基于IBA-KELM模型的TCT光伏阵列故障诊断研究

Research on Fault Diagnosis of TCT Photovoltaic Array Based on IBA-KELM Model

针对光伏阵列的开路故障、短路故障、老化故障和局部阴影故障,提出了一种基于核极限学习机KELM(kernel extreme learning machine)的光伏阵列故障诊断方法,并采用改进的蝙蝠算法IBA(improved bat algorithm)对核极限学习机模型的参数进行优化来提高模型的诊断准确率。为避免蝙蝠算法陷入局部最优并加快在参数寻优过程中的收敛速度,引入Levy飞行策略并在速度更新公式中引入指数递减的惯性权重。通过全连接TCT(total-crosstied)结构光伏阵列的故障数据验证表明,与BA-KELM,PSO-KELM、PSO-ELM模型相比,IBA-KELM模型在参数优化过程中收敛速度更快,优化后模型诊断精度也更高。

Aimed at the open circuit fault,short circuit fault,degradation fault and partial shadow fault of a photo-voltaic(PV)array,a fault diagnosis method based on kernel extreme learning machine(KELM)is proposed,and the im-proved bat algorithm(IBA)is used to optimize the parameters of the KELM model,thus improving the model’s diagnos-tic accuracy.To avoid the bat algorithm from falling into a local optimum and accelerate the convergence speed in the parameter optimization process,the Levy flight strategy is introduced,and an exponential decreasing strategy is intro-duced into the speed update formula.The verification of fault data of a total-cross-tied(TCT)PV array indicates that com-pared with the BA-KELM,PSO-KELM,and PSO-ELM models,the proposed IBA-KELM model converges faster in the pa-rameter optimization process,and the fault diagnostic accuracy after optimization is higher.