基于神经网络的烧结终点预报的研究

Simulation study on predicting of burning through point of sintering based on artificial neural network

针对烧结终点非线性的特点,采用误差反向传播算法的多层前馈神经网络(BP)来建立其模型,用自适应算法确定学习参数,用改进的BP神经网络的算法求出结构适宜的自适应网络。提出并实践了提高烧结终点BP神经网络预报速度的数据处理方法,基于现场数据采用计算机仿真的结果表明该方法的有效性。

The BP network is used in modeling of the burning through point(BTP) of sintering with nonlinear characteristics. The adaptive algorithm is used to determine iterative learning parameters. The adaptive network with appropriate structure is given by the modified algorithm for the BP network. The data-handling method for shortening the prediction time of the BTP is advanced and applied to the prediction of a practical case. The simulation result illustrates the effectiveness of the presented method.