二层分解技术在电价预测中的应用研究

针对电价波动幅度大且预测精度低的问题，提出了二层分解技术与神经网络组成的电价多步预测模型。该模型首先采用集合经验模式分解将原始电价序列分解为一系列分量，变分模态分解将第一层分解产生的最高频率分量进一步分解为一系列模态分量，所有分量采用神经网络模型进行预测，并使用纵横交叉算法对神经网络的参数进行优化，最后叠加所有子序列，得出预测电价值。仿真结果表明，所提出的模型相比其他混合模型具有更好的预测性能，且实用价值高。

Research on two-layer decomposition technique for predicting electricity price

To solve the problems that the electricity price has a lot of volatility and a low accuracy , this paper proposes a multi-step prediction model composed of two-layer decomposition technique and back propagation neural network(BP) . Firstly, variational mode decomposition(VMD) is specifically applied to further decompose the high frequency intrinsic mode functions (IMFs) generated by ensemble empirical mode decomposition(EEMD) into a number of modes. Next, BP model optimized by crisscross optimization algorithm(CSO) algorithm is utinized to forecast all sub-sequences, and the forecast series of electricity price is obtained by adding up all sub-sequences. The simulation result shows that the proposed model has superior performances for other hybrid models , and it has a great practical values.