Abstract

A time-varying mathematical model of an oil-fired boiler unit, with estimated parameters, is used to represent a poorly defined multivariable physical process subjected to input disturbances and measurement noise.The model is simulated in discrete form and is used to generate on-line control action related to a quadratic performance criterion.A parameter-estimation technique is included in the overall control logic, based on a time-domain sensitivity analysis of the system-transition and -driving matrixes, and is used for continuously updating the model to the process system which is subjected to continuous change and random input disturbances.A form of predictive on—off and bounded linear control is included within the control programme to give suboptimal performance based on minimisation of a two-period and current-performance index, respectively. Relatively close model following and/or improved control can be obtained with inputs based on the model reference system, and some slight improvement in performance is possible using dual-mode control action.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call