The invention discloses a method for estimating multipath parameters based on an improved differential evolution algorithm of generalized mutual entropy, and belongs to the field of signal processing.The method mainly comprises the following steps: after converting a multipath parameter estimation problem into an optimization problem with constraint conditions, designing an objective function byusing generalized mutual entropy, restricting the range of multipath parameters by using priori knowledge, and finally, iteratively estimating the multipath parameters by using an improved differential evolution (DE) algorithm. The improved DE algorithm used is a class constrained DE algorithm that determines three values, namely, the scaling factor Fi, the cross-over rate CRi and the number of difference components ki based on individual classes. The multipath estimation method provided by the invention can improve the precision and speed of multipath estimation based on the intelligent optimization method under non-Gaussian noise.