基于机器学习的球团矿质量预测模型研究现状

Research status of pellet ore quality prediction model based on machine learning

链篦机-回转窑是一种生产球团矿的新工艺,因其能生产出满足高炉需求的高质量球团矿而被各球团矿生产企业广泛采用。然而链篦机-回转窑生产球团矿的工艺复杂,成品球团矿的质量又与各生产工艺和操作密切相关,这使得球团矿的质量控制和调整滞后并呈现较强的周期性。当前大多数球团矿生产企业采取现场人工取成品球团矿样送检,这使得球团矿检测结果不能实时指导生产且具有很大的不确定性。因此,建立成品球团矿的质量预测算法模型就显得尤为重要。国内外学者在球团矿质量预测算法做了大量工作,主要有基于案例的推理算法、基于神经网络的算法、遗传算法和其他算法。本文对球团矿质量预测算法进行了分析,并对算法的性能和该领域的未来发展进行了评价和反思。

The chain grate machine-rotary kiln is a new process for producing pellet ore, because it can produce high-quality pellet ore to meet the needs of blast furnaces, therefore which is widely used by pellet ore manufacturing enterprise.However, the production process of pellet ore with chain grate machinerotary kiln is complex, and the quality of finished pellet ore is closely related to each production technics and operation, which makes the quality control and adjustment of pellet ore lag behind, and appearing strong periodicity. At present, most pellet ore production enterprises take finished pellet ore samples manually on site for inspection, which makes the pellet ore testing results not real-time guidance for the production process and has a great deal of uncertainty. Therefore, it shows up particularly important to establish the quality prediction algorithm model for the finished pellet ore. The scholars at home and abroad have done a lot of work on the quality prediction algorithm for the pellet ore,there are mainly case-based inference algorithms, neural network-based algorithms, genetic algorithms and other algorithms. In this paper, the pellet ore quality prediction algorithm is analyzed, and the performance of the algorithm and the future development of this field are evaluated and reflected.