Optimizing Two-Dimensional Cutting Algorithms for the TFT-LCD Industry using Multi-Objective Strategies

Thin-film transistor (TFT) displays, celebrated for their high resolution and color accuracy, dominate the current display market.However, during the Thin-film transistor liquid crystal display (TFT-LCD) panel cutting process, reliance on traditional batch production methods persists among manufacturers.This approach results in several drawbacks, including excessive consumption of glass substrates, extended cutting paths, and increasing costs due to inefficient utilization of residual materials.To address these challenges, this study draws on concepts from machine learning to introduce a multi-objective adaptive (MOA) cutting stock algorithm.The algorithm employs a state selection matrix to optimize the sequence of blank selection at each cutting step and incorporates a blank cutting criterion to maximize the value of leftover materials.Experimental results indicate that the algorithm reduced production costs by approximately 15% on real-world data from a TFT-LCD factory.