Dilip B. Kotak

Agent-based collaborative project management system for distributed manufacturing

In order to address the dynamic requirements of project-based productions, this paper proposes a collaborative project scheduling and management system framework using a distributed multi-agent approach. This framework consists of three types of agents: L-Agents to enable integration with MRP/ERP and other legacy systems; P-Agents to schedule and manage the projects; and D-Agents to schedule, monitor and coordinate actual productions. These generic agents are dynamically deployed and customized at each location where a portion of the project is executed. An overall coordination is achieved through the communication and negotiation among these distributed agents. The proposed system enables project managers and others to effectively accommodate frequent engineering changes and other uncertainties while making an effective use of available distributed resources and fulfilling the customer requirements. A company manufacturing tools for sheet metal fabrications is used as an example to test and validate our prototype system. The prototyped system architecture is well suited for distributed problem solving applications and is adaptable to other application domains.

GRIDBSCAN: GRId Density-Based Spatial Clustering of Applications with Noise

Clustering is one of the basic data mining tasks that can be used to extract hidden information from data in the absence of target classes. One of the most well-known density based clustering algorithms for processing spatial data is Density-Based Spatial Clustering of Application with Noise (DBSCAN) that uses learning parameters epsiv and minPts to define the density that will be sought in the data set while forming the clusters. The major drawbacks of the DBSCAN algorithm are its sensitivity to user input required to execute the algorithm, inability to recognize clusters with different densities, and computational complexity. In this study, we propose a three-level clustering method to address the second issue. The first level selects appropriate grids so that the density is homogeneous in each grid. The second stage merges cells with similar densities and identifies the most suitable values of epsiv and minPts in each grid that remain after merging. The third step of the proposed method executes the DBSCAN method with these identified parameters in the dataset. The proposed method is tested in three artificial benchmark data sets to demonstrate that the clusters are correctly identified.

Holonic manufacturing system for distributed control of automated guided vehicles

Automated guided vehicle (AGV) systems traditionally utilize a central controller to assign transportation tasks among vehicles. Because of its rigidity, a centralized architecture may not suitable for modern manufacturing with its rapidly changing markets, demands for products with higher quality, smaller batch sizes, customer specifications, and shorter delivery times. This research applies concepts from holonic manufacturing systems to develop a distributed AGV control system that is flexible and robust in the presence of disturbances. We present the architecture of the holonic manufacturing system, a coordination strategy based on contract nets, and simulation results that indicate an improvement in utilization when compared with traditional centralized control.

Distributed coordination of micro-grids using bilateral contracts

More than 1 billion people worldwide living in rural communities without access to electricity could benefit from distributed generation. A major difficulty with any distributed generation solution, however, is coordinating the distributed energy resources to meet demands. Distributed control techniques based on resource allocation can be used to solve this problem. One such technique utilizes bilateral contracts between loads and sources, which can be facilitated by auctions. Four major auction types - first-price sealed bid, Vickrey, English and Dutch - can be used. Simulations that mimic distributed control have been conducted in order to compare the four auction types. The simulations show which auction types should be used under different conditions.

Operational scheduling for rough mills using a virtual manufacturing environment

Production scheduling in wood product plants, such as furniture factories, is highly complex due to its sensitivity to defects occurring in wood. Due to the huge number of choices and uncertainty of the outcome an optimum schedule cannot be computed mathematically. These choices, with complex interdependencies, have to be made on an hourly basis at different machines and must be coordinated to achieve the overall production,objectives. A decision support tool, using a 3D simulation model, is provided to predict the production performance of the null under various operational schedules, before a selected schedule is implemented. A 3D simulation model was developed accurately depicting the layout and operation of a rough mill. A historical database, some of it maintained via human input and some of it collected automatically, was analyzed to provide the basis for predicting the behavior of the mill when processing different type of raw materials and when producing different combination of products. The validated model, which links to the scheduling database (i.e., raw material available, products required) is used to test-fly alternate production schedules as often as required, thus identifying potential production problems before they occur in the mill. The next step will be to incorporate intelligent scheduling agents using techniques, such as fuzzy logic and genetic algorithms, to provide scheduling recommendations on individual machines. In the future these will be coordinated in a holonic manufacturing execution system to achieve better global performance. Performance measurement will require assessment based upon multiple criteria (e.g., productivity, recovery, grade utilization, and satisfaction of order files).

Holonic system architecture for scalable infrastructures

In a previous paper Gruver, Kotak, van Leeuwen and Norrie proposed a Holonic Systems Architecture for manufacturing co-ordination. In another project Daimler-Chrysler demonstrated that a holonic approach resulted in significant improvements in productivity and robustness of their engine assembly line design and the technique was found to be scalable and could be implemented in a stepwise fashion. Presently NRC is examining the ways of evolving a hydrogen infrastructure for fuelling hydrogen in fuel cell vehicles. Such an infrastructure will need to be scalable, reliable and cost competitive. Therefore we chose to use the holonic approach for modelling and simulation of such an infrastructure, which presently does not exist and will initially be sparse and highly distributed. This paper describes the original design features of the holonic system architecture, the hydrogen infrastructure requirements and the enhancements we made to the architecture in the process of implementing the test application. We used a 3-D simulation technology to create a virtual hydrogen infrastructure, created holons to proxy its elements (i.e., hydrogen generation, storage and dispensing stations, vehicles and highways with alternate routes) and the interface between the virtual and holonic worlds. We systematically introduce holons which proxy real physical devices, to demonstrate how this architecture could be phased in from a virtual model to the physical world without changing the holonic control.

Rough mill component scheduling: heuristic search versus genetic algorithms

Rough mill production systems cut lumber into smaller components needed to produce wood products. Because of the systems limited sorting capacity, rough mill operators need to schedule when different component sizes are made, a process called part scheduling and replacement. This scheduling process is significant because it greatly affects system performance. Three component scheduling algorithms are examined in this paper: a heuristic method that mimics how human operators manually schedule components; and two methods based on genetic algorithms, the simple genetic algorithm and the ordering messy genetic algorithm. The performance of the algorithms is analyzed and tested on four cutting bills. Results show that the ordering messy genetic algorithm outperformed the simple genetic algorithm, and heuristic component replacement performed better than replacement based on the genetic algorithms objective function. Also, heuristic cut-list selection performed better on cutting bills with more short pieces, whereas GA cut-list selection performed better on bills it with longer pieces.

Coordination of distributed energy resources

A major problem with any solution for distributed generation (DG) is the coordination of distributed energy resources (DER) to meet demand. Taking a new approach to the problem, we consider off-grid implementations that would have the greatest impact in remote and developing regions. Our proposed system creates a micro-grid by linking together a group of individual communities for backup and peak-shaving. As such, the DER must be coordinated (and scheduled) to meet the total demand of the micro-grid. Two distributed coordination strategies are currently being examined - bilateral contracts and market-based techniques. Employing fuzzy costs allows the system to be optimized for a range of values. Finally, we describe research in progress to develop a holonic system model of a distributed testbed and implement it using the JADE development platform to evaluate coordination strategies under different micro-grid configurations.

Comparison of hierarchical and holonic shop floor control using a virtual manufacturing environment

Shop floor control is typically implemented with a centralized supervisory control system. Holonic control, a form of distributed control, provides a more robust control environment and improves production performance. The authors create a 3D virtual manufacturing environment to compare hierarchical control with holonic control. We use this virtual environment to analyze alternative approaches and quantify their effects on the system behavior in a controlled environment. Finally, we present a measure of robustness derived from reliability theory and utilize it for comparisons.

A distributed decision support system for lumber jag selection in a rough mill

A rough mill production system is an unpredictable dynamic system for which the effects on production of many factors are strongly interrelated. A distributed decision support system for dynamic jag (a load of lumber) selection in defect sensitive production is proposed. Case-Based Reasoning and heuristic rules determine recommended jags for cut-lists in the decision process. Using a statistical method, cases are grouped by categorizing problems and solutions. Heuristic rules measure the similarity between current and past cases. The framework for a decision support system has two layers in which jag types are selected at the top level and specific jags are chosen on the bottom level. Decisions from different sources influence both top level and bottom level. The dynamic character of the system is taken into account by collaboration between distributed decision points that can change indices by retrieving previous cases and storing new cases. The recommended jags are tested by simulation and used for feedback to the decision support system. Finally, the system is validated by comparison with actual production.