项目名称: 超磁致伸缩材料的非线性动力学及驱动器控制研究
项目编号: No.11272229
项目类型: 面上项目
立项/批准年度: 2013
项目学科: 数理科学和化学
项目作者: 竺致文
作者单位: 天津大学
项目金额: 86万元
中文摘要: 本课题基于超磁致伸缩材料的动态迟滞非线性特性,首次考虑磁致伸缩率与磁场强度、磁场频率、温度、应力4个变量的耦合关系,将人工智能法(如支持向量机法、规则集成法)和机器学习理论(如随机梯度法、随机森林法)等多元统计回归法,引入到超磁致伸缩材料的多场耦合非线性本构模型的建立中;研究超磁致伸缩驱动器的非线性动力学特性,提出改进的复规范形方法求解系统的强非线性高精度解;首次提出基于增量型多变量支持向量机的逆模型补偿策略,设计出滑模变结构PID控制器,实现精密快速控制;首次从工程实际中存在的随机干扰出发,研究驱动器在随机扰动下的随机动力学特性,分析系统的全局稳定性、随机分岔特性和安全域的首次穿越时间,再应用随机最优控制策略使系统始终处于安全域,从而提高系统的可靠性。这对于推动超磁致伸缩材料的本构模型的建立和超磁致伸缩驱动器的动力学特性理论研究及广泛应用有着重要的理论意义和工程应用价值。
中文关键词: 超磁致伸缩材料;非线性;磁滞;多场耦合;随机最优控制
英文摘要: In this project, the coupling relationship among magnetic field intensity, frequency, stress and temperature of giant magnetostrictive materials was studied based on the hysteretic nonlinear characteristics and the multi-variable coupling characteristics of giant magnetostrictive materials. Artificial intelligence method (such as support vector machine method and rule integration method) and machine learning theory (such as the stochastic gradient method and random forest method) were introduced to set up the multi-field coupling nonlinear constitutive model of giant magnetostrictive materials. The nonlinear dynamic characteristics of giant magnetostrictive actuator were studied, and strong-nonlinear solution with high precision was obtained in improved complex normal form method. Incremental multivariate support vector machine inverse model compensation strategy was proposed, and the sliding -mode variable structure PID controller was designed to control rapidly and accurately. The stochastic nonlinear dynamic characteristics of giant magnetostrictive actuator subject to stochastic excitation were studied, the global stochastic bifurcation characteristics and the first-passage time of security domain were analyzed. The stochastic optimal control strategy was proposed to delay the first-passage time and improve
英文关键词: giant magnetostrictive materials;nonlinear;hysteresis;multi-field coupling;stochastic optimal control