Controlando umidade relativa em câmaras frias: melhorias no sistema de medição e uma estratégia de controle baseada em sistemas de eventos discretos

Abstract

Apple cultivation is an agroindustrial activity that depends essentially on the existence of cold chambers to store fruits after harvest, under temperatures around 1°C and Relative Humidity (RH) between 92 and 95%rh. Controlling environments like these is challenging because of the size of the chamber, its entrances, and distinct zones, and especially because of the effect that temperature and RH impose on each other when orchestrated in conjunction. Typically, the refrigeration cycles associated with these conditions are expected to cause condensation in the RH sensor, affecting the monitoring and control of RH. This work proposes a strategy to improve RH control in cold chambers. It first introduces a novel Measurement System (MS) that tolerates high RH rates, condensation events, and low ventilation. The system encapsulates the measurement devices in a closed structure that is automatically controlled to detect saturation and respond to it accordingly. Then, based on the event map observable from the proposed MS, a formal event-driven model emerges as a way to represent the real system logically, as well as its rules for expected coordination under control. This model is processed using the Supervisory Control Theory (SCT) of Discrete Event Systems (DESs) and leads to a control strategy that effectively captures the relationship between refrigeration and relative humidity. The controller is then converted into C code using an automated generation tool, whose development is also part of the contributions of this paper. Experiments are conducted to show that the proposed MS reduces the sensor response time under low ventilation and improves the accuracy of detecting variations in RH. Furthermore, the control strategy, in conjunction with MS, enables the controller to anticipate variations of RH, ensuring more precise process control and reduced Root Mean Squared Error (RMSE) and standard deviation.