文档下载 文档下载导航
设为首页 | 加入收藏
搜索 请输入内容:  
 导航当前位置: 文档下载 > 所有分类 > V. Agent-Based Framework for Simulation and Support of Dynamic Engineering
侵权投诉

V. Agent-Based Framework for Simulation and Support of Dynamic Engineering

Abstract. The paper reports on the first results of the Productivity Simulation Initiative (PSI) project of Cadence Design Systems GmbH. The project addresses the problem of fine-grained modeling and simulation of dynamic engineering design processes in or

the design objective, sudden unavailability of the team members, the changes in the workload of the designers according to the influence of the other projects, etc.

Based on these usage modes PSI testbed comprises the following two parts. The first part (the initial testbed) contains the detailed records of 1 – 2 ongoing design projects to extract the knowledge of the acting humans and log every event during the execution of the project as accurately as possible. DEDP model and the adequacy of the implemented interaction mechanisms are evaluated based on these “logs of DEDP execution”. As the result the corrective factors are extracted to improve the quality of the simulation. The performance of the process can be accessed based on this improved simulation of a design process.

The second part of the testbed is under creation and covers the prediction capability. A set of 3 to 5 design artifacts will be used to create detailed project plans by experienced project managers (2-3 per artifact). As the part of this planning process all decisions and their reasons will be recorded to extract the know-how of the project manager.

Initial set of experiments has been performed on the two simplified scenarios: the process of the design of a digital multimedia encoder [15] and the process of the design of an analog controlled amplifier [23]. These processes have been described according to the data collected by lead designers of Cadence Design Systems GmbH in their previous design projects. Execution logs have been created for the respective DEDPs through filling in the DEDP questionnaires [16]. These logs formed the initial testbed for DEDP-MAS prototype.

The scenarios were simplified to keep the complexity at a low to medium level. For example, the digital scenario is characterized by:

– 5 designers

– Design artifact comprising 4 functional blocks

– The process from RTL design up to tape-out in GDSII format resulted in 36 activities.

5 Experiments for Framework Prototype Checking

Experiments with the DEDP-MAS prototype are performed in frame of the real world project of low complexity and amount of work under some simplifications. It was assumed that the execution log does not exist at the beginning of the process, but is gradually developed in line with the DEDP flow. The objective of the experiments was to develop the so called prediction-correction methodology. The methodology should predict the development of the process up to the next checkpoint through the simulation based on the initial task and existing agents’ mental models. General view of the experiments scenario is outlined in Fig. 3.

DEDP-MAS prototype configuration (the set of software agents with the accordingly prepared mental models) corresponds to the project design team. It is assumed in the experiments that only one agent of the Designer class performing the Task manager role in DEDP-MAS prototype assists the human Project leader and the rest of the agents of this class simulate the activities of the other design team members.

上一页第6页

你可能喜欢

热门文档

相关文档

站点地图 | 文档上传 | 侵权投诉 | 手机版
本站所有资源均来自互联网,本站只负责收集和整理,均不承担任何法律责任,如有侵权等其它行为请联系我们.
文档下载 Copyright 2013 doc.guandang.net All Rights Reserved.  -  email