The invention discloses a method for carrying out rapid probability modeling on large-scale renewable energy source data. The method comprises the following steps of: building a distributed parallel framework for rapid new energy source modeling on the basis of a Spark and an HDFS, and enabling the framework to be compatible with an existing storage system of renewable energy sources; constructinga parentage map for solving Wakeby by utilizing RDD fault tolerance and memory calculation-based characteristics, so as to realize an RDD-based rapid novel Wakeby distribution estimation algorithm; and deducing a function relationship between a calculation time of the parallel estimation algorithm and a sample data size and two important adjustable parameters of the Spark. According to the method, the deduced function relationship between the time consumption of the algorithm and the sample size, a degree of parallelism and a number of partitions is beneficial for helping program adjustment and optimization of programmers on a Spark platform, time consumption prediction of the programmers under given configuration and hardware configuration of the programmers under an expected time undera given sample data size.