The role of process identification in optimization of manufacturing systems is discussed in general terms. Optimization of a work-restricted forging process is considered in detail and it is shown that its identification consists of estimating the parameters of a normal regression model. Since forging parameters are variable, routine updating of identification estimates is required. The response of the sequence of updated estimates obtained after a process change has occurred is of importance to the effectiveness of optimization. The problem is examined by applying two methods of updating to data which stimulate the occurrence of a change in the parameters of the forging model. The first is valid for stationary processes only and yields long sequences of inaccurate estimates after a process discontinuity. The second method recovers the accuracy in fewer updating steps but does this at the expense of a greater sampling variability of the estimates when the process is steady. A test for process homogeneity is employed to decide at every stage of updating whether the sequence should be continued or a new one started from fresh.

This content is only available via PDF.
You do not currently have access to this content.