Nonlinear Systems Identification via an Input-Output Model Based on a Feedforward Neural Network

This paper develops a feedforward neural network based input output model for a general unknown nonlinear dynamic system identification when only the inputs and outputs are accessible observations. In the developed model, the size of the input space is directly related to the system order. By monitoring the identification error characteristic curve, we are able to determine the system order and subsequently an appropriate network structure for systems identification. Simulation results are promising and show that generic nonlinear systems can be identified, different cases of the same system can also be discriminated by our model.

Identification and PID Control for a Class of Delay Fractional-order Systems

In this paper, a new model identification method is developed for a class of delay fractional-order system based on the process step response. Four characteristic functions are defined to characterize the features of the normalized fractional-order model. Based on the time scaling technology, two identification schemes are proposed for parameters U+02BC estimation. The scheme one utilizes three exact points on the step response of the process to calculate model parameters directly. The other scheme employs optimal searching method to adjust the fractional order for the best model identification. The proposed two identification schemes are both applicable to any stable complex process, such as higher-order, under-damped U+002F over-damped, and minimum-phase U+002F nonminimum-phase processes. Furthermore, an optimal PID tuning method is proposed for the delay fractional-order systems. The requirements on the stability margins and the negative feedback are cast as real part constraints U+0028 RPC U+0029 and imaginary part constraints U+0028 IPC U+0029. The constraints are implemented by trigonometric inequalities on the phase variable, and the optimal PID controller is obtained by the minimization of the integral of time absolute error U+0028 ITAE U+0029 index. Identification and control of a Titanium billet heating process is given for the illustration. © 2014 Chinese Association of Automation.

Dynamic Stability Analysis of Linear Time-varying Systems via an Extended Modal Identification Approach

The problem of linear time-varying（LTV） system modal analysis is considered based on time-dependent state space representations, as classical modal analysis of linear time-invariant systems and current LTV system modal analysis under the ＂frozen-time＂ assumption are not able to determine the dynamic stability of LTV systems. Time-dependent state space representations of LTV systems are first introduced, and the corresponding modal analysis theories are subsequently presented via a stabilitypreserving state transformation. The time-varying modes of LTV systems are extended in terms of uniqueness, and are further interpreted to determine the system＇s stability. An extended modal identification is proposed to estimate the time-varying modes, consisting of the estimation of the state transition matrix via a subspace-based method and the extraction of the time-varying modes by the QR decomposition. The proposed approach is numerically validated by three numerical cases, and is experimentally validated by a coupled moving-mass simply supported beam exper- imental case. The proposed approach is capable of accurately estimating the time-varying modes, and provides anew way to determine the dynamic stability of LTV systems by using the estimated time-varying modes.

SYSTEM IDENTIFICATION OF ADAPTIVE TRUSS STRUCTURES USING PIEZOELECTRIC ACTIVE MEMBERS

Adaptive truss structures are a new kind of structures with integrated active members,whose dynamic characteristies can be beneficially modified to meet mission requirements.Active members containing actuating and sensing units are the major components of adaptive truss structures.Modeling of adaptive truss structures is a key step to analyze the structural dynamic characteristics.A new experimental modal analysis approach,in which active members are used as excitatiDn sources for modal test,has been proposed in this paper.The excitation forces generated by the active members, which are different from the excitation forces exerted on structures in the conventional modal test,are internal forces for the truss structures.The relation between internal excitation forces and external forces is revealed such that the traditional identification method can be adopted to obtain modal parameters of adaptive structures.Placement problem of the active member in adaptive truss structures is also discussed in this work. Modal test and analysis are conducted with a planar adaptive truss structure by using piezoelectric active members in order to verify the feasibility and effectiveness of the proposed method.

Identification of Hammerstein Model Using Hybrid Neural Networks

The identification problem of Hammerstein model with extension to the multi input multi output (MIMO) case is studied. The proposed identification method uses a hybrid neural network (HNN) which consists of a multi layer feed forward neural network (MFNN) in cascade with a linear neural network (LNN). A unified back propagation (BP) algorithm is proposed to estimate the weights and the biases of the MFNN and the LNN simultaneously. Numerical examples are provided to show the efficiency of the proposed method.

FMLSMETHODOFPARAMETERESTIMATIONANDITSAPPLICATIONTOHEAT┐EXCHANGERIDENTIFICATION

Using the inversion of the auto correlation function Toeplitz matrix of pseudo random binary sequence (PRBS) derived in this paper and the theorem of partitioned matrix inversion, a fast multistage least squares (FMLS) method is developed. Its performances are theoretically analyzed and digital simulation is made to compare FMLS with multistage least squares (MSLS), correlation least squares(COR LS) and LS for their computer speed and identification accuracy. Finally, FMLS is applied to identifying the heat excharger dynamics. It is shown that FMLS is a good and effective identification technique.

Continuous-time System Identification with Nuclear Norm Minimization and GPMF-based Subspace Method

To improve the accuracy and effectiveness of continuous-time (CT) system identification, this paper introduces a novel method that incorporates the nuclear norm minimization (NNM) with the generalized Poisson moment functional (GPMF) based subspace method. The GPMF algorithm provides a simple linear mapping for subspace identification without the timederivatives of the input and output measurements to avoid amplification of measurement noise, and the NNM is a heuristic convex relaxation of the rank minimization. The Hankel matrix with minimized nuclear norm is used to determine the model order and to avoid the over-parameterization in subspace identification method (SIM). Furthermore, the algorithm to solve the NNM problem in CT case is also deduced with alternating direction methods of multipliers (ADMM). Lastly, two numerical examples are presented to evaluate the performance of the proposed method and to show the advantages of the proposed method over the existing methods. © 2014 Chinese Association of Automation.

An Iterative Learning Approach to Identify Fractional Order KiBaM Model

This paper discusses the parameter and differentiation order identification of continuous fractional order KiBaM models in ARX (autoregressive model with exogenous inputs) and OE (output error model) forms. The least squares method is applied to the identification of nonlinear and linear parameters, in which the Grünwald-Letnikov definition and short memory principle are applied to compute the fractional order derivatives. An adaptive P-type order learning law is proposed to estimate the differentiation order iteratively and accurately. Particularly, a unique estimation result and a fast convergence speed can be arrived by using the small gain strategy, which is unidirectional and has certain advantages than some state-of-art methods. The proposed strategy can be successfully applied to the nonlinear systems with quasi-linear characteristics. The numerical simulations are shown to validate the concepts. © 2017 Chinese Association of Automation.

Effective Methods and Performance Analysis on Data Transmission Security with Blind Source Separation in Space-Based AIS

In space-based Automatic Identification Systems(AIS), due to high satellite orbits, several Ad Hoc cells within the observation range of the satellite are vulnerable to interference by an external signal.To increase efficiency in target detection and improve system security, a blind source separation method is adopted for processing the conflicting signals received by satellites. Compared to traditional methods, we formulate the separation problem as a clustering problem. Since our algorithm is affected by the sparseness of source signals, to get satisfactory results, our algorithm assumes that the distance between two arbitrary mixed-signal vectors is less than the doubled sum of variances of distribution of the corresponding mixtures. Signal sparsity is overcome by computing the Short-Time Fourier Transform, and the mixed source signals are separated using the improved PSO clustering. We evaluated the performance and the robustness of the proposed network architecture by several simulations. The experimental results demonstrate the effectiveness of the proposed method in not only improving satellite signal receiving ability but also in enhancing space-based AIS security.

On-Line Real Time Realization and Application of Adaptive Fuzzy Inference Neural Network

In this paper, a modeling algorithm developed by transferring the adaptive fuzzy inference neural network into an on-line real time algorithm, combining the algorithm with conventional system identification method and applying them to separate identification of nonlinear multi-variable systems is introduced and discussed.

Clothing identification via deep learning:forensic applications

Attribute-based identification systems are essential for forensic investigations because they help in identifying individuals.An item such as clothing is a visual attribute because it can usually be used to describe people.The method proposed in this article aims to identify people based on the visual information derived from their attire.Deep learning is used to train the computer to classify images based on clothing content.We first demonstrate clothing classification using a large scale dataset,where the proposed model performs relatively poorly.Then,we use clothing classification on a dataset containing popular logos and famous brand images.The results show that the model correctly classifies most of the test images with a success rate that is higher than 70%.Finally,we evaluate clothing classification using footage from surveillance cameras.The system performs well on this dataset,labelling 70%of the test images correctly.

Subspace-based identification of discrete time-delay system

We investigate the identification problems of a class of linear stochastic time-delay systems with unknown delayed states in this study. A time-delay system is expressed as a delay differential equation with a single delay in the state vector. We first derive an equivalent linear time-invariant(LTI) system for the time-delay system using a state augmentation technique. Then a conventional subspace identification method is used to estimate augmented system matrices and Kalman state sequences up to a similarity transformation. To obtain a state-space model for the time-delay system, an alternate convex search(ACS) algorithm is presented to find a similarity transformation that takes the identified augmented system back to a form so that the time-delay system can be recovered. Finally, we reconstruct the Kalman state sequences based on the similarity transformation. The time-delay system matrices under the same state-space basis can be recovered from the Kalman state sequences and input-output data by solving two least squares problems. Numerical examples are to show the effectiveness of the proposed method.

Application of a Standard Procedure to Avoid Errors When Comparing Fingerprints with Their Reversals in Fake Documents

Laterally reversed(mirrored)fingerprints are difficult to detect by applying routine search procedures.One suggestion to avoid errors when dealing with probable reversals is to perform comparisons with both direct and reversed fingerprints.This simple procedure has been applied and led to the detection of two more cases of reversed fingerprint usage in fake documents.In one of the reported cases,experts found on the web the same fingerprints used by criminals in fake documents.This finding is important because it indicates that matched fingerprints do not necessarily link different criminal cases.