融合知识图谱与多元神经网络的发动机故障预测方法

阴艳超; 王宵; 徐成现

doi:10.13433/j.cnki.1003-8728.20220184

融合知识图谱与多元神经网络的发动机故障预测方法

doi: 10.13433/j.cnki.1003-8728.20220184

昆明理工大学机电工程学院，昆明　650500

基金项目: 国家自然科学基金项目（52065033）

详细信息

作者简介:
阴艳超（1977−），教授，博士生导师，博士，研究方向为智能制造、知识服务，yinyc@163.com

中图分类号: TP391
计量
- 文章访问数: 152
- HTML全文浏览量: 72
- PDF下载量: 21
- 被引次数: 0
出版历程
- 收稿日期: 2021-12-07
- 刊出日期: 2023-12-25

Engine Fault Prediction Method Integrating Knowledge Graph and Multivariate Neural Network

Faculty of Mechanicaland Electrical Engineering, Kunming University of Scienceand Technology, Kunming 650500, China

摘要

摘要: 针对汽车发动机故障率高且种类多，故障征兆与故障之间存在多对多的复杂耦联关系，故障溯源难度大、准确率低等问题，提出了融合知识图谱和多元神经网络的发动机故障智能预测方法。将发动机运行状态、故障现象、故障原因和维修记录作为输入信息，通过知识抽取、消歧和加工形成为可表示、可推理的结构化知识网络，并进行特征向量转换；建立了包含故障记录嵌入层、卷积层、GRU门控层和注意力机制的多元神经网络通路，通过特征向量训练形成了发动机故障预测模型，实现了发动机定性故障现象到定量故障推理，再到定性故障预测输出的映射变换；通过实际维修案例验证了所提KG-CNN-GRU-Att方法的可行性和有效性。
- 知识图谱 /
- 多元神经网络 /
- 发动机故障 /
- 向量转换 /
- 智能预测
Abstract: Aiming at the problems of high incidence rate and multiple types of automobile engine faults, complex coupling relationship between fault symptom and faults, and difficulty in fault traceability and low accuracy, an intelligent prediction method of engine fault based on knowledge graph and multiple neural network is proposed. Firstly, the engine running state, fault phenomena, fault causes and maintenance records are taken as input information, and the structured knowledge network which can be represented and inferred is formed through knowledge extraction, disambiguation and processing. The eigenvectorsare carried out for the knowledge network. Secondly, a multi-neural network pathway including fault record embedding layer, convolutional layer, GRU (Gate Recurrent Unit) gating layer and attention mechanism is established, and the engine fault prediction model is formed through eigenvector training, which will realize the mapping transformation from qualitative fault phenomena to quantitative fault reasoning, and then to qualitative fault prediction output.Finally, the feasibility and effectiveness of the proposed KG-CNN-GRU-Att (Knowledge Graph-Convolutional Neural Network- Gate Recurrent Unit-Attention) method are verified by a practical maintenance case.
- knowledge graph /
- multivariate neural network /
- engine fault /
- vector transformation /
- intelligent prediction

HTML全文

图 1 发动机故障智能预测框架

Figure 1. Intelligent engine fault prediction framework

下载: 全尺寸图片幻灯片

图 2 发动机故障知识图谱构建流程

Figure 2. Engine fault knowledge graph construction procedures

下载: 全尺寸图片幻灯片

图 3 汽车发动机故障维修知识图谱构建部分实例

Figure 3. Some examples of automotive engine fault maintenance knowledge graph construction

下载: 全尺寸图片幻灯片

图 4 部分字向量结果图

Figure 4. Partial results of word vector

下载: 全尺寸图片幻灯片

图 5 多元混合神经网络模型

Figure 5. Multivariate fusion neural network model

下载: 全尺寸图片幻灯片

图 6 GRU神经网络结构

Figure 6. GRU neural network structure

下载: 全尺寸图片幻灯片

图 7 注意力机制运行机理图

Figure 7. Attention mechanism operating mechanisms

下载: 全尺寸图片幻灯片

图 8 不同模型融合知识图谱预测效果对比图

Figure 8. Comparison of prediction effect of different models fusion knowledge graph

下载: 全尺寸图片幻灯片

图 9 未融合知识图谱模型精确率对比图

Figure 9. Comparison of precision rates of unfused knowledge graph models

下载: 全尺寸图片幻灯片

表 1 实验参数设置

Table 1. Experimental parameter setting

超参数	参数名	参数值
h	隐藏层节点数	198
droupout	丢弃率	0.2
lr	学习率	2 × 10⁻⁴
epoch	训练次数	100
loss	损失函数	Sigmod

下载: 导出CSV

表 2 不同模型融合知识图谱的故障预测效果对比

Table 2. Comparison of fault prediction effect of different models in fusion knowledge graph

模型	P/%	R/%	F₁/%
KG-CNN	75.21	68.02	71.43
KG-LSTM	76.84	70.31	73.43
KG-Bi-lstm	77.04	71.23	74.02
KG-CNN-LSTM	79.37	72.65	75.86
KG-CNN-GRU-Att	80.64	74.59	77.50

下载: 导出CSV

表 3 未融合知识图谱的不同模型的故障预测效果对比

Table 3. Comparison of fault prediction effects of different models without fusion knowledge graph

模型	P/%	R/%	F₁/%
CNN	74.37	67.49	70.76
LSTM	74.61	69.82	72.14
Bi-lstm	75.17	70.35	72.68
LSTM	77.24	71.26	74.13
CNN-GRU-Att	79.51	72.03	75.59

下载: 导出CSV

参考文献(18)

[1]	王金鑫, 王忠巍, 马修真, 等. 柴油机润滑系统多故障耦合关系分析及解耦技术[J]. 哈尔滨工程大学学报, 2021, 42(6): 825-832. WANG J X, WANG Z W, MA X Z, et al. Multi-fault coupling analysis and decoupling approach for diesel engine lubrication system[J]. Journal of Harbin Engineering University, 2021, 42(6): 825-832. (in Chinese)
[2]	陈鲲, 茆志伟, 张进杰, 等. 基于和声搜索优化栈式自编码器的柴油发动机故障诊断[J]. 机械工程学报, 2020, 56(11): 132-140. doi: 10.3901/JME.2020.11.132 CHEN K, MAO Z W, ZHANG J J, et al. Diesel engine fault diagnosis based on stack autoencoder optimized by harmony search[J]. Journal of Mechanical Engineering, 2020, 56(11): 132-140. (in Chinese) doi: 10.3901/JME.2020.11.132
[3]	柳长源, 车路平, 毕晓君. 基于TWSVM算法的发动机故障识别方法[J]. 内燃机学报, 2019, 37(1): 84-89. LIU C Y, CHE L P, BI X J. Engine fault identification based on TWSVM algorithm[J]. Transactions of CSICE, 2019, 37(1): 84-89. (in Chinese)
[4]	杜党党, 贾晓亮, 王卓. 基于故障彩色图谱的航空发动机故障模式识别[J]. 计算机集成制造系统, 2018, 24(9): 2297-2305. DU D D, JIA X L, WANG Z. Fault model identification method for aero engines based on fault color picture[J]. Computer Integrated Manufacturing Systems, 2018, 24(9): 2297-2305. (in Chinese)
[5]	刘梓权, 王慧芳. 基于知识图谱技术的电力设备缺陷记录检索方法[J]. 电力系统自动化, 2018, 42(14): 158-164. LIU Z Q, WANG H F. Retrieval method for defect records of power equipment based on knowledge graph technology[J]. Automation of Electric Power Systems, 2018, 42(14): 158-164. (in Chinese)
[6]	徐增林, 盛泳潘, 贺丽荣, 等. 知识图谱技术综述[J]. 电子科技大学学报, 2016, 45(4): 589-606. XU Z L, SHENG Y P, HE L R, et al. Review on knowledge graph techniques[J]. Journal of University of Electronic Science and Technology of China, 2016, 45(4): 589-606. (in Chinese)
[7]	刘峤, 李杨, 段宏, 等. 知识图谱构建技术综述[J]. 计算机研究与发展, 2016, 53(3): 582-600. LIU Q, LI Y, DUAN H, et al. Knowledge graph construction techniques[J]. Journal of Computer Research and Development, 2016, 53(3): 582-600. (in Chinese)
[8]	陈果, 杨默晗, 于平超. 基于深度学习的航空发动机不平衡故障部位识别[J]. 航空动力学报, 2020, 35(12): 2602-2615. CHEN G, YANG M H, YU P C. Aero-engine unbalanced fault location identification method based on deep learning[J]. Journal of Aerospace Power, 2020, 35(12): 2602-2615. (in Chinese)
[9]	姜洁, 李秋红, 张高钱, 等. 基于NN-ELM的航空发动机燃油系统执行机构故障诊断[J]. 航空动力学报, 2016, 31(2): 484-492. JIANG J, LI Q H, ZHANG G Q, et al. Fault diagnosis for actuators of aero-engine fuel system based on NN-ELM[J]. Journal of Aerospace Power, 2016, 31(2): 484-492. (in Chinese)
[10]	李本威, 林学森, 杨欣毅, 等. 深度置信网络在发动机气路部件性能衰退故障诊断中的应用研究[J]. 推进技术, 2016, 37(11): 2173-2180. LI B W, LIN X S, YANG X Y, et al. Research on application of deep belief networks on engine gas path component performance degradation defect diagnostics[J]. Journal of Propulsion Technology, 2016, 37(11): 2173-2180. (in Chinese)
[11]	张善文, 王振, 王祖良. 结合知识图谱与双向长短时记忆网络的小麦条锈病预测[J]. 农业工程学报, 2020, 36(12): 172-178. ZHANG S W, WANG Z, WANG Z L. Prediction of wheat stripe rust disease by combining knowledge graph and bidirectional long short term memory network[J]. Transactions of the Chinese Society of Agricultural Engineering, 2020, 36(12): 172-178. (in Chinese)
[12]	WANG H F, LIU Z Q. An error recognition method for power equipment defect records based on knowledge graph technology[J]. Frontiers of Information Technology & Electronic Engineering, 2019, 20(11): 1564-1578.
[13]	ZHANG C H, ZHOU M, HAN X, et al. Knowledge graph embedding for hyper-relational data[J]. Tsinghua Science and Technology, 2017, 22(2): 185-197. doi: 10.23919/TST.2017.7889640
[14]	WANG H T, HE J, ZHANG X H, et al. A short text classification method based on N-gram and CNN[J]. Chinese Journal of Electronics, 2020, 29(2): 248-254. doi: 10.1049/cje.2020.01.001
[15]	LI N, CAO J D. Synchronization criteria for multiple memristor-based neural networks with time delay and inertial term[J]. Science China Technological Sciences, 2018, 61(4): 612-622. doi: 10.1007/s11431-017-9189-3
[16]	DURRANI S P, BALLUFF S, WURZER L, et al. Photovoltaic yield prediction using an irradiance forecast model based on multiple neural networks[J]. Journal of Modern Power Systems and Clean Energy, 2018, 6(2): 255-267. doi: 10.1007/s40565-018-0393-5
[17]	WANG T C, WANG J Y, WU Y, et al. A fault diagnosis model based on weighted extension neural network for turbo-generator sets on small samples with noise[J]. Chinese Journal of Aeronautics, 2020, 33(10): 2757-2769. doi: 10.1016/j.cja.2020.06.024
[18]	HU X D, WANG X Q, MENG F J, et al. Gabor-CNN for object detection based on small samples[J]. Defence Technology, 2020, 16(6): 1116-1129. doi: 10.1016/j.dt.2019.12.002