Energy-saving method and performance analysis of chiller plants group control based on Kernel Ridge Regression and Genetic Algorithm

Abstract

The upgrading of traditional chiller plants to the high-efficiency refrigerator room is conducive to effective utilization and conservation of energy. An energy-saving method of chiller plants group control based on Kernel Ridge Regression and Genetic Algorithm are presented. Combined with machine learning and the classic model of the chiller plants, the multiple nonlinear problems between the system parameters of the chiller plants are solved. A more detailed group control model of chiller plants is established, which is suitable for multiple chillers, variable-speed pumps, and variable-frequency fans. Incorporate the start-up time of the chiller, the difference between the actual cooling capacity and the target cooling capacity into the penalty function, and use the Genetic Algorithm for multi-dimensional optimization to find the best efficiency point of the entire refrigeration room and achieve the goal of energy saving. A comparative analysis of the actual operation of the cold room system of a large electronics factory in Guangdong for three months is carried out. Compared with the traditional PID algorithm, the three-month power saving rate of this group control energy saving method is: 31.34%, 14.33%, 19.22%.