Electric power generation market has game characteristics such as players with incomplete information and bounded rationality. To meet such requirements, evolutionary game theory is applied. Based on the differences among units status of generation companies, modeling and simulation are carried out for the biding strategy of each generating unit by random pairing with another member of game population with differential bounded rationality. A practical case study shows that model developed by multi-population replicator dynamic of evolutionary game can help power generating companies to predict competitorspsila strategies tendency quantitatively so as to adjust own strategies to achieve maximum payoff. The model is also helpful to the administrators of electric power market, who care for the performance of the whole market, to pursuit for better market management.