Teknik Tiga Paras Terimbuh untuk Kawalan Optimum Penyesuaian Keadaan Pegun Proses Berskala Besar

The augmented three level technique is developed with the aim to extend the application of the three level structures without information exchange between local control units to a non-linear problem. The algorithm is a hierarchical in structure and the solution obtained is the control set-points to satisfy the Kuhn-Tucker necessary condition via integration of the model based optimization with model parameter estimation. The algorithm structured such a manner that it utilizes efficiently the available process mathematical and real process measurement information. The advantages it has over the same class of hierarchical control algorithms are that it does not require real process derivative information and there is no information exchange between local control units. These control algorithms are solved in two stages which are internal and external. Internal optimization consists of two levels which are Price coordinator level and constrainted optimization level while Powell conjugate direction method is used in the external optimization to calculate Lagrange coefficient, λ This control structures have been tested using four simulation problems. Simulation results show that the augmented technique converges faster thus reduce the set point changes than normal technique method at chosen values of Lagrange multiplier, p and update coefficient, ε for convex problems, ε range from 1.0 to 1.2 is suitable for the structures. However augmented Lagrangian function can only solve non-convex problem by using large value of (rho). Generally, control structure with input-output feed back has less control set points changes and converge faster compared with control structure with output feed back only.