COMPARISON STUDY OF SUBSPACE IDENTIFICATION METHODS APPLIED TO FLEXIBLE STRUCTURES

A novel nonlinear modelling approach has been developed and implemented on Alstom gasifier using Wiener model. The linear element of the Wiener model was identified by a combined subspace state space method, which integrated MOESP (Multivariable Output-Error State Space) and N4SID (Numerical algorithms for subspace state space system identification) method in the estimation of system matrices. Then a single layer neural network was chosen as the nonlinearity of the model. The proposed model identification method was used to model Alstom gasifier with strong nonlinearity and multivariable couples. The results compared to a combined linear subspace identification method demonstrate that the nonlinear method proposed in this paper behave better approximation.

Summary form only given. In this paper, the authors use the numerical algorithm for subspace state space system identification (N4SID) to extract dynamic parameters from phasor measurements collected on the western North American Power Grid. The data were obtained during tests on June 7, 2000, and they represent wide area response to several kinds of probing signals including low-level pseudo-random noise (LLPRN) and single-mode square wave (SMSW) injected at the Celilo terminal of the Pacific HVDC Intertie (PDCI). An identified model is validated using a cross validation method. Also, the obtained electromechanical modes are compared with the results from Prony analysis of a ringdown and with signal analysis of ambient data measured under similar operating conditions. The consistent results show that methods in this class can be highly effective even when the probing signal is small.

This paper presents a method for the identification of multiple-input-multiple-output (MIMO) Hammerstein systems for the goal of prediction. The method extends the numerical algorithms for subspace state space system identification (N4SID), mainly by rewriting the oblique projection in the N4SID algorithm as a set of componentwise least squares support vector machines (LS-SVMs) regression problems. The linear model and static nonlinearities follow from a low-rank approximation of a matrix obtained from this regression problem.

System identification based structural damage aggravation detection in a large masonry building

The performance of machines and equipment degrades as a result of aging and wear. This decreases performance reliability and increases the potential for faults and failures. To ensure proper functionality of complex systems, advanced technologies for performance diagnosis and control are being incorporated into engineering designs, which requires an ever-increasing number of sensors and measurement devices. Nevertheless, a sensor, just as any other dynamic system, degrades and fails. A faulty sensor may cause process performance degradation, process shut down, or even a fatal accident because it is no longer able to deliver accurate information about the monitored system. Therefore, it is essential to assess sensor performance to ensure system reliability. In this paper, a method is proposed to detect, isolate, and compensate sensor degradation. The numerical algorithm for subspace state space system identification is used to track the changes of the time constants and gains of the sensor and the monitored system. Without imposing requirements for redundant sensors and measurement devices, this method utilizes the fact that sensor readings depict dynamic characteristics of the sensors as well as those of the monitored system. The newly proposed method is verified in angular sensor degradation detection using high-fidelity simulations of an automotive electronic throttle system.

Modal parameter identification of a multiple-span post-tensioned concrete bridge using hybrid vibration testing data

This work is placed on the scope of characterization and modeling of structures and the central objective is to establish a computer code capable to provide models for structures of interest from practical experimentation, to be used to refine the computer models a posteriori. To realize that project, the experimental modal analysis was applied in the Eigensystem Realization Algorithm – ERA. As well as most of the techniques of identification of dynamic of systems, the ERA also creates the compute modes, being necessary the application of procedures for filtering the real modes before the noises and/or computers modes. In this work, two techniques for the distinction were used: Modal Amplitude Coherence (MAC) and Modal Singular Value (MSV). Beside this, a novel approach to help interpret and filter based in MSV deviation was proposed and tested, producing good results. To achieve this, a computer code that receives experimental data and returns the global properties of the system, i.e., the natural frequencies, the damping factors and the mode shapes was developed. The preliminary results, developed in the know structure, indicate the code has the potential to handle with complex geometries.

A comparative study of the MAX and SUM machine-learning algorithms using virtual fuzzy sets

This paper introduces a new adaptive algorithm in system identification. This adaptive algorithm is a variation of greedy algorithm.

Eigensystem Realization Algorithm (ERA), Stochastic Subspace Identification (SSI), Continuous Wavelet Transform (CWT), and Enhanced Frequency Domain Decomposition (EFDD) are four widely used damping identification methods. Their performance remains unclear in previous reviews and comparative discussions. This uncertainty can be attributed to three critical factors: the varying behavior of different methods under different excitations, the lack of a clear benchmark for evaluating accuracy, and the unquantified influence of parameter tuning on results. This study revisits and evaluates these methods under controlled conditions and addresses the key challenges that hinder reliable damping identification. A major challenge identified is the sensitivity of the results to parameter settings, which significantly impacts the stability and accuracy of the identification. Based on the evaluation, recommended methods and corresponding parameter guidelines are provided for three common excitation scenarios: impulse, white noise, and earthquakes. This study offers practical guidance for the selection and application of damping identification methods in structural dynamics.

A computational-fluid-dynamics-based aeroelastic analysis method is proposed to simulate control-surface limit-cycle oscillation (LCO) induced by freeplay gap. The present method is based on the previously proposed aeroelastic reduced-order model (ROM), in which an unsteady aerodynamic state-space model is generated from aerodynamic responses to step excitation of individual mode using the eigensystem realization algorithm, and connected to a structural dynamic state-space model within the MATLAB/SIMULINK environment. The aeroelastic ROM is extended to treat structural nonlinearity due to the control-surface freeplay by generating an additional feedback line of generalized residual forces in the SIMULINK model. To reduce the problem size and the computation time, a fictitious-mass modal approach is used, which can afford the possible local change of stiffness. The present method is first validated for its capability to simulate aeroelastic responses of a regional-jet horizontal-tail (HT) model without freeplay, and shows good prediction of flutter characteristics measured in the transonic flutter wind-tunnel testing. The present method is then applied to the HT model with the elevator freeplay gaps. Although there are some differences between the computed and measured limit-cycle-oscillation ranges, the present method well predicts the nonlinear limit-cycle-oscillation behavior in its amplitude and frequency measured in the test, and shows promise toward accurate prediction of control-surface LCO due to its freeplay gap.

In this paper, the inverted pendulum is controlled by Quanser universal benchtop equipment-direct current (QUBE DC) motor, and the motor inverted pendulum system is identified when the sinusoidal signal is input. However, there is a large error when the angle of the pendulum is pre-identified by numerical algorithms for subspace state space system identification algorithm. Therefore, this paper designs a multi strategy optimization simulated annealing particle swarm optimization algorithm, which can accurately identify the complex system with sinusoidal input signal. In the identification experiments of sinusoidal signals with multiple frequencies and amplitudes, this paper found that the algorithm performs relatively best at a frequency of 8 rad/s. Moreover, at a frequency of 8 rad/s, the algorithm can quickly reduce the error to 1.69% within 100 generations. Finally, based on the hardware model identified by particle swarm optimization, this paper designs an explicit model predictive control method to control the inverted pendulum, and tests the constraint processing and anti-disturbance performance of the system under the voltage pulse interference with different duty cycles, which realizes the inverted pendulum balance and has robustness under the interference.

This dissertation presents a new method for the statistical performance analysis of subspace system identification algorithms. A perturbation expansion of the singular value decomposition is used to approximate the effects of process and measurement noise on the identified system poles. Eigenvalue variance results are obtained that do not depend on asymptotic statistics and are therefore applicable for short data records. These performance measures are conditional upon the specific input and the parameters of the actual system being identified. This method is demonstrated on two variations of the Numerical Subspace State-Space System Identification (N4SID) algorithm. The accuracy of these theoretical performance measures is validated using simulation. Finally, this new method is applied to the first step of three well known subspace identification algorithms to determine the effects of row and column weighting on the variance of identified poles.

A updating method using strain frequency response function with emphasis on local structure

An algorithm to estimate multiple-input, multiple-output, time-series models for structural system identification using short time histories is presented. The method uses order overspecification to reduce estimation bias and automatically sorts identified system modes from computational modes. The system natural frequencies, damping factors, mode shapes, and modal participation factors are estimated and can be used to reconstruct singleinput, single-output models for every input-output pair. Experimental data from a large cantilevered flexible truss are used to demonstrate the technique. Modal estimates are comparable to those obtained using accepted methods such as the eigensystem realization algorithm.