Application of Neural Networks for the Safety and Performance Evaluations of Infrastructure Systems

9402049周神经网络即使在输入不完整、有噪声或以前看不到的情况下也能做出合理的响应(输出)的能力在工程应用中具有重大影响。很多时候，由于自然或环境的限制，例如在自然灾害之后，并不是所有形式或类型的数据都被收集用于安全分析。在这种情况下，传统的可靠性分析可能会中断、终止或严重受损。此外，神经网络接受输入和响应之间的关系是确定性的、概率的或模糊的。女性科学家和工程师职业发展奖为研究人员提供了一个机会，利用概率、统计和模糊集理论将她目前的结构安全研究领域扩展到土木工程，特别是结构工程。神经网络是一种人工智能工具，已被证明比专家系统更灵活。从其他工程学科的研究来看，它们显示出这个概念也适用于土木工程问题的巨大潜力。***

9402049 Chou The neural network's capability of making reasonable response (output) even when it was presented with incomplete, noisy or previously unseen input has significant impact in engineering applications. Many times, due to physical or environmental limitations such as after a natural disaster, not all forms or types of data are collected for safety analysis. In this situation, tradition reliability analysis could be halted, terminated or greatly compromised. In addition, neural network accepts relationships between input and response that are deterministic, probabilistic or fuzzy. The Career Advancement Award for Women Scientists and Engineers provides the Investigator an opportunity to expand her current research area of structural safety utilizing probabilistic, statistical and fuzzy set theory to civil engineering, particularly structural engineering. Neural network is an artificial intelligence tool that has shown to be more flexible than expert systems. From studies performed in other engineering disciplines, they show great potential that the concept is applicable to civil engineering problems as well. ***