摘要
为掌握液压泵全寿命周期健康状态,提出一种基于改进VMD算法的液压泵寿命状态检测方法。针对变分模态分解中难以确定分解层数和分解带宽的问题,引入萤火虫算法对VMD初始参数组合进行优化,通过仿真分析验证该方法的有效性。开展液压泵加速寿命试验,提取液压泵不同寿命阶段出口压力脉动信号,应用改进VMD算法进行分解,计算各IMF分量的能量占比、IMF重构信号的能量熵及时域指标作为12维状态特征样本库,建立结构为12-20-5的DBN神经网络进行液压泵寿命状态识别。分析结果表明,该方法在泵全部寿命阶段均能保证较高的识别准确率,平均准确率达到97.4%,为液压泵寿命状态检测提供了新的方法。
In order to grasp the operating status of hydraulic pump in time,a method for monitoring the life of hydraulic pump based on the improved VMD algorithm is proposed.Aiming at the difficulties to select the parameters of decomposition layers and decomposition bandwidth in variational mode decomposition,the firefly algorithm is used to optimize the initial parameters combination of variational modal decomposition,and the validity of the method is verified by simulation signals.Accelerated life test of hydraulic pump is platformed and pressure ripple signals of pump outlet at different life stages are extracted.Appling improved VMD algorithm,energy ratio of each IMF component and energy entropy are calculated and time domain index of IMF reconstruction signal are extracted as 12-dimensional state feature sample database.A DBN neural network with a structure of 12-20-5 is established to identify the life status of the hydraulic pump.Results show that this method distinguish different stages of the pump life effectively with an average accuracy rate of 97.4%,which provides a new method for the life state detection of the hydraulic pump.
作者
宣元
何琳
陈宗斌
廖健
XUAN Yuan;HE Lin;CHEN Zong-bin;LIAO Jian(Institution of Vibration & Noise, Naval University of Engineering, Wuhan, Hubei 430033;National Key Laboratory on Ship Vibration & Noise, Naval University of Engineering, Wuhan, Hubei 430033)
出处
《液压与气动》
北大核心
2020年第10期69-77,共9页
Chinese Hydraulics & Pneumatics
基金
国防科技重点实验室基金(6142204180301)。
关键词
变分模态分解
萤火虫算法
加速寿命试验
特征识别
variational mode decomposition
firefly algorithm
accelerated life test
state characteristics
