中文摘要：

为了提高光伏系统的发电效率,同时降低人工维护的成本,提出了一种基于BP（back propagation）神经网络的光伏组件在线故障诊断策略;分析了光伏组件短路和异常老化故障的成因,并在Matlab中对光伏组件故障状态下的输出特性进行了仿真研究。根据仿真结果并结合光伏组件的数学模型,总结了光伏组件的故障规律,建立了BP神经网络故障诊断模型及模拟光伏组件各种故障的仿真模型。用该模型采集了适合神经网络训练的样本,并对神经网络诊断模型进行了训练。结合光伏功率优化器,进行了组件在线故障诊断的仿真和实验研究,结果验证了文中方法的正确性、有效性和环境适应性。

英文摘要：

To improve generating efficiency of photovoltaic（PV） generation system and decrease the cost for its artificial maintenance,based on back propagation（BP） neural network（NN） an online fault diagnosis strategy for PV modules is proposed.The contributing factors causing short-circuit fault and abnormal aging of PV modules are analyzed and using Matlab the output characteristics of PV module under fault condition is simulated.According to simulation results and combining with mathematical model of PV module,the fault patterns of PV module are summarized,and a BPNN based fault diagnosis model for PV modules is built and a model to simulate various faults occurred in PV modules is established.The samples suitable for the training of BPNN are collected and the BPNN based fault diagnosis model is trained.Combining with PV power optimizer the simulation and experimental research on online fault diagnosis of PV modules are performed,and results from simulation and research show that the proposed fault diagnosis strategy is correct,effective and possesses environmental suitability.