Construction Quality Hazard Management with Deep Learning–Based Multimodal Storage Strategy–Enabled Blockchain

Abstract

Hazard-related data are a critical component in construction quality hazard management (CQHM). However, data security and latency issues in CQHM cannot be guaranteed in centralized systems currently and prevent it from achieving the goals of secure and efficient hazard analysis and further real-time quality process control. Focusing on these goals, a decentralized CQHM framework is proposed by introducing blockchain (BC) and deep learning (DL) technology. Moreover, considering the blockchain’s limited storage capacity and block size, a deep learning–based multimodal storage strategy is designed with smart contracts and InterPlanetary File System (IPFS) for data lightweight. In accordance with the proposed framework, comparative experiments were conducted to demonstrate its feasibility by analyzing related metrics like accuracy, cost, and throughput. This study deepens the understanding of data security and latency issues in CQHM and offers technical guidance in establishing BC and DL solutions. Besides, the DL-based multimodal storage strategy provides a substantial data-driven advancement for lightweight on-chain data storage. Moreover, the proposed framework is promising to smooth the quality hazard analysis progress in improving on-site decision efficiency, promoting cooperation and standardizing quality process control.