《计算机应用》唯一官方网站 ›› 2026, Vol. 46 ›› Issue (1): 305-313.DOI: 10.11772/j.issn.1001-9081.2025010070
李亚男1, 郭梦阳1, 邓国军2, 陈允峰2, 任建吉1, 原永亮3(
)
收稿日期:2025-02-07
修回日期:2025-03-29
接受日期:2025-04-02
发布日期:2026-01-10
出版日期:2026-01-10
通讯作者:
原永亮
作者简介:李亚男(1983—),男,河南温县人,副教授,博士, CCF会员,主要研究方向:大数据分析、隐私计算基金资助:
Yanan LI1, Mengyang GUO1, Guojun DENG2, Yunfeng CHEN2, Jianji REN1, Yongliang YUAN3(
)
Received:2025-02-07
Revised:2025-03-29
Accepted:2025-04-02
Online:2026-01-10
Published:2026-01-10
Contact:
Yongliang YUAN
About author:LI Yanan, born in 1983, Ph. D., associate professor. His research interests include big data analysis, privacy computing.Supported by:摘要:
针对发动机运行数据的多模态以及难以实现有效的发动机寿命预测问题,提出一种融合图像和发动机运行时间数据潜在关系的多模态融合特征并分支发动机寿命预测方法。首先,利用滑动窗口对发动机运行数据进行分割,以构造发动机运行数据的序列样本,并采用格拉姆角场(GAF)将构造的序列样本转化为图像;其次,用序列样本和图像分别通过双向长短期记忆(BiLSTM)网络和卷积神经网络(CNN)获取趋势和周期等传感器之间的潜在关系特征;最后,引入交叉注意力机制(CAM)实现2种模态特征的融合并实现发动机寿命的预测。在公开的C-MAPSS数据集上的实验结果表明,该预测方法的R-squared (R2)高于0.99,而均方根误差(RMSE)在1以内。可见,该方法能在保证预测精度的同时改善计算效率。
中图分类号:
李亚男, 郭梦阳, 邓国军, 陈允峰, 任建吉, 原永亮. 基于多模态融合特征的并分支发动机寿命预测方法[J]. 计算机应用, 2026, 46(1): 305-313.
Yanan LI, Mengyang GUO, Guojun DENG, Yunfeng CHEN, Jianji REN, Yongliang YUAN. Method for life prediction of parallel branching engine based on multi-modal fusion features[J]. Journal of Computer Applications, 2026, 46(1): 305-313.
| 数据集 | 发动机数 | 操作状态分类序号 | 故障类别分类序号 | |
|---|---|---|---|---|
| 训练集 | 测试集 | |||
| FD001 | 100 | 100 | 1 | 1 |
| FD002 | 260 | 259 | 6 | 1 |
| FD003 | 100 | 100 | 1 | 2 |
| FD004 | 249 | 248 | 6 | 2 |
表1 C-MAPSS 数据集
Tab. 1 C-MAPSS dataset
| 数据集 | 发动机数 | 操作状态分类序号 | 故障类别分类序号 | |
|---|---|---|---|---|
| 训练集 | 测试集 | |||
| FD001 | 100 | 100 | 1 | 1 |
| FD002 | 260 | 259 | 6 | 1 |
| FD003 | 100 | 100 | 1 | 2 |
| FD004 | 249 | 248 | 6 | 2 |
| 数据集 | 方法 | R2 | RMSE | Score |
|---|---|---|---|---|
| FD001 | MPCNN-BiLSTM | 0.998 7 | 0.106 5 | 0.996 5 |
| CNN | 0.987 6 | 6.346 6 | 0.995 9 | |
| BiLSTM | 0.974 2 | 9.180 5 | 0.995 7 | |
| FD002 | MPCNN-BiLSTM | 0.997 6 | 0.888 6 | 0.994 3 |
| CNN | 0.979 7 | 8.841 7 | 0.992 2 | |
| BiLSTM | 0.966 1 | 11.437 2 | 0.991 1 |
表2 数据集FD001和FD002上MPCNN-BiLSTM的消融实验结果
Tab. 2 Ablation experiment results of MPCNN-BiLSTM on datasets FD001 and FD002
| 数据集 | 方法 | R2 | RMSE | Score |
|---|---|---|---|---|
| FD001 | MPCNN-BiLSTM | 0.998 7 | 0.106 5 | 0.996 5 |
| CNN | 0.987 6 | 6.346 6 | 0.995 9 | |
| BiLSTM | 0.974 2 | 9.180 5 | 0.995 7 | |
| FD002 | MPCNN-BiLSTM | 0.997 6 | 0.888 6 | 0.994 3 |
| CNN | 0.979 7 | 8.841 7 | 0.992 2 | |
| BiLSTM | 0.966 1 | 11.437 2 | 0.991 1 |
| 数据集 | 方法 | R2 | RMSE | Score |
|---|---|---|---|---|
| FD001 | MPCNN-BiLSTM | 0.998 7 | 0.106 5 | 0.996 5 |
| CMGRU | 0.987 2 | 6.445 5 | 0.996 1 | |
| MSTPP | 0.989 8 | 5.757 8 | 0.996 3 | |
| sLSTM | 0.984 8 | 7.040 2 | 0.996 0 | |
| GRU | 0.976 6 | 8.737 4 | 0.995 9 | |
| XGBoost | 0.961 8 | 11.172 7 | 0.990 8 | |
| FD002 | MPCNN-BiLSTM | 0.997 6 | 0.888 6 | 0.994 3 |
| CMGRU | 0.964 2 | 11.742 6 | 0.991 3 | |
| MSTPP | 0.983 6 | 7.928 5 | 0.990 8 | |
| sLSTM | 0.981 8 | 8.357 7 | 0.990 7 | |
| GRU | 0.957 3 | 12.835 1 | 0.990 5 | |
| XGBoost | 0.953 1 | 13.445 8 | 0.989 3 |
表3 数据集FD001和FD002上不同方法的预测性能对比结果
Tab. 3 Comparison results of prediction performance of different methods on datasets FD001 and FD002
| 数据集 | 方法 | R2 | RMSE | Score |
|---|---|---|---|---|
| FD001 | MPCNN-BiLSTM | 0.998 7 | 0.106 5 | 0.996 5 |
| CMGRU | 0.987 2 | 6.445 5 | 0.996 1 | |
| MSTPP | 0.989 8 | 5.757 8 | 0.996 3 | |
| sLSTM | 0.984 8 | 7.040 2 | 0.996 0 | |
| GRU | 0.976 6 | 8.737 4 | 0.995 9 | |
| XGBoost | 0.961 8 | 11.172 7 | 0.990 8 | |
| FD002 | MPCNN-BiLSTM | 0.997 6 | 0.888 6 | 0.994 3 |
| CMGRU | 0.964 2 | 11.742 6 | 0.991 3 | |
| MSTPP | 0.983 6 | 7.928 5 | 0.990 8 | |
| sLSTM | 0.981 8 | 8.357 7 | 0.990 7 | |
| GRU | 0.957 3 | 12.835 1 | 0.990 5 | |
| XGBoost | 0.953 1 | 13.445 8 | 0.989 3 |
| 数据集 | 方法 | 训练时间/s | 测试时间/s |
|---|---|---|---|
| FD001 | MPCNN-BiLSTM | 47.532 2 | 0.232 5 |
| CMGRU | 49.173 2 | 0.284 2 | |
| MSTPP | 63.257 1 | 0.372 5 | |
| sLSTM | 45.125 7 | 0.193 5 | |
| GRU | 43.132 4 | 0.175 1 | |
| XGBoost | 15.398 2 | 0.112 4 | |
| FD002 | MPCNN-BiLSTM | 105.147 6 | 0.523 3 |
| CMGRU | 118.231 1 | 0.634 2 | |
| MSTPP | 126.165 3 | 0.775 9 | |
| sLSTM | 92.776 1 | 0.452 8 | |
| GRU | 88.335 4 | 0.425 5 | |
| XGBoost | 19.882 6 | 0.173 5 |
表4 在不同数据集上不同方法的训练和测试时间对比
Tab. 4 Comparison of training time and test time for different methods on different datasets
| 数据集 | 方法 | 训练时间/s | 测试时间/s |
|---|---|---|---|
| FD001 | MPCNN-BiLSTM | 47.532 2 | 0.232 5 |
| CMGRU | 49.173 2 | 0.284 2 | |
| MSTPP | 63.257 1 | 0.372 5 | |
| sLSTM | 45.125 7 | 0.193 5 | |
| GRU | 43.132 4 | 0.175 1 | |
| XGBoost | 15.398 2 | 0.112 4 | |
| FD002 | MPCNN-BiLSTM | 105.147 6 | 0.523 3 |
| CMGRU | 118.231 1 | 0.634 2 | |
| MSTPP | 126.165 3 | 0.775 9 | |
| sLSTM | 92.776 1 | 0.452 8 | |
| GRU | 88.335 4 | 0.425 5 | |
| XGBoost | 19.882 6 | 0.173 5 |
| 方法 | R2 | RMSE | Score |
|---|---|---|---|
| MPCNN-BiLSTM | 0.977 5 | 6.720 4 | 0.993 5 |
| CMGRU | 0.967 4 | 15.307 4 | 0.987 9 |
| MSTPP | 0.975 2 | 13.374 1 | 0.991 4 |
| sLSTM | 0.963 4 | 16.221 6 | 0.985 2 |
| GRU | 0.942 6 | 20.327 5 | 0.970 5 |
| XGBoost | 0.966 3 | 15.563 1 | 0.987 2 |
表5 不同方法在刀具数据集上的对比结果
Tab. 5 Comparison results of different methods on tool dataset
| 方法 | R2 | RMSE | Score |
|---|---|---|---|
| MPCNN-BiLSTM | 0.977 5 | 6.720 4 | 0.993 5 |
| CMGRU | 0.967 4 | 15.307 4 | 0.987 9 |
| MSTPP | 0.975 2 | 13.374 1 | 0.991 4 |
| sLSTM | 0.963 4 | 16.221 6 | 0.985 2 |
| GRU | 0.942 6 | 20.327 5 | 0.970 5 |
| XGBoost | 0.966 3 | 15.563 1 | 0.987 2 |
| [1] | 牟含笑,郑建飞,胡昌华,等.基于CDBN与BiLSTM的多元退化设备剩余寿命预测[J].航空学报, 2022, 43(7): 308-319. |
| MOU H X, ZHENG J F, HU C H, et al. Remaining useful life prediction of multivariate degradation equipment based on CDBN and BiLSTM [J]. Acta Aeronautica et Astronautica Sinica, 2022, 43(7): 308-319. | |
| [2] | YUAN Y, YANG Q, WANG G, et al. Combined improved tuna swarm optimization with graph convolutional neural network for remaining useful life of engine [J]. Quality and Reliability Engineering International, 2025, 41(1): 174-191. |
| [3] | HU J, QIAN X, CHENG H, et al. Remaining useful life prediction for aircraft engines based on phase space reconstruction and hybrid VNS-SVR model [J]. Journal of Intelligent and Fuzzy Systems: Applications in Engineering and Technology, 2021, 41(2): 3415-3428. |
| [4] | CAI L, YIN H, LIN J, et al. An update-strategy-based Gaussian process regression method for aeroengines fault prediction [J]. IEEE Transactions on Industrial Informatics, 2023, 20(2): 1941-1951. |
| [5] | WU Z, WANG Z, WEI H, et al. Remaining useful life prediction for equipment based on RF-BiLSTM [J]. AIP Advances, 2022, 12(11): No.115209. |
| [6] | MOHRIL R S, SOLANKI B S, KULKARNI M S, et al. XGBoost based residual life prediction in the presence of human error in maintenance [J]. Neural Computing and Applications, 2023, 35(4): 3025-3039. |
| [7] | YANG X, ZHENG Y, ZHANG Y, et al. Bearing remaining useful life prediction based on regression shapelet and graph neural network [J]. IEEE Transactions on Instrumentation and Measurement, 2022, 71: No.3505712. |
| [8] | 李路云,王海瑞,朱贵富.改进1D-CNN和LSTM的涡扇发动机剩余寿命预测[J].热能动力工程, 2023, 38(7): 194-202. |
| LI L Y, WANG H R, ZHU G F. Remaining Useful Life Prediction of Turbofan Engine based on Improved 1D-CNN and LSTM [J]. Journal of Engineering for Thermal Energy and Power, 2023, 38(7): 194-202. | |
| [9] | 秦子轩,张晓东,白广芝,等.基于多尺度特征融合的航空发动机剩余寿命预测[J].航空发动机, 2024, 50(4): 114-120. |
| QIN Z X, ZHANG X D, BAI G Z, et al. Aeroengine remaining useful life prediction based on multi-scale feature fusion [J]. Aeroengine, 2024, 50(4): 114-120. | |
| [10] | 陈俊英,席月芸,李朝阳.多尺度局部特征和Transformer全局学习融合的发动机剩余寿命预测[J].自动化学报, 2024, 50(9): 1818-1830. |
| CHEN J Y, XI Y Y, LI Z Y. Prediction of aeroengine remaining life by combining multi-scale local features and Transformer global learning [J]. Acta Automatica Sinica, 2024, 50(9): 1818-1830. | |
| [11] | 占可,王寅杰,董路南,等.基于改进格拉姆角场和注意力机制的滚动轴承故障诊断[J].轴承, 2024(8): 80-85. |
| ZHAN K, WANG Y J, DONG L N, et al. Fault diagnosis for rolling bearings based on improved Gramian angular field and attention mechanism [J]. Bearing, 2024(8): 80-85. | |
| [12] | 张国栋,尹强,羊柳.基于格拉姆角场和PSO-CNN的滚动轴承故障诊断方法[J].兵器装备工程学报, 2024, 45(4): 301-308. |
| ZHANG G D, YIN Q, YANG L. Rolling bearing fault diagnosis method based on Gram angle field and PSO-CNN [J]. Journal of Ordnance Equipment Engineering, 2024, 45(4): 301-308. | |
| [13] | 刘志坚,何蔚,刘航,等.基于格拉姆角场变换和深度压缩模型的变压器故障识别方法[J].电网技术, 2023, 47(4): 1478-1490. |
| LIU Z J, HE W, LIU H, et al. Fault identification method for power transformer Based on Gramian angular field transformation and deep compression model [J]. Power System Technology, 2023, 47(4): 1478-1490. | |
| [14] | REN J, WANG Z, PANG Y, et al. Genetic algorithm-assisted an improved AdaBoost double-layer for oil temperature prediction of TBM [J]. Advanced Engineering Informatics, 2022, 52: No.101563. |
| [15] | MA Y, WANG Z, GAO J, et al. A novel method for remaining useful life of solid-state lithium-ion battery based on improved CNN and health indicators derivation [J]. Mechanical Systems and Signal Processing, 2024, 220: No.111646. |
| [16] | DE PATER I, MITICI M. Developing health indicators and RUL prognostics for systems with few failure instances and varying operating conditions using a LSTM autoencoder [J]. Engineering Applications of Artificial Intelligence, 2023, 117(Pt A): No.105582. |
| [17] | ZHU Q, ZHOU Z, LI Y, et al. Contrastive BiLSTM-enabled health representation learning for remaining useful life prediction [J]. Reliability Engineering and System Safety, 2024, 249: No.110210. |
| [18] | 张研,景超,王慧民,等.基于周期注意力机制的中长期光伏发电功率预测[J].太阳能学报, 2024, 45(10): 298-308. |
| ZHANG Y, JING C, WANG H M, et al. Medium and long term photovoltaic power prediction based on periodic attention mechanism [J]. Acta Energiae Solaris Sinica, 2024, 45(10): 298-308. | |
| [19] | XU T, HAN G, ZHU H, et al. Multi-resolution LSTM-based prediction model for remaining useful life of aero-engine [J]. IEEE Transactions on Vehicular Technology, 2024, 73(2): 1931-1941. |
| [20] | ZHOU J, QIN Y, CHEN D, et al. Remaining useful life prediction of bearings by a new reinforced memory GRU network [J]. Advanced Engineering Informatics, 2022, 53: No.101682. |
| [21] | 彭文闯,郭晟楠,万怀宇,等.面向交通事故预测的时空多模态点过程[J].计算机应用研究, 2023, 40(8): 2340-2345. |
| PENG W C, GUO S N, WAN H Y, et al. Multimodal spatial-temporal point processes for traffic accident event prediction [J]. Application Research of Computers, 2023, 40(8): 2340-2345. | |
| [22] | WANG Z, MA Y, GAO J, et al. Cured memory RUL prediction of solid-state batteries combined progressive-topologia fusion health indicators [J]. IEEE Transactions on Industrial Informatics, 2025, 21(5): 4051-4060. |
| [23] | YUAN Y, YANG Q, REN J, et al. Attack-defense strategy assisted osprey optimization algorithm for PEMFC parameters identification [J]. Renewable Energy, 2024, 225: No.120211. |
| [24] | JIANG J, ZHANG X, YUAN Z. Feature selection for classification with Spearman’s rank correlation coefficient-based self-information in divergence-based fuzzy rough sets [J]. Expert Systems with Applications, 2024, 249(Pt B): No.123633. |
| [25] | FERREIRA C, GONÇALVES G. Remaining useful life prediction and challenges: a literature review on the use of machine learning methods [J]. Journal of Manufacturing Systems, 2022, 63: 550-562. |
| [26] | LI X, LIU X, YUE C, et al. Systematic review on tool breakage monitoring techniques in machining operations [J]. International Journal of Machine Tools and Manufacture, 2022, 176: No.103882. |
| [1] | 王菲, 陶冶, 刘家旺, 李伟, 秦修功, 张宁. 面向智慧家庭空间的时空知识图谱的双模态融合构建方法[J]. 《计算机应用》唯一官方网站, 2026, 46(1): 52-59. |
| [2] | 张宏俊, 潘高军, 叶昊, 陆玉彬, 缪宜恒. 结合深度学习和张量分解的多源异构数据分析方法[J]. 《计算机应用》唯一官方网站, 2025, 45(9): 2838-2847. |
| [3] | 石超, 周昱昕, 扶倩, 唐万宇, 何凌, 李元媛. 基于骨架和3D热图的注意缺陷多动障碍患者动作识别算法[J]. 《计算机应用》唯一官方网站, 2025, 45(9): 3036-3044. |
| [4] | 彭鹏, 蔡子婷, 刘雯玲, 陈才华, 曾维, 黄宝来. 基于CNN和双向GRU混合孪生网络的语音情感识别方法[J]. 《计算机应用》唯一官方网站, 2025, 45(8): 2515-2521. |
| [5] | 林进浩, 罗川, 李天瑞, 陈红梅. 基于跨尺度注意力网络的胸部疾病分类方法[J]. 《计算机应用》唯一官方网站, 2025, 45(8): 2712-2719. |
| [6] | 陶永鹏, 柏诗淇, 周正文. 基于卷积和Transformer神经网络架构搜索的脑胶质瘤多组织分割网络[J]. 《计算机应用》唯一官方网站, 2025, 45(7): 2378-2386. |
| [7] | 张英俊, 闫薇薇, 谢斌红, 张睿, 陆望东. 梯度区分与特征范数驱动的开放世界目标检测[J]. 《计算机应用》唯一官方网站, 2025, 45(7): 2203-2210. |
| [8] | 王艺涵, 路翀, 陈忠源. 跨模态文本信息增强的多模态情感分析模型[J]. 《计算机应用》唯一官方网站, 2025, 45(7): 2237-2244. |
| [9] | 吴宗航, 张东, 李冠宇. 基于联合自监督学习的多模态融合推荐算法[J]. 《计算机应用》唯一官方网站, 2025, 45(6): 1858-1868. |
| [10] | 龙雨菲, 牟宇辰, 刘晔. 基于张量化图卷积网络和对比学习的多源数据表示学习模型[J]. 《计算机应用》唯一官方网站, 2025, 45(5): 1372-1378. |
| [11] | 王丹, 张文豪, 彭丽娟. 基于深度学习的智能反射面辅助通信系统信道估计[J]. 《计算机应用》唯一官方网站, 2025, 45(5): 1613-1618. |
| [12] | 耿海军, 董赟, 胡治国, 池浩田, 杨静, 尹霞. 基于Attention-1DCNN-CE的加密流量分类方法[J]. 《计算机应用》唯一官方网站, 2025, 45(3): 872-882. |
| [13] | 袁宝华, 陈佳璐, 王欢. 融合多尺度语义和双分支并行的医学图像分割网络[J]. 《计算机应用》唯一官方网站, 2025, 45(3): 988-995. |
| [14] | 王地欣, 王佳昊, 李敏, 陈浩, 胡光耀, 龚宇. 面向水声通信网络的异常攻击检测[J]. 《计算机应用》唯一官方网站, 2025, 45(2): 526-533. |
| [15] | 张翰林, 王俊陆, 宋宝燕. 融合衍生特征的时间序列事件分类方法[J]. 《计算机应用》唯一官方网站, 2025, 45(2): 428-435. |
| 阅读次数 | ||||||
|
全文 |
|
|||||
|
摘要 |
|
|||||