Technological evolution, heading for industry 4.0, makes companies tend to automate their management and operation, ideally defining it through business process models. To describe policies or rules related to the execution order of the activities in an organization, Declarative Business Process Models permit a relaxed description of activity order, which needs monitoring to detect non-conforming behaviors. Commonly, the detection of a violation implies that the malfunction has already occurred, being better to avoid the violation in advance. To predict future violations, prognosis is required.

To allow the modeling of real business behavior, an extension of declarative business process models including both time patterns and multiple instances is proposed. This new model can be used to prognosticate if current process instances may violate a defined model in the future, according to the analysis of the robustness of the process instances evolution. The proposed Model-Based Prognosis is based on analyzing the event traces that represent the current instances and propagate their possible progression through the Constraint Programming paradigm. To ascertain if the model could be violated, it is analyzed how its robustness can tackle unexpected behaviors.

To complete the formalization and modeling, an implementation applied to a real medical example is included in the paper. The prognosis of concurrent instances is addressed, dealing with formalized time and activity patterns even considering the resource availability, and getting acceptable execution times.

The automatic verification and prognosis of declarative business processes are addressed considering concurrency and synchronization of multiple instances, performing well in terms of execution time.

走向工业4.0的技术演进使公司倾向于自动化其管理和操作，理想情况下是通过业务流程模型对其进行定义。为了描述与组织中的活动的执行顺序相关的策略或规则，声明式业务流程模型允许对活动顺序进行宽松的描述，需要对其进行监视以检测不符合要求的行为。通常，检测到违规意味着故障已经发生，最好事先避免违规。为了预测将来的违规行为，需要进行预后。

为了允许对真实的业务行为进行建模，提出了对声明性业务流程模型的扩展，包括时间模式和多个实例。根据对流程实例演化的鲁棒性的分析，该新模型可用于预测当前流程实例将来是否会违反定义的模型。所提出的基于模型的预测基于对表示当前实例的事件跟踪进行分析，并通过约束编程范例传播其可能的进程。为了确定是否可能违反该模型，分析了模型的鲁棒性如何解决意外行为。

为了完成形式化和建模，本文包含了应用于实际医学示例的实现。解决了并发实例的预后，甚至考虑了资源可用性来处理形式化的时间和活动模式，并获得了可接受的执行时间。

考虑到多个实例的并发性和同步性，解决了声明性业务流程的自动验证和预后问题，在执行时间方面表现良好。