针对船用LNG双燃料发动机设备复杂，故障预测效率低、准确度差的问题，提出一种长短期记忆网络与改进粒子群优化算法优化支持向量机融合的预测模型。利用LSTM模型时间序列变化的能力对设备未来的运行状态进行预测，然后采用非线性自适应惯性权重改进PSO算法对SVM参数进行寻优，以提高其寻优能力和收敛速度；改进的LSTM-PSO-SVM融合模型可实现对设备故障状态的快速、准确预测。通过对某船用LNG双燃料发动机的故障预测仿真，结果表明上述模型具有更高的故障识别准确率和更快的识别速度，能够准确预测船用LNG双燃料发动机潜在故障。
In order to solve the problems of complex equipment, low efficiency and poor accuracy of fault prediction for marine LNG dual-fuel engine, a prediction model combining long-short term memory network (LSTM) and improved particle swarm optimization algorithm (PSO) to optimize the model of support vector machine (SVM) was proposed. The time series capability of the LSTM model is used to predict the future operating state of the device, and then the SVM parameters are optimized using a non-linear adaptive inertia weight improvement PSO algorithm to improve the optimization power and convergence rate. The improved LSTM-PSO-SVM fusion model can predict the equipment fault state quickly and accurately. Through the fault prediction simulation of a low-speed Marine LNG dual-fuel engine, the results show that the above model has accurate and efficient prediction ability, and can accurately identify potential faults of Marine LNG dual-fuel engine.
2024,46(4): 120-126 收稿日期：2023-03-01
DOI：10.3404/j.issn.1672-7649.2024.04.023
分类号：TP206+.3；U664.1
基金项目：江苏省重点研发计划资助项目（BE2021075）；江苏省科技成果转化专项资金项目（BA2022066）
作者简介：姜峰(1996-)男,硕士研究生,研究方向为船舶智能运维与保障
参考文献：
[1] 雷伟, 甘少炜, 周国强. 船用双燃料发动机关键技术分析[J]. 船海工程, 2016, 45(3): 86-89+98.
LEI W, GAN S W, ZHOU G Q. Analysis of key technologies of marine dual-fuel engine[J]. Ship and Sea Engineering, 2016, 45(3): 86-89+98.
[2] 杨奕飞, 冯静. 基于HMM-SVR的船舶动力设备故障模式识别与状态预测研究[J]. 船舶工程, 2018, 40(3): 68-72+97.
YANG Y F, FENG J. Study on fault pattern recognition and state prediction of marine power equipment based on HMM-SVR[J]. Marine Engineering, 2018, 40(3): 68-72+97.
[3] YANG Y, TAN M, DAI Y. An improved CS-LSSVM algorithm-based fault pattern recognition of ship power equipments[J]. Plos One, 2017, 12(2): e0171246.
[4] 王文波, 彭伟, 顾寄南, 等. 基于深度学习与增量更新的滚动轴承故障预测方法研究[J]. 江苏科技大学学报(自然科学版), 2022, 36(5): 54-59.
WANG W B, PENG W, GU J N, et al. Research on rolling bearing fault prediction method based on deep learning and incremental updating[J]. Journal of Jiangsu University of Science and Technology (Natural Science Edition), 2022, 36(5): 54-59.
[5] CHUI KT, GUPTA BB, VASANT P. A genetic algorithm optimized RNN-LSTM model for remaining useful life prediction of turbofan engine[J]. Electronics, 2021, 10(3): 285.
[6] 石彪, 王海燕, 焦品博. 基于改进GWO-LSTM的船舶主机性能预测模型[J]. 上海海事大学学报, 2022, lancet(2): 96-102.
SHI B, WANG H Y, JIAO P B. Based on improved GWO comes-LSTM ship host performance prediction model[J]. Journal of Shanghai maritime university, 2022, lancet(2): 96-102.
[7] 于飞, 樊清川, 宣敏. 基于蝗虫优化Bi-LSTM网络的电机轴承故障预测[J]. 电机与控制学报, 2022, 26(6): 9-17.
YU F, FAN Q C, XUAN M. Based on locust optimization Bi-LSTM network of motor bearing fault prediction[J]. Journal of motor and control, 2022, 26(6): 9-17.
[8] 顾吉峰, 王蓓. 基于改进粒子群算法的孪生支持向量机[J]. 计算机工程与设计, 2020, 9(11): 3078-3082.
GU J F WANG B. Based on improved particle swarm optimization twin support vector machine[J]. Computer engineering and design, 2020, 9(11): 3078-3082.
[9] WANG Y, WANG D, TANG Y. Clustered hybrid wind power prediction model based on ARMA, PSO-SVM, and clustering methods[J]. IEEE Access, 2020, 8: 17071-17079.
[10] PENG W, WANG J, YING S. Short-term load forecasting model based on attention-LSTM in electricity market[J]. Power System Technology, 2019, 43(5): 1745-1751.
[11] LIU J, PAN C, LEI F, et al. Fault prediction of bearings based on LSTM and statistical process analysis[J]. Reliability Engineering & System Safety, 2021, 214: 107646.
[12] 李元, 张昊展, 唐晓初. 基于多模态数据全信息的概率主成分分析故障检测研究[J]. 仪器仪表学报, 2021, 42(2): 75-85.
LI Y, ZHANG H Z, TANG X C. Modal data based on probabilistic principal component analysis of comprehensive information fault detection study [J]. Journal of Instruments and Meters, 2021 (2): 75-85.
[13] 王超, 李大中. 基于LSTM网络的风机齿轮箱轴承故障预警[J]. 电力科学与工程, 2020, 36(9): 40-45.
WANG C, LI D Z. Fault warning of fan gearbox bearing based on LSTM network[J]. Electric Power Science and Engineering, 2020, 36(9): 40-45.
[14] YANG Y, ZHANG M, DAI Y. A fuzzy comprehensive CS-SVR model-based health status evaluation of radar[J]. Plos one, 2019, 14(3): e0213833.
[15] 高金武, 贾志桓, 王向阳, 等. 基于PSO-LSTM的PEMFC退化趋势预测[J/OL]. 吉林大学学报(工学版), 1-10[2022-12-01].
GAO J W, JIA Z H, WANG X Y, et al. Based on PSO-LSTM PEMFC degradation trend prediction [J/OL]. Journal of Jilin University (Engineering Science), 1-10[2022-12-01].
[16] 张鑫, 徐遵义, 何慧茹, 等. 基于RBM和SVM的风电机组叶片开裂故障预测[J]. 电力系统保护与控制, 2020, 13(15): 134-140.
ZANG X, XU Z Y, HE H R, et al. Matter and the SVM based wind turbine blade crack fault prediction[J]. Power System Protection and Control, 2020, 13(15): 134-140.
[17] 梁川. 基于一维卷积神经网络的风机故障预测[J]. 科技创新与应用, 2020(32): 132-135.
LIANG C. Fan fault prediction based on one-dimensional convolutional neural network[J]. Science and Technology Innovation and Application, 2020(32): 132-135.
