摘要
This paper introduces a bilevel programming approach to electricity tariff optimization for the purpose of demand response management(DRM)in smart grids.In the multi-follower Stackelberg game model,the leader is the profit-maximizing electricity retailer,who must set a time-of-use variable energy tariff in the grid.Followers correspond to the groups of prosumers(simultaneous producers and consumers of the electricity).They respond to the observed tariff,schedule controllable loads and determine the charging/discharging policy of their batteries to minimize the cost of electricity and to maximize the utility at the same time.A bilevel programming formulation of the problem is defined,and its fundamental properties are proved.The primal-dual reformulation is proposed in this paper to convert the bilevel optimization problem into a single-level quadratically constrained quadratic program(QCQP),and a successive linear programming(SLP)algorithm is applied to solve it.It is demonstrated in computational experiments that the proposed approach outperforms earlier typical methods based on the KarushKuhn-Tucker(KKT)reformulation regarding both solution quality and computational efficiency on practically relevant problem sizes.Besides,it also offers more flexible modeling capabilities.
This paper introduces a bilevel programming approach to electricity tariff optimization for the purpose of demand response management(DRM) in smart grids.In the multi-follower Stackelberg game model,the leader is the profit-maximizing electricity retailer,who must set a time-of-use variable energy tariff in the grid.Followers correspond to the groups of prosumers(simultaneous producers and consumers of the electricity).They respond to the observed tariff,schedule controllable loads and determine the charging/discharging policy of their batteries to minimize the cost of electricity and to maximize the utility at the same time.A bilevel programming formulation of the problem is defined,and its fundamental properties are proved.The primal-dual reformulation is proposed in this paper to convert the bilevel optimization problem into a single-level quadratically constrained quadratic program(QCQP),and a successive linear programming(SLP)algorithm is applied to solve it.It is demonstrated in computational experiments that the proposed approach outperforms earlier typical methods based on the KarushKuhn-Tucker(KKT) reformulation regarding both solution quality and computational efficiency on practically relevant problem sizes.Besides,it also offers more flexible modeling capabilities.
基金
supported by the GINOP János Bolyai Research Fellowship.grant(No.2.3.2-15-2016-00002)
the NKFIA grant(No.129178)
the János Bolyai Research Fellowship.
