基于机器学习的短期电力负荷预测方法研究

针对短期电力负荷数据具有明显周期性的特点,将基于机器学习引入到短期电力负荷预测领域,提出一种基于岭回归估计的RBF神经网络短期电力负荷预测方法,该方法利用机器学习算法RBF在非线性拟合方面的优势,结合岭回归对RBF神经网络输出层权值进行参数估计,有效消除输入多重共线性问题,采用广义交叉验证法对构建的模型进行评估,寻找最优岭参数,提高了电力负荷预测精度。通过实际负荷预测案例,与传统BP神经网络负荷预测方法进行比对,验证了提出的电力负荷预测方法较传统方法具有较好的稳定性和较高的预测精度,为电力负荷预测提供了新思路。

Research on Short-term Power Load Forecasting Method Based on Machine Learning

According to the characteristics of short-term electric power load data has obvious cyclical, the machine learning is introduced into the short-term power load forecasting field, and a the RBF neural network short-term power load forecasting method based on ridge regression estimates is proposed, the method using the advantages in nonlinear fitting of RBF machine learning algorithm, combined with ridge regression parameters to estimate RBF neural network output layer weights, effectively eliminate the input multicollinearity problem, the generalized cross-validation method is adopted to evaluate the load forecasting model, to find the optimal ridge parameter, improve the power load forecasting accuracy.By comparing the actual load forecasting case with the traditional BP neural network load forecasting method, it is verified that the power load forecasting method proposed in this paper has better stability and higher prediction accuracy than the traditional method.