Numerical simulation and physical model experiment are important methods for studying nearshore wave.Before implementing numerical simulation and physical experiments,it is necessary to determine the incident waves elements based on the characteristics of nearshore waves.Nearshore waves are typically dominated by mixed waves,which contain multimodal independent wave systems,each of which requires several wave characteristic parameters for complete description.It is of great significance to realize the selection of representative sea states for multimodal wave systems.Firstly,this paper introduces popular data mining algorithms,and then according to the measured wave data of Hambantota Port,it selects the representative wave conditions defined by two-dimensional,three-dimensional and multi-dimensional wave characteristic parameters respectively using maximum dissimilarity algorithm(MDA).The results show that the outcome of the MDA is better than the manual results for the two-dimensional wave parameters that can be artificially selected.If the wave conditions are defined by three-dimensional or even multidimensional wave characteristic parameters,the representative sea states selected by MDA algorithm are even distribution in data space and have high representativeness,which proves the MDA algorithm is suitable for selecting representative wave characteristic parameters under mixed sea conditions.