中国经济发展水平区域差异的人工神经网络判定

Analysis on Regional Differentiations of Economic Development by Using Artificial Neural Networks

本文在对目前经济发展水平度量方法进行分析的基础上,运用人工神经网络(ANN)的理论和方法,构建了ANN模型分析中应用最为广泛的BP网络,并对2000年中国31个省、市(自治区)的经济发展水平进行了评价。网络运行结果表明,中国经济发展水平的区域差异显著,评价结果与专家的判断基本近似。根据评价结果,采用最短聚类分析法,将中国区域经济发展水平分为5级,经济发展水平较高的省(市、区)主要分布在东部沿海地区,经济发展水平较低及落后的省(市、区)主要分布在中部和西部地区,中国经济发展水平的区域差异主要表现为东部和中西部及沿海和内地的差异。可见,人工神经网络用于评价经济发展水平简便、实用,且避免了人工确定指标权重的主观性,是一条具有发展和应用前景的途径。

Artificial neural networks, originally inspired by their biological namesakes, are composed of many simple intercommunicating elements, or neurons, working in parallel to solve a problem. Applied present situation of artificial neural network theory for geological study is analyzed in this paper. Back-propagation (BP) neural network is a widely used artificial neural network model with the character of self-training and strong capability in solving nonlinear problems. On the basis of economic development level evaluation techniques, application of ANN theory and method, adopting 10 economic indexes, we develop BP neural network model with 10 neurons in input layer, 6 neurons in hidden layer and 1 neuron in output layer in economic development level evaluation with which Chinese economic development level in 2000 is evaluated. The run results of BP show that the regional differentiations of economic development level in China are obvious which accords with reality and the opinions of experts. The first is Shanghai with the highest economic evaluation value 4.538 and the second is Beijing, with 4.153.Guizhou province with the lowest economic evaluation value 1.146 is the last. The economic evaluation value of Shanghai is 3.96 times that of Guizhou province. According to the economic evaluation value, the shortest clustering method is adopted and 31 provinces (cities) or autonomous regions are classified into 5 groups. The first groups with the highest evaluation value include Shanghai and Beijing two cities and the last groups with the lowest evaluation value include Yunnan, Gansu and Guizhou three provinces. The provinces (cities) or autonomous regions with higher evaluation value are mainly distributed in the eastern coastal areas and the provinces (cities) or autonomous regions with lower evaluation value are mainly distributed in the middle and western areas. The economic disparities between the eastern and middle-western areas and the littoral and hinterland are the main regional economic gaps in China. The evaluation results also indicate that application of BP neural network to assessing regional differentiations of economic development level without assuming parametric relationship is convenient, precise and feasible which can be an alternative approach of assessing regional differentiations of economic development level.