Browsers provide a structured view of model data, which you can use to review, modify, create, and manage
the contents of a model. In addition to visualization, browsers offer features like search, filtering, and sorting,
which enhance your ability to navigate and interact with the model data.
FE geometry is topology on top of mesh, meaning CAD and mesh exist as a single entity. The purpose of FE geometry
is to add vertices, edges, surfaces, and solids on FE models which have no CAD geometry.
Part label information must be provided to successfully train a machine learning model. Use the Classifier: Label tool to create classifiers and labels, which can be assigned to parts later.
Manage all of the IDs for the entities that you create, and define ID ranges for all of the entities in each Include
file in relation to the full model in order to avoid ID duplication.
Perform automatic checks on CAD models, and identify potential issues with geometry that may slow down the meshing
process using the Verification and Comparison tools.
Local coordinate systems can be used for setting up loads/boundary conditions that do not act in the global axis direction,
transforming results, defining material orientation, and many other operations.
Use the Auto Contacts tool to determine contact interfaces between selections of components or elements. Based on
the user-specified options like proximity tolerance, surface creation method, main surface type, and secondary type,
the tool generates contacts based on set segments or node and element combinations.
Tools and workflows that are dedicated to rapidly creating new parts for specific use cases, or amending existing
parts. The current capabilities are focused on stiffening parts.
Use PhysicsAI to build fast predictive models from CAE data. PhysicsAI can be trained on data with any physics or
remeshing and without design variables.
Explore, organize and manage your personal data, collaborate in teams, and connect to other data sources, such as
corporate PLM systems to access CAD data or publish simulation data.
Train a machine learning model based on labeling information it contains.
Successfully training a machine learning model requires at least two labels to be
defined within the selected classifier. All labels must be assigned at least three
parts to train the model.
Tip: When training fails
due to one or both of these reasons, use the messages displayed in the status
bar to provide the missing label information and proceed with the model
training.
From the Assembly ribbon, click the Classify > Train tool.
From the guide bar, select the
Classifier drop-down menu to select an existing
classifier to train.
Click Train.
After training a machine learning model, a binary file
(<classifier_name>_<train_iteration>.aic)
containing all training information is saved in the Classifier folder. If you
add new labels to a classifier or you modify existing label assignments, retrain
the machine learning model to apply the changes.
Note: The
value of <train_iteration> increments by one after
every successful training operation. The number of train iterations starts
at 0 and the <train_iteration> suffix is appended to
the file names only when train_iteration > 0.
You can
use the resultant binary (.aic) files from model
training on different machines. As a result, an advanced user can create
the classifiers and train the machine learning models to provide files
for other users to perform part classification, eliminating the need to
train a classifier (See Add Trained Model File).
Tip: From the guide bar, click and select Generate Log
File, if necessary. The log files
(<classifier_name>_<train_iteration>_log.txt)
are saved to the Classifier folder after training a machine learning
model.
After the classifier has been trained, the confusion matrix
is displayed. The confusion matrix displays the accuracy of the trained machine learning
model versus the labels that you provided. The confusion matrix is displayed by default.
Click to toggle the matrix on/off.
Each row shows the label name and the total number of parts given that label by
the user. This sum is displayed under Support.
Each column shows the label
predicted by the machine learning model.
Precision
The ratio of the number of correctly predicted labels of a given type to
the total number of predictions of that same label. Provides insight
into the ability of the machine learning model to predict the label
correctly.
Support
The total number of parts with each label (user defined).
Overall Accuracy
The total ratio of correct predictions to the total number of
predictions.
The Floor row
displays a "4" under the Floor column and "1" under the Lights column. This means
there were five Floor labels given to parts by the user, as confirmed by the total
in the Support column.
A "4" in the Floor column indicates that the machine
learning model predicted the Floor label to have four parts. The precision is 1
because each time it predicted that a part was a Floor, it was correct.
For
the Lights row, the number "7" in the Support column indicates that there were seven
assigned Light labels to parts. However, the Lights column display "1" under Floor
and "7" under Lights. This means the machine learning model predicted eight parts to
be lights; seven times it was correct, but once it predicted a part to be a light
that was assigned the label Floor. Therefore, the precision of predicting Lights is
7/8, or 0.88.
The overall accuracy value is calculated as the ratio of the sum
of the confusion matrix’s diagonal entries to the column sum of the support column,
therefore:(1)
Note: Training results can be affected by
the order that parts are labeled. This effect is much less pronounced as the
number of parts assigned to each label increases.