Smart Inference Engines for Model Validation and Updating: Advances in Service Life Prediction and Cost-Effective Decisions
Abstract
Extensive research during the past two decades has been devoted to developing risk-based management and service life extension methods for critical structural systems using physics-based and data-driven system response models, supported by data obtained from structural health monitoring (SHM) systems. State-of-the-art decision-making approaches for risk-based structural integrity management demonstrate when to inspect and repair structural systems. However, when and how to update the numerical models under random condition changes (e.g., climate change, aging, and post-event system deterioration) have yet to be examined in the recent studies. Addressing the high computational cost of the updating process, this project will develop a costeffective methodology to validate and update the numerical models for a complex system against sporadic events. The project tasks will first to build a high-fidelity physics-based model based on data-driven model validation approaches (Task 1); build a data-driven response model (Task 2) through machine learning approaches to plan on risk-based maintenance strategies at the modeling phase and determine the need for in-service actions during and after an event; and perform hypothesis tests on post-event changes in the structural performance to determine the need for model updating (Task 3). If successful, the research will produce smart pre- and post-event inference engines to implement into a smart risk-based decision-making framework that will predict the remaining service life and cost-effective maintenance strategies for naval systems by implementing distinct physical features of a ship~s structure. The inference engines will be a key driver to create a critical civil infrastructure management paradigm shift to IoT (Internet of Things)-based, real-time automated systems, which can improve the operational integrity of unmanned naval systems.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Feb 17, 2020
- Source ID
- N000142012055
Entities
People
- Yeongae Heo
Organizations
- Case Western Reserve University
- Office of Naval Research
- United States Navy