考虑认知动态变化的多技能项目调度模型

Multi-Skill Project Scheduling Model Considering Cognitive Dynamic Change

摘要: 许多互联网企业并未对人力资源实施精密细致的规划管理，在开发复杂软件项目时面临着人才调度困难，开发效率低下的困境，选择高效的多技能员工调度策略成为现代软件项目管理的新趋势。现有研究较少对员工的认知能力建模且未考虑企业人才培养战略目标的实现。因此，本文针对员工认知能力动态变化的多技能项目调度问题，以项目工期、项目成本以及技能增长效益为目标建立了考虑变动工作效率的优化模型，基于学习–遗忘曲线理论(LFCM)设计了工作效率随时间动态变化的计算方法，并提出一种考虑工作效率的新型工资分配模式。设计了改进的NSGA-III求解模型，算法引入了包含优先级和员工配置属性的多维染色体编码，使用一种新型进度生成策略(SSGS-SMTS)获得相应编码下调度方案的目标函数值，对不同维度的染色体分别采用特殊的交叉和变异操作以保证解的合理性和多样性。最后，结合工程实例进行验证，表明了所提模型及算法的有效性。

Abstract: Most internet companies have not implemented precise and meticulous planning and scheduling of human resources, and are facing the dilemma of difficult talent scheduling and low development efficiency when implementing complex software projects. Choosing an efficient multi-skilled staff scheduling strategy has become a new trend in modern software project management. Existing research seldom models the cognitive ability of staff scientifically and often ignores the realization of corporate strategic objectives in talent cultivation. Hence, aiming at the multi-skill project scheduling problems in which the cognitive ability of staff changes dynamically, an optimization model considering variable perform efficiency with the objectives of project duration, project cost and skill growth benefit is established. Based on the LFCM (Learning and Forgetting Curve Model) theory, the calculation method of perform efficiency which changes with time dynamically is designed. The improved NSGA-III is adopted to solve the model. The algorithm introduces a multi-dimensional chromosome coding which includes the attributes of priority and staff allocation with the SSGS-SMTS (Serial Schedule Generation Scheme Combining with Semaphore Mechanism and Topological Sorting) to obtain the values of objective function under corresponding scheduling scheme. Special crossover and mutation are adopted for chromosomes of different dimensions in order to ensure the rationality and diversity of solutions. Finally, the excellence of the proposed model and algorithm is verified through a project example.