摘要
在复杂工况下,偏导射流伺服阀可采集的故障信号有限且易受噪声干扰,导致其特征提取困难。针对该问题,提出了一种基于海星优化变分模态分解、时域卷积网络、引入自注意力机制的双向门控循环单元的故障诊断方法。首先,利用海星优化算法自适应确定变分模态分解参数,提高信号分解的准确性与鲁棒性;随后结合最小包络熵原则选取关键固有模态函数,从中提取主要特征;最后,将提取的特征融入时域卷积网络与自注意力机制增强的双向门控循环单元网络,提高故障特征的表达能力与分类性能。为验证所提方法的有效性,构建了偏导射流伺服阀故障仿真平台和开展了多种典型故障工况试验;结果表明,所提模型的故障识别准确率达到97.33%,具有较强的鲁棒性和诊断精度。
Deflector jet servo valve fault signals are limited and easily affected by noise under complex conditions,resulting in difficult feature extraction.This paper presents a fault diagnosis method combining starfish optimization algorithm-based variational mode decomposition,temporal convolutional network,and a self-attention bidirectional gated recurrent unit network.The starfish optimization algorithm selects variational mode decomposition parameters to improve decomposition accuracy and robustness.Main features are extracted from key intrinsic mode functions based on minimum envelope entropy.These features are entered into a temporal convolutional network and a self-attention-based bidirectional gated recurrent unit network to enhance fault classification.A fault simulation platform and dataset are built,with experiments under typical fault conditions.Results show that the fault recognition accuracy of the method achieves 97.33%,demonstrating strong robustness and high diagnostic performance.
作者
张帅印
陶建峰
吴兆宇
陈方飞扬
谭浩洋
ZHANG Shuaiyin;TAO Jianfeng;WU Zhaoyu;CHEN Fangfeiyang;TAN Haoyang(School of Mechanical and Power Engineering,Shanghai Jiao Tong University,Shanghai 200240;State Key Laboratory of Mechanical System and Vibration,Shanghai Jiao Tong University,Shanghai 200240;Shanghai Tunnel Engineering Co.,Ltd.,Shanghai 200137)
出处
《液压与气动》
北大核心
2025年第8期1-11,共11页
Chinese Hydraulics & Pneumatics
基金
国家重点研发计划(2024YF0505303)
上海隧道工程有限公司项目(2022-SK-01-3)。
关键词
偏导射流伺服阀
变分模态分解
双向门控循环单元
故障诊断
deflector jet servo valve
variational mode decomposition
bidirectional gated recurrent unit
fault diagnosis
