变遗忘因子递推最小二乘Hammerstein系统辨识算法

摘.要:本论文研究了单输入单输出非线性Hammerstein系统的辨识问题,提出了一种具有变遗忘因子的递推最小二乘算法.由于Hammerstein系统模型的非线性特征,传统的递推最小二乘算法无法直接用来解决该系统的辨识问题.为此,论文将Hammerstein系统参数进行了映射变换,使得变换后的系统参数与Hammerst...     

A Multi-innovation Recursive Least Squares Algorithm with a Forgetting Factor for Hammerstein CAR Systems with Backlash

This study addresses the identification of Hammerstein CAR systems with backlash, where the nonlinear backlash is described as one regression identification model using a two switching function mathematical model. In such a case, the Hammerstein CAR systems with backlash can be transformed into a piecewise linearized model. Then, a novel multi-innovation recursive least squares algorithm with a forgetting factor is applied to estimate the parameters of the proposed model. Finally, numerical examples are presented to test the performance of the proposed algorithm.     

最少元件的多输入多输出MOCCII电流模式滤波器

本文提出了两种基于MOCCⅡ（多端输出的第二代电流传输器）的多输入多输出的电流模式滤波器。两种电路均由2个MOCCⅡ及4个接地RC元件构成。每一种电路除了实现出单输出的低通、带通、高通、带组、全通电流模式滤波器外，还能实现三种不同类型的具有同时多输出的电流模式滤波器，提出的电路具有很低的无源灵敏度；同时应用基本电流镜技术实现出结构简单的高精度CMOS MOCCⅡ，并对MOCCⅡ及提出的滤流器电路进行了PSICE仿真。     

Least Squares-Based Iterative Identification Methods for Linear-in-Parameters Systems Using the Decomposition Technique

By extending the least squares-based iterative (LSI) method, this paper presents a decomposition-based LSI (D-LSI) algorithm for identifying linear-in-parameters systems and an interval-varying D-LSI algorithm for handling the identification problems of missing-data systems. The basic idea is to apply the hierarchical identification principle to decompose the original system into two fictitious sub-systems and then to derive new iterative algorithms to estimate the parameters of each sub-system. Compared with the LSI algorithm and the interval-varying LSI algorithm, the decomposition-based iterative algorithms have less computational load. The numerical simulation results demonstrate that the proposed algorithms work quite well.     

一种基于最小二乘法的经典辩识的新方法

系统辩识方法分为经典和现代两大类.经典辩识方法采用非参数模型,其表现形式是以时间或频率为自变量的实验曲线.在对以时间或频率为自变量的实验曲线或者实验数据进行处理时,由于传统方法本身固有的缺陷,导致辩识过程的随意性较大,不能利用所有数据,从而造成不能充分利用所测数据的全部信息,不但浪费了信息,而且辩识出来的结果准确度很差.本文提出了一种用最小二乘法解决上述问题的方法,实践证明具有较高的精度,取得了较好的辩识效果.     

Maximum Likelihood Recursive Least Squares Estimation for Multivariable Systems

This paper discusses parameter estimation problems of the multivariable systems described by input–output difference equations. We decompose a multivariable system to several subsystems according to the number of the outputs. Based on the maximum likelihood principle, a maximum likelihood-based recursive least squares algorithm is derived to estimate the parameters of each subsystem. Finally, two numerical examples are provided to verify the effectiveness of the proposed algorithm.     

基于最小二乘的时差定位算法

时差测量方程是非线性双曲线方程,可以通过引入中间变量将其转化为线性方程,对近年来国内外学者关于这种方法的时差定位算法进行了总结。当已知测量误差的先验信息时,可以采用两步加权最小二乘法和约束加权最小二乘法,当测量误差的先验信息未知时,还可以采用约束总体最小二乘的方法。在求解约束最小二乘问题时,采用常规的拉格朗日法计算复杂、运算量大,而采用高斯一牛顿法不仅可以大为降低运算量,还能提高解的精度和稳定性。此外,对约束加权最小二乘法和约束总体最小二乘法之间的关系进行了探讨,得到了它们等价性的条件。     

Fuzzy Least Squares for Identification of Individual Pharmacokinetic Parameters

In this paper, we examined the added value of fuzzy nonlinear regression to identify individual pharmacokinetic parameters in the case of noisy fuzzy data and/or small sample sizes. We first described three approaches that use least squares of errors as a fitting criterion for parameter estimation by fuzzy regression. Next, we compared the estimation and prediction capability of fuzzy least squares (FLS) and ordinary least squares (OLS) regressions via a simulation experiment, so as to determine the conditions of data size and variability under which one approach could be deemed superior over the other. We considered two empirical pharmacokinetic models. Our results showed that OLS regression outperformed FLS regression when the sample size was larger and/or there existed more outliers in the data. Overall, FLS regression was more powerful as the dataset size decreased. When the data were smaller in size and contained more variability, FLS regression's performance remained better than that of the OLS regression. Although the accuracy of the three FLS regression approaches was very close in almost all instances, those that estimated fuzzy parameters were superior in terms of predictive capability. These findings could aid in selecting the proper regression technique to employ in the presence of fuzzy data.      

A Least Squares Solution for Use in the Six-Port Measurement Technique

Although based on the use of simple amplitude detectors, it is possible to obtain complex values of reflection coefficient, via the six-port technique, from the intersection of three circles in the complex plane. In a typical case, the circle centers are detemined primarily by the six-port design and are nominally constant, while the radii are proportional to the square root of the ratio of the output of three of the detectors to a fourth one. As a practical matter, however, these circles will not intersect in a point because of noise or other errors in the detectors.This paper develops a presedure for choosing Gamma in this context. Moreover, the question of what may be inferred about the system performance from the extent of this intersection failue is briefly considered.     

Convergence Analysis of the Hierarchical Least Squares Algorithm for Bilinear-in-Parameter Systems

This paper studies the convergence of the hierarchical identification algorithm for bilinear-in-parameter systems. By replacing the unknown variables in the information vector with their estimates, a hierarchical least squares algorithm is derived based on the model decomposition. The proposed algorithm has higher computational efficiency than the over-parameterization model-based recursive least squares algorithm. The performance analysis shows that the parameter estimation errors converge to zero under persistent excitation conditions. The effectiveness of the proposed algorithm is verified by simulation examples.     

Recursive Least Squares Parameter Estimation for a Class of Output Nonlinear Systems Based on the Model Decomposition

In this paper, we study the parameter estimation problem of a class of output nonlinear systems and propose a recursive least squares (RLS) algorithm for estimating the parameters of the nonlinear systems based on the model decomposition. The proposed algorithm has lower computational cost than the existing over-parameterization model-based RLS algorithm. The simulation results indicate that the proposed algorithm can effectively estimate the parameters of the nonlinear systems.     

Consensus Based Least Squares Estimation for Single-Integrator Multi-Agent Systems with a Time-Varying Reference State

This paper discusses consensus tracking of single-integrator multi-agent systems with a time varying reference state based on the least squares estimation theory to deal with the case of communications disturbance. Since none of the followers can communicate with the leader within the accepted time due to communications disturbance, least squares estimation is applied for online estimation of the time varying reference state and reducing the tracking errors occurred due to communications disturbance. A theoretical proof is provided under some assumptions while the necessary and sufficient conditions are demonstrated for which consensus is reached on the time-varying reference state in case of communications disturbance. Finally, a Matlab example is given to validate the effectiveness of the proposed approach.     

基于改进的稀疏最小二乘双子支撑向量回归的数字预失真技术

为了补偿大容量卫星通信射频前端的功率放大器的非线性,传统的数字预失真(DPD)模型需要更多的系数和更高的阶次,严重影响预失真前馈路径的资源消耗。为了解决这一问题,该文提出一种基于改进的稀疏最小二乘双子支撑向量回归(ISLSTSVR)的低复杂度DPD方法。首先通过构建原空间的决策函数解决最小二乘双子支撑向量回归(LSTSVR)模型解不稀疏的问题;同时引用截断最小二乘损失函数增加模型的鲁棒性;然后采用Nystrom逼近方法得到核矩阵的低秩近似,进一步采用Cholesky分解降低核矩阵的运算复杂度;最后由低秩的核矩阵求得模型稀疏解。实验选用基于单管氮化镓(GaN)器件的宽带AB类功率放大器,以40 MHz的32QAM信号进行激励。预失真实验表明,该方法能在保证模型精度的情况下大幅减少DPD模型系数和计算复杂度,为星载射频前端的预失真技术提供了有效的系数降维思路和方法。     

哈默斯坦非线性时变系统的加权学习辨识方法

针对有限区间哈默斯坦(Hammerstein)非线性时变系统,该文提出一种加权迭代学习算法用以估计系统时变参数。首先将Hammerstein系统输入非线性部分进行多项式展开,采用迭代学习最小二乘算法辨识系统的时变参数。为了防止数据饱和,采用带遗忘因子的迭代学习最小二乘算法,进而引入权矩阵,采用加权迭代学习最小二乘算法改进系统跟踪误差,以提高辨识精度。该文分别给出3种算法的推导过程并进行仿真验证。结果表明,与迭代学习最小二乘算法和带遗忘因子迭代学习最小二乘算法相比,加权迭代学习最小二乘算法具有辨识精度高、跟踪误差小以及迭代次数少等优点。     

加权最小二乘法和MUSIC法联合测向技术研究

针对传统相位干涉仪测向法精度不高和MUSIC算法空间谱峰搜索耗时长的问题,提出了一种基于加权最小二乘法和MUSIC算法相结合的均匀圆阵测向技术。首先根据阵列短基线组求解相位模糊,并通过引入中间变量将相位差测量方程转化为线性方程组,然后运用加权最小二乘法对其进行求解,并利用信号到达角与中间变量的关系得到信号到达角初始估计,最后MUSIC算法依据信号到达角初始估计进行空间谱峰搜索,得到高精度信号到达角估计。仿真实验对比了所提方法与现有相位干涉仪及MUSIC算法的角度估计精度和耗时,证实了所提方法的正确性和有效性。     

基于总体最小二乘的红外图像去噪

针对红外图像存在的加性、乘性及混合噪声,采用从图像中截取图像块,再用图像块的线性结合对原图像进行去噪,总体最小二乘算法用来求解其中的系数向量,充分考虑了噪声图像中存在的不确定性,通过这组系数得到去噪后的红外图像.在对像素点空间关系权重的求解上,采用模糊核聚类算法将红外图像粗略进行聚类,归为同一类的像素点之间存在较强空间约束关系,否则认为它们之间存在较弱空间约束.通过与维纳滤波算法比较,仿真结果证明了总体最小二乘去噪算法在红外图像的视觉质量和信噪比改善两个方面的有效性.最后通过比较无噪红外图像与去噪红外图像的直方图表明总体最小二乘去噪算法的优越性.     

Networked Predictive Control Systems Based on the Hammerstein Model

In this paper, a novel predictive control-based approach is proposed for a networked control system with random delays containing an input nonlinear process based on a Hammerstein model. The method uses a time-delay two-step generalized predictive control scheme, which consists of two parts: one is to deal with the input nonlinearity of the Hammerstein model and the other is to compensate for the network-induced delay in the networked control system. A theoretical result using the Popov criterion is presented for the closed-loop stability of the system in the case of a constant delay. Simulation examples illustrating the validity of the approach are also presented.     

基于最小二乘法的吸咀吸取压力的研究

表面贴装技术生产线中的电子元件实装机通过吸咀内的负压将电子元件吸附,从而实现对电子元器件的拾取和贴装。为便于准确掌握吸咀取料是否成功,通过将负压传感器检测到的压力变化与PMAC板卡上的A/D转换值进行对应并将其量化,对得到的数据利用最小二乘法的原理在MATLAB软件环境下进行多项式拟合,得到最优的数据曲线和数据结果。     

总体最小二乘拟合的盖氏圆盘信源数估计法

针对调整因子设置不当造成在信源数估计时盖氏圆盘法（GDE）性能下降的问题,提出一种基于总体最小二乘拟合的盖氏圆盘法（TLS-GDE）。该方法以圆盘半径作为拟合点进行直线拟合,若拟合点含信号圆盘半径,则拟合偏差较大,利用这一特性制定了比值阶跃准则,进行信源数判决,解决了盖氏圆盘法对调整因子依赖的问题。最后,仿真结果表明,该方法鲁棒性较好,在低信噪比下性能要优于GDE算法,更具有实用价值。