Scheme generation and optimal selection of principal demensions for large naval ships were studied based on chaotic differntial evlution for multi-objective(CDEMO) and instuitionistic fuzzy multiple attribute group decision making(IFMAGDM). An optimization model was established using objective attribute indexes. The pareto-optimal scheme set was generated using CDEMO. The model of IFMAGDM was introduced for scheme optimal selection. Aimed at the problem of consisency correction, the concept of adjustment of biggest degree was proposed to improve the iterative algorithm, which was efficient and satisfying to the initial preference of experts. The case study indicates that  the method of consisency correction is highly reliable and the proposed mothod for scheme generation and optimal selection is highly practical, which offers a novel, simple and efficient avenue for large naval ship preliminary design.