Reinforcement Learning for Maintenance Decision-Making of Multi-State Component Systems with Imperfect Maintenance

2021 ◽  
Author(s):  
Van-Thai Nguyen ◽  
Phuc Do ◽  
Alexandre Voisin ◽  
Benoit Iung
Keyword(s):  
Decision Making ◽  
Reinforcement Learning ◽  
Imperfect Maintenance ◽  
Component Systems ◽  
Maintenance Decision ◽  
State Component

scholarly journals Bayesian maintenance decision optimisation based on computing the information value from condition inspections

2021 ◽  
pp. 1748006X2097812
Author(s):  
Guang Zou ◽  
Kian Banisoleiman ◽  
Arturo González
Keyword(s):  
Decision Making ◽  
Structural Integrity ◽  
Fatigue Cracking ◽  
Cost Effective ◽  
Offshore Structures ◽  
Information Value ◽  
Bayesian Decision ◽  
Maintenance Strategies ◽  
Optimal Maintenance ◽  
Maintenance Decision

A challenge in marine and offshore engineering is structural integrity management (SIM) of assets such as ships, offshore structures, mooring systems, etc. Due to harsh marine environments, fatigue cracking and corrosion present persistent threats to structural integrity. SIM for such assets is complicated because of a very large number of rewelded plates and joints, for which condition inspections and maintenance are difficult and expensive tasks. Marine SIM needs to take into account uncertainty in material properties, loading characteristics, fatigue models, detection capacities of inspection methods, etc. Optimising inspection and maintenance strategies under uncertainty is therefore vital for effective SIM and cost reductions. This paper proposes a value of information (VoI) computation and Bayesian decision optimisation (BDO) approach to optimal maintenance planning of typical fatigue-prone structural systems under uncertainty. It is shown that the approach can yield optimal maintenance strategies reliably in various maintenance decision making problems or contexts, which are characterized by different cost ratios. It is also shown that there are decision making contexts where inspection information doesn’t add value, and condition based maintenance (CBM) is not cost-effective. The CBM strategy is optimal only in the decision making contexts where VoI > 0. The proposed approach overcomes the limitation of CBM strategy and highlights the importance of VoI computation (to confirm VoI > 0) before adopting inspections and CBM.

scholarly journals Individual differences in experienced and observational decision-making illuminate interactions between reinforcement learning and declarative memory

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Batel Yifrah ◽  
Ayelet Ramaty ◽  
Genela Morris ◽  
Avi Mendelsohn
Keyword(s):  
Decision Making ◽  
Reinforcement Learning ◽  
Declarative Memory ◽  
Contextual Information ◽  
Memory Performance ◽  
Relevant Information ◽  
Subjective Memory ◽  
Types Of Information ◽  
Reinforcement Learning Models ◽  
Implicit And Explicit

AbstractDecision making can be shaped both by trial-and-error experiences and by memory of unique contextual information. Moreover, these types of information can be acquired either by means of active experience or by observing others behave in similar situations. The interactions between reinforcement learning parameters that inform decision updating and memory formation of declarative information in experienced and observational learning settings are, however, unknown. In the current study, participants took part in a probabilistic decision-making task involving situations that either yielded similar outcomes to those of an observed player or opposed them. By fitting alternative reinforcement learning models to each subject, we discerned participants who learned similarly from experience and observation from those who assigned different weights to learning signals from these two sources. Participants who assigned different weights to their own experience versus those of others displayed enhanced memory performance as well as subjective memory strength for episodes involving significant reward prospects. Conversely, memory performance of participants who did not prioritize their own experience over others did not seem to be influenced by reinforcement learning parameters. These findings demonstrate that interactions between implicit and explicit learning systems depend on the means by which individuals weigh relevant information conveyed via experience and observation.

scholarly journals STANDARDISING MAINTENANCE JOBS TO IMPROVE GROUPING DECISION MAKING

2021 ◽  
Vol 1 ◽  
pp. 2701-2710
Author(s):  
Julie Krogh Agergaard ◽  
Kristoffer Vandrup Sigsgaard ◽  
Niels Henrik Mortensen ◽  
Jingrui Ge ◽  
Kasper Barslund Hansen ◽  
...  
Keyword(s):  
Decision Making ◽  
Tacit Knowledge ◽  
True Variance ◽  
Complex Decision ◽  
Maintenance Decision ◽  
Maintenance Activities

AbstractMaintenance decision making is an important part of managing the costs, effectiveness and risk of maintenance. One way to improve maintenance efficiency without affecting the risk picture is to group maintenance jobs. Literature includes many examples of algorithms for the grouping of maintenance activities. However, the data is not always available, and with increasing plant complexity comes increasingly complex decision requirements, making it difficult to leave the decision making up to algorithms.This paper suggests a framework for the standardisation of maintenance data as an aid for maintenance experts to make decisions on maintenance grouping. The standardisation improves the basis for decisions, giving an overview of true variance within the available data. The goal of the framework is to make it simpler to apply tacit knowledge and make right decisions.Applying the framework in a case study showed that groups can be identified and reconfigured and potential savings easily estimated when maintenance jobs are standardised. The case study enabled an estimated 7%-9% saved on the number of hours spent on the investigated jobs.

scholarly journals Optimal Policies for Quantum Markov Decision Processes

2021 ◽  
Author(s):  
Ming-Sheng Ying ◽  
Yuan Feng ◽  
Sheng-Gang Ying
Keyword(s):  
Decision Making ◽  
Reinforcement Learning ◽  
Quantum Systems ◽  
Sequential Decision Making ◽  
Mathematical Framework ◽  
Sequential Decision ◽  
Learning Techniques ◽  
Optimal Policies ◽  
Markov Decision ◽  
Programming Algorithms

AbstractMarkov decision process (MDP) offers a general framework for modelling sequential decision making where outcomes are random. In particular, it serves as a mathematical framework for reinforcement learning. This paper introduces an extension of MDP, namely quantum MDP (qMDP), that can serve as a mathematical model of decision making about quantum systems. We develop dynamic programming algorithms for policy evaluation and finding optimal policies for qMDPs in the case of finite-horizon. The results obtained in this paper provide some useful mathematical tools for reinforcement learning techniques applied to the quantum world.

Sign in / Sign up

Export Citation Format

Share Document