鉴于目前人工神经网络主要应用于自然科学和工程技术等领域上,选取了以长江流域97个地级市为研究对象,以反映城市经济发展水平、辐射能力、吸引能力、生活方便舒适程度以及文化娱乐水平等指标进行城市实力等级的研究,来弥补人工神经网络在人文社科方面应用的不足。结果表明,城市等级分为六级与实际情况非常近似,这说明白组织人工神经网络具有良好的聚类功能;同时,还得出长江流域不同区段内都存在一至两个具有大区意义甚至全国意义的特大城市,从而为构建合理的城市群、城市密集地区等城市体系提供了更有说服力的依据;此外,长江流域东中西部城市等级存在的较强的非均衡性也得到了较为详细的分析。
Since artificial neural network (ANN) is a powerful tool applied in the field of natural science and engineering technology, this paper explored, using ANN, the urban strength hierarchy in the Yangtze Basin to build up a bridge to link ANN to humanities. 20 indexes were selected including economic development level, radiation and attractive forces, the level of life convenience and cultural establishments, and the data in 2004 city statistical yearbook of China were analyzed with the software Matlab 6.5. The results indicated that to classify the hierarchy of urban comprehensive strength to six groups is accordant with the reality, showing the excellent clustering function of self-organizing artificial neural network; there is one or two metropolis with great significance to a greater region even to the whole country in different sections of the basin-namely, Chongqing and Chengdu in the upper reaches, Wuhan in the middle reaches, and Shanghai, Nanjing and Hangzhou in the lower reaches-providing persuasive reasons for constructing reasonable urban agglomerations in the basin. Also, the nonequilibrium status of urban hierarchy in the Yangtze Basin was analyzed in the paper.