Production Planning Decision Based on the Optimized Fuzzy Time-series Clustering

Dynamic market uncertainties lead to the complexity of production planning in semiconductor manufacturing factory. Interrelationships instead of independency among different products planning significantly impact the decision-making process and results. Moreover, such interaction-based production planning almost stays at quantitatively study. To well characterize such interactions caused by market dynamic uncertainties, clustering method is proposed for quantitative analysis, which paves the way for a low risk production planning. This paper presents an optimized Fuzzy Short Time-series (FSTS) Clustering method to study the tendency of production data, where Fuzzy Subtractive Clustering is introduced to identify the categories numbers, Genetic Algorithm (GA) is applied to confirm the initialization of fuzzy central matrix, and Weight-FSTS Clustering is employed for better trend description. Numerical experimental data in a semiconductor manufacturing factory show the feasibility and effectiveness of this optimized FSTS Clustering method.