Showing papers on "Robustness (computer science) published in 1994"

Recurrent neural networks and robust time series prediction

Abstract: We propose a robust learning algorithm and apply it to recurrent neural networks. This algorithm is based on filtering outliers from the data and then estimating parameters from the filtered data. The filtering removes outliers from both the target function and the inputs of the neural network. The filtering is soft in that some outliers are neither completely rejected nor accepted. To show the need for robust recurrent networks, we compare the predictive ability of least squares estimated recurrent networks on synthetic data and on the Puget Power Electric Demand time series. These investigations result in a class of recurrent neural networks, NARMA(p,q), which show advantages over feedforward neural networks for time series with a moving average component. Conventional least squares methods of fitting NARMA(p,q) neural network models are shown to suffer a lack of robustness towards outliers. This sensitivity to outliers is demonstrated on both the synthetic and real data sets. Filtering the Puget Power Electric Demand time series is shown to automatically remove the outliers due to holidays. Neural networks trained on filtered data are then shown to give better predictions than neural networks trained on unfiltered time series. >

A robust MIMO terminal sliding mode control scheme for rigid robotic manipulators

Abstract: In this paper, a robust multi-input/multi-output (MIMO) terminal sliding mode control technique is developed for n-link rigid robotic manipulators. It is shown that an MIMO terminal switching plane variable vector is first defined, and the relationship between the terminal switching plane variable vector and system error dynamics is established. By using the MIMO terminal sliding mode technique and a few structural properties of rigid robotic manipulators, a robust controller can then be designed so that the output tracking error can converge to zero in a finite time, and strong robustness with respect to large uncertain dynamics can be guaranteed. It is also shown that the high gain of the terminal sliding mode controllers can be significantly reduced with respect to the one of the linear sliding mode controller where the sampling interval is nonzero. >

On adaptive spatial-temporal processing for airborne surveillance radar systems

Abstract: Implementing the optimum spatial-temporal (angle-Doppler) processor involves two crucial issues: the selection of processing configurations, and the development of adaptive algorithms which can efficiently approach the performance potential of the selected configuration. Among the three available configurations, the joint-domain, the cascade space-time, and the cascade time-space, this work shows that, in contrast to a popular belief, the detection performance potentials of both cascade configurations can fall far below that of the joint-domain optimum. In addition, this work presents a new adaptive algorithm, called the Joint-Domain Localized Generalized Likelihood Ratio detection (JDL-GLR), which is data efficient i.e., with fast convergence to the joint-domain optimum, as well as computationally efficient, together with such desirable features as the embedded constant false-alarm rate (CFAR) and robustness in non-Gaussian interference. >

Efficiency versus robustness : the case for minimum Hellinger distance and related methods

Abstract: It is shown how and why the influence curve poorly measures the robustness properties of minimum Hellinger distance estimation. Rather, for this and related forms of estimation, there is another function, the residual adjustment function, that carries the relevant information about the trade-off between efficiency and robustness. It is demonstrated that this function determines various second-order measures of efficiency and robustness through a scalar measure called the estimation curvature. The function is also shown to determine the breakdown properties of the estimators through its tail behavior. A 50% breakdown result is given. It is shown how to create flexible classes of estimation methods in the spirit of $M$-estimation, but with first-order efficiency (or even second-order efficiency) at the chosen model, 50% breakdown and a minimum distance interpretation.

On the development of discontinuous observers

Abstract: A framework is proposed for the development and analysis of robust observers for uncertain dynamical systems. A variable structure systems approach is used. Emphasis is placed upon the numerical tractability of the associated synthesis procedure. A selected numerical example is used to illustrate the proposed algorithm.

Enhancement of robustness in observer-based fault detection†

Abstract: A prerequisite for the feasibility of the technique of observer-based fault detection and isolation (FDI) in dynamic systems is a satisfactory robustness with respect to modelling uncertainties. This paper surveys the most relevant methods to increase the robustness in both the stage of residual generation and residual evaluation. Among these methods are the generalized observer scheme, the robust parity space check, the unknown input and H ∞ observer scheme, the decorrelation filter, and the concept of adaptive threshold selection. It is pointed out that the unknown input observer concept, which provides perfect decoupling from the modelling errors or its optimal approximation with the aid of H ∞ techniques, constitutes a general framework of robust residual generation that generalizes and unifies numerous other approaches, among them the parity space and detection filter approach. It is further shown that this FDI method can even be applied to a certain class of nonlinear systems.

Algorithm 733: TOMP–Fortran modules for optimal control calculations

Abstract: A great number of analysis and synthesis problems of modern processes can be written as state and control constrained optimal control problems governed by ordinary differential equations with multipoint boundary values. As the software tools for following this attractive approach are still missing or can be used only by experts, the structure and usage of an easy-to-use software package is described which efficiently solves the given problem. Among its features are user-orientation, applicability on personal computers and mainframes, and robustness with respect to model changes and inaccurate starting values. It has been tested on a number of complex engineering tasks, including aerospace and robotic trajectory planning.

Voltage phasor and local system frequency estimation using Newton type algorithm

Abstract: A new approach to the design of a digital algorithm for voltage phasor and local system frequency estimation is presented. The estimation problem is considered as an unconstrained optimization problem. The algorithm is derived using Newton's iterative method, very commonly used in load-flow studies. The algorithm showed a very high level of robustness as well as high measurement accuracy over a wide range of frequency changes. The algorithm convergence of order two provided fast response and adaptability. To demonstrate the performance of the algorithm developed, computer simulated, experimentally obtained and real-life data records are processed. The presented work is a part of a project concerning the application of microprocessors in frequency relaying. >

Review of parity space approaches to fault diagnosis for aerospace systems

Abstract: This paper provides a tutorial review of the state of the art in parity space fault diagnosis approaches with particular emphasis on aerospace systems. The basic concepts and definitions are given and a consistent framework is presented to draw together the important links amongst the known methods for fault diagnosis. Residual generation in the parity space has been recognized as a core element in this framework. The robustness and isolation problems are the main focus of the paper. Recent research topics on robust fault diagnosis are outlined, and new ideas as to how the parity space approach can be used to deal with robustness are discussed.

Decision trees for real-time transient stability prediction

Abstract: The ability to rapidly acquire synchronized phasor measurements from around the system opens up new possibilities for power system protection and control. This paper demonstrates how decision trees can be constructed offline and then utilized online for predicting transient stability in real-time. Primary features of the method include building a single tree for all fault locations, using a short window of realistic-precision post-fault phasor measurements for the prediction, and testing robustness to variations in the operating point. Several candidate decision trees are tested on 40,800 faults from 50 randomly generated operating points on the New England 39 bus test system. >

On a robust nonlinear servomechanism problem

Abstract: A notion of kth-order robust control for the nonlinear servomechanism problem is introduced. A transmission zeros condition is given for the existence of a kth-order robust control law such that, regardless of small parameter perturbations of the plant and control law, the closed-loop system will induce a stable invariant manifold with the error map zero up to kth-order at each point of the manifold. Such a control law incorporates an internal model of up to k subsystems determined by the exosystem. A special case of the result of this paper recovers the well-known linear robust servomechanism result. >

Robust time-optimal control - Frequency domain approach

Abstract: The design of nonrobust and robust time-optimal controllers for linear systems in the frequency domain is presented. The bang-bang profile is represented as the superposition of time-delayed step inputs or the output of a time-delay filter subject to a step input. A parameter optimization problem is formulated to minimize the final time of the maneuver with the constraint that the time-delay filter cancels all of the poles of the system. The issue of robustness to errors in the model is addressed by placing multiple zeros of the time-delay filter at the estimated locations of the poles of the system. The design technique is illustrated on representative models of large space structures, for rest-to-rest, time-optimal, and robust time-optimal maneuvers. Spin-up maneuvers are shown to be special cases of the general formulation. IME-OPTIMAL control of flexible spacecraft is a topic of current interest.1 Many computational approaches and analyses have been presented in the recent literature to deal with the effects of flexibility. Most of these works deal with planar (singleaxis) rest-to-rest maneuvers under two categories: near-minimumtime control and exact-minimum-time control. The first category of methods is based on smooth approximations to minimum-time control for an equivalent rigid body. This class of methods has been shown to be well suited when applied moments or torques are produced by either throttlable thrusters or reaction wheels.2'3 Higher modes of the system are not excited due to the smoothness of the control profile. The second category of methods deals with on-off thrusters directly. Rajan4 formulates the problem including

Robust learning algorithm for blind separation of signals

Abstract: The authors present a novel, efficient, self-normalising, unsupervised adaptive learning algorithm for the on-line (real-time) separation of statistically independent unknown source signals from a linear mixture of them. In contrast to the known algorithms the new algorithm allows the separation (or extraction) of extremely badly scaled signals (i.e. some or even all of the source and/or sensor signals can be very weak). Moreover, the mixing matrix can be very ill-conditioned. >

Nonlinear predictive controllers for continuous systems

Abstract: In this paper a new technique for the design of nonlinear feedback controllers to track desired response histories is presented. The controller is developed based on continuous minimization of predicted tracking errors. Some properties on the tracking performance and robustness of the controller are established. Both stateand output-tracking problems are considered in a unified framework. The effectiveness of this approach is demonstrated by an application in missile autopilot design.

Robust Optimization for Power Systems Capacity Expansion under Uncertainty

Abstract: We develop a robust optimization model for planning power system capacity expansion in the face of uncertain power demand. The model generates capacity expansion plans that are both solution and model robust. That is, the optimal solution from the model is ‘almost’ optimal for any realization of the demand scenarios (i.e. solution robustness). Furthermore, the optimal solution has reduced excess capacity for any realization of the scenarios (i.e. model robustness). Experience with a characteristic test problem illustrates not only the unavoidable trade-offs between solution and model robustness, but also the effectiveness of the model in controlling the sensitivity of its solution to the uncertain input data. The experiments also illustrate the differences of robust optimization from the classical stochastic programming formulation.

On-line fault detection in uncertain nonlinear systems using diagnostic observers: a survey

Abstract: The paper discusses the principles of model-based fault detection and isolation (FDI) in nonlinear and time-varying uncertain dynamic systems. Such systems are typical for such complex plants as, for example, in the chemical process industries or in advanced transportation technology. For a model-based fault diagnosis in such situations, robust or even adaptive strategies are needed. In this paper the theory of robust linear observer-based residual generation for FDI is reviewed from a general point of view. The structural equivalence between the parity space approach and observer-based approach is shown in a new simple graphical way by showing that the observer-based FDI concept can easily be transformed into an equivalent extended parity space configuration, without claiming, however, equivalence of the underlying design techniques. The unknown input observer approach known as a most powerful and comprehensive framework for robust residual generation for FDI in uncertain linear systems is extended to cl...

On the P-type learning control

Abstract: Sufficient conditions for the robustness and convergence of P-type learning control algorithms for a class of time-varying, nonlinear systems are presented. The authors prove the uniform boundedness of the system state and the input control with respect to the existence of errors of initialization, measurement noises, and fluctuations of system dynamics. Furthermore, the system output converges uniformly to the desired one in absence of all disturbances. Finally, specialization of the results to linear systems are presented. >

Minimum disparity estimation for continuous models: Efficiency, distributions and robustness

Abstract: A general class of minimum distance estimators for continuous models called minimum disparity estimators are introduced. The conventional technique is to minimize a distance between a kernel density estimator and the model density. A new approach is introduced here in which the model and the data are smoothed with the same kernel. This makes the methods consistent and asymptotically normal independently of the value of the smoothing parameter; convergence properties of the kernel density estimate are no longer necessary. All the minimum distance estimators considered are shown to be first order efficient provided the kernel is chosen appropriately. Different minimum disparity estimators are compared based on their characterizing residual adjustment function (RAF); this function shows that the robustness features of the estimators can be explained by the shrinkage of certain residuals towards zero. The value of the second derivative of theRAF at zero,A 2, provides the trade-off between efficiency and robustness. The above properties are demonstrated both by theorems and by simulations.

Fuzzy model based control: stability, robustness, and performance issues

Abstract: A nonlinear controller based on a fuzzy model of MIMO dynamical systems is described and analyzed. The fuzzy model is based on a set of ARX models that are combined using a fuzzy inference mechanism. The controller is a discrete-time nonlinear decoupler, which is analyzed both for the adaptive and the fixed parameter cases. A detailed stability analysis is carried out, and the main result is that the closed loop is globally stable and robust with respect to unstructured uncertainty, which may include modeling error and disturbances. In addition, bounds on the asymptotic and transient performance are given. The main assumptions on the system and model are that they must not have strong nonminimum-phase effects, except time-delay, and the unstructured uncertainty must not be too large. A simulation example illustrates some of the properties of the modeling method and model based control structure. >

Performance analysis and improvement in model reference adaptive control

Abstract: The purpose of this paper is to address the issue of performance by using two additional criteria to assess performance in the ideal and nonideal situations. They are the mean square tracking error criterion and the L/sub /spl infin// tracking error bound criterion. We use these criteria to examine the performance of a standard model reference adaptive controller and motivate the design of a modified scheme that can have an arbitrarily improved nominal performance in the ideal case and in the presence of bounded input disturbances. It is shown that for these cases the modified scheme can provide an arbitrarily improved zero-state transient performance and an arbitrary reduction in the size of possible bursts that may occur at steady state. As in every robust control design, the nominal performance has to be traded off with robust stability and therefore the improvement in performance achieved by the proposed scheme is limited by the size of the unmodeled dynamics, as established in the paper. >

Model-Based Object Recognition - A Survey of Recent Research

Abstract: We survey the main ideas behind recent research in model-based object recognition. The survey covers representations for models and images and the methods used to match them. Perceptual organization, the use of invariants, indexing schemes, and match verification are also reviewed. We conclude that there is still much room for improvement in the scope, robustness, and efficiency of object recognition methods. We identify what we believe are the ways improvements will be achieved.

A cluster estimation method with extension to fuzzy model identification

Abstract: We present an efficient method for estimating cluster centers of numerical data. This method can be used to determine the number of clusters and their initial values for initializing iterative optimization-based clustering algorithms such as fuzzy C-means. Here were combine this cluster estimation method with a least squares estimation algorithm to provide a fast and robust method for identifying fuzzy models from input/output data. A benchmark problem involving the prediction of a chaotic time series shows this method compares favourably with other more compositionally intensive methods. >

Face locating and tracking for human-computer interaction

Abstract: Effective human-to-human communication involves both auditory and visual modalities, providing robustness and naturalness in realistic communication situations. Recent efforts at our lab are aimed at providing such multimodal capabilities for human-machine communication. Most of the visual modalities require a stable image of a speaker's face. We propose a connectionist face tracker that manipulates camera orientation and room, to keep a person's face located at all times. The system operates in real time and can adapt rapidly to different lighting conditions, cameras and faces, making it robust against environmental variability. Extensions and integration of the system with a multimodal interface are presented. >