In the textile industry (understood as yarns and fabrics), production planning faces particular challenges in operational production planning due to the variability of weaves and fabrics depending on the seasons.
An additional challenge is fast fashion, which forces textile producers to work outside of orders (“in stock,” meaning they sell what has already been produced), with very tight delivery times. The greatest difficulty lies in identifying the colors to stock, as they depend on fluctuating trends, with variability of just a few weeks even within the same sales “season.”
The project involves creating an algorithm that, starting from sales data, forecasts orders for the following weeks, with the aim of automating operational production planning. Specifically, the algorithm will leverage hysteresis phenomena in purchasing behaviors, anticipating color/variant trends in the short term and updating operational production plans in the short term. Artificial Intelligence techniques will be used to develop the algorithm.
From a technical perspective, the algorithm will be engineered as a compute container, capable of communicating input and output with the main production MES systems.