In order to accurately assess the SDN network security status of power IOT business traffic, this paper proposes an SDN-oriented network security sensing method. The method extracts network security posture indicators based on the attack characteristics suffered by data, control and application planes. On this basis, a network security posture assessment method based on CS-BPNN is proposed, which uses the cuckoo search algorithm to find the optimal weights and thresholds of the model, and then applies the BP algorithm to adjust the error. On the basis of network security posture assessment, in order to solve the nonlinear problem in security posture prediction, the parameters of the LSTM prediction model are calculated using the gray wolf optimization algorithm to control the direction of the optimization search. The results show that the overall error of the proposed CS-BPNN assessment model is small and closest to 0. The prediction accuracy of the GWO-LSTM model in different tasks reaches more than 95%, which confirms the accuracy and stability of the method proposed in this paper. It provides methodological support for network feature security response and strategy design in the environment of electric power IoT.
