基于PSO-Elman神经网络的短期电力负荷预测

为了解决BP神经网络在短期电力负荷预测中存在局部极小、收敛速度慢等问题,本文采用粒子群算法(Particle Swarm Optimization,PSO)优化Elman动态神经网络进行精准预测。根据输入输出参数个数确定Elman神经网络结构,利用PSO算法优化网络的权值和阈值,并将优化后的最优个体赋给Elman动态神经网络作为初始权值、阈值进行网络训练,从而建立基于PSO-Elman的电力负荷预测模型。采用某钢厂实测电力数据对该方法和模型进行验证,并与传统的BP、Elman网络模型预测方法进行对比,结果表明该方法和模型在有效缩短网络收敛时间的同时,具备更高的负荷预测精度和稳定性。

Short-term Electric Load Forecasting Based on PSO-Elman Neural Network

To solve the problems such as slow convergence speed and local minimum of in the training process of BP neural network, the method using the particle swarm optimization algorithm (PSO) to optimize the Elman neural network is adopted for short-term power load forecasting in this paper. The Elman neural network structure has been built according to the number of the input and output parameters, and then PSO algorithm has been used to optimize the weights and thresholds of the network. The optimal individual is assigned to the Elman neural network as the initial weights and thresholds of the training process. Based on the upper work, the power load forecasting model based on PSO-Elman neural network is established. The PSO-Elman model has been tested by using the real electricity data, and the performance of this model has been compared with that of the traditional BP and Elman neural network forecasting model. The results show that the method and model of this paper can effectively shorten the convergence time of network, and have higher load forecasting accuracy and stability then the others.