Comparative Analysis of Machine Learning Algorithms for Heat Exchangers Diagnosis in Electrified Aircraft

Optimizing performance in aircraft heat exchangers is crucial, especially for future electrified aircraft where effectively managing the heat produced by high-power electronics stands out as one of the most significant challenges. This study proposes leveraging advanced diagnostic techniques based on Artificial Intelligence (AI) algorithms to enhance efficiency and safety in heat exchangers. These techniques are proficient in assessing their state of health, thereby enabling the implementation of a Condition-Based Maintenance (CBM) approach. As fouling constitutes a primary failure mode for heat exchangers, this paper presents a comparative analysis of Machine Learning (ML) algorithms aimed at evaluating the severity of this specific failure mechanism. Due to the limited availability of data from flight operations related to the behavior of this device, the data for training and testing the data-driven algorithms was generated using a high-fidelity model of an aircraft’s Thermal Management System (TMS), capable of simulating its behavior under both healthy and degraded conditions.