Machine learning-based state prediction and optimization of orthogonal iterative abort strategy for unbalanced power grids

Different generation profiles and electricity demand are likely to cause imbalance in the grid state, which puts pressure on the intelligent scheduling of power networks. In this paper, we utilize the grid cross-section data for training through machine learning methods and predict the grid state from both macro and micro perspectives. Subsequently, a multi-objective optimization model of grid balance is constructed, and the genetic algorithm is optimized based on the orthogonal iteration method, and the model is optimization-seeking in order to make real-time adjustments to the grid state. The accuracy of the two prediction results of this paper’s method is higher than 85% by similar day matching calculation. In addition, using the optimization method of this paper achieves a reduction of grid fluctuation of 203w and 44var respectively on the basis of the traditional method, which effectively alleviates the imbalance state of the grid and helps to guarantee the security of the national electricity consumption.