论文部分内容阅读
针对铅锌烧结过程中具有强非线性、时滞的特点,提出一种基于变学习率的烧结块产量质量神经网络预测模型。通过分析过程特性和工况参数的相关性,确定影响产量和质量的操作参数;采用普通的BP(Back Propagation,简称BP)神经网络结构,建立铅锌烧结块产量质量预测模型;在网络训练的过程中,采用变学习率的方法对BP算法进行改进,获得了满意的预测效果,该算法具有较快的收敛速度。将改进的神经网络模型进行仿真实验,结果表明,该模型具有较高的预测精度和较强的自学习功能,从而验证了方法的有效性。
Aiming at the characteristics of strong nonlinearity and time lag in the process of lead-zinc sintering, a predictive model based on variable learning rate is proposed to predict the mass production of sinter block based on neural network. By analyzing the correlation between the process characteristics and the working parameters, the operating parameters affecting the yield and quality were determined. The production model of lead-zinc sinter mass production was established using the common BP (Back Propagation) neural network structure. In the network training In the process, BP algorithm is modified by changing the learning rate, and the satisfactory prediction result is obtained. The algorithm has faster convergence speed. The improved neural network model is simulated, the results show that the model has high prediction accuracy and strong self-learning function, which verifies the effectiveness of the method.