The invention discloses a method for estimating sea surface channel parameters based on an improved support vector machine and grey modeling, and relates to wireless communication. The method comprises the steps of 1) acquiring original channel parameter data by using a first acquisition unit, and preprocessing the original channel parameter data by using a second preprocessing unit; 2) inputting a channel parameter estimation sample into a third estimation unit to perform channel parameter estimation; 3) enabling a prediction result to act as input of a grey model GM (1, 1), performing accumulation and equalization processing on the input data to acquire a once-accumulation sequence and a whitened background sequence, building a grey differential equation according to a grey model, solving a prediction function, performing discretization on the grey differential equation to acquire a grey difference equation, solving prediction parameters a^, b^ by using a least square method, and substituting the prediction parameters into the prediction equation to solve a prediction result; 4) acquiring a prediction equation about a regulation factor gamma by using an optimization algorithm, building an optimization model, and finding out a value capable of enabling a prediction error value to be the minimum of the regulation factor gamma; and 5) substituting the value of the regulation factor gamma into the step 2) and the step 3) so as to acquire accurate channel parameter prediction values.