论文部分内容阅读
提出了一种基于小波变换和改进萤火虫优化极限学习机的短期负荷预测方法.通过小波分解和重构,对原始负荷序列进行降噪;在模型训练阶段利用改进的萤火虫算法优化极限学习机参数,获得各序列的最优模型;针对各子序列分别预测叠加得到最终预测值.通过在两种时间尺度的数据序列上进行数值计算,与传统的ARMA、BP神经网络、支持向量机及LSSVM等多种经典预测模型相比,模型预测效果更优.
This paper proposes a short-term load forecasting method based on wavelet transform and improved firefly optimization limit learning machine.With wavelet decomposition and reconstruction, the original load sequence is denoised. In the model training stage, the improved firefly algorithm is used to optimize the parameters of extreme learning machine, Obtain the optimal model of each sequence, and predict the superposition of each sub-sequence to get the final prediction value.Compared with the traditional ARMA, BP neural network, support vector machine and LSSVM, by numerical calculation on the data sequences of two time scales, Compared with the classical prediction model, the model prediction is more effective.