A CMA-ES and LightGBM-Based Misoperation Prevention Method for Secondary Systems in Smart Substations

With the widespread adoption of smart substations, the secure operation of power systems imposes stricter requirements on secondary system maintenance. Addressing the issue of potential misoperations during state transitions of secondary equipment, this paper proposes a misoperation prevention method for secondary systems in smart substations based on the LightGBM model, with parameters optimized using the CMA-ES algorithm. Considering that the original signals of smart substation secondary systems are predominantly binary {O, 1} state quantities, this paper employs switch quantity encoding to define data groups. The CMA-ES algorithm is then applied to optimize the hyperparameter combinations of the LightGBM model, enhancing the efficiency of parameter tuning in an automated manner and overcoming the limitations of manual tuning. Finally, a misoperation prevention model for secondary systems in smart substations is constructed based on station-end information. Simulation tests using historical operational data demonstrate a misoperation prevention accuracy rate of 98.34%, thereby validating the effectiveness of the proposed approach.