一、绪论国内外已不少学者通过构建不同模型对经济周期进行监测预警,人工神经网络是方法之一。神经网络实质是一个大规模的连续的非线性时间动力系统,它具有并行分布处理、连续时间非线性动力学、网络全局作用及学习联想能力和高度的鲁棒性等特点。[1]与此同时,它也具有不可预测性、自适应性、耗散性和吸引性等非线性动力系统的所共有的特性。[2]对于宏观经济这样一个复杂的“黑箱”运作系统,人工神经网络体现出其优良性。 I. INTRODUCTION Many scholars at home and abroad have been monitoring and warning the economic cycle by constructing different models. Artificial neural network is one of the methods. Neural network is essentially a large-scale continuous non-linear time-driven power system. It has the characteristics of parallel distributed processing, continuous time nonlinear dynamics, network global effect, learning associative ability and high robustness. [1] At the same time, it also has the properties common to nonlinear dynamical systems such as unpredictability, adaptability, dissipation and attraction. [2] Artificial neural networks show its superiority to such a complex “black box” operating system as macroeconomics.