Graduation date: 1996

The main focus of this work is on the problem of existence of nonlinear optimal controllers realizable by artificial neural networks. Theoretical justification, currently available for control applications of neural networks, is rather limited. For example, it is unclear which neural architectures are capable of performing which control tasks. This work addresses applicability of neural networks to the synthesis of approximately optimal state feedback. Discrete-time setting is considered, which brings extra regularity into the problem and simplifies mathematical analysis. Two classes of optimal control problems are studied: time-optimal control and optimal control with summable quality index. After appropriate relaxation of the optimization problem, the existence of a suboptimal feedback mapping is demonstrated in both cases. It is shown that such a feedback may be realized by a multilayered network with discontinuous neuron activation functions. For continuous networks, similar results are obtained, with the existence of suboptimal feedback demonstrated, except for a set of initial states of an arbitrarily small measure. The theory developed here provides basis for an attractive approach of the synthesis of near-optimal feedback using neural networks trained on optimal trajectories generated in open loop. Potential advantages of control based on neural networks are illustrated on application to stabilization of interconnected power systems. A nearly time-optimal controller is designed for a single-machine system using neural networks. The obtained controller is then utilized as an element of a hierarchical control architecture used for stabilization of a multimachine power transmission system. This example demonstrates applicability of neural control to complicated, nonlinear dynamic systems. / Graduation date: 1995

We propose a general methodology based on robust optimization to address the problem of optimally controlling a supply chain subject to stochastic demand in discrete time. The attractive features of the proposed approach are: (a) It incorporates a wide variety of phenomena, including demands that are not identically distributed over time and capacity on the echelons and links; (b) it uses very little information on the demand distributions; (c) it leads to qualititatively similar optimal policies (basestock policies) as in dynamic programming; (d) it is numerically tractable for large scale supply chain problems even in networks, where dynamic programming methods face serious dimensionality problems; (e) in preliminary computation experiments, it often outperforms dynamic programming based solutions for a wide range of parameters. / Singapore-MIT Alliance (SMA)

Most manufacturing processes produce parts that can only be correctly measured after the process cycle has been completed. Even if in-process measurement and control is possible, it is often too expensive or complex to practically implement. In this paper, a simple control scheme based on output measurement and input change after each processing cycle is proposed. It is shown to reduce the process dynamics to a simple gain with a delay, and reduce the control problem to a SISO discrete time problem. The goal of the controller is to both reduce mean output errors and reduce their variance. In so doing the process capability (e.g. Cpk) can be increased without additional investment in control hardware or in-process sensors. This control system is analyzed for two types of disturbance processes: independent (uncorrelated) and dependent (correlated). For the former the closed-loop control increased the output variance, whereas for the latter it can decrease it significantly. In both cases, proper controller design can reduce the mean error to zero without introducing poor transient performance. These finding were demonstrated by implementing Cycle to Cycle (CtC) control on a simple bending process (uncorrelated disturbance) and on an injection molding process (correlated disturbance). The results followed closely those predicted by the analysis. / Singapore-MIT Alliance (SMA)

Timothy L. Johnson, Martin E. Kaliski. / Final Technical Report. / Bibliography: leaf 17a. / Prepared for Air Force Office of Scientific Research Contract F49620-80-C-0002.

Timothy Johnson, Martin E. Kaliski. / Interim report. / Bibliography: leaf 3. / "June 8, 1981." / AFOSR Contract F49620-80-C-0002

by Paul Moroney, Alan S. Willsky, Paul K. Houpt. / Bibliography: leaves 32-34. / "October, 1979." / NASA Ames Grant NGL-22-009-124

by T. Pappas, A.J. Laub, N.R. Sandell, Jr. / Bibliography: leaf 38. / "May 1979." / Contract ERDA-E(49-18)-2087 Contract No. DAAG29-79-C-0031

D.G. Wimpey, T.L. Johnson, M.E. Kaliski. / Bibliography: p. 6. / "April, 1981." / U.S. Air Force Office of Scientific Research contract F49620-80-C-0002

Timothy L. Johnson. / Bibliography: p. 7. / "April, 1981"--report documentation page. "20th CDC." / U.S. Air Force Office of Scientific Research contract F49620-80-C-0002