A Novel Control Method of Sugar Boiling Based on Model-Free Adaptive Control and Neural Networks

To enhance sugar absorption and improve crystal uniformity, a novel adaptive control method for sugar boiling that integrates model-free adaptive control and neural networks is proposed to address the challenge of effective control owing to the complex nonlinear relationships. This method leverages the sugar brix deviation and sugar level deviation in the vessel to drive neural networks, which calculate the opening coefficients of hot water valves, syrup valves, and steam valves. Consequently, adaptive control was achieved during the sugar boiling process. The experimental results demonstrate that the apparent purity obtained by the proposed method reached 22.52%, surpassing both the manual method (with an apparent purity of 21.53%) and the PID-based method (with an apparent purity of 21.31%). Compared with these methods, the proposed approach excels in crystal uniformity, color value, and apparent purity difference. It is strongly recommended that this method be adopted for automatic control of sugar boiling to further enhance these critical parameters.