Anomaly, Incident and Change Detection

Automated detection of anomalies, incidents, and changes is another fundamental component of "intelligent", automated/autonomous systems.

On the one hand, it serves to ensure the quality and plausibility of data in any form and from any source.

On the other hand, anomalies, fault, incident and change detection is an important direct and active component in various safety functions or systems for monitoring and predicting system states.


Generic applications

  • Quality assurance of data from any kind of tests and simulation
  • Automated identification of harmful or conspicuous data in extensive test series
  • Automated identification of failed tests already during the test execution
  • Implement of diagnostic systems for automatic and learning machine monitoring
  • Efficient identification and classification of type and frequency of
    • real system states and operational loads of systems for improved layouts in continuous product improvement
    • abnormal and seldom operational states, system use cases and scenarios

Specific applications

  • Evaluation of traffic flows and automatically detect accidents or other special events
  • Implementation of warning systems for environmental pollutants
  • Quality assurance in simulation data management and scenario management
  • Monitoring of operational design domains in automated driving


ANDATA's technological approach to anomaly and incident detection consists of a specific expert system in which the "expectations of normal data" and systems are formulated in the form of a committee of criteria criteria with different complexities. These criteria consists amongst others of simply, statistical metrics, arbitrary signal metrics, complex pattern recognition techniques or sophisticated adaptive machine learning methods for signal classification. The comparison of forecast models with actual measurements can also be used as a criteria.


Anomalies, fault and incident detection is a standard procedure in all ANDATA applications. At the most important automotive customers the following exemplary systems are in productive use:

  • Expert system for quality assurance and result plausibility in simulation data management (e.g. CAE-Bench)
  • Quality assurance in scenario management for the development and validation of automated driving functions
  • Plausibilization of crash signals from tests and simulations


Brainer Stipulator Brainer