Condition-aware supervisory control for automated particulate matter calibration: A simulation study of the ManDust system

Accurate particulate matter (PM) calibration requires not only stable concentration regulation but also reliable measurement validation, abnormal-condition recognition, and traceable supervisory operation. This study proposes a condition-aware supervisory framework for the automated ManDust PM calibration system. The framework integrates feedforward control with proportional–integral–derivative (PID) feedback, measurement validation, actuator constraints, condition monitoring, operational-state classification, alarm management, maintenance-oriented decision support, and programmable logic controller (PLC) and supervisory control and data acquisition (SCADA) functions for operational traceability. The framework was evaluated through simulation at PM concentration targets of 400, 800, 1000, and 1600 µg m−3 using 144 six-hour sessions comprising 51,840 one-minute observations. Compared with conventional PID control, the full condition-aware framework reduced normalized root-mean-square error from 3.089% to 1.453%, increased time within the ±5% target band from 87.05% to 99.54%, and reduced median disturbance-recovery time from 91.8 to 5.5 min, while total control variation changed only from 0.524 to 0.519. Under simulated abnormal conditions, detection rates were 95.8% for feeder-efficiency loss and 100% for reference-PM measurement dropout and PLC-SCADA communication interruption, with false-alarm rates of 0.7–2.1%, and median detection delays of 1–3 min. Functional ablation distinguished concentration-control functions from the additional supervisory capabilities provided by measurement validation, constraint handling, and fault/state supervision. These results demonstrate the simulation-level feasibility of the proposed architecture; experimental validation of closed-loop ManDust performance remains necessary before physical implementation and metrological validation.