关键词: 产品开发, 过程管理, 顺序迭代, 过程模型, 奖赏马尔科夫链

Abstract: To effectively manage and improve the product development processes, a reward-Markov-chain-based quantitative model was proposed for sequential iterative processes. This model was different from existing sequential iterative process models in several aspects. The influencing factors for rework in the model not only influenced the duration of the rework task, but also influenced the durations of the tasks following the rework task, i.e. it influneced the remaining time of the rework task. The rework amount of a task decreased as the increase of iteration times at each stage. The model differentiated the task rework amount caused by different tasks. To estimate the expectant development cycle, an analysis model and a simulation model were established for sequential iterative process estimation, and a simulation algorithm was also presented. The process time estimation models were illustrated by a software testing process. Analysis and comparisons between the proposed models and other existing models were presented according to the experiment results. Finally, the possible causes for deviation from the practical process time of estimated result were discussed.

Key words: product development, process management, sequential iteration, process model, reward Markov chain, design structure matrix