基于光伏阵列输出预测模型的故障诊断方法

Fault diagnosis method based on an output prediction model of a photovoltaic array

本文提出了一种基于光伏组件阵列输出预测模型的故障诊断方法用于识别阵列直流侧的四种故障。该方法选择I-V(电流-电压)和P-V(功率-电压)曲线上的关键点作为诊断特征量,并将无故障情况下的特征量作为参考值;通过组件的双二极管模型建立了一个光伏阵列输出预测模型来预测不同工作条件下的参考值,在MATLAB/Simulink平台搭建光伏阵列仿真无故障和不同故障实验并将对应特征量与预测参考值对比。无故障实验中,三种工作条件下预测与仿真测量参考值的误差率低于2%,表明了阵列输出预测模型对参考值预测的准确性,并将预测值作为后续故障实验的参考值;计算几种故障实验的仿真测量特征量与参考值的相对误差,分析特征量的变化,给出了不同故障类型诊断的方法。

A fault diagnosis method was proposed based on an output prediction model of a photovoltaic(PV)module array to identify four kinds of faults on the DC side.Key points on the I-V(curren-tvoltage)and P-V(power-voltage)curves were selected as diagnostic features,and features without faults were used as reference features.A PV array output prediction model was setup by using double-diode model of a module to predict references under different operating conditions.The photovoltaic array simulation with no faults and different faults were conducted using MATLAB/Simulink platform,and the corresponding features were compared with predicted references.In the experiment without faults,the error rate between predicted and simulated measured reference features was less than 2%under three operating conditions,indicating a good prediction accuracy for reference features by the prediction model of a PV array.The predicted values were then used as references for subsequent fault experiments.By calculating relative error for simulated measured features to references in various fault experiments,changes in features were analyzed and diagnostic methods for different types of faults were summarized.