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.

Constructing Bayesian Network in a Changing World.

Adaptive Learning of Dynamic Bayesian Networks with Changing Structures by Detecting Geometric Structures of Time Series

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

Advances in Bayesian network modelling: Integration of modelling technologies

Research On Modeling The Nonhomogenous Markov Decision Systems With Dynamic Bayesian Network

Bayesian Network for Dynamic Variable Structure Learning and Transfer Modeling of Probabilistic Soft Sensor

Hierarchical Bayesian Network Modeling Framework for Large-Scale Process Monitoring and Decision Making.

Integrating Model Construction and Evaluation

Dynamic Bayesian Network for Robust Latent Variable Modeling and Fault Classification

Adaptive user support in educational environments: A Bayesian Network approach

From knowledge bases to decision models

A Hybrid Approach for Identifying the Structure of a Bayesian Network Model

Using Dynamic Bayesian Networks to Model User-Experience

Dynamic Network Updating Techniques For Diagnostic Reasoning

Using Bayesian Network to Develop an Approach for Construction Safety Risk Assessment

Bayesian modeling of networks in complex business intelligence problems

The Bounded Bayesian

A Bayesian network model integrating organizational, individual and psychological factors for strengthening construction worker safety behavior

Bayesian dynamic financial networks with time-varying predictors

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

Environmental Assessment Model of Green Ecological Residential Buildings Based on Bayesian Network