摘要
提出基于粒子群优化(Particle Swarm Optimization,PSO)的极限学习机(Extreme Learning Machine,ELM)的液压油性能衰退预测方法。以L-HM46抗磨液压油为研究对象,设计液压油性能衰退实验,检测油液的黏度、张角、水分含量、衰退度。基于提出的液压油性能衰退预测方法,利用遍历搜索和PSO算法分别对ELM的外部、内部参数进行优化选取,从而建立最优的性能衰退预测模型。将油液的黏度、张角、水分含量作为模型输入特征向量,衰退度作为模型输出,采用PSO-ELM性能衰退预测模型对液压油性能进行仿真分析。结果表明:PSO-ELM算法计算结果与实验数据吻合较好;PSO-ELM算法预测精度达到了98.47%,高于ELM算法的预测精度,表明PSO-ELM算法能更准确地预测液压油的衰退情况,为确定换油时机提供参考。
A prediction method of hydraulic oil performance degradation based on extreme learning machine(ELM)with particle swarm optimization(PSO)was proposed.Taking L-HM46 anti-wear hydraulic oil as an example,the oil performance degradation experiment was designed to detect the viscosity,opening angle,moisture content and decline degree of the hydraulic oil.Based on the proposed performance degradation prediction method of hydraulic oil,the external and internal parameters of ELM were optimized by ergodic search and PSO algorithm respectively,and the optimal performance degradation prediction model was established.Taking viscosity,opening angle and,moisture content as the input eigenvector and decline degree of hydraulic oil as the output of the model,the performance of hydraulic oil was simulated and analyzed by PSO-ELM performance degradation prediction model.The results show that the calculation results of PSO-ELM algorithm are in good agreement with the experimental data.The prediction accuracy of PSO-ELM algorithm reaches 98.47%,which is higher than that of elm algorithm.It shows that PSO-ELM algorithm can more accurately predict the decline of hydraulic oil,and provide a reference for determining the oil change time.
作者
宋新成
杨洁
王崴
李恒威
郝俊杰
郭亮亮
SONG Xincheng;YANG Jie;WANG Wei;LI Hengwei;HAO Junjie;GUO Liangliang(Air Defense and Missile Defense College,Air Force Engineering University,Xi'an Shaanxi 710051,China;Unit 93285 of PLA,Gongzhuling Jilin 136100,China)
出处
《润滑与密封》
CAS
CSCD
北大核心
2021年第12期131-135,共5页
Lubrication Engineering
基金
军内科研项目(KJ2018-2019C060)。
关键词
粒子群优化
极限学习机
液压油
性能衰退
particle swarm optimization
limit learning machine
hydraulic oil
performance degradation
