Usually traditional optimization method for composite materials’ (laminates, sandwich structure) started from choosing some design variables like ply number, ply angle, stacking sequence, and the thickness of face and core materials. Second, set up the constrains and objective functions. Finally, apply these design variables and constrains into the optimization algorithm to solve. If design variables and objective functions are not smoothly and monotonically related, such as ply angle as a variable, the evolutionary algorithms are usually adopted since the gradient-based optimization algorithms could only search for local optimal solution. Although the evolutionary algorithms have the ability to search for global solutions, a number of the iterations is needed. In the same time, if the evolutionary algorithms are combined with Finite Element Method to solve the constrains and objective functions due to the complexity of the problem, the time consumed would be considerable. This research proposes a hybrid algorithm for composites based on lamination parameters. The optimization problem is divided into two. First, set the ply number as variable and search for the optimal solution by gradient-based optimization algorithms. Second, set the ply angle as the variable and fit the optimal ply number by evolutionary algorithms. Since the Finite Element Analysis is induced in the first step, the consumed time to converge would be substantially reduced. This thesis first set up a feasible region for extension, bending, couple ply numbers by two-dimensional fitting inequalities. Then the feasible region is induced to the optimization method mentioned above to substantially limit the boundary to avoid the does not exist solutions. Additionally, the accuracy of analysis methods, the accuracy of optimization methods, the efficiency, and the versatility are sequentially verified and found out that about 80% of iteration in Finite Element computing is reduced when applying this method on laminates. Furthermore, the sandwich equivalent stiffness estimation is proposed so that this method could also apply to sandwich structure. Finally, the method is verified not only more efficient but also possible to find a better solution than the traditional method. Finally, an example is used to demonstrate the benefit the optimization method in this research: the structure design of the Personal Lightweight Electric Vehicle.