Abstract: To address the problems of complex and difficult-to-distinguish fault types and low fault identification accuracy in photovoltaic arrays, this paper proposes a photovoltaic array fault diagnosis method based on a coati optimization algorithm (COA)-optimized long short-term memory network (LSTM). A photovoltaic array model is constructed in MATLAB/Simulink to simulate various single and compound fault operating conditions, and fault features are extracted by analyzing the output characteristic curves. The COA is then used to optimize the key parameters of the LSTM, and a COA-LSTM fault diagnosis model is established. Simulation comparison results indicate that the proposed method can accurately identify various types of faults, achieving a diagnostic accuracy of 97.6%. and that its performance is superior to that of RF, SVM, CPO-LSTM, WOA-LSTM, and other models, demonstrating its high diagnostic accuracy and application potential.