貌似是要求

Maximum Likelihood estimation for a mixture of Gaussians

For a mixture of different normal distributions (a Gaussian Mixture Model),
the correct parameters for mean and standard deviation for each Gaussian
cannot be computed by simply taking mean and std of the entire data set.
The observed data must be divided into several Gaussians,
each of with its own mean and standard deviation.
One approach for obtaining the paramaters for each Gaussian is to
apply the algorithm for Expectation Maximization.

  http://www.mathworks.com/help/toolbox/stats/gmdistribution.fit.html

假设

    X:mu1=3;sd1=7

      数据在matlab模拟产生

    Y:mu2=9;sd2=2

      数据在matlab模拟产生.

请问你要求甚么?