基于改进PSO-BP网络的配电网故障选线与测距

针对人工智能算法在解决配电网故障选线和测距问题时容易陷入局部最优解并难以满足精确性和鲁棒性要求的问题,提出了一种基于改进粒子群优化神经网络的配电网故障选线与测距算法.该算法结合混沌优化算法和粒子群优化算法得到收敛能力更强的粒子群优化算法,通过提取配电网的零序电压与电流的暂态及稳态特征来构成特征向量,并分别使用训练集训练改进粒子群优化神经网络算法,从而能更精确地预测配电网的故障线路及其距离.仿真测试结果表明,所提出的算法能获得更精确的选线和测距结果,具有一定的实用性.

Fault line selection and location for distribution network based on improved PSO-BP neural network

In order to solve the problems that during the fault line selection and location for the distribution network, the artificial intelligence algorithms are easily trapped into the local optimal solutions and are difficult to meet the requirements in both accuracy and robustness, a fault line selection and location algorithm for distribution network based on the improved particle swarm optimization(PSO)neural network was proposed. The proposed algorithm combines the chaos optimization and PSO algorithms to obtain a new PSO algorithm with stronger convergence ability. Through extracting the characteristics of both transient and steady states of zero-sequence voltage and current of distribution network, the feature vector was formed, and the PSO neural network algorithm was trained and improved with the training set. Therefore, the fault line and location of distribution network can be predicted more accurately. The results of simulation test show that the proposed algorithm can obtain more accurate line selection and location results, and has certain practicality.