Aiming at the multi-index collaborative energy dispatch problem of grid-connected wind, solar and storage microgrids under the condition of high proportion of renewable energy access, an optimal dispatch method based on deep deterministic policy gradient (DDPG) is proposed. First, a microgrid dispatching model consisting of wind power, photovoltaic, energy storage, user load and upper-level power grid is established; Second, a composite reward function is constructed that takes into account operating income, power balance, new energy consumption and energy storage operation constraints, and power imbalance, non-consumption of new energy and energy storage state-of-charge out-of-bounds are set as penalty items; Finally, DDPG is used to continuously adjust the energy storage charging and discharging power and grid interactive power, and adaptive attenuation noise is introduced to take into account early exploration of training and later stable convergence. The proposed method can significantly improve the training convergence speed and online decision-making efficiency while ensuring dispatching economy, and is suitable for microgrid energy dispatching under conditions of wind and solar output, load and electricity price fluctuations.