Using mega-fuzzification and data trend estimation in small data set learning for early FMS scheduling knowledge

Provided with plenty of data (experience), data mining techniques are widely used to extract suitable management skills from the data. Nevertheless, in the early stages of a manufacturing system, only rare data can be obtained, and built scheduling knowledge is usually fragile. Using small data sets, this research's purpose is improving the accuracy of machine learning for flexible manufacturing system (FMS) scheduling. The study develops a data trend estimation technique and combines it with mega-fuzzification and adaptive-network-based fuzzy inference systems (ANFIS). The results of the simulated FMS scheduling problem indicate that learning accuracy can be significantly improved using the proposed method involving a very small data set. [PUBLICATION ABSTRACT]