Min-Sen Chiu

Robust PID controller design via LMI approach

Abstract In this paper, a method which allows explicit incorporation of the description of system uncertainties in the problem formulation for designing robust proportional-integal-derivative (PID) controller is presented. The multiple-model paradigm is employed to represent the uncertainties and provides the basis for the PID controller design. Using standard techniques, the robust PID controller design is reduced to a convex constraint problem which can be efficiently solved with linear matrix inequalities (LMIs) approach. Various practical specifications are considered simultaneously within this framework. Extension to system with time-delays, which follows naturally from the preceding development, is also discussed. Two examples are used to demonstrate the effectiveness of the proposed method and a comparison with the existing PID tuning methods is made.

Non-interacting control design for multivariable industrial processes

Abstract In this paper, a systematic method is proposed for the design of general multivariable controller for complex processes to achieve the goal of fast loop responses with acceptable overshoots and minimum loop interaction while maintaining low complexity of the feedback controller. The design of general transfer function type controller is based on the fundamental relations under decoupling of a multivariable process, and the characterization of the unavoidable time delays and non-minimum phase zeros that are inherent in the decoupled loops. The objective loop transfer functions are then suitably specified to achieve fast loop response taking into account the performance limitation imposed by those non-minimum phase zeros and time delays. The ideal controller is then obtained which is in general a complicated irrational transfer matrix, for which model reduction with recursive least squares is applied in the frequency domain to obtain a much simpler transfer matrix with its elements in the form of rational transfer function plus delay. Simulations show that very satisfactory control performance is achieved.

Further work on pulse-jet fabric filtration modeling

Further results of the performance of pulse-jet fabric filtration based on a recently developed model were presented. These results include filtrate rates, the built-up of particle cakes over filter surfaces and their coverage and the extent of filter cleaning from cycle to cycle until a steady state is reached. In addition, a simplified procedure of calculating the steady state filter performance was proposed. The applicability and limitations of the simplified procedure were established by comparing the results from the simplified procedure with those based on the more complete model.

An Effective Technique for Batch Process Optimization with Application to Crystallization

An iterative dynamic optimization methodology is developed for on-line optimization of batch processes in the presence of plant-model mismatch and measurable error. In the proposed method, the plant-model mismatch is effectively eliminated by using information from previous batches to modify the trajectories that are applied to the subsequent ones. In addition, the effect of modelling error on the convergence of this algorithm is investigated. The utility of the proposed method is illustrated through the end-point optimization problem in the batch crystallization process, and comparisons to other optimization methods are made.

Further Results on Expansion Design of Partially Decentralized Controllers: 2×2 Plant Cases

An expansion procedure to design partially decentralized controllers was recently proposed. This method requires the operation of pseudo-inverse of non-square matrices, which is not unique. Thus, it introduces an extra design freedom, which was not addressed in the previous studies. In this paper, the effect of the pseudo-inverse on the stability and performance of 2×2 partially decentralized control systems is studied. Specifically, the range of the pseudo-inverse factors arising from the stability condition of the control system is derived. In addition, an ‘ideal’ pseudo-inverse factor, which minimizes interaction in the control system, is obtained based on the concept of non-square block relative gain. Simulation results show that a partially decentralized controller with the ‘ideal’ pseudoinverse factor gives an improved controller performance.

A multiple-model approach to decentralized internal model control design

This paper presents a new decentralized internal model control (IMC) design method for controlling the multi-input multi-output plants that are subject to multiple operating regimes. Under this situation, the existing decentralized IMC design methods may become inadequate and conservative to be used for the resulting control problem mainly because they are developed based on a single model. To reduce the inherent conservatism in the existing design methods, a global multiple-model that accurately describes the plant in the operating space is incorporated into the proposed decentralized IMC design method. Simulation results illustrate that the proposed method is indeed less conservative than its conventional counterpart in controlling a plant with a range of operating points.

A Model for Pulse Jet Fabric Filters

ABSTRACT A new model for pulse jet fabric filtration is proposed. In contrast to the earlier model of Ravin and Humphries,1 which was formulated on the steady state assumption, the present study is aimed at developing a predictive capability for both transient and steady state operations, taking into account the compression effect of filter cakes. The models relative simplicity allows frequent updating of the model parameter values, thus improving the accuracy of predictions. As a result, the model is particularly useful in developing control algorithms and designing controllers of pulse jet fabric filtration systems.

The identification of neuro-fuzzy based MIMO Hammerstein model with separable input signals

A novel identification method of neuro-fuzzy based MIMO Hammerstein model by using the correlation analysis method is presented in this paper. A special test signal that contains independent separable signals and uniformly random multi-step signal is adopted to identify the MIMO Hammerstein process, resulting in the identification problem of the linear model separated from that of nonlinear part. As a result, the identification of the dynamic linear element can be separated from the static nonlinear element without any redundant adjustable parameters. Moreover, it can circumvent the problem of initialization and convergence of the model parameters discussed in the existing iterative algorithms used for identification of MIMO Hammerstein model. Examples are used to illustrate the effectiveness of the proposed method.

Robust decentralized controller design for unstable systems

Recently established methodologies for independent design of robust decentralized controllers are limited to open-loop stable processes. To facilitate independent design for open-loop unstable processes, new formulations are developed using the stable factorization approach (Vidyasagar, 1985, Control System Synthesis: A Factorization Approach. The MIT press, Cambridge, MA.). Application of the proposed design procedure to examples reveals that controllers for unstable systems can be designed to contain robustness properties. This result gives the control system designer more insight into the control of unstable processes.

Multiple-Objective Based Model Predictive Control of Pulse Jet Fabric Filters

With the increasingly stringent emissions regulations of fine particulates and air toxics, pulse jet fabric filters have become an attractive particulate collection option for utilities. Despite their wide application, the present control algorithm used in fabric filtration systems can best be described as rudimentary. In this paper, a model predictive control (MPC) technique is employed owing to its distinct advantages. Unlike its conventional counterpart, the proposed MPC algorithm takes the economic factor into consideration, which is formulated in a multiple-objective optimization framework. To avoid the local optimum, a global optimization technique is in corporated into the proposed MPC design. Simulation results show that the proposed multiple-objective optimization based MPC design method is especially suitable to the pulse jet fabric filtration process, where the set point change and process disturbance occur frequently.