University of Pretoria
Browse
IMAGE
FIG1.png (59.1 kB)
IMAGE
FIG2.png (174.3 kB)
DOCUMENT
FIG3.pdf (22.02 kB)
IMAGE
FIG4.png (111.61 kB)
IMAGE
FIG5.png (86.17 kB)
IMAGE
FIG6.png (42.56 kB)
IMAGE
FIG7.png (68.29 kB)
IMAGE
FIG8.png (77.88 kB)
IMAGE
FIG9.png (65.68 kB)
IMAGE
FIG10.PNG (313.88 kB)
IMAGE
FIG11.PNG (199.47 kB)
IMAGE
FIG12.PNG (9.2 kB)
IMAGE
FIG13.PNG (47.58 kB)
IMAGE
FIG14.png (54.26 kB)
IMAGE
FIG15.png (34.51 kB)
IMAGE
FIG16.png (42.09 kB)
IMAGE
FIG17.png (20.02 kB)
IMAGE
FIG18.png (58.76 kB)
IMAGE
FIG19.png (12.36 kB)
IMAGE
FIG20.png (12.35 kB)
1/0
80 files

The performance of surrogate models, data assimilation and hybrid models when applied to the water-cooled freeboard of a process converter.

dataset
posted on 2023-03-07, 10:00 authored by Karl HellbergKarl Hellberg
  • This dataset contains the figures and tables underpinning a dissertation on incorporating sensor measurements using data assimilation and machine learning to improve the accuracy of thermal finite element methods, as well as the data that these figures and tables portray. The filenames of the figures and tables correspond to the numbering applied to them in the dissertation. The data portrayed by these figures and tables are provided by means of the following *.mat files:surrogate_cs1.mat contains the performance of different types of surrogate models of a simplified model of a process converter freeboard, on which Tables 8 and 9 are based.
  • data_assimilation_cs1.mat contains the performance of different data assimilation algorithms applied to a simplified model of a process converter freeboard, on which Figures 14-17 and 41-60 are based.
  • hybrid_cs1.mat contains the performance of hybrid models of a simplified model of a process converter freeboard, on which Figures 19, 20 and 21-23, as well as Table 13 are based.
  • surrogate_cs2.mat contains the performance of different types of surrogate model of a half model of a process converter freeboard, on which Table 15 is based.
  • data_assimilation_simulated_cs2.mat contains the performance of different data assimilation algorithms applied using simulated measurements of a process converter freeboard, on which Figures 27-29 and Table 17 are based.
  • hybrid_simulated_cs2.mat contains the performance of hybrid models trained using simulated measurements of a process converter freeboard, on which Figures 30-33 and Table 18 are based.
  • data_assimilation_real_cs2.mat contains the performance of different data assimilation parameters for an application using real measurements of a process converter freeboard, on which Figure 35 is based.
  • hybrid_real_cs2.mat contains the performance of hybrid models trained using real measurements of a process converter freeboard, on which Figures 37-40 and Table 20 are based.

Each *.mat file (a Matlab file format which can be used within other software environments like Octave and Python ) contains a variable called ‘description’ that provides a short description of each of the other variables contained in the file.

Funding

Centre for Asset Integrity Management bursary, University of Pretoria Masters Research bursary

History

Department/Unit

Department of Mechanical and Aeronautical Engineering

Usage metrics

    Engineering, Built Environment and Information Technology

    Licence

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC