Implementation of AI-Based Chemical Dispersion Prediction and Safety Management Information System

Purpose: This study proposes an implementation framework for an intelligent safety management information system that integrates public big data and AI to preemptively respond to hazardous chemical leakage accidents in industrial complexes. The research aims to bridge the gap between technical analytics and managerial practice, establishing a new paradigm for intelligent disaster management within the framework of sustainable corporate governance. Research design, data and methodology: The research design utilizes public APIs from the Korea Meteorological Administration to collect multivariate time-series data. The methodology employs the Long Short-Term Memory (LSTM) algorithm to predict dynamic pollutant dispersion paths. A key methodological distinction lies in the standardization of complex predictive data into a unified 'Environmental Risk Index (ERI)' and the automation of real-time response processes through user-centered UI/UX design. Results: The findings demonstrate that the proposed system provides management with objective, data-driven evidence for high-stakes decision-making through risk quantification. Furthermore, it offers a cost-effective ESG management model for small and medium-sized enterprises (SMEs) by leveraging public data infrastructure instead of costly proprietary sensors, proving its practical efficiency in resource-limited environments. Conclusions: This system functions as a strategic information asset that significantly enhances organizational resilience and ensures industrial site safety. The study concludes that the integration of AI-driven predictive models into managerial information systems is essential for proactive risk control and sustainable corporate governance, providing a robust strategic mechanism for modern industrial safety.