Model referenced monitoring and diagnosis — application to the manufacturing system

This paper describes “model referenced monitoring and diagnosis” a systematic method of monitoring and diagnosis. In this method, a model is used either to generate standard values for the monitoring parameters, or to derive a systematic diagnostic algorithm. As examples of the model referenced monitoring in manufacturing systems, the cutting torque prediction system and tool breakage detection system are described. In the former example, the cutting torque values at each point in time, which can be used as reference values in monitoring the abnormal cutting condition, are evaluated based on a solid modelling system. In the latter, an autoregressive (AR) model is adaptively fitted to the cutting torque signal in order to detect any sudden change in the cutting state due to tool breakage. Two examples are also described for the case of model referenced diagnosis; the diagnosis of sequentially controlled machines using the state graph model, and diagnosis by means of a failure causality model. The former method is applicable to machines controlled by sequence control. Based on the state graph of the machine and the controller, the diagnostic programme can be generated in combination with the control programme. The failure causality model represents the propagation of the effects of failures in the machine. All possible combinations of failure causes are obtained by solving the simultaneous Boolean equations derived from the model.     

Adaptive sequential segmentation of piecewise stationary time series

The problem of adaptive segmentation of time series with abrupt changes in the spectral characteristics is addressed. Such time series have been encountered in various fields of time series analysis such as speech processing, biomedical signal processing, image analysis and failure detection. Mathematically, these time series often can be modeled by zero mean gaussian distributed autoregressive (AR) processes, where the parameters of the process, including the gain factor, remain constant for certain time intervals and then jump abruptly to new values. Identification of such processes requires adaptive segmentation: the times of parameter jumps have to be estimated thoroughly to constitute boundaries of “homogeneous” segments which can be described by stationary AR processes. In this paper, a new effective method for sequential adaptive segmentation is proposed, which is based on parallel application of two sequential parameter estimation procedures. The detection of a parameter change as well as the estimation of the accurate position of a segment boundary is effectively performed by a sequence of suitable generalized likelihood ratio (GLR) tests. Flow charts as well as a block diagram of the algorithm are presented. The adjustment of the three control parameters of the procedure (the AR model order, a threshold for the GLR test and the length of a “test window”) is discussed with respect to various performance features. The results of simulation experiments are presented which demonstrate the good detection properties of the algorithm and in particular an excellent ability to allocate the segment boundaries even within a sequence of short segments. As an application to biomedical signals, the analysis of human electroencephalograms (EEG) is considered and an example is shown.     

Visual Tracking For References Generated By A Stochastic Model

This paper describes a visual tracking system for an unknown reference signal. A time‐varying reference signal is realized as a random process generated by an auto‐regressive (AR) model, which is identifed by a recursive algorithm. Based on the obtained AR model, the future value of reference signal is predicted. We propose a new visual tracking system using generalized minimum variance control (GMVC) and illustrate its properties through experiments.     

A blind separation method of overlapped multi-components based on time varying AR model

A method utilizing single channel recordings to blindly separate the multicomponents overlapped in time and frequency domains is proposed in this paper. Based on the time varying AR model, the instantaneous frequency and amplitude of each signal component are estimated respectively, thus the signal component separation is achieved. By using prolate spheroidal sequence as basis functions to expand the time varying parameters of the AR model, the method turns the problem of linear time varying parameters estimation to a linear time invariant parameter estimation problem, then the parameters are estimated by a recursive algorithm. The computation of this method is simple, and no prior knowledge of the signals is needed. Simulation results demonstrate validity and excellent performance of this method.     

基于SVD-AR模型与VPMCD的轴承故障诊断方法

针对强噪声背景下振动信号故障特征难以提取的问题,提出了基于奇异值分解的自回归(SVD-AR)模型,用于提取振动信号的特征,并与变量预测模型模式识别(VPMCD)方法相结合应用于轴承故障诊断.对轴承振动信号进行SVD;然后,利用奇异值差分谱对分量信号进行筛选,对能够反映故障信息的分量信号建立AR模型,提取轴承振动信号的特征信息;采用VPMCD对滚动轴承运行状态进行识别.实验证明了方法的合理性和有效性.     

基于AR(1)模型的阵列传感器信号融合算法研究

本文将时间序列理论成功应用于阵列传感器信息融合和模式识别中,用AR(1)模型针对三个样本对不同传感器的响应信号进行建模,用Durbin-Levinson方法估计出AR(1)模型的自回归参数 ,依据不同的样本数据得到不同的模型参数,不同的参数即融合了不同样本的特征信息.实验结果表明该方法有效的解决了工程实际问题,对时间序列理论在信息融合和模式识别中的应用有一定的参考价值.     

Signal segmentation using changing regression models with application in seismic engineering

The change detection and segmentation methods have gained considerable attention in scientific research and appear to be the central issue in various application areas. The objective of the paper is to present a segmentation method, based on maximum a posteriori probability (MAP) estimator, with application in seismic signal processing; some interpretations and connections with other approaches in change detection and segmentation, as well as computational aspects in this field are also discussed. The experimental results obtained by Monte Carlo simulations for signal segmentation using different signal models, including models with changes in the mean, in FIR, AR and ARX model parameters, as well as comparisons with other methods, are presented and the effectiveness of the proposed approach is proved. Finally, we discuss an application of segmentation in the analysis of the earthquake records during the Kocaeli seism, Turkey, August 1999, Arcelik station (ARC). The optimal segmentation results are compared with time–frequency analysis, for the reduced interference distribution (RID). The analysis results confirm the efficiency of the segmentation approach used, the change instants resulted by MAP appearing clear in energy and frequency contents of time–frequency distribution.     

卫星姿控系统基于模糊基函数网络与自回归模型的故障预测

针对卫星姿态控制系统的故障预测问题,给出了模糊基函数网络(FBFN)与自回归模型(AR)相结合的故障预测方法,并提出了预测置信因子的概念,对故障预测的准确性进行评价.首先利用卫星正常运行时的姿态数据训练FBFN,将训练好的FBFN作为卫星姿控系统的标准输出模型;然后把卫星实时姿态数据与FBFN输出数据之间的差值作为残差...     

基于时间序列分析的压气机喘振检测

喘振检测对于保障压气机安全运转具有重要意义。以压气机出口总压作为喘振检测特征信号,采用时间序列分析方法建立了AR模型。多步预测的误差和滞后现象显示AR模型难以预测突发性的喘振现象。在喘振发生时,残差方差急剧增大,模型参数变化显著,设定合适的门限可及时可靠地检测出喘振信号。     

On the use of the dual-frequency ENVISAT altimeter to determine snowpack properties of the Antarctic ice sheet

The primary purpose of ice-sheet altimetry is to monitor the changes in ice-sheet topography which may impact on global sea-level. However, the altimetric signal is sensitive to different properties of the snowpack, and therefore can also be used to determine these properties. The radar altimeter onboard the European Space Agency's ENVISAT satellite provides a dual-frequency dataset at Ku (13.6 GHz) and S band (3.2 GHz). In this paper, these signals are studied over the Antarctic ice-sheet during the 4 first years of the mission (2002-2006), in order to retrieve snowpack properties.The altimeter signal can be described by 4 classical waveform parameters. The 4 year time-series of all these parameters are decomposed into a linear and a seasonal time component. The linear component is almost constant. The distribution of the mean parameters over the Antarctic ice-sheet shows that the altimeter signal is sensitive to small-scale (mm) surface roughness.For the first time, the amplitudes and phases of the seasonal variations are characterized. The S band amplitudes are greater than the Ku band, and the phase varies over the entire ice-sheet. Previous studies suggested that the seasonal variations of the altitude from the altimeter are created by a decrease of the snowpack height through compaction. The dual-frequency observations shown here suggest that this hypothesis is too simple. Instead, the altitude variations observed in the altimetric signal are not created by the snowpack height change, but are more likely caused by the seasonal change of the snow properties, which cause a different response between the S and Ku bands. Therefore, both the linear and the seasonal variations of the altimetric signal can be used to retrieve snowpack properties.Here, we compare the dual-frequency ENVISAT signal with a model of the altimetric echo over the Antarctic ice-sheet. The model combines a surface model with a sub-surface model, for both the S and Ku bands. The Brown model [Brown G. S. (1977). The average impulse response of a rough surface and its applications. IEEE Transactions on Antennas and Propagation, 25, 1.] is used to describe the interaction of the radar wave with the snow surface. The backscatter coefficient of the surface is derived using the IEM method [Fung, A. K. (1994). Microwave scattering and emission models and their applications, Boston, MA: Artech House.]. The sub-surface signal takes into account both the layering effects and the scattering caused by the homogeneous media which is composed of small snow grains. The model is tested in two areas of the Antarctic plateau which present very different waveform parameters. The sensitivity of the radar signal to the different snowpack properties is investigated. The analysis of the waveform behaviours shows that the sub-surface signal can be completely masked by the small-scale surface roughness signal.Finally, the temperature and surface density effects are investigated in order to explain the seasonal variations of the altimetric signal. Both the temperature and the compaction rate of the snow change seasonally. Temperature is shown to impact on the Ku band signal. Furthermore, the compaction rate of the snow surface can explain all of the seasonal variation characteristics observed at both the S and Ku bands. The seasonal change of compaction rate in the snow creates a change in the waveform shape that can bias the altitude. In particular, the snow compaction can induce a bias in the retrieved altimetric altitude of more than 80 cm for the Ku band and 1.5 m for the S band. This work underlines that the altitude time-series needs to be corrected for the shape of the altimetric echo over ice-sheets.     

基于AR模型和神经网络的舰船水压信号检测方法

为了有效地从风浪背景中检测舰船水压场信号，根据舰船水压场信号和波浪噪声信号的差异，以时间序列的AR模型理论为依据，采用基于AR模型和神经网络的舰船水压信号检测方法。该检测算法的核心是将检测问题转化为模式识别问题，首先对接收信号建立AR模型并提取AR模型系数作为特征向量，然后利用人工神经网络对信号进行检测。在此基础上，通过不同浪级情况下海浪水压力场的仿真信号数据，对某型目标舰船的水压力信号进行了检测计算，验证了该方法的有效性，尤其是达到了在高海况、低信噪比条件下，对目标信号检测率比较高、虚警率比较低的效果。     

卡尔曼滤波在心电信号QRS波检测中的尝试

心电信号QRS波的检测方法很多，但在准确性与实时性方面都不太好，该文中将心电信号按照QRS波周期进行分割，利用RLS算法的自适应AR建模，为心电信号建立模型，再利用kalman滤波算法对心电信号进行滤波和预测，在保证R波探测率的同时提高了探测的速度。针对心率不齐或者QRS波周期产生波动的情况，程序中利用各QRS波周期的相似性，求其互相关，以确定周期T，同时对T进行自适应建模，以便对下一周期预测。经过试验，取得了比较好的效果。     

基于高阶统计量的心音信号分析

对高阶统计量方法应用于心音信号分析进行了研究。建立了心音信号的AR双谱模型,获得了心音的双谱幅度图,并采用模型参数作为特征参量对心音信号进行了二类模式识别。实验结果表明,高阶统计量在心音信号分析和处理中具有较大的应用潜力。     

舰船水压信号的预测方法研究

提出了一种能从海浪水压信号背景下提取舰船水压信号的预测异常检测(PAD)法。模型预测值与测量值相比较所得的差值被作为检测舰船水压信号是否存在的判据。讨论了作为PAD中预测模型的线性的自回归(AR)模型和非线性的神经网络(NN)模型，并用模拟数据和实测数据对二者进行了比较。仿真结果表明，PAD效果良好，预测模型中，NN模型要优于AR模型。     

基于神经网络重力固体潮信号的建模与预测

重力固体潮信号主要是由于太阳、月亮等天体轨道相对位置变化而产生的,同时受地质、水文、大气等地理条件变化的影响,所以既是一个有规律、周期性变化的信号,也包含反映地质、水文、大气等地理条件变化的异常信息。通过对重力固体潮信号的建模,可反映、预测重力固体潮信号中周期性变化的基本规律,通过对比其理论计算值,可进一步提取重力固体潮信号中的异常变化信息。基于一种具有强鲁棒性、纯随机搜索的新群体智能优化算法,改进径向基神经网络学习算法,避免学习算法进入局部最优,提高网络训练的有效性和所建网络模型的可靠性。在实验中,利用重力固体潮信号训练改进的径向基神经网络,得到了重力固体潮信号的有效径向基神经网络模型。利用上述模型预测重力固体潮信号的估计值,并与传统径向基神经网络模型、AR模型预测结果进行对比,表明改进训练算法的径向基网络模型预测的结果更加精确,说明改进训练算法在重力固体潮信号的径向基网络建模中是有效的,可推广应用于其它时间信号序列的建模与预测中。     

FIR方法在维纳滤波器设计中的应用与仿真

基于最小均方误差准则,得到维纳-霍夫方程,并利用FIR(有限冲击响应)方法求解,进而得到维纳滤波器的传递函数。在MatLab环境下,基于一阶AR(自回归)模型生成原始信号,对维纳滤波器进行设计和仿真,并分析抽样点数、AR模型参数、信噪比和滤波器阶数对滤波效果的影响。仿真结果表明,增大抽样点数和信噪比以及减小AR模型参数,滤波效果增强;增大滤波器阶数,滤波效果先增强后减弱。     

基于AR模型的MEMS陀螺误差补偿方法研究

针对MEMS(Microelectro Mechanical Systems）陀螺具有成本低、体积小但误差较大的问题,探讨MEMS陀螺的误差补偿方法。基于AR模型方法,采集MEMS陀螺原始信号,对原始信号进行预处理,利用预处理后的数据建立陀螺的AR(Auto Regressive)模型,辨识出模型参数。利用该模型对陀螺信号进行误差补偿,计算出陀螺的较精确值。通过对某MEMS陀螺误差补偿的静态和动态试验表明,提出的方法能够有效地减小误差,提高陀螺的测量精度。     

分层向量自回归的多通道脑电信号的特征提取研究

有效的特征提取方法能提高脑机接口（Brain-computer interface,BCI）系统对脑电（Electroencephalogram,EEG）信号的识别率.因脑电信号都是多通道的,本文将分层向量自回归（Hierarchical vector autoregression,HVAR）模型用于脑电信号的特征提取,并结合传统的线性支持向量机（Support vector machine,SVM）用于脑电信号识别.该模型不仅克服了自回归（Autoregression,AR）模型只能用来提取单通道特征的局限性,而且不再采用传统VAR（Vector autoregression）模型所有通道共用一个时滞的处理方法.创新之处在于在传统的VAR模型基础上添加正则化思想,有效地压缩参数空间,实现合理的分层结构.本文首次将HVAR模型用于由Keirn等采集并整理的脑电数据中.实验结果证明HVAR模型在阶数较小的情况下（2阶）与阶数较大（6阶）的AR模型效果相当,可见低阶的HVAR能很好地刻画脑电信号的时空关联关系,这说明HVAR可能是刻画EEG信号的一种新颖的方法,这对其他多通道时间序列分析都有借鉴意义.     

A smoothness priors time-varying AR coefficient modeling of nonstationary covariance time series

A smoothness priors time varying AR coefficient model approach for the modeling of nonstationary in the covariance time series is shown. Smoothness priors in the form of a difference equation constraint excited by an independent white noise are imposed on each AR coefficient. The unknown white noise variances are hyperparameters of the AR coefficient distribution. The critical computation is of the likelihood of the hyperparameters of the Bayesian model. This computation is facilitated by a state-space representation Kalman filter implementation. The best difference equation order-best AR model order-best hyperparameter model locally in time is selected using the minimum AIC method. Also, an instantaneous spectral density is defined in terms of the instantaneous AR model coefficients and a smoothed estimate of the instantaneous time series variance. An earthquake record is analyzed. The changing spectral analysis of the original data and of simulations from a time varying AR coefficient model of that data are shown.