基于前馈与混合自适应算法的振动主动控制方法研究

发布时间：2018-05-23 12:57

本文选题：振动主动控制 + 自适应控制　；参考：《上海大学》2014年博士论文

【摘要】：振动主动控制涉及到结构动力学、声学、材料科学技术、电子信息与技术、控制科学与工程等诸多学科,是目前振动工程领域最重要的研究方向。作为振动主动控制研究的核心内容,当前振动主动控制方法及其算法的研究如火如荼,近乎包含了所有现存的控制理论方法与控制技术。但总体而言,多数的控制方法和技术多多少少都存在适用性和实践性的问题,目前学术界还没有提出一种能够适用于多数振动主动控制系统控制器设计的范式方法与技术,相关的理论方法和技术均亟待进一步的深入研究与探索。基于国家自然科学基金重大研究计划科研课题背景,围绕“在保证受控系统稳定的基础上,提高整个振动主动控制系统的鲁棒性与控制性能”这一核心问题,本文选取当前研究热点之一的自适应振动主动控制方法与技术为切入点,着重面向自适应前馈振动控制方法与自适应混合振动控制方法,进行相关方法研究、控制算法推导、算法分析与实验验证工作。在理论方法上给出相关控制算法的稳定性与收敛性分析范式方法,在技术方法上给出相关控制算法实施的一般技术指引。论文主要研究工作与学术贡献如下： (1)使用惯性作动器作为抑振作动器与模拟扰动作动器,使用加速度传感器作为参考传感器与残差传感器,构建具有典型机械结构振动正反馈的实验平台和实验环境。形成一套半实物振动主动控制仿真分析与实时振动主动控制实验平台,用于实现面向工程应用的振动主动控制方法研究与验证。 (2)针对现有参数自适应算法的局限性,研究和导出一种范式改进型递推最小二乘算法；给出长时间实时自适应控制过程中,使用U-D分解技术更新自适应增益矩阵的具体过程与详细方法。 (3)针对振动主动控制系统面临的系统辨识问题,研究和推导递推最小二乘算法、扩展递推最小二乘算法和扩展预测模型输出误差算法等三种辨识方法。确定白化测试校验准则与模型校验方法,给出振动主动控制系统参数辨识的一般方法与技术流程。 (4)基于改进型递推最小二乘参数自适应算法族,提出自适应前馈振动主动控制算法的范式描述方法；基于超稳定性理论,给出了完整的控制算法推导过程与算法分析过程。通过分析给出了确定环境下算法稳定与随机环境下算法收敛的严格正实条件,以及相应的放松严格正实条件的要求。针对确定环境满足完美匹配条件、随机环境满足完美匹配条件以及不满足完美匹配条件的情况,分别给出了相关控制算法的仿真算例。仿真结果表明：通过引入特定的预滤波器,能够显著改善自适应前馈振动控制算法的控制效果；在测量环境存在较大噪声时以及参数数目不能满足完美匹配条件时,所给出的自适应前馈振动控制算法依然有效,同时具有优良的抑振效果与收敛性能。 (5)为改善单纯前馈自适应振动控制算法的收敛性能与控制效果,推导了一种基于混合自适应控制算法的振动主动控制方法,提出了混合自适应振动主动控制算法的范式描述与分析方法,给出了完整的自适应混合振动控制算法的推导过程与算法分析过程。通过分析给出了确定环境下算法稳定与随机环境下算法收敛的严格正实条件,以及相应的放松严格正实条件的要求。同样针对确定环境满足完美匹配条件、随机环境满足完美匹配条件以及不满足完美匹配条件的情况,分别给出了相关控制算法的仿真算例。仿真结果表明：混合自适应控制器能够有效改善前馈自适应振动控制器的性能,以更少的控制器阶次,得到更佳的控制效果。 (6)针对所提出的自适应前馈振动控制算法、自适应混合振动控制算法进行实验分析与验证工作,并将相关结果与前馈FULMS算法、混合FULMS算法予以比较。相关实验结果表明：基于改进型递推最小二乘参数自适应算法族的所有前馈振动控制算法,其控制效果均明显优于前馈FULMS算法；基于改进型递推最小二乘参数自适应算法族的所有混合自适应振动控制算法,其控制效果均明显优于混合FULMS算法、结合极点配置反馈的混合前馈自适应控制算法,以及基于改进型递推最小二乘参数自适应算法族的前馈振动控制算法。
[Abstract]:Active vibration control, which involves structural dynamics, acoustics, material science and technology, electronic information and technology, control science and engineering, is the most important research direction in the field of vibration engineering. As the core content of the active vibration control research, the current research on the main vibration control method and its algorithm is in full swing. All the existing control theory methods and control techniques are included, but in general, most of the control methods and techniques are both practical and practical. At present, the academic circle has not put forward a paradigm and technique that can be applied to most of the controller design of the active vibration control system. The technology needs further research and exploration.
Based on the background of the scientific research project of the National Natural Science Foundation of China, this paper focuses on the core problem of "improving the robustness and control performance of the whole vibration active control system on the basis of ensuring the stability of the controlled system". This paper selects the adaptive vibration active control method and technology as one of the hot spots in the current research. The adaptive forward feed vibration control method and the adaptive hybrid vibration control method are used to study the related methods, the control algorithm derivation, the algorithm analysis and the experimental verification. In the theoretical method, the stability and convergence analysis paradigm method of the related control algorithm is given, and the general implementation of the related control algorithm is given in the technical method. The main research work and academic contributions are as follows:
(1) using an inertial actuator as an anti vibration actuator and an analogue actuator, the acceleration sensor is used as a reference sensor and a residual sensor, the experimental platform and the experimental environment for the positive feedback of a typical mechanical structure vibration are constructed. A set of simulation analysis for the active control system of the semi physical vibration and the experiment of active control of the real time vibration are formed. The platform is used to research and verify the active vibration control method for engineering application.
(2) in view of the limitations of the existing parameter adaptive algorithm, a new type of recursion least squares algorithm is developed and derived. The specific process and detailed method of updating adaptive gain matrix using U-D decomposition technique in the process of long time adaptive control are given.
(3) aiming at the system identification problem of the active vibration control system, we study and deduce the recursive least square algorithm, expand the three identification methods, such as the recursive least square algorithm and the output error algorithm of the extended prediction model, and determine the whitening test verification criterion and the model verification method, and give the general square of the parameter identification of the active vibration control system. The process of law and technology.
(4) based on the improved recursive least square parameter adaptive algorithm family, a paradigm description method of adaptive feedforward vibration active control algorithm is proposed. Based on the super stability theory, the complete control algorithm derivation process and the algorithm analysis process are given. The algorithm convergence under the environment and random environment is given by analysis. In order to determine the perfect match condition of the environment, the random environment satisfies the perfect matching condition and the condition that the perfect match is not satisfied, the simulation example of the correlation control algorithm is given. The simulation results show that the specific prefilter is introduced. The adaptive feedforward vibration control algorithm can significantly improve the control effect of the adaptive feedforward vibration control algorithm. When the measurement environment has large noise and the number of parameters can not meet the perfect matching condition, the adaptive feedforward vibration control algorithm is still effective, and it has excellent vibration suppression effect and convergence performance.
(5) in order to improve the convergence performance and control effect of the simple feedforward adaptive vibration control algorithm, a vibration active control method based on the hybrid adaptive control algorithm is derived. The model description and analysis method of the hybrid adaptive vibration active control algorithm is proposed, and the complete adaptive hybrid vibration control algorithm is derived. The process and algorithm analysis process. Through the analysis, the strict positive real conditions of algorithm convergence under the environment algorithm stability and random environment are given, as well as the corresponding requirements for the relaxation of strict positive real conditions. It is also aimed at satisfying the perfect matching conditions for the environment to meet the perfect matching conditions, and the random environment satisfies the perfect matching condition and does not satisfy the perfect matching condition. The simulation results show that the hybrid adaptive controller can effectively improve the performance of the feedforward adaptive vibration controller, with less controller order and better control effect.
(6) in view of the proposed adaptive feedforward vibration control algorithm, the adaptive hybrid vibration control algorithm is used for experimental analysis and verification, and the correlation results are compared with the feedforward FULMS algorithm and the hybrid FULMS algorithm. The experimental results show that all the feedforward vibration control based on the improved recursive minimum two multiplication parameter adaptive algorithm is based on the experimental results. The control effect of the algorithm is obviously better than the feedforward FULMS algorithm, and all the hybrid adaptive vibration control algorithms based on the improved recursive least squares parameter adaptive algorithm are obviously better than the hybrid FULMS algorithm, combined with the hybrid feedforward adaptive control algorithm with the pole assignment feedback, and the improved recursive algorithm. Feedforward vibration control algorithm for small multiplicative parameter adaptive algorithm family.
【学位授予单位】：上海大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TB535

【参考文献】