Coupled evaluation and forecasting of smart city sustainability with Kolmogorov-Arnold networks

Smart cities face the challenge of balancing infrastructure development, pollution management and ecological sustainability, especially in unevenly developed regions in China’s urban areas. A specific gap in current practice is the absence of integrated forecasting and evaluation tools capable of addressing regional disparities and optimising sustainable urban planning. To fill the gap, this study proposes a novel integrated evaluation framework that combines TOPSIS, Coupled Coordination and Kolmogorov-Arnold Networks (KANs) for time-series forecasting of indicators for systematic sustainability evaluation. The results indicate that the overall coordination level of China's 21 smart cities has gradually improved, exhibiting progressive coordination characteristics. The comprehensive evaluation framework effectively reflects the long-term changes in smart city development and reveals potential synergies between regions. This study provides an evaluation basis and a practical path for regional differentiation policy and optimal resource allocation. • Multidimensional framework combines static evaluation with dynamic city forecasting • KANs forecasting and spatiotemporal dynamic feature capture for smart cities • Integrated evaluation reveals deeper relationships in regional urban synergies