Aiming at the integrated WG- PV- ESS planning problem of distribution network considering the uncertainty of distributed power and load. Firstly, the uncertainty of distributed power and load is characterized by a probabilistic model, and the probabilistic power flow was solved based on the improved semi-invariant method. Then, considering the investment cost, carbon cost, economic benefits and voltage improvement, and aiming at maximize the comprehensive benefit, the comprehensive planning model of WG- PV- ESS comprehensive planning model considering reactive power compensation device is established. A carnivorous plant algorithm based coevolutionary algorithm is proposed to solve the model. In this algorithm, the multi-agent comprehensive planning problem to be solved is decomposed into three sub-planning problems:distributed generation, energy storage and reactive power compensation devices, and each sub-problem is solved based on CPA, and the optimal planning scheme is obtained by coordinating the evolution of various groups through the ecosystem. Finally, the rationality of the proposed model and the efficiency and universality of the solution algorithm are verified by IEEE33 and IEEE69.