Rahmat A. Shoureshi

Neural networks for system identification

Two approaches are presented for utilization of neural networks in identification of dynamical systems. In the first approach, a Hopfield network is used to implement a least-squares estimation for time-varying and time-invariant systems. The second approach, which is in the frequency domain, utilizes a set of orthogonal basis functions and Fourier analysis to construct a dynamic system in terms of its Fourier coefficients. Mathematical formulations are presented, along with simulation results. >

Intelligent Control Systems: Are They for Real?

The fundamental concept of feedback to control dynamic systems has played a major role in many areas of engineering. Increases in complexity and more stringent requirements have introduced new challenges for control systems. This paper presents an introduction to and appreciation for intelligent control systems, their application areas, and justifies their need. Specific problem related to automated human comfort control is discussed. Some analytical derivations related to neural networks and fuzzy optimal control as elements of proposed intelligent control systems, along with experimental results, are presented. A brief glossary of common terminology used in this area is included

Failure Detection Diagnostics for Thermofluid Systems

This paper presents experimental results for an on-board microprocessor-based failure detection package designed to assist in the diagnosis of heat pump failures. A model-free limit and trend checking scheme, and a model-based innovations detection formulation operate in parallel to detect anomalous behavior. This dual approach permits the study of tradeoffs between failure detection performance and method complexity. A series of typical anomalies are experimentally simulated in a heat pump, and results are presented to demonstrate the performance of each detection strategy

Robust control for manipulators with uncertain dynamics

Abstract The increase in applications of manipulators in industry and automated manufacturing systems calls for more robust manipulator controllers. A major problem for most of the present control schemes is the inability to deal with dynamic uncertainties associated with a manipulator or its working environment. This paper considers such uncertainties and presents a robust control scheme for industrial manipulators. To illustrate the feasibility of this controller, it has been applied to a General Electric P50 robot. The control design procedure and the implementation results on the GE-P50 robot are presented.

A Modern Control Approach to Active Noise Control

Active noise control systems currently in use and/or described in the research literature are typically bassed on adaptive signal processing theory or, equivalently, adaptive feedforward control theory. This paper presents a modern control aproach to the problem of active noise cancellation in a three-dimensional space. The controller is designed based on a direct self-tuning regulator. Two forms of adaptive control, namely, pole placement and minimum variance controls are considered and compared in simulation

A robust failure diagnostic system for thermofluid processes

Abstract Recent advances in microprocessor technology have enabled the application of process diagnostics to a variety of systems for improved performance and reliability. Failure detection and isolation strategies monitor a systems operation for degradations, and if detected, classify the failure source. In this paper, an on-board diagnostic system will be presented for small-scale thermofluid processes. A model-free limit and trend checking scheme, and a model-based innovations failure detection strategy monitor the system in parallel to detect anomalous behaviour. An experimental-based multiple hypothesis failure isolation strategy statistically classifies the degradations using an a priori failure database. To demonstrate the diagnostic systems performance, a series of failures have been experimentally induced, detected and classified in a residential refrigerator.

Derivation and application of an expert fuzzy optimal control system

Abstract This paper presents derivation of a fuzzy optimal controller which may be used to develop a knowledge based expert control system for supervising and tuning of conventional feedback control loops. A general theory, based on the conceptual framework for optimization in a fuzzy environment, is presented for the development of linear and nonlinear fuzzy optimal control systems. This technique is based on integration of the optimal control theorem and fuzzy variables. The resulting expert controller is expressed in the form of linguistic fuzzy rules, representing the knowledge base for tuning of the feedback control loops. Tuning inferences are made from the response of local feedback control loops by applying the compositional rule of inference to the knowledge base.

Vision-based intelligent control for automated assembly

Development of an automated assembly system requires integration of different engineering modules and coordination of interactions between these modules. This paper presents some of the results of an effort in developing an assembly automation schene for an automated work cell. Integration of the control scheme with the vision module and an on-line trajectory planner is presented. Special characteristics of this automation scheme are dynamic integration of vision and feedback control, real-time operation, uncertainty compensation and error recovery. The hardware required for implementation of this scheme is described. Results of implementing this scheme on a PUMA robot performing a carburetor and gasket mating operation are presented.

Active noise control: a marriage of acoustics and control

In a response to a growing demand for environments of 70 dB or less noise levels, many industrial sectors have focused their attention on the development of quality products with some form of noise control (cancellation) system. Automotive manufacturers, aircraft designers; engine manufacturers; appliance industries; producers of heating, ventilating, and air conditioning systems, etc. all have been involved in formulation and development of noise control mechanisms for their products. Active noise control (ANC) has proven to be the most effective technology, especially for low frequency excitations that the human is more sensitive to. Active noise control requires integration of control theory, acoustic, on-line system identification, and signal processing. Thus, ANC is a cross-disciplinary subject that requires expertise from several areas, especially control and acoustics. The authors research activity in this area over the past eight years has gone through many challenges in order to develop ANC systems that have been feasible and accurate from both control and acoustics viewpoints. This paper summarizes lessons learned from the marriage of acoustics and control, and provides an overview of progresses made in ANC by investigating active noise cancellation from a control point of view. This has been in contrast to almost all other approaches taken from acoustics, modal analysis, and/or adaptive signal processing viewpoint.

Learning and decision making for intelligent control systems

The features of intelligent control systems are briefly described, and early research on machine intelligence is summarized. The development of learning schemes for intelligent control systems is examined, focusing on three current studies in which the author has been involved. >