基于小波包去噪和深度学习的电力行业碳排放预测模型研究

Carbon Emission Prediction Model for Power Industry Based on Wavelet Packet Denoising and Deep Learning

针对当前电力行业碳排放预测模型精度不高、参数优化困难等问题,提出一种基于小波包分解(Wavelet Packet Decomposition,WPD)去噪的碳排放预测模型PSO-CNN-LSTM.首先利用小波包分解将建模数据进行去噪,然后构建CNN-LSTM模型对碳排放数据进行预测.为了解决模型超参数选取困难的问题,利用粒子群优化(Particle Swarm Optimization,PSO)算法对模型进行迭代寻优,寻求最优的超参数组合.经验证可知,所提出的基于WPD去噪的PSO-CNN-LSTM模型的4种评价指标均最优且模型泛化能力更强,说明该模型可以应用于国家或区域尺度电力行业碳排放预测.

Aiming at the problems of low accuracy and difficult parameter optimization of the current carbon emission prediction model in the power industry,a carbon emission prediction model PSO-CNN-LSTM based on wavelet packet decomposition(WPD)denoising is proposed.Firstly,wavelet packet decomposition is used to denoise the modeling data,and then CNN-LSTM model is constructed to predict the carbon emission data.In order to solve the problem of difficult selection of model hyperparameters,particle swarm optimization(PSO)algorithm is used to iteratively optimize the model to seek the optimal hyperparameter combination.The empirical results show that the four evaluation indexes of the proposed PSO-CNN-LSTM model based on WPD denoising are all the best and the model has stronger generalization ability,indicating that the model can be applied to the prediction of carbon emissions in the power industry at national or regional scale.