Enhanced Fault Detection in Solar Photovoltaic Modules Using VMD-LSTM Model

Detection of solar PV faults in an Accurate and versatile technique are essential because the safety and efficiency of solar photovoltaic (PV) modules greatly depend on efficient fault identification. However, because it can be challenging to identify complex operating patterns and detect tiny faults, currently methods frequently have low accuracy. This may make it more difficult to validate the models, which would limit their usefulness in the actual world. This paper presents a new fault detection method that combines Empirical Mode Decomposition (EMD) with the power of Long Short-Term Memory (LSTM) networks. Critical features are efficiently extracted from the data by use of an adaptive decomposition of voltage and current signals into Intrinsic Mode Functions k(IMFs) through the use of EMD. An LSTM network that has been trained to recognize complex patterns and periodic connections then processes this information. Our model which has been validated using a PSCAD simulation model, shows notable improvements in accuracy and durability of more than 92% after undergoing thorough testing on a simulated PV system that allows for several fault types and their severities when compared to existing methods.