Sachin C. Patwardhan

Fault detection and isolation using correspondence analysis

In this paper, a new approach to fault detection and diagnosis that is based on correspondence analysis (CA) is proposed. CA is a powerful multivariate technique based on the generalized singular value decomposition. The merits of using CA lie in its ability to depict rows as well as columns as points in the dual lower dimensional vector space. CA has been shown to capture association between various features and events quite effectively. The key strengths of CA, for fault detection and diagnosis, are validated on data involving simulations as well as experimental data obtained from a laboratory-scale setup.

Data-driven model based control of a multi-product semi-batch polymerization reactor

Generic model control (GMC) has been successfully used for achieving tight control of batch/semi-batch processes. As the requirement to developing a mechanistic model can prove to be a bottle-neck while implementing GMC, many researchers have recently proposed GMC formulations based on black box models developed using artificial neural networks (ANN). The applicability of most of these formulations is limited to continuously operated systems with relative degree one. In addition, these formulations cannot handle constraints on inputs systematically. In the present study, ANN based GMC (ANNGMC) approach is extended to semi-batch processes with relative order higher than one. The nonlinear time-varying behaviour of batch/semi-batch processes is approximated using ANN model developed in the desired operating region. The ANN model is further used to formulate a nonlinear controller using GMC framework for solving trajectory-tracking problems associated with semi-batch reactors. The control problem at each sampling instant is formulated as a constrained optimization problem whereby the constraints on manipulated inputs can be handled systematically. The proposed controller formulation is used for solving trajectory-tracking problems associated with semi-batch reactors. The performance of the proposed control algorithm is evaluated by simulating the challenge problem proposed by Chylla and Haase (1993), which involves temperature control of a multi-product semi-batch polymerization reactor under widely varying operating conditions. The simulation exercise reveals that the performance of proposed ANNGMC formulation is comparable to the performance of the GMC formulation based on the exact mechanistic model, and is much better than PID controller performance.

SOFT SENSING AND STATE ESTIMATION: REVIEW AND RECENT TRENDS

Abstract Abstract Process monitoring and control requires estimation of quality variables, which are often not measurable directly. A cost effective approach to monitor these variables in real time is to employ model based soft sensing and state estimation techniques. Dynamic model based state estimation is a rich and highly active area of research and many novel approaches have emerged over last over last few years. In this paper, we review recent developments in the area of recursive linear and nonlinear Bayesian state estimation techniques.

Multi-Scenario-Based Robust Nonlinear Model Predictive Control with First Principle Models

A robust nonlinear model predictive control (NMPC) algorithm is developed based on a multi-scenario formulation. The uncertainties are characterized by different scenarios so that the calculated control action is feasible over the entire uncertainty region. We show that this multi-scenario formulation is Input-to-State practically Stable (ISpS), and can be easily extended to the recently proposed advanced-step NMPC (as-NMPC), which is able to reduce the online computational delay. We demonstrate the advantages of this strategy on a large-scale air separation unit.

Control of an Autonomous Hybrid System Using a Nonlinear Model Predictive Controller

Abstract State estimation and estimator based predictive control of nonlinear autonomous hybrid systems poses a challenging problem as these systems involve discontinuities that are introduced by switching of the discrete variables. In this paper, we propose a state estimation scheme for an autonomous hybrid system using an ensemble Kalman filter (EnKF), which belongs to the class of particle filters and is a derivative free nonlinear state estimator. We then proceed to develop a novel nonlinear model predictive control scheme that inherits the approach used in EnKF formulation for future trajectory predictions. The efficacy of the proposed state estimation and control scheme is demonstrated by conducting simulation studies on a benchmark hybrid three-tank system.

Constrained State Estimation Using Particle Filters

Abstract Recursive estimation of constrained nonlinear dynamical systems has attracted the attention of many researchers in recent years. For nonlinear/non-Gaussian state estimation problems, particle filters have been widely used. As pointed out by Daum (2005), particle filters require a proposal distribution and the choice of proposal distribution is the key design issue. In this paper, a novel approach for generating the proposal distribution based on a Constrained Unscented Kalman filter is proposed. The efficacy of the proposed constrained state estimation algorithm using a particle filter (CUPF) is illustrated via a successful implementation on a simulated gas-phase reactor.

Design and implementation fault tolerant model predictive control scheme on a simulated model of a three-tank hybrid system

Significant research has been carried out over the past three decades in the area of fault tolerant control. Most methods available in the chemical engineering literature are capable of detecting, identifying, estimating and accommodating faults for nonlinear processes with continuous states without state dependent and controlled switching. This work is aimed at developing a method for diagnosing and accommodating faults in a nonlinear three-tank hybrid system (autonomous switching of dynamics) using an Ensemble Kalman filter based fault diagnosis scheme. The online accommodation scheme described by Deshpande et al. [2] for nonlinear systems has been extended to nonlinear autonomous hybrid systems using the proposed method. The efficacy of the proposed Fault-tolerant derivative free state estimation based model predictive control scheme is demonstrated by conducting simulation studies on a benchmark three-tank hybrid system.

Robust extended Kalman filter based nonlinear model predictive control formulation

The analysis of the EKF error sequence in [1] is expanded to the case with non-vanishing perturbations. The robust stability of NMPC and EKF pair is established. In addition, we propose an output-feedback NMPC formulation based on both the prior and posterior errors of the EKF to achieve offset-free regulatory control behavior.

Model-Plant Mismatch Detection in MPC Applications using Partial Correlation Analysis

Abstract In model predictive control of processes, the process model plays an important role. The performance of the controller depends on the quality of the model and hence on the model-plant mismatch. Although model-plant mismatch is inevitable, it is highly desirable to minimize it. For processes with large number of inputs and outputs, re-identification of the model is a costly exercise as keeping a large number of inputs in a perturbed or excited state for a long time means loss of normal production time. Hence, it would be highly desirable to detect the precise location of the mismatch so that only a few inputs would have to be perturbed and only the degraded portion of the model updated. In this work, a methodology is proposed for the detection of mismatch from closed-loop operating data. The proposed methodology is based on the analysis of partial correlations between the model residuals and the manipulated variables. Its efficacy is demonstrated on two simulation case studies as well as its application to data from an industrial process.

Internal model control of dc-dc boost converter exhibiting non-minimum phase behavior

This paper investigates the application of two degree-of-freedom (2DOF) internal model controller (IMC) design approach for output voltage regulation of representative boost type dc-dc converter operated in continuous conduction mode (CCM). This system exhibits non-minimum phase behavior due to occurrence of a RHP zero, which poses limitation in the bandwidth available for any control scheme. The IMC structure provides an alternate parameterization of the conventional feedback controllers and is relatively easy to tune to achieve satisfactory servo and regulatory behavior simultaneously. To demonstrate the effectiveness of this 2DOF-EVIC control scheme, simulation studies have been conducted using SEVIULINK platform under different servo and regulatory scenarios. To begin with, simulations are carried out with plant dynamics simulated using linear transfer functions. To assess the feasibility of using the proposed EVIC controller on an experimental setup, the plant dynamics are later simulated using a nonlinear dynamic model. The simulation results clearly imply that the proposed EVIC performs better than the PID controller in linear as well as nonlinear simulations. Moreover, the performance of the IMC tuned using the linear simulation does not change significantly when used for operating the nonlinear plant.