Probabilistic realtime failure detection for safety enforcement workflows

Abstract

It is possible to streamline Environmental, Health and Safety (EH&S) duties through the use of workflow management technology. This approach requires to specify workflow models which among others consist of activities. In order to meet given safety regulations these activities are to be completed correctly and within given deadlines. Otherwise, activity failures emerge which may lead to breaches against safety regulations. A novel domain-specific workflow meta data model is proposed. The model enables a system to detect and predict activity failures through the use of data about the company, failure statistics, and activity proxies. Since the detection and prediction methods are based on the evaluation of constraints specified on EH&S regulations a system approach is proposed that builds on the integration of a Workflow Management System (WFMS) with an EH&S Compliance Information System. Main principles of the failure detection and prediction are described. For EH&S managers the system shall provide insights into the current failure situation. This can help to prevent and mitigate critical situations such as safety enforcement measures that are behind their deadlines.