Scholarlink

<p>The dynamic characteristics of electro-hydraulic proportional valve are determined by the mechanical, electromagnetic, fluid, control and their coupling effects. It is also difficult to accurately predict the transient behavior at a design state through the simplified model, empirical formula, simple coupling analysis, multi-software joint simulation, bond diagram and others. The dynamic grid model including all subsystems has present the good potential. As a result, integrating the electrical, electromagnetic, fluid, mechanical and control subsystems of an electro-hydraulic proportional valve with high performance, a coupled finite element model is established for presenting dynamic characteristics under different driving strategies, such as high-low voltage switching excitation, direct current, pulse width modulation and adaptive pulse width modulation. Further considering excellent dynamic characteristics and driving convenience, the adaptive pulse width modulation strategy has been considered as the better one. Secondly, the partial dataset of finite element model under such a strategy is statistically analyzed, and the feedforward neural network model is further utilized to reveal dynamic characteristics under more conditions, thus predicting the optimal value of adaptive pulse width modulation. Comparing with results from the finite element model under the optimal strategy, maximum absolute errors of steady-state displacement and response time from the feedforward neural network are 0.013&#xa0;mm and 2&#xa0;ms, respectively, thus determining the best driving strategy with the shortest response time and unobvious overshooting. Consequently, the optimization method combining the coupled finite element model with the neural network model has high accuracy, which avoids the repeated calculation of the finite element model under more conditions and reduces computing costs.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

The driving strategy optimization for an electro-hydraulic valve based on finite element model and feedforward neural network

  • 期刊论文

  • 发表日期：

    2025年5月2日

查看全文
  • Mingming Sun,
  • Sixiu Shi,
  • Xianju Yuan,
  • Junjie Chen,
  • Xiaobing Chen,
  • Dongxiong Wang,
  • Zhongqiang Feng

摘要

The dynamic characteristics of electro-hydraulic proportional valve are determined by the mechanical, electromagnetic, fluid, control and their coupling effects. It is also difficult to accurately predict the transient behavior at a design state through the simplified model, empirical formula, simple coupling analysis, multi-software joint simulation, bond diagram and others. The dynamic grid model including all subsystems has present the good potential. As a result, integrating the electrical, electromagnetic, fluid, mechanical and control subsystems of an electro-hydraulic proportional valve with high performance, a coupled finite element model is established for presenting dynamic characteristics under different driving strategies, such as high-low voltage switching excitation, direct current, pulse width modulation and adaptive pulse width modulation. Further considering excellent dynamic characteristics and driving convenience, the adaptive pulse width modulation strategy has been considered as the better one. Secondly, the partial dataset of finite element model under such a strategy is statistically analyzed, and the feedforward neural network model is further utilized to reveal dynamic characteristics under more conditions, thus predicting the optimal value of adaptive pulse width modulation. Comparing with results from the finite element model under the optimal strategy, maximum absolute errors of steady-state displacement and response time from the feedforward neural network are 0.013 mm and 2 ms, respectively, thus determining the best driving strategy with the shortest response time and unobvious overshooting. Consequently, the optimization method combining the coupled finite element model with the neural network model has high accuracy, which avoids the repeated calculation of the finite element model under more conditions and reduces computing costs.