Optimum operation point search system of the air conditioner components using deep reinforcement learning algorithm

摘要

In the air conditioner field, even if the optimum design point or thermodynamic simulation (Ph lead point, refrigerant flow design, air flow design) of each component of the product is not known, only the specification accuracy of each component is given. If matched, it relates to a system that finds the operating point of each component to exhibit optimal product energy efficiency in the assembled state of each component of the product using a deep reinforcement learning algorithm. In particular, an air conditioner assembled including components of a compressor, an expansion valve, an indoor unit fan, and an outdoor unit fan to enable cooling or heating operation; A first computer receiving the number of revolutions (COM) of the compressor, the number of revolutions of the expansion valve (EEV), the number of revolutions of the indoor fan (IF), and the number of revolutions of the outdoor fan (OF) from the air conditioner; And power consumption (P(t)), cooling/heating capacity (C(t)), cooling/heating efficiency (EER(t)), slope of cooling/heating efficiency according to time of the air conditioner running from the air conditioner (dEER(t )/dt), and a second computer that receives data of a cooling/heating capability (Cset) set in advance as a target value of the air conditioner and provides it to the first computer, wherein the first computer is provided from the second computer. Through the respective data and each data provided from the air conditioner, in-depth reinforcement learning is performed through a neural network structure previously installed, and the number of revolutions (COM) of the compressor capable of operating the air conditioner at the highest efficiency, and the opening degree of the expansion valve The present invention relates to an air conditioner component operation point search system using an in-depth reinforcement learning algorithm that automatically searches for (EEV), the number of revolutions of the indoor fan (IF), and the number of revolutions of the outdoor fan (OF).