Component-based software engineering is currently a development strategy used to improve complex embedded systems. The engineers have to deal with a large number of quality requirements (e.g. safety, security, availability, reliability, maintainability, portability, performance, and temporal correctness requirements), hence the development of complex embedded systems is becoming a challenging task. Enhancement of the quality prediction in component-based software engineering systems using soft computing techniques is the foremost intention of the research. Therefore, this paper proposes an extreme learning machine (ELM) classifier with the ant colony optimization algorithm and Nelder–Mead (ACO–NM) soft computing approach for component quality prediction. To promote efficient software systems and the ability of the software to work under several computer configurations maintainability, independence, and portability are taken as three core software components metrics for measuring the quality prediction. The ELM uses AC–NM for updating its weight to transform the quality constraints into objective functions for providing a global optimum quality prediction. The experimental results have shown that the proposed work gives an improved performance in terms of Sensitivity, Precision, Specificity, Accuracy, Mathews correlation coefficient, false positive rate, negative predictive value, false discovery rate, and rate of convergence.

中文翻译：

---

一种使用 ACO-NM 优化的极限学习机的基于组件的软件系统的新质量预测模型。

基于组件的软件工程是目前用于改进复杂嵌入式系统的开发策略。工程师必须处理大量的质量要求（例如安全性、安全性、可用性、可靠性、可维护性、可移植性、性能和时间正确性要求），因此复杂嵌入式系统的开发正成为一项具有挑战性的任务。使用软计算技术增强基于组件的软件工程系统的质量预测是本研究的首要目的。因此，本文提出了一种具有蚁群优化算法和 Nelder-Mead (ACO-NM) 软计算方法的极限学习机 (ELM) 分类器，用于组件质量预测。为了促进高效的软件系统和软件在多种计算机配置下工作的能力，可维护性、独立性和可移植性被视为衡量质量预测的三个核心软件组件指标。ELM 使用 AC-NM 更新其权重，将质量约束转换为目标函数，以提供全局最优质量预测。实验结果表明，所提出的工作在灵敏度、精确度、特异性、准确度、Mathews 相关系数、假阳性率、阴性预测值、错误发现率和收敛速度方面具有改进的性能。ELM 使用 AC-NM 更新其权重，将质量约束转换为目标函数，以提供全局最优质量预测。实验结果表明，所提出的工作在灵敏度、精确度、特异性、准确度、Mathews 相关系数、假阳性率、阴性预测值、错误发现率和收敛速度方面具有改进的性能。ELM 使用 AC-NM 更新其权重，将质量约束转换为目标函数，以提供全局最优质量预测。实验结果表明，所提出的工作在灵敏度、精确度、特异性、准确度、Mathews 相关系数、假阳性率、阴性预测值、错误发现率和收敛速度方面具有改进的性能。