基于KELM的光伏组件故障诊断方法

摘要

提出了一种基于核极限学习机(KELM)的光伏组件故障诊断方法.分析了各类型光伏组件故障与光伏组件等效电路模型参数的关系.将光伏组件模型参数和参数辨识的最优均方根误差(RMSE)引入作为诊断局部固有阴影的特征变量,并对KELM故障模型输入特征向量作了整合优化.通过搭建的仿真模型和实例分析证实,与直接将等效电路模型参数作为神经网络输入的方法相比,所提方法可以更快速、精确地识别出短路、老化及阴影故障.%A fault diagnosis method for photovoltaic module based on kernel extreme learning machine (KELM) was presented.The relationship between the fault of various types of photovoltaic modules and photovoltaic module equivalent model parameters were analyzed.The optimal root mean square error (RMSE) for parameter identification was introduced as the characteristic variable of the local intrinsic shadow diagnosis,and the input characteristic vector of KELM fault diagnosis model was formulated and optimized.The simulation model and experimental analysis show that compared to the method directly using the equivalent model parameters as the input to the neural network,the proposed method can be more rapid and precise to identify the conventional short circuit,aging and shadow fault of the photovoltaic module.