International. A group of Italian researchers presented a review of the performance of 19 refrigeration plants in which artificial intelligence algorithms were implemented.
Artificial intelligence (AI), including machine learning (ML), artificial neural networks (ANN), and deep learning (DL), which are subfields of AI, can generate significant benefits in terms of operation and maintenance of refrigeration and air conditioning systems.
Fault detection and diagnosis (FDD) for predictive maintenance and improvement of machine operations is one of the most studied AI applications in these fields. As predictable failures, the authors give examples of refrigerant charge defects, in terms of overload or leakage/insufficient load, and heat exchanger failures in terms of fouling, secondary fluid mass flow, and reduced heat transfer.
Using AI algorithms instead of physics-based ones can be interesting if a reduced calculation time is required and if the cumulative operational effects over time are considered. FDD enables better scheduling of maintenance operations to avoid serious failures and limit costs related to minor failures.
AI is also being widely used for predicting the energy performance of refrigeration equipment. Energy performance prediction is useful for establishing the optimal control strategy and maximizing energy savings. For example, predicting the COP of a geothermal heat pump with an ANN algorithm using soil temperature and air temperature of the inlet and outlet condenser as input variables resulted in an error of only 1% between the measured and predicted COP.
Predictive control is an important application of AI. Thanks to their simplicity and reduced computing time, ANN algorithms can be implemented in the main control boards of operating systems in real time. These novel control systems can push the boundaries of widely used thermostatic and PID controls by providing a shorter response time during transient operation, achieving significant energy savings.
As an example, the authors present the modeling of a 4-cooler plant with a two-level algorithm to minimize operating costs. At the first level, a genetic algorithm is used to predict the on/off state of the cooler based on the cooling charge. At the second level, particle swarm optimization is used to minimize system energy consumption using cooling capacity, water temperature difference, and enthalpy as main inputs. During two days of operation, energy savings of 14% were achieved; In addition, thanks to a high calculation rate, it is possible to implement this algorithm to control the entire plant in real time.
On the other hand, little work has been done on the prediction of frost formation and on the techniques of optimizing thaw control, which shows that this area of research is still to be explored.
Source: International Refrigeration Institute.
