Abstract Among the biggest challenges in manufacturing, we can mention the difficulty of locating points that most impact performance; and predict the time required to perform certain tasks. This paper presents a new approach for completion time prediction and performance analysis in manufacturing, considering the individual behavior of process activities. The framework addresses process mining techniques to support the development of a probabilistic model in Bayesian Networks and predictive models. The integration of these models makes it possible to calculate the occurrence probability of all activities and future behavior of the process. Two case studies are carried out to validate the framework, compared against FSM Analyzer. Results show estimates with good accuracy, helping managers in their action plans.

中文翻译：

---

制造系统中的性能分析和时间预测

摘要 在制造业面临的最大挑战中，我们可以提到定位对性能影响最大的点的难度；并预测执行某些任务所需的时间。考虑到过程活动的个体行为，本文提出了一种新的制造完成时间预测和性能分析方法。该框架解决了过程挖掘技术，以支持贝叶斯网络和预测模型中概率模型的开发。这些模型的集成使得计算过程的所有活动和未来行为的发生概率成为可能。进行了两个案例研究以验证该框架，并与 FSM Analyzer 进行比较。结果显示的估算非常准确，有助于管理人员制定行动计划。