Model-Based Development of a Diagnostic Algorithm for Central Inverter Thermal Management System Fault Detection and Isolation

This paper presents a method for developing an on-field diagnostic algorithm for a Thermal Management System (TMS) of a Central Inverter (CI) for solar fields. The first step relies on an in-depth qualitative analysis, making use of the physical knowledge of the system, through the Fault Tree Analysis (FTA), whose outcome is an inferential isolation tool called Fault Signature Matrix (FSM), which univocally links every fault to a set of symptoms detected during the on-field monitoring. A quantitative fault-to-symptoms sensitivity study is performed using a stationary thermodynamic model, previously developed by the authors, to simulate the system under faulty operating conditions. Four faults are simulated, like heat exchanger fouling or cold plate blockage, and in addition, the variation of the maximum junction temperature of the electronic components is examined to assess the impact of the analyzed faults on the performance of the plant’s thermal management system. The resilience of the FSM is evaluated by testing several fault magnitudes and symptom thresholds and thus a final modified matrix is obtained.