Objective “National Fitness” is a national strategy that can help promote the development of green and convenient “new fitness carrier”. Waterfront public spaces can offer a comfortable environment for physical exercise, and can provide free fitness opportunities anytime and anywhere. Therefore, they have the potential to serve as an excellent “new fitness carrier”. However, the fitness resources available in many waterfronts have not been fully exploited. Based on the multi-agent behavior simulation technology and waterfront fitness quality assessment system, this research conducts a quality measurement of nine typical waterfront public spaces along the Huangpu River, and explores the possibility of further optimizing the spatial quality of the Minsheng Wharf area, aiming to provide a systematic method for quality assessment and optimization prediction. Two key aspects are highlighted, one is the simulation of waterfront fitness behavior, including model selection, operating mechanism, and fitting method, and the other is the measurement indicators related to blue-green fitness, which are derived from the measurement of blue-green exercise space, blue-green exercise behavior, and public services for fitness.

Methods The first is the development of an indicator system, which include dimensions of venue characteristics, facility service levels, and environmental support for physical well-being. To address fitness demands and waterfront space issues, eight primary indicators are identified, including accessibility, openness, diversity, completeness, adaptability, efficiency, water proximity, and comfort. Additionally, 21 secondary indicators are selected based on waterfront space characteristics, fitness behavior, and simulation outputs. Their calculation formulas and weights are determined through expert questionnaires and the analytic hierarchy process. The second is simulation. On-site surveys are conducted to collect spatial and pedestrian activity data, while eye-tracking and perceived restoration data are obtained through recovery experiments. The initial and corrected recovery weights for different age groups and exercise types are calculated, forming the operational mechanism of the simulation model adopted. The agent-based social force models are then developed. After qualitative and quantitative adjustments, the simulation model is validated, and the final results are obtained through data analysis and indicator output. The above methods are then applied to the quality assessment and optimization prediction for Minsheng Wharf. Initially, the current quality is measured, showing that its overall quality is only superior to the Y2 sample among the nine samples selected along the Huangpu River. The primary issues are found in the service levels of facilities and environmental support for physical well-being. The research identifies specific problematic indicators within each dimension and analyzes related issues. Following this, eight optimization elements and 19 detailed improving points are proposed. Simulations are conducted for different detailed levels of individual elements, and the optimization schemes are streamlined. Finally, taking into account crowd flow and pre-established conditions, various optimization schemes are rehearsed and compared.

Results The comprehensive quality assessment indicates that Y1 and Y2 samples rank in the top tier, with X1, X2, and C2 samples following in the second tier. The alignment between the dimension-specific and overall measurement results shows that the environmental support for physical well-being was the greatest. The highest-performing individual indicators include the average open hours of fitness spaces, proximity to sanitary facilities, shoreline openness, and perceived restoration. Among the three measurement dimensions, the most significant contributing indicators are the per capita open fitness space area, fitness facility utilization rate and density of people near the water. Based on the quality measurements of the nine samples, reference values for the 21 secondary indicators are derived using the normal distribution method, serving as a benchmark for the quality assessment of other waterfront fitness spaces. The research on Minsheng Wharf provides optimal and suboptimal schemes under all conditions, indicating that the comprehensive quality scores of all eight optimization schemes are significantly higher than the current condition.

Conclusion The quality assessment system for public fitness spaces in urban waterfronts should be based on fitness demands, and take into account the characteristics of waterfront environments and outputs from behavior simulations. This study integrates spatial, behavioral, and psychological dimensions, and combines static and dynamic indicators. Three key measurement dimensions, namely characteristics of fitness venues, service levels of facilities, and environmental support for physical well-being, are identified along with eight primary and 21 secondary indicators. Formulas, weights, and reference values are established to provide a quantitative basis for evaluating both the current status and potential optimization of waterfront fitness spaces. Improvement requires identifying problematic dimensions, indicators, and elements, and making accurate prediction for future usage of optimized schemes. Leveraging full-time dynamic simulation, multi-scheme comparison, and virtual scheme rehearsals, this research develops methods for fitness preference analysis, model construction, and simulation fitting. The optimization prediction process involves assessing current quality, simulating single-element optimizations, simplifying schemes, and generating multidimensional recommendations. These methods are applied to assess the quality of nine typical waterfront sections along the Huangpu River, producing results that may serve as a benchmark for other similar areas.