The time average reward for some dynamic fuzzy systems

In this paper, by using a fuzzy relation, we define a dynamic fuzzy system with a bounded convex fuzzy reward on the positive orthant Rⁿ₊ of an n-dimensional Euclidean space. As a measure of the system's performance, we introduce the time average fuzzy reward, which is characterized by the limiting fuzzy state under the contractive properties of the fuzzy relation. In the one-dimensional case, the average fuzzy reward is expressed explicitly by the functional equations concerning the extreme points of its α-cuts. Also, a numerical example is given to illustrate the theoretical results.     

Semidynamic Algorithms for Maintaining Single-Source Shortest Path Trees

We consider the problem of updating a single-source shortest path in either a directed or an undirected graph, with positive real edge weights. Our algorithms for the incremental problem (handling edge insertions and cost decrements) work for any graph; they have optimal space requirements and query time, but their performances depend on the class of the considered graph. The cost of updates is computed in terms of amortized complexity and depends on the size of the output modifications. In the case of graphs with bounded genus (including planar graphs), graphs with bounded arboricity (including bounded degree graphs), and graphs with bounded treewidth, the incremental algorithms require O(log n) amortized time per vertex update, where a vertex is considered updated if it reduces its distance from the source. For general graphs with n vertices and m edges our incremental solution requires O( log n) amortized time per vertex update. We also consider the decremental problem for planar graphs, providing algorithms and data structures with analogous performances. The algorithms, based on Dijkstra's technique [6], require simple data structures that are really suitable for a practical and straightforward implementation. Received January 1995; revised February 1997.     

Input disturbance suppression for a class of feedforward uncertain nonlinear systems

This paper deals with the problem of asymptotically rejecting bounded unknown disturbances affecting the input channel of a feedforward uncertain nonlinear system. The problem is solved assuming that the matched disturbance belongs to the class of signals generated by an autonomous neutrally stable exosystem whose state is not accessible. We design an internal model-based regulator capable on one hand to reject the matched disturbance for any initial state of the exosystem and, on the other hand, to robustly globally asymptotically stabilize the system using state feedback.     

Simulation-based optimization of Markov decision processes: An empirical process theory approach

We generalize and build on the PAC Learning framework for Markov Decision Processes developed in Jain and Varaiya (2006). We consider the reward function to depend on both the state and the action. Both the state and action spaces can potentially be countably infinite. We obtain an estimate for the value function of a Markov decision process, which assigns to each policy its expected discounted reward. This expected reward can be estimated as the empirical average of the reward over many independent simulation runs. We derive bounds on the number of runs needed for the convergence of the empirical average to the expected reward uniformly for a class of policies, in terms of the V-C or pseudo dimension of the policy class. We then propose a framework to obtain an ?-optimal policy from simulation. We provide sample complexity of such an approach.     

Synthesis of a Multi-Connected Digital Controller for a Robotized Vibration Isolation Platform Based on <Emphasis Type="Italic">H</Emphasis><Subscript>∞</Subscript>-Optimization

We consider the problem of constructing multi-connected control of a robotic platform designed to protect technological objects and human operators from low-frequency influences on part of the moving base. The platform includes six drive mechanisms with stepper motors. The problem is solved by the methods of the modern theory of robust stabilization and optimal control based on H_∞-optimization in the state space. We construct a mathematical model of the multidimensional system, taking into account the characteristics of electromechanical drives and using signals of feedback sensors as state variables. We give an example of synthesizing a multidimensional optimal stabilizing controller in the form of state feedback for a system with disturbances bounded in L₂-norm. We define the feedback control structure and obtain the matrix of feedback coefficients. We also show the results of mathematical modeling.     

Semi-global robust output regulation of minimum-phase nonlinear systems based on high-gain nonlinear internal model

We consider the assumption of existence of the general nonlinear internal model that is introduced in the design of robust output regulators for a class of minimum-phase nonlinear systems with rth degree (r ≥ 2). The robust output regulation problem can be converted into a robust stabilisation problem of an augmented system consisting of the given plant and a high-gain nonlinear internal model, perfectly reproducing the bounded including not only periodic but also nonperiodic exogenous signal from a nonlinear system, which satisfies some general immersion assumption. The state feedback controller is designed to guarantee the asymptotic convergence of system errors to zero manifold. Furthermore, the proposed scheme makes use of output feedback dynamic controller that only processes information from the regulated output error by using high-gain observer to robustly estimate the derivatives of the regulated output error. The stabilisation analysis of the resulting closed-loop systems leads to regional as well as semi-global robust output regulation achieved for some appointed initial condition in the state space, for all possible values of the uncertain parameter vector and the exogenous signal, ranging over an arbitrary compact set.     

Robust H∞ output‐feedback control of retarded state‐multiplicative stochastic systems

Linear, state‐delayed, continuous‐time systems are considered with both stochastic and norm‐bounded deterministic uncertainties in the state–space model. The problem of robust dynamic H_∞ output‐feedback control is solved, for the stationary case, via the input–output approach where the system is replaced by a nonretarded system with additional deterministic norm‐bounded uncertainties. A delay‐dependent result is obtained which involves the solution of a simple linear matrix inequality. In this problem, a cost function is defined which is the expected value of the standard H_∞ performance cost with respect to the stochastic parameters. A practical example taken from the field of guidance control is given that demonstrates the applicability of the theory. Copyright © 2010 John Wiley & Sons, Ltd.     

Distributed Explicit Bounded LTL Model Checking

Automated formal verification becomes a significant part of an industrial design process. Favourite formal verification method  model checking  is strongly limited by the size of the model of the verified system. It suffers from the so called state explosion problem. We propose to fight this problem by applying the idea of bounding the examined state space in explicit model checking. Moreover, we combine this approach with the distribution of the computation among the network of workstations. We consider several distributed bounded LTL model checking algorithms and carry out a series of experiments to evaluate them and to compare their behaviour.     

Learning action probabilities from delayed reinforcement

A reinforcement scheme for learning automata, applicable to real situations where the reinforcement received from the environment is delayed, is presented. The scheme divides the state space into regions following the boxes approach of Michie and Chambers. Each region maintains estimates of the reward characteristics of the environment and contains a local automaton that updates action probabilities whenever the system state enters it. Estimates of reward characteristics are obtained using reinforcement received during the period of eligibility. Results obtained through computer simulation of the inverted pendulum problem are compared with the adaptive critic learning developed by Barto et al. (1983).     

Semi-global stabilization of minimum phase nonlinear systems in special normal form via linear high-and-low-gain state feedback

We provide an alternative solution to the problem of semi-global stabilization of a class of minimum phase nonlinear systems which is considered in Reference 17. Our method yields a stabilizing linear state feedback law in contrast to a nonlinear state feedback law proposed in Reference 17. We eliminate the peaking phenomenon by inducing a specific time-scale structure in the linear part of the closed-loop system. This time-scale structure consists of a very slow and a very fast time scale. The crucial component in our method is the relation between the slow and the fast time scales. Our proposed linear state feedback control law has a single tunable gain parameter that allows for local asymptotic stability and regulation to the origin for any initial condition in some a priori given (arbitrarily large) bounded set.     

Symbolic execution with abstraction

We address the problem of error detection for programs that take recursive data structures and arrays as input. Previously we proposed a combination of symbolic execution and model checking for the analysis of such programs: we put a bound on the size of the program inputs and/or the search depth of the model checker to limit the search state space. Here we look beyond bounded model checking and consider state matching techniques to limit the state space. We describe a method for examining whether a symbolic state that arises during symbolic execution is subsumed by another symbolic state. Since the number of symbolic states may be infinite, subsumption is not enough to ensure termination. Therefore, we also consider abstraction techniques for computing and storing abstract states during symbolic execution. Subsumption checking determines whether an abstract state is being revisited, in which case the model checker backtracks—this enables analysis of an under-approximation of the program behaviors. We illustrate the technique with abstractions for lists and arrays. We also discuss abstractions for more general data structures. The abstractions encode both the shape of the program heap and the constraints on numeric data. We have implemented the techniques in the Java PathFinder tool and we show their effectiveness on Java programs. This paper is an extended version of Anand et al. (Proceedings of SPIN, pp. 163–181, 2006).     

Synthesizing <Emphasis Type="Italic">H</Emphasis><Subscript>∞</Subscript>-controllers for multidimensional systems with given accuracy and degree of stability

We consider linear multidimensional systems with output controllers subject to external perturbances from the class of polyharmonic functions with unknown amplitudes and bounded powers. We formulate the synthesis problem for continuous and discrete output controllers that guarantee a given accuracy with respect to the object’s controlled variables. We introduce the notion of a stabilized state radius of the closed system with respect to controlled variables and reformulate the problem of guaranteeing a given accuracy as a problem of guaranteeing the necessary or minimal possible stabilized state radius. Controller synthesis reduces to a standard H _∞-problem of suppressing external perturbances, and its numerical solution is based on the linear matrix inequalities (LMI) approach implemented in the MATLAB package LMI Control Toolbox. We show a way to take into account a given degree of stability of the closed system which determines control time. We show a sample controller synthesis for an interconnected electric drive.     

H∞control of interconnected nonlinear sampled-data systems with parametric uncertainty

This paper studies the problem of robust control design for a class of interconnected uncertain systems under sampled measurements. The class of system under consideration is described by a state space model containing unknown cone bounded nonlinear interaction and time-varying norm-bounded parameter uncertainties in both state and output equations. Our attention is focused on the design of linear dynamic output feedback controllers using sampled measurements. We address the problem of robust H_∞ control in which both robust stability and a prescribed H_∞ performance are required to be achieved irrespective of the uncertainties and nonlinearities. The H_∞ performance measure involves both continuous-time and discrete-time signals. It has been shown that the above problems can be recast into H_∞ syntheses for related N decoupled linear sampled-data systems without parameter uncertainties and unknown nonlinearities, which can be solved in terms of Riccati differential equations with finite discrete jumps. A numerical example is given to show the potential of the proposed technique.     

H∞ Control of Linear Discrete‐Time Systems With Norm‐Bounded Nonlinear Uncertainties

In this paper, robust H_∞ control of a class of discrete‐time uncertain systems in state‐space form with linear nominal parts and norm‐bounded nonlinear uncertainties in both state and output equations is discussed. Such systems have a unique characterisic; that is, the two norm‐bounded nonlinear uncertainties have the equivalent representation by means of time‐varying and norm‐bounded linear uncertainties. To overcome the conservativenss of [5], the two nonlinear uncertainty sets are considered to be different. Then, by converting such systems into related discrete‐time linear systems with time‐varying and norm‐bounded linear uncertainties, we obtain that a sufficient condition for robust H_∞ control of such systems is equivalent to the solvability of the same problem of the related linear uncertain systems, which is solvable by means of a linear algebraic Riccati inequality.     

Average Case Analysis of Bounded Space Bin Packing Algorithms

We consider the one-dimensional bin packing problem and analyze the average case performance of bounded space algorithms. The analysis covers a wide variety of bin packing algorithms including Next-K Fit, K-Bounded Best Fit and Harmonic algorithms, as well as of others. We assume discrete item sizes, an assumption which is true in most real-world applications of bin packing. We present a Markov chains method which is general enough to calculate results for any i.i.d. discrete item size distribution. The paper presents many results heretofore unknown or conjectured from simulation. Some of the results are surprising as they differ considerably from results for continuous distributions.     

H∞ Adaptive tracking control for switched systems based on an average dwell-time method

This paper investigates the H_∞ state tracking model reference adaptive control (MRAC) problem for a class of switched systems using an average dwell-time method. First, a stability criterion is established for a switched reference model. Then, an adaptive controller is designed and the state tracking control problem is converted into the stability analysis. The global practical stability of the error switched system can be guaranteed under a class of switching signals characterised by an average dwell time. Consequently, sufficient conditions for the solvability of the H_∞ state tracking MRAC problem are derived. An example of highly manoeuvrable aircraft technology vehicle is given to demonstrate the feasibility and effectiveness of the proposed design method.     

Stochastic switched controller design of networked control systems with a random long delay

The problem of stabilizing Networked Control Systems (NCSs) with random but bounded delays is discussed in this paper. By using an augmented state‐space method, this class of problem can be modeled as a discrete‐time jump linear system governed by finite‐state Markov chains. As the network‐induced time vary delay of NCSs changes along with the network transferring route and the network load, results in systems becoming instable with controller designs based on a fixed transition matrix, we firstly make use of the V‐K iteration algorithm to design m groups of stabilizing controllers that satisfy different m transition matrixes, and then constitute a switched controller for them and a switch. The simulation shows that if the switched controller is used to stabilize the discrete‐time jump linear system, this system not only has a larger stabilizing span, but also has better dynamic stabilizing characters compared to those with only one group of controllers. Copyright © 2010 John Wiley and Sons Asia Pte Ltd and Chinese Automatic Control Society     

Exponential H∞ filtering for switched linear systems with interval time‐varying delay

This paper deals with the problem of exponential H_∞ filtering for a class of continuous‐time switched linear system with interval time‐varying delay. The time delay under consideration includes two cases: one is that the time delay is differentiable and bounded with a constant delay‐derivative bound, whereas the other is that the time delay is continuous and bounded. Switched linear filters are designed to ensure that the filtering error systems under switching signal with average dwell time are exponentially stable with a prescribed H_∞ noise attenuation level. Based on the free‐weighting matrix approach and the average dwell technology, delay‐dependent sufficient conditions for the existence of such a filter are derived and formulated in terms of linear matrix inequalities (LMIs). By solving that corresponding LMIs, the desired filter parameterized matrices and the minimal average dwell time are obtained. Finally, two numerical examples are presented to demonstrate the effectiveness of the developed results. Copyright © 2008 John Wiley & Sons, Ltd.     

Transforming Unbalanced Multiway trees into a practical external data structure

Summary Unbalanced Multiway Trees (UM-trees) of degree m are external data structures wherein each node may be linked to at most m subtrees. Although they allow fast searches and updates UM-trees show poor space efficiency. In order to improve their time and space performance an overflow technique similar to that used previously for B ^*-trees is adapted to UM-trees. Analytical results show that when compared with UM-trees the new data structure improves average searching time and substantially increases space efficiency. Asymptotic space efficiency, measured as the ratio of space used to space generated, is 50% in the worst case and in the average case it is upper bounded by at least 85.7% for any m. Simulations suggest that this upper bound is tight. Compared with a homologous variant of B ^*-trees, simulation results indicate that the data structure proposed is highly balanced with comparable average space efficiency and lower average searching time. We conclude that overflow techniques can be applied not only to bottom-up type trees (e.g., B-trees) but also to top-down type trees such as UM-trees, in cases when the performance of an external data structure needs to be improved with little overhead.     

Feedback invariance of SISO infinite-dimensional systems

We consider a linear single-input single-output system on a Hilbert space X, with infinitesimal generator A, bounded control element b, and bounded observation element c. We address the problem of finding the largest feedback invariant subspace of X that is in the space c ^⊥ perpendicular to c. If b is not in c ^⊥, we show this subspace is c ^⊥. If b is in c ^⊥, a number of situations may occur, depending on the relationship between b and c.