9 th International Conference on DEVELOPMENT AND APPLICATION SYSTEMS, Suceava, Romania, May 22-24, 2008 16 Abstract—Cooperation become more and more popular in software industry and the emergent agent technology was also applied to industrial systems. The paper attempts to summarize some convergent developments in power system area that exploit agent technology and present a simple simulation based on Matlab and Java Agent DEvelopment Kit. Index Terms—Power Systems Control, Multi-Agent Systems I. INTRODUCTION Power networks are operated by thousands of devices following simple rules with local information. Some of these control devices are already pre-programmed for anticipated situations, but the liberalization of electricity markets or new trends (such as distributed generation, renewable energy) increase interconnectivity between the components and the centralized real-time control becomes more difficult. As networking and embedding technologies advance, it is more possible to design autonomous entities for such process automation functions where the tasks require cooperative distributed problem solving. Agents can improve the control devices, such as relays, voltage regulators or Flexible AC Transmission Systems (FACTS) devices. The connection between software entities and automation subsystems are fixed (generally defined at design-time), but the systems should deal with unanticipated requests. The models for the architecture of future electricity systems consider new concepts: primary the entire system (transmission and distribution) can be designed and used as an integrated unit. The network must provide connectivity (multiple links) between points of power supply and demand. The network must interact permanently with the customers who can optimize his options and should pay for required services. II. MULTI-AGENT SYSTEM TECHNOLOGY An agent is an abstraction object (software or hardware) capable of autonomous action in some environment; in other words “is anything that can be viewed as perceiving its environment through sensors and acting upon that environment through effectors,” [25]. The agent concept is similar with software object, but Object-Oriented Programming (OOP) defines the object in terms of methods and attributes, an agent is defined in terms of behavior and ontology. Behaviors implement the tasks (or intentions) of an agent. They are logical activity units that can be composed in various ways to achieve complex execution patterns and that can be concurrently executed. The ontology indicates the vocabulary of the symbols used in the messages content. The agents involves in a communication must ascribe the same meaning to these symbols for the communication to be effective. A stronger notion of agent include mental properties [31], like knowledge, belief or intention; additionally maybe consider mobility- agents can move from one host to another, veracity- agents do not knowingly communicate false information, benevolence- agents always try to do what they are asked of, rationality- agents will try to achieve their goals. In Multi-Agent System (MAS) tasks are deployed by interacting agents that can cooperate with each other. Some characteristics generally adopted: “each agent has incomplete information or capabilities for solving the problem and, thus, has a limited viewpoint; there is no system global control; data are decentralized; and computation is asynchronous” [27]. MAS can be considered "self-organized systems" as they tend to find the best solution for their problems without external intervention. Some challenges in developing MAS are task decomposition, defining the agent behavioral rules, agent coordination, setting the environment where agents live. III. MAS IN POWER SYSTEM The industrial processes are susceptible to power quality problems outside the scope of conventional interruptions. The information is critical in detecting, resolving and preventing power quality problems and is essential for evaluating total system performance. A power-quality agent can read the measured data from the process, compute power quality indexes, and shared the processed data with other agents. A distributed system is more available then one centralized to detect incidents like transients (frequency and amplitude changes in voltage and/or current waveforms), sags and swells, flicker, noise, harmonics or frequency variations. The actually electric networks are operated local and distributed by heterogeneous agents that can range from simple devices (like relays) to intelligent entities (like human operators). We can say that each agent can sense only a few of the network’s state variables and influence only a few of its control variables. Several researches have been made using agents in power system applications, the typical areas referring to restoration and expansion planning, cascading failures, optimal power flow, secondary voltage control, oscillation damping control, adaptive relaying, fault diagnosis [32], power market or Strategic Power Infrastructure Defense [16]. Similar to classic hierarchical operations, in [1] [11] the agents form layers: the reactive layer consists of agents that Agent-Based Systems in Power System Control Cristinel COSTEA 1 , Adrian PETROVAN 2 North University of Baia Mare str.V.Babes nr.62A, RO-430083 Baia Mare 1 ccostea@ubm.ro , 2 adrian.petrovan@ubm.ro