Research On Trading Agent SCM Model And Optimization Method

Supply chain management, which enhance the competitiveness of the whole enterpriseby optimalizing of the production of commodity, information sharing and service betweenbuyers and suppliers, is widely adopted by enterprises as management thought and method.However, with the rapid development of electronic commerce and artificial intelligence,traditional supply chain management mode can’t meet the needs of the enterprises today. Ine-commerce, how to make agent make strategies according to the change of marketenvironment is the research focus in the e-commerce field. Trading agent competition,referred as TAC, is co-sponsored by Carnegie Mellon University and other schools. TAC isdesigned to simulate the real market behavior and applied the current theoretical research inartificial intelligence to implementation the transaction process. TAC/SCM is the supply chainmanagement platform, the researchers can apply the studies to this platform and verify theeffectiveness of the idea. This paper focus on TAC/SCM platform, design a kind of Agentmodel which have specific adaptive, high competitive strategy, and puts forward the improvedPSO algorithm, thus effectively solved the sales module constrained optimization problems.Firstly this paper introduces the research status of supply chain management in recentyears, and briefly explain the TAC platform design thought and the rules of the game, andthen studies the reverse auction mechanism, Agent theory and the optimized predictionalgorithm’s related theory. Based on the TAC supply chain management platform, this paperdesign HEU2012Agent model, and compare HEU2012model with other Agent thatparticipated in TAC/SCM game Agent to verify the effectiveness of the model. But due to theincomplete information in the environment of high complexity, high adaptability and highuncertainty, this paper model adaptability is not too ideal. So according to the sales module ofthe constrained optimization problems, this paper puts forward a new kind of penalty functionmodel, and based on the biological parasitics theory to improve the basic particle swarmoptimization (PHPSO). In the experiment, We compare PHPSO with SPSO algorithm andFDR-PSO algorithm. Experiments show that PHPSO algorithm has better solutions for the local convergence problem. Finally, we applied the optimized sales module to the model,where the effect have been significantly tested.