Toshiba AI Technology Catalog

  • Anomaly detection

Facilities failure sign detection technology: Detects deviations from normal behavior

This technology can detect signs of failure, even in facilities data with complex changes resulting from controls.

  • Based on discussions with the client, among facilities data with multi-dimensional timelines, we designate data from sensors and devices where failure signs often arise as “data requiring particular attention.”
  • This technology uses AI to detect deviations from “normal behavior” in data requiring attention, and interprets these deviations as signs of failure.
  • Because only normal data is needed for AI learning, the technology can be applied even in cases where there is no data on abnormalities.
  • Because it adopts deep learning as an AI method, it can be applied even in cases where the facilities data demonstrates complex changes due to controls.


  • Detect signs of failure in facilities and equipment; e.g., plants, manufacturing equipment, and building facilities

Benchmarks, strengths, and track record

  • Accommodate even data with complex changes.
  • There are examples of applications in products, and examples of PoC in detecting failure signs.