C.W. Chan

Performance evaluation of UKF-based nonlinear filtering

The performance of the modified unscented Kalman filter (UKF) for nonlinear stochastic discrete-time system with linear measurement equation is investigated. It is proved that under certain conditions, the estimation error of the UKF remains bounded. Furthermore, it is shown that the design of noise covariance matrix plays an important role in improving the stability of the algorithm. Error behavior of the UKF is then derived in terms of mean square error (MSE), and the Cramer-Rao lower bound (CRLB) is introduced as a performance measure. The modified UKF is found to approach the CRLB if the difference between the real noise covariance matrix and the selected one is small enough. These results are verified by using Monte Carlo simulations on two example systems.

Molecular design of luminescent dinuclear gold(I) thiolate complexes: From fundamentals to chemosensing

Abstract A number of dinuclear gold(I) phosphine thiolates have been synthesized and characterized. Detailed spectroscopic and luminescence studies have provided a fundamental understanding on their spectroscopic origins, which serves as the basis for the design of versatile spectrochemical and luminescence chemosensors as well as molecular optoelectronic ‘on-off’ switching devices based on the switching on and off of weak metal⋯metal interactions using the dinuclear gold(I) phosphine thiolate as the basic building block. The binding characteristics have been studied by both UV–vis and emission spectroscopic measurements, and the identities of the ion-bound species have been confirmed by electrospray-ionization mass spectrometric studies.

On the modelling of nonlinear dynamic systems using support vector neural networks

Abstract Though neural networks have the ability to approximate nonlinear functions with arbitrary accuracy, good generalization results are obtained only if the structure of the network is suitably chosen. Therefore, selecting the ‘best’ structure of the neural networks is an important problem. Support vector neural networks (SVNN) are proposed in this paper, which can provide a solution to this problem. The structure of the SVNN is obtained by a constrained minimization for a given error bound similar to that in the support vector regression (SVR). After the structure is selected, its weights are computed by the linear least squares method, as it is a linear-in-weight network. Consequently, in contrast to the SVR, the output of the SVNN is unbiased. It is further shown here that the variance of the modelling error of the SVNN is bounded by the square of the given error bound in selecting its structure, and is smaller than that of the SVR. The performance of the SVNN is illustrated by a simulation example involving a benchmark nonlinear system.

Model-reference adaptive control based on neurofuzzy networks

Model reference adaptive control (MRAC) is a popular approach to control linear systems, as it is relatively simple to implement. However, the performance of the linear MRAC deteriorates rapidly when the system becomes nonlinear. In this paper, a nonlinear MRAC based on neurofuzzy networks is derived. Neurofuzzy networks are chosen not only because they can approximate nonlinear functions with arbitrary accuracy, but also they are compact in their supports, and the weights of the network can be readily updated on-line. The implementation of the neurofuzzy network-based MRAC is discussed, and the local stability of the system controlled by the proposed controller is established. The performance of the neurofuzzy network-based MRAC is illustrated by examples involving both linear and nonlinear systems.

Correspondence: Author's reply to Comments on 'Performance evaluation of UKF-based nonlinear filtering'

It is stated in the above-mentioned comment that the main result of the paper Xiong, Zhang, et al. [(2006). Performance evaluation of UKF-based nonlinear filtering. Automatica 42(2), 261-270] can be extended to a class of filters, such as the extended Kalman filter (EKF). As we show here, this belief can be justified in a rigorous way, even for the nonlinear stochastic system with a nonlinear measurement equation.

NACHWEIS VON KALIUMIONEN DURCH LUMINESZENZ AUF DER BASIS SCHWACHER GOLD-GOLD-WECHSELWIRKUNGEN IN ZWEIKERNIGEN GOLD(I)-KOMPLEXEN

Zwei vielseitig anwendbare Lumineszenzionensonden fur Kaliumionen wurden auf der Grundlage des „Ein- und Ausschaltens” einer Gold-Gold-Wechselwirkung entwickelt. Nach Zugabe von Kaliumionen weisen die Gold(I)-Komplexe [Au2(R2PCH2PR2)(S-benzo[15]krone-5)2] (siehe Bild; R=Ph, Cyclohexyl; M+=K+) eine intensive rote Lumineszenz auf.

Online fault detection and isolation of nonlinear systems

This paper describes an online fault detection scheme for a class of nonlinear dynamic systems with modelling uncertainty and inaccessible states. Only the inputs and outputs of the system can be measured. The faults are assumed to be functions of the state, instead of the output and the input of the system. A nonlinear online approximator using dynamic recurrent neural network is utilised to monitor the faults in the system. The construction and the learning algorithm of the online approximator are presented. The stability, robustness and sensitivity of the fault detection scheme under certain assumptions are analysed. An example demonstrates the efficiency of the proposed fault detection scheme.

Brief Fault detection of systems with redundant sensors using constrained Kohonen networks

The Kohonen self-organizing map (KN) was developed for pattern recognition, and has been extended to fault classification. However, the KN cannot be applied to classify faults from the system output if it contains other factors, such as system state and sensor mounting errors. To overcome this problem, a constrained KN (CKN) is proposed. To eliminate the effect of the system state and the mounting errors, it is proposed that the weight vectors of the CKN are constrained in the parity space. The training algorithm of the CKN is derived, and its convergence discussed. Application of the CKN to fault classification is presented, and its performance is illustrated by an example involving a redundant sensor system with six sensors.

UNSCENTED KALMAN FILTER FOR FAULT DETECTION

Abstract In this paper, the approximation of nonlinear systems using unscented Kalman filter (UKF) is discussed, and the conditions for the convergence of the UKF are derived. The detection of faults from residuals generated by the UKF is presented. As fault detection often reduced to detecting irregularities in the residuals, such as the mean, the local approach, a powerful statistical technique to detect such changes, is used to detect fault from the residuals generated from the UKF. The properties of the proposed method are also presented. To illustrate the performance of the proposed method, it is applied to detect faults in the attitude sensors of a satellite.

Fault estimation for a class of nonlinear dynamical systems

Model based fault estimation for a class of nonlinear dynamical systems is investigated. The state of the system is assumed unavailable, and a nonlinear observer is used to estimate the state. In the observer, a neurofuzzy network is used as the approximator to estimate faults. The network is trained online and the convergence of the proposed learning algorithm is established. Abrupt faults and incipient faults are analyzed in the paper and they can be estimated accurately using a neurofuzzy network with the proposed learning algorithm.