| Popis: |
The typical process modeling and control approach of dealingwith each manufacturing process stage in isolation is re-evaluated anda new approach is developed for the case of a series ofprocess stages and the case of several similar processesoperating in parallel at different sites. The serialmulti-stage modeling framework uses the possible correlationbetween stages as a tool to achieve tighter control of anonstationary process line. The case of several similarprocesses operating in parallel at different sites is termedglobal manufacturing. Global manufacturing allows updatedestimation of the sensitivity or slope parameters over thecourse of the process in some cases since the site effectcauses a spread in the range of the inputs. This spreadin the inputs provides an opportunity for stable sensistivityestimates without perturbing the process as is usuallynecessary. The Bayesian parallel site estimation procedure isshown to have broad application to any scenario where thedata is collected at various related but not identical sites.Specifically, uniformity modeling is explored using theBayesian estimation procedure. The multi-systems operation and controlmethodologies developed for the serial and global manufacturingcases provide valuable tools for the improvement ofmanufacturing processes in many industries. |
