Res Block）Conv1d_2+Conv1d_3 卷积层2
（Conv1d_2）6 15816 最大池化层2
（Maxpool_2）2 2 卷积层3
（Conv1d_3）6 2/1 14976 最大池化层3
（Maxpool_3）2 2 自注意力机制
（Attention）1 1 双向门控循环
（BiGRU）
下载: 导出CSV
表 4 故障类型统计表
Table 4. Statistics of Fault Type
故障位置 故障名称 水泵电机端轴承 外圈故障 水泵叶轮端轴承 外圈故障 内圈故障 滚动体故障 保持架故障 复合故障 电机轴承外圈+叶轮轴承内圈故障
下载: 导出CSV
表 5 水泵故障数据集
Table 5. Pump Failure Dataset
故障类型 训练样本数 测试样本数 标签 正常状态 1000 1000 0 水泵电机端轴承故障 1000 1000 1 水泵叶轮端轴承故障 1000 1000 2 复合故障 1000 1000 3
下载: 导出CSV
表 6 三种模型的故障诊断准确率对比 %
Table 6. Comparison of Fault Diagnosis Accuracy of Three Methods
工况编号 信号 RNN模型 GRU模型 本文模型 1 FZ 100 100 100 MZ 100 100 100 2 FZ 90 97.50 100 MZ 100 100 100 3 FZ 100 100 100 MZ 100 100 100 4 FZ 100 100 100 MZ 100 100 100 5 FZ 95 100 100 MZ 100 100 100 6 FZ 90 100 100 MZ 100 100 100 7 FZ 100 100 100 MZ 80 90 100 8 FZ 100 100 100 MZ 100 100 100 9 FZ 100 100 100 MZ 94.75 97 100 平均结果 97.20 99.13 100
下载: 导出CSV
-
[1] 潘军,黎义斌,瞿泽晖,等. 华龙一号主泵卡轴事故工况瞬态过渡过程数值分析[J]. 核动力工程,2024, 45(1): 201-209. [2] 张秀春,夏虹,刘永康,等. 基于移动窗核主成分分析的核电厂主泵故障检测[J]. 核动力工程,2024, 45(3): 234-240. [3] HANG J, ZHANG J Z, CHENG M, et al. An overview of condition monitoring and fault diagnostic for wind energy conversion system[J]. Transactions of China Electrotechnical Society, 2013, 28(4): 261-271. [4] 朱少民,夏虹,吕新知,等. 基于ARIMA和LSTM组合模型的核电厂主泵状态预测[J]. 核动力工程,2022, 43(2): 246-253. [5] LI D D, LIU Y H, ZHAO Y, et al. Fault diagnosis method of wind turbine planetary gearbox based on improved generative adversarial network[J]. Proceedings of the CSEE, 2021, 41(21): 7496-7506. [6] DEY S, STORI J A. A Bayesian network approach to root cause diagnosis of process variations[J]. International Journal of Machine Tools and Manufacture, 2005, 45(1): 75-91. doi: 10.1016/j.ijmachtools.2004.06.018 [7] 刘子铭,罗能,艾琼. 基于频域数据注意力机制的核电厂水泵故障模式识别模型研究[J]. 核动力工程,2021, 42(6): 203-208. [8] HOU J J, LU X K, ZHONG Y D, et al. A comprehensive review of mechanical fault diagnosis methods based on convolutional neural network[J]. Journal of Vibroengineering, 2024, 26(1): 44-65. doi: 10.21595/jve.2023.23391 [9] TANG S N, ZHU Y, YUAN S Q. Intelligent fault identification of hydraulic pump using deep adaptive normalized CNN and synchrosqueezed wavelet transform[J]. Reliability Engineering & System Safety, 2022, 224: 108560. [10] XIA M, LI T, XU L, et al. Fault diagnosis for rotating machinery using multiple sensors and convolutional neural networks[J]. IEEE/ASME Transactions on Mechatronics, 2018, 23(1): 101-110. doi: 10.1109/TMECH.2017.2728371 [11] XING Z K, LIU Y B, WANG Q, et al. Multi-sensor signals with parallel attention convolutional neural network for bearing fault diagnosis[J]. AIP Advances, 2022, 12(7): 075020. doi: 10.1063/5.0095530 [12] SRISAILA A, RAJANI D, MADHAVI M V D N S, et al. Modelling a dense hybrid network model for fake review analysis using learning approaches[J]. Soft Computing, 2024, 28(4): 3519-3532. doi: 10.1007/s00500-023-09609-4 -
计量
- 文章访问数: 43
- HTML全文浏览量: 15
- PDF下载量: 1
- 被引次数: 0
