基于卷积神经网络的暂稳极限功率计算

Calculation of Transient Stability Limit Based on Convolutional Neural Network

目前求取联络线暂态稳定传输功率极限的时域仿真法和基于李雅普诺夫稳定性理论的直接法计算过程复杂,针对此问题,提出基于卷积神经网络的输电断面暂态稳定极限功率计算方法。首先将系统运行的数据与实验仿真数据相结合,转化为输电断面特征属性,选择输电断面关键特征,将其作为神经网络的输入层向量,然后经过多次训练构建出系统关键特征与输电断面暂态稳定极限功率的非线性映射关系。最后以IEEE14节点进行算例分析,验证了计算方法的可靠性以及有效性。

Current procedures to calculate transient stability limit of interface tie line power transfer,using either time domain simulation method or direct method based on Lyapunov stability theory,are very time-consuming and complex.In view of this problem,a new method to compute transient stability limit of interface power transmission is proposed based on convolutional neural network.Firstly,the system operation data and the experimental simulation data are combined together to formulate the characteristic attributes of the transmission interface.Then certain key features of the transmission interface are selected as the input layer vector of the neural network.And next the nonlinear mapping relationship between the key features of the system and the transient stability limit of interface power transmission is constructed after multiple rounds of training processes.Finally,the reliability and effectiveness of the proposed calculation method are verified by case studies of IEEE14 bus system.