Autocorrection of engineering diagrams

Automatic error correction is standard for text documents. We develop generative AI methods that detect errors in engineering diagrams such as P&IDs and process flow diagrams (PFDs) and suggest corrections.

Challenge

Process engineering relies on PFDs and P&IDs to represent process flows and equipment configurations. These diagrams can contain errors that cause safety hazards, inefficient operation, and unnecessary costs. Checking and correcting them is a tedious, manual process.

Our approach

Inspired by the success of large language models (LLMs) for grammatical error correction of human language, we investigate generative models for the autocorrection of flowsheets. The input to the model is a potentially erroneous flowsheet, and the output is a suggestion for a corrected flowsheet. In a first study, an autocorrection model trained in a supervised manner on synthetic flowsheets reached a top-1 accuracy of 80% and a top-5 accuracy of 84% on an independent test set. Follow-up work uses a graph-based generative AI methodology to autocorrect P&IDs of utility systems.

From research to practice

  • Automated error correction of engineering documents is one of the further developments of the Digitization Companion (DigiCo) prototype software for Smart P&IDs.

Key publications

  • Schulze Balhorn, L., Goldstein, D. P., Seijsener, N., Dao, K., Driessen, G. H. M., & Schweidtmann, A. M. (2026). A graph-based generative artificial intelligence methodology for autocorrection of utility-system P&IDs. Chemie Ingenieur Technik, e70135. doi:10.1002/cite.70135
  • Schulze Balhorn, L., Caballero, M., & Schweidtmann, A. M. (2024). Toward autocorrection of chemical process flowsheets using large language models. Computer Aided Chemical Engineering, 53, 3109–3114. doi:10.1016/B978-0-443-28824-1.50519-6 · Details
  • Vogel, G., Hirtreiter, E., Schulze Balhorn, L., & Schweidtmann, A. M. (2023). SFILES 2.0: An extended text-based flowsheet representation. Optimization and Engineering, 24, 2911–2933. doi:10.1007/s11081-023-09798-9 · Details
← All projects