FDIA Detection in Power Grid Based on Opposition-Based Whale Optimization Algorithm and Multi-layer Extreme Learning Machine

At present, the existing false data injection attack (FDIA) detection methods for cyber-physical power system can not precisely obtain the location of the attack due to its limited ability of feature expression. Therefore, this paper proposes a FDIA location detection method based on opposition-based learning whale optimization algorithm and multi-layer extreme learning machine (OWOA-ELMML). The proposed method not only extends the extreme learning machine into a multi-layer neural network to solve the problem of its limited ability of feature expression, but also introduces the whale optimization algorithm to optimize the number of neurons of the multi-layer extreme learning machine, and uses the opposition-based learning strategy to improve its convergence speed and detection accuracy so as to prevent the influence of randomly determining the number of neurons in each hidden layer on the generalization performance and location detection results of the detection method. Through a large number of simulation tests on IEEE-14 and 57-bus test systems under different scenarios, it is verified that the proposed method can automatically identify the exact position of the attacked system state through the historical data. Compared with other comparative methods, the proposed method has better precision, recall rate and F1 value.