Journal of Engineering, Project, and Production Management, 2026, 16(5), 2025-366
Multi-Node Supply Chain Demand Forecasting and Fluctuation Modeling Based on Transformer-LSTM Hybrid Model
Lecturer, School of
Economics and Management, Tianjin Vocational Institute, Tianjin, 300341,
China,
E-mail: jingdaihuakai138@163.com
Project and Production Management
Received December 29, 2025; revised March 4, 2026; accepted April 7, 2026
Available online June 17, 2026
Abstract: Existing research has limitations in co-modeling and quantifying uncertainties in demand sequences across multiple nodes in a supply chain, particularly in addressing local temporal patterns and global dynamic correlations. This paper proposes a Transformer-Long Short-Term Memory (Transformer-LSTM) hybrid model based on a dual-attention gated fusion architecture, denoted as Dual-Attention Gated Fusion Hybrid Model (DGFM). This model extracts local dependency features and global correlation features of the sequence through a parallel-running Long Short-Term Memory (LSTM) network and a Transformer encoder, respectively. A learnable gating weight matrix is applied to adaptively weight and fuse these two types of features, generating a hybrid representation with multi-scale spatiotemporal awareness. Using a quantile regression output layer, the representation generates a nonparametric conditional probability distribution that effectively captures demand fluctuations. The experimental results indicate that the global feature weight increases to 0.85 during promotional periods. The correlation coefficients of the DGFM model remain below 0.31 across all node pairs. This research provides a useful quantitative analysis tool for understanding demand perception and risk-taking in such complex supply chains.
Keywords: Supply chain demand forecasting, Transformer-LSTM hybrid model, volatility modeling; quantile regression, multi-node time series analysis.
Copyright © Journal of Engineering, Project, and Production Management (EPPM-Journal).
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License.
Requests for reprints and permissions at eppm.journal@gmail.com.
Citation: Feng, J. (2026). Multi-Node Supply Chain Demand Forecasting and Fluctuation Modeling Based on Transformer-LSTM Hybrid Model. Journal of Engineering, Project, and Production Management, 16(5), 2025-366.
