基于环境与神经网络的软件自适应建模

1)深圳市检验检疫科学研究院,广东深圳 518045; 2)河北师范大学数学与信息科学学院,河北石家庄050024

软件工程; 软件自适应; 环境需求; 软件建模; 神经网络; 环境用例预测

Software adaptive modeling method based on environment and neural network
Qin Zhiwu1, Xie Jinxiong1, Cai Yi'na1, and Yan Yixuan1,2

Qin Zhiwu1, Xie Jinxiong1, Cai Yi'na1, and Yan Yixuan1,21)Shenzhen Academy of Inspection and Quarantine, Shenzhen 518045, Guangdong Province, P.R.China2)College of Mathematics and Information Science, Hebei Normal University, Shijiazhuang 050024, Hebei Province, P.R.China

software engineering; software adaptive; environment requirement; software modeling; neural network; environment use case prediction

DOI: 10.3724/SP.J.1249.2017.06570

备注

软件需求模型的建模是保证软件可靠运行的基础.传统方法在建模过程中对环境需求考虑较少,对环境的变化无法有效识别和合理应对,导致软件生命周期缩短.现有自适应建模过程属于被动感知需求,欠缺对未来可持续发展的需求加以预测和应对,同样无法延长软件生命周期.为了尽可能延长软件生命周期,提高重构开发效率,提出一种针对环境变化的软件自适应建模方法.该方法将软件运行所处环境作为需求单独分析处理,首先识别环境用例,其次构建环境用例并将功能指标进行量化处理,采用BP神经网络预测环境需求变化并作出应对策略.与HAN-YANG-XING模型比较,该方法可主动感知需求,对环境变化进行预测并做出适应性判断,有效延长软件生命周期.

The software requirement model is the basis to improve the development efficiency and ensure the reliable operation. However, the traditional methods can neither distinguish the environment requirements during the modeling process, nor effectively identify and give reasonable response to the environment changes. Meanwhile, the existing software adaptive modeling process belongs to passively sensing requirements and does not effectively predict and deal with future requirements. In order to solve the above-mentioned problems, a novel software adaptive modeling method is proposed to adapt to the environment changes. The method firstly utilizes the software environment in which the software is running as a separate requirement analysis. Then, the environment use case is identified and constructed and the function index is quantified. Finally, the BP neural network is used to predict the change of environment requirements and make the corresponding strategy. Compared with the HAN-YANG-XING model,the proposed method can actively sense requirements,predict the environmental changes and make adaptive judgments,which can effectively extend the software life cycle.

·