仪器非线性特性的在线递推估计

本文介绍了递推最小二乘法在仪器非线性特性估计中的应用，与非递推理最小二北法相比递推最小二乘法具有计算简单、占用内存少，能实现在线估计的特点，特别适合于由微处理器构成的数字化仪器。     

基于前向神经网络的时变非线性结构系统辨识快速递推最小二乘法

针对基于前向神经网络的普通递推最小二乘估计存在着自适应跟踪慢和精度低的问题，提出了一种可对非线性时变系统进行快速辨识的新方法，因该方法有类似递推最小二乘算法的形式，称其为基于前向神经网络的快速递推最小二乘算法。该算法对传统的递推最小二乘算法的递推方式进行了改变，以更好的跟踪非线性时变系统的动态特性。针对典型的系统辨识仿真算例，通过与现有常用方法的比较研究显示了这种算法具有计算简单、收敛速度快和辨识精度高的良好性能。最后将方法用于一个三自由度时变非线性振动系统，结果同样验证了方法的良好特性。     

基于前向神经网络的非线性时变系统辨识的改进递推最小二乘算法

标准的递推最小二乘算法随着递推次数的增加,增益矩阵将逐渐趋于零,致使递推算法慢慢失去修正能力,出现所谓的“数据饱和”现象。为了克服“数据饱和”问题,本文首先对递推最小二乘算法进行改进,得到了改进的最小二乘算法(IRLS),并给出了收敛性证明,然后将该算法应用于基于前向神经网络的非线性时变系统辨识。通过对两个非线性时变系统进行有效验证,仿真结果表明本文算法计算精度高、计算速度快、数值稳定性好,并能有效克服“数据饱和”。     

聚丙烯熔融指数的最小二乘建模与自适应预报

针对工业双环管丙烯聚合装置熔融指数(MI)在线测量的困难,进行了聚丙烯熔融指数的在线最小二乘建模与自适应预报方法的研究。首先结合聚合反应机理分析,以Hammerstein模型描述聚丙烯熔融指数的动态特性,其次,利用渐消记忆递推最小二乘法在线更新Hammerstein模型参数。在此基础上,通过降维观测器的设计提出并实现了工业双环管聚丙烯装置的一种熔融指数的在线最小二乘建模与自适应预报方法。最后,利用工业数据验证了此算法的可行性和有效性。     

最小二乘估计的方差和不确定度的应用

本文讨论了用最小二乘法估计的回归系数和方差的无偏性，以及用最小二乘法得到的测试结果和不确定度的表露，并举例示之。     

递推最小二乘法在角位置误差检定中的应用

全组合方法能够把测角系统的角位置误差和棱体的工作角偏差有效地分离出来。在满足角位置误差测量精度的情况下,为了提高测试效率,将递推最小二乘法应用于组合测角中。本文以23面棱体与391齿盘的偏差检定为例,设计了递推最小二乘法的检测软件,实现了每增加一个量测序列,就可以显示新的工作角偏差向量,新的测试的残余误差与计算的随机误差的标准差,提高了测试效率。此方法不仅可以用于棱体一齿盘的检定中,而且可以用于圆感应同步器、光栅等测角系统中。     

基于自适应递推LSSVM的硅锰合金成分在线预测

针对硅锰合金埋弧熔炼过程的特点,提出了一种基于自适应递推最小二乘支持向量机(ARLSSVM)的合金成分在线预测模型.该模型以实测工况参数为数据集,当新增一个样本时,分别采用增长记忆递推算法、限定记忆递推算法和缩减记忆递推算法训练最小二乘支持向量机( LSSVM),有效避免高维矩阵的求逆,加快模型更新的速度.然后通过自适...     

精确测定低对称晶系多昌材料点阵常数的X射线衍射方法

点阵常数是多晶材料的重要物理参数之一，陶瓷材料多数属非立方晶系化合物，研究低对称晶系点阵常数的测定方法和提高精度的途径对材料研究有重要的意义，本文提出了联立方程法、联立方程外推法、最小二乘法、最小二乘外推法、线对法、线对最小二乘法等六种测定方法和稳定性判 法、抛弃 平均法、最小二乘判法等三种数据处理方法及其适用条件、用本文的测定低对称晶系多晶材料点阵常数的方法可以得到较精确可靠的结果。     

单静止站纯方位目标航向估计

单静止站纯方位被动声纳系统是不完全可观测系统,关于该类系统的既往研究表明,对于匀速直线运动的目标,利用目标方位角信息可以估计出目标的航向,这具有重要的实际应用意义.通过对目标运动几何模型分析,推导得到目标方位角与目标航向的关系公式,并提出依据几何关系的目标航向估计算法.通过进行坐标变换引入最小二乘估计思想,提高了估计方法的有效性.使用Monte-Carlo计算机仿真测试了算法在不同目标方位角和不同数据长度下的性能,并与拟线性递推法和递推最小二乘法进行了比较,结果表明该算法具有更高的估计精度和更好的适应性.     

基于MATLAB仿真的配网故障定位研究

本文采用基于模型参数识别原理的递推最小二乘法对小电流接地系统单相接地故障进行定位。利用MATLAB建立故障定位模型,并仿真经不同电阻接地。对故障时故障相母线电压和故障线路电流进行采样作为定位算法程序的输入数据。利用电压和电流数据求解算法下电感L的估计值,从而计算出故障点到保护安装处的距离。由于篇幅有限,本文将直接给出直接解参数方程及最小二乘法两种算法的仿真结果,通过比较来确定最小二乘法在故障定位中更具有可行性。     

A novel intelligent fuzzy estimator for uniform load determination on a beam with varying boundary conditions

A novel intelligent fuzzy input estimation method that estimates the uniform input load on a beam structural system is presented in this article. The uniform input load acting on a beam structural system is estimated from the measured dynamic responses using the inverse method. The algorithm includes the fuzzy Kalman filter and the fuzzy-weighted recursive least square method. This study presents an efficient estimator accelerated and weighted using fuzzy-accelerating and -weighting factors proposed based on the fuzzy logic inference system. The capability of the proposed algorithm is demonstrated through several beam structural system examples with different types of boundary conditions. The simulation results are compared by alternating between the constant, adaptive, and fuzzy weighting factors. The results demonstrate that the presented method applied to beams with various structural system boundary conditions is successful.     

时变参数识别方法研究

本文重点研究了现有各种时变参数识别方法的内在联系及各自特点。在引起统一的模型结构以后，利用随机逼近原理，提出了各种在线识别算法的一般递推形式。然后基于变物理参数微分方程模型，推导出了不同辨识算法各自的递推形式。用六种方法对参数缓变，快变及突变三种系统作了计算机仿真识别，以比较各种的特点。研究结果对时变参数识别具有理论和应用价值。     

用考虑置信区间长度影响的最小二乘法拟合S－N曲线

S－N曲线是用名义应力法预测结构疲劳寿命的基础。基于试验疲劳寿命分散的物理机制,提出了一个S－N曲线试验数据处理的加权最小二乘法,此方法中每组数据的权重与该组数据的置信区间成反比。计算结果表明用考虑置信区间长度影响的最小二乘法拟合得到的S－N曲线比用一般最小二乘法得到的S－N曲线具有更高的可靠度,给出的寿命预测结果也更安全。     

Study on Prediction Method of Structural Three-parameter Power Function P-S-N Curve

This paper applies weighted least square method to estimate the three-parameter power function equation of the fatigue life curve,and uses comprehensive fatigue life coefficient to correct the equation,and at the same time combines probability statistics method to bring out the prediction method of structure's three-parameter power function P-S-N curve,finally applies the prediction method to a ship's frame-type elevate,based on the fatigue test data of it's material-5A06 aluminium alloy,to obtain it's structure's three-parameter power function P-S-N curve. Compared with the conventional least square method,the presented method can give a rational estimation due to it's a best linear unbiased estimator,and this method has stronger practicality in engineering,and can be used in many structures of different materials,for achieving their three-parameter power function P-S-N curve.     

移动最小二乘支持向量机

基于移动最小二乘逐点逼近思想,移动权被引入到最小二乘支持向量机的误差变量中,得到新算法的模型.此外,证明了用移动最小二乘支持向量机作函数估计与在特征空间中用移动最小二乘法得到的解是一致的,揭示了移动最小二乘支持向量机所选择的核函数相当于移动最小二乘法所选择基函数组.数值试验与实例进一步验证所提出方法的优越性.     

Reliability analysis of exponentiated Poisson-exponential constant stress accelerated life test model

Accelerated life testing is an efficient tool frequently adopted for obtaining failure time data of test units in a lesser time period as compared to normal use conditions. We assume that the lifetime data of a product at constant level of stress follows an exponentiated Poisson-exponential distribution and the shape parameter of the model has a log-linear relationship with the stress level. Model parameters, the reliability function (RF), and the mean time to failure (MTTF) function under use conditions are estimated based on eight frequentist methods of estimation, namely, method of maximum likelihood, method of least square and weighted least square, method of maximum product of spacing, method of minimum spacing absolute-log distance, method of Cramér-von-Mises, method of Anderson–Darling, and Right-tail Anderson–Darling. The performance of the different estimation methods is evaluated in terms of their mean relative estimate and mean squared error using small and large sample sizes through a Monte Carlo simulation study. Finally, two accelerated life test data sets are considered and bootstrap confidence intervals for the unknown parameters, predicted shape parameter, predicted RF, and the MTTF at different stress levels, are obtained.     

Finite elements formulated by the weighted discrete least squares method

By use of the least squares error criterion, an alternate finite element formulation is presented. The method is based on the discrete or element-wise minimization of square and weighted differential equation residuals which are expressed in terms of element nodal quantities. In order to overcome the stringent inter-element continuity requirement, a major stumbling block, on the element trial functions two practical schemes are proposed. One is the reduction of the original governing differential equation to a system of equivalent first order differential equations; the other is a method of smoothing discontinous trial functions. The latter essentially relaxes the continuity requirement and yields efficient non-conforming finite elements. This paper also demonstrates the use of constant weights which significantly improves the rates of convergence. Several numerical examples illustrate the proposed method. From these examples, it may be concluded that the use of constant weights and the relaxation of the inter-element continuity requirement are two indispensable features of the weighted discrete least square method.     

Prediction of Time Series Empowered with a Novel SREKRLS Algorithm

For the unforced dynamical non-linear statespace model, a new Q1 and efficient square root extended kernel recursive least square estimation algorithm is developed in this article. The proposed algorithm lends itself towards the parallel implementation as in the FPGA systems. With the help of an ortho-normal triangularization method, which relies on numerically stable givens rotation, matrix inversion causes a computational burden, is reduced. Matrix computation possesses many excellent numerical properties such as singularity, symmetry, skew symmetry, and triangularity is achieved by using this algorithm. The proposed method is validated for the prediction of stationary and non-stationary MackeyGlass Time Series, along with that a component in the x-direction of the Lorenz Times Series is also predicted to illustrate its usefulness. By the learning curves regarding mean square error (MSE) are witnessed for demonstration with prediction performance of the proposed algorithm from where it’s concluded that the proposed algorithm performs better than EKRLS. This new SREKRLS based design positively offers an innovative era towards non-linear systolic arrays, which is efficient in developing very-large-scale integration (VLSI) applications with non-linear input data. Multiple experiments are carried out to validate the reliability, effectiveness, and applicability of the proposed algorithm and with different noise levels compared to the Extended kernel recursive least-squares (EKRLS) algorithm.     

基于模糊规则的非线性系统快速建模方法

描述了一个通过学习从输入输出采样数据中提取MISO模糊规则的具有插值性能的新型模糊系统。提出的模糊模型继承了Sugeno模型及其变化形式的许多优点。对于新的模糊模型采用递推最小二乘估计方法代替启发式误差反馈学习方法辨识结论参数，整个辨识过程所需的计算时间相对于其他方法最短，适用于在线辨识。仿真结果表明了该方法的有效性与实用性。