考虑风险相关性的软件风险多目标优化控制研究

Multi-object risk control model for software project considering risk correlation

研究软件风险控制的理论和方法,对提高软件开发成功率有着重要作用.面对软件风险管理的精细化要求,已有的单目标风险控制模型难以有效管理软件风险.本文将软件风险控制成本和软件风险暴露值作为控制目标,提出软件风险多目标优化控制模型.进一步将风险损失层面的相关性纳入风险控制模型,建立了考虑风险损失非可加性的风险控制多目标优化模型,刻画软件风险管理实践中存在的风险相关性问题.采用多目标粒子群算法对风险控制多目标优化模型进行求解,并采用一个软件开发项目的风险控制问题进行案例分析.分析结果表明,在软件风险暴露和软件风险控制成本两个目标之间近似呈现非线性置换关系,可以根据项目中风险控制成本的实际情况,快速找到对应的最优风险控制策略.考虑风险相关性的风险控制模型能刻画出软件风险管理实践中更加复杂的关系,给出更加符合实际情况的风险控制策略,对提高软件风险管理水平有着重要意义.

Research on software risk control theories and methods play an important role in improving the success rate of software project. With the requirement of software risk refined management, the existing single-objective risk control model is difficult to manage software risks effectively. This paper regards software risk control cost and software risk exposure as control objectives, and proposes a multi-object risk control model for software project. Furthermore, it puts forward a multi-object risk control model for software project considering risk correlation. The multi-objective particle swarm optimization is applied to solve the risk control model. Moreover, the example is applied to a medium-sized software project. Case analysis results show that there is a nonlinear trade-off relationship approximately between the software risk exposure and software risk control effort. The software manager can identify the corresponding optimal risk control strategy according to the actual risk control cost. The results also indicate that the consequence from the multi-object risk control model for software project considering risk correlation can better describe the actual situation of risk management.