Considering the increment of air conditioning electricity consumption, air conditioning load aggregators（ALA）are prone to participate in bidding electricity market by managing enormous quantity and dispersive air conditioning load （AL）,which not only reduces load peak valley difference, but also provides the opportunity for AL to take part in power market bidding.This work adopts reference price effect from behaviour economics to depict electricity consumption preference. A bilevel optimization method is established. The upper level problem aims to maxi mize the social welfare, while the lower level aims to maximize the ALA benefits based on consumers’preference. Karush Kuhn Tucker condition is applied to transfer the complex bilevel problem to a solvable single level problem. Case studies are provided to verify the feasibility of the proposed model.