Main Content

AKITA

AKITA is a project on time series analytics at the Hasso-Plattner-Institute in collaboration with Rolls Royce. The goal of this project is to support aircraft engineers in the development of new jet turbine technologies that reduce CO2 and noise emissions. For this purpose, we develop novel anomaly detection systems for the automatic and scalable analysis of engine test data. In this way, test cycles can be shortened and modern technology released in a more timely manner.

On the technical side, we conduct research on more efficient, more robust and more precise automatic subsequence anomaly detection algorithms for time series data. Our AI algorithms are supposed to find, classify and report suspicious sensor recordings in huge corpora of measured time series data. Challenges in this analysis are the complexity of the time series, their size (lengths and width), the lack of training data, and the variability of expected anomalies and recording settings.

Summary of project details:

• Subprojects:
  • Efficient Subsequence Anomaly Detection on Time Series Data
• Funding:
  • Luftfahrtforschungsprogramm (LuFo)
• Partners:
  • Hasso Plattner Institute
  • Rolls Royce
• Selected Publications:
  • Anomaly Detection in Time Series: A Comprehensive Evaluation (VLDB 2022)
  • TimeEval: A Benchmarking Toolkit for Time Series Anomaly Detection Algorithms (VLDB 2022)
  • Distributed detection of sequential anomalies in univariate time series (VLDBJ 2021)