INNOVATIVE APPROACHES TO ADDRESS SCALABILITY CHALLENGES IN WORKFLOW AUTOMATION FOR LARGE-SCALE ENTERPRISE BUSINESS PROCESS MANAGEMENT SYSTEMS

Abstract

The paper describes a workflow orchestration framework that supports the automation of the incident management process in enterprise IT using scalable and AI-powered workflow orchestration. With the help of a real-world ITSM event log set of data made available by Kaggle and containing more than 240,000 entries, we will introduce a predictive mechanism to forecast the number of tickets coming and proactively make simulated autoscaling choices. Due to intensive data preparation, event trace generation, and discovering patterns, we can identify temporal patterns in workflow activities and focus incident processing on volume versus priority. A specific model, Facebook Prophet, is applied to predict daily ticket inflow, and autoscaling logic is prompted based on the prediction of predicted thresholds reaching 400 events/day. The simulation justifies using the model to facilitate the redistribution of resources before the busy season. Important performance tools, such as average resolution time, the number of escalations, and satisfaction ratings averaged by the impact level, are scrutinised extensively. The proposed framework is more effective than the usual static orchestration systems because it presents the idea of forecasting-based provisioning, which provides both the enforcement of the SLA and optimisation of operational costs. Another research gap that is closed in our study is that we integrate process mining and predictive autoscaling, in contrast with reactive triggers. The contribution of this work to the literature is a practical, modular and enterprise-capable solution of intelligent BP M orchestration. This model can also include LSTM and Transformer-based architectures in real-time streaming systems.