The generation expansion planning is an important problem in electric system, especially complex and urgent in the background of electric market. To this issue, this paper proposes a generation investment planning model for power companies, which is based on mutli-agent technology. Firstly, a double-layer optimization construction for the generation-side in electric market is developed in order to combine the price competition with generation expansion planning. Secondly, multiple agents such as generation groups, generation market and generation plants are designed with genetic algorithm and Q-learning algorithm applied to improve their optimization ability. The result presents that the model is feasible and effective and could provide support for decision making to plant expansion planning. This work is supported by National Natural Science Foundation of China (No. 61273144 and No. 61533013).