金属颗粒在交变磁场中的磁化机理及其基于复阻抗信号的识别方法

On a ship, metallic particle distribution and composition within lubrication and hydraulic systems are of great significance for condition monitoring and fault diagnosis. At present, detection mechanism of the coil-type particle sensor has limitations, which leads to the contradiction between the detection efficient and the sensitivity, the rough classification of particles into ferromagnetic and non-ferromagnetic metal, and failure on size estimation for the unknown particle properties. Therefore, this project takes the magnetization response of metal particles under alternating magnetic field as the entry point, the relationship among particle complex magnetic permeability, the scattering field, the particle material properties and the external excitation is researched. The effects of metal particle scattering field on complex impedance of sensor is explored, the model of complex impedance and particle properties is established, then particle material identification and particle size recognition are realized based on permeability and conductivity. By obtaining kernel function and its properties based on complex impedance output model, the kernel function is used to perform signal recovery of particle sparse spike sequence, thereby separating the superimposed particle signals. This project is the fundamental theory for optimization and sensitivity improvement of online inductive particle sensor.

船舶润滑与液压系统中，金属颗粒在油液中的粒径分布及成分对机械设备的状态监测和故障诊断具有重要意义。目前线圈式磨粒传感器检测机理研究不深入导致其存在局限性，如检测通量和灵敏度形成矛盾、只能将金属颗粒粗略分为铁磁和非铁磁金属两大类以及无法在未知颗粒属性情况下估计粒度等。为此本项目以金属颗粒在交变磁场下磁化响应研究为切入点，探索颗粒复数磁导率、散射场、颗粒材料属性和外部激励的关系。探索金属颗粒散射场对线圈式传感器的复阻抗影响规律，从而构建复阻抗与颗粒属性的模型，实现基于磁导率电导率两个参数的颗粒材质识别和粒度测量。根据复阻抗输出模型，探索线圈式传感器的核函数获取方法及其性质，利用核函数进行颗粒离散脉冲序列的信号复原处理，从而分离叠加的颗粒信号。本课题是线圈式颗粒传感器的优化以及高灵敏度的在线检测的基础理论。

船舶润滑与液压系统中，金属颗粒在油液中的粒径分布及成分对机械设备的状态监测和故障诊断具有重要意义。当前线圈式磨粒传感器检测机理研究不深入导致其存在局限性：如检测通量和灵敏度形成矛盾，检测结果只能将金属颗粒粗略分为铁磁和非铁磁金属两大类，无法在颗粒材质未知情况下估计其尺寸。针对上述问题，本项目以金属颗粒在交变磁场下磁化响应研究为切入点，掌握了颗粒复数磁导率、散射场、颗粒材料属性和外部激励的关系。探索了金属颗粒散射场对线圈式传感器的复阻抗影响规律，从而构建复阻抗与颗粒属性的模型，实现了基于颗粒的磁导率电导率两个参数的颗粒材质识别和粒度测量。根据上述机理模型，探索了线圈式传感器的核函数获取方法及其性质，提出了基于机器学习的颗粒信号复原方法，实现多种金属颗粒混合时颗粒材质辨识和粒度测量。