Long Short Term Memory utilized Photovoltaic Inverter Humidity Controller for Capacitor Reliability Enhancement
Conference
Roy, S, Khan, AS, Abu Taher, M et al. (2023). Long Short Term Memory utilized Photovoltaic Inverter Humidity Controller for Capacitor Reliability Enhancement
. 10.1109/ETFG55873.2023.10407373
Roy, S, Khan, AS, Abu Taher, M et al. (2023). Long Short Term Memory utilized Photovoltaic Inverter Humidity Controller for Capacitor Reliability Enhancement
. 10.1109/ETFG55873.2023.10407373
Photovoltaic (PV) string inverters, an expensive component, suffers from reliability and long-term stability issue due to ambient factors and cycling condition. Specifically, daily variations in humidity pose a detrimental effect on the film capacitor's health, potentially reducing its lifespan. This study starts with a comprehensive failure analysis of PV inverter film capacitors. Then To mitigate this, a Long Short-Term Memory (LSTM) model is used for time series humidity forecasting. The real field data fed prediction model's output is then linearly mapped to set a speed reference for a DC motor to allow the variable speed ventilation for the inverter. It effectively removes excess moisture from the encloser maintaining the internal humidity below a specified threshold. The accuracy in prediction and concerned dynamic tracking is found highly effective to satisfy the needed humidity conditioning. The benefit is to prevent the degradation of internal components (i.e. film capacitor) within the inverter suffered from humidity absorption. This extension of component lifespan ensures higher reliability of the inverter-based system.
