基于迁移学习的2，6-二甲酚纯度近红外光谱在线检测

使用近红外光谱分析技术,对2,6-二甲酚（2,6-DMP）单体分离工段精馏提纯过程的产品纯度进行在线检测。由于2,6-DMP产品塔的产品纯度较高（通常为99.10%~99.95%）,样本的纯度值集中分布在一个较小的范围,样本数据之间的变异系数小,样本之间的区分度低,导致所采集光谱与产品纯度之间缺乏相关性,因而无法建立可靠的近红外模型。为了准确检测2,6-DMP产品塔的产品纯度,从而对产品质量进行实时调控,引入迁移学习算法,充分利用2,6-DMP精馏提纯过程中不同塔之间近红外光谱数据的相似性,借助于其他塔中较低2,6-DMP纯度的近红外光谱数据,改善产品塔中较高2,6-DMP纯度的近红外模型性能。通过对某合成材料公司2,6-DMP精馏提纯过程的产品纯度建立近红外在线检测模型,验证了所提方法的有效性。分析结果表明,借助于不同样本数量以及纯度范围的近红外光谱数据,所建模型精度有所不同,说明迁移学习算法不仅可以有效解决样本区分度低情形下的建模问题,且建模精度与所借助光谱的样本数量以及纯度范围密切相关。     

基于迁移变分自编码器-标签映射的湿式球磨机负荷参数软测量

在工况改变时，湿式球磨机的实时数据和建模数据分布不一致，不满足传统软测量建模方法要求的数据同分布假设，导致模型失准和性能恶化。为此，引入迁移学习思想，提出一种基于迁移变分自编码器-标签映射的软测量模型，实现多工况下湿式球磨机负荷参数的准确测量。首先，迁移目标域数据编码得到的隐变量分布参数，对源域数据对应隐变量进行拟合，再解码得到迁移数据；然后采用相似性度量选取相似样本构建标签映射模型，并得到映射标签；最后使用迁移数据和映射标签构建出最终的软测量模型。实验结果表明，该软测量方法显著优于现有方法，适用于多工况下的软测量建模。     

基于多尺度核JYMKPLS迁移模型的间歇过程产品质量的在线预测方法

针对过程数据不足,且具有强非线性和多尺度特性的新间歇过程,结合迁移学习方法与多尺度核学习方法的优势,提出了一种基于多尺度核JYMKPLS（Joint-Y multi-scale kernel partial least squares）迁移模型的间歇过程产品质量在线预测方法。该方法首先通过迁移学习利用相似源域的旧过程数据提高新间歇过程建模效率和质量预测的精度。然后,针对间歇过程数据的非线性和多尺度特性问题,引入了多尺度核函数以更好地拟合数据变化的趋势,从而提高模型的预测精度。此外,提出模型在线更新和数据剔除,通过在线持续改善迁移模型对新间歇过程的匹配程度,以消除相似过程间的差异性给迁移学习带来的不利影响,从而不断地提升预测精度。最后,通过仿真验证了所提方法的有效性,结果表明,与传统的数据驱动建模方法相比,本文所提方法能够有效提高建模效率和预测精度。     

基于迁移学习工况划分的裂解炉收率PSO-LS-SVM建模

乙烯裂解炉收率的实时预测对于生产的先进控制及节能降耗具有重要意义。实际生产过程中,不同工况的收率具有较大差别,采用单一工况、单一模型无法满足生产需要。考虑到裂解炉不同运行过程中的相似性,同时为了减小建模过程中典型样本的采集成本,有效利用历史数据,辅以迁移学习算法实现工况的高精度划分。不同工况采用泛化能力强、训练速度高的最小二乘支持向量机建模,并利用粒子群算法对LS-SVM的参数寻优,进一步提高模型精度,从而实现了多工况、多模型的高精度收率预测。基于某乙烯厂现场数据的实验结果表明,多工况、多模型的预测效果更准确合理,PSO优化LS-SVM建立的裂解炉收率模型预测精度更高,趋势跟踪性能良好。     

溶剂油生产两塔分馏单元的建模与应用

本文以溶剂油生产装置的两塔分馏单元为研究对象,针对两塔分馏单元的实际情况,分析其生产工艺特点,如产品模式切换频繁、产品更新换代快、装置运行负荷变化幅度大等,在结合分馏单元实际生产数据的基础上,以化工工艺过程模拟软件Aspen HYSYS V7.1为工具,选用Aspen HYSYS V7.1软件提供的成熟的分馏塔严格计算模块,并对Peng-Robinson、Chao-Seader、Grayson Streed状态方程进行了模拟比较,根据比较结果选择Peng-Robinson为模型的状态方程,最终建立了两塔分馏单元的过程模型。通过对所建立的模型进行验证与应用,数据证明所建立的模型是可行、有效的,能够满足指导实际生产的需要,用于分析产品模式切换动态过程、预测产品性能、预测分馏塔运行状态、降低生产能耗等等,使企业实际生产做到低能耗、高效率,提高企业的市场竞争力。     

基于迁移学习工况划分的裂解炉收率PSO-LS-SVM建模

乙烯裂解炉收率的实时预测对于生产的先进控制及节能降耗具有重要意义。实际生产过程中,不同工况的收率具有较大差别,采用单一工况、单一模型无法满足生产需要。考虑到裂解炉不同运行过程中的相似性,同时为了减小建模过程中典型样本的采集成本,有效利用历史数据,辅以迁移学习算法实现工况的高精度划分。不同工况采用泛化能力强、训练速度高的最小二乘支持向量机建模,并利用粒子群算法对LS-SVM的参数寻优,进一步提高模型精度,从而实现了多工况、多模型的高精度收率预测。基于某乙烯厂现场数据的实验结果表明,多工况、多模型的预测效果更准确合理,PSO优化LS-SVM建立的裂解炉收率模型预测精度更高,趋势跟踪性能良好。     

基于迁移变分自编码器-标签映射的湿式球磨机负荷参数软测量

在工况改变时,湿式球磨机的实时数据和建模数据分布不一致,不满足传统软测量建模方法要求的数据同分布假设,导致模型失准和性能恶化。为此,引入迁移学习思想,提出一种基于迁移变分自编码器-标签映射的软测量模型,实现多工况下湿式球磨机负荷参数的准确测量。首先,迁移目标域数据编码得到的隐变量分布参数,对源域数据对应隐变量进行拟合,再解码得到迁移数据;然后采用相似性度量选取相似样本构建标签映射模型,并得到映射标签;最后使用迁移数据和映射标签构建出最终的软测量模型。实验结果表明,该软测量方法显著优于现有方法,适用于多工况下的软测量建模。     

基于迁移学习的裂解炉产率建模

乙烯裂解炉通常以石油分馏产品为原料,并在高温条件下使长链分子的烃断裂成各种短链的气态烃和少量液态烃,从而获得乙烯、丙烯等烯烃及其他产品.建立这些主要产品的产率模型对裂解炉的先进控制、操作优化等任务在理论和实际上都具有重要意义.尽管在不同裂解原料、不同炉型的裂解炉状况下产品收率均存在差异,但由于裂解炉运行具有半连续性、周期性特征,裂解温度、停留时间及烃分压等因素对裂解产率的影响具有共性,因此为减小建模过程中典型样本采集等成本,有效利用历史数据提高建模精度,有效利用这些不同运行过程中存在的相似性,辅以迁移学习算法实现不同工况下裂解产率的快速建模.相比较以前的研究,此建模方法在少量新数据的情况下充分挖掘了历史数据中包含的信息.最后,以某乙烯厂为研究实例进行裂解产率建模,结果显示能够获得较好的产率预测精度,验证了该建模方法的有效性.     

填料塔改装成乳化塔

我厂有机硅精馏工段,主要分离甲基氯硅烷合成制得的粗单体,粗单体中除二甲基二氯硅烷用连续乳化塔提纯以外,其余组份单体,长期以来,一直使用间歇填料塔。为了平衡分馏设备,提高蒸馏效率,将φ300间歇填料塔改装成乳化塔(间歇式)。该乳化塔的试验成功,进一步证明乳化塔的优越性,以下作简单介绍。一、乳化塔与填料塔对比1.处理能力提高1.4倍     

直径1.2米波纹填料塔试验总结

近代石油化工生产中,原料、半成品、成品的分离提纯往往成为生产上的关键问题。如石油气C_3,馏分的分离,甲基有机硅单体分离等,由于它们沸点十分接近,要分离它们,需要几十块、甚至上百块理论板。这就推动了高效填料塔的研究与应用。 兰州炼油厂气体分馏装置在大闹技术革命、增大处理量的过程中,发现原脱丁烷塔(有40层板的浮阀塔)已处饱和状态,再提高炼量会造成产品质量不合格,故拟建—新脱丁烷塔。在分析了国内外各种传质设备发展状况的基础上,选择了由我国工人阶级和革命技术人     

Modeling of energy consumption in organosilicon monomer fractionation process based on transfer learning

Organosilicon monomer fractionation process needs to separate many substances and the separation requirements are high, leading to high energy consumption. The modeling of energy consumption is of great significance to the optimization of fractionation process. Although the type of equipment and operating conditions of each purification unit in the fractionation process are different, each purification unit has certain similarity according to the dynamic characteristics of the distillation process. Therefore, making full use of these similarities in modeling can reduce the cost of data acquisition in the modeling process. By introducing the transfer learning algorithm, the similarity between the data domains of the source unit and the target unit is increased, the knowledge transfer in different data domains is realized, the knowledge contained in the existing data is fully utilized, and a new energy consumption model is established. Finally, the simulation experiment was carried out with the field data of an organosilicon factory, and the results showed that when the sample size of target domain is little, the transfer learning can effectively improve the performance of the model, so the effectiveness and the practical application of the algorithm were verified.     

Modeling and predictive control of nonlinear processes using transfer learning method

This work develops a transfer learning (TL) framework for modeling and predictive control of nonlinear systems using recurrent neural networks (RNNs) with the knowledge obtained in modeling one process transferred to another. Specifically, transfer learning uses a pretrained model developed based on a source domain as the starting point, and adapts the model to a target process with similar configurations. The generalization error for TL-based RNN (TL-RNN) is first derived to demonstrate the generalization capability on the target process. The theoretical error bound that depends on model capacity and the discrepancy between source and target domains is then utilized to guide the development of pretrained models for improved model transferability. Subsequently, the TL-RNN model is utilized as the prediction model in model predictive controller (MPC) for the target process. Finally, the simulation study of chemical reactors via Aspen Plus Dynamics is used to demonstrate the benefits of transfer learning.     

Multi-source domains transfer learning strategy based on similarity measurement for batch process quality prediction

Generally, when multiple source processes are available for transfer modelling, an inappropriate transfer learning strategy may lead to the deterioration of model accuracy and even cause negative transfer. For this issue, a multi-source domains transfer learning strategy based on similarity measurement is proposed in this work for selecting the optimal transfer modelling strategy according to different scenarios. By counting the data amount and judging the similarity between each source domain and the target domain, the rich information contained in multiple source processes is well utilized to assist the modelling process in the target domain. The proposed strategy provides three alternative specific transfer mechanisms by setting the judgement conditions so that the possibility of negative transfer can get reduced and the accuracy of prediction can be guaranteed. To ensure the reliability of the transfer model and adapt to the change in conditions, the updating scheme based on online data is adopted. Finally, the mixed-effects Gaussian processes modelling method is taken as an example to verify the effectiveness of the proposed multi-source transfer learning strategy through the simulation experiment of the cobalt oxalate synthesis process.     

Modeling of oil near-infrared spectroscopy based on similarity and transfer learning algorithm

Near-infrared spectroscopy mainly reflects the frequency-doubled and total-frequency absorption information of hydrogen-containing groups (0 -H, C-H, N-H, S-H) in organic molecules for near-infrared lights with different wavelengths, so it is applicable to testing of most raw materials and products in the field of petrochemicals. However, the modeling process needs to collect a large number of laboratory analysis data. There are many oil sources in China, and oil properties change frequently. Modeling of each raw material is not only unfeasible but also will affect its engineering application efficiency. In order to achieve rapid modeling of near-infrared spectroscopy and based on historical data of different crude oils under different detection conditions, this paper discusses about the feasibility of the application of transfer learning algorithm and makes it possible that transfer learning can assist in rapid modeling using certain historical data under similar distributions under a small quantity of new data. In consideration of the requirement of transfer learning for certain similarity of different datasets, a transfer learning method based on local similarity feature selection is proposed. The simulation verification of spectral data of 13 crude oils measured by three different probe detection methods is performed. The effectiveness and application scope of the transfer modeling method under different similarity conditions are analyzed.     

高含硫天然气净化装置 分析

针对高含硫天然气净化装置运行能耗高的问题,本文建立了高含硫天然气净化过程中各类 值计算方法,并对离子溶液体系的 值计算方法进行了修正,使其适用于酸气吸收过程中醇胺溶液的 值计算。在天然气净化过程模拟软件ProMax建立的净化过程全流程模型的基础上,采用 分析方法对高含硫净化装置的全流程进行用能分析。分析结果表明,净化装置全流程的 效率为54.2%,其中硫黄回收单元和尾气处理单元 效率最高,分别为66.8%和66.1%;脱酸气单元的 损失最高,占全流程总 损失的43.5%,这是由于净化装置处理的原料气中H2S含量很高,需要更大溶剂循环量才能使净化气达到商品气标准,这导致吸收溶剂再生过程的能耗大大增加。本文研究成果可指导高含硫天然气净化装置的用能评价及节能改造。     

Modification and optimization of benzene alkylation process for production of ethylbenzene

In this paper, an industrial ethylbenzene production unit has been simulated and the results are compared against five-day experimental data. According to prevailing unit condition, i.e. recycled ratio of benzene, benzene selectivity, and energy consumption, the unit is not working under its optimum conditions for minimum cost of ethylbenzene production. In the current design, high amount of benzene recycle (6:1) causes to have an additional cost due to fractionation of ethylbenzene from benzene. A new approach is proposed to modify the benzene alkylation process and reduce the unit's energy consumption. In the newly designed scheme, two double-bed alkylation reactors converted into four single-bed reactors. The amount of injected ethylene, alkylation reactors temperature, and recycled stream are regulated as adjustable parameters for the optimization of the process. In the modified process, the reflux ratio reduced to 1.87 and the benzene selectivity increased. The optimized process shows a considerable decrease in the unit's energy consumption in compare to the current process. Also, the mass fraction of ethylbenzene would reach to 99.12% of purity before entering to the transalkylation reactor for further purification. Therefore, if the presented purity is acceptable for the final application, the transalkylation reactor could be eliminated from the new design.     

ZIF-8浆液法分离CH4/N2的双吸收-吸附塔工艺流程建模与模拟

由于ZIF-8浆液独特的可流动性，可以借鉴传统的吸收-解吸工艺，实现煤层气中甲烷的多级连续高效富集。在单吸收-吸附塔工艺的基础上，为了进一步降低能耗，提出了高低压双吸收-吸附塔新型分离工艺，并对该工艺进行了全流程建模及模拟。采用平衡级法，建立了工艺流程中各单元传质设备的数学模型，包括吸收-吸附塔、闪蒸罐、解吸塔。此外，还进行了灵敏度分析，探究了平衡级数、进料位置、气液比、解吸压力等因素对产品气中甲烷浓度以及回收率等工艺性能的影响。模拟结果表明，当产品气中甲烷浓度达到90.13%（mol）时，回收率为90.25%。并且单位原料气能耗为0.445 kW·h∙m^-3（原料气），低于单塔能耗（0.510 kW·h∙m^-3）。由此，改进的双塔工艺在满足甲烷纯度和回收率的同时，相较于单塔工艺进一步降低了能耗。     

基于深度学习的复杂化工过程软测量模型研究与应用

针对复杂化工生产过程中的一些原材料消耗量难以直接测量的问题，提出了一种基于深度学习的软测量方法。该方法基于一段时间的历史数据，利用平稳小波变换提取历史数据中的多尺度信息，然后与每一个时间点的可观测数据进行合并得到完整的数据集，再划分出训练集和测试集，用带有注意力机制的深度学习算法进行训练和泛化，进而建立软测量模型。最后将提出的方法应用到对苯二甲酸（PTA）生产装置乙酸消耗的软测量中。通过与极限学习机 (extreme learning machine，ELM)、多层感知器 (multi-layer perceptron，MLP)以及普通长短期记忆网络(long short-term memory，LSTM)方法比较，结果表明，该模型的预测准确度较高，具有一定的有效性和适用性，同时对PTA生产装置的乙酸消耗量进行预测分析，从而提高产能和降低能耗。     

原油储运库的全能流模型

将过程系统分析理论应用于原油储运库的用能分析,提出原油储运库全能流模型概念.按照分解-综合-分解的模式,将系统分解为最基本的单元设备,分析用能实质、划分用能过程,建立单元模型;以原油、能量在系统中的转移为线索,结合工艺流程将单元设备综合为系统结构模型;最终给出系统计算框图,结合实际运行数据,确定系统各个节点的水力、热力...