电力系统供电负荷短期预测模型设计

Prediction Model design of Short-Term Load for Power System

为了提高供电系统电量负荷预测的精度,加快收敛速度,改进陷入局部最小值的可能性以及优化过程单一等问题,建立了电力系统供电负荷短期预测模型,提出粒子群算法对反向神经网络初始化,并引入改进的遗传算法,在交叉过程中用父代最佳值与下一代种群结合,优化网络权值,提升模型性能。将模型应用于辽宁某地区短时电量预测中,结果表明,上述方法加快了模型的收敛速度,提高了预测精度;电力系统供电负荷短期预测模型很好地解决了电量预测精度不高等问题。

In order to improve power load prediction accuracy,fasten convergence,avoid local minima,and optimize the process,a prediction model of short-term load for power system is established. The Particle Swarm Optimization is used for model initialization,and the improved genetic algorithm is introduced. The improved model is applied to predict the short-time power in a region of Liaoning. Results show that this method accelerates the convergence of BP neural network and improve the prediction accuracy.