Optimization of Color Stability of Chinese Garden-Style Printed Fabrics: Dyeing and Finishing Process Based on GraphSAGE-BiLSTM

Lin Zhou

Article
2026 / Volume 9 / Pages 326-350
Received 30 July 2025; Accepted 22 September 2025; Published 28 February 2026
https://doi.org/10.31881/TLR.2026.326

Abstract
This paper addresses the technical challenges of dye migration (ink bleeding), attenuation of color fastness in high-saturation color blocks, and pattern blurring—unintended negative consequences arising from the pursuit of the artistic “ink-wash diffusion effect” in the dyeing and finishing of Chinese garden-style printed fabrics. An intelligent optimization method based on GraphSAGE-BiLSTM is proposed. By constructing a dyeing and finishing knowledge graph, the compatibility constraints of reactive dyes and color fixatives within the dye-auxiliary-fiber ternary interaction network, as well as the influence of process parameters, are quantified. A dual-channel cross-modal learning model is designed, utilizing GraphSAGE (Graph Sample and Aggregation) to learn the characteristics of the dye auxiliary network. BiLSTM (Bidirectional Long Short-Term Memory) decodes the dynamic effects of the evaporation temperature curve and washing sequence and optimizes feature weights via an attention mechanism. Finally, the Pareto optimal process solution is generated by combining the NSGA-II (Non-dominated Sorting Genetic Algorithm II) multi-objective optimization algorithm. Experimental results show that this method improves color fastness from level 3.0 (control group) to level 4.5 (on a 1–5 scale) and reduces color difference from 1.80 to 1.20 (a decrease of 33.3%). Additionally, it reduces the amount of auxiliary agents in group B with medium-intensity process parameters by 10% and the energy consumption of washing by 14.5%. This study enables collaborative modeling of structured rules and temporal dynamics in dyeing and finishing processes and provides an industrializable pathway for improving color stability in cultural and creative textiles.

Keywords
Chinese garden-style textiles, color stability optimization, dyeing and finishing process, GraphSAGE-BiLSTM model, knowledge graph