Modernizing CAD-Based Infrastructure Design Workflows Through Internal Automation, Data Standards, and Process Modelling
To improve the current practices in modernizing CAD workflows within infrastructure projects, there is a need for transitioning from traditional practices characterized by inefficiencies towards intelligent automation and standardization of processes. In this regard, a three-stage framework consisting of intelligent dependency graph generation, optimization of design processes and automation of tasks within processes is proposed in this study. The first stage consists of the creation of the intelligent dependency graph utilizing knowledge graph transformers, which use graph neural networks along with transformer attention. The knowledge graph transformer learns the relationship between various CAD entities, layers, design standards and components of infrastructure to generate an intelligent dependency graph. Subsequently, the Process Mining Evolution Engine makes use of process discovery algorithms to detect inefficiencies, repetitious engineering activities, and approval processes to come up with an optimized workflow. Lastly, the Reinforcement Automation Orchestrator uses reinforcement learning to suggest the appropriate automation, task ordering, and validation timing of tasks within the workflow. As a result, the framework transforms CAD models, design standards, and process records into actionable recommendations. Consequently, the generated optimized infrastructure design workflow has an accuracy of 97.4%.