高炉煤气流分布过程的多算法融合预测模型

Multi-algorithm fusion prediction model for the blast furnace gas flow distribution process

中文关键词  高炉  煤气流分布  自回归移动平均模型  限定记忆  极限学习机  预测

英文关键词  blast furnace  gas flow distribution  auto-regressive moving average model  finite memory  extreme learning machine  prediction

中文摘要

      在高炉生产中, 准确的预测高炉煤气流分布状况将有助于保证高炉的稳定顺行. 针对传统高炉煤气流预测 模型的缺陷, 本文提出了一种将带遗传因子的自回归移动平均模型(FF??ARMAX)和基于限定记忆的正则化极限学 习机(RFMLS??RELM)相结合的高炉煤气流多步预测模型. 在数据预处理方面, 建立FF??ARMAX模型消除原始数 据中的测量误差, 同时采用傅里叶变换法消除数据中叠加的环境噪声. 最后采用RFMLS??RELM算法进行多步预 测, 对比试验表明, 该算法在应用于煤气流预测时, 预测精度更高, 适用于对煤气流分布状况的多步预测. 多步预测 实验结果表明, 该模型虽然仍旧无法完全解决预测误差随预测步数的增加而不断叠加的问题, 但相较于其他传统预 测模型能够实现更好的预测效果和更高的预测精度, 为高炉操作人员分析炉况提供了有效的帮助和支持.

英文摘要

      In blast furnace production, the prediction of gas flow will help to ensure the stability of the blast furnace. Aiming at the shortcomings of the traditional prediction model of blast furnace gas flow, this paper proposes a multistep prediction model of gas flow based on auto-regressive moving average model based on recursive least squares with forgetting factor (FF??ARMAX) and regularized extreme learning machine based on recursive finite memory least squares (RFMLS??RELM). In terms of data correction, construct FF??ARMAX model to eliminate the measurement error in the data. At the same time, the Fourier transform method is used to denoising. The RFMLS??RELM algorithm is used for multi-step prediction. The comparison experiments show that the proposed algorithm is faster in modeling and higher in prediction accuracy when applied to blast furnace gas flow prediction. It is suitable for the multi-step prediction of gas flow distribution of blast furnace. The experimental results show that the model can accurately predict the distribution of blast furnace gas flow, which provides effective help and support for blast furnace operators to analyze furnace conditions.