论文部分内容阅读
电网负荷具有非线性的特点,可预测性较差。提出一种基于神经网络的电网负荷超载预测模型,设计了一个由输入层、隐含层以及输出层构成的网络模型,采用该模型对电网负荷超载进行预测,并基于贝叶斯定理训练该预测模型,对预测过程中的超载负荷不确定因素进行全面分析,采用求极大似然估计和迭代算法得到预测模型的最优参数,然后将各步迭代求解得到的神经网络输入当作随机变量输入,最后依据塑造的最佳预测模型得到最优电网负荷超载的预测结果。实验结果说明,所设计模型的预测结果同实际更接近,避免了不确定性对电网负荷超载预测的干扰,具有较高的预测准确度。
Grid load has the characteristics of non-linear, poor predictability. A network overload forecasting model based on neural network is proposed. A network model consisting of input layer, hidden layer and output layer is designed. The model is used to predict the load of power grid. The model is trained based on Bayes’ theorem Model, a comprehensive analysis of the uncertainty of the overload load during the forecasting process is performed. The optimal parameters of the prediction model are obtained by using the maximum likelihood estimation and the iterative algorithm. Then, the input of neural network obtained through the iteration of each step is taken as the input of random variables Finally, based on the best prediction model, the forecast result of the optimal grid load overload is obtained. The experimental results show that the predicted results of the designed model are closer to the actual ones, which avoids the interference of the uncertainties on the load forecast of power grid and has higher prediction accuracy.