Artificial intelligence design framework for optical backplane engineering.

摘要

The core contribution of this work is a new framework for architecture-driven software engineering of large-scale, agent-based, complex systems for Knowledge-Based Engineering (KBE). This reconfigurable and scalable framework fills a niche between traditional requirements elicitation and design tool implementation. The new architectural framework model, referred to as the Artificial Intelligence Design Framework (AIDF), allows KBE system architects to achieve intellectual control over the high-level development process. Therefore, structural modelers using object-oriented languages can specify the coding requirements based on hierarchical decomposition of functional requirements using axiomatic design principles. The engine block defined by the framework has the capability to utilize networked knowledge repositories available through intelligent agents acting on Web Services for the purpose of design risk mitigation for reliability engineering.;KBE systems produced based on the new architecture framework automates the design and inference processes for reliability engineering using an interlaced dual engine block, developed during the National Aeronautics and Space Administration (NASA) Fellowship. This type of intelligent automation of design support for product engineering can save on development cost and time, while improving on quality. A comprehensive solution is proposed to address this risk mitigation need for reliability engineering using a System-of-Systems (SoS) approach, which consists of a synergistic overlap of many broad topics such as design, agent modeling, and systems engineering. A case study implementation of the AIDF is developed using Acclaro Design for Six Sigma (DFSS) architectural development tool for configuring an optical backplane engineering application with design matrix, optimization, and verification techniques.;The Generic Architecture for Upgradeable Real-time Dependable Systems validation framework is introduced as a validation strategy for post-deployment expansion, after applying DFSS front-end validation during pre-deployment development. In addition to architecture validation, a comprehensive validation approach for a KBE SoS applications using the Synergistic Validation Methodology (SVM) for the AIDF has been developed. In conclusion, an AIDF-SVM is introduced as an architectural framework with a recommended validation methodology that functions as a platform for developing large-scale, reconfigurable and scalable KBE SoS applications.