Misfit_data_set
Contents
Data Structure: Misfit_data_set |
|
Description |
Data defining the Misfit Evaluation |
Usage |
Misfit_data_set NUM=ival where ival is the data structure number |
Description |
Overview The Misfit_data_set data structure is used to define the history result files containing the target and model solutions, together with the variable and weighting data to be used in computing the misfit between the experimental data and model results.
The minimum data requirement is the name of the target data file and the associated model history set number. In this case the misfit is computed using a normalised root mean square error for all common variables (with the same name) between the target and the model history result files while non-common variables will be ignored. Alternatively, specific variable names in the target may be defined to be associated to specific model variable names for comparison and computation of misfit (e.g. this may be required when the variable name in the target result file and model history file In addition individual variables or specific time intervals may be assigned weighting. This allows for example the misfit evaluation to be focused on a particular section of the results in a history file (e.g. a particular time range).
Both the experimental data (target) and model results are defined in standard ParaGeo format in comma separated value (csv) format. The target data should contain as a minimum a Header section with the variable IDs and variable tags and a section with the values for each variable according to the following format:
Where the words in bold blue need to be specified exactly as shown and the two first columns of values are always considered Step and Time values respectively. It is expected that the model history file will contain additional data than the minimum shown above.
Notes •A Misfit_data_set data structure must be defined for each target result data file to be used in the inverse procedure.
|
 | Experiment_filename Target results file name (Compulsory) |
Usage |
||||
|
||||
Description |
||||
The Experiment_filename is compulsory and defines the name of the history file that contains the target results. The Experiment_filename should be located in the target directory defined within File_data data structure.
Notes •When running an inverse analysis from MATINA the default Experiment_filename name is set as project_name_SetNo.hdh , where SetNo is the set number associated with the Misfit_data_set data structure.
|
| Model_set_number Model results history set (Compulsory) |
Usage |
||||
|
||||
Description |
||||
The Model_set_number is compulsory and defines the number of the history set generated from each model simulation that is associated to the present Misfit_data_set. Hence Model_set_number in conjunction with Experiment_filename define which files should be compared between the target and model results.
Notes •When running an inverse analysis from MATINA the default Model_set_number number is SetNo , where SetNo is the set number associated with the Misfit_data_set data structure.
|
| Error_type Type of function for computing misfit |
Usage |
||||
|
||||
Description |
||||
The Error_type keyword is used to define the method for computing the misfit between the target (experimental) and model results. Valid options are: •"RMS" or "Root_mean_square" - Root mean square of the differences normalised by the root mean square of the target values. •"NAV" or "Normalised_average" - Average of the absolute differences normalised by the corresponding individual target values.
|
| Experiment_variable_IDs List of variables to be used in the Misfit evaluation |
Usage |
|||||||||||||||
|
|||||||||||||||
Description |
|||||||||||||||
Experiment_variable_IDs defines a list of variable names in the the target (or experimental) history file that should be used in the misfit calculation; i.e. it allows exclusion of variables that are present in the file but should not be taken into account or are not important. By default the target (experimental) and model history variable names are considered to be identical. If the variable names in the target and model history files differ then the list of equivalent variable names in the model history file is defined using Model_variable_IDs. If Experiment_variable_IDs is not defined then all common variables in the target and model history file will be used in the misfit evaluation (all non-common variables will be ignored).
|
|||||||||||||||
| Experiment_variable_tags List of tags for target variables |
Usage |
|||||||||||||||
|
|||||||||||||||
Description |
|||||||||||||||
Defines a list of variable tags for each variable name listed in Experiment_variable_IDs. Variable tags are optional and can be used to identify a particular column of data when the primary variable ID occurs more than once in the target/experimental data set; e.g. for axial strain values at multiple locations the specific location would be identified with the tag.
|
|||||||||||||||
| Model_variable_IDs List of variables to be used in the Misfit evaluation |
Usage |
|||||||||||||||
|
|||||||||||||||
Description |
|||||||||||||||
The Model_variable_IDs keyword is used to define the list of model variable names in the model history file associated to the target variable names listed in Experiment_variable_IDs . Model_variable_IDs is not required if the target variables and model variables names are identical.
|
|||||||||||||||
| Model_variable_tags List of tags for the model variables |
Usage |
|||||||||||||||
|
|||||||||||||||
Description |
|||||||||||||||
Defines a list of variable tags for each variable name listed in Model_variable_IDs. Variable tags are optional and can be used to identify a particular column of data when the primary variable ID occurs more than once in the model history set; i.e. when the history data set has more than one point.
|
|||||||||||||||
| Variable_weights List of weights for each variable misfit |
Usage |
|||||||||||||||
|
|||||||||||||||
Description |
|||||||||||||||
Variable_weights keyword allows to list and assign weights to each variable listed in Experiment_variable_IDs for misfit calculation purposes. These weights are then used when evaluating the total misfit which is defined as the weighted average of the individual variables misfits. If Variable_weights is not defined all variables are assigned unit weighting.
Notes •Experiment_variable_IDs must be defined if Variable_weights is defined.
|
|||||||||||||||
| Time_curve_weights List of time ranges and corresponding weights |
Usage |
|||||||||||||||
|
|||||||||||||||
Description |
|||||||||||||||
The Time_curve_weights allows to assign different weights to different time ranges from the history results via a start time, end time and weight for misfit calculation purposes. Several time ranges may be specified. If time ranges overlap then the largest weighting in the overlap zone will be used. Time_curve_weights applies to all variables in the misfit evaluation. The specific data for each time range weight is: •Location 1 - Start time for the range •Location 2 - End time for the range •Location 3 - Weight
Notes •The weights are defined relative to the default weight of 1.0; i.e. to increase the weighting within a time range requires, weight > 1.0. Conversely to decrease the weighting for a time range, weight < 1.0. •Time_curve_weights may be used for example to exclude the results at initial times for misfit calculation purposes by assigning a weight of 0 to the time range to be excluded.
|
|||||||||||||||
| Set_weigth Weight for the current misfit data set |
Usage |
||||
|
||||
Description |
||||
Weight assigned to the current misfit data set.
|