This study intended to provide an identification method to estimate the thickness of sea ice with the temperature of sea ice and sea water, which avoids the limitations of using thickness data directly. Firstly, a quasi-linear thermodynamic system with unknown thickness of sea ice is established, which proved the existence and uniqueness of its solution. Secondly, a parameter identification model is constructed which takes the thickness of sea ice as the identified variable, and takes the temperature deviation between the system output and practical observations as an objective function. Finally, a new algorithm, containing of the semi-implicit difference scheme, Genetic Algorithm and Hooke-Jeeves Algorithm, is established for  obtaining the thickness of sea ice. We proved our method is effective and feasible by analysing the sensitivity of identification variable.