S. Thompson

Modelling and prediction of NOx emission in a coal-fired power generation plant

Abstract In this paper NOx emissions modelling for real-time operation and control of a 200 MWe coal-fired power generation plant is studied. Three model types are compared. For the first model the fundamentals governing the NOx formation mechanisms and a system identification technique are used to develop a grey-box model. Then a linear AutoRegressive model with eXogenous inputs (ARX) model and a non-linear ARX model (NARX) are built. Operation plant data is used for modelling and validation. Model cross-validation tests show that the developed grey-box model is able to consistently produce better overall long-term prediction performance than the other two models.

A comparison of idle speed control schemes

Abstract This paper examines the idle speed regulation control problem in multi-point spark ignited petrol engines. Several possible solutions are presented, including proportional plus integral control, fuzzy logic control, adaptive fuzzy logic control, adaptive fuzzy logic control in conjunction with Smith prediction and dynamic matrix control. All of the controllers are compared in simulation and, where possible, on a production vehicle. The performance measures used for comparison purposes were mean-square error and maximum error. It is shown that there are several possible alternatives to the existing proportional plus integral control used on the air bypass valve of production vehicles.

A case study of fundamental grey-box modeling

Abstract Fundamental grey-box modeling is an approach that can be used to model complex nonlinear dynamic systems for which the underlying mechanisms are either too complex or only partially known a priori. In this paper, further discussion is made on the motivations behind this method and the framework of this method is also extended. As a case study, it is used to model the NOx emission in a coal-fired power generation plant.

Disturbance decoupling in descriptor systems via output feedback-a parametric eigenstructure assignment approach

Based on a complete parametric approach for eigenstructure assignment in descriptor linear systems via output feedback, disturbance decoupling using output feedback in descriptor linear systems is investigated. Both the dynamical and static disturbance decoupling problems are tackled. Necessary and sufficient conditions for both problems are proposed in terms of the closed-loop eigenvalues and eigenvectors. By arranging these conditions into constraints on the design parameters provided by eigenstructure assignment, the disturbance decoupling problems are converted into eigenstructure assignment problems with extra parameter constraints. The approach guarantees closed-loop regularity, offers certain flexibility and can provide all the degrees of design freedom. An example is investigated to show the effect of the proposed approach.

Eigenstructure assignment design for proportional-integral observers: the discrete-time case

A complete parametric design approach for proportional-integral observers of multivariable discrete-time linear systems is proposed based on an eigenstructure assignment technique. Complete parameterizations for all the observer gains as well as the eigenvector matrix of the observer system matrix are established in terms of three sets of design parameters that satisfy three basic and simple constraints. The proposed approach provides all the degrees of freedom and has great potential in applications. An illustrative example shows the effect of the proposed approach.

Separation principle for robust pole assignment-an advantage of full-order state observers

A separation principle for robust pole assignment in full-order observer-based control system designs is proposed, which reveals the fact that pole assignment with minimum sensitivities in a n-dimensional linear multivariable system using a full-order observer-based state feedback controller can easily be realized by solving two separate n-dimensional state feedback robust pole assignment problems.

Modelling of Combustion and Emissions for a Cyclic Power Plant

Abstract Within a power plant, it is highly desirable to improve efficiency and reduce pollutant emissions. Consequently this interest has led to the modelling of combustion and emissions which can be used for boiler operation and control system design. This paper presents a model of combustion and NOx emissions for a power plant boiler. The major inputs are fuel flow, air flow, and burner tilt positions. The outputs are concentrations of the components in the flue gas. The model is verified by experiments on a 300 MW boiler.

Heat transfer models for boiler fouling monitoring

This paper uses heat transfer models to monitor fouling from ash in a coal-fired power generation boiler. In particular, a counterflow heat exchanger model for the economiser and a periodic flow model for the airheater are developed. Model inputs are local measurements of flow, temperature, etc, and the outputs are fouling indices in real time. Results from tests on a transient-load full-scale power station are piesented for both models. Recurring fouling patterns are shown to suggest relationships with time, load and with changes in the in-service cleaning system.

Root-locus algorithms for microcomputers

root-finding routine (Melsa, 1970; N-Nagy and AlTickriti, 1972; Richard et al, 1979; Thomas, 1976; and Vernon, 1967). However, if a microcomputer is used, attempts to generate root-locus plots with a root-finding routine usually prove unsuccessful. The reason is due primarily to the architecture of the microcomputer, which makes its mode of operation relatively slow. Typically, a root-locus problem which may take a few seconds to solve

GA BASED NEURAL NETWORK MODELING OF NOX EMISSION IN A COAL-FIRED POWER GENERATION PLANT

Abstract Genetic algorithm-based neural network modeling is studied. A MLP model for predicting NOx emission in a coal-fired power generation plant is trained using genetic algorithms. In order to avoid over-training, two data sets are involved, i.e. one data set is used for searching the weights and bias, the other set is used for validation. The fitness function for the GA based training is the combination of the training error and validation error. The GA-based MLP model has been tested over different periods of operation, showing the merits of this modeling technique.