Research On Adaptive Negotiation Strategy Based On Multi-agent Supply Chain Production-marketing Collaboration Conflict Resolution

Negotiation is an effective way to solve collaborative conflict for multi-Agentsupply chain. Especially, in the case that customer demand of downstream market isconstantly changing, the production-marketing cooperative conflict between themanufacturer and retailer for ordered price, quantity of products inevitably occurs.Therefore, research on how to adapt to the opponent and the changing market demandbased on intelligent negotiation strategy is gradually becoming an important problemfor production-marketing collaboration of multi-Agent supply chain. The adaptivenegotiation strategy based on machine learning to resolve the production-marketingcooperative conflict of multi-Agent supply chain is deeply studied.First, to solve the problem of the parallel complexity of multi-Agent negotiation,it sets a negotiation framework of multi-Agent production-marketing collaboration.The blackboard Agent can be used as an intermediary to coordinate the supply chaininformation communication, and can real time induce the changes of marketingcustomers needs, it makes the both Agents timely give effective methods.Secondly, to solve the problem of poor adaptability and solving conflict lowefficiency of traditional negotiation, three adaptive negotiation methods based onoptimized machine learning algorithms which are applicable to solve theproduction-marketing cooperative conflict of supply chain are designed. The firstmethod is using grey relational degree and genetic algorithm to optimize case-basedreasoning(CBR), rival proposal is regarded as target case, grey relational degree isused as the calculation method of case matching, intelligent search mechanism ofgenetic algorithm can optimize the weights of case issues. The second method is usingsimulated annealing(SA) algorithm to optimize particle swarm optimization(PSO),adaptive PSO factor is designed, independently searches the global optimalnegotiation result, the SA temperature controls avoiding to fall into local optimum.The third method is using Adaboost algorithm to optimize the Q-reinforcementlearning algorithm, it transforms the historical values from weak classifier into strongclassifier to gradually obtain a more accurate forecasting value of rival proposal.Then, to solve the problem of diversity of production-marketing cooperativeconflict of supply chain and weak conflict resolution intelligence, a negotiation modelthat takes ordering price and batch as issues by establishing a model forproduction-marketing cooperation is set. Respectively aiming at three types ofconflicts, adaptive negotiation strategies of conflict resolution are given. First, to achieve their value maximum utility, conflicts occurs between the two sides. Anegotiation strategy adapted to rival proposal changes based on optimized CBRmethod is proposed. Secondly, to achieve minimization of the cost for supply chain,conflicts occurs between the two sides. A self-learning negotiation strategy that setsthe minimized cost as the fitness function based on the optimized PSO method isproposed. Thirdly, the market customer demand changing leads to theproduction-marketing cooperative conflict. An adaptive negotiation strategy that isadapted to changes and predicts rival proposal based on optimized Q-reinforcementlearning method.Finally, experiments are performed to verify the reasonabiltiy and effectivenessof the strategies proposed. It designs an order model between the manufacturer Agentand retailer Agent, and assigns variables and parameters according to the negotiationmodel. Aiming at the three types of production and marketing cooperative conflictsproposed abve, simulates on the process of negotiation adopting different adaptivemethods. It compares the negotiation results using the adaptive strategies based onoptimized methods proposed in this paper and the negotiation results usingunoptimized methods.The self-adaptive negotiation strategy that can be adapted to the opponentproposal is discussed, achieve the optimal cost, and adapted to changes of marketdemand based on the production-marketing cooperative conflicts of multi-Agentsupply chain. Research results can eliminate production-marketing cooperativeconflict, improve the negotiation efficiency, strengthen the intelligent level of supplychain, not only provide theory support for supply chain collaboration, and makemulti-Agent negotiation towards more efficient development.