基于联合典型变量矩阵的多阶段发酵过程质量相关故障监测

为考虑发酵过程的质量变量和动态特征对于阶段划分的影响,提出了一种基于联合典型变量矩阵的多阶段发酵过程质量相关故障监测方法。首先,将历史三维数据沿批次方向展开,对每个时间片矩阵进行典型相关分析（canonical correlation analysis, CCA）,得到融合过程变量和质量变量信息的联合典型变量矩阵,对其进行K均值聚类,实现基于静态特征的第1步划分;然后采用慢特征分析（slow feature analysis, SFA）算法提取表征过程动态性的慢特征,对其进行聚类实现第2步划分。最后综合分析两步划分结果,将生产过程划分为不同的稳定阶段和过渡阶段,并在划分的子阶段中分别建立CCA监测模型进行质量相关故障监测。该方法通过静态和动态特征的变化实现两步划分,准确区分强动态变化与阶段切换,有效提高质量相关的故障监测模型精度。青霉素仿真平台与大肠杆菌实际生产数据验证了所提方法的可行性和有效性。     

基于证据合成的高斯过程回归多模型软测量方法

针对生物发酵过程,提出了一种基于证据理论的高斯过程回归多模型软测量方法,其中多模型融合策略同时考虑了数据聚类特性和软测量子模型统计特性。首先,对聚类后的各子类建立高斯过程回归子模型;然后,基于聚类隶属度函数和高斯过程回归子模型后验概率分别设计子模型权值,并利用证据合成规则将两类权值进行证据合成得到融合权值;最后,将该融合权值作为加权因子对子模型进行融合。通过青霉素发酵过程仿真数据和红霉素发酵过程工业数据研究表明, 相比单一模型和传统多模型高斯过程回归软测量方法,本文所提方法具有较高的预测精度和较小的预测不确定度。     

缺失数据下基于EM算法的非线性过程建模

线性时变参数系统(LPV)将多阶段、非线性的过程建模转化为线性多模型的辨识问题,近年来得到了极大关注。考虑缺失数据下LPV系统的离线建模问题,首先引入一个二进制变量表征输出样本缺失状态,选取过程关键变量作为调度变量,确定主要工况点;然后围绕不同工况点建立局部子模型,将输出缺失部分和采样数据的模型归属当作隐藏变量,利用EM算法进行参数估计,再采用高斯权重函数融合各子模型。最后分别针对典型二阶过程和连续搅拌反应釜(CSTR),运用提出的多模型和算法进行仿真实验,表明有效性。     

PDPSO优化多阶段AR-PCA间歇过程监测方法

针对间歇过程固有的多阶段特性和动态性,提出基于种群多样性的自适应惯性权重粒子群算法（PDPSO）优化的多阶段自回归主元分析（AR-PCA）间歇过程监测方法。该方法引入了PDPSO算法指导AP聚类偏向参数的选取,避免了传统方法依据聚类评价指标选取参考度时的盲目性。对PDPSO优化AP聚类的多阶段发酵过程的数据样本建立AR-PCA模型能够消除各阶段的动态性及变量之间的自相关和互相关影响。最后,对自回归（AR）模型的残差矩阵建立主成分分析（PCA）模型用于发酵过程监测。将该方法应用到青霉素发酵过程,并与传统方法进行对比,结果表明,该方法能够有效进行间歇过程阶段划分并降低故障的漏报和误报。     

PDPSO优化多阶段AR-PCA间歇过程监测方法

针对间歇过程固有的多阶段特性和动态性,提出基于种群多样性的自适应惯性权重粒子群算法(PDPSO)优化的多阶段自回归主元分析(AR-PCA)间歇过程监测方法。该方法引入了PDPSO算法指导AP聚类偏向参数的选取,避免了传统方法依据聚类评价指标选取参考度时的盲目性。对PDPSO优化AP聚类的多阶段发酵过程的数据样本建立AR-PCA模型能够消除各阶段的动态性及变量之间的自相关和互相关影响。最后,对自回归(AR)模型的残差矩阵建立主成分分析(PCA)模型用于发酵过程监测。将该方法应用到青霉素发酵过程,并与传统方法进行对比,结果表明,该方法能够有效进行间歇过程阶段划分并降低故障的漏报和误报。     

基于Gath-Geva算法和核极限学习机的多阶段间歇过程软测量

间歇过程具有较强的非线性,多阶段、慢时变及批次间存在变化,采用单一预测模型不能反映间歇过程的多阶段特性及阶段间过渡特性。提出一种基于Gath-Geva聚类和核极限学习机（kernel extreme learning machine,KELM）的多模型软测量方法。首先采用主成分分析（principal component analysis,PCA）对输入做特征提取,然后利用Gath-Geva算法对间歇过程进行多阶段工况划分,根据生产工况特性划分为不同的操作阶段后,分别建立局部KELM模型。对任一待预测样本,分别计算其对应各局部模型的预测值,最后采用贝叶斯集成,将其隶属于各局部模型的模糊隶属度作为权重和预测值融合得到最终预测值。以青霉素发酵数据进行实验测试,结果表明所提多模型算法相较于单一模型,具有更高的预测精度。     

基于在线聚类和关联向量机的多模型软测量建模

针时软测量样本具有按工况点聚类的特性,提出一种基于在线聚类和关联向量机的多模型软测量建模方法.聚类算法通过设定各辅助变量的权重、按引力原理聚类以及合并子聚类,可把样本按照不同的工作点进行聚类.子模型通过关联向量机实现概率化预测,并采用一种更加有效的核参数选择算法提高算法速度.该建模方法在加氢裂化分馏塔装置的轻石脑油终馏点在线预测系.统中取得了良好的效果.     

基于改进DPC的青霉素发酵过程多模型软测量建模

青霉素发酵过程具有较强的非线性、时变性、阶段性和不确定性,基于单一的软测量模型对产物浓度进行在线估计,难以满足系统对模型精度的要求。针对上述问题,提出一种改进密度峰值聚类的多模型软测量建模方法来估计青霉素发酵过程中的产物浓度。首先,引入相似度函数代替欧氏距离计算样本点的k近邻,并且计算样本点与其k近邻之间的共享近邻,进而利用样本点的k近邻及共享近邻重新定义样本点的局部密度。其次,利用样本点之间的k近邻关系来重新定义样本点的分配策略;通过改进的聚类算法得到各聚类子集,分别建立基于最小二乘支持向量机的软测量模型。Pensim仿真平台的验证结果表明,改进的聚类算法能够更加准确地对样本数据进行聚类,从而有效提高青霉素发酵过程软测量模型的估计精度。     

基于Gath-Geva算法和核极限学习机的多阶段间歇过程软测量

间歇过程具有较强的非线性,多阶段、慢时变及批次间存在变化,采用单一预测模型不能反映间歇过程的多阶段特性及阶段间过渡特性。提出一种基于Gath-Geva聚类和核极限学习机(kernel extreme learning machine,KELM)的多模型软测量方法。首先采用主成分分析(principal component analysis,PCA)对输入做特征提取,然后利用Gath-Geva算法对间歇过程进行多阶段工况划分,根据生产工况特性划分为不同的操作阶段后,分别建立局部KELM模型。对任一待预测样本,分别计算其对应各局部模型的预测值,最后采用贝叶斯集成,将其隶属于各局部模型的模糊隶属度作为权重和预测值融合得到最终预测值。以青霉素发酵数据进行实验测试,结果表明所提多模型算法相较于单一模型,具有更高的预测精度。     

基于改进的快速搜索聚类算法和高斯过程回归的催化重整脱氯前氢气纯度多模型建模方法

氢气是催化重整反应的重要副产物之一,建立氢气纯度软测量模型有助于指导生产。针对催化重整过程工况复杂多变、单一软测量模型难以满足精度要求,提出了一种基于改进的快速搜索聚类算法和高斯过程回归的多模型软测量建模方法。首先,针对快速搜索聚类算法中截断距离是由人为设定的问题,提出了一种截断距离确定方法。并用该改进算法对历史数据进行自动分类,建立各个数据子集的高斯过程回归模型,使各子模型在最大程度上反映不同工况点。然后,针对聚类后得到的带有类别标签的历史数据,建立类别辨识模型,与各子模型相结合,形成开关模式的组合模型。最后,将该建模方法应用于连续催化重整装置,建立了脱氯前氢气纯度的在线计算模型。结果表明,该多模型建模方法具有较高的预测精度,优于传统的单一模型,有一定的实用价值。     

Operating condition diagnosis based on HMM with adaptive transition probabilities in presence of missing observations

A new approach for modeling and monitoring of the multivariate processes in presence of faulty and missing observations is introduced. It is assumed that operating modes of the process can transit to each other following a Markov chain model. Transition probabilities of the Markov chain are time varying as a function of the scheduling variable. Therefore, the transition probabilities will be able to vary adaptively according to different operating modes. In order to handle the problem of missing observations and unknown operating regimes, the expectation maximization algorithm is used to estimate the parameters. The proposed method is tested on two simulations and one industrial case studies. The industrial case study is the abnormal operating condition diagnosis in the primary separation vessel of oil‐sand processes. In comparison to the conventional methods, the proposed method shows superior performance in detection of different operating conditions of the process. © 2014 American Institute of Chemical Engineers AIChE J, 61: 477–493, 2015     

Dynamic modelling and performance optimization of an extractive fermentation

A dynamic model for a continuous extractive fermentation process is proposed. Using this model and the Iterative Dynamic Programming algorithm, optimal operating profiles are identified for process start-up from conventional fermentation operation. Under the optimal operating profile, profit during the transient operation accumulates at a rate approaching that of steady state operation. A sensitivity analysis investigating the effect of economic variations on the overall profit function shows that optimization of manipulated variable profiles primarily reduces the impact of substrate cost, so that the profit function approaches a linear function of the product value.     

一种用于时变系统的复位变参数递推最小二乘算法研究

Based on the idea of the set-membership identification,a modified recursive least squares algorithm with variable gain, variable forgetting factor and resetting is presented.The concept of the error tolerance level is proposed.The selection criteria of the error tolerance level are also given according to the min-max principle.The algorithm is particularly suitable for tracing time-varying systems and is similar in computational complexity to the standard recursive least squares algorithm.The superior performance of the algorithm is verified via simulation studies on a dynamic fermentation process.     

用于催化裂化装置产率预测的Filter-Wrapper特征变量选择方法

催化裂化过程是重质油轻质化的重要手段,为了研究操作条件、原料性质等因素对产品分布的影响,通常需要对催化裂化过程建立准确可靠的数学模型。选择合适的输入变量对模型预测效果有着较大的影响,而在现有的催化裂化装置模型中,输入变量的选取主要依赖于对催化裂化机理的理解。本文从数据驱动建模的角度出发,提出一种Filter法与Wrapper法联合使用的特征子集选择方法。该方法在输入变量选取的过程中不依赖于催化裂化的先验知识,是一种数据驱动的自发的特征变量选择过程。以某炼油厂催化裂化装置为研究对象,利用该装置的生产数据分别选择用于干气和焦炭产率预测模型的输入变量,建立了预测精度高、输入变量数目适中的模型。此外,该方法为催化裂化装置建模的变量选取提供了新角度。     

Multi-objective modeling and optimization for scheduling of cracking furnace systems

Cracking furnace is the core device for ethylene production. In practice, multiple ethylene furnaces are usual y run in parallel. The scheduling of the entire cracking furnace system has great significance when multiple feeds are simultaneously processed in multiple cracking furnaces with the changing of operating cost and yield of product. In this paper, given the requirements of both profit and energy saving in actual production process, a multi-objective optimization model contains two objectives, maximizing the average benefits and minimizing the average coking amount was proposed. The model can be abstracted as a multi-objective mixed integer non-linear programming problem. Considering the mixed integer decision variables of this multi-objective problem, an improved hybrid encoding non-dominated sorting genetic algorithm with mixed discrete variables (MDNSGA-I ) is used to solve the Pareto optimal front of this model, the algorithm adopted crossover and muta-tion strategy with multi-operators, which overcomes the deficiency that normal genetic algorithm cannot handle the optimization problem with mixed variables. Finally, using an ethylene plant with multiple cracking furnaces as an example to illustrate the effectiveness of the scheduling results by comparing the optimization results of multi-objective and single objective model.     

Optimization of rice wine fermentation process based on the simultaneous saccharification and fermentation kinetic model☆

Chinese rice wine making is a typical simultaneous saccharification and fermentation(SSF) process.During the fermentation process,temperature is one of the key parameters which decide the quality of Chinese rice wine.To optimize the SSF process for Chinese rice wine brewing,the effects of temperature on the kinetic parameters of yeast growth and ethanol production at various temperatures were determined in batch cultures using a mathematical model.The kinetic parameters as a function of temperature were evaluated using the software Origin8.0.Combing these functions with the mathematical model,an appropriate form of the model equations for the SSF considering the effects of temperature were developed.The kinetic parameters were found to fit the experimental data satisfactorily with the developed temperature-dependent model.The temperature profile for maximizing the ethanol production for rice wine fermentation was determined by genetic algorithm.The optimum temperature profile began at a low temperature of 26 °C up to 30 h.The operating temperature increased rapidly to 31.9 °C,and then decreased slowly to 18 °C at 65 h.Thereafter,the temperature was maintained at18 °C until the end of fermentation.A maximum ethanol production of 89.3 g·L~(-1)was attained.Conceivably,our model would facilitate the improvement of Chinese rice wine production at the industrial scale.     

基于改进PSO-RBFNN的海洋蛋白酶发酵过程软测量

针对海洋蛋白酶（marine protease,MP）发酵过程中某些关键参量难以在线检测,离线测量存在大滞后、易染菌的问题,提出了一种基于改进的粒子群-径向基神经网络（PSO-RBFNN）的MP发酵过程软测量建模方法。首先采用指数下降惯性权重（exponential decreasing inertia weight,EDIW）策略对粒子群算法进行改进,克服了固定惯性权重和自适应惯性权重的粒子群算法易于陷入局部极小,进化后期收敛速度慢以及全局搜索能力弱的缺点;然后,采用改进后的粒子群算法对径向基神经网络连接权值进行在线优化,确定RBFNN拓扑结构;最后,根据MP发酵过程的输入/输出向量构建RBFNN软测量模型。实验仿真结果表明,EDIW策略改进的PSO-RBFNN软测量模型训练时间缩短了40%左右,模型预测精度提高了3%以上。