Resumo: We did some research on heuristically solving simultaneous lotsizing and scheduling problems by means of aggregation and decomposition. The idea came up to select the aggregation and disaggregation rules that fit best to a certain problem instance using a neural network. However, to start with, we simplified the problem setting and tried to forecast the CLSP heuristic that would "best" fit a certain CLSP problem instance:
https://www.researchsquare.com/article/rs-3709286/v1