Arne Thesen

SEMI-MARKOV DECISION MODELS FOR REAL-TIME SCHEDULING

Abstract We first present a class of real-time scheduling problems and show that these can be formulated as semi-Markov decision problems. Then we discuss two practical difficulties in solving such problems. The first is that the resulting model requires a large amount of data that is difficult to obtain; the second is that the resulting model usually has a state space that is too large for analytic consideration. Finally, we present a non-intrusive ‘knowledge acquisition’ method that identifies the states and transition probabilities that an expert uses in solving these problems. This information is then used in the semi-Markov optimization problem. A circuit board production line is used to demonstrate the feasibility of this method. The size of the state space is reduced from 2035 states to 308 by an empirical procedure. An ‘optimal’ solution is derived based on the model with the reduced state space and estimated transition probabilities. The resulting schedule is significantly better than the one use...

Design and Evaluation of Tabu Search Algorithms forMultiprocessor Scheduling

Using a simple multiprocessor scheduling problem as a vehicle, we explore the behavior of tabu search algorithms using different tabu, local search and list management strategies. We found that random blocking of the tail of the tabu list always improved performance; but that the use of frequency-based penalties to discourage frequently selected moves did not. Hash coding without conflict resolution was an effective way to represent solutions on the tabu list. We also found that the most effective length of the tabu list depended on features of the algorithm being used, but not on the size and complexity of the problem being solved. The best combination of features included random blocking of the tabu list, tasks as tabus and a greedy local search. An algorithm using these features was found to outperform a recently published algorithm solving a similar problem.

An expert scheduling system for material handling hoists

Abstract A new approach is introduced to the scheduling of hoists on an automated electroplating line. The problem is seen as one of making a sequence of hoist movement decisions in a slowly changing environment. Prior to making any one of these decisions, a knowledge base is consulted to identify the best decision rules for the current system state. This rule is then used to make detailed hoist movement decisions. Simulation experiments were used to identify good rules for different conditions. The resulting information was then used to build the knowledge base. The effectiveness of the resulting system was evaluated using simulation experiments involving real time considerations such as changing work loads and breakdowns. The proposed procedure always performed at least as well as the best of the individual rules in its rule base, and, on the average, a 10% increase in throughput was observed.

Introduction to simulation

An overview of simulation modeling and analysis is presented, from the perspective of prospective users wanting to use simulation as a decision aid. Important considerations in building simulation models and analyzing their outputs are discussed. A brief overview of available software is given. The following guidelines for the beginning analyst are presented: define the objectives before simulating; use the correct level of detail-begin with a simple model; select software that is appropriate for the problem, level of experience, and time frame; and remember that simulation results are observations of random variables, and learn to interpret the results accordingly. >

Design and evaluation of a selective assembly station for high precision scroll compressor shells

Certain automotive parts call for assemblies to be produced to tolerances that cannot be economically reached using standard high volume machining practices. Shims are used instead. We show that the required precision may be reached by using selective assembly. An efficient selective assembly system is proposed. Simulation is used to evaluate the performance of this system, and configurations capable of tolerance improvements of up to 1/20 are suggested.

Some notes on systems models and modelling

Abstract This paper discusses the basic concepts of models and modelling. The difference in information contents between a model and the real or imagined situation it is intended to represent is pointed out, and the fact that a model should not be divorced from its intended purpose is strongly emphasized. The act of modelling is defined and discussed. It is concluded that even though formalized approaches to modelling (such as the systems approach) exists ; modelling, as well as the use of models, remains very much an art rather than a science.

Simulation For Decision Making. An Introduction

We give an overview of simulation modeling and analysis from the perspective of prospective users wanting to use simulation as a decision aid. Important considerations in building simulation models and analyzing their outputs are discussed. A brief overview of available software is given. At the end of this tutorial, you should have a general understanding of simulation and an understanding of its applicability to your situation.

Knowledge aquisition methods for expert scheduling systems

The concept of expert scheduling systems is introduced. The problem of knowledge acquisition in this content is discussed. Two specific techniques used for this purpose are illustrated.

An efficient generator of uniformly distributed random variates between zero and one

A random number generator particularly well suited for simula tion work on microcomputers is presented. The generator dif fers from conventional generators in that the floating point representation of the variate is constructed directly without first going the way of an equivalent integer representation of the variate. Implementations of the generator in FORTRAN-77 and PASCAL/MT+ for the IBM PC are presented, and the results of extensive empirical tests of the generator are provided. The generator is shown to be about as fast as conventional 16-bit congruential generators but with substantially better statistical properties and at least one order of magnitude longer period.

Writing simulations from scratch: Pascal implementations

Techniques for implementing simulation models in Pascal are discussed. Special emphasis is placed on the development of efficient data structures and random number generators. Source codes for efficient but not commonly available algorithms are provided. A floppy disk containing all procedures discussed in the paper is available from the author.