基于BP神经网络的企业技术创新能力评价及应用研究

【作者】 吕晓强；

【导师】 夏维力；

【作者基本信息】 西北工业大学 ， 企业管理， 2005， 硕士

【摘要】 技术创新被认为是经济增长的源泉和推动知识经济发展的有效途径。今天,企业的技术创新活动已经成为决定企业生存和发展的动力和源泉。因此,如何客观、科学、有效、定量地评价企业技术创新能力,对于企业在同行业竞争中客观的认识自身的技术创新能力,采取适当的技术创新战略,提供自身的竞争优势,获得最佳的经济效益和社会效益具有特别重要的现实意义。本文主旨是对企业技术创新能力进行分析,建立一套基于技术创新过程的技术创新能力评价指标体系,并在此基础上探讨对其进行综合评价的方法。 文章首先对企业技术创新和技术创新能力理论进行阐述,界定了企业技术创新能力的定义和构成要素,并在前人研究的基础上构建了基于技术创新过程的企业技术创新能力评价指标体系;其次,针对当前评价企业技术创新能力的方法所存在的不足,本文提出一种基于BP神经网络的企业技术创新能力评价方法。根据已建立的评价指标体系,设计BP神经网络模型,并给出简单可操作的评价程序:再次,在计算方法上,用MATIAB的神经网络工具箱(NNT)来进行网络设计和计算。通过大量学习样本的训练和测试,使模型的误差达到预定的范围内,至此一个完整的网络模型建立完毕;最后,以实例来验证这种方法的准确性和可操作性。更多还原

【Abstract】 Technologic innovation is viewed as the source of economy growth and an effective way of promoting knowledge-based economy development. Today, the activities of enterprise’s technologic innovation have already been the power and headspring that decide the subsistence and development of the enterprise. So, it is realistically significant for an enterprise to evaluate its TI capacity objectively, scientifically, efficiently and quantitatively. Only in this way can an enterprise scientifically realize its TI capacity in its competition with its counterpart, employ appropriate innovation strategy, make full use of its advantages and achieve the best economic and social benefits. The present paper makes an analysis of the enterprise’s TI capacity and sets up an index system based on the technologic innovation process to evaluate the technologic innovation capacity, and tries to find a way to synthesize it.In the first part, the author expounds the theories related to the TI, introduces its concept and the composing elements, and sets up an index system of evaluating the enterprise’s TI capacity which is based on the technologic innovation process according to former research. In the second part, because of the shortage of the evaluating the enterprise’s TI capacity currently, the author presents a new method of evaluating TI capacity, which is based on BP neural network. According to the index system that has been already set up, design the BP neural network model, and present the doable evaluating programs. In the third part, on the method of compute, the author uses the Neural Network Toolbox (NNT) based on MATLAB to make network design and compute. Through the training and testing of lots of studying samples, the error of the model will limit in a preconcerted range. Thus, a integrative network model has been established. In the end, the article proves the accuracy feasibility of the method with an example.更多还原