基于BP网络对微波加热温度的预测

Prediction of Microwave Heating Temperature Based On BP Neural Network

针对微波加热腔内媒质温度难实时测量的问题,给出了采用BP神经网络算法对加热媒质温度进行实时预测。本论文中首先完成了对媒质温度预测的三层BP网络的设计。其中针对隐含层节点数多少的问题,文中首先利用经验公式估计出隐含层节点数目的范围,然后通过设置循环对隐含层节点数目进行自动筛选,选出最佳的隐含层节点数。最后,在MATLAB的平台上对其进行仿真,并对手动和自动筛选隐含层节点的网络性能以及温度预测结果进行对比分析。

There is a difficult problem that the temperature of medium is measured in the microwave heating cavity. First of all, the BP neural network algorithm is used to predict the real-time temperature of heating medium. In this thesis, the design of three layer BP network for the prediction of medium＇s temperature is completed. About the node number problem of hidden layer, firstly the empirical formula is used to estimate the range of the node number of hidden layer, then the node number of hidden layer is screened automatically by setting the circulation. Finally, the simulation is done on MATLAB platform. The network performance and the temperature prediction results of the manual screening or the automatic screening are compared and analyzed.