摘要：探究产业结构的一般演进规律一直是经济学家关注的焦点。目前关于产业结构的研究较多，但在产业划分上过于粗略，在预测方法上对数据的要求苛刻而与现实不符。针对上述问题，本文首先搜集整理了1988-2005年中国制造业内部细分行业的产值数据，细致地刻画了近20年来中国制造业内部结构的变动趋势，结果发现我国制造业结构一直处于高度化的演进过程。此外，针对产业结构变动的影响因素错综复杂而难以建立有效的计量经济预测模型的难题，本文提出了一种基于微分进化算法的支持向量机预测模型，并利用中国制造业结构数据对该预测方法进行了检验。结果表明，基于微分进化算法的支持向量机预测模型能够有效地把握趋势信息而克服不确定及短期因素的影响，具有较强的可操作性和较高的预测精度。

Abstract: It is a focal issue in the literature to explore the general rule of industrial structure evolution. Although many researchers pay attention to the studies on industrial structure, there are two deficiencies, namely, one is the classification of industrial structure is too rough, another is the data required by forecasting model doesn’t match the real world. Therefore, the evolution of Chinese manufacturing industrial structure from 1988 to 2005 is depicted. It is revealed that Chinese manufacturing industrial structure is on the trend of upgrading all the time. Besides, a new forecasting model - support vector machine is proposed based on a differential evolution algorithm - to forecast Chinese manufacturing industrial structure. This model is tested as a good forecasting model; it effectively solves the problem that the econometrics forecasting model of industrial structure is difficult to be established.