DHGE and DMG MORI Develop AI Applications for Predicting Machine Tool Failure Risks
DHGE Plans to Expand Use of MATLAB in Future Projects
“MATLAB simplified our prototyping process and required no intermediate steps or data transfers, which was a benefit compared to Python.”
Key Outcomes
- Using MATLAB apps accelerated data processing tasks and improved time between prototyping and deployment
- Statistics and Machine Learning Toolbox enabled researchers to predict machine tool failure risks
- MATLAB apps were developed and deployed for predictive maintenance use in industrial contexts
DHGE developed a prediction model to identify potential field failures by analyzing the highly nonlinear and complex relationships between system configurations and failure risks.
Gera-Eisenach Corporate University (DHGE) is a technical and applied sciences institution that collaborates with industry to find efficient and innovative solutions to persistent challenges, such as predicting the risk of machine tool failures. The university has collaborated with international machining company DMG MORI Seebach and used MATLAB® tools to develop and test apps designed to use AI to predict when maintenance is needed in industrial machines, helping prevent breakdowns and improve efficiency.
To create an intuitive app for end users, DHGE first used MATLAB tools for feature engineering. This looked like standardizing and processing key data from different sources besides designing experiment methods to identify features. Next, by using code developed in MATLAB App Designer and compiled using MATLAB Compiler™, the researchers processed data into a common vector that could be used to train a shallow neural network. This model was created using neural network functions from Statistics and Machine Learning Toolbox™. Using MATLAB tools helped ease the collaboration between DHGE and DMG MORI by providing a seamless way to implement the model.
DHGE was looking for a solution to efficiently process data into a common format for predictive maintenance models. By using MATLAB tools exclusively to develop the predictive model, researchers benefited from a smooth iterative design process as well as data consistency, safety, and security. The development process also required fewer intermediate steps and data transfers compared to similar work done using Python®, which reduced time between prototyping and deployment. Using these tools, DHGE researchers created an app for DMG MORI to implement that created a more robust and efficient process for the production and configuration of tool machines.
DHGE researchers plan to continue using MATLAB tools—as well as MathWorks support—in future projects, which include evaluating the function of this predictive maintenance app as well as developing a new MATLAB toolbox to support machine data imports to meet a growing need in the AI landscape.
