A photovoltaic new energy station network attack quantity prediction model based on GA-BP

The detection and forecasting of abnormal traffic in photovoltaic (PV) new energy stations, along with the analysis of different features in network traffic, are crucial for preventing network attacks. This paper investigates a network attack quantity forecast method with Backpropagation (BP) network based on Genetic Algorithm (GA). First, the attack data is analyzed and statistically processed. Then, a detailed study of the basic concepts of BP is conducted, including the specific steps and parameter determination for the BP neural network. Simulations using Matlab are performed to obtain prediction results and highlight issues inherent in the BP algorithm. Next, to address the deficiencies of BP, GA algorithm is proposed for predicting attack events. By analyzing the forecasting performance and fitting effects, the introduction of the GA results in improved model performance. Specifically, MAPE increased by 0.032&percnt;, RMSE decreased by 95.6858, and <i>R</i>² improved by 0.08245.