In order to avoid risks and improve the market share and sustainability of electricity sales companies in the future electricity retail market

it is necessary to provide end-users who have retail options with scientific

reasonable and competitive electricity sales package. A new idea for sales package design optimization is proposed. The decision model for consumers' internal reference price(IRP)is built considering the preference behavior of consumers and introducing classification assimilation coefficient. The sticky behavior(or inertia behavior)of a single user is defined

and the sticky function for a single user is established. The user decision matrix is established to reflect the package selection by group users and the total utility function of electricity sales companies is formulated based on user stickiness and decision matrix. To search for the optimal object

the package structure combining fixed costs and unit costs is taken as a basis

while the package price as a variable. The cross particle swarm optimization(PSO)algorithm is used to improve the traditional method for searching multiple populations simultaneously by taking the cell array as a unit. At the same time

two amendments are made to the decision matrix to simulate the absolutely rational consumption under the condition of no preference and the irrational consumption under the condition of preference. Finally

a case study is conducted to verify the feasibility and optimization effect of the model by comparing packages selection by the potential customers before and after optimization.