In order to cope with the impact of wind power and photovoltaic uncertainty on the safe and stable operation of the power grid and to make up for the shortcomings of the traditional single-objective optimal power flow model, a transient stability constrained multi-objective optimal power flow (TSCMOOPF) model and a solution method are proposed to take into account the wind and solar uncertainty. Firstly, an ensemble learning method based on artificial neural network (ANN), deep neural network (DNN) and surprisal-driven zoneout long short-term memory (SZLSTM) are adopted to construct a wind and photovoltaic output prediction model to improve the prediction accuracy and robustness. Secondly, considering the economy and stability of the system, a multi-objective function including the minimization of active network loss, the minimization of fuel cost, and the optimization of the voltage stability index is established to construct a TSCMOOPF model. Then, an improved reference vector guided evolutionary algorithm (RVEA) is designed for the solution. Finally, simulation experiments are carried out on the improved IEEE 39-bus system. The results show that the proposed ensemble learning method performs well in wind and photovoltaic output prediction, the multi-objective optimization model ensures transient stability while active network loss and fuel cost are reduced significantly, and the improved RVEA algorithm is better than the traditional multi-objective algorithm in terms of convergence and diversity.