Abstract:

Objective To address the challenges in control of airborne hazardous substances, such as toluene diisocyanate (TDI), during adhesive manufacturing, including the management of fugitive emissions, the low efficiency of conventional protective facilities, and the insufficient monitoring and early warning capabilities, an intelligent control system against hazardous substances was developed, and its engineering application effectiveness was evaluated in order to serve as a technical reference for occupational hazard prevention and control in similar workplaces.

Methods The field investigation was conducted to identify the major sources of hazardous substances and to evaluate the deficiencies of existing protective facilities. An integrated intelligent protection system was then established by incorporating dynamic regulation of health, safety, and environment (HSE) parameters, multi-sensor data fusion, a CNN-LSTM-Attention hybrid model for concentration prediction, and digital twin technologies. Finally, the system, s effectiveness was evaluated through on-site actual measurements, data analysis, and comparative evaluations before and after the retrofit, with performance indicators including TDI concentration at workplaces, the local exhaust ventilation control velocity, the energy consumption, and the management performance.

Results After the retrofit, the control velocity of the local exhaust ventilation system increased from (0.15 to 0.21) m/s to (0.83 to 1.02) m/s, meeting the requirements of relevant technical specifications. The short-term average concentration of TDI at workplaces in the adhesive preparation room decreased from 1.53 mg/m³ to 0.14 mg/m³, representing a reduction rate of 90.85%. The CNN-LSTM-Attention model demonstrated good predictive performance for concentration changes of airborne hazardous substances, with a coefficient of determination (R²) of 0.954. Following the intelligent upgrade, the average operating frequency of ventilation fans decreased by 30.0%, the average daily runtime of air-conditioning systems decreased by 66.7%, and the energy consumption per unit of captured substance decreased by 54.0%. The sensor data accuracy increased from 82.5% to 96.2%, and the emergency response time was shortened from 6 min to 2 min.

Conclusions The intelligent retrofit of control facilities against hazardous substances can effectively improve the capture efficiency of airborne hazards in adhesive production workplaces. It enabled a closed-loop management system encompassing real-time monitoring, trend prediction, automatic regulation, and visualized management. This approach not only reduces occupational hazard risks but also achieves energy savings and operational optimization, demonstrating the considerable potential for good engineering application.