Real-Time Prediction and Prevention of Worker Accidents and Safety Hazards on Construction Sites using Mobile Machine Learning Framework

Abstract

Construction sites are dynamic, complicated situations where maintaining safety is both important and difficult. Due to their reliance on labor-intensive and error-prone manual monitoring and intervention, traditional safety measures frequently fail. Furthermore, complicated indoor conditions are common on construction sites, making it difficult to precisely track the locations of workers and equipment. To overcome these issues, this study proposes a real-time prediction and prevention of worker accidents and safety hazards on construction sites using a mobile machine learning framework (RWASHCS-MMLF). On construction sites, mobile technology is utilized to gather real-time data from sensors, cameras, and wearable technology, allowing machine learning models to instantaneously forecast possible worker accidents and safety risks. In order to capture activity, the RWASHCS-MMLF model first collects data from the mobile sensors. The data from the sensors is then cleaned and filtered using data preprocessing. After that, temporal and spatial characteristics are extracted, including movement patterns, speed, position, and proximity to hazards. Mobile data is then incorporated into a trained machine learning model to instantaneously predict possible mishaps or dangerous occurrences. Lastly, the alert and decision system suggests preventive measures and allows push notifications or alarms on mobile devices for site supervisors’ visual dashboards. The efficiency of the RWASHCS-MMLF strategy in enhancing the usage of construction safety management systems and lowering the likelihood of future accidents and fatalities is demonstrated by the extensive on-site trials that validate the proposed RWASHCS-MMLF model. The result is a system with improved responsiveness and speed, which is essential for time-sensitive applications like worker safety prediction.