Electric Motors Predictive Maintenance (Automotive Client)

about
Our client is a leading automotive manufacturer in Turkey.
Problem
  • Fast-paced manufacturing pace, allows no time for stoppage
  • Maintenance engineers are missing "hard-to-detect" defects resulting in failures and severe interruptions of production
  • Conventional predictive maintenance applications are costly to scale as they require very high frequency of data (i.e. milliseconds) and data collection is typically hardware-dependent (sensors to be placed)
Action
  • AI detects subtle machine anomalies/defects that might be an indicator of failures – a simple yet capable approach
  • Implemented plug & play, highly scalable solution that does not require any special sensors or any kind of hardware (utilizes data collected by built-in driver)
  •  
  • Single column of data is enough, no redundant complexity (e.g., electricity current with 500 ms frequency)
  • Adjustable to different type of machines and equipments that contain an electric motor.
Tool Stack
  • Python, Spark, Cassandra, Grafana, Prometheus, Kafka, Airflow

6

Hrs

Predict Failures Hours In Advance

$

50

K

Avoided Cost per Failure

15

%

Overall Equipment Efficiency Increase