【摘要】 大批分布式新能源接入，使得配电网运行环境日趋复杂，对无功优化提出了更高要求。为应对该挑战，提出了一种数据驱动的无功优化新方法。使用带残差结构的改进卷积神经网络提取配电网负荷和新能源出力数据的高维特征；进行特征位置编码并引入自注意力机制进一步处理，建立数据驱动模型。利用历史运行数据和无功优化最优策略库训练所提无功优化模型，发掘配电网节点数据和无功设备补偿策略间的非线性映射规律。通过IEEE 33节点算例验证了本文所提方法的有效性。更多还原

【Abstract】 The operational complexity of distribution networks is increasingly exacerbated due to the integration of numerous distributed new energy sources, which imposes more stringent requirements for reactive power optimization. To address this challenge, this paper proposes a new data-driven approach for reactive power optimization. Initially, an enhanced convolutional neural network with a residual structure is employed to extract high-dimensional features from the load data and new energy output of the distribution network. Subsequently, encoding for feature location is performed, and a self-attention mechanism is integrated to refine these features, thereby establishing a data-driven model. The proposed reactive power optimization model is trained using historical operating data and a repository of optimal reactive power optimization strategies,enabling it to discern the intricate nonlinear relationship between distribution network node data and reactive power equipment compensation strategies. The validity of the proposed method is demonstrated through a case study involving the IEEE33-node test system.更多还原