A Data-Based Augmented Model Identification Method for Linear Errors-in-Variables Systems Based on EM Algorithm

Guo, Fan; Wu, Ouyang; Ding, Yongsheng; Huang, Biao

doi:10.1109/tie.2017.2703680

Cited by 19 publications

(7 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The performance of EM algorithm for BDLM heavily relies on the initialization. 40 In general, deterministic and stochastic strategies are usually adopted to determine the initial value of EM approach. 38 The deterministic strategy selects a set of established values as starting points but fails to try other initial values when it is trapped to the undesired solution.…”

Section: Framework For Online Anomaly Detection With Bdlmmentioning

confidence: 99%

Anomaly detection of structural health monitoring data using the maximum likelihood estimation-based Bayesian dynamic linear model

Zhang

Wang

Wan

et al. 2020

Structural Health Monitoring

View full text Add to dashboard Cite

Enormous data are continuously collected by the structural health monitoring system of civil infrastructures. The structural health monitoring data inevitably involve anomalies caused by sensors, transmission errors, or abnormal structural behaviors. It is important to identify the anomalies and find their origin (e.g. sensor fault or structural damage) to make correct interventions. Moreover, online anomaly identification of the structural health monitoring data is critical for timely structural condition assessment and decision-making. This study proposes an online approach for detecting anomalies of the structural health monitoring data based on the Bayesian dynamic linear model. In particular, Bayesian dynamic linear model, consisting of various components, is implemented to characterize the feature of real-time measurements. Expectation maximization algorithm and Kalman smoother are combined to estimate the Bayesian dynamic linear model parameters and generate log-likelihood functions. The subspace identification method is introduced to overcome the initialization issue of the expectation maximization algorithm. The log-likelihood difference of consecutive time steps is then used to determine thresholds without introducing extra anomaly detectors. The proposed Bayesian dynamic linear model-based approach is first illustrated by the simulation data and then applied to the structural health monitoring data collected from two long-span bridges. The results indicate that the proposed method exhibits good accuracy and high computational efficiency and also allows for reconstructing the strain measurements to replace anomalies.

show abstract

Section: Framework For Online Anomaly Detection With Bdlmmentioning

confidence: 99%

Anomaly detection of structural health monitoring data using the maximum likelihood estimation-based Bayesian dynamic linear model

Zhang

Wang

Wan

et al. 2020

Structural Health Monitoring

View full text Add to dashboard Cite

show abstract

“…The EM algorithm is an iterative optimization algorithm actually and it is particularly suitable for the issues involve hidden variable or missing data (Liu et al, 2018;Yang et al, 2017cYang et al, , 2018aHuang et al, 2017). Hence, it is widely applied for system identification, process monitoring and so on (Yang et al, 2018c;Zhang et al, 2015;Chen et al, 2013;Xie et al, 2012;Guo et al, 2017). The main idea behind the EM algorithm is to iteratively search for the desired estimates by maximizing the corresponding cost function.…”

Section: Expectation-maximization Algorithm Revisitedmentioning

confidence: 99%

“…SI mainly aims at obtaining a good quality model which can accurately describe the practical process dynamics based on the informative process data set (Yang and Yang, 2018;Ji et al, 2012;Zhang et al, 2018;Yuan et al, 2017;Huang et al, 2018b). Since majority of the practical systems exhibit complicated nonlinear property, the approaches for nonlinear system identification with various types of models have been widely studied in the existing literature (Yang et al, 2018c;Yang and Yin, 2017;Chen et al, 2013;Xie et al, 2012;Guo et al, 2017). Among various kinds of different model structures, the main advantages of the state-space model (SSM) can be concluded as follows: The hidden inner dynamics of the system can be reflected by the latent state, while the latent state variable of the system is reflected by the SSM; The SSM is convenient for the system controller design and suitable for describing the multiple input and multiple output (MIMO) systems (Marconato et al, 2014).…”

Section: Introductionmentioning

confidence: 99%

Identification of nonlinear state-space time-delay system

Liu

Zhang

Peng

et al. 2019

View full text Add to dashboard Cite

Purpose This paper aims to investigate an identification strategy for the nonlinear state-space model (SSM) in the presence of an unknown output time-delay. The equations to estimate the unknown model parameters and output time-delay are derived simultaneously in the proposed strategy. Design/methodology/approach The unknown integer-valued time-delay is processed as a latent variable which is uniformly distributed in a priori known range. The estimations of the unknown time-delay and model parameters are both realized using the Expectation-Maximization (EM) algorithm, which has a good performance in dealing with latent variable issues. Moreover, the particle filter (PF) with an unknown time-delay is introduced to calculated the Q-function of the EM algorithm. Findings Although amounts of effective approaches for nonlinear SSM identification have been developed in the literature, the problem of time-delay is not considered in most of them. The time-delay is commonly existed in industrial scenario and it could cause extra difficulties for industrial process modeling. The problem of unknown output time-delay is considered in this paper, and the validity of the proposed approach is demonstrated through the numerical example and a two-link manipulator system. Originality/value The novel approach to identify the nonlinear SSM in the presence of an unknown output time-delay with EM algorithm is put forward in this work.

show abstract

“…So, if we could use the information regarding the inputs, system identification performance can be improved. Hence, in this work, we also consider an input generation dynamics [24,27] in conjunction with the EIV model. Here, we attribute the following nonlinear model to the input generation dynamics and assume that the structure of the input generation dynamics is known.…”

Section: The Robust Errors-in-variables Modelmentioning

confidence: 99%

“…EIV modelling is one of the approaches to consider the errors present in the measurements of both the input and output variables [19,25,26]. Even though there are various approaches for the linear EIV identification [20][21][22][23][24], none of them have addressed the problem of robust EIV identification in presence of outliers.…”

Section: Introductionmentioning

confidence: 99%

Robust identification for nonlinear errors-in-variables systems using the EM algorithm

Guo

Hariprasad

Huang

et al. 2017

Journal of Process Control

View full text Add to dashboard Cite

A Data-Based Augmented Model Identification Method for Linear Errors-in-Variables Systems Based on EM Algorithm

Cited by 19 publications

References 30 publications

Anomaly detection of structural health monitoring data using the maximum likelihood estimation-based Bayesian dynamic linear model

Anomaly detection of structural health monitoring data using the maximum likelihood estimation-based Bayesian dynamic linear model

Identification of nonlinear state-space time-delay system

Robust identification for nonlinear errors-in-variables systems using the EM algorithm

Contact Info

Product

Resources

About