I built a time series forecasting system for eight tank sensors, incorporating external variables like weather and manual inputs. The goal was to improve reliability in environments with partial automation.

• 8 models trained individually for each sensor
• MAE between 0.5–0.8 mm (out of ~50 mm range)
• Max error under 5 mm even in edge cases
• Enables proactive maintenance and anomaly detection

• ScopeTime series forecasting for industrial IoT
• ImpactImproved planning and reduced maintenance risks