Thomas Stanford

Control of nonisothermal CSTR with time varying parameters via dynamic neural network control (DNNC)

Dynamic neural network control (DNNC) is a model predictive control strategy potentially applicable to nonlinear systems. It uses a neural network to model the process and its mathematical inverse to control the process. The advantages of single hidden layer DNNC are threefold: First, the neural network structure is very simple, having limited nodes in the hidden layer and output layer for the SISO case. Second, DNNC offers potential for better initialization of weights along with fewer weights and bias terms. Third, the controller design and implementation are easier than control strategies such as conventional and hybrid neural networks without loss in performance. The objective of this paper is to present the basic concept of single hidden layer DNNC and illustrate its potential. In addition, this paper provides a detailed case study in which DNNC is applied to the nonisothermal CSTR with time varying parameters including activation energy (i.e., deactivation of catalyst) and heat transfer coefficient (i.e., fouling). DNNC is compared with PID control. Although it is clear that DNNC will perform better than PID, it is useful to compare PID with DNNC to illustrate the extreme range of the nonlinearity of the process. This paper represents a preliminary effort to design a simplified neural network-based control approach for a class of nonlinear processes. Therefore, additional work is required for investigation of the effectiveness of this approach for other chemical processes such as batch reactors. The results show excellent DNNC performance in the region where conventional PID control fails.

Model identification of nonlinear time variant processes via artificial neural network

This paper demonstrates that neural networks in conjunction with recursive least squares can be used effectively for model identification of nonlinear time variant processes. The developed approach updates the process model partially at any given sampling time. By updating only a subset of parameters at a given time sample, rather than all network parameters, convergence time is significantly reduced. In addition, meeting the convergence criteria and over-parametrization are less of a problem. The updating approach is applied to a nonisothermal CSTR with time varying parameters and its performance is demonstrated. The resulting approach predicts the process output extremely well and has the ability to learn on-line.

Optimal control for nonlinear systems via dynamic neural network control (DNNC)

Design techniques for nonlinear dynamic systems are closely related to their stability properties. Stability results can be used to design an optimal controller. The paper discusses the stability analysis of the dynamic neural network control (DNNC). The results from the DNNC stability analysis are used to define the neural network stability index (NNSI). NNSI can be used to determine the optimal DNNC network structure and controller. An optimal DNNC is designed for the nonisothermal CSTR as an example of a wide class of nonlinear processes.<<ETX>>

Dynamic neural network control for non-linear systems: optimal neural network structure and stability analysis

Abstract Design techniques for non-linear dynamic systems are closely related to their stability properties. Stability results can be used to design a reliable controller. This paper discusses the stability analysis of the dynamic neural network control (DNNC). The results from DNNC stability analysis will be used to define the neural network stability index (NNSI). The NNSI is a practical index which in current form can only be used with DNNC structures. The NNSI can be used to determine the optimal DNNC network structure. In addition, we will provide guidelines for the design of an optimal DNNC network structure for the conventional neural network structure for model-based control strategies. In this study, DNNC will be designed for a non-isothermal CSTR as an example of a wide class of non-linear processes.

Linear sweep voltammetry in flooded porous electrodes at low sweep rates

Abstract A theoretical analysis of linear sweep voltammetry (LSV) in flooded-porous electrodes is treated for reversible (Nernstian) and first-order irreversible reactions. At low sweep rates, the ohmic potential drop within the electrode is negligible and concentration gradients are predominantly in the axial direction. The solution to the reversible case is mathematically simple, but the results are presented to understand the influence external mass-transfer resistance has on the voltammogram. For irreversible kinetics, a Green’s function technique is used to obtain an analytical solution to the diffusion equation. An analytical solution for the current as a function of the electrode dimensions, sweep rate and reaction kinetic parameters allows one to predict the voltammogram over a wide range of conditions. The analytical solution is used to develop correlations that enable the kinetic parameters (i.e. exchange current density per unit volume and the transfer coefficient) to be easily extracted from experimental data.

Graduate Professional Education of Engineers in Industry for Innovation and Technology Leadership.

Innovation and the development of technology are recognized worldwide as the driving force for competitiveness and economic prosperity. Without diminishing the importance of scientific research, it is now evident that innovation and the development of technology constitute primarily a needs-driven creative professional practice which requires engineering leadership. Experience shows that developing technology and engineers simultaneously is a very effective way of increasing industrys capacity to innovate effectively. In order to contribute most effectively to industrial innovation and the creation of wealth, graduate engineers need to develop their practical and leadership skills as well as their technical knowledge and theoretical understanding throughout their careers. This requires an educational approach driven by the satisfaction of real industrial needs which is commensurate with the professional dimensions of engineering leadership.

LIQUID DISPERSION MECHANISMS IN AGITATED TANKS PART III. LOW VISCOSITY DISCRETE PHASE INTO HIGH VISCOSITY CONTINUOUS PHASE

The dispersion or a low viscosity liquid into a high viscosity liquid was investigated in an agitated tank using a pitched blade turbine. The trailing vortex system was found to be responsible for the formation of ligaments and sheets of the low viscosity liquid. Dispersion, though, was found to occur due to: 1) the break-up of ligaments and 2) small drop production from large drops in a recirculation flow; both dispersion mechanisms were a classical Rayleigh type break-up. The drop size produced in the recirculation flow from large drops was on the order of those observed in the turbulent fragmentation mechanism. The flow, though, was entirely laminar.

Evaluation and implementation of control strategies for moving-bed coal gasifiers using mgas

Abstract The mgas (Morgantown Energy Technology Center gasifier advanced simulation) model can be used to simulate dynamically standard moving bed gasifiers and novel gasifier configurations. The mgas code is written in fortran and takes into account specifications such as reactor geometry, reaction schemes, kinetics and thermodynamics. In this paper, well-proven control methods are demonstrated on a standard moving-bed gasifier using mgas and its suitability for meaningful control studies is evaluated. Steady state analysis tools are used to uncover the optimal control strategies. Dynamic study is also undertaken and three controller schemes found to be the most suitable for the coal gasifier are implemented. The performance of the controller mechanism with optimal tuning parameters is shown for disturbance rejection and for setpoint tracking. With minor modification the mgas model is found to be very useful for designing control strategies for the conventional coal gasifier and therefore should be applied to evaluate the control of more complex novel gasifier configurations.

Constrained multivariable control of a diaphragm-type chlorine/caustic electrolyzers using dynamic matrix control

A multivariable linear dynamic matrix control (DMC) strategy is implemented for a highly nonlinear, electrochemical process with significant dead time. The goal is for the control system to improve energy efficiency by returning the process to the desired operating point or to within control limits in an optimal manner. A comparison of the DMC strategy to traditional control strategies shows improved controller performance with DMC.

Optimal Control of Batch Solution Polymerization of Styrene

Batch solution polymerization is an exothermic chemical reaction described by a highly nonlinear set of state equations. Optimal control of this process is a trajectory tracking problem which is not amenable to solution by classical frequency-domain methods. Using the styrene-toluene -AIBN system as a model, a control algorithm involving a combination of nonlinear model predictive feed-forward and PID feedback control has been examined. Employing this control algorithm, the process is made to follow the optimal temperature trajectory very accurately. Application of the algorithm to other polymerization reactions, including those for copolymers, should be equally effective.