Design of Public Space Virtual Scene Layout Optimization System Based on Convolutional Neural Network

This paper presents a design for a virtual scene layout optimization system for public spaces, utilizing convolutional neural networks (CNN) to enhance spatial efficiency and user experience. The system leverages CNN algorithms to analyze spatial data and optimize layout configurations by considering factors such as traffic flow, accessibility, and aesthetic appeal. Through real-time analysis and iterative adjustments, the system generates optimized virtual layouts that can be applied to diverse public spaces. The experimental results show that the space utilization rate of the design method is as high as 99.54%, which is significantly improved compared to other literature methods (up to 97.82%). When the total number of facilities increased from 3 to 18, the optimization time increased from 10.28 seconds to 14.26 seconds, with a growth rate of only 38.7%, far lower than the growth rate of other methods. The internal environmental diversity index remains above 0.80, which outperforms the fluctuation performance or low values of other literature methods. The separation degree of functional area spatial distribution is reasonable, effectively avoiding the problem of underutilized areas. By flexibly adjusting algorithm parameters for different types of public spaces, a dual improvement in cost control and resident satisfaction has been achieved. For example, in cultural venues, commercial complexes, and high-density urban centers, costs have been reduced by 500,000 yuan, 800,000 yuan, and 1.5 million yuan, respectively, and resident satisfaction has increased by 15%, 12%, and 10%, respectively. At the same time, based on actual survey data and expert evaluations, the satisfaction of residents in optimized public spaces has significantly improved, especially with the high weight value of seating facilities leading to a significant increase in satisfaction. These specific experimental data and quantitative indicators fully demonstrate the effectiveness and applicability of our method in the planning and design of internal environments in public spaces.