Digitization of chemical process flow diagrams using deep convolutional neural networks
Advances in deep convolutional neural networks led to breakthroughs in many computer vision applications. In chemical engineering, a number of tools have been developed for the digitization of Process and Instrumentation Diagrams. However, there is no framework for the digitization of process flow d...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2023-03-01
|
| Series: | Digital Chemical Engineering |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2772508122000631 |
| _version_ | 1811159878404669440 |
|---|---|
| author | Maximilian F. Theisen Kenji Nishizaki Flores Lukas Schulze Balhorn Artur M. Schweidtmann |
| author_facet | Maximilian F. Theisen Kenji Nishizaki Flores Lukas Schulze Balhorn Artur M. Schweidtmann |
| author_sort | Maximilian F. Theisen |
| collection | DOAJ |
| description | Advances in deep convolutional neural networks led to breakthroughs in many computer vision applications. In chemical engineering, a number of tools have been developed for the digitization of Process and Instrumentation Diagrams. However, there is no framework for the digitization of process flow diagrams (PFDs). PFDs are difficult to digitize because of the large variability in the data, e.g., there are multiple ways to depict unit operations in PFDs. We propose a two-step framework for digitizing PFDs: (i) unit operations are detected using a deep learning powered object detection model, (ii) the connectivities between unit operations are detected using a pixel-based search algorithm. To ensure robustness, we collect and label over 1000 PFDs from diversified sources including various scientific journals and books. To cope with the high intra-class variability in the data, we define 47 distinct classes that account for different drawing styles of unit operations. Our algorithm delivers accurate and robust results on an independent test set. We report promising results for line and unit operation detection with an Average Precision at 50 percent (AP50) of 88% and an Average Precision (AP) of 68% for the detection of unit operations. |
| first_indexed | 2024-04-10T05:48:10Z |
| format | Article |
| id | doaj.art-04af0364fe854b2aa5e58820a9eac9ff |
| institution | Directory Open Access Journal |
| issn | 2772-5081 |
| language | English |
| last_indexed | 2024-04-10T05:48:10Z |
| publishDate | 2023-03-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Digital Chemical Engineering |
| spelling | doaj.art-04af0364fe854b2aa5e58820a9eac9ff2023-03-05T04:26:09ZengElsevierDigital Chemical Engineering2772-50812023-03-016100072Digitization of chemical process flow diagrams using deep convolutional neural networksMaximilian F. Theisen0Kenji Nishizaki Flores1Lukas Schulze Balhorn2Artur M. Schweidtmann3Delft University of Technology, Department of Chemical Engineering, Van der Maasweg 9, Delft, 2629 HZ, NetherlandsDelft University of Technology, Department of Chemical Engineering, Van der Maasweg 9, Delft, 2629 HZ, NetherlandsDelft University of Technology, Department of Chemical Engineering, Van der Maasweg 9, Delft, 2629 HZ, NetherlandsCorresponding author.; Delft University of Technology, Department of Chemical Engineering, Van der Maasweg 9, Delft, 2629 HZ, NetherlandsAdvances in deep convolutional neural networks led to breakthroughs in many computer vision applications. In chemical engineering, a number of tools have been developed for the digitization of Process and Instrumentation Diagrams. However, there is no framework for the digitization of process flow diagrams (PFDs). PFDs are difficult to digitize because of the large variability in the data, e.g., there are multiple ways to depict unit operations in PFDs. We propose a two-step framework for digitizing PFDs: (i) unit operations are detected using a deep learning powered object detection model, (ii) the connectivities between unit operations are detected using a pixel-based search algorithm. To ensure robustness, we collect and label over 1000 PFDs from diversified sources including various scientific journals and books. To cope with the high intra-class variability in the data, we define 47 distinct classes that account for different drawing styles of unit operations. Our algorithm delivers accurate and robust results on an independent test set. We report promising results for line and unit operation detection with an Average Precision at 50 percent (AP50) of 88% and an Average Precision (AP) of 68% for the detection of unit operations.http://www.sciencedirect.com/science/article/pii/S2772508122000631Process flow diagrams (PFD)Flowsheet digitizationObject detectionDigitalizationMachine learningDeep convolutional neural network |
| spellingShingle | Maximilian F. Theisen Kenji Nishizaki Flores Lukas Schulze Balhorn Artur M. Schweidtmann Digitization of chemical process flow diagrams using deep convolutional neural networks Digital Chemical Engineering Process flow diagrams (PFD) Flowsheet digitization Object detection Digitalization Machine learning Deep convolutional neural network |
| title | Digitization of chemical process flow diagrams using deep convolutional neural networks |
| title_full | Digitization of chemical process flow diagrams using deep convolutional neural networks |
| title_fullStr | Digitization of chemical process flow diagrams using deep convolutional neural networks |
| title_full_unstemmed | Digitization of chemical process flow diagrams using deep convolutional neural networks |
| title_short | Digitization of chemical process flow diagrams using deep convolutional neural networks |
| title_sort | digitization of chemical process flow diagrams using deep convolutional neural networks |
| topic | Process flow diagrams (PFD) Flowsheet digitization Object detection Digitalization Machine learning Deep convolutional neural network |
| url | http://www.sciencedirect.com/science/article/pii/S2772508122000631 |
| work_keys_str_mv | AT maximilianftheisen digitizationofchemicalprocessflowdiagramsusingdeepconvolutionalneuralnetworks AT kenjinishizakiflores digitizationofchemicalprocessflowdiagramsusingdeepconvolutionalneuralnetworks AT lukasschulzebalhorn digitizationofchemicalprocessflowdiagramsusingdeepconvolutionalneuralnetworks AT arturmschweidtmann digitizationofchemicalprocessflowdiagramsusingdeepconvolutionalneuralnetworks |