| Demand response,as a way to regulate the balance between power supply and demand,plays an moderately significant part in keeping the stable operation of the power grid.Over the past few years,as the size of the data center expands,the growth of power consumption and its flexible power regulation capability,the data center has become one of the main roles in power demand response.In order to meet the power demand response,the traditional data center meets the demand by regulating the energy consumption of the computing equipment.However,this regulation method is no longer applicable in the multi-tenant data center facing the incentive separation.To dispose of the problem of incentive separation,the chief work of the thesis is to research and design a reasonable incentive mechanism to encourage tenants to participate in demand response.When tenants participate in demand response,design an efficient tenant selection method to solve the optimal tenant combination and achieve the purpose of minimizing costs.The specific research work is as follows:(1)Aiming at the lack of incentive mechanism in multi-tenant data centers,a power demand response incentive method for electricity pricing based on Stackelberg game is proposed.First,in light of the power supply and demand relationship,the pricing communication between operators and tenants is viewed as a Stackelberg game demand response model with a single leader and multiple followers.Next,a cost-benefit model is established for tenants and operators,and proves that there is a Nash equilibrium point under this model through theoretical analysis.Then the Nash equilibrium point is figured out by the proposed pricing incentive method.This equilibrium point can not only make the tenants actively participate in the demand response but make the operators and tenants obtain the best income.Finally,through simulation experiments,the algorithm is contrasted with other typical optimization methods.The consequences show that the proposed pricing incentive method not only increases the initiative of tenants to participate in demand response,but also realizes the optimization of the cost of both supply and demand,and shortens the solving time.(2)In order to meet the demand response and reasonably select tenants to reduce costs,a cost optimization method based on improved polar bear algorithm is proposed.First,the interaction between the tenant’s bid and the operator is modeled as a cost optimization model of reverse auction,which encourages the tenant to actively participate in the reverse auction,and submits the corresponding energy-saving plan and expected reward.Then,the thesis formally describe the cost optimization problem,and use the improved polar bear algorithm to solve the optimal cost and tenant combination.The sigmoid function is used to discretize the location of individual polar bears,so as to simulate whether the tenant’s bid is selected.The mutation strategy is added to the standard polar bear algorithm to increase individual diversity and the possibility of finding the optimal solution.And the thesis replace fixed field of view with adaptive field of view to dynamically adjust the radius of local search,thus improving the efficiency of finding the optimal solution.Finally,the improved polar bear algorithm is contrasted with other typical intelligent optimization algorithms.The experimental solutions show that the improved algorithm can not only reasonably select tenants to meet the needs,but also reduce the time cost of searching the best tenant.This thesis conducts in-depth research on pricing incentives and cost optimization in multi-tenant data centers,and provides new ideas and methods for further research on demand response,rational planning of resources on the demand side and response side,and ensuring the stable operation of the electricity market.It also extends the application of the polar bear algorithm to this field. |
