Microgrid Scheduling Optimization under False Data Injection Attack

False data injection attacks (FDIAs) pose a great threat to the data security of microgrid. The day-ahead scheduling strategy of microgrid is in turn mainly based on microgrid operation data. Under FDIAs, it is important for the efficiency and economic cost of microgrid scheduling to properly repair the attacked data and optimize the scheduling using predictive data from deep learning. In this work, we model the economic cost of microgrid scheduling, use deep learning prediction algorithm to predict data based on historical values of microgrid operation, and repair the attacked data by the predicted data, and combine the optimization algorithm to optimize microgrid day-ahead power scheduling to make scheduling more economical and efficient. By this way, the malicious action was corrected and system recover as normal performance to avoid the economical lost that FDIAs aimed to attrite.