Neil A. Duffie

Real-time distributed scheduling of heterarchical manufacturing systems

Abstract Lack of master-slave relationships and absence of a centralized supervisor in heterarchical manufacturing systems prohibits the use of centralized scheduling. A fully distributed scheduling method for these systems is described in this paper. Entities in the system continually generate tentative local schedules in real time using a cooperative scheduling heuristic and select the beast schedule for execution. Tentative schedules are evaluated in time-scaled, distributed simulations using software that is a replica of the real system software. The state of the real system is copied into this virtual system at the beginning of each simulation, eliminating cumulative uncertainties and immediately incorporating equipment failures and other unexpected events into schedule evaluation. Experimental results illustrate the real-time distributed scheduling method and its response to faults.

Nonhierarchical control of manufacturing systems

Abstract An alternative to the high cost of development, maintenance, operation and modification of complex manufacturing systems with hierarchical architectures is under study at the University of Wisconsin-Madison. Nonhierarchical control architectures offer prospects of reduced complexity by localizing information and control, reducing software development costs, increasing maintanability and modifiability, and improving reliability.

Fault-tolerant heterarchical control of heterogeneous manufacturing system entities

Abstract This paper describes a new control architecture and a set of undelying design principles for developing and implementing fault-tolerant manufacturing systems consisting of Processing machines, robots, control computers and human operators. Significant innovations in the approach include the concept of “intelligent manufactured parts”, the use of a “flat” heterarchical architecture as opposed to the widely used hierarchical architecture, unification of simulation and control in system development and operation, integration of human as “colleagues” of other entities rather than “masters”, and achievement of implicit modifiability and fault-tolerance. An experimental heterarchically controlled manufacturing system has been developed in which robotic cells manipulat and manufacture parts. The system is comprised of independent robot, part processing, manufactured part, and human entities that cooperatively control the system through messages exchanged on a communication network. Successful design and implementation of the experimental system has shown that the approach presented in this paper has attractive properties that should be considered in future system designs, addressing a need for new system architectures and design philosophies that result in reduced complexity, higher fault tolerance, shorter development times, and lower development costs.

Non-hierarchical control of a flexible manufacturing cell

Abstract As hierarchically controlled computer-integrated manufacturing systems growthey tend to become complex and their designability, maintainability, expandability and fault tolerance deteriorate. As an alternative, herterarchical control architectures offer prospects of reduced compexity, reduced software development costs, high modularity, high flexibility, and improved fault tolerance. By locating decision making where information originates, global information is reduced to a minimum, scheduling becomes dynamic, machines and parts become “intelligent” entities that cooperatively interact, and the overall system is decomposed into functionally simplified, modular parts. Three flexible machining cell control systems have been constructed at the University of Wisconsin-Madison and are described in this paper: a centralized controller, a hierarchical controller with dynamic scheduling, and a fully distributed heterarchical controller with “intelligent parts”. Comparative results are reported showing that the heterarchical approach possesses a number of advantages including increased fault-tolerance, inherent adaptability and reconfigurability, decreased complexity, and reduced software development cost.

Synthesis of heterarchical manufacturing systems

Abstract The complexity of computer-integrated manufacturing systems with hierarchical architectures grows rapidly with size, resulting in accompanying high costs of development, installation, operation, maintenance, and modification. It is also difficult to introduce fault tolerance into hierarchical structures without significantly increasing system complexity. In the place of a hierarchical structure for manufacturing systems, “cooperative heterarchies” have been suggested in which there are no “higher level” controllers in the system, and each member conforms to certain rules in order to obtain certain privileges. Heterarchical architures offer prospects of: reduced complexity and improved fault tolerance by localizing information and control; reduced software development costs by eliminating supervisory levels; and higher main tainability and modifiability due to improved modularity and self-configurability. Unfortunately, they do not resemble hierarchical management and production control structures, and traditional deterministic design philosophies and methodologies are not necessarily appropriate. This paper is devoted to a discussion of design objectives and philosophies for heterarchical systems. An automated mold production system with a heterarchical architecture is discussed as an example.

Local Capacity

This paper considers the problem of local capacity H∞ control for a class of production networks of autonomous work systems with time-varying delays in the capacity changes. The system under consideration is modeled in a discrete-time singular form. Attention is focused on the design of a controller gain for the local capacity adjustments which maintains the work-in-progress (WIP) in each work system in the vicinity of planned levels and guarantees the asymptotic stability of the system and reduces the effect of the disturbance input on the controlled output to a prescribed level. In terms of a matrix inequality, a sufficient condition for the solvability of this problem is presented using an appropriate Lyapunov function, which depends on the size of the delay and is solved by existing convex optimization techniques. When this matrix inequality is feasible, the controller gain can be found by using LMI Toolbox Matlab. Finally, numerical results are provided to demonstrate the proposed approach.

H_{\infty}

Heterarchical control architectures with fully distributed control have been developed in order to improve responsiveness and effectiveness of manufacturing shop-floor control systems. The dynamics of these highly distributed systems have been difficult to predict particularly when control is based on heuristics. In this paper a dynamical model is developed for a single machine processing an arbitrary number of parts. The structure of the system, which requires queuing of parts when they arrive at a machine, leads to nonlinearities such as dead-zone and discontinuities. A continuous arrival time controller of the integrating type is used that results in a system that can be modeled using nonlinear differential equations that can be solved using a method due to Filippov (1960, 1988). This enables prediction of trajectories of part arrival times and derivation of closed form expressions for steady-state values. The analytical model for the dynamics is validated and the dynamic response of the system is illustrated using numerical simulation.

Control for Production Networks of Autonomous Work Systems With Time-Varying Delays

Abstract A method is described in this paper for obtaining values of coefficients in a kinematic model can be developed Next, the kinematic modal is fitted to position data. Then, coefficients of the model are analyzed so that a diagnosis of the source of errors in the mechanism can be made The results can be used for mechanical or software error correction of machine tools, robots, linkages in sensing mechanisms, etc.

Nonlinear dynamics in distributed arrival time control of heterarchical manufacturing systems

Abstract An analysis of a production planning and control system with closed-loop control of backlog and work-in-progress is presented in this paper, illustrating integration of methods of control engineering with methods of production engineering. The architecture of the system is described and a control-theoretic dynamic model is developed that includes uncertainties in capacity and work input that result from equipment failures, rush orders, etc. Transfer function analysis is used to model dynamic relationships between system inputs and variables including backlog and work-in-progress. The results are used to select control laws for desired system performance and to calculate system response.

Error Diagnosis and Compensation Using Kinematic Models and Position Error Data

Abstract In this paper a discrete dynamic model of a single workstation is used to design and analyse control algorithms for closed-loop PPC that improve performance, especially response to disturbances such as rush orders and periodic fluctuations in capacity, while ensuring that dynamic behavior remains favorable and robust. The presence of delays in adjusting capacity presents challenges in both dynamic analysis and control algorithm design that are addressed in the paper. Methods of control engineering, such as transfer function and frequency response analysis, are used to make analysis of fundamental system properties tractable and to improve control of dynamic behavior.