偏差补偿最小二乘参数估计方法研究

发布时间：2018-09-13 14:44

【摘要】：课题以复杂机械系统动力学建模与设计为工业应用背景,在国家自然科学基金项目《连续非线性动力学系统参数模型小波调制直接辨识》(项目号:50875028)资助下,拟定偏差补偿最小二乘参数估计方法为研究内容。主要进行了以下几个方面的研究: (1)介绍了调制函数法的基本原理和四种调制函数的性质和特点,并根据其原理建立了基于调制函数法的连续系统辨识模型。 (2)引入了一种偏差补偿最小二乘算法。研究了该方法在离散系统中的辨识原理以及其递推过程。将该方法应用到有色噪声干扰的离散系统模型参数辨识的算例中,验证了该算法的有效性。 (3)建立了Hermite调制函数法连续系统等价模型。将离散模型辨识中偏差补偿原理应用于线性连续系统模型辨识,以有色噪声干扰下的线性连续系统为例,证明基于调制函数法的偏差补偿最小二乘法对辨识线性连续系统模型参数有较好的噪声抑制效果。 (4)将本文的方法应用于某冷轧平整机液压AGC压力闭环系统的2/2、1/2阶动态参数模型辨识,与测试结果比较,验证了本文方法的实用性。研究表明:偏差补偿原理无论是对离散模型辨识还是连续模型辨识,均有较好的抗干扰能力,并有广泛的适用性。
[Abstract]:Under the background of complex mechanical system dynamic modeling and design, supported by National Natural Science Foundation project "Direct Identification of Parameter Model of continuous nonlinear dynamic system by Wavelet Modulation" (project number: 50875028), The least square parameter estimation method of deviation compensation is proposed as the research content. The main contents are as follows: (1) the basic principle of the modulation function method and the properties and characteristics of the four modulation functions are introduced. According to its principle, a continuous system identification model based on modulation function method is established. (2) A deviation compensated least square algorithm is introduced. The identification principle and recursive process of this method in discrete systems are studied. The method is applied to the parameter identification of discrete system with colored noise interference, and the validity of the algorithm is verified. (3) the equivalent model of continuous system with Hermite modulation function method is established. The principle of deviation compensation in discrete model identification is applied to linear continuous system model identification. The linear continuous system with colored noise is taken as an example. It is proved that the deviation compensated least square method based on modulation function method has better noise suppression effect on identifying the parameters of linear continuous system model. (4) the proposed method is applied to the hydraulic AGC pressure closed loop of a cold rolling mill. 2 / 2 / 2 / 2 dynamic parameter model identification of the system, Compared with the test results, the practicability of the proposed method is verified. The results show that both discrete model identification and continuous model identification have good anti-interference ability and wide applicability.
【学位授予单位】：长沙理工大学
【学位级别】：硕士
【学位授予年份】：2011
【分类号】：TH113

【相似文献】