Smart Abnormal Trend Analysis System (SATAS)

Industrial partner:  DSP Construction & Engineering Works Pte.Ltd

• The operational process of SATAS

• Data Filtering Module (DFM)

  • It is found that some dates have NULL entries or data corresponding to certain time slots are missing. Initially such instances are identified by DFM and filtered out on daily basis.

• Feature Extraction Module (FEM)

 

• • The FEM is used to extract significant features for abnormal event detection.

    

    

• Abnormal Trajectory Detection Module (ATDM)

  • The trajectories of the principle components extracted in the FEM are used as features to detect the dates with abnormal events.

• Detailed Abnormal Event Detection Module (DAEDM)

  • The raw data corresponding to the suspected abnormal trajectories are analyzed to extract more details of the

    abnormality.