ART.I.CO. (ARTificial Intelligence for COoling): An innovative method for optimizing the control of refrigeration systems based on Artificial Neural Networks

Advanced control methods proved they effectiveness in reducing energy consumption of refrigeration systems equipped with a variable-speed compressor, but they could not be suitable for fixed-speed compressors, which are usually controlled by a simple ON/OFF logic with a mechanical thermostat, which does not allow to optimize the performance of such devices. Hence, a novel control method based on the use of Artificial Neural Networks to optimize the operations of refrigeration systems equipped with a fixed-speed compressor is proposed. This technique uses an Artificial Neural Network, which stem from a three-step process, able to provide the ON/OFF control loop with the optimal hysteresis value accordingly to the requirement of the user, in terms of set-point temperature and optimization priority, and the ambient temperature. The proposed control method was encoded in a microcontroller to test its effectiveness with a refrigeration system. The results of the experimental tests demonstrated the great potential of this approach showing a reduction of energy consumption of 6.8% and 2.2% with no stored material and ambient temperatures of 25 °C and 32 °C, respectively. Then, the introduction of 45 kg of stored material led to energy savings up to 13.4% and 6.6% with ambient temperatures of 25 °C and 32 °C, respectively. Furthermore, it was evidenced that door openings and pick-and-place operations can reduce the positive effect of this approach, reducing the energy saving to 3.7%. The results show that Artificial Neural Networks can be successfully applied to optimize the ON/OFF control loop of refrigeration systems, considering both plug-in and built-in solutions.