基于粒子群算法的辛烷值损失减少过程的优化  

Optimization of Octane Number Loss Reduction Process Based on Particle Swarm Algorithm

作　　者：丁佳铭 

机构地区：[1]上海理工大学机械工程学院,上海

出　　处：《建模与仿真》2023年第3期2027-2039,共13页Modeling and Simulation

摘　　要：汽油是小型汽车的主要燃料,但其燃烧产生的尾气排放对大气环境有着重要的影响,将汽油清洁化已是迫在眉睫,但同时也要保证其辛烷值的含量。本文运用神经网络的方法针对处理辛烷值(ON)损失预测模型的问题展开研究。对给定的原始采集数据中部分样本进行预处理,通过拉依达准则(3σ准则)去除异常值,采用最大最小的限幅方法剔除一部分不在此范围的样本;然后对于残缺值大于20%的位点可以通过巴莱多定律进行删除,为了选取出对产品属性影响较大的操作变量与因素,使用SPSS软件通过主成分分析法对原有的367个变量进行降维处理,得到了具有典型性和独立性的22个主要变量,通过BP神经网络建立预测模型。为了满足硫含量的条件,利用粒子群算法对硫含量再次预测,得出硫含量与主要变量的表达式,能够得到使得辛烷值损失取得最小值的最优解。Gasoline is the main fuel for small cars, but the exhaust emissions from its combustion have an im-portant impact on the atmospheric environment. It is urgent to clean gasoline, but at the same time, it is necessary to ensure its octane number. In this paper, neural network is used to study the prob-lem of dealing with the prediction model of octane number (ON) loss. Pre-process some samples in the given original collected data, and pass the Laida criterion (3σ Criterion) to remove the outliers, and use the maximum and minimum limiting method to remove some samples that are not in this range;Then, sites with incomplete values greater than 20% can be deleted through the Baredo’s law. In order to select the operational variables and factors that have a greater impact on the prod-uct attributes, the original 367 variables are reduced by using the SPSS software through the prin-cipal component analysis method, and 22 main variables with typicality and independence are ob-tained. The prediction model is established through BP neural network. In order to meet the condi-tions of sulfur content, the particle swarm algorithm is used to predict the sulfur content again, and the expression of sulfur content and main variables is obtained, which can obtain the optimal solu-tion to minimize the loss of octane number.

关 键 词：辛烷值损失 粒子群算法 BP神经网络 主成分分析法 拉依达准则 SPSS软件 预测模型 小型汽车 

分 类 号：TE6[石油与天然气工程—油气加工工程]