A. Galip Ulsoy

Reconfigurable manufacturing systems: Key to future manufacturing

Presented in this article is a review of manufacturing techniques and introduction of reconfigurable manufacturing systems; a new paradigm in manufacturing which is designed for rapid adjustment of production capacity and functionality, in response to new market conditions. A definition of reconfigurable manufacturing systems is outlined and an overview of available manufacturing techniques, their key drivers and enablers, and their impacts, achievements and limitations is presented. A historical review of manufacturing from the point-of-view of the major developments in the market, technology and sciences issues affecting manufacturing is provided. The new requirements for manufacturing are discussed and characteristics of reconfigurable manufacturing systems and their key role in future manufacturing are explained. The paper is concluded with a brief review of specific technologies and research issues related to RMSs.

Trends and perspectives in flexible and reconfigurable manufacturing systems

To better understand future needs in manufacturing and their enabling technologies, a survey of experts in manufacturing has been conducted. The survey instrument (i.e., questionnaire) tries to assess the experience to date with the use of flexible manufacturing systems (FMS) and to examine the potential roles and enabling technologies for reconfigurable manufacturing systems (RMS). The results show that two-thirds of respondents stated that FMSs are not living up to their full potential, and well over half reported purchasing FMS with excess capacity (which was eventually used) and excess features (which in many cases were not eventually used). They identified a variety of problems associated with FMS, including training, reconfigurability, reliability and maintenance, software and communications, and initial cost. However, despite these issues, nearly 75% of respondent expressed their desire to purchase additional, or expand existing FMSs. The experts agreed that RMS (which can provide exactly the capacity and functionality needed, exactly when needed) is a desirable next step in the evolution of production systems. The key enabling technologies for RMS were identified as modular machines, open-architecture controls, high-speed machining, and methods, training and education for the operation of manufacturing systems.

Analysis of a system of linear delay differential equations

A new analytic approach to obtain the complete solution for systems of delay differential equations (DDE) based on the concept of Lambert functions is presented. The similarity with the concept of the state transition matrix in linear ordinary differential equations enables the approach to be used for general classes of linear delay differential equations using the matrix form of DDEs. The solution is in the form of an infinite series of modes written in terms of Lambert functions. Stability criteria for the individual modes, free response, and forced response for delay equations in different examples are studied, and the results are presented. The new approach is applied to obtain the stability regions for the individual modes of the linearized chatter problem in turning. The results present a necessary condition to the stability in chatter for the whole system, since it only enables the study of the individual modes, and there are an infinite number of them that contribute to the stability of the system. @DOI: 10.1115/1.1568121#

Principal Developments in the Adaptive Control of Machine Tools

Although adaptive control (AC) systems for machine tools have a tremendous potential for improving productivity in manufacturing, their acceptance by industry has been slow. This paper identifies the major research and development areas for AC machine tools, and summarizes the principal developments of the last two decades. Current research at The University of Michigan, which is aimed at the development of stable yet high performance AC systems for turning and milling, is also described.

Reconfigurable manufacturing systems and their enabling technologies

A reconfigurable manufacturing system (RMS) is designed for rapid adjustment of production capacity and functionality in response to new market conditions and new process technology. It has several distinct characteristics including modularity, integrability, customisation, convertibility and diagnosability. There are a number of key interrelated technologies that should be developed and implemented to achieve these characteristics. This paper examines and identifies these technologies. After a brief description of the RMSs and their goals, aspects of reconfiguration (reconfigurable system, software, controller, machine, and process) are explained; this provides one with a better understanding of the enabling technologies of RMSs. Some of the issues related to the technology requirements of RMSs at the system and machine design levels, and ramp -up time reduction are then explained. The paper concludes with descriptions of some of the future research directions for RMSs.

Control of machining processes

This paper reviews the important recent research contributions for control of machining processes (e.g., turning, milling, drilling, and grinding). The major research accomplihsments are reviewed from the perspective of a hierarchical control system structure which considers servo, process, and supervisory control levels. The use and benefits ad advanced control methods (e.g., optimal control, adaptive control) are highlighted and illustrated with examples from research work conducted by the authors

An Approach to Control Input Shaping With Application to Coordinate Measuring Machines

Dimensional measurements obtained with Coordinate Measuring Machines (CMMs) are negatively affected by self-induced structural vibrations. In this paper, a control strategy that reduces the structural vibrations in a CMM is outlined and experimentally demonstrated. The control strategy, designated the Feedforward Filter, is developed by establishing the relationship between contemporary controller input shaping techniques and traditional notch filtering methods. Issues on both robustness and multiple mode vibrations are addressed. Controller input development takes place in the discrete time domain. This method provides results identical to those for optimal command input preshaping obtained through non-linear programming methods and requires considerably less computational effort. Experimental results show a 50 percent reduction in the peak-to-peak magnitude of structural vibrations as compared to unshaped bang-bang trajectories.

Adaptive Sinusoidal Disturbance Rejection in Linear Discrete-Time Systems—Part I: Theory

An adaptive regulation approach against disturbances consisting of linear combinations of sinusoids with unknown and/or time varying amplitudes, frequencies, and phases for SISO LTI discrete-time systems is considered. The new regulation approach proposed is based on constructing a set of stabilizing controllers using the Youla parametrization of stabilizing controllers and adjusting the Youla parameter to achieve asymptotic disturbance rejection. Three adaptive regulator design algorithms are presented and their convergence properties analyzed. Conditions under which the on-line algorithms yield an asymptotic controller that achieves regulation are presented. Conditions both for the case where the disturbance input properties are constant but unknown and for the case where they are unknown and time varying are given. In the case of error feedback, the on-line controller construction amounts to an adaptive implementation of the Internal Model Principle. The performance of the adaptation algorithms is illustrated through a simulation example. A companion paper [4] describes the implementation and evaluation of the algorithms for the problem of noise cancellation in an acoustic duct.

Flexural motion of a radially rotating beam attached to a rigid body

Flexural motion of a radially rotating beam attached to a rigid body is investigated. Fully coupled non-linear equations of motion are derived by using the extended Hamiltons Principle. Spatial dependence is suppressed by using the extended Galerkin method. A torque profile is used to drive the rotating body so that the rigid body motion is not known a priori . The effect of the coupling terms upon the vibration waveforms are investigated by using both a linearized analysis and numerical solution of the differential equations. It is found that for small values of the ratio of the flexible beam and rigid shaft inertia uncoupled equations can lead to substantially incorrect results, particularly with regard to frequencies.

Identification of a Driver Steering Model, and Model Uncertainty, From Driving Simulator Data

Vehicle active safety systems are designed to improve driving safety while the driver is still in control of the vehicle. For the design of such systems, driver-controller interaction can be significant and should not be neglected. Pilutti @1# shows that a lane departure warning system can be improved by considering variations in driver state. An active controller that uses direct intervention of vehicle motion will lend to driver-controller interaction issued, and generally requires a thorough investigation of driver behavior before active safety controllers can be implemented @2‐4 # For vehicle lateral control, the steering wheel angle is the primary means for control actuation. Many driver models try to approximate the real driver’s road tracking performance, assuming certain driver inputs and outputs. Several driver models have been developed in the literature ~e.g., @5‐9#!. A well-known result from human factors research is the ‘‘crossover’’ model @5,6#. The crossover model states that the open loop frequency response of the driver-vehicle combination approximates that of transfer function v c /s around the crossover frequency, where v c is the crossover frequency. Many driver models can be regarded as different realizations of the crossover model ~e.g., @7,8#!. These models have similar characteristics around the crossover frequency and differ more at higher and lower frequency ranges. Models of the driver steering control, based on system identification, which can be used for on-line implementation have also been reported in @10,11#. Although these models approximate the driver behavior well, no driver model is expected to represent the real driver completely. Furthermore, for the purpose of controller design, it is common practice to use a low-order driver model. Therefore, it is reasonable to expect that significant driver model uncertainty exists. This driver model uncertainty can have a significant effect on the performance of the designed control system. To successfully design an active safety controller, including warning and direct intervention, it is necessary to obtain a reasonable representation of this driver model uncertainty. The uncertainty can be used to represent the difference between one real driver and the driver model at any instant in time, or to represent the change in driving behavior with time. On a larger scale, the uncertainty can also be used to represent the variation across several different drivers if the designed system is to be used by different drivers. However, studies of driver model uncertainty have not been reported in the literature. This article aims to develop a model for the driver model uncertainty using the measured data from real drivers driving a fixed-base driving simulator. In general, model uncertainty can be divided into structured uncertainty ~e.g., parametric uncertainty! and unstructured uncertainty ~e.g., additive uncertainty due to unmodeled dynamics!. Driver model uncertainty includes contributions from: model order, parameter uncertainty, and nonlinearity @12#. In this study, the parametric uncertainty is used to represent the variation of driver behavior during a period of time. The unstructured uncertainty is used to account for unmodeled dynamics and nonlinearity of the real driver. Intuitively, higher order models can capture additional dynamic characteristics of the real driver, and can be considered more ‘‘complete.’’ If a linear low order model is used for the driver, the difference in model order will contribute to the model uncertainty. From the controller design viewpoint, the parametric uncertainty may be dealt with using adaptive control techniques and the unstructured uncertainty may be addressed using robust control techniques. The contribution of this research work is a new approach to compute the driver model with parametric and unstructured uncertainty from driving simulator data. 2 Identification of Driver Model From Simulator Data In this study it is proposed to use system identification techniques and driving simulator data to obtain a driver model and the model uncertainty. We consider a black-box driver model where the lateral deviation from the centerline of the road ~y! is treated as the input of the driver. The driver’s output is the steering wheel angle ~d!. The idea of this model is shown schematically in Fig. 1. The objective of this research is to obtain a nominal driver model (Gd) with parametric uncertainty and unstructured model uncertainty ~D! from the driving simulator data. This idea is shown in Fig. 2, where an additive model uncertainty is used. It is noted that the measured d is not completely related to the input signal y. Therefore, a disturbance term e8 is used to represent the portion of the data that cannot be included in D. In order to understand the identification results more clearly, it is helpful to examine the data in more detail.