摘要
气流床气化技术受到广泛关注,一般工艺要求液态排渣,而灰渣粘度决定着气化炉排渣能否顺利。在灰渣粘度预测中,粘度与灰渣呈复杂的非线性映射关系,而目前尚未有成熟的模型。本文拟从模糊模型人手,采用遗传算法(GA)优化神经网络(BP)的初始权阈值,优化后的神经网络模型,再预测灰渣粘度值。预测过14组样本,将预测值同三种不同机理模型预测值比较,证明GA-BP模型预测值同实验值最接近,且精度明显较其它模型高。
The gasification of coal in entrained flow gasifier to produce clean syngas for use in the industry is attracting considerable interest worldwide, where the slag is tapped in liquid state, so the viscosity of which should be adequate for achieving successful slag tapping. In viscosity predicting, affecting factors of the slag viscosity are so complicated, and also at present, a precise mathematical model has not been established. In this paper, a new neural network optimized by genetic algorithm for use in predicting the viscosity of the slag is presented. The viscosity of fourteen samples are predicted with optimized network model. After taking result to compare with the value predicted by other models, it indicates that the predicting value using GA-BP model is closer to the experiment one, and also with a higher precision than other models.
出处
《计算机与应用化学》
CAS
CSCD
北大核心
2007年第5期609-613,共5页
Computers and Applied Chemistry
基金
国家重点基础研究发展计划(2004CB217703)
上海市曙光计划(06SC34)
关键词
灰渣粘度
神经网络
遗传算法
煤气化
slag viscosity, neural network, Genetic Algorithm （GA）, coal gasification
