Journal of Engineering, Project, and Production Management, 2026, 16(5), 2025-337
Improving Spatiotemporal Accuracy of Port Air Quality Monitoring Using GCN-Att Fusion Model
1 Deputy
Manager and Safety Director, First Stevedoring Branch Company, SPG
Rizhao Port Group Co., Ltd. 276800, China, E-mail: jdy7141319@163.com
(corresponding author).
2 Deputy Manager, First Stevedoring
Branch Company, SPG Rizhao Port Group Co., Ltd. 276800, China, E-mail:
zpw306@163.com
3 Deputy Director, Technology
Innovation Center, SPG Rizhao Port Group Co., Ltd. 276800, China,
E-mail: xuguannan_rzg@163.com
4 Deputy Department Manager,
Technology Innovation Center, SPG Rizhao Port Group Co., Ltd. 276800,
China, E-mail: qq909_long@163.com
5 Executive Director, Shandong Port
Technology Group Rizhao Co., Ltd. 276826, China, E-mail:
wd521cmm@163.com
Project Management
Received December 16, 2025; revised January 29, 2026; accepted March 1, 2026
Available online June 17, 2026
Abstract: The contradiction between dynamic pollution events in ports and the assumption of a static graph structure leads to a significant decline in the prediction performance of existing Graph Convolutional Network-Attention (GCN-Att) models under abrupt change scenarios. To address this issue, this paper develops a dynamic topology-sensing mechanism that constructs a dynamic spatial adjacency matrix integrating ship Automatic Identification System (AIS) trajectories and real-time meteorological data. A real-time edge-weight update unit driven by abrupt change events is designed to enable the graph structure to adaptively respond to the pollution diffusion process. Experiments show that in ship berthing events, the Root Mean Square Error (RMSE) of PM₂.₅ prediction is 4.21±0.63 μg/m³, with an R² of 0.89±0.04. Under extreme weather conditions such as typhoons, the model response latency is reduced to 8.1 seconds; the data missing tolerance reaches 86%; the Mean Absolute Error (MAE) of PM₂.₅ is only 4.7 μg/m³, significantly outperforming existing methods and validating its high accuracy, low latency, and strong robustness.
Keywords: Spatiotemporal monitoring, port air quality, graph attention networks, dynamic topology adaptation.
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: Zhang, S., Zhou, P., Xu, G., Mu, Z., and Wang, D. (2026). Improving Spatiotemporal Accuracy of Port Air Quality Monitoring Using GCN-Att Fusion Model. Journal of Engineering, Project, and Production Management, 16(5), 2025-337.
