摘要
根据皮江法炼镁工艺生产过程中的影响因素特点,针对标准BP神经网络存在的收敛速率慢、易陷入局部极小值等缺陷,建立了基于遗传算法优化的BP神经网络镁还原率预报模型。利用筛选后的生产数据对模型进行训练和预测,结果显示该预报模型能够较为精确地预报镁还原率,最大误差小于1.3%,一定程度上可用于指导皮江法炼镁工艺中工艺参数的选择。
Based on the characteristics of influencing factors in the smelting process of Pidgeon process, in view of the standard BP neural network's disadvantages including slow - rate convergence and trapped easily into local minima value, the magnesium reduction rate prediction model based on BP neural network optimized by genetic algorithm is established. The model is rehearsed and tested by the screening production data. The results show that the prediction model could relatively precisely predict the magnesium reduction rate and the maximum error is less than 1.3%. To a certain extent the technique parameters selection in Pidgeon process can he directed by the model.
出处
《轻金属》
CSCD
北大核心
2013年第12期43-46,共4页
Light Metals
关键词
遗传算法
BP神经网络
镁还原率
预报
genetic algorithm
BP neural network
magnesium reduction rate
prediction