In order to support intelligent bidding in multi-market, this paper proposes a joint bidding model of risk preference modeling under price limit constraint. The electric energy market is divided into day-ahead and real-time sub-markets, and the auxiliary services are divided into standby and frequency regulation sub-markets. The framework integrates temporal state encoding, preference utility estimation, feasible region modification and iterative policy search to generate quotes based on historical prices, renewable forecasts, load trajectories and settlement signals. The experimental results on 17,520 hours of samples collected from 4 trading zones show that compared with the three benchmark strategies, the expected operating cost of the proposed method is reduced by 11.8%, the average bidding profit is increased by 9.6%, the downside risk based on CVaR is reduced by 13.4%, the average absolute bidding deviation is 2.7%, and the average computation time per round is 0.41 s. The results show that the proposed model maintains the stability of revenue under the limit price constraint, and improves the bidding feasibility from 91.2% to 97.8%, which can provide decision support for the computer-driven bidding system in the actual operation scenario of the electricity market.
