基于软测量的化工精馏过程推断控制策略

针对化工精馏过程产品成分无法在线检测及其用温度间接控制产品成分的常规控制策略存在着控制精度低的问题,提出基于软测量的精馏过程成分非线性串级推断控制策略。该控制策略首先提出核岭回归的实时软测量方法,即利用满足Mercer条件的核函数改进线性岭回归算法,实现精馏过程产品成分的在线检测;然后在此基础上,提出一种新的非线性串级推断控制策略,即副环采用常规的温度间接控制,主环采用基于核岭回归软测量的推断控制策略。通过Mejedell等建立的精馏塔动态模型分别对单端和双端成分非线性串级推断控制策略性能进行分析,仿真结果表明,与传统控制方案比较,新控制策略的控制质量有了较大提高,控制结构简单,易于实施。     

Optimal selection of soft sensor inputs for batch distillation columns using principal component analysis

In this paper, a novel methodology based on principal component analysis (PCA) is proposed to select the most suitable secondary process variables to be used as soft sensor inputs. In the proposed approach, a matrix is defined that measures the instantaneous sensitivity of each secondary variable to the primary variables to be estimated. The most sensitive secondary variables are then extracted from this matrix by exploiting the properties of PCA, and they are used as input variables for the development of a regression model suitable for on-line implementation.This method has been evaluated by developing a soft sensor that uses temperature measurements and a process regression model to estimate on-line the product compositions for a simulated batch distillation process. The identification of the optimal soft sensor inputs for this case study has been discussed with respect to the definition of the sensitivity matrix, the data sampling interval, the presence of measurement noise, and the size of the input set. The simulation results demonstrate that the proposed approach can effectively identify the size and configuration of the input set that leads to the optimal estimation performance of the soft sensor.     

基于核岭回归估计器的常压塔航煤干点推断控制

提出了一种基于核岭回归推断估计器的新型推断控制策略，来实现常压塔航煤干点的在线检测和控制.首先，对支持向量机与最小二乘支持向量机回归算法进行了分析，并提出一种直接优化核岭回归算法.其次，通过采集的二次变量数据和化验数据，用核岭回归方法建立了航煤干点的估计器模型．最后进行了仿真，结果表明，在相同样本集下，与支持向量机、RBF网络模型比较，所提建模方法调节参数少，预测精度高．     

基于最小二乘支持向量机的航煤干点软测量应用研究

传统支持向量机是近几年发展起来的一种基于统计学习理论的学习机器,在非线性函数回归估计方面有许多应用。最小二乘支持向量机用等式约束代替传统支持向量机方法中的不等式约束,利用求解一组线性方程得出对象模型,避免了求解二次规划问题。本文采用最小二乘支持向量机解决了航空煤油干点的在线估计问题,结果表明,最小二乘支持向量机学习速度快、精度高,是一种软测量建模的有效方法。在相同样本条件下,比RBF网络具有较好的模型逼近性和泛化性能,比传统支持向量机可节省大量的计算时间。     

基于核岭回归方法的定位算法研究

针对无线传感器网络环境下的定位问题,提出了一种基于核岭回归(Kernel ridge regression,KRR)的定位算法。核岭回归算法是在岭回归算法的基础上加入了核函数,该算法在离线阶段采用核岭回归方法提取所有位置指纹数据间的非线性关系,训练出非线性回归定位模型;在线阶段采集目标点的接收信号强度指示（Received signal strength indicator,RSSI）值,利用非线性定位模型估计目标点的物理位置。仿真分析了影响算法性能的各个因素,并在室内典型办公环境下进行了定位实验。实验结果表明,该算法在不同因素的影响下,相比传统加权K近邻算法(Weight K-nearest neighbor,WKNN)算法能达到更好的定位精度,在位置网格间距1.8 m时,WKNN算法平均定位误差为2.53 m,而该算法误差为1.58 m。     

基于塔板组成线的反应精馏图解设计

图解设计法,可以相对较少的已知数据快速、直观地获得反应精馏塔的结构和操作参数。本文在转换组成变量的基础上,系统地推导了反应精馏塔板组成线方程及设计可行性判据,提出了基于塔板组成线进行反应精馏图解设计的策略,并对理想反应体系进行了实例设计计算。研究结果表明：利用塔板组成线进行反应精馏设计可同时获得多组可行设计方案,比现有精馏线和剩余曲线设计方法更为有效,并有利于设计的优化和经济效益评价。     

Inferential control system of distillation compositions using dynamic partial least squares regression

In order to control product compositions in a multicomponent distillation column, the distillate and bottom compositions are estimated from on-line measured process variables. In this paper, inferential models for estimating product compositions are constructed using dynamic Partial Least Squares (PLS) regression, on the basis of simulated time series data. It is found that the use of past measurements is effective for improving the accuracy of the estimation. The influence of selection of measurements and sampling intervals on the performance is also investigated. From the detailed dynamic simulation results, it is found that the cascade control system based on the proposed dynamic PLS model works much better than the usual tray temperature control system.     

Plant-wide design and control of acetic acid dehydration system via heterogeneous azeotropic distillation and divided wall distillation

Energy-saving plant-wide design and plant-wide control of an acetic acid dehydration system with the feed containing methyl acetate and p-xylene are investigated in the study. A heterogeneous azeotropic distillation using isobutyl acetate as an entrainer is designed to obtain high-purity acetic acid at the column bottom and to keep a small acetic acid loss through the top aqueous draw. The accumulation of p-xylene in the column is avoided by adding a side product stream. The mixture in the aqueous phase of decanter, containing mostly water, methyl acetate, and isobutyl acetate is separated using a divided wall distillation column. The whole acetic acid dehydration system includes a heterogeneous azeotropic distillation column and a divided wall distillation column.The control strategies using temperature loops are proposed for this acetic acid dehydration system. For the heterogeneous azeotropic distillation column, the requirements for acetic acid compositions in both the aqueous phase of the decanter and the column bottom can be satisfied by designing entrainer inventory temperature control and cascade temperature control simultaneously. The stages of controlled temperatures are chosen by singular value decomposition and closed-loop analysis methods based on the criteria of minimum entrainer makeup. For the divided wall distillation column, steady-state analysis methods are used for the selection of proper controlled and manipulated variables and the determination of their pairings. Dynamic simulation results demonstrate that the proposed plant-wide control strategy can maintain product purities and reject external disturbances in feed flow and composition changes as well as internal disturbances such as changes in liquid and vapor splits.     

State estimation and inferential control for a reactive batch distillation column

An optimal reflux ratio profile is obtained for a reactive batch distillation system utilizing the capacity factor as the objective function in a nonlinear optimization problem. Then, an Artificial Neural Network (ANN) estimator system, which utilizes the use of several ANN estimators, is designed to predict the product composition values of the distillation column from temperature measurements inferentially. The network used is an Elman network with two hidden layers. The designed estimator system is used in the feedback inferential control algorithm, where the estimated compositions and the reflux ratio information are given as inputs to the controller to see the performance of the ANN. In the control law, a scheduling policy is used and the optimal reflux ratio profile is considered as pre-defined set-points. It is found that, it is possible to control the compositions in this dynamically complex system by using the designed ANN estimator system with error refinement whenever necessary.     

基于核岭回归的非线性内模控制

提出一种基于核蛉回归(KRR)建模的内模控制策略.该方法充分利用基干结构风险最小化为学习规则的回归方法的非线性拟合性能,建立内模控制系统,从理论上分析了内模控制系统的稳定性和稳态误差同逆模与内模估计误差的关系问题.仿真表明,在训练样本有限和有噪声污染情况下,该系统较神经网络方法具有更好的控制性能.     

Identification of a binary distillation column

In this paper the identification of a binary distillation column simulated in a digital computer is described. The pseudo-random binary signal method has been used to disturb two inputs of the column. The well-known cross-correlation technique was used to compute the impulse response matrix G(s), and with a non-linear programming technique a parametric model has been obtained. These results were discretized in order to get a digital transfer matrix, and a discrete state space model suitable for digital control purposes was derived. The reduced-order linear model obtained was compared with the complete non-linear model of the distillation column. Problems involved with correction of the reduced model for any operating conditions are examined and a model-correction algorithm is suggested.     

Modelling of a surface marine vehicle with kernel ridge regression confidence machine

This paper describes the use of Kernel Ridge Regression (KRR) and Kernel Ridge Regression Confidence Machine (KRRCM) for black box identification of a surface marine vehicle. Data for training and test have been obtained from several manoeuvres typically used for marine system identification. Thus, a 20/20 degrees Zig-Zag, a 10/10 degrees Zig-Zag, and different evolution circles have been employed for the computation and validation of the model. Results show that the application of conformal prediction provides an accurate model that reproduces with large accuracy the actual behaviour of the ship with confidence margins that ensure that the model response is within these margins, making it a suitable tool for system identification.     

基于神经网络的解耦控制新方法及其应用

本提出两种基于神经网络的多变量解耦控制方法。方法１通过设计神经网络补偿装置，使得包括补偿神经网络在内的广义对象的Ｂｒｉｓｔｏｌ第一系数矩阵为对角阵；方法２首先定义了神经网络的串联，并联和反馈运算，然后在此基础上设计一个神经网络补偿装置，使得包括补偿神经网络在内的广义对象矩阵为对角阵。将其用于某二元精馏塔的塔顶和塔底组分控制，仿真结果证实了本方法的有效性。     

Two-point control of a reactive distillation column for composition and conversion

Reactive distillation is a hybrid process with dual process objectives: reactant conversion and product composition. Control schemes for reactive distillation frequently neglect the effect of the principal operating parameters on the reactant conversion, and this has a detrimental effect on the overall process profitability. An ETBE reactive distillation column has been used as a case study to show how a two-point control configuration, which recognises the importance of both composition and conversion, can be developed and implemented for a reactive distillation process. The combined composition and conversion control configuration was tested using SpeedUp dynamic simulations and proved to be effective in maintaining a high isobutylene conversion despite process disturbances. The two-point control scheme also had superior disturbance rejection capability, especially for feed rate changes, and composition set-point sensitivity compared with a one-point control scheme.     

Adaptive dynamic matrix control of a distillation column with closed-loop online identification

This paper presents an online identification technique where a process is identified in terms of pseudo impulse response coefficients and subsequently used to update convolution type models to accommodate process-model mismatch. As an example, dynamic matrix control has been applied adaptively to control the top product composition of a distillation column for both servo and regulatory problems. The algorithm automatically detects a large step-like disturbance requiring fresh identification of the process and subsequently adapts the controller to the new model. Simulation studies using an analytical dynamic full order model of a distillation column demonstrated the usefulness of the adaptation scheme. Experimentation on a pilot scale distillation unit vindicated the simulation results.     

On-line tuning strategy for model predictive controllers

This paper presents an intuitive on-line tuning strategy for linear model predictive control (MPC) algorithms. The tuning strategy is based on the linear approximation between the closed-loop predicted output and the MPC tuning parameters. By direct utilization of the sensitivity expressions for the closed-loop response with respect to the MPC tuning parameters, new values of the tuning parameters can be found to steer the MPC feedback response inside predefined time-domain performance specifications. Hence, the algorithm is cast as a simple constrained least squares optimization problem which has a straightforward solution. The simplicity of this strategy makes it more practical for on-line implementation. Effectiveness of the proposed strategy is tested on two simulated examples. One is a linear model for a three-product distillation column and the second is a non-linear model for a CSTR. The effectiveness of the proposed tuning method is compared to an exiting offline tuning method and showed superior performance.     

软测量技术的数据预处理方法研究

针对软测量技术在线实施时的数据预处理问题,提出了基于聚类分析的过失误差侦破方法。该方法不需过程的先验知识和假设,直接面向数据,可十分方便地在线实现。将该方法与滑动平均滤波算法相结合,可以有效处理过程测量数据的过失误差和随机误差,从而提高软仪表估计的精度。在二元精馏塔底产品组分浓度软测量仪表在线进行的仿真中,应用该方法进行数据预处理,使进入软测量模型的过程数据更接近真实值,取得了很好的效果。     

Selftuning control of some pilot plant processes

This paper reviews the results of some experimental tests carried out to evaluate the performance of selftuning (ST) controllers for temperature control of a continuous stirred tank heater, composition control of a binary distillation column and pH control of an acidic effluent. All the pilot plant units have been controlled using a single variant selftuning control, and a newly developed multivariant ST controller was used for simultaneous control of the terminal compositions of the distillation column. The control performance of the units operating under ST control is compared to that obtained using very well tuned proportional plus integral (PI) or proportional plus integral plus derivative (PID) conventional controllers. Control of pH is shown using a technique of electrochemical neutralisation coupled with a single variant ST controller. The control algorithms have been programmed on a number of microprocessor- and minicomputer-based systems. Z80 for stirred tank heater control, LS1-11 for pH control and HP1000 distributed computer system for distillation control.     

大型炼油蒸馏塔的多输入预报自校正控制

本文针对一个大型炼油蒸馏塔的控制,设计了一种适用于具有多个确定性扰动对象的多 输入预报自校正控制器,其特点是综合了在线辨识、多输入前馈、最小方差和PI规律各自的优 点,并采用变遗忘因子的最小二乘估计算法在线修正参数,使控制器的收敛性、稳定性与自适应能力较一般自校正控制器大为提高.实践表明,该控制器很好地克服了生产中的干扰和波 动,平稳了操作,成功地实现了对蒸馏塔的产品质量控制.     

Advanced-multi-step nonlinear model predictive control

Nonlinear model predictive control (NMPC) has gained widespread attention due to its ability to handle variable bounds and deal with multi-input, multi-output systems. However, it is susceptible to computational delay, especially when the solution time of the nonlinear programming (NLP) problem exceeds the sampling time. In this paper we propose a fast NMPC method based on NLP sensitivity, called advanced-multi-step NMPC (amsNMPC). Two variants of this method are developed, the parallel approach and the serial approach. For the amsNMPC method, NLP problems are solved in background multiple sampling times in advance, and manipulated variables are updated on-line when the actual states are available. We present case studies about a continuous stirred tank reactor (CSTR) and a distillation column to show the performance of amsNMPC. Nominal stability properties are also analyzed.