液压与气动2025，Vol.49Issue(12)：48-56,9.DOI:10.11832/j.issn.1000-4858.2025.12.005
基于物理信息神经网络的气动控制阀阀芯不平衡力计算方法
Physics-informed Neural Network-based Computational Method for Unbalanced Force Calculation of Pneumatic Control Valving Element
摘要
Abstract
To address the issues of low efficiency in calculating unbalanced forces on pneumatic control valving element and the difficulty in accurately predicting dynamic characteristics.Applying a physics-informed neural network to calculations of valving element unbalanced force is proposed.By analyzing the working principle of pneumatic control valves and the generating mechanism of valving element unbalanced forces,the neural network model integrating fundamental fluid mechanics equations and simulation data is constructed.This enables efficient prediction of valving element unbalanced forces,with the method's validity ultimately verified through CFD simulations.Research findings demonstrate that the physics-informed neural network predictions closely match CFD simulations,with computation time reduced to under 10 seconds.The maximum error(＜8％)occurs at the valve's critical closing point.This approach overcomes the high computational costs of traditional methods,enabling real-time,precise prediction of dynamic unbalanced forces.It provides a novel technical pathway for intelligent design and dynamic characteristic optimization of pneumatic control valves.关键词
气动控制阀/阀芯不平衡力/物理信息神经网络/数值模拟Key words
pneumatic control valve/valving element unbalanced force/physics-informed neural network/numerical simulation分类
机械制造引用本文复制引用
郝洪涛,田源,麦学武,王月磊..基于物理信息神经网络的气动控制阀阀芯不平衡力计算方法[J].液压与气动,2025,49(12):48-56,9.基金项目
国家自然科学基金(52465014) （52465014）
宁夏回族自治区自然科学基金(2025AAC020022) （2025AAC020022）
