Dynamic Scheduling Environment Research Articles

A DRL-Based Reactive Scheduling Policy for Flexible Job Shops With Random Job Arrivals

In real-life production systems, arrivals of jobs are usually unpredictable, which makes it necessary to develop solid reactive scheduling policies to meet delivery requirements. Deep reinforcement learning (DRL) based scheduling methods are capable of quickly responding to dynamic events by learning from the training data. However, most of policy networks in DRL algorithms are trained to choose priority dispatching rules (PDR), thus, to some extent, the efficiency of obtained scheduling plans is limited by the performance of PDRs. This paper investigates a dynamic flexible job shop scheduling problem with random job arrivals for the total tardiness minimization. A DRL-based reactive scheduling method, proximal policy optimization with attention-based policy network (PPO-APN), is proposed to make real-time decisions for the dynamic scheduling environment, where the attention-based policy network (APN) is able to directly select pending jobs distinguished from the action space that consists of PDRs. Additionally, a global/local reward function (GLRF) is designed to address the reward sparsity issue during training processes. The proposed PPO-APN is tested on randomly generated instances with different production configurations, and is compared with frequently-used PDRs and DRL-based methods. Numerical experimental results indicate that APN and GLRF components significantly improve the training efficiency, and the PPO-APN shows better overall performance compared with other methods. <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Note to Practitioners</i> —This work is motivated by a typical production scenario in discrete manufacturing systems, where orders randomly arrive at the shop floor and require to be scheduled in a short time to ensure the on-time delivery. Previous research work tends to apply DRL algorithms to choose suitable dispatching rules for the ease of implementation. Nevertheless, the jobs that can be selected by dispatching rules are rather limited, thus many possible high-quality scheduling plans are ignored. This work first sorts all the unscheduled jobs by a heuristic algorithm, and puts some of top-ranked jobs to a pool. When a machine becomes available, it will directly choose a job from the pool as the next processing task. The job selection policy is represented by a novel attention-based network, and is trained by a powerful DRL algorithm. The aforementioned process is repeatedly executed in a simulation environment to collect the training data. Therefore, after being trained for a certain period of time, the policy will become smarter and can be applied to make right decisions in real-time. The proposed reactive scheduling method has been proved to be more efficient than dispatching rules and DRL-based approaches, and is effective in the production scheduling for a wide variety of discrete manufacturing scenarios, such as automobile and electronics industries. Moreover, the proposed method can be further extended to address dynamic scheduling problems with some production characteristics via adding constraints for the job selection or re-defining calculations for the completion time of operations accordingly.

Knowledge extraction for solving resource-constrained project scheduling problem through decision tree

PurposeProject scheduling plays an essential role in the implementation of a project due to the limitation of resources in practical projects. However, the existing research tend to focus on finding suitable algorithms to solve various scheduling problems and fail to find the potential scheduling rules in these optimal or near-optimal solutions, that is, the possible intrinsic relationships between attributes related to the scheduling of activity sequences. Data mining (DM) is used to analyze and interpret data to obtain valuable information stored in large-scale data. The goal of this paper is to use DM to discover scheduling concepts and obtain a set of rules that approximate effective solutions to resource-constrained project scheduling problems. These rules do not require any search and simulation, which have extremely low time complexity and support real-time decision-making to improve planning/scheduling.Design/methodology/approachThe resource-constrained project scheduling problem can be described as scheduling a group of interrelated activities to optimize the project completion time and other objectives while satisfying the activity priority relationship and resource constraints. This paper proposes a new approach to solve the resource-constrained project scheduling problem by combining DM technology and the genetic algorithm (GA). More specifically, the GA is used to generate various optimal project scheduling schemes, after that C4.5 decision tree (DT) is adopted to obtain valuable knowledge from these schemes for further predicting and solving new scheduling problems.FindingsIn this study, the authors use GA and DM technology to analyze and extract knowledge from a large number of scheduling schemes, and determine the scheduling rule set to minimize the completion time. In order to verify the application effect of the proposed DT classification model, the J30, J60 and J120 datasets in PSPLIB are used to test the validity of the scheduling rules. The results show that DT can readily duplicate the excellent performance of GA for scheduling problems of different scales. In addition, the DT prediction model developed in this study is applied to a high-rise residential project consisting of 117 activities. The results show that compared with the completion time obtained by GA, the DT model can realize rapid adjustment of project scheduling problem to deal with the dynamic environment interference. In a word, the data-based approach is feasible, practical and effective. It not only captures the knowledge contained in the known optimal scheduling schemes, but also helps to provide a flexible scheduling decision-making approach for project implementation.Originality/valueThis paper proposes a novel knowledge-based project scheduling approach. In previous studies, intelligent optimization algorithm is often used to solve the project scheduling problem. However, although these intelligent optimization algorithms can generate a set of effective solutions for problem instances, they are unable to explain the process of decision-making, nor can they identify the characteristics of good scheduling decisions generated by the optimization process. Moreover, their calculation is slow and complex, which is not suitable for planning and scheduling complex projects. In this study, the set of effective solutions of problem instances is taken as the training dataset of DM algorithm, and the extracted scheduling rules can provide the prediction and solution of new scheduling problems. The proposed method focuses on identifying the key parameters of a specific dynamic scheduling environment, which can not only reproduces the scheduling performance of the original algorithm well, but also has the ability to make decisions quickly under the dynamic interference construction scenario. It is helpful for project managers to implement quick decisions in response to construction emergencies, which is of great practical significance for improving the flexibility and efficiency of construction projects.

A Dynamic Scheduling Method for Carrier Aircraft Support Operation under Uncertain Conditions Based on Rolling Horizon Strategy

The efficient scheduling of carrier aircraft support operations in the flight deck is important for battle performances. The supporting operations and maintenance processes involve multiple support resources, complex scheduling process, and multiple constraints; the efficient coordination of these processes can be considered a multi-resource constrained multi-project scheduling problem (MRCMPSP), which is a complex non-deterministic polynomial-time hard (NP-hard) problem. The renewable resources include the operational crews, resource stations, and operational spaces, and the non-renewable resources include oil, gas, weapons, and electric power. An integer programming mathematical model is established to solve this problem. A periodic and event-driven rolling horizon (RH) scheduling strategy inspired by the RH optimization method from predictive control technology is presented for the dynamic scheduling environment. The periodic horizon scheduling strategy can track the changes of the carrier aircraft supporting system, and the improved event-driven mechanism can avoid unnecessary scheduling with effective resource allocation under uncertain conditions. The dual population genetic algorithm (DPGA) is designed to solve the large-scale scheduling problem. The activity list encoding method is proposed, and a new adaptive crossover and mutation strategy is designed to improve the global exploration ability. The double schedule for leftward and rightward populations is integrated into the genetic process of alternating iterations to improve the convergence speed and decrease the computation amount. The computational results show that our approach is effective at solving the scheduling problem in the dynamic environment, as well as making better decisions regarding disruption on a real-time basis.

Comparison of ensemble learning methods for creating ensembles of dispatching rules for the unrelated machines environment

Dispatching rules are often the method of choice for solving various scheduling problems, especially since they are applicable in dynamic scheduling environments. Unfortunately, dispatching rules are hard to design and are also unable to deliver results which are of equal quality as results achieved by different metaheuristic methods. As a consequence, genetic programming is commonly used in order to automatically design dispatching rules. Furthermore, a great amount of research with different genetic programming methods is done to increase the performance of the generated dispatching rules. In order to additionally improve the effectiveness of the evolved dispatching rules, in this paper the use of several different ensemble learning algorithms is proposed to create ensembles of dispatching rules for the dynamic scheduling problem in the unrelated machines environment. Four different ensemble learning approaches will be considered, which will be used in order to create ensembles of dispatching rules: simple ensemble combination (proposed in this paper), BagGP, BoostGP and cooperative coevolution. Additionally, the effectiveness of these algorithms is analysed based on some ensemble learning parameters. Finally, an additional search method, which finds the optimal combinations of dispatching rules to form the ensembles, is proposed and applied. The obtained results show that by using the aforementioned ensemble learning approaches it is possible to significantly increase the performance of the generated dispatching rules.

A Prism-Based and Gap-Based Approach to Shopping Location Choice

In this paper we present a prism-based and gap-based approach to model shopping location choice. Location choice is a fundamental decision in the activity scheduling process. We propose a simple yet robust model to capture shopping location choice behaviour. In this model, an individual first chooses a time window (or gap); the choice of the shopping location depends on the gap chosen. This notion arises from our understanding that shopping location choice behaviour depends on shopping type, scheduling constraints, time of day, and day of week. Or quite simply, where you shop depends on when you shop. The gap-based approach to destination choice is envisioned as a small but significant step towards a more comprehensive location choice model in a dynamic scheduling environment.

Reoptimization of the minimum total flow-time scheduling problem

We consider reoptimization problems arising in dynamic scheduling environments, such as manufacturing systems and virtual machine managers. Due to changes in the environment (out-of-order or new resources, modified jobs’ processing requirements, etc.), the schedule needs to be modified. That is, jobs might be migrated from their current machine to a different one. Migrations are associated with a cost – due to relocation overhead and machine set-up times. In some systems, a migration is also associated with job extension. The goal is to find a good modified schedule, with a low transition cost from the initial one. We consider the objective of minimizing the total flow-time.We present optimal algorithms for the problem of achieving an optimal solution using the minimal possible transition cost. The algorithms and their running times depend on our assumptions on the instance and the allowed modifications. For the modification of machines’ addition, we also present an optimal algorithm for achieving the best possible schedule using a given limited budget for the transition.

Advanced dispatching rules for large-scale manufacturing systems

Dispatching rules have been successfully applied to job sequencing and scheduling in large-scale manufacturing systems such as wafer fabrication plants, automatic guided vehicle systems, etc. Because they can be easily communicated and implemented, and because they can be speedily applied, dispatching rules are also one of the most prevalent approaches in this field. However, naysayers often criticize the sluggish performance levels of traditional dispatching rules. Furthermore, in many large-scale factories, scheduling systems have been installed and operational for more than 5 years with “satisfactory” results, but managers still believe that more beneficial modifications are possible. Specifically, better scheduling methods, dispatching rules, test environments, and reporting tools are needed. Over the years, a few new solutions have been proposed to address these issues. For instance, most traditional dispatching rules are based on historical data. With the emergence of data mining and online analytic processing, dispatching rules can now take predictive information into account. Further, rather than concentrating on a single performance measure, some dispatching rules are designed to optimize multiple objectives at the same time. Moreover, the content of a dispatching rule can be optimized for a largescale manufacturing system. In light of advanced computing systems, dispatching rules continue to be one of the most promising technologies for practical applications. This special issue focuses on innovative but practical dispatching rules rather than complex algorithms. This type of dispatching rule will continue to drive the mainstream of practical applications in factories for the foreseeable future. This special issue is intended to provide the details of advanced dispatching rule development and applications of those rules to job sequencing and scheduling in large-scale manufacturing systems. We are very grateful for the positive responses we have received from the authors who submitted papers and the marvelous help provided by a number of referees in the paper reviewing process. After a strict review, 25 papers were finally accepted for publication in this special issue. Zhang et al. used a genetic algorithm (GA) to optimize a set of dispatching rules for scheduling a job shop. Bayesian networks were also utilized to model the distribution of high-quality solutions in the population and to produce each new generation of individuals. In addition, some selected individuals were further improved by a special local search. One advantage of their method is that it can be readily applied in various dynamic scheduling environments which must be investigated with simulation. Lu and Romanowski considered a dynamic job shop problem in which job shops are disrupted by unforeseen events such as job arrivals and machine breakdowns. They used multi-contextual functions (MCFs) to describe the unique characteristics of a dynamic job shop at a specific time and examined 11 basic dispatching rules and 33 composite rules made with MCFs that describe machine idle time and job waiting time. The experimental data showed that schedules made by the composite rules outperformed schedules made by conventional rules. Lin et al. integrated an ant colony optimization (ACO) algorithm with a number of new ideas (heuristic initial solution, machine reselection step, and local search procedure) and T. Chen (*) Department of Industrial Engineering and Systems Management, Feng Chia University, Taichung, Taiwan e-mail: tolychen@ms37.hinet.net

Evaluation of the impact of information delays on flexible manufacturing systems performance in dynamic scheduling environments

The significance of the adverse effects of information delays (IDs) on flexible manufacturing system (FMS) performance is getting increasing attention from manufacturing systems managers, planners, schedulers, Enterprise Resource Planning software developers, and researchers because of their potential to disrupt production schedules. In this paper, we examine the extent of the adverse impact that IDs have on FMS performance. The FMSs are assumed to operate in a “review period” mode; i.e., the control decisions (e.g., sequencing and dispatching) are taken based on information monitored at predetermined intervals of time called “review periods”. The performance deterioration occurs due to the obsolescence of system status information. Key empirical findings based on extensive simulation experiments are: (1) IDs significantly degrade FMS performance for due date-based measures (mean tardiness and percent of jobs tardy); (2) IDs also degrade FMS performance for non-due date-based measures (mean flowtime and average machine utilization), albeit to a less severe degree; (3) routing flexibility, often regarded as a significant factor to influence FMS performance, is superseded by status review information delay.

A dispatching rule-based hybrid genetic algorithm focusing on non-delay schedules for the job shop scheduling problem

Job shop scheduling is an important decision process in contemporary manufacturing systems. In this paper, we aim at the job shop scheduling problem in which the total weighted tardiness must be minimized. This objective function is relevant for the make-to-order production mode with an emphasis on customer satisfaction. In order to save the computational time, we focus on the set of non-delay schedules and use a genetic algorithm to optimize the set of dispatching rules used for schedule construction. Another advantage of this strategy is that it can be readily applied in a dynamic scheduling environment which must be investigated with simulation. Considering that the rules selected for scheduling previous operations have a direct impact on the optimal rules for scheduling subsequent operations, Bayesian networks are utilized to model the distribution of high-quality solutions in the population and to produce the new generation of individuals. In addition, some selected individuals are further improved by a special local search module based on systematic perturbations to the operation processing times. The superiority of the proposed approach is especially remarkable when the size of the scheduling problem is large.

Optimization of schedule stability and efficiency under processing time variability and random machine breakdowns in a job shop environment

AbstractThe ability to cope with uncertainty in dynamic scheduling environments is becoming an increasingly important issue. In such environments, any disruption in the production schedule will translate into a disturbance of the plans for several external activities as well. Hence, from a practical point of view, deviations between the planned and realized schedules are to be avoided as much as possible. The term stability refers to this concern. We propose a proactive approach to generate efficient and stable schedules for a job shop subject to processing time variability and random machine breakdowns. In our approach, efficiency is measured by the makespan, and the stability measure is the sum of the variances of the realized completion times. Because the calculation of the original measure is mathematically intractable, we develop a surrogate stability measure. The version of the problem with the surrogate stability measure is proven to be NP‐hard, even without machine breakdowns; a branch‐and‐bound algorithm is developed for this problem variant. A tabu search algorithm is proposed to handle larger instances of the problem with machine breakdowns. The results of extensive computational experiments indicate that the proposed algorithms are quite promising in performance. © 2011 Wiley Periodicals, Inc. Naval Research Logistics, 2011

Robust airline schedule design in a dynamic scheduling environment

In the past decade, major airlines in the US have moved from banked hub-and-spoke operations to de-banked hub-and-spoke operations in order to lower operating costs. In Jiang and Barnhart (2009) [1], it is shown that dynamic airline scheduling, an approach that makes minor adjustments to flight schedules in the booking period by re-fleeting and re-timing flight legs, can significantly improve utilization of capacity and hence increase profit. In this paper, we develop robust schedule design models and algorithms to generate schedules that facilitate the application of dynamic scheduling in de-banked hub-and-spoke operations. Such schedule design approaches are robust in the sense that the schedules produced can more easily be manipulated in response to demand variability when embedded in a dynamic scheduling environment. In our robust schedule design model, we maximize the number of potentially connecting itineraries weighted by their respective revenues. We provide two equivalent formulations of the robust schedule design model and develop a decomposition-based solution approach involving a variable reduction technique and a variant of column generation. We demonstrate, through experiments using data from a major U.S. airline that the schedule generated can improve profitability when dynamic scheduling is applied. It is also observed that the greater the demand variability, the more profit our robust schedules achieve when compared to existing ones.

A Dynamic Resource Management and Scheduling Environment for Embedded Multimedia and Communications Platforms

We present a framework, OpenCLosE, for dynamic resource management and scheduling of applications written in open compute language (OpenCL) for heterogeneous multimedia and graphics platforms, such as those found in multimedia smartphones and automotive infotainment clusters. We describe the design of a resource manager and master scheduler for the OpenCLosE environment, that allows efficient realization of multiple applications within a multitasked platform.

Intelligent Job Shop Scheduling Based on MAS and Integrated Routing Wasp Algorithm and Scheduling Wasp Algorithm

Dynamic scheduling algorithms are gaining more and more special attention for their satisfying robustness when confronted with unexpected events as well as their considerably high performance in scheduling. An organization structure for intelligent job shop scheduling based on MAS (Multiple Agents System) is put forward to achieve effective and efficient production. A hybrid multilayer agent structure is put forward to facilitate constructing various agents. Consequently, all resources in a manufacturing system are reorganized into an agile manufacturing network of agents of autonomous and cooperative characteristics. Regarding wasps as a specific kind of agents, wasp colony algorithms are used to solve job shop dynamic scheduling problem. Based on the principle of the wasp colony algorithm, two different algorithms, namely the routing wasp algorithm and the scheduling wasp algorithm, are combined to solve the job shop dynamic scheduling problem. The algorithms are modified to better adapt to job shop dynamic scheduling environment. The algorithms are developed based on Eclipse 3.2 and J2SE 6.0. Simulation experiments are accomplished and experimental data are analyzed. The results show that the principle of the algorithms is simple, their computational quantity is small, and they can be applied to multi-batch dynamic scheduling with unpredictable entry time due to their favorable potential.

FAST

Energy is a valuable resource in embedded systems as the lifetime of many such systems is constrained by their battery capacity. Recent advances in processor design have added support for dynamic frequency/voltage scaling (DVS) for saving energy. Recent work on real-time scheduling focuses on saving energy in static as well as dynamic scheduling environments by exploiting idle time and slack because of early task completion for DVS of subsequent tasks. These scheduling algorithms rely on a priori knowledge of worst-case execution times (WCET) for each task. They assume that DVS has no effect on the worst-case execution cycles (WCEC) of a task and scale the WCET according to the processor frequency. However, for systems with memory hierarchies, the WCEC typically does change under DVS because of requency modulation. Hence, current assumptions used by DVS schemes result in a highly exaggerated WCET. This paper contributes novel techniques for tight and flexible static timing analysis, particularly well-suited for dynamic scheduling schemes. The technical contributions are as follows: (1) We assess the problem of changing execution cycles owing to scaling techniques. (2) We propose a parametric approach toward bounding the WCET statically with respect to the frequency. Using a parametric model, we can capture the effect of changes in frequency on the WCEC and, thus, accurately model the WCET over any frequency range. (3) We discuss design and implementation of the frequency-aware static timing analysis (FAST) tool based on our prior experience with static timing analysis. (4) We demonstrate in experiments that our FAST tool provides safe upper bounds on the WCET, which are tight. The FAST tool allows us to capture the WCET of six benchmarks using equations that overestimate the WCET by less than 1%. FAST equations can also be used to improve existing DVS scheduling schemes to ensure that the effect of frequency scaling on WCET is considered and that the WCET used is not exaggerated. (5) We leverage three DVS scheduling schemes by incorporating FAST into them and by showing that the energy consumption further decreases. (6) We compare experimental results using two different energy models to demonstrate or verify the validity of simulation methods. To the best of our knowledge, this study of DVS effects on timing analysis is unprecedented.

Dynamic Scheduling Using the Mixture of a Genetic Algorithm and a Priority Rule (The Effect of Priority Rule with Consideration to the Due-Date Allowance)

We have proposed a dynamic scheduling method using the mixture of a genetic algorithm and a priority rule. In the method, a dynamic scheduling problem is assumed to be a series of static problems and scheduled on a rolling basis. In this paper, we examine the effect of priority rule used in the method. For this purpose, two types of the ATC rules having two parameters are used. The first type is the rule whose parameters are tuned for the dynamic scheduling environment. The second type is the rule whose parameters are tuned for the imminent static scheduling problem. Numerical experiments show the followings : The proposed scheduling method searches effective schedules guided by the priority rule and the rule has important effect to the scheduling performance. For the dynamic job shop scheduling, the scheduling method using the first type of ATC rule greatly outperforms the one using the second type of ATC rule. The reason can be explained by considering the due-date allowance contained in the optimized schedules.

Concepts for Dynamic Scheduling in the Laboratory

Laboratory environments are controlled more and more by automated systems. Written procedures and lab journals are replaced by workflow description languages and electronic notebooks, which not only describe the processes but are used also for the control of the entire workbench, data acquisition, and documentation. Dynamic scheduling is needed in such an environment. Multiple samples with different procedures are processed in parallel, competing for the instruments. The whole environment may be also underlaid by optimization strategies like throughput or minimum sample processing time. The modeling of all components interacting on the workbench—samples, devices, sensors, results, database systems, and so on—needs to rely on concepts building a consistent framework. This article gives a set of terms and definitions used in a dynamic scheduling environment. It describes most of these entities in detail, including their functionality and attributes as well as their logical and physical interactions. It also describes concepts such as workflows with activities and constraints, functional libraries, hidden transport, and dynamic execution. Maintenance, calibration, and error management also are included. Finally, it discusses how the entities interact with the different components in the scheduling system.

Job shop scheduling techniques in semiconductor manufacturing

This paper presents a brief review on job shop scheduling techniques in semiconductor manufacturing. The manufacturing environment in a semiconductor industry is considered a highly complex job shop, involving multiple types of work centers, large and changing varieties of products, sequence-dependent setup times, reentrant process flow, etc., in a dynamic scheduling environment. Due to the stubborn nature of the deterministic job shop scheduling problem itself, many of the solutions proposed are of hybrid construction cutting across the traditional disciplines. The problem has been investigated from a variety of perspectives resulting in several analytical techniques combining generic as well as problem-specific strategies. In this paper, we seek to provide a brief overview of the problem, the techniques used and the researchers involved in solving this problem.

Adaptive scheduling in random flexible manufacturing systems subject to machine breakdowns

In a FMS dynamic scheduling environment, frequent rescheduling to react to disruptions such as machine breakdowns can make the behaviour of the system hard to predict, and hence reduce the effectiveness of dynamic scheduling. Another approach to handle the disruptions is to update the job ready time and completion time, and machine status on a rolling horizon basis, and consider the machine availability explicitly in generating schedules. When machine downtime has a small variation, the operation completion time is estimated by using limiting (steady-state) machine availability. However, steady-state analysis is sometimes unlikely to provide a complete picture of the system when there is a large variation in machine downtime and repair time, and frequent disruptions (e.g. tool failures) exist. Transient analysis of machine availability will be more meaningful in such a situation during a finite observation period. In this paper, an adaptive scheduling approach is proposed to make coupled decisions about part/machine scheduling and operation/tool assignments on a rolling horizon basis, while the operation completion time is estimated by a transient machine availability analysis based on a two-state continuous time Markov process. The expected tool waiting time is explicitly considered in the job machine scheduling decision. The effectiveness of the proposed method is compared with other approaches based on various dispatching heuristics such as Apparent Tardiness Cost, Cost OVER Time, and Bottleneck Dynamics, etc under different shop load and machine downtime levels.

Dynamic Job Shop Scheduling Using a Neural Network. 1st Report. A Two-Stage Training Method to Minimize Tardiness Cost.

We consider dynamic job shop scheduling problems with a tardiness cost criterion. The most practical scheduling method in dynamic scheduling environments is using dispatching rules. However, there is no dispatching rule that performs best under all scheduling conditions. In this paper, we apply a multi-layer feed-forward neural network to the scheduling, and propose a two-stage training method to minimize the average tardiness cost of jobs under various scheduling conditions. In the first stage, each neural network for its specified scheduling condition is trained separately by a simulated annealing method. In the second stage, a final neural network for all scheduling conditions is trained by a back-propagation algorithm. Simulation results show that the trained neural network outperforms the best dispatching rules in any levels of due-date tightness and job arrival rate.

A computational study on design and performance issues of multi-agent intelligent systems for dynamic scheduling environments

Multi-Agent Intelligent Systems (MAIS) are loosely-coupled network of problem solving systems that, whenever needed, work together with each other to dynamically solve problems that none of the system can individually solve. Among the advantages of the MAIS, when compared to the centralized systems, are increased reliability, faster problem solving, decreased communication, and more flexibility. Learning to coordinate the actions is one of the most important task in MAIS. In the current research, we use a widely reported dynamic job shop scheduling simulation model that uses distributed genetic learning of job scheduling strategies (Pendharkar, P.C., 1997. Doctoral Dissertation, Graduate School, Southern Illinois University at Carbondale; Pendharkar, P.C., 1998. Distributed learning of objectives for adaptive scheduling (in review); Pendharkar, P.C., Bhattacharyya, S., 1997. Multi-agent learning in distributed artificial intelligence. Proc. 2nd INFORMS Conference on Information Systems and Technology. San Diego, CA, p.156–163; Bhattacharyya, S., Koehler, G.J., 1997. Learning by objectives for adaptive shop-floor learning. Decision Sciences (to appear). Aytug, H., Koehler, G.J., Snowdon, J.L., 1994. Genetic learning of dynamic scheduling within a simulation environment, Computers and Operations Research, 21 (8), 909–925; Aytug, H., Bhattacharyya, S., Koehler, G.J., Snowdon, J.L., 1994. A review of machine learning in scheduling, IEEE Transactions on Engineering Management 41 (2) ) and study the performance and design issues in multi-agents information systems for dynamic scheduling in manufacturing. Among the design issue and performance issues considered in this research are coordination between agents, number of agents, and frequency of learning. Our results indicate that coordination between agents, and learning frequency play a significant role in the performance of multi-agent intelligent systems.