Research and application of the impact of distributed photovoltaic power supply access on distribution network automation technology

This paper inquiries into the impact of photovoltaic source intermittency on behaviour of existing distribution networks. Real PV power output profile are added into detailed simulations conducted on 33-bus IEEE radial distribution system. Load flow study was conducted in evaluating steady-state operational conditions before integration of PV Systems. Subsequently, PV power sources was introduced to a networks, and the effect of Photo Voltaic power outputs intermittency were examined over 24-hour time durations. Result indicate that PV integration improves overall voltage profiles by an average of 5% and reduces power losses by 10%. Additionally, active power productions from traditional source decreases by 15% due to PV integration. However, the intermittent nature of PV lead to significant fluctuations in daily bus voltage profiles of up to 8% and modify the power production plan by 20% and also that BiLSTM outperforms MLP and SVM across all metrics, with RMSE of 7.536, MAE of 4.369, MAPE of 15.87, and PCC of 0.961. In comparison, MLP has RMSE of 12.181, MAE of 7.526, MAPE of 34.71, and PCC of 0.914, while SVM shows RMSE of 14.923, MAE of 11.549, MAPE of 38.96, and an unusually high PCC of 9.928. Thus, BiLSTM exhibits the lowest error rates and highest correlation, indicating its superior predictive performance. To mitigate these issues, this study can assist planning operators in adjusting regulation programs and daily power generation plans and electrical power systems deregulation, conducting numerous thematic simulations of this nature can provide comprehensive insights into system behavior in the presence of intermittent power sources, thereby guiding planners in deciding on protection and defense actions. To enhance the prediction of PV power output intermittency, Bi-directional Long Short-Term Memory (Bi-LSTM) network can be used. Bi-LSTM models have shown promising results in capturing temporal dependencies and can be used to forecast PV power output profiles with higher accuracy, thereby aiding in better understanding and managing the impact of PV intermittency on distribution networks.