Model-Based Fault Diagnostic Strategy for Microgrids

Microgrids are a topic of interest in recent years, largely due to their compatibility with the integration of distributed renewable resources, capability for bidirectional power flow, and ability to reconfigure to mitigate the effects of faults. Fault diagnosis algorithms are a foundational technology for microgrids. These algorithms must have two primary capabilities. First, faults must be detectable; it is known when the fault occurs. Second, faults must be isolable; the type and location of detected faults can be determined. However, most fault handling research considering microgrids has focused on the protection algorithm. Protection algorithms seek to quickly extinguish dangerous faults which can damage components. However, these algorithms may not sufficiently capture less severe faults, or provide comprehensive monitoring for the microgrid. This is particularly relevant when considering applications involving fault tolerant control or dynamic grid reconfiguration. Although well-accepted in the automotive and aviation field, model-based diagnostics have not been extensively applied to microgrid or power grid systems. Therefore, this work proposed a diagnostic concept relying on model-based diagnostics. Structural analysis is used to develop model-based residuals, and data from a Simulink microgrid model is used to define and calibrate diagnostic tests. It is shown that the use of model-based diagnostics yields better detection and isolation performance compared to overcurrent protection or differential current protection alone.