A Fault Diagnosis Method for Photovoltaic Modules Based on Transfer Long Short-Term Memory Neural Network

As the core component of solar power station, the reliability of photovoltaic (PV) module affects the safety and stable operation of the whole system. PV modules are mostly installed in harsh environment, which are prone to various failures. It is very important to find the faults in time for improving the service life and normal operation efficiency of PV modules. However, the fault data of PV modules is insufficient, which makes the diagnosis very challenging. Aiming at the challenge, a fault diagnosis method based on transfer long short-term memory (LSTM) neural network is proposed for PV modules. Firstly, the LSTM-based power generation prediction model of PV module under normal state is trained. Then, by using a small amount of fault data to fine-tune the parameters, the power generation prediction model under fault state is obtained, which effectively solves the problem of insufficient fault data to realize fault diagnosis. The experimental results prove that the proposed fault diagnosis method in this paper can accurately and effectively detect the fault of PV modules under different fault conditions.