AI for Production Data – Automatically Recognizing Relationships
When the volume of data becomes too large for manual analysis.
A production problem can be influenced by many machine values. Temperature, pressure, speed, machine states, and other process variables change simultaneously – and relevant relationships are often difficult to discern with the naked eye.
Machine Designer utilizes Machine Learning to systematically analyze recorded production data and provide insights into relevant influencing factors and patterns.
You define the production problem. Machine Designer analyzes the data.
You know your production. You know which event needs to be investigated – for example, a downtime, a quality deviation, or a specific machine condition.
Machine Designer takes over the data-driven investigation of the recorded machine values and assesses which information might be particularly relevant for the event in question.
You do not need to program a machine learning model or be a data scientist to do this.
1. Set Event
Define what should be examined – for example, a downtime or the achievement of a specific process state.
2. Analyze Production Data
Machine Designer generates suitable features from the recorded machine data and trains a model for the selected question.
3. Evaluate Results
You will receive insights into which machine values and patterns are related to the event being analyzed and how significant the model is.
Production knowledge meets Machine Learning
AI does not know your production. However, it can recognize patterns in data that are easily overlooked in a manual analysis.
The production manager provides process knowledge and questions. Machine Designer takes on a significant part of the technical data analysis. It is only the combination of production knowledge and data-driven analysis that makes the results valuable.
What can be examined?
Downtime
What changes regularly occur before a shutdown?
Quality deviations
What process variables distinguish good production periods from notable ones?
Process States
What machine values are associated with a specific condition?
Unusual Behavior
Which combinations of process values differ from normal production behavior?
AI supports decision-making – it does not replace process knowledge.
The results of a machine learning model are indications of statistical correlations. They should be evaluated in conjunction with knowledge about the machine and the production process. Machine Designer is intended to assist production managers in quickly arriving at the right questions and potential influencing factors – not to hand over decisions unexamined to an AI.