Abstract
Mobile, small-scale refrigeration applications for last-mile deliveries have gained increased importance in recent years. As they face disturbances and frequent door openings more extensively than long-distance transport, reliable temperature information is crucial for control concepts to comply with temperature regulations and increase efficiency. For desired ruggedness, sensors are typically integrated within the cooling unit, yielding substantial deviations between actual air temperature in the cooling chamber and measured one - particularly in periods of altered airflow conditions resulting from fan switching and the actual door opening status. This article introduces and compares two hybrid, online estimation procedures to overcome this issue: firstly, a Kalman-filter approach based on a simple lumped physical heat transfer model, and secondly, a graybox-model approach resulting from a realizable inversion of the physical model. Experimental investigations of typical operating profiles provide 14h of real-world data to parameterize (11.75 h data length), validate (2.25 h data length), and compare both estimators. The proposed concepts are based on a sensor setup available in state-of-the-art system architectures and provide satisfactory temperature estimates with more than 83 % overall fit. As the algorithms provide comprehensive process insight independent of the actual operating condition, sophisticated control schemes can be built upon the concept.
NSTL
Elsevier