基于FCN和LSTM深度学习模型的电力终端负荷预测方法

Power terminal load forecasting method based on FCN and LSTM deep learning model

针对目前研究的电力终端负荷预测方法在预测过程中,未考虑负荷终端时序性和非线性的特点,存在预测精度较低,速度较慢的问题,提出了基于FCN和LSTM深度学习模型的电力终端负荷预测方法。利用循环神经网络,建立电力终端负荷预测模型,分析记忆状态,对电力终端数据进行预处理,并编码非数字特征,确定预测模型。利用均方误差公式得到的电力终端负荷预测模型的损失函数,通过数据预处理、优化数据参数、训练电力终端负荷预测模型完成负荷预测。实验结果表明,基于FCN和LSTM深度学习模型的电力终端负荷预测方法充分分析了终端时序性和非线性特点,有效确保预测精度,提高预测速度。

Aiming at the problems of low prediction accuracy and slow speed in the current power terminal load forecasting method,which does not consider the characteristics of load terminal timing and nonlinearity,a power terminal load forecasting method based on FCN and LSTM deep learning model is proposed.Using the cyclic neural network,the power terminal load forecasting model is established,the memory state is analyzed,the power terminal data is preprocessed,and the non digital features are encoded to determine the forecasting model.The loss function of power terminal load forecasting model obtained by mean square error formula is used to complete load forecasting through data preprocessing,optimizing data parameters and training power terminal load forecasting model.The experimental results show that the power terminal load forecasting method based on FCN and LSTM deep learning model fully analyzes the timing and nonlinear characteristics of the terminal,effectively ensures the forecasting accuracy and improves the forecasting speed.