Ahmet Palazoglu

Nolinear model predictive control using Hammerstein models

Abstract Nonlinear models that are composed of a linear dynamic element in series with a nonlinear static element prove to be very attractive in describing the behaviour of many chemical processes. In this paper, a model predictive control scheme is proposed using the Hammerstein model structure. Two simulation examples, a pH neutralization process and a binary distillation column, are used to demonstrate the effectiveness of the method.

Model predictive control based on Wiener models

Wiener models, consisting of a linear dynamic element followed in series by a static nonlinear element, are considered to be ideal for representing a wide range of nonlinear process behavior. They are relatively simple models requiring little more effort in development than a standard linear model, yet offer superior characterization of systems with highly nonlinear gains. Wiener models may be incorporated into model predictive control (MPC) schemes in a unique way which effectively removes the nonlinearity from the control problem, preserving many of the favorable properties of linear MPC. This paper examines various model structures including ARX and step-response models with polynomial or spline nonlinearities and their corresponding identification strategies. These techniques are then applied to an experimental pH neutralization process where the performance of Wiener MPC is compared with that of the linear MPC and the benchmark PID control to showcase the salient features of this new approach.

Application of Wiener model predictive control (WMPC) to a pH neutralization experiment

pH control is recognized as an industrially important, yet notoriously difficult control problem. Wiener models, consisting of a linear dynamic element followed in series by a static nonlinear element, are considered to be ideal for representing this and several other nonlinear processes. Wiener models require little more effort in development than a standard linear step-response model, yet offer superior characterization of systems with highly nonlinear gains. These models may be incorporated into model predictive control (MPC) schemes in a unique way which effectively removes the nonlinearity from the control problem, preserving many of the favorable properties of linear MPC. In this paper, Wiener model predictive control (WMPC) is evaluated experimentally, and also compared with benchmark proportional integral derivative (PID) and linear MPC strategies, considering the effects of output constraints and modeling error.

Process data de-noising using wavelet transform

The recovery of process information from noisy data de-noising is studied by investigating the classical solution of the estimation problem first. Next, the effectiveness of wavelet-based algorithms for data recovery is considered. A novel method based on coefficient de-noising according to WienerShrink method of wavelet thresholding is proposed. Simulation results are presented, highlighting the advantages of the de-noising method over the classical approaches based on the mean square error criterion.

Prediction of 24-hour-average PM2.5 concentrations using a hidden Markov model with different emission distributions in Northern California

Prediction of air pollutant levels plays an important role in the regulatory plans aimed at the control and reduction of airborne pollutants such as fine particulate matter (PM). Deterministic photochemical air quality models, which are commonly used for regulatory management and planning, are computationally intensive and also expensive for routine predictions. Compared to deterministic photochemical air quality models, data-driven statistical models are simpler and may be more accurate. In this paper, hidden Markov models (HMM) are used to forecast daily average PM(2.5) concentrations 24 h ahead. In conventional HMM applications, observation distributions emitted from certain hidden states are assumed as having Gaussian distributions. However, certain key meteorological factors and most PM(2.5) precursors exhibit a non-Gaussian distribution in reality, which would degrade the HMM performance significantly. In order to address this problem, in this paper, HMMs with log-normal, Gamma and generalized extreme value (GEV) distributions are developed to predict PM(2.5) concentration at Concord and Sacramento monitors in Northern California. Results show that HMM with non-Gaussian emission distributions is able to predict PM(2.5) exceedance days correctly and reduces false alarms dramatically. Compared to HMM with Gaussian distributions, HMM with log-normal distributions can improve the true prediction rate (TPR) by 37.5% and reduce the false alarms by 78% at Concord. And HMM with GEV distribution can improve TPR by 150% and reduce false alarms by 63.62% at Sacramento Del Paso Manor. Comparisons between different distributions used in HMM show that the closer the distribution employed in HMM is to the observation sequence, the better the model prediction performance.

Refinery scheduling of crude oil unloading, storage and processing using a model predictive control strategy

Abstract A model predictive control (MPC) strategy is presented to determine the optimal control decisions for the short-term refinery scheduling problem. For cases where process disturbances occur or new plans need to be implemented during the scheduling period, the moving horizon strategy allows control decisions to be updated effectively to maintain an optimal operation. Furthermore, this strategy takes advantage of information regarding the system and disturbance prediction over the moving horizon to be used in obtaining the control decisions for the given time interval. To demonstrate the performance of the MPC strategy, especially for various moving horizon lengths, three different case studies concerning scheduling problem in a crude oil refinery were used. The refinery includes the shipping vessels, the storage and charging tanks, and the crude distillation units. Several disturbance scenarios regarding mixed oil demands were constructed to illustrate the performance of the proposed strategy.

A multiobjective approach to design chemical plants with robust dynamic operability characteristics

Abstract In chemical plants, operability problems arise mainly due to poor process designs, inaccurate models and/or the control system designs that are unable to cope with process uncertainties. In this paper, a process design methodology is presented that addresses the issue of improving dynamic operability in the present of process uncertainty through appropriate design modifications. The multiobjective nature of the design problem is carefully exploited in the subsequent formulations and a nonlinear programming approach is taken for the simultaneous treatment of both steady-state and dynamic constraints. Scope—Today, a chemical engineer faces the challenge of designing chemical plants that can operate safely, smoothly and profitably within a dynamic process environment. For a typical chemical plant, major contributions to such an environment originate from external disturbances such as variations in the feedstock quality, different product specifications and/or internal disturbances like catalyst poisoning and heat-exchanger fouling. To guarantee a flexible operation despite such upsets, traditionally, the procedure was either to oversize the equipment or to place large storage tanks between the processing units. Proposed design methods attempted to find optimal operating regimes for chemical plants while compensating for process uncertainty through empirical overdesign factors. Studies concerned with the interplay between the process design and operation aspects have appeared recently [1, 2] and focused on achieving better controllability upon modifying the plant design, without explicitly considering process uncertainty. Nevertheless, maintaining satisfactory dynamic operability in an environment of uncertainty remained as a pressing issue and the need was raised quite frequently for a rigorous treatment of the topic [3]. The development of new analytical tools [4, 5] made it possible to consider dynamic operability at the process design stage and modify the plant design accordingly. In this paper, a methodology is presented, that systematically guides the designer towards process designs with better dynamic operability and economics, The problem is formulated within a multiobjective optimization framework and makes extensive use of singular-value decomposition and nonlinear semi-infinite programming techniques. Conclusions and Significance—A multiobjective optimization problem is proposed for designing chemical processes with better dynamic operability characteristics. Robustness indices are used as the indicators of dynamic operability and placed as constraints within the optimization scheme. A semi-infinite nonlinear programming problem results due to the frequency-dependent nature of such constraints. A discretization procedure is suggested to handle the infinite number of constraints and an ellipsoid algorithm allows an interactive solution of the process design problem. A process consisting of three CSTRs is treated as an example, illustrating the potential of the methodology in solving design-related operability problems.

Classification of abnormal plant operation using multiple process variable trends

Abstract This paper illustrates two strategies for the detection and classification of abnormal process operating conditions in which multiple process variable trends are available. The first strategy uses a hidden Markov model (HMM) for overall process classification while the second method uses a back-propagation neural network (BPNN) to determine the overall process classification. The methods are compared in terms of their ability to detect and correctly diagnose a variety of abnormal operating conditions for a non-isothermal CSTR simulation. For the case study problem, the BPNN method resulted in better classification accuracy with a moderate increase in training time compared with the HMM approach.

Robust H∞ control of nonlinear plants based on multi-linear models: an application to a bench-scale pH neutralization reactor

This work is aimed at developing a methodology to design controllers for nonlinear plants where desirable robustness and performance properties must be maintained across a large range of operating conditions. The approach is based on the multi-linear model representation and the H∞ control design that allows inclusion of plant nonlinearities by representing the original system as a set of local uncertain linear plants. To assess the merits of the proposed technique, experiments are performed on a bench-scale pH neutralization reactor. The results demonstrate robust performance and robust stability in the presence of disturbances and set-point variations.

Classification of process trends based on fuzzified symbolic representation and hidden Markov models

Abstract This paper presents a strategy to represent and classify process data for detection of abnormal operating conditions. In representing the data, a wavelet-based smoothing algorithm is used to filter the high frequency noise. A shape analysis technique called triangular episodes then converts the smoothed data into a semi-qualitative form. Two membership functions are implemented to transform the quantitative information in the triangular episodes to a purely symbolic representation. The symbolic data is classified with a set of sequence matching hidden Markov models (HMMs), and the classification is improved by utilizing a time correlated HMM after the sequence matching HMM. The method is tested on simulations with a non-isothermal CSTR and compared with methods that use a back-propagation neural network with and without an ARX model.