Office of Technology Transfer – University of Michigan

High Performance Anomoly Detection

Technology #2972

Questions about this technology? Ask a Technology Manager

Download Printable PDF

Dragan Djurdjanovic
Managed By
Drew Bennett
Associate Director - Software Licensing 734-615-4004


In the current paradigm of product development, the quality of a product, its production, and service is mainly designed, tested, and implemented during its development. As such, once a product is released, any quality problems may be difficult to identify. Particularly in the automotive industry, engineering is often at the root cause of problems, which lead to warranty repairs, resulting in loss of a company’s profit. Anomaly detection is employed in complex non-linear systems, such as an automotive system, and requires a high-fidelity model or representation of nominal system behavior that can be compared to actual system behavior to detect deviations. Such systems often require expert guidance or substantial computation time, due to which real-time monitoring becomes difficult. Furthermore due to the large number of inputs, environmental factors, and complex interrelationships in many such systems, the root cause for one or more anomalies is difficult to determine.


University of Michigan researchers have developed a system and method for detecting anomalies, based on learning model-based lifecycle software and systems. Such software and systems are self-diagnosing and typically include embedded diagnostic agents, which can include anomaly detection agents and knowledge-based agents. The systems can include an integrated development environment (IDE) and a run-time environment (RTE) that are linked together. The IDE contains a set of development tools linked within the IDE and to the RTE. The RTE includes a number of diagnostic agents linked within the RTE and to the IDE. Thereby, the development tools and the diagnostic agents communicate with each other to detect anomalies based on the deviation between learned normal operating behavior and monitored current operating behavior.

Applications and Advantages


  • Distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network


  • Does not require detailed knowledge of-nl-the system dynamics
  • Detects gradual changes of the system
  • Detects and isolates controller (software)-nl-anomalies from plant (hardware) anomalies