Click machines below to toggle manual jam/shock
Q-Agent Exploring (ε=0.15)
Cycle: 0
Items Produced: 0
Delivery Rate: 0.0/min
Line Stall Loss: 0%
Avg Q-Value: 0.00
Live Factory Benchmarks Q-LEARNING
Efficiency Yield
94.2%
Energy Draw
28.4 kW
Buffer Overflow Loss
0 units
Reward Score
+142.0
Agent dynamically routes raw iron/coal toward whichever smelter is falling below starvation threshold.
Production Shocks
RL Policy Hyperparameters
0.15
0.15
0.90
1.5
State-Action Q-Heatmap
| Lattice State | Route S1 | Route S2 | Route Split |
|---|---|---|---|
| S1 < 30% Low | +4.12 | -1.20 | +0.50 |
| S2 < 30% Low | -0.85 | +4.80 | +0.10 |
| Balanced Queues | +1.40 | +1.35 | +2.90 |
| High Backpressure | -1.10 | -1.05 | +0.85 |