Hiromu Ohno

Comparison of Multivariate Statistical Process Monitoring Methods with Applications to the Eastman Challenge Problem

Abstract To improve the performance of multivariate statistical process control (MSPC), two advanced methods, moving principal component analysis (MPCA) and DISSIM, have been proposed. In MPCA and DISSIM, an abnormal operation can be detected by monitoring the directions of principal components and the dissimilarity index, respectively. Another important extension of MSPC was made with Multiscale PCA (MS-PCA). The present work investigates the characteristics of several statistical monitoring methods. The monitoring performance is compared with applications to simulated data obtained from a 2×2 process and the Tennessee Eastman process. The superiority of MPCA and DISSIM over the conventional methods comes from the fact that those methods focus on changes in the distribution of process data. Furthermore, the advantages of MPCA or DISSIM over the conventional MSPC and that of MS-PCA are combined, and new methods, termed MS-MPCA and MS-DISSIM, are proposed.

A new multivariate statistical process monitoring method using principal component analysis

Abstract Principal component analysis (PCA) has been used successfully as a multivariate statistical process control (MSPC) tool for detecting faults in processes with highly correlated variables. In the present work, a novel statistical process monitoring method is proposed for further improvement of monitoring performance. It is termed ‘moving principal component analysis’ (MPCA) because PCA is applied on-line by moving the time-window. In MPCA, changes in the direction of each principal component or changes in the subspace spanned by several principal components are monitored. In other words, changes in the correlation structure of process variables, instead of changes in the scores of predefined principal components, are monitored by using MPCA. The monitoring performance of the proposed method and that of the conventional MSPC method are compared with application to simulated data obtained from a simple 2×2 process and the Tennessee Eastman process. The results clearly show that the monitoring performance of MPCA is considerably better than that of the conventional MSPC method and that dynamic monitoring is superior to static monitoring.

Comparison of statistical process monitoring methods: application to the Eastman challenge problem

Abstract Multivariate statistical process control (MSPC) has been successfully applied to chemical procesess. In order to improve the performance of fault detection, two kinds of advanced methods, known as moving principal component analysis (MPCA) and DISSIM, have been proposed. In MPCA and DISSIM, an abnormal operation can be detected by monitoring the directions of principal components (PCs) and the degree of dissimilarity between data sets, respectively. Another important extension of MSPC was made by using multiscale PCA (MS-PCA). In the present work, the characteristics of several monitoring methods are investigated. The monitoring performances are compared with using simulated data obtained from the Tennessee Eastman process. The results show that the advanced methods can outperform the conventional method. Furthermore, the advantage of MPCA and DISSIM over conventional MSPC (cMSPC) and that of the multiscale method are combined, and the new methods known as MS-MPCA and MS-DISSIM are proposed.

Theory and practice of optimal control in continuous fermentation process

In this paper a mathematical model of aminoacid fermentation using bacteria is built based on the kinetic study of experimental data of the pilot plant. This model makes it possible to formulate mathematically two optimal control problems in the actual process: (1) the problem of optimal start-up to a desired steady state, involving the optimization of the steady state, and (2) the problem of dynamical operation aiming at the maximum production of the amino-acid within a specified operation period. Both problems are solved by the maximum principle and Greens theorem. The numerical results suggest one of the most practical patterns of operation.

Multirate multivariable model predictive control and its application to a polymerization reactor

Abstract This paper describes the results of a joint university-industry study to control a semi-commercial impact copolymerization reactor that produces ethylene-propylene-rubber, and mixes it with polypropylene homo-polymer at the molecular level. By changing operating conditions, the reactor can produce numerous grades of polymer according to consumer requests. An appropriate multivarible control system is necessary to achieve swift grade changeovers by changing pressure and concentrations of monomers in the reactor. Because measurements of concentrations, available every 6 minutes, occur less frequently than pressure measurements, we extend the Model Predictive Control (MPC) scheme to deal with this multirate output-sampled process. A 3×3 multirate MPC is developed and installed in the reactor. A thorough description of the features of this control algorithm is also included as well as experimental results to illustrate its performance.

Dissimilarity of Process Data for Statistical Process Monitoring

Abstract For monitoring chemical processes, multivariate statistical process control (MSPC) has been widely used. In the present work, a new process monitoring method is proposed. The proposed method utilizes a change in distribution of process data, since the distribution reflects the corresponding operating condition. In order to quantitatively evaluate the difference between two data sets, the dissimilarity index is defined. The proposed method and the conventional SPC methods are applied to monitoring problems of the Tennessee Eastman process. The results have clearly shown that the monitoring performance of the proposed method is considerably better than that of the conventional methods.

Model predictive control with adaptive disturbance prediction and its application to fatty acid distillation column control

Abstract This paper describes the results of a joint university-industry study to control a fatty acid distillation sequence, which is plagued with severe disturbance problems. In order to solve the disturbance problem, a model predictive control algorithm is modified in terms of disturbance prediction. Assuming that the dynamics of the unmeasured disturbances is generated by an auto-regressive form, the dynamics of the disturbance can be adaptively identified by using time series data of prediction errors and inputs. Using an identified disturbance model with a process model, future outputs are predicted. Control actions are determined so that the predicted output is as close to the target value as possible. This modified model predictive control aglorithm is applied to a ratio control scheme for three distillation columns. The control system developed has been in use sucessfully for more than six years to produce commercial products.

Combined Multivariate Statistical Process Control

Abstract Multivariate statistical process control (MSPC) based on principal component analysis (PCA) has been widely used in chemical processes. Recently, the use of independent component analysis (ICA) was proposed to improve monitoring performance. In the present work, a new method, referred to as combined MSPC (CMSPC). is proposed by integrating PCA-based SPC and ICA-based SPC. CMSPC includes both MSPC methods as its special cases and thus provides a unified framework for MSPC. The effectiveness of CMSPC was demonstrated with its applications to a multivariable system and a CSTR process.

A study on robust stability of model predictive control

Abstract The robust stability of Model Predictive Control (MPC) is analyzed without performing any numerical computations. Some robust stability theorems are newly derived. They assure that a stable MPC system is easily realized only by tuning a couple of control parameters even if some large plant-model mismatch exist. A guideline for the tuning the control parameters is also provided.

Operation Profile Optimization for Batch Process through Wavelet Analysis and Multivariate Analysis

A new regression method, wavelet coefficient regression (WCR), based on wavelet analysis and multivariate analysis is proposed. It can build a statistical model that relates operation profiles with product quality in a batch process. In WCR, selected wavelet coefficients of operation profiles are used as input variables of a statistical model; thus time-related information such as timing of manipulation can be successfully modeled. In addition, by integrating multivariate analysis and wavelet analysis, WCR can cope with correlation of input variables. As a result, WCR enables us to build an accurate statistical model of a batch process. On the basis of WCR, a data-driven method for improving product quality in a batch process is also proposed. The proposed method can determine operation profiles that can achieve the desired product quality and optimize the operation profiles under a given performance index and various constraints. The usefulness of the proposed WCR and profile optimization method is demonstrated through a case study of lysine production based on a semi-batch fermentation process