GMV PitIA®
gmv
GMV PitIA® is an Artificial Intelligence tool for monitoring, predicting, and optimizing complex systems, capable of detecting anomalies, predicting critical variables, and optimizing processes from operational data. It is adaptable to various domains, including space, where it is used for satellite
Technical specifications
- Application demonstrated in domains with
- high operational demand and large volumes of data
- Anomaly detection
- without the need for labeled historical failure data
- Multiple variables
- one operational view (converts complex operational data into actionable indicators, alerts, and explanations)
- Adaptable to
- multiple domains (industrial processes to space systems)
- Based on
- real operational data (models complex systems from historical and operational data without altering the process)
- Explainability of results
- helps identify causes of anomalous behaviors and deviations
- Designed to support
- decision-making (intuitive interfaces and visualizations)
- Deployment
- flexible and scalable (compatible with cloud and on-premise, adaptable to integration needs)
- Architecture
- modular and configurable (based on independent components)
- Space Use Case
- Anomaly detection in satellite telemetry
- Validation
- ESA Anomaly Detection Benchmark (based on real and annotated ESA mission data)
- Performance
- Solid balance between model performance, scalability, usability, and robustness compared to Global STD, iForest, KNN, or Telemanom ESA
- Suitable for
- high telemetry volumes, limited labeled data, and explainable alerts
- Benefits (Space)
- Detection without labeled data, scalability in complex systems, explainable alerts, ready for operation (low computational requirements, easy integration)
- Industrial Use Case
- Prediction of critical variables in industrial processes
- Benefits (Industry)
- Reduced analyzer usage, continuous prediction, updated model, increased plant availability, more proactive decisions
About
GMV PitIA® is GMV’s Artificial Intelligence tool designed for the monitoring, prediction, and optimization of complex systems. It excels at detecting anomalies, predicting critical variables, and optimizing processes by leveraging operational data. The system’s ability to model system behavior and anticipate deviations leads to more efficient, secure, and reliable operations. Its flexible approach allows GMV PitIA® to be adapted to diverse domains and use cases, ranging from industrial applications to space systems. It transforms raw data into actionable knowledge, aiming to improve system performance, reduce operational costs, and support decision-making.
Key features include its applicability to multiple domains, including industrial processes and space systems, and its reliance on real operational data to model complex systems without requiring process alteration. It offers explainability of results, helping to identify the causes of anomalous behaviors and deviations from normal operating conditions. Designed to support decision-making, it provides intuitive interfaces and visualizations for operators and analysts. The deployment is flexible and scalable, compatible with both cloud and on-premise environments, and adaptable to various integration needs. Its modular and configurable architecture, based on independent components, facilitates adaptation to different use cases, incorporation of new functionalities, and integration with existing data sources and systems.
In the space domain, GMV PitIA® addresses the challenge of detecting anomalies in large volumes of continuous satellite telemetry. It learns the normal behavior of multivariate systems and identifies relevant deviations, generating alerts and explaining the origin of anomalies by indicating contributing parameters. The solution was validated against the ESA Anomaly Detection Benchmark, using real and annotated data from European Space Agency missions, demonstrating a robust balance between model performance, scalability, usability, and robustness compared to reference algorithms like Global STD, iForest, KNN, or Telemanom ESA. It is particularly suitable for high telemetry volumes, limited labeled data, and the need for explainable alerts. Benefits include detection without labeled data, scalability in complex systems, explainable alerts, and readiness for operation with low computational requirements and easy integration into monitoring processes.
Documentation
Need the full ICD, test reports or a specific revision? Ask the supplier directly.