基于数据驱动的化工过程参数报警阈值优化

为了提高化工过程报警系统的性能,需要对过程参数的报警阈值进行优化设置。针对传统阈值方法存在的问题,采用核密度估计方法、基于历史数据对过程报警状态进行估计,从最小化误报警和漏报警概率的角度建立了优化过程报警阈值的目标函数,并采用数值优化的方法进行求解。应用于TE过程的仿真结果表明,此方法能够有效地减少过程误报警的次数,并且对漏报警的次数进行抑制。     

Optimal supply chain design and operations under multi‐scale uncertainties: Nested stochastic robust optimization modeling framework and solution algorithm

Although strategic and operational uncertainties differ in their significance of impact, a “one‐size‐fits‐all” approach has been typically used to tackle all types of uncertainty in the optimal design and operations of supply chains. In this work, we propose a stochastic robust optimization model that handles multi‐scale uncertainties in a holistic framework, aiming to optimize the expected economic performance while ensuring the robustness of operations. Stochastic programming and robust optimization approaches are integrated in a nested manner to reflect the decision maker's different levels of conservativeness toward strategic and operational uncertainties. The resulting multi‐level mixed‐integer linear programming model is solved by a decomposition‐based column‐and‐constraint generation algorithm. To illustrate the application, a county‐level case study on optimal design and operations of a spatially‐explicit biofuel supply chain in Illinois is presented, which demonstrates the advantages and flexibility of the proposed modeling framework and efficiency of the solution algorithm. © 2016 American Institute of Chemical Engineers AIChE J, 62: 3041–3055, 2016     

在线压缩KECA的自适应算法在故障检测中的应用

在复杂的大规模工业过程系统中,实时过程监视、优化计算时间和降低运行内存是实现最终产品质量的最关键和最具挑战性的任务,提出一种在线压缩核熵成分分析（online reduced kernel entropy component analysis, ORKECA）的自适应故障检测算法。首先计算训练样本的核矩阵,根据保留的特征值与特征向量选择有代表性的观测值,构造一个符合全局数据信息特征的压缩集,计算监测统计数据的平方预测误差（squared prediction error, SPE）,并利用核密度估计确定控制限。对于在线实时采集的数据,计算该数据的统计量并与压缩集的控制限比较,根据过程状态分析核熵成分分析（kernel entropy component analysis, KECA）模型是否需要进行更新,可以有效提高实时监测过程数据的性能。最后,以一个非线性数值案例及TE过程数据对该方法进行仿真数值分析。结果表明,所提的方法具有有效的可行性。     

A data‐driven multistage adaptive robust optimization framework for planning and scheduling under uncertainty

A novel data‐driven approach for optimization under uncertainty based on multistage adaptive robust optimization (ARO) and nonparametric kernel density M‐estimation is proposed. Different from conventional robust optimization methods, the proposed framework incorporates distributional information to avoid over‐conservatism. Robust kernel density estimation with Hampel loss function is employed to extract probability distributions from uncertainty data via a kernelized iteratively reweighted least squares algorithm. A data‐driven uncertainty set is proposed, where bounds of uncertain parameters are defined by quantile functions, to organically integrate the multistage ARO framework with uncertainty data. Based on this uncertainty set, we further develop an exact robust counterpart in its general form for solving the resulting data‐driven multistage ARO problem. To illustrate the applicability of the proposed framework, two typical applications in process operations are presented: The first one is on strategic planning of process networks, and the other one on short‐term scheduling of multipurpose batch processes. The proposed approach returns 23.9% higher net present value and 31.5% more profits than the conventional robust optimization method in planning and scheduling applications, respectively. © 2017 American Institute of Chemical Engineers AIChE J, 63: 4343–4369, 2017     

Multi‐objective optimization superimposed model‐based process design of an enzymatic hydrolysis process

The concepts of green process engineering and rigorous model‐based approaches have proven to be highly beneficial in process engineering. Although a combination of these two principles thus appears extremely promising, it is not found very commonly in literature. The very high complexity resulting from this combination poses great challenges for the process design and design engineers. Therefore, this work presents an innovative methodology for the model‐based process design with superimposed multi‐objective optimization for an exemplary process. This process for the enzymatic hydrolysis of fatty acid methyl ester combines several aspects of green process engineering and represents an exemplary process with an enzymatic liquid‐liquid‐solid reaction system. The optimization results based on operating and investment costs reveal important insights on the exemplary process and highlight the great advantages of the developed methodology as a profound basis for academic and industrial process design purposes. © 2017 American Institute of Chemical Engineers AIChE J, 63: 1974–1988, 2017     

基于热扩散核密度确定密度峰值法的历史工况识别

在工业生产过程中,生产决策的调整或生产状况的变化会导致生产过程多模态化,常用的数据聚类方法进行工况识别时存在参数选取困难或需要先验知识等限制。基于此,提出一种将人工智能领域的热扩散核密度确定密度峰的技术与高斯混合模型相结合的方法,可有效克服目前方法的缺点。该方法首先利用热扩散核密度确定密度峰的技术估算每个数据点的密度及其与局部密度较大点的距离,获取数据集的聚类中心并完成聚类;其次,利用高斯混合模型获取不同工况的特征参数：平均值、协方差和先验概率,从而对多工况历史过程进行准确的描述;最后,利用文献中仿真数据和Tennessee Eastman过程两个案例进行验证,并与K-均值法和F-J改进的高斯混合模型进行比较,证明了本文提出方法可更加方便、有效地对历史工况进行准确识别。     

基于核熵独立成分分析的故障检测方法

传统核独立成分分析(KICA)依据特征值的大小进行降维,但是特征值大并不一定取得的信息熵贡献度也是最大的。针对这个问题,提出一种基于核熵独立成分分析(KEICA)的故障检测方法。将训练数据集投影在高维核空间,通过对数据信息熵的贡献大小选取核主成分,并建立独立成分分析(ICA)模型。对训练样本求I²和SPE统计量,并利用核密度估计计算统计量的控制限。计算测试数据对训练数据的核矩阵,将其投影在ICA模型上并计算测试样本的统计量,统计量超出控制限的样本即可被识别为故障样本。将该方法用于非线性数值例子和Tennessee Eastman (TE)过程的故障检测,并与传统的核主成分分析(KPCA)、核熵成分分析(KECA)和KICA方法进行对比,表明KEICA的监测效果优于其他三种方法。     

基于核熵独立成分分析的故障检测方法

传统核独立成分分析（KICA）依据特征值的大小进行降维,但是特征值大并不一定取得的信息熵贡献度也是最大的。针对这个问题,提出一种基于核熵独立成分分析（KEICA）的故障检测方法。将训练数据集投影在高维核空间,通过对数据信息熵的贡献大小选取核主成分,并建立独立成分分析（ICA）模型。对训练样本求和统计量,并利用核密度估计计算统计量的控制限。计算测试数据对训练数据的核矩阵,将其投影在ICA模型上并计算测试样本的统计量,统计量超出控制限的样本即可被识别为故障样本。将该方法用于非线性数值例子和Tennessee Eastman（TE）过程的故障检测,并与传统的核主成分分析（KPCA）、核熵成分分析（KECA）和KICA方法进行对比,表明KEICA的监测效果优于其他三种方法。