A search method for process optimization with designed experiments and some observations

What is it about?

Many products or processes have factors, which are to be set at particular values or states so that the product/ process is optimized, in the desired way. (For example, in making a cake, the oven temperature and the time for which the cake is baked in oven are two possible factors; these are to be set at decided values, so that the cake is "good".) However, alongside the control factors, there may be some other things or factors, which may result into some noise (or fluctuations, or errors) in the product/ process. We need to do some experiments to arrive at the optimum level of the factors, in the presence of such noise. Often, there may be a large number of factors, and each having many possible values. In this article, we suggest an approach, which has a search method as a part of it, to arrive at optimized/ satisfactory levels of the factors.

Featured Image

Why is it important?

The approach takes into account many relevant issues. The search method, as is seen in numerical experiments, appears to work well in quite generalized set-ups. The approach may be well-suited in many practical instances.

Perspectives