• A wasserstein generative adversal network with ladder punishment is proposed to deal with multiple uncertainties of scenery. The maximum fluctuation range of generated scenery output is more than 90 % consistent with the maximum fluctuation range of actual output, which can well reflect the volatility of scenery output. At the same time, the improved ISODATA (Iterative self-organizing data analysis techniques algorithm, ISODATA) algorithm is combined to reduce the scene. The effect of scene reduction is improved by 74.1 % compared with the traditional method, and the representativeness of typical scenes is higher. • In the aspect of safety, the risk of voltage exceeding the limit, branch power exceeding the limit and system load loss are proposed. In the aspect of economy, the risk of line loss and operating profit and loss of the system are considered. At the same time, an improved subjective and objective combination weighting method is proposed. The calculation results take into account the knowledge and experience of experts and the data characteristics of indicators, which makes the weight calculation results more convincing, determines the comprehensive operation risk of distribution network, and can be used as a reference for measures to improve the operation risk of distribution network. • The wind-wind output volatility - line failure rate model in lightning weather is established, and the distribution network operation risk under the consideration of wind-wind volatility and lightning weather is analyzed, and the impact of the wind-wind output fluctuation on the system operation risk is considered, and a variety of factors affecting the distribution network operation risk are comprehensively considered to improve the comprehensibility and accuracy of the assessment. Aiming at the problem that the operation state of distribution network is affected by the output fluctuation of renewable energy and the topology of distribution network, a comprehensive risk assessment method of distribution network based on WGAN-GP (Wasserstein generative adversarial network with gradient penalty, WGAN-GP) and improved subjective and objective combination weights is proposed. Firstly, a Wasserstein generation adversative network with ladder penalty is proposed to deal with multiple uncertainties in the landscape. At the same time, combined with the improved ISODATA (Iterative self-organizing data analysis techniques algorithm, ISODATA) scenario reduction method, and considering the change of the topology of distribution network caused by lightning weather, the wind and landscape output fluctuation-lightning weather line failure rate model is established. Secondly, considering the economy and safety of the operation of the renewable energy access system, the risks of voltage overstep, branch power overstep and system load loss are proposed in terms of safety, and economic risk indicators are proposed in terms of economy, considering the risk of system operation line loss and operating profit and loss, so as to comprehensively consider the source of system risk. Thirdly, the improved subjective and objective combination of weights is used to assign reasonable weights. The calculation results take into account the knowledge and experience of experts and the data characteristics of indicators, which makes the weight calculation results more convincing, determines the comprehensive operation risk of distribution network, and serves as a reference for measures to improve the operation risk of distribution network. Finally, CPLEX was used to solve the IEEE33-node distribution network optimization model to verify the effectiveness and feasibility of the proposed model and method, and further analyzed the distribution network operation risk considering wind and lightning weather, and considered the influence of flexible load balance wind and wind output fluctuation on the system operation risk. [ABSTRACT FROM AUTHOR]