Browsing by Author "Zafiriou, E."
Now showing 1 - 20 of 25
Results Per Page
Sort Options
Item Application of Neural Networks on the Detection of Sensor Failure During the Operation of a Control System.(1989) Naidu, S.R.; Zafiriou, E.; McAvoy, Thomas J.; ISRNeural computing is one of the fastest growing branches of artificial intelligence. Neural Nets, endowed with inherent parallelism hold great promise owing to their ability to capture highly nonlinear relationships. This paper discusses the use of the back propagation neural net for failure cognition in chemical process systems. The backpropagation paradigm along with traditional fault detection algorithms such as the finite integral square error method and the nearest neighbor method are discussed. The algorithm is applied to an IMC controlled first order linear time invariant plant subject to high model uncertainty. Compared to traditional methods, the backpropagation technique is shown to be able to accurately discern the supercritical failures from their subcritical counterparts. The use of backpropagation fault detection systems in on-line adaptation of nonlinear plants has been investigated.Item Control System Sensor Failure Detection via Networks of Localized Receptive Fields(1990) Yao, S.C.; Zafiriou, E.; ISRThis paper investigates the use of local receptive field networks (LRFN) in detecting sensor failures of a control system in the presence of model-plant mismatch. Simulation results indicate that LRFNs hold significant promise in sensor failure detection. Another issue discussed in this paper is a method to prune redundant nodes. A simple scheme which uses singular value decomposition (SVD) is developed to identify and remove excess nodes. Comparable classification performance is obtained using reduced and standard LRFN.Item Design of Robust Digital Controllers and Sampling-Time Selection for SISO Systems.(1987) Zafiriou, E.; Morari, M.; ISRThe stability of a digital control system and its performance in terms of the continuous plant output are studied. A two-step controller design is proposed. In the first step, the assumption of no modelling error is made and a controller that combines properties of the algorithm that minimizes the sum of squared errors and a deadbeat-type algorithm is designed so that no intersample rippling appears. In the second step, a filter is designed so that appropriate conditions which guarantee robust stability and performance in the presence of model-plant mismatch are satisfied. The effect of the sampling time on the achievable performance and the robustness properties of the system is examined and the results are incorporated in a complete procedure for sampling-time selection and robust controller design. Finally, the procedure and some theoretical implications are illustrated with examples.Item Design of the IMC Filter by Using the Structured Singular Value Approach.(1987) Zafiriou, E.; Morari, M.; ISRThe Internal Model Control (IMC) structure has been widely recognized as very useful in clarifying the issues related to the mismatch between the model used for controller design and the actual process. The structure also gives rise to a two step controller synthesis procedure, of which the second step deals with the design of a low pass filter such that robustness with respect to model-plant mismatch is guaranteed. The Structured Singular Value (SSV) was introduced recently and it allowed the non-conservative quantification of the concept of robust performance. This paper deals with the design of the IMC filter by uaing the SSV and it demonstrates how this approach can be used with either an H{SUB 2^-} or an H{SUB INFINITY} optimal controller.Item Digital Controller Design for Multivariable Systems with Structural Closed-Loop Performance Specifications.(1988) Zafiriou, E.; Morari, M.; ISRThe problem of the direct design of the closed-loop transfer function matri'; is addressed for multivariable discrete systems. The limitations imposed by unstable zeros, time delays and the structure associated with these are quantified. A design procedure is formulated that provides the designer with quantitative measures for evaluating the tradeoffs between different closed-loop interaction structures and durations. The problem of intersample rippling is also considered. The procedure requires only linearalgebra operations, includes the eventual construction of the feedback controller in state space, and is presented in a way that allows its straightforward computer implementation.Item Digital Controller Design for Multivariable Systems with Structural Closed-Loop Performance Specifications.(1987) Zafiriou, E.; Morari, M.; ISRThe problem of the direct design of the closed-loop transfer function matrix is addressed for multivariable discrete systems. The limitations imposed by unstable zeros, time delays and the structure associated with these are quantified. A design procedure is formulated that provides the designer with quantitative measures for evaluating the tradeoffs between different closed-loop interaction structures and durations. The problem of intersample rippling is also considered. The procedure requires only linear algebra operations, includes the eventual construction of the feedback controller in state space and is presented in a way that allows its straightforward computer implementation.Item Digital Controllers for SISO Systems: A Review and a New Algorithm.(1987) Zafiriou, E.; Morari, M.; ISRSeveral digital control algorithms for linear single-input single-output systems are examined and the effect of the sampling period on their performance is analyzed in terms of rippling, overshoot and settling time. The problem is addressed in the frequency domain (ztransform) and it is shown that each controller works for some classes of systems but that none works for all. The similarities and differences of these controllers are established and an explanation of their deficiencies is given based on the location of the zeros of the discrete system. The insight gained leads to a simple new rule for the design of a controller which combines the advantages of the different algorithms but at the same time is free of their problems. A single tuning parameter is included which directly affects the closed-loop speed of response and bandwidth. The parameter can be used to detune the controller in the event that the real system differs from the model on which the controller design is based. No tuning is necessary when the available model is exact, unless smaller values for the manipulated variable, at the cost of a slower response, are preferred.Item Internal Model Control: Robust Digital Controller Synthesis for Multivariable Open-Loop Stable or Unstable Processes(1990) Zafiriou, E.; Morari, M.; ISRThe two-step Internal Model Control (IMC) procedure is presented for the synthesis of multivariable discrete controllers. This paper adds the following features to the IMC design methodology: (i) Extension to open-loop unstable plants. (ii) Design of the first-step (no model error) IMC controller so that the L2-error (sum of squared errors) is minimized for every setpoint or disturbance vector in a designer-specified set and their linear combinations. (iii) The second-step (model-plant mismatch) multivariable low-pass filter is designed for robust stability and performance by minimizing a non-conservative robustness measure, the Structured Singular Value. (iv) The potential problem in intersample rippling is avoided by introducing a modification in the first-step controller and formulating the robust performance objective for the continuous plant output.Item Internal Model Control: Robust Digital Controller Synthesis for Multivariable Open-Loop Stable or Unstable Systems.(1987) Zafiriou, E.; Morari, M.; ISRThe two-step Internal Model Control procedure is used for the synthesis of multivariable discrete controllers for open-loop stable or unstable plants. The plant models used in the proposed method are transfer function matrices. In the first step the controller is designed so that the L{SUB 2^-}error (sum of squared errors) is minimized for every setpoint or disturbance vector in a set and their linear combinations. A modification is then introduced to avoid the potential problem of intersample rippling. In the second step a low-pass filter is designed so that stability and good performance characteristics are maintained in the presence of model-plant mismatch. The continuous plant output is considered in order to avoid bad intersample behavior. The filter parameters are obtained as the result of a minimization of a non-conservative robustness measure, the Structured Singular Value. Special filter structures have to be used for open-loop unstable or ill-conditioned plants.Item Iterative Batch-to-Batch Input Profile Determination for Semi- Batch Processes.(1989) Zhu, J.M.; Zafiriou, E.; ISRPolymerization processes are very complex and high nonlinear. Their modeling often involves reaction mechanism analysis combined with empirical tests. Mismatch between the model and the industrial plant often exists and it can be the cause of bad performance when optimal input profiles computed for a particular model are applied to the actual plant. The approach followed in this paper is to directly modify the input profile from batch to batch, and it has been applied through computer simulations to the determination of the minimum-time temperature profile for the batch bulk polymerization of sqrene under modeling error. The results demonstrate that the approach has strong robustness characteristics and fast convergence properties.Item A Local Small Gain Theorem and Its Use for Robust Stability of Uncertain Feedback Volterra Systems(1993) Zheng, Q.; Zafiriou, E.; ISRThe requirement to evaluate a gain over the whole signal space is one of the restrictions in the well-known small gain theorem. Using the concepts of local gain and strict causality a local form of small gain theorem is proposed, which can be used to analyze input magnitude dependent stability problems of feedback nonlinear systems, such as a Volterra system. Since only finite order Volterra series can be handled in practice, an uncertainly model is derived to address the robustness issue of approximating a nonlinear system by a finite Volterra series in the context of closed-loop control. The local small gain theorem is then used to analyze the feedback properties of the uncertain Volterra system and a sufficient condition for robust stability is obtained.Item Model Reduction for RTCVD Optimization(1996) Theodoropoulou, A.; Adomaitis, Raymond A.; Zafiriou, E.; ISRA model of a three-zone Rapid Thermal Chemical Vapor Deposition (RTCVD) system is developed to study the effects of spatial wafer temperature patterns on polysilicon deposition uniformity. A sequence of simulated runs is performed, varying the lamp power profiles so that different wafer temperature modes are excited. The dominant spatial wafer thermal modes are extracted via Proper Orthogonal Decomposition and subsequently used as a set of trial functions to represent both the wafer temperature and deposition thickness. A collocation formulation of Galerkin's method is developed to discretize the original modeling equations, giving a low-order model which looses little of the original, high-order model's fidelity. We make use of the excellent predictive capabilities of the reduced model to optimize power inputs to the lamp banks to achieve a desired polysilicon deposition thickness at the end of a run with minimal deposition spatial nonuniformity.Item On the Closed-Loop Stability of Constrained QDMC(1991) Zafiriou, E.; ISRThe presence of constraints in the on-line optimization problem solved by Model Predictive Control algorithms results in a nonlinear control system, even if the plant and model dynamics are linear. This is the case both for physical constraints, like saturation constraints, as well for performance or safety constraints on outputs or other variables of the process. This paper discusses how constraints affect the stability properties of the closed-loop nonlinear system. In particular we concentrate on presenting a formulation that allows one to relate hard as well as soft constraints to stability. The degree of softening can be determined to guarantee stability.Item On the Effect of Constraint Softening on the Stability and Performance of Model Predictive Controllers(1992) Zafiriou, E.; Chiou, Hung-Wen; ISRThe presence of constraints in the on-line optimization problem solved by Model Predictive Control algorithms results in a nonlinear control system, even if the plant and model dynamics are linear. This is the case both for physical constraints, like saturation constraints, as well for performance or safety constraints on outputs or other variables of the process. Performance constraints can usually be softened by allowing violation if necessary. This is advisable, as hard constraints can lead to stability problems. The determination of the necessary degree of softening is usually a trial-and-error matter. This paper utilizes a theoretical framework that allows to relate hard as well as soft constraints to closed-loop stability. The problem of determining the appropriate degree of softening is addressed by treating the parameters (weights) affecting the amount of softening as one-sided real-valued uncertainty and solving a robust stability problem.Item On the Tuning of Nonlinear Model Predictive Control Algorithms(1993) Ali, Emad; Zafiriou, E.; ISRNonlinear Model Predictive Controllers determine appropriate control actions by solving an on-line optimization problem. A nonlinear process model is utilized for on-line prediction, making such algorithms particularly appropriate for the control of chemical reactors. The algorithm presented in this paper incorporates an Extended Kalman Filter, which allows operations around unstable steady-state points. The paper proposes a formalization of the procedure for tuning the several parameters of the control algorithm. This is accomplished by specifying time-domain performance criteria and using an interactive multi- objective optimization package off-line to determine parameter values that satisfy these criteria. A reactor example is used to demonstrate the effectiveness of the proposed on-line algorithm and off-line tuning procedure.Item Optimal Control of Semi-Batch Processes in the Presence of Modeling Error(1990) Zafiriou, E.; Zhu, J.M.; ISRBatch processes are usually complex and highly nonlinear systems. Modeling error can be the cause of bad performance when optimal input profiles computed for a particular model are applied to the actual plant. The approach followed in this paper uses the available model and actual plant measurements to modify the operation of the next batch, without requiring the remodeling of the process. The effect of model error on the convergence of the iterative batch to batch input profile determination is investigated. The method is applied through computer simulations to the determination of the optimal feedrate profile for a cell mass production process. A model parameter update scheme is also proposed, based on the convergence analysis. This is applied to the determination of the optimal temperature profile of bulk polymerication of the optimal temperature profile of styrene.Item Optimal Feed Rate Profile Detamination for Fed-Batch Fermentations in the Presence of Model Plant Mismatch.(1989) Zafiriou, E.; Zhu, J.M.; ISRModelling error can be the cause of bad performance when optimal feed-rate profiles computed for a particular model are applied to the actual plant. This paper suggests the modification of the input trajectory from batch to batch, by using information from previous batches to modify the trajectories that are applied to the subsequent ones. The proposed approach does not require the remodeling of the process, but instead it redetermines the input profile directly, so that a steady improvement is accomplished from batch to batch.Item Optimization-based Tuning of Nonlinear Model Predictive Control with State Estimation(1993) Ali, Emad; Zafiriou, E.; ISRNonlinear Model Predictive Controllers determine appropriate control actions by solving an on-line optimization problem. A nonlinear process model is utilized for on-line prediction, making such algorithms particularly appropriate for the control of chemical reactors. The algorithm presented in this paper incorporates an Extended Kalman Filter, which allows operations around unstable steady-state points. The paper proposes a formalization of the procedure for tuning the several parameters of the control algorithm. This is accomplished by specifying time-domain performance criteria and using an interactive multi- objective optimization package off-line to determine parameter values that satisfy these criteria. Three reactor examples are used to demonstrate the effectiveness of the proposed on-line algorithm and off-line procedure.Item Recent Advances in the Use of the Internal Model Control Structure for the Synthesis of Robust Multivariable Controllers.(1987) Zafiriou, E.; ISRThis paper presents the following recent theoretical developments in the IMC methodology: DOT Multivariable controller design for the minimization of the Integral Squared error (ISE) for every input direction in a set and their linear combinations. DOT Treatment of open-loop unstable plants; use of the two-degree-of- freedom controller. DOT Minimization of the Structured Singular Value (SSV) for robust performance over the IMC Filter parameters; unconstrained problem; analytic computation of the gradients. DOT Computation of the worst (over all possible plants) ISE for a particular setpoint or disturbance input. The paper deals with continuous systems. Extension to sampled-data systems is straight forward but not included here for lack of space.Item Robust Control of Processes with Hard Constraints.(1989) Zafiriou, E.; ISRA significant number of Model Predictive Control algorithms solve on-line an appropriate optimization problem and do so at every sampling point. The major attraction of such algorithms, like the Quadratic Dynamic Matrix Control (QDMC), lies in the fact that they can handle hard constraints on the inputs (manipulated variables) and outputs of a process. The presence of such constraints results in an on-line optimization problem that produces a nonlinear controller, even when the plant and model dynamics are assumed linear. This paper provides a theoretical framework within which the stability and performance properties of such algorithms can be studied. Necessary and/or sufficient conditions for nominal and robust stability are derived and two examples are used to demonstrate their effectiveness.