基于RBFNN混合粒子群算法的电力负荷短期预测

根据电力系统负荷预测的不同目的,提出一种基于RBFNN混合粒子群优化算法(HPSO)预报电力系统短期负荷,即首先采用改进的粒子群优化算法(MPSO)全局优化网络模型参数然后在MPSO全局搜索模型参数基础上利用梯度下降法局部优化网络模型参数,建立电力系统短期负荷的时序人工神经网络模型。仿真结果表明,该方法与传统的预测方法相比,减少了训练时间,提高了精度和适应性。

Power Load Forecasting by HPSO Based on RBFNN

A hybrid particle swarm optimization (HPSO) based on radical basis function neural networks (RBFNN) is proposed according to the different purpose of power system load forecasting. First, using modified particle swarm optimization (MPSO) searches for the parameters in the multidimensional complex space through cooperation and competition among the individuals in a population of particles to speed up global search. Furthermore, gradient decreasing algorithm searches for the model parameter of RBFNN to speed up the local search. The prediction of power system short-term load is simulated using the proposed HPSO algorithm. Result shows that it can get a better forecasting effect compared with conventional BP algorithm. This approach reduces the training time, accelerates the speed of PSO algorithm, and improves the adaptability of the artificial neural networks system.