Research on frequency rise of photovoltaic data based on improved Enhanced Deep Residual Networks

Converting low-frequency photovoltaic (PV) data into high-frequency data is an effective means to improve the accuracy of PV output prediction. Since the traditional reconstruction algorithm is difficult to achieve high-frequency data reconstruction, an improved enhanced deep residual network (I-EDRN) is introduced to reconstruct low-frequency PV data into high-frequency data. The data redundancy is improved by converting the time series data into a two-dimensional matrix representation. I-EDRN builds a deep residual network based on the improved residual block structure to improve feature learning ability. At the same time, spatial attention (SA) mechanism is introduced to strengthen the local perception ability of the network. A PV power station is taken as an example to verify the algorithm. The verification results show that the proposed method can achieve better photovoltaic high-frequency reconstruction effect than the traditional method. The verification results show that compared with the traditional methods, the proposed method can achieve lower average absolute error and average absolute error percentage.