基于机器学习的降低汽油辛烷值的损失模型
Loss Model of Reducing Gasoline Octane Number Based on Machine Learning

现阶段，即使在电力驱动汽车迅猛发展的今天，从实用角度看，以汽油为燃料驱动的车辆仍然是公路出行、运输的主流；从保有量，更换周期考虑，汽油驱动汽车在很长的时间内仍不会退出历史舞台。而汽油燃烧性能的好坏是汽油综合质量重要指标，其中辛烷值(RON)的含量则是能直接反应汽油燃烧性能的重要指标。因此，在汽油的炼化和精制的过程中，减少辛烷值的损失，不仅能够提高燃烧性能，也能带来巨大的经济效益，这以成为了全世界各国研究的重点。本文的工作就是在收集了大量的某石化企业的催化裂化汽油精制脱硫的长期运行而产生的历史数据上，通过数据挖掘技术，分析354个操作变量和最终产品辛烷值损失之间的关系，建立辛烷值损失的预测模型。进一步的通过建立模型，给出每个样本的优化操作条件，在保证汽油脱硫效果的情况下，降低汽油辛烷值的损失。
At this stage, even today with the rapid development of electric-powered vehicles, from a practical point of view, gasoline-powered vehicles are still the mainstream of road travel and transportation; from the perspective of inventory and replacement cycle, gasoline-powered cars will not withdraw from the stage of history for a long time. The combustion performance of gasoline is an important indicator of the overall quality of gasoline, and the octane number (RON) content is an important indicator that can directly reflect the combustion performance of gasoline. Therefore, in the pro-cess of gasoline refining and refining, reducing the loss of octane can not only improve the combus-tion performance but also bring huge economic benefits, which has become the focus of research in countries all over the world. The work of this paper is to collect a large number of historical data generated by the long-term operation of FCC gasoline refinement and desulfurization of a petro-chemical enterprise, through data mining technology, analyze the relationship between 354 oper-ating variables and the loss of octane number of the final product, and establish a prediction model for octane loss. Further through the establishment of a model, the optimized operating conditions of each sample are given, and the loss of gasoline octane number is reduced under the condition of ensuring the desulfurization effect of gasoline.