Date:             Wednesday, March 11, 2020

Time:            6:00 pm

Where:         MS&T – Rolla, MO

Address:      301 W 16th Street, Rolla MO 65409

Parking:

Park in the parking lot right in front of the EE building (Area L). Parking enforcement is only until 4:30 PM, so no need to use the metered parking.

Room:

EE 237, 2nd floor. Enter the main lobby from parking, take last door on your left, take the stairs up, enter the 1st door on your right.

In addition to conducting some Section business we will also have a special guest speaker, Dr. N. Pappa.  See details below.

****************************************************************************

Deep Learning for Control of Bioprocess

Abstract:

The global market demand for many bioproducts has been increasing significantly over the last years due to their application in the human health, food and cosmetics. Hence, the approach to next generation bioproduction has focused on quality of the products and the control of such bioprocesses has gained extensive interest. Most of the study related to applying advanced control and optimization for bioprocess reported challenges in modeling the metabolic reactions associated with the microbial activity requiring state-of-the-art tools for modeling and control.  Due to the uncertainty and lack of knowledge on the system dynamics in many practical bioreactor, the neural network based online closed-loop controller can be employed.

A Deep adaptive online neural network based control scheme that has been proposed for the tracking control of a nonlinear bioprocess with unknown internal dynamics. A feedback linearization controller along with a DNN based function approximation is proposed for a nonlinear lutein production bioprocess. The setpoint trajectory to yield maximum lutein production is efficiently tracked by the proposed online NN based adaptive controller.

Also the Deep neural networks (DNN) based online optimal adaptive control of a class of affine nonlinear systems with uncertain dynamics is introduced. The proposed approach uses two DNNs to solve the optimal regulation over infinite time horizon of nonlinear discrete time systems with known control coefficient matrix and unknown system internal dynamics. With Lyapunov technique, it is shown that all the generated signals are bounded and the estimated optimal control approaches the optimal input with smaller error.

Bio:

Dr. N. Pappa is Professor in the Department of Instrumentation Engineering at MIT Campus, Anna University, India and Currently She is Fulbright Scholar pursuing research under Dr.Jagannathan Sarangapani at Missouri University of Science and Technology on “Deep Learning for Modeling and Control of Bioprocess”.

She holds Ph.D from Anna University and M.Tech from Cochin University. She has excelled in academics throughout and received gold medals in both UG and PG for her academic achievements. She was awarded a project under young scientist FAST TRACK scheme of DST. She has visited foreign countries such as USA, Germany, Australia and Singapore to carry out joint research work and to participate in leading Control Conferences such as ACC, IEEE, IFAC, ALCOSP etc.

Dr. Pappa’s interest includes developing Artificial Neural Network (ANN) based soft sensors, image based measurement, Design of advanced controllers and Realization of soft sensors and controllers in Embedded hardwares.