High-precision temperature control of DC energy metering device PCB based on an asymmetric fuzzy PID-Smith predictor

Temperature drift in metering-sensitive components can degrade the accuracy of DC energy metering devices, making PCB temperature stability an important design requirement. This study develops an asymmetric fuzzy gain-scheduled PID controller with Smith prediction for TEC-based PCB temperature regulation. The hardware platform uses an STM32F103C8T6 controller, an SHT30 temperature sensor, and a BTS7960B-based H-bridge TEC driver. The normalized MATLAB model is used only for controller-structure comparison; Fluent-UDF provides a three-dimensional numerical reference for thermal-field regulation; and hardware experiments establish the actual closed-loop response and steady-state accuracy. An effective delay of 1.5 s was estimated from preliminary Fluent transient simulations. To examine the reduced-order plant representation, paired Fluent open-loop cases with identical initial and boundary conditions were differenced to isolate the incremental TEC response. FOPDT fitting yielded a gain magnitude K F  = 247.35 °C per normalized duty-ratio unit and an equivalent time constant T F  = 176.06 s, with RMSE  = 0.0833 °C and R 2  = 0.9933. In matched-parameter ablation with common base gains ( K p , K i , K d ) = (8,1,4), the disturbance peak decreased from 1.349 °C for symmetric fuzzy PID to 1.261 °C after asymmetric regulation and to 1.183 °C for the complete asymmetric fuzzy PID-Smith structure. Under common re-tuning conditions, the proposed structure produced the smallest disturbance peak and the lowest composite objective among the evaluated controllers. Hardware tests at 25, 30, 35, 40 and 45 °C all satisfied the ±0.2 °C steady-state temperature-control requirement.