Spatiotemporal decoupling of actuator dead zones in high throug 2027 Mineral

The processing capacity of high-throughput vision-based intelligent ore sorting systems is strictly constrained by the physical dead zones of mechanical actuators. Under random Poisson arrivals, particle clustering in space translates to severe time-domain congestion that degrades sorting performance. While algorithmic optimization cannot overcome this physical bound and homogeneous multi-stage configurations suffer from superimposed dead-zone penalties, this paper proposes a heterogeneous dual-stage sorting architecture and a corresponding adaptive collaborative scheduling strategy (ACSS). By cascading a primary high-torque actuator (HTA, modeled as M/D/1/0 loss queue) and a secondary high-dynamic actuator (HDA, modeled as M/G/1/0 loss queue), the architecture achieves spatiotemporal decoupling of congestion via multi-stage spatial stratification. The ACSS integrates dynamic time windows, state-holding logic and a predictive abort mechanism to proactively mitigate sorting losses arising from combined dead-zone effects across the two stages. Discrete-event simulations (DES) and hardware-in-the-loop (HIL) experiments demonstrate that, at 50 % target particle proportion, the system raises the 80 %-recovery threshold from 2 particles/s (single-stage HDA) to 4 particles/s, doubling the processing capacity. The proposed system exhibits an asymmetric U-shaped performance characteristic.