Prediction of Kansei image for flight simulator cockpit based on back propagation neural network&genetic algorithm

To address the insufficient consideration of pilots' Kansei image needs in flight simulation cockpit design, this study proposes a Kansei image design method that integrates the Backpropagation Neural Network (BP) and Genetic Algorithm (GA). Representative cockpit samples are selected, and eye-tracking experiments are conducted to extract and classify visual field design feature factors, reducing subjective partition interference. Using the KJ method and text mining techniques, key Kansei image indicators are identified through multidimensional analysis of evaluation words, further refined using the NASA-TLX scale to establish a comprehensive evaluation index system. A BP neural network is employed to map visual field features to Kansei image demand indicators, generating a high-precision predictive model. Based on this model, a genetic algorithm optimizes the combination of visual field features, providing solutions tailored to specific Kansei image needs. Validation through design practices and evaluation experiments demonstrates that the BP-GA-based method enables quantitative analysis and optimization of cockpit Kansei image design. The proposed approach effectively addresses pilots' Kansei image demands, reduces task load, and offers a novel technical pathway and theoretical foundation for optimizing future flight simulation cockpit design.