Advanced Control Design Research Articles

Hardware Synthesis of Explicit Model Predictive Controllers

The general solution to constrained linear and piecewise linear model predictive control (MPC) has recently been explicitly characterized in terms of piecewise-linear (PWL) state feedback control. This means that a PWL controller can be precomputed using parametric programming, and the exact explicit MPC implementation amounts to the evaluation of a PWL function in the control unit. It has recently been shown that PWL function evaluation can be accelerated by searching a binary tree data structure, leading to highly efficient, accurate, and verifiable software implementation in low-cost embedded control units. In this work, we report hardware synthesis results for this type of PWL control, and show that explicit MPC solutions can be implemented in an application specific integrated circuit (ASIC) with about 20 000 gates, leading to computation times in the microsecond scale. This opens the way for the use of highly advanced control designs such as constrained MPC in small-scale industrial and consumer electronics application areas that are characterized by fast sampling or low cost, including mechatronics, microelectromechanical systems (MEMS), automotive control, power electronics, and acoustics. The main limitation of the approach is that the memory requirements increase rapidly with the problem dimensions

Comparative evaluation of nonlinear model predictive and flatness-based two-degree-of-freedom control design in view of industrial application

Two advanced nonlinear model-based control design methods – nonlinear model predictive control (NMPC) and a two-degree-of-freedom control-scheme with flatness-based feedforward control design and decentralised PI-controllers (FB-2DOF) – are compared in view of industrial application. The comparative evaluation is carried out on a setpoint-transition of the Klatt–Engell reactor model. Based on an analysis of simulation scenarios, the controllers are compared with respect to controller performance, robustness criteria, and implementation issues. Thereby, the choice of the control task and the comparison methodology are oriented on industrial practice. In the considered comparative evaluation, NMPC exhibits performance advantages when it comes to time-efficient setpoint-transitions in the nominal case, in which FB-2DOF control design is restricted by the existing input constraints. In return, robustness of stability of the FB-2DOF controller is determined only by the feedback controll part; it is therefore independent from the setpoint-transition performance – determined by the feedforward controll part – whereas the NMPC suffers from degradation of robustness properties if it is tuned for time-efficiency only. NMPC allows direct incorporation of process models and constraints, but, as it employs computationally expensive online optimisation, has to be connected to the digital control system (DCS) via some standard interface. The FB-2DOF controller in contrast can be directly implemented in a DCS, whereby the feedforward-part can be realised as an extension of an already existing feedback-part.

Control of stationary and transportation fuel cell systems: Progress and opportunities

Fuel cell power systems have witnessed intense development in the recent years as they offer a clean and efficient alternative for power generation in stationary and transportation applications. In this paper, we present an overview of recent work on dynamic modeling and control design for stationary and transportation fuel cell systems including the research at United Technologies Corporation. The use of equation oriented modeling framework for system level dynamic modeling enabled by tools, such as Dymola and gPROMS is described. It is demonstrated that the non-linear system level models readily allow linear model derivation and the application of advanced control analysis and design techniques that provide more insight into the system level dynamic and control issues in fuel cell systems. The dynamic modeling and advanced control design research performed at UTC for a hydrogen fed PEMFC power plant for a bus and a natural gas fed PEMFC power plant for stationary power generation is described. We also identify, for future research, the challenges and opportunities in several areas relating to fuel cell systems ranging from power management and freeze startup to system level modeling, control, diagnostics and hardware-in-the-loop validation of the control system. The emphasis is on polymer electrolyte membrane fuel cell (PEMFC) systems but phosphoric acid fuel cell (PAFC) systems are also briefly discussed.

Adaptive controller design for a sensorless IPMSM drive system with a maximum torque control

An advanced controller design for a sensorless interior permanent-magnet synchronous motor control system is proposed. A maximum torque/ampere control algorithm is used to improve the torque performance of the drive system. In addition, an adaptive controller combined with the maximum torque/ampere control is proposed to obtain a good transient response and a good load disturbance rejection capability. The proposed control method can be applied to both a sensorless adjustable speed control system and a sensorless position control system. Using the Lyapunov stability theory and Barbalat's lemma it is shown that the closed-loop sensorless control system is an asymptotical stable system. A 32-bit digital signal processor is used to execute the rotor position estimation algorithm and the control algorithm. Experimental results validate the theoretical analysis and show the correctness and feasibility of the proposed system

Nonrigid, Linear Plasma Response Model Based on Perturbed Equilibria for Axisymmetric Tokamak Control Design

Tokamak control design relies on an accurate linear model of the plasma response, which can often dominate the local field variations in regions under active feedback control. For example, when fluxes at selected points on the plasma boundary are regulated in DIII–D [1], the plasma response to a change in a coil current gives rise to a flux change which can be larger than and opposite to the flux change caused by the coil alone.In the past, rigid plasma models have been used for linear stability and shape control design. In a rigid model, the plasma current profile is considered fixed and moves rigidly in response to control coils to maintain radial and vertical force balance. In a nonrigid model, however, changes in the plasma shape and current profile are taken into account. Such models are expected to be important for future advanced tokamak control design. The present work describes development of a nonrigid plasma response model for high-accuracy multivariable control design and provides comparisons of model predictions against DIII-D experimental data. The linear perturbed plasma response model is calculated rapidly from an existing equilibrium solution.

RANGE CONTROL MPC APPROACH FOR TWO-DIMENSIONAL SYSTEM

This paper deals with model predictive control of a distributed parameters system which is described by a linear two-dimensional (dependent on two spatial directions) parabolic partial differential equation. This partial differential equation is transformed to the discrete state space description using the finite difference approximation. A model with a large dimension is obtained and has to be reduced for an advanced control design. The balanced truncation method is used for the model dimension reduction. For this low dimension model, the range control approach is applied.

IMPROVING THE CONTROL AND RELIABILITY OF AN ELECTROMECHANICAL ACTUATOR

By combining the disciplines of control design, condition monitoring and fault detection identification and accommodation, it should be possible to provide actuators that possess: consistent and tuneable closed-loop performance, a tolerance to sensor faults and the ability to detect actuator deterioration early. This will lead to improved reliability and availability of the actuator. This paper discusses research carried-out to enable the inclusion of advanced control design, fault detection and condition monitoring schemes in an electro-mechanical actuator. The results in the paper were obtained on a laboratory scale actuator in test laboratories at the ALSTOM Power Technology Centre, Whetstone, UK.

Advanced controller design for aircraft gas turbine engines

In this paper, NARMAX and neural network representations are shown to be suitable for modelling a gas turbine engine. These models then provide the basis for the model-based control strategies. Due to the nonlinearities of the engine and constraints on the fuel feed, a proportional, integral and derivative (PID) controller cannot cope with its whole operating range and therefore a gain-scheduling PID controller is required. Since the parameters in the gain-scheduling PID controller need to be changed with the operating range, there is a need for a global nonlinear controller. Model predictive control is then proposed as a solution to this problem. Two methods are considered, one is approximate model predictive control (AMPC) using the instantaneous linearisation of a nonlinear model incorporating the generalised predictive control and the other is the nonlinear model predictive control (NMPC). Their properties are presented and discussed. Both provide good control performance for the engine across its whole operating range. Since AMPC can be obtained analytically, requires less computation time and avoids the local minimum, it also provides the better control performance in the face of disturbance and model uncertainties. So it is preferred to NMPC. The results illustrate the improvements in control performance that can be achieved to that of gain-scheduling PID controllers.

Reducing Brownian Motion in an Electrostatically Tunable MEMS Laser

Brownian motion of a tunable vertical-cavity surface-emitting lasers (VCSEL) movable microelectromechanical systems (MEMS) mirror is shown to broaden the spectral width of the laser emission line. A low-noise wavelength feedback system is presented and evaluated. This feedback system is employed to reduce measured linewidth from 1050 to 400 MHz fullwidth at half-maximum (FWHM), an improvement of 62%. The effects of device design parameters on closed-loop Brownian motion levels are analyzed. The potential increases in performance resulting from vacuum packaging and advanced controller design are discussed.

Current‐mode control of DC–DC power converters: a backstepping approach

AbstractCurrent‐mode control is a robust industrial DC–DC power converter control scheme that has been successfully tested, is widely accepted, and is commonly implemented with conventional linear P and PI controllers. The need of more systematic designs with enhanced performance has motivated recent studies with linear and nonlinear advanced techniques, but the rigorous understanding and substantiation of the underlying key robust functioning capability are still lacking. In this paper, a constructive control approach is employed to study the current‐mode control problem of a class of DC–DC power converters, yielding a cascade control design methodology with: a robust convergence criterion coupled with a systematic and transparent construction‐tuning procedure, a rationale to explain the robust controller functioning, and a unified framework to interpret and compare the proposed controller with the existing conventional and advanced control designs. The implementation and functioning of the proposed control design is illustrated experimentally with a boost converter. Copyright © 2003 John Wiley & Sons, Ltd.

The poloidal fields control system of the globus-m tokamak

Abstract The Poloidal Fields Control System of the Globus-M tokamak is described in this article and some attention is focussed on design of the feedback controllers, which is key to tokamak operation stability. Two different controllers can be used for achieving the requisite piasma parameters. The most advanced shape controller design methods were recommended. Two types of controller (Proportional differential - PD and Linear Quadratic Gaussian - LQG), based on the described controller synthesis methodology, were examined and compared by using analysis of the important dynamie characteristics that determine the system stability. The choice of the controller that is intended to be used in the project was made.

Linear MIMO controller design for an industrial reactive distillation column

Tbe aim of this paper is to demonstrate the use of advanced linear controller design methods for an industrial scale reactive distillation column. The open-loop analysis of the process behavior indicates that linear control of the process is promising. Therefore, a control structure selection is made by an analysis of the physical behavior of the system and a RGA-analysis. Afterwards a frequency response approximation technique is used for the design of linear MIMO-controllers. Robustness of the closed loop is observed during the controller design. Finally, the controller is tested on the nonlinear process model.

Evaluation of Fin Roll Stabiliser Controller Design

The control of ship roll motion using fins or rudders has been extensively studied in the last decade. Many different control design techniques have been investigated but very few have found application in practice. The majority of fin stabilisation control systems are still based on PID controllers. The objective of this paper is to identify the benefits of advanced control design techniques and possible advantages, which they might have over their classical counterparts. Three control design techniques namely, classical PID, optimised PID and H∞ controllers are compared and frequency and time domain performance indices are calculated to demonstrate the advantages and disadvantages of each method.

Use of Modified Relay Techniques for the Design of Model-Based Controllers for Chemical Processes

In this paper the possibility of using relay techniques to build parametric models for the design of advanced controllers is analyzed. A modified relay technique (ATV+ method), which requires more experimental runs and can lead to the design of high-order controllers, is compared with the “two-channel relay” (TCR) technique, which requires simpler experimental tests and leads to PI/PID control. The comparison is carried out for a large number of typical process dynamics by considering ease of application, duration of experimental tests, and achievable performance. For a quantitative evaluation of these features, several indexes are defined and simulations of time responses are analyzed. As a synthetic conclusion, performance can be largely improved by adopting the ATV+ method, with model-based controllers, at the expense of a longer duration of experiments. The time required for TCR depends on initial knowledge of the process. This paper gives more details to quantify the tradeoff between performance and times. The presented results can be used as a guideline for the process control engineer to help in the choice of the more suitable technique, according to the real needs of the plant.

LFT models of continuous biotechnological processes

The aim of the paper is to obtain suitable state-space models of continuous biotechnological processes (CBTP) in the framework of Linear Fractional Transformation (LFT). The LFT models are starting point in most of the advanced robust control design and analysis methods. Therefore, a linearized process model in the state-space is used whose elements are supposed to vary within certain bounds to represent the nonlinear behaviour of the real plant. The performance specifications are defined in the frequency domain through weighting functions. Two LFT models of CBTP are obtained ready for controller design aimed to optimize robust stability margins and robust performance, respectively.

Design of industrial automated systems via relay ladder logic programming and Petri nets

For the past few decades, programmable logic controllers (PLCs) using relay ladder logic (RLL) programming have been the workhorse for controlling event-driven industrial automated systems. RLL proved to be flexible compared to the hardwired RLL control implementation, due to its feature of software implementation. As automated systems become more complex, they also become more difficult to understand and maintain. It takes tremendous effort to accommodate specification changes (which are becoming frequent) to meet today's flexible automation needs. Several methods are emerging to overcome the shortcomings of RLL. Petri nets (PNs), initially proposed as a modeling tool, have been developed as such a method. This paper adopts an industrial-scale system to compare RLL and PN design methods so that the advantages of PN-like approaches are fully recognized. The criteria are: (i) the understandability that relates to the ability to evaluate the programmed logic, to verify its correctness and to maintain the control system, and (ii) the flexibility that relates to the easy modification of logic when the specification changes. This network takes an existing industrial system, conducts discrete event control designs by using both RLL programming and PN methods, and performs a comparative study on them. Together with previous comparison results using small-scale systems, the results of this study support that PN-like advanced discrete event control design methods are better than RLL in terms of the understandability and flexibility of the resulting control design.

Control of flexible spacecraft using nonlinear approximation of input shape dependence on reorientation maneuver parameters

This paper presents a design of a novel advanced feedforward controller for a large-angle reorientation maneuver of a spacecraft with flexible appendages, such as a telecommunication satellite. The controller is capable of fast computation of an optimal feedforward input shape for a nonlinear flexible system. This is achieved through radial basis function (RBF) network approximation of the feedforward dependence on the maneuver parameters. The paper describes a flexible spacecraft testbed and studies, the application of the controller to the testbed control. It presents a few insights into the design, and performance analysis of the RBF network approximation. The designed controller is demonstrated to combine excellent performance in the compensation of residual flexible vibrations for the nonlinear system under consideration, as well as computational simplicity.

Integrated Environmental Control Modeling of Coal-Fired Power Systems

The capability to estimate the performance and costs of advanced environmental control systems for coal-fired power plants is critical to a variety of planning and analysis requirements faced by utilities, regulators, researchers, and analysts in the public and private sectors. This paper describes a computer model developed for the U.S. Department of Energy (DOE) to provide an up-to-date capability for analyzing a variety of pre-combustion, combustion, and post-combustion options in an integrated framework. A unique feature of the model allows performance and costs of integrated environmental control concepts to be modeled probabilistically as a means of characterizing uncertainties and risks. Examples are presented of model applications, comparing conventional and advanced emission control designs. The magnitude of technological risks associated with advanced technologies now under development are seen to vary markedly across applications. In general, however, integrated environmental control concepts show significant potential for more cost-effective methods of emissions control.

State-space adaptive H∞ controller with application to a ship control system

This paper is concerned with the design of a state-space H∞ self-tuning surface ship auto-pilot. A state-space approach is employed for the on-line recursive identification which lends itself readily to the H∞ problem formulation. To improve the robustness properties of the identification scheme, vector variable forgetting factor and UD factorisation of the covariance matrix are used. Two different methods for the controller design are described and conditions are given that allow suspension of parameter estimation and controller updating. Both approaches are used to design controllers that are robust to changes in ship speed and sea-states. The benefits of an advanced control design and the advantages of self-tuning capabilities are illustrated by means of non-linear simulations.

Development of Semi-Active Truck Suspension

Within the EC project Copernicus SADTS (Semi-Active Damping of Truck’s Suspension and Its Influence on Driver and Road Loads) the semi-active suspension for Liaz truck prototype is being developed. The objective is to decrease the dynamic road-tyre forces in order to reduce the road damage. It was necessary to develop a suitable truck prototype, its simulation model, different control concepts and to implement the semi-active suspension. The paper describes the development stages and the so far achieved results. The full mathematical model was the basis for simplified models for control designs and served as a reference model for comparison of different control concepts. The spatial three dimensional simulation model has been developed and verified. The base model includes also the reference inputs of excitation and the different criteria for performance evaluation. The purpose of the model was to investigate the vertical dynamics though with the influence of antisymmetry of uneveness and lateral motion. A number of different control concepts have been considered. Possible suitable control concepts have been selected from existing classical as well as advanced control design methods. Also several within this project newly developed approaches have been added. There have been considered the control concepts of simple approaches (sky hook), classical approaches (LQR), multi-objective multi-parameter optimization (ΜΟΡΟ), fuzzy and neural networks, robust control design. The newly developed control concepts are ground-hook and combination sky/ground-hook, robust multi-objective optimization based on Taguchi approach, fuzzy control based on optimization, structural approach (separate control of frame and cabine suspension). The performance criteria for evaluation have been road forces, driver’s seat and load accelerations, robustness against change of parameters. The comparison has been done on the simulation model in SIMPACK program package connected with MAT-LAB/SIMULINK. Very promising results have been achieved up to now by combination of sky-ground hook and by the classical LQR control. The evaluation of other concepts is still in progress. Hardware and field tests are scheduled for 1997.