基于改进粒子群-模糊神经网络的短期电力负荷预测

Short-term load forecasting based on modified particle swarm optimizer and fuzzy neural network model

为了提高短期电力负荷预测精度,提出了改进的粒子群-模糊神经网络混合优化算法.用改进的粒子群训练神经网络,实现了模糊神经网络参数优化.建立了基于该优化算法的短期负荷预测模型,综合考虑气象、天气、日期类型等影响负荷的因素,利用贵州电网历史数据进行短期负荷预测.仿真表明,该方法的收敛速度和预测精度优于传统模糊神经网络法、BP神经网络法、粒子群-BP算法和粒子群-模糊神经网络方法,该优化算法克服了神经网络和粒子群优化方法的缺点,改善了模糊神经网络的泛化能力,提高了电网短期负荷预测的精度,各日预测负荷的平均百分比误差可控制在1.2%以内.该算法可有效用于电力系统的短期负荷预测.

To improve short-term load forecasting accuracy, a modified particle swarm optimizer （MPSO） and fuzzy neural network （FNN） hybrid optimization algorithm is proposed. In which the FNN is trained by MPSO to implement the optimization of FNN parameters. The short-term load-forecasting model is established based on the modified particle swarm optimizer and fuzzy neural network hybrid optimization algorithm. In load forecasting such factors impacting loads as meteorology, weather and date types are comprehensively considered. Using the method and history load data of Guizhou power system, the short- term load forecasting was carried out. The result shows the convergence of method is faster and forecast accuracy is more accurate than that of the traditional fuzzy neural network, BP neural network, the particle swarm optimizer （PSO） and BP neural networks, PSO and fuzzy neural networks. The hybrid algorithm improves the fuzzy neural network generalization capacity, and overcomes the traditional PSO algorithm and fuzzy neural network that exist in some of the shortcomings. The short-term load-forecasting accuracy is improved in Guizhou power system, which average percentage error is not more than 1.2%. The hybrid algorithm can be used efficaciously in short time load forecasting of the power system.