Basic concept of multi-agent system (1)

xiaoxiao2021-03-06  24

Software agent definition:

A software entity that can be continuously, spontaneous, and is a software entity associated with the relevant agent and processes.

It features:

Autonomous: Most of the functions can be completed without manual intervention.

Social capabilities: Can interact with other agents and people.

Passive response capability: can sense changes in the environment to generate real-time responses

Active response capability: Active execution gives a breeding belief and goals.

Time continuity: The process is continuous.

Adaptive: Adapt to the environmental changes by learning mechanisms.

Mobility: It can carry your own context to migrate to other environments.

The three characteristics, interactivity, intelligence, and mobility of the software agent. These three dimensions constitute a type of space. The software agent can obtain its location in space by evaluating the above three factors.

Theoretical knowledge required by the agency system

An agent must be intelligent, and you need to reason to reason for your knowledge, so you need to express and reasoning mechanisms for knowledge.

In the multi-agent system, knowledge is local, and the problem is global. In most cases, it is necessary to solve a problem with other agents, so that the information between the agent is inevitable, so there is a proxy communication language (ACL) KQML is a relatively perfect ACL, 3 levels: content, communication and message.

A coordinated model is required to handle activities such as creation, deletion, and communications. The purpose of coordination model is to provide a situation of formalized frames, including 3 elements: coordinates, coordinating media, and coordination rules.

The social evolution of multi-agent systems is a multi-agent consultation model. Its requires agency must have three functions: 1 Express yourself 2: Specify others 3: Impact others.

转载请注明原文地址:https://www.9cbs.com/read-41211.html

New Post(0)