AgriSense · Rice Mill Predictive Analytics

🏭 Process Line — Live Asset Status

Paddy in → milled rice out. Color = real-time machine health.

🩺 Asset Fleet — Health & Remaining Useful Life

Click a machine to drill into its predictive model

🚨 Predictive Alerts

Ranked by failure risk & impact

📈 Plant Vibration Index (all rotating assets)

RMS velocity, mm/s · rising trend on WP-01 & RH-01 driving alerts

WP-01 Whitener RH-01 Husker Fleet avg Alarm (7.1)

🌡 Bearing Temperature Trend

°C · early thermal signature of bearing degradation

WP-01 RD-01 Dryer Warning (75°C)

⚙ Predictive Maintenance Workbench

Physics + ML model output per asset · RUL, failure probability, fault signature

❤ Health Index

Composite condition score

⏳ Remaining Useful Life

Model-estimated days to functional failure

⚠ 30-Day Failure Risk

Probability of failure within horizon

📉 Failure Probability Forecast (next 60 days)

Survival model — crosses the 50% line near projected failure date

Failure probability Maintenance trigger (40%)

🎚 Vibration vs Alarm Thresholds

ISO 10816 zones — Good / Warning / Alarm

Measured Warning Alarm

🔊 Vibration Frequency Spectrum (FFT)

Bearing fault frequencies — a spike at BPFO/BPFI confirms outer/inner race defect, not just imbalance

🧩 Component Wear

Estimated remaining life by part

✅ Recommended Actions

⚡ Plant Power Draw (live)

Total real power, kW · streaming from motor meters

🔌 Power Factor

Reactive load efficiency

Below 0.90 incurs utility penalty

📊 Specific Energy Consumption by Asset

kWh per tonne of paddy · highlights energy-inefficient machines

💰 Energy Cost & Avoidable Waste

Daily ₹ · anomaly detection flags abnormal draw

Actual cost Optimal baseline

🏆 Head Rice Yield

% whole grains — the profit metric

💧 Dryer Outlet Moisture

% — target 12–13% for safe storage

✨ Polish / Whiteness

Degree of milling index

📉 Broken Rice % — Trend & Forecast

Husker roller wear (RH-01) is pushing breakage above target

Broken % Target (4.0%) Forecast

🔬 Root-Cause Correlation

Husker rubber-roll gap vs broken rice — ML found the driver

Correlation r = 0.86 · Optimal gap ≈ 0.65 mm · current drift to 0.92 mm is increasing breakage.

🛠 Auto-Generated Work Orders

Created from predictive triggers — prioritised by risk × downtime cost

Priority	Asset	Predicted Issue / Action	Due	Est. Cost	Status

📦 Spare Parts Readiness

Stock vs predicted demand (next 30 days)

🗓 Next 14-Day Schedule

📈 Reliability Trend — Reactive vs Predictive Era

Unplanned downtime hours/month since sensor program went live (month 4)

Unplanned downtime (hrs) Planned maintenance (hrs)

💡 What predictive analytics delivers for the rice mill

Each capability is driven by the same sensor stream, layered into increasing business value.

🔧 How it works — from sensor to decision

Vibration, temperature, motor-current, RPM, power and moisture sensors stream from each machine into an edge gateway on the plant floor. Data lands in a time-series database; ML models (anomaly detection, survival/RUL, regression) score it continuously and push alerts, work orders and this dashboard.

📡 Sensors on each machine→ 🔌 Edge gateway (Modbus/MQTT)→ 🗄 Time-series DB→ 🧠 ML models (RUL · anomaly · yield)→ 📊 Dashboard · alerts · auto work-orders

Typical sensor set per machine: tri-axial vibration (bearing housing), bearing & motor-winding temperature (RTD/IR), motor current/power (CT clamp), shaft speed (proximity), plus process sensors — moisture (dryer/paddy), roller gap (husker), throughput (load cell).

AgriSense Predictive Analytics — demonstration build with simulated sensor data. Numbers refresh live to mimic a real plant feed. Replace the simulator with your historian/MQTT feed to go production.

AgriSense | Predictive Analytics