| 摘要 |
PURPOSE:To realize an excellent optimizing control by setting a limit time to the convergent calculation of a computer in response to the control period and repeating the converging calculation within the limit time so as to calculate the best evaluation function value thereby obtaining the excellent optimum solution. CONSTITUTION:When an initial value is set and the state where no optimizing solution is obtained is discriminated, the converging calculation is executed only once and the 1st optimizing solution is obtained. When the optimum solution exists, it is compared with the preceding calculation result, and when said solution is a better solution, it is preserved and the converging calculation is repeated until it is discriminated that the time reaches the limit time. The arrival of the limit time is discriminated by counting a unit calculation time (k) and discriminating that it reaches the limit count value k1. An object function value (f) (shown in dotted lines) is decreased as the calculation is repeated, a limit error (c) is decreased sequentially and it reaches a limit permissible error epsilon1 or below at k=5. In selecting k1=11 for example, the error (c) is decreased up to a value S2 and the better optimizing control is attained. |
