AIT-400 Artificial Intelligence Industrial Trainer

Create | Train | Apply

CREATE: Use pre-existing AI models or develop your own.

TRAIN: Fit the model with data from previous operations to improve AI performance.

APPLY: Deploy the trained AI models to solve the industrial problems, testing the model accuracy and generating more data.

AI software suite

A comprehensive software platform for building, training, and deploying AI models—either through a no-code interface or using Python in descriptive Jupyter notebooks.

Core applications and functions

⇢ Artificial vision – Identify objects.

⇢ Movement optimization – Predict the acceleration of an electric actuator.

⇢ Outcome forecast – Score the success probability in a process.

⇢ Predictive maintenance – Anomaly detection in the system performance.

⇢ Quality control validation – Classify between valid/faulty objects.

AIT-400 Process

The problems to be solved with AI revolve around optimising the movement of an electric actuator: What is the maximum acceleration that can be applied to the load without falling and being damaged?

-To do this, the type of load must be recognised, so the AI will make a prediction based on a photo of the tower (1).

Then, there are two types of problems related to acceleration:

-Regression: Predicting the appropriate acceleration given a tower of pieces (2).

-Classification: Predicting whether or not the tower will fall, given a tower of pieces and an acceleration (3).

Finally:

-After launching the actuator, another photo is taken at the final position, and it is evaluated whether the test was valid or not. In other words, it is evaluated whether the tower arrived in the correct position or not (4).

-The speed and force data from the electric shaft are analysed to detect anomalies in the movement of the actuator that may indicate a need for predictive maintenance (5).