Smart Occupational Safety Management Through IOT Sensor Networks, Machine Learning, and Real-Time Risk Assessment in Chemical Processing Plants

Abstract

This study addresses the persistent problem that conventional occupational safety systems in chemical processing plants often rely on periodic inspection, fixed alarms, and post-incident review, which are insufficient for detecting rapidly evolving hazards in high-risk environments. The purpose of the research was to examine whether smart occupational safety management can be improved through the integration of IoT sensor networks, machine learning, and real-time risk assessment in chemical processing plants using a quantitative, cross-sectional, case-based design. Data were collected from 200 valid respondents drawn from enterprise plant cases, including operations, maintenance, safety and compliance, process control, and supervisory roles, out of 220 distributed questionnaires, yielding a 90.9% valid response rate. The key independent variables were IoT sensor networks, machine learning, and real-time risk assessment, while smart occupational safety management was the dependent variable. Analysis was conducted using SPSS through descriptive statistics, Cronbach’s alpha reliability testing, Pearson correlation, and multiple regression. The findings showed high mean scores for all core constructs, including IoT sensor networks (M = 4.18, SD = 0.61), machine learning (M = 4.06, SD = 0.67), real-time risk assessment (M = 4.24, SD = 0.58), and smart occupational safety management (M = 4.29, SD = 0.55). Reliability was strong, with Cronbach’s alpha ranging from 0.84 to 0.90 and an overall instrument reliability of 0.91. Correlation results indicated significant positive associations with smart occupational safety management for IoT sensor networks (r = .71, p < .001), machine learning (r = .68, p < .001), and real-time risk assessment (r = .76, p < .001). The regression model was highly significant, F (3,196) = 145.27, p < .001, explaining 69% of the variance (R² = .69), with real-time risk assessment emerging as the strongest predictor (β = .38), followed by IoT sensor networks (β = .31) and machine learning (β = .27). The study implies that chemical plants should adopt integrated, data-driven safety architectures to strengthen hazard detection, emergency readiness, and reliability-oriented plant safety performance.