Abstract:In the context of modern engineering, environmental, and societal concerns, there is an increasing demand for methods able to identify rational management strategies for civil engineering systems, minimizing structural failure risks while optimally planning inspection and maintenance (I&M) processes. Most available methods simplify the I&M decision problem to the component level, often assuming statistical, structural, or cost independence among components, due to the computational complexity associated with global optimization methodologies under joint system-level state descriptions. In this paper, we propose an efficient algorithmic framework for inference and decision-making under uncertainty for engineering systems exposed to deteriorating environments, providing optimal management strategies directly at the system level. In our approach, the decision problem is formulated as a factored partially observable Markov decision process, whose dynamics are encoded in Bayesian network conditional structures. The methodology can handle environments under equal or general, unequal deterioration correlations among components, through Gaussian hierarchical structures and dynamic Bayesian networks, decoupling the originally joint system state space to component networks conditional on shared random variables. In terms of policy optimization, we adopt a deep decentralized multi-agent actor-critic (DDMAC) reinforcement learning approach, in which the policies are approximated by actor neural networks guided by a critic network. By including deterioration dependence in the simulated environment, and by formulating the cost model at the system level, DDMAC policies intrinsically consider the underlying system-effects. This is demonstrated through numerical experiments conducted for both a 9-out-of-10 system and a steel frame under fatigue deterioration. Results demonstrate that DDMAC policies offer substantial benefits when compared to state-of-the-art heuristic approaches. The inherent consideration of system-effects by DDMAC strategies is also interpreted based on the learned policies.

A comparison of one-stage and two-stage Bayesian processes in engineering decision-making

Variational Bayesian State Space Model for Dynamic Process Fault Detection

A Mixed Bayesian Network for Two-Dimensional Decision Modeling of Departure Time and Mode Choice

Inference and dynamic decision-making for deteriorating systems with probabilistic dependencies through Bayesian networks and deep reinforcement learning

Two-stage Approach to Causality Analysis-Based Quality Problem Solving for Discrete Manufacturing Systems

Hybrid Machine Learning Modeling of Engineering Systems -- A Probabilistic Perspective Tested on a Multiphase Flow Modeling Case Study

Bayesian design of experiments for generalised linear models and dimensional analysis with industrial and scientific application

A two-level Bayesian early fault detection for mechanical equipment subject to dependent failure modes

A variational Bayesian neural network for structural health monitoring and cost-informed decision-making in miter gates

Human-algorithm collaborative Bayesian optimization for engineering systems

Bayesian two-stage structural identification with equivalent formulation and EM algorithm

Bayesian Optimization for Non-Convex Two-Stage Stochastic Optimization Problems

Two-dimensional Bayesian monitoring method for nonlinear multimode processes

Online probabilistic operational safety assessment of multi-mode engineering systems using Bayesian methods

A comprehensive review of Bayesian statistics in natural hazards engineering

Bayesian Variable Selection in Double Generalized Linear Tweedie Spatial Process Models

Bayesian Benefit-Risk Analysis for Sustainable Process Design

A new perspective on Bayesian Operational Modal Analysis

Bayesian sequential update for monitoring and control of high-dimensional processes

A comparison of mixed-variables Bayesian optimization approaches