基于GA-PSO混合算法的钢杆磁特性参数识别方法

发布时间：2018-04-01 21:08

本文选题：磁特性曲线　切入点：参数识别　出处：《仪器仪表学报》2017年04期

【摘要】：测量轴类零件的磁滞回线,利用其特征参数的变化表征零件表面硬度及硬化层深度,是具有工程应用前景的电磁无损检测新技术之一,其关键是轴类零件磁特性曲线测量装置的研制和磁特性参数高精度识别方法的研究。设计出一种基于闭环磁路的钢杆磁滞回线测量实验装置,并基于J-A磁滞模型,提出了一种遗传粒子群(GA-PSO)混合算法,实现了钢杆磁滞回线全局与局部特征参数的快速、高精度识别。实验测得的3种不同材质钢杆磁滞回线,对比分析了混合优化算法与单一算法(遗传、粒子群、模拟退火)的参数识别速度与精度,结果表明,混合算法全局识别结果的最小均方根误差仅为0.004 7,低于单一算法的相应结果;混合算法对局部特征参数(矫顽力、剩余磁感应强度)识别的相对误差均小于0.35%,优于单一算法识别精度。上述实验测试和磁特性参数识别方法,有望应用于销钉、螺栓等轴类构件表面硬化层的无损检测。
[Abstract]:Measuring the hysteresis loop of shaft parts and using the change of characteristic parameters to characterize the surface hardness and hardened layer depth of the parts is one of the new electromagnetic nondestructive testing techniques with engineering application prospect.The key is the development of measuring device for magnetic characteristic curve of shaft parts and the research of high precision identification method of magnetic characteristic parameters.An experimental device for measuring hysteresis loop of steel rod based on closed loop magnetic circuit is designed. Based on J-A hysteresis model, a hybrid genetic particle swarm optimization (GA-PSO) algorithm is proposed. The global and local characteristic parameters of the hysteresis loop of steel rod are realized quickly.High precision recognition.The hysteresis loops of three kinds of steel bars with different materials are measured experimentally. The parameter identification speed and accuracy of hybrid optimization algorithm and single algorithm (genetic, particle swarm optimization, simulated annealing) are compared and analyzed. The results show that,The minimum root mean square error of the global recognition result of hybrid algorithm is only 0.004, which is lower than that of single algorithm.The relative error of the residual magnetic induction intensity) is less than 0.35, which is superior to the recognition accuracy of the single algorithm.The above experimental test and magnetic characteristic parameter identification method are expected to be used in nondestructive testing of surface hardening layer of axial components such as pins bolts and so on.
【作者单位】：北京工业大学机械工程与应用电子技术学院;
【基金】：国家自然科学基金(11527801)项目资助
【分类号】：TG115;TP18

【相似文献】