| 摘要 |
<p>An industrial process plant has data concerning its inputs and outputs fed to a control apparatus which optimizes operation by an optimization procedure using linear programming. The plant includes components which can be switched between idle and run states. A process model representing the plant includes a plurality of constraint functions and an objective function which is to be minimized or maximized, the latter function including terms for each idle/run component representing a change in its status. A model updating procedure updates the parameters of the process model in accordance with the current process state, including setting the signs of the coefficients of the idle/run component terms in dependence on the current statuses of those components so that each such term makes a positive cost contribution to the objective function in the event of a change of status. A scheduling procedure operates the optimization procedure at fixed intervals and/or whenever the current state differs sufficiently from the most recent state set by the optimization procedure, or under operator control.</p> |
