Mining activities in Côte d'Ivoire have led to significant environmental degradation, necessitating effective monitoring systems. A combination of machine learning algorithms and sensor fusion techniques was employed to design and test the proposed system. The system achieved a detection accuracy of 95% with minimal false positives, indicating its effectiveness in identifying environmental anomalies in real-time. The developed sensors and IoT systems have demonstrated significant potential for improving environmental management at mining sites. Further field tests are recommended to refine the system before full-scale deployment across all mining operations. Environmental Monitoring, Mining Sites, Sensors, Internet of Things (IoT), Machine Learning The maintenance outcome was modelled as $Y_{it}=\beta_0+\beta_1X_{it}+u_i+\varepsilon_{it}$, with robustness checked using heteroskedasticity-consistent errors.