Every cellular glass factory has his “temperature curve expert”. These experts observe a cellular glass foam and know in a lot of cases how it can be improved by changing temperatures, foaming agents, …. . Their knowledge and availability have a strong influence on the quality and production yield of a factory.
In the past, managers have tried to describe their knowledge in a ISO9000 system or another system with procedures. This has never worked properly because the “book” knows always less than this person. This was always clear when this person was not present in the factory, the yield drops temporarily.
Ten years ago, we needed Kitt in Knight Rider to see a self driven car, but today Tesla and many others have converted this fiction into reality. And I guess that this “machine learning” technology will enter the (cellular) glass factories soon. I expect that neural networks will do the job, because the data for input and output are already available.
Neural networks (NN), programmed in Python, based on Numpy are a good candidate because they are open source and will be fast further developed. Indeed, the NN can be trained with the massive amount of data from the past and present. To obtain a new quality, the NN will find a new solution with data from the past like today the operator is doing by using his experience. Moreover, the NN of collaborating factories can be coupled and the operator experience will not be lost after retirement.