论文部分内容阅读
对于间歇蒸馏过程,提前准确判断从低馏分到主馏分的转馏分点是影响最终产品质量和产量的关键环节,设计基于数据挖掘技术的转馏分点在线预报软测量系统是一项重要的过程质量控制手段,为提高生产的综合自动化水平创造了重要条件。根据混沌理论,温度能较高程度的反映体系内反应及分离情况,因此选取间歇蒸馏上升气温度为考察变量。针对数据非线性、动态、数据长度短、不同批次数据不等长等特点,提出了将不同批次数据按照随机的顺序首尾相接组成长数据集的数据重构策略;采用自回归求和滑动平均方法和最小二乘支持向量机方法建立了组合时间序列预测模型;通过对理论转馏分温度与实际转馏分温度的差值和预测曲线近似斜率的统计分析,建立了转馏分点在线预报系统,经过在实际生产中的验证,实现了对转馏分点提前1min的准确预报。
For the batch distillation process, it is an important process quality to design an accurate on-line forecasting soft measurement system based on data mining technology in order to accurately judge in advance the distillation fraction from the low fraction to the main fraction, which is the key point affecting the final product quality and yield. Control measures have created important conditions for improving the overall level of automation of production. According to the chaos theory, the temperature can reflect the reaction and separation in the system to a higher degree. Therefore, the rising temperature of intermittent distillation is selected as the investigation variable. Aiming at the characteristics of non-linearity, dynamic data length, short data length and unequal data of different batches, a data reconstructing strategy was proposed to reconstruct the data of different batches of data into groups according to random order. Sliding average method and least square support vector machine method to establish the combined time series forecasting model. Through the statistical analysis of the difference between the theoretical temperature and actual distillate temperature and the approximate slope of the predictive curve, an online forecasting system , After the actual production of the verification, to achieve a fraction of 1 minutes ahead of the distillate accurate forecast.