Orthogonal-Least-Squares Forward Selection for Parsimonious Modelling from Data

The objective of modelling from data is not that the model simply fits the training data well. Rather, the goodness of a model is characterized by its generalization capability, interpretability and ease for knowledge extraction. All these desired properties depend crucially on the ability to construct appropriate parsimonious models by the modelling process, and a basic principle in practical nonlinear data modelling is the parsimonious principle of ensuring the smallest possible model that explains the training data. There exists a vast amount of works in the area of sparse modelling, and a widely adopted approach is based on the linear-in-the-parameters data modelling that include the radial basis function network, the neurofuzzy network and all the sparse kernel modelling techniques. A well tested strategy for parsimonious modelling from data is the orthogonal least squares (OLS) algorithm for forward selection modelling, which is capable of constructing sparse models that generalise well. This contribution continues this theme and provides a unified framework for sparse modelling from data that includes regression and classification, which belong to supervised learning, and probability density function estimation, which is an unsupervised learning problem. The OLS forward selection method based on the leave-one-out test criteria is presented within this unified data-modelling framework. Examples from regression, classification and density estimation applications are used to illustrate the effectiveness of this generic parsimonious modelling approach from data.

Design of Radial Basis Function Network Using Adaptive Particle Swarm Optimization and Orthogonal Least Squares

This paper presents a two-level learning method for designing an optimal Radial Basis Function Network (RBFN) using Adaptive Velocity Update Relaxation Particle Swarm Optimization algorithm (AVURPSO) and Orthogonal Least Squares algorithm (OLS) called as OLS-AVURPSO method. The novelty is to develop an AVURPSO algorithm to form the hybrid OLS-AVURPSO method for designing an optimal RBFN. The proposed method at the upper level finds the global optimum of the spread factor parameter using AVURPSO while at the lower level automatically constructs the RBFN using OLS algorithm. Simulation results confirm that the RBFN is superior to Multilayered Perceptron Network (MLPN) in terms of network size and computing time. To demonstrate the effectiveness of proposed OLS-AVURPSO in the design of RBFN, the Mackey-Glass Chaotic Time-Series as an example is modeled by both MLPN and RBFN.

基于OLS的径向基函数神经网络实现多种数字信号调制方式自动识别

基于决策论的信号调制样式自动识别方法具有简单易行、适合在线分析的优点,针对一些参数的计算进行了改进,并提出了基于该方法,利用正交最小二乘法(OLS)的径向基函数(RBF)神经网络,实现数字信号调制样式自动识别的方法.提高了该方法的识别能力,对信噪比(SNR)为6～30dB的测试信号识别得到了较好的结果.识别的数字信号为2ASK、4ASK、2PSK、4PSK(QP-SK)、2FSK、4FSK与16QAM.

基于OLS与EPSO算法的RBF企业订单预测模型研究

提出了一种最小正交二乘算法(OLS)和进化粒子群优化算法(EPSO)相结合构建RBF神经网络的企业订单预测模型。OLS采用前向回归算法,从输入数据中选取适当的中心,动态地避免网络规模过大和随机选择中心带来的数值病态问题;EPSO方法调整网络中的参数,如RBF中心位置,RBF宽度和隐层与输出层之间的权值,以提高网络的泛化能力。

改进的OLS算法选择RBFNN中心的方法

提出了一种优化选择径向基神经网络数据中心的算法,该算法结合了Kohonen网络的模式分类能力,将初步分类结果用做RBFNN的初始数据中心,然后采用OLS算法进行优化选择,对比仿真实验表明该算法效果比单独使用OLS算法生成的RBFNN性能更好。

基于OLS-RBF神经网络的指挥信息系统效能评估

针对指挥信息系统评估体系中存在的不准确、不完善的问题,提出了一种基于OLS-RBF神经网络的指挥信息系统的评估方法。利用RBF神经网络结构简单、收敛速度快、逼近精度高的优点,同时弱化了人为因素对评估过程的影响,使评估模型结构更加合理,评估结果更加准确。仿真结果表明,与其他方法相比,基于RBF神经网络的作战指挥信息系统模型结果的误差更小,与真实值更加接近。此外,RBF神经网络还可以广泛应用于其他模型的预测,能收到较好的效果。

基于Kohonen网络和OLS算法的RBFNN中心选择方法

提出了一种优化选择径向基神经网络数据中心的算法,该算法结合了Kohonen网络的模式分类能力,将初步分类结果用作RBFNN的初始数据中心,然后采用OLS算法进行优化,对比仿真实验表明该算法效果比单独使用OLS算法生成的RBFNN性能更好。

Cancellation for frequency offset in OFDM system based on TF-LMS algorithm

In an orthogonal frequency division multiplexing(OFDM) system,a time and frequency domain least mean square algorithm(TF-LMS) was proposed to cancel the frequency offset(FO).TF-LMS algorithm is composed of two stages.Firstly,time domain least mean square(TD-LMS) scheme was selected to pre-cancel the frequency offset in the time domain,and then the interference induced by residual frequency offset was eliminated by the frequency domain mean square(FD-LMS) scheme in frequency domain.The results of bit error rate(BER) and quadrature phase shift keying(QPSK) constellation figures show that the performance of the proposed suppression algorithm is excellent.

Wavelet Neural Networks for Adaptive Equalization by Using the Orthogonal Least Square Algorithm

Equalizers are widely used in digital communication systems for corrupted or time varying channels. To overcome performance decline for noisy and nonlinear channels, many kinds of neural network models have been used in nonlinear equalization. In this paper, we propose a new nonlinear channel equalization, which is structured by wavelet neural networks. The orthogonal least square algorithm is applied to update the weighting matrix of wavelet networks to form a more compact wavelet basis unit, thus obtaining good equalization performance. The experimental results show that performance of the proposed equalizer based on wavelet networks can significantly improve the neural modeling accuracy and outperform conventional neural network equalization in signal to noise ratio and channel non-linearity.

A nonlinear PCA algorithm based on RBF neural networks

Traditional PCA is a linear method, but most engineering problems are nonlinear. Using the linear PCA in nonlinear problems may bring distorted and misleading results. Therefore, an approach of nonlinear principal component analysis (NLPCA) using radial basis function (RBF) neural network is developed in this paper. The orthogonal least squares (OLS) algorithm is used to train the RBF neural network. This method improves the training speed and prevents it from being trapped in local optimization. Results of two experiments show that this NLPCA method can effectively capture nonlinear correlation of nonlinear complex data, and improve the precision of the classification and the prediction.

LMMSE-based SAGE channel estimation and data detection joint algorithm for MIMO-OFDM system

A new channel estimation and data detection joint algorithm is proposed for multi-input multi-output （MIMO） - orthogonal frequency division multiplexing （OFDM） system using linear minimum mean square error （LMMSE）- based space-alternating generalized expectation-maximization （SAGE） algorithm. In the proposed algorithm, every sub-frame of the MIMO-OFDM system is divided into some OFDM sub-blocks and the LMMSE-based SAGE algorithm in each sub-block is used. At the head of each sub-flame, we insert training symbols which are used in the initial estimation at the beginning. Channel estimation of the previous sub-block is applied to the initial estimation in the current sub-block by the maximum-likelihood （ML） detection to update channel estimatjon and data detection by iteration until converge. Then all the sub-blocks can be finished in turn. Simulation results show that the proposed algorithm can improve the bit error rate （BER） performance.

REQUIRED NUMBER OF ITERATIONS FOR SPARSE SIGNAL RECOVERY VIA ORTHOGONAL LEAST SQUARES

In countless applications,we need to reconstruct a K-sparse signal x∈R n from noisy measurements y=Φx+v,whereΦ∈R^(m×n)is a sensing matrix and v∈R m is a noise vector.Orthogonal least squares(OLS),which selects at each step the column that results in the most significant decrease in the residual power,is one of the most popular sparse recovery algorithms.In this paper,we investigate the number of iterations required for recovering x with the OLS algorithm.We show that OLS provides a stable reconstruction of all K-sparse signals x in[2.8K]iterations provided thatΦsatisfies the restricted isometry property(RIP).Our result provides a better recovery bound and fewer number of required iterations than those proposed by Foucart in 2013.

Blind cancellation for frequency offset in OFDM system based on MCMA-RLS algorithm

Modified constant modulus and recursive least squares （MCMA-RLS） algorithm is proposed to cancel interference caused by the variable frequency offset （FO） in the orthogonal frequency division multiplexing （OFDM） system. The MCMA-RLS algorithm is composed of two stages including MCMA scheme and RLS scheme. MCMA is selected to pre-cancel the variable frequency offset firstly, and then the residual interference has been canceled by the RLS scheme. BR error rate is simulated to demonstrate that the proposed method is robust for canceling the variable frequency offset.

An MMSE Decoding Algorithm without Matrix Inversion in QSTBC

The matrix inversion operation is needed in the MMSE decoding algorithm of orthogonal space-time block coding （OSTBC） proposed by Papadias and Foschini. In this paper, an minimum mean square error （MMSE） decoding algorithm without matrix inversion is proposed, by which the computational complexity can be reduced directly but the decoding performance is not affected.

不同容量下并网模式交流微电网短路故障早期检测与区域定位

随着分布式电源(distributed generation,DG)的容量变化,微电网原有的供电结构发生改变,使得潮流大小、方向和功率结构发生变化,对快速检测和定位微电网中的短路故障区域提出了挑战。在MATLAB/Simulink中搭建低压交流微电网模型;通过高尺度小波能量谱算法对微电网与大电网公共连接点(point of common coupling,PCC)处检测到的电流进行分解,提取适应不同容量情况的短路故障特征值,实现了不同容量下微电网短路故障的早期检测;利用小波能量谱特征结合基于正交最小二乘法(orthogonal least square,OLS)的径向基函数(radial basis function,RBF)神经网络算法提出一种适用于不同容量微电网的短路故障区域定位方法,并进行仿真验证;在此基础上设计并网模式微电网短路故障保护硬件系统,并进行实验验证。结果表明,所设计的保护系统能够快速、准确地同时实现并网模式下交流微电网短路故障的早期检测与区域定位。

权重化QR分解的正交匹配追踪算法硬件实现

为在小型化、低成本的硬件平台实现正交匹配追踪(Orthogonal Matching Pursuit,OMP)算法,针对OMP算法中最小二乘法的问题,该文构造一个确定性的传感矩阵,提出一种低复杂度、低资源的权重化QR分解的OMP(Weighted QR decomposition OMP,WQR-OMP)算法硬件结构,在ZYNQ 7020型号芯片上搭建WQR-OMP SOC系统.WQR-OMP算法在传感矩阵进行QR分解后,根据三角矩阵R中元素的分布特性,通过权重化运算只保留主对角线上的元素而其他余元素归零,得到对角矩阵D,然后近似计算稀疏向量的解.实验结果表明:与基于QR分解的OMP(QR decomposition OMP,QR-OMP)和Batch-OMP算法的硬件结构相比,WQR-OMP算法硬件结构的重构速度更快、存储资源更少.在压缩率为0.25的条件下,WQR-OMP SOC系统对256×256分辨率图像的重构时间为400 ms左右,其速率比仅使用ARM处理器的重构速率提高了约6.3倍.与其他现有研究对比,该系统在Block RAM存储资源消耗较少的情况下,进一步提升了重构速度,适用于存储资源受限的硬件平台.

Randomized Algorithms for Orthogonal Nonnegative Matrix Factorization

Orthogonal nonnegative matrix factorization(ONMF)is widely used in blind image separation problem,document classification,and human face recognition.The model of ONMF can be efficiently solved by the alternating direction method of multipliers and hierarchical alternating least squares method.When the given matrix is huge,the cost of computation and communication is too high.Therefore,ONMF becomes challenging in the large-scale setting.The random projection is an efficient method of dimensionality reduction.In this paper,we apply the random projection to ONMF and propose two randomized algorithms.Numerical experiments show that our proposed algorithms perform well on both simulated and real data.

基于正交最小二乘的声源高分辨识别定位方法研究

针对小孔径平面阵下的低频相干源定位分辨率低的问题,提出一种基于正交最小二乘(OLS)的协方差拟合定位法。首先,将OLS运用到协方差拟合成本函数中;然后,利用OLS能在迭代识别源时能逐渐减少点扩散函数(PSF)分量影响的特点,在每次迭代后,利用当前源信息重新校正先前识别源,并通过指数参数提高声源选择标准以缩减各声源主瓣宽度提高空间分辨率。仿真和试验表明,所提方法显著降低了声源间距固定的双相干声源可分辨的最低频率。

基于参数自修正的配电网故障定位数字孪生技术研究

配电网参数受天气条件和负载条件等因素影响会发生变化。由于传感装置安装有限、数据延时传输等因素,无法实时获得配电网准确参数,进而给传统故障定位方法的精度带来影响。针对以上问题,通过建立配电网数字孪生模型,基于配电网数字孪生模型的参数自修正技术,提出了一种定位模型随参数变化动态校正的配电网故障定位方法。同时,搭建了基于数字孪生服务器和实时数字仿真系统(real time digital system, RTDS)的数字孪生平台,实现了配电网实时的物理模型和数字孪生模型的同步运行。在算例仿真中,利用该数字孪生平台,验证了基于数字孪生技术的配电网故障定法方法。结果表明,该方法可在各类系统运行条件下实时修正配电网参数,显著提高配电网故障定位的速度和精度。

基于RBF神经网络的光纤陀螺启动补偿及应用

光纤陀螺对温度比较敏感,由于温度引起的零偏漂移是光纤陀螺工作尤其是启动过程中的一种较大误差。文中为了减小光纤陀螺启动过程的零偏漂移、缩短启动时间,提出了对光纤陀螺启动过程进行补偿的方案。该方案以光纤陀螺温度和温度变化率为输入、光纤陀螺漂移为输出建立二输入单输出的RBF神经网络,用于陀螺启动过程补偿。在室温下对某型号光纤陀螺启动漂移进行了补偿,试验结果表明该方法能有效减小陀螺的启动温度漂移,缩短陀螺启动时间。将该方案运用到某型号的光纤陀螺寻北仪上,常温试验表明,该方案大大缩短了寻北仪的准备时间,提高了寻北精度。