摘要
针对径向基函数(radial basis function,RBF)神经网络收敛速度慢、易于陷入局部极小点的问题,提出了基于蚁群优化算法(ant colony optimization,ACO)的RBF神经网络线损计算新方法。通过引入交叉和变异改进后的ACO训练BRF神经网络,使其具有神经网络广映射能力、ACO快速全局收敛以及启发式学习等特点。利用优化后的RBF神经网络算法拟合配电线路线损与特征参数之间的复杂关系,实现配电网线损计算。仿真结果表明,优化后的BRF神经网络算法的线损计算误差基本在1%以内,具有良好的收敛能力和较快的计算速度。
Aiming at low convergence of neural networks of radial basis function （RBF） and relapse into local minimum, the paper proposes a new method for calculating line loss with RBF neural networks on the basis of ant colony optimization （ACO） algorithm. By introducing crossed and variant ACO to train BRF neural networks, it is enabled to be capable of wide-mapping, fast overall convergence and heuristic learning. Improved RBF neural network algorithm is used to fit the complex relationship between line loss of distribution networks and characteristic parameter to achieve line loss calculation. The simulation result shows that the calculation error of optimized BRF neural network algorithm is approximately smaller than 1% and it is of favorable convergence and high calculation speed.
出处
《广东电力》
2012年第2期72-76,共5页
Guangdong Electric Power
基金
广东省教育厅重点自然科学基金资助项目(040094)
