This study was carried out for the purpose of practicing the concept of green, healthy and sustainable urban development, and building a healthy and effective regional economy. The urban sprawl theory is the theoretical foundation in this study. Creatively, deep learning and neural networks were combined to propose an algorithm. Afterwards, a theoretical model of urban sprawl growth was built, with urban industrial development as the principal indicator. According to cities’ economic data in the Yangtze River Economic Belt from 2005 to 2019, the model was optimized through data learning to obtain the final measurement system. On this basis, the urban sprawl levels in different regions could be analyzed. Results demonstrate that the model using the FEPA (Financial Time Series - Empirical Mode Decomposition - Principal Component Analysis - Artificial Neural Network) fusion algorithm has higher errors than other models, with an average accuracy of 82.61% and a total evaluation score of 0.778. Compared with the latest urban sprawl analysis models, the proposed model improves the performance by 7.9%, and keeps the hit rate above 70%, which is suitable for the urban sprawl measurement. In addition, the urban spatial pattern of the Yangtze River Economic Belt is always greater than 1, and the entire space is relatively narrow. From 2005 to 2019, the region's urban sprawl degree has increased by more than 50%. In general, the proposed model can well indicate the urban sprawl situation, so it has a significant reference value for research on the urban sprawl measurement system.
Read full abstract