A fair and reasonable electricity settlement mechanism in the electricity market is a key factor for the successful operation of the market. Based on the systematic analysis of the trading mechanism of the electricity retail market, the study proposes a distributed computing framework based on the settlement of the electricity retail market. In view of the settlement price problem in the retail market, a mixed-dimension particle swarm algorithm is used to design a settlement price prediction model for the retail electricity market and to realize the construction of a mathematical model for electricity retail settlement. Comparison experiments of different prediction models are carried out with relevant data from two regions in a certain place, and it is found that the MD-PSO model in this paper has superior retail settlement price prediction accuracy and its performance in terms of RMSE values (210.614 and 126.832), MAE values (22.567 and 7.157), and MAPE values (20.729% and 8.614%) is better than that of the other models. The empirical analysis of the electricity retail settlement model is also carried out, and the settlement deviation is below 10% in all months except February and March, which proves the applicability and practical value of the electricity retail settlement model in this paper.