Health status evaluation of photovoltaic array based on deep belief network and Hausdorff distance

Photovoltaic (PV) arrays, as the core part of PV plants, are sensitive to the complex environment that can lead to fluctuations in their power generation performance. The health status evaluation (HSE) of PV arrays is beneficial for routine maintenance and economic value evaluation. In this paper, a method for evaluating the health status of PV array based on deep belief network (DBN) and Hausdorff distance (HD) is proposed. First, the I-V curves of the PV array are preprocessed, including curve filtering and points redistribution. Then, the practical features of I-V characteristics are extracted by DBN. Next, the health indicator (HI) of the PV array is constructed by HD and Logistic function. Finally, the triangular fuzzy membership function is used to build the mapping relationship between the HI values and the health grades of the PV array. The proposed method enables fully extracting the features from the I-V characteristics of PV arrays and gives an accurate evaluation of different states of PV arrays. The experimental results show that the proposed HSE method can realize the expected objectives.

Automated summary

This paper proposes a method for evaluating the health status of PV arrays using deep belief network (DBN) and Hausdorff distance (HD). By preprocessing and extracting features from the I-V curves, constructing a health indicator using HD and Logistic function, and mapping the HI values to the health grades of the PV array using triangular fuzzy membership function, the proposed method enables accurate evaluation of different states of PV arrays. The experimental results demonstrate the effectiveness of the HSE method.