Auction-based decision making for distributed real-time control of flexible manufacturing systems

The study presents the development, implementation, and benchmarking of two auction-based decision making models in a distributed real-time control framework for random flexible manufacturing systems. The framework applies an auction mechanism for facilitating part allocation on machine tools based on an exchange economy. Two auction models have been implemented. The first model, Modified Sealed-Bid auction, is built on seal-bid auction with the addition of iterative behavior. Parts and machines are represented by agents. The fundamental motivation of each agent in the system is to increase the reward. The Modified Sealed-Bid auction model has been benchmarked against typical dispatching rules using simulation in order to evaluate its performance. The simulation results indicate that the framework outperforms the dispatching rules on most of the performance measures.; The second auction model is built on the ascending auction concept in which part agents alternately bid on preferred machines and compete on the basis of increasing price. Valuation function is used as a reward indication in order to facilitate the auction. The Ascending Auction model has also been tested using the same simulation model and the performance of the two auction models has been compared using additional production scenarios. Simulation results reveal that the performance patterns of the two auction models vary. Both models have different strengths and weaknesses under different production scenarios. In conclusion, both simulation studies indicate that the proposed auction-based models have proven to be effective decision making tools in distributed real-time control of random flexible manufacturing systems.