The randomness of distributed photovoltaic (PV) power, as one of the important characteristics of uncertainty, brings challenges to the risk assessment and control of system operation. In this paper, a randomness evaluation method for distributed PV power based on Copula function is proposed. Firstly, randomness decoupling analysis is carried out with the degree of disorder in two dimensions of amplitude and frequency, and the high-risk events of distributed PV power disorder change such as strong disorder and high-frequency disorder are focused and defined. Secondly, the joint distribution model of distributed PV power randomness elements is constructed based on Copula function, which reveals the comprehensive characteristics of randomness under the background of joint elements. Finally, through the introduction of the joint recurrence period and the co-occurrence recurrence period, the data support is provided for the configuration of station reserve energy storage operation hours and the execution cycle of source-grid-load-storage interaction projects for distributed PV power stations. Combined with the historical data of a rooftop PV power station, the feasibility and effectiveness of the method are verified by a simulation arithmetic examples.