To solve the problem of approximating an arbitrary probability density in the G/G/1 queue system, an approach based on the use of hyperexponential distributions is proposed. An H2/H2/1 type system is considered as an approximation of the G/G/1 system. It is taken into account that the initial analyzed sequences have correlation properties. The paper proposes to use an EM algorithm adapted to obtain a mixture of exponential distributions when determining a hyperexponential model for correlated flows. Two approaches to determining the parameters of hyperexponential distributions based on the use of the EM-algorithm are considered: an approach based on the use of copulas and an approach based on the use of a regression equation.