存在返工迭代和大量风险因素是复杂系统研发项目的典型特征。有些返工迭代是由于研发流程不合理导致任务间信息反馈过多造成的，应优化研发流程；有些返工迭代是由于存在技术风险因素，但为了获得预期技术性能水平而产生的，这时的返工迭代是积极的，但应评估由此带来的风险。考虑到流程结构和风险对鲁棒项目调度结果和过程的影响，提出研发项目流程优化、风险评估与鲁棒调度三者在建模、仿真、评估与优化上的有效集成，以该集成框架为核心开展深入的理论、方法和应用研究。深入研究研发项目中各类返工迭代及数学描述，建立基于DSM的研发流程多目标仿真优化模型，设计智能优化算法求解Pareto解；建立基于技术风险考虑返工迭代的研发项目风险评估模型；研究DSM向典型活动网络的转换方法，综合考虑任务参数、流程结构和项目范围的不确定性，提出新的鲁棒性度量方法，建立主动型／响应型调度模型，设计面向大规模问题的高效的混合智能优化启发式算法。

The existence of iterations and a large number of risk factors is the characteristic of complex systems research and development (R&D) projects. Some iteration is caused by excessive feedback between tasks produced by unreasonable project process, where process optimization is called for. While the other is the consequence of technical risk factors accompanied by pursuing desired level of systems performance and corresponding risks of such positive iteration should be assessed. Taking into account the impacts of project process architecture on procedures and results of robust project scheduling, the effective integration of modeling, simulation, evaluation and optimization for R&D projects process optimization, risk assessment and robust scheduling is proposed and is utilized to guide the in-depth theory, methodology and application research. Various types of iterations of R&D projects and their mathematical description are studied. Then, the multi-objective simulation optimization model based on the design structure matrix (DSM) is developed and intelligent optimization algorithms are designed to obtain Pareto optimum solutions. Further, considering iterations, the R&D projects risk assessment model is constructed based on technical risks. And the methods for transforming DSM into classic activities network are explored. Last but not least, new robustness measurements combing uncertainties, such as task parameters, process architecture and project scope uncertainties are presented; proactive/reactive project scheduling models are developed, and efficient and robust hybrid intelligent optimization heuristic algorithms for large-scale problems are also designed.