基于专家知识的ANFIS软件质量评价建模与仿真

通过研究目前广泛用于软件质量评判的模糊综合评判法,分析了该方法存在的不足,并提出了基于专家知识的自适应神经推理(adaptive neural-fuzzy inference system,ANFIS)软件质量评价模型。在建模过程中,利用专家知识获取输入指标的不同重要度,对指标进行修正作为软件质量评判模型的输入,结合ANFIS的自适应、自学习特性,实现了软件质量评价过程中推理规则适应度、隶属度函数形状及参数的自适应调节,提高了质量评判模型的可信度、准确度。最后运用该模型对嵌入式软件质量进行评价,仿真和试验结果说明了该方法的可行性和有效性。

Modeling and simulation of ANFIS to evaluate software quality based on expert knowledge

On the basis of analyzing the fuzzy synthetic evaluation which now usually used in software quality evaluation,the disadvantages of this method and proposes an adaptive neural-fuzzy inference system(ANFIS) software quality evaluation model which found on expert knowledge is analyzed.The expert knowledge is used to get the weight of the input index,then the index is modified so that it can be the input of the model.With combining the characters of adaptive and self-study of ANFIS,the self-adjustability of inference rules' fitness and membership function's shape are realized during the evaluation of the software quality.At last,this method is used to evaluate the embedded software quality,the simulation and testing results show that this method is correct and available.