基于ANN削减负荷的发输电组合系统可靠性评估

RELIABILITY EVALUATION OF BULK POWER SYSTEM USING ARTIFICIAL NEURAL NETWORK FOR LOAD SHEDDING

为提高计算效率 ,提出了基于人工神经网络 ( ANN)的发输电组合系统可靠性评估模型。该模型为一个 3层前向神经网络 ,其中输入层为参与可靠性计算的元件的信息 ,输出层为系统中节点负荷的信息。用改进的 BP算法训练该网络 ,经训练后的网络具有负荷削减计算功能。由于该模型不需进行在线的故障状态潮流计算以及负荷削减计算 ,大大提高了计算效率 ,在缓解“计算灾”方面取得了较大进展。该 ANN模型考虑了发电机出力、变压器和线路容量以及负荷等的变化 ,比通常的可靠性计算模型具有更强的实用性。用实例验证了方法的正确性和有效性。

For improving the calculation efficiency of reliability evaluating(RE) for bulk power system,a novel RE model based on artificial neural network(ANN) is proposed. The ANN has three layers. The inputs into the input laper are the information of components and the output is the parameters of the curtailed loads on buses. The ANN,which is trained by error back propagation(BP) algorithm with variable learning rate and variable momentum,has the function of load shedding in failure states. As there is no need to calculate power flow online and load shedding load under all events,the algorithm has remarkable efficiency. The proposed model and algorithm greatly alleviates the 'calculation catastrophe'. The variation of generator output, capability of transformer and line, and demand of load are considered in the ANN model. So the model has the extensive applicability. The effectiveness,correctness and practicability of the model have been verified by some actual power systems.