With the increasing diversification and personalization of market demands, traditional product styling design methods face significant limitations in dealing with complex parameters and subjective user evaluation. In this paper, a method based on fuzzy neural network (FNN) and genetic algorithm (GA) is proposed. Firstly, a fuzzy inference system based on Takagi–Sugeno model is constructed to fuzzify the design parameters into linguistic variables and optimize the weights by fuzzy rules and GA to realize the quantitative mapping of user requirements. Secondly, an improved FNN architecture combining Gaussian Radial Basis Function (RBF) and ReLU activation function with Adam algorithm for training is designed to enhance the nonlinear modeling capability. Finally, the FNN is combined with GA to construct a closed-loop optimization system, in which the FNN quickly predicts the design output, and the GA optimizes the parameters and iteratively updates them, and ultimately achieves the globally optimal design that meets the user’s needs. Experimental results show that the hybrid FNN-GA method outperforms single FNN, GA and other optimization algorithms in multi-objective design tasks. The method shows stronger robustness and generalization ability when dealing with high-dimensional design parameters and complex user requirements, providing a new path to solve multi-objective optimization problems. Future research will extend the application of the method in industrial design, architectural design and user experience optimization, and explore the combination of more machine learning techniques and real-time data processing to further enhance the adaptability and scalability of the system and provide support for intelligent design.