摘要
电液伺服泵控系统具备功重比高、响应快等优点,在多领域得到广泛应用,但如何针对该系统开展更有效健康评估,进一步保障系统的安全性和可靠性成为必须面对的问题。按照明确原理、建立数学模型、建立仿真模型、仿真分析的思路针对健康评估方法开展研究,提出油液体积含气量、气隙磁密、泄漏系数3个健康评估指标并确定阈值,构建了LGA(LSTM-GRNN-ANN)深度神经网络健康评估方法并进行仿真分析,结果显示该方法准确率约为97.48%,比LSTM、GRNN健康评估方法具有更高的准确率,为继续深入开展电液伺服泵控系统健康评估的研究提供了理论支持。
The electro-hydraulic servo pump control system has the advantages of high power to weight ratio and fast response,and is widely used in various fields.But how to conduct more effective health assessments for the system and further ensure its safety and reliability has become a necessary issue to face.Research on health assessment methods was conducted according to clearing principles,establishing mathematical models,establishing simulation models,and making simulation experiments.Three health assessment indicators as oil volume gas content,air gap magnetic density,and leakage coefficient,were proposed and their thresholds were determined.Then a LGA(LSTM-GRNN-ANN)deep neural network health assessment method was constructed and simulation analysis was conducted.The results show that the accuracy of this method is about 97.48%,which is higher than LSTM and GRNN health assessment methods.This provides theoretical support for further research on health assessment of electro-hydraulic servo pump control systems.
作者
刘克毅
李渊
王飞
陈革新
王梦
张亚欧
LIU Keyi;LI Yuan;WANG Fei;CHEN Gexin;WANG Meng;ZHANG Ya’ou(Key Laboratory of Xinjiang Coal Resources Green Mining,Ministry of Education,Xinjiang Institute of Engineering,Urumqi Xinjiang 830023,China;School of Mechanical and Electrical Engineering,Xinjiang Institute of Engineering,Urumqi Xinjiang 830023,China;School of Mechanical and Aerospace Engineering,Jilin University,Changchun Jilin 130022,China;School of Mechanical Engineering,Yanshan University,Qinhuangdao Hebei 066004,China)
出处
《机床与液压》
北大核心
2024年第9期173-179,共7页
Machine Tool & Hydraulics
基金
新疆维吾尔自治区自然科学基金资助项目(2022D01A51,2022D01B135,2022D01A244)
新疆煤炭资源绿色开采教育部重点实验室自主课题(KLXGY-Z2402)。
关键词
电液伺服泵控系统
健康评估
LGA深度神经网络
仿真分析
electro-hydraulic servo pump control system
health assessment
LGA deep neural network
simulation analysis
