KI-Glossar
Das vollständige Wörterbuch der Künstlichen Intelligenz
Maintenance Policy
Strategy defining maintenance actions to be taken in every possible state of a system, aiming to optimize an objective such as cost minimization.
Maintenance Agent
Algorithmic entity in RL that observes equipment state and decides on maintenance actions (inspection, repair, replacement) to undertake.
Cost Reward
Negative reward signal in RL for maintenance, penalizing repair costs, production downtime, and unnecessary maintenance actions.
Degradation State
Representation of the current state of equipment in the MDP, including indicators such as wear, vibrations, temperature, and failure history.
Failure Simulation
Digital model reproducing equipment degradation behavior, used to generate training data for the RL agent without real risk.
Transfer Learning
Technique allowing the reuse of knowledge acquired by an RL agent on one type of equipment to accelerate learning on similar equipment.
RL-Based Condition-Based Maintenance
Approach where maintenance decisions are triggered by degradation thresholds learned dynamically by the RL agent rather than predefined.
Multi-Agent for Interconnected Systems
Extension of RL where multiple agents collaborate or compete to optimize maintenance of interdependent equipment in a complex system.