WareFlowTwin Optimization Advisor
Lightweight optimization advisor that recommends warehouse improvement actions from process mining features and bottleneck signals.
Purpose
Given event-log features and detected bottlenecks, rank actions:
| Action | When to use |
|---|---|
| Reduce Transfers | High transfer ratio or repeated transfer bottleneck |
| Relocate Storage | Slow aisle dwell times, high picking density |
| Reprioritize Picking | Excessive waiting or shipment SLA risk |
| Rebalance Workers | Worker load imbalance across zones |
| Batch Picking | High picking density, compatible order mix |
Input Features (12)
avg_waiting_min, transfer_ratio, rework_rate, slow_aisle_share, shipment_delay_rate, case_volume, picking_density, storage_utilization, conformance_fit, variant_entropy, peak_hour_load, worker_load_imbalance
Files
| File | Description |
|---|---|
config.json |
Model configuration |
advisor_weights.json |
Linear policy weights and biases |
References
- Source movements: VillanovaAI/Temporal_Logistics_Inventory_Movements
- Benchmark: warehouse-process-mining-benchmark
- Simulator: warehouse-process-mining-digital-twin
License
Apache-2.0