Laurence R. Phillips

Agent-Based Control of Distributed Infrastructure Resources

Managing distributed infrastructure resources is usually accomplished by telephone calls among the managers and operators. This works reasonably well under ordinary circumstances but breaks down—often catastrophically—under stress. Individual motivation, long response times, and poor situation awareness interfere with operation and can even cause breakdowns. Broadly distributed operations would more robust and fail more gracefully than centralized systems, but remain unlikely given the difficulty in operating large infrastructures even with modern Supervisory Control And Data Acquisition (SCADA) systems. The solution is management by distributed software that maintains normal operation, enforces operational and security policy, deals with contingencies, and protects against malicious indsiders, errors, and outright attacks. We specify a distributed agent coalition able to accomplish this for distributed electric power and describe a prototype implementation based on Sandia-developed technology.

A Multi-agent System for Coordinating International Shipping

Moving commercial cargo across the US-Mexico border is currently a complex, paper-based, error-prone process that incurs expensive inspections and delays at several ports of entry in the Southwestern US. Improved information handling will dramatically reduce border dwell time, variation in delivery time, and inventories, and will give better control of the shipment process. The Border Trade Facilitation System (BTFS) is an agent-based collaborative work environment that assists geographically distributed commercial and government users with transshipment of goods across the US-Mexico border. Software agents mediate the creation, validation and secure sharing of shipment information and regulatory documentation over the Internet, using the World-Wide Web to interface with human actors. Agents are organized into Agencies. Each agency represents a commercial or government agency. Agents perform four specific functions on behalf of their user organizations: (I) agents with domain knowledge elicit commercial and regulatory information from human specialists through forms presented via web browsers; (2) agents mediate information from forms with diverse ontologies, copying invariant data from one form to another thereby eliminating the need for duplicate data entry; (3) cohorts of distributed agents coordinate the work flow among the various information providers and they monitor overall progress of the documentation and the location of the shipment to ensure that all regulatory requirements are met prior to arrival at the border; (4) agents provide status information to human actors and attempt to influence them when problems are predicted.

The Role of Conversation Policy in Carrying Out Agent Conversations

Structured conversation diagrams, or conversation specifications, allow agents to have predictable interactions and achieve predefined information-based goals, but they lack the flexibility needed to function robustly in an unpredictable environment. We propose a mechanism that dynamically combines conversation structures with separately established policies to generate conversations. Policies establish limitations, constraints, and requirements external to specific planned interaction and can be applied to broad sets of activity. Combining a separate policy with a conversation specification simplifies the specification of conversations and allows contextual issues to be dealt with more straightforwardly during agent communication. By following the conversation specification when possible and deferring to the policy in exceptional circumstances, an agent can function predictably under normal situations and still act rationally in abnormal situations. Different conversation policies applied to a given conversation specification can change the nature of the interaction without changing the specification.

Agents and islands: managing a power system before, during, and after transition to the islanded state

This paper presents a fictional scenario describing the effects of a category four hurricane on a metropolitan area, accompanied by a challenge: describe, and eventually realize, a system able to carry out the necessary power system operations without human participation. We outline the capabilities of an automated system for managing electric power. The overarching task of the automated system is islanding: To separate the metro areas power system from the primary power grid and manage its operation during several hurricane-induced contingencies, with the power system operational throughout. The essential technology needed to support this automation is agents. We address the roles agents play in the transition to the islanded state and in power system operation within the island; features of an appropriate agent substrate; the way the agents are organized; and information exchange among agents, the power system, and human operators

Software Requirements Specification for Information Management for Grid Control

This Software Requirements Specification defines the functions of a simulation power grid model. The model defines grid control functions that focus on real-time control and related communication of information among entities that share the operation of the power grid. Deregulation of the power markets necessitates increased communications among entities who have economic motivation to restrict access to important information from other market participants. New power market concepts will impact how planning and real-time control are performed. The simulation power grid model provides the tool for investigating issues of distributed computing, data sharing, data access, communication system capacity, and communications reliability. The model enables researchers to develop intelligent distributed control agents for managing Area Control Error and transmission security. The software requirements specification is defined using a subset of the Unified Modeling Language, a class diagram describing all of the objects along with their attributes, methods, and some modified use cases. The work described in this report was coordinated by the Consortium for Electric Reliability Technology Solutions on behalf of the Assistant Secretary of Energy Efficiency and Renewable Energy, Offce of Power Technologies of the U S . Department ofEnergy under Contract No. DE-ACO4-94AL85000.

Managing Microgrids Using Grid Services

In order for electric power generating capacity to be supplanted to a meaningful extent by sources smaller than 200 kW, an automated means of managing systems of small sources must be found or their sheer number will over-whelm the power production community. Micro grids - power systems comprising multiple small interconnected generators - are a promising response to this need, but an automated micro grid management system has not been demonstrated. This paper describes the energy management task and its execution in a standardized grid services environment.

Algorithmic approaches to reducing risk in power system operation

This paper identifies and describes algorithms of interest for reducing risk in power system operation. Risk is the probability of an event multiplied by the cost of the events consequences. For a power system, the primary consequence is failure of the system to meet demand. Risk reduction is examined from three perspectives: protection, reduced consequences, and improved operation. In particular we examine distributed algorithms from a risk perspective. This work provides direction for algorithm R&D in the area of risk reduction and indicates underlying principles of algorithm utilization for distributed system operation. Our prior work identified a framework for distributed operation and described the framework tasks, but did not identify suitable algorithms, properties, and metrics. This paper addresses this need by pointing out algorithms of particular interest for power system operation and relating these to risk. Unifying principles for distributed power system operation are indicated. Future work will demonstrate functioning instantiations of the most relevant algorithms in a power system context.

Knowledge Discovery and Data Mining (KDDM) survey report.

The large number of government and industry activities supporting the Unit of Action (UA), with attendant documents, reports and briefings, can overwhelm decision-makers with an overabundance of information that hampers the ability to make quick decisions often resulting in a form of gridlock. In particular, the large and rapidly increasing amounts of data and data formats stored on UA Advanced Collaborative Environment (ACE) servers has led to the realization that it has become impractical and even impossible to perform manual analysis leading to timely decisions. UA Program Management (PM UA) has recognized the need to implement a Decision Support System (DSS) on UA ACE. The objective of this document is to research the commercial Knowledge Discovery and Data Mining (KDDM) market and publish the results in a survey. Furthermore, a ranking mechanism based on UA ACE-specific criteria has been developed and applied to a representative set of commercially available KDDM solutions. In addition, an overview of four R&D areas identified as critical to the implementation of DSS on ACE is provided. Finally, a comprehensive database containing detailed information on surveyed KDDM tools has been developed and is available upon customer request.

Agent-Based Mediation and Cooperative Information Systems

This report describes the results of research and development in the area of communication among disparate species of software agents. The two primary elements of the work are the formation of ontologies for use by software agents and the means by which software agents are instructed to carry out complex tasks that require interaction with other agents. This work was grounded in the areas of commercial transport and cybersecurity.