【摘要】 在大规模导航软件系统中 ,软件评估的主要任务就是要精确地估测程序模块中存在的缺陷数 .介绍了组合导航软件的评估方法和步骤 ,着重介绍了JM模型和G -O模型 ,并提出了神经网络评估软件可靠性的具体方法 .对组合导航软件的特点及测试方法进行了讨论 .根据软件模块测试报告 ,分别用JM模型、G-O模型和神经网络对软件模块的缺陷数进行预测 ,并给出了组合导航软件 3种模型的评估质量的比较 .结果表明 :神经网络方法优于传统模型方法更多还原

【Abstract】 A method is proposed for evaluation of software reliability of an integrated navigation system, especially JM and G-O model, and the evaluation for software reliability is based on a neural networks model. According to the software test record, the number of faults in the program is estimated using the model of JM, G-O and neural networks. The comparison of the three model quality for evaluation of software reliability, indicates that neural networks model has better quality than traditional model.更多还原