Article: A construction behaviour prediction method for cable tunnels based on Qwen3-VL temporal enhancement Journal: International Journal of Information and Communication Technology (IJICT) 2026 Vol.27 No.79 pp.1 - 19 Abstract: To address the problems of low prediction accuracy and poor real-time performance in traditional cable tunnel construction behaviour prediction methods, a prediction method based on Qwen3-VL temporal enhancement is proposed. First, multi-source data related to cable tunnel construction (including construction equipment operation parameters, environmental monitoring data and process execution records) are collected and preprocessed to eliminate noise and data redundancy. Then, the temporal enhancement module of Qwen3-VL is optimised to enhance the model's ability to capture time-series features of construction behaviour, effectively mining the temporal correlation and evolution rules among multi-dimensional construction data. Finally, the enhanced feature set is input into the prediction network to realise accurate prediction of key construction behaviours such as non-standard operation, process deviation and equipment failure risk. Experimental results show that compared with traditional machine learning methods and general deep learning models, the proposed method improves the prediction accuracy by 8.3%-12.7% and shortens the prediction response time by 15%-22%, which can provide reliable technical support for safe and efficient construction of cable tunnels. Inderscience Publishers - linking academia, business and industry through research

Title: A construction behaviour prediction method for cable tunnels based on Qwen3-VL temporal enhancement

Authors: Weilun Xu; Jian Liu; Hao Wang

Addresses: State Grid Nanjing Power Supply Company, Nanjing, 210019, China ' State Grid Nanjing Power Supply Company, Nanjing, 210019, China ' State Grid Nanjing Power Supply Company, Nanjing, 210019, China

Abstract: To address the problems of low prediction accuracy and poor real-time performance in traditional cable tunnel construction behaviour prediction methods, a prediction method based on Qwen3-VL temporal enhancement is proposed. First, multi-source data related to cable tunnel construction (including construction equipment operation parameters, environmental monitoring data and process execution records) are collected and preprocessed to eliminate noise and data redundancy. Then, the temporal enhancement module of Qwen3-VL is optimised to enhance the model's ability to capture time-series features of construction behaviour, effectively mining the temporal correlation and evolution rules among multi-dimensional construction data. Finally, the enhanced feature set is input into the prediction network to realise accurate prediction of key construction behaviours such as non-standard operation, process deviation and equipment failure risk. Experimental results show that compared with traditional machine learning methods and general deep learning models, the proposed method improves the prediction accuracy by 8.3%-12.7% and shortens the prediction response time by 15%-22%, which can provide reliable technical support for safe and efficient construction of cable tunnels.

Keywords: cable tunnel; construction behaviour; prediction optimisation; Qwen3-VL; temporal enhancement.

DOI: 10.1504/IJICT.2026.155056

International Journal of Information and Communication Technology, 2026 Vol.27 No.79, pp.1 - 19

Received: 09 Jan 2026
Accepted: 16 Mar 2026
Published online: 24 Jul 2026 *

Title: A construction behaviour prediction method for cable tunnels based on Qwen3-VL temporal enhancement

Keep up-to-date