Research on Computational Offloading Strategies in Industrial Quality Inspection Scenarios

The study investigates the optimization of execution efficiency for interdependent task in industrial quality inspection systems through the proposal of a Fault-aware Dynamic Deep Deterministic Policy Gradient (FD-DDPG) based computational offloading framework. Initially, task dependencies are formally modeled using a Directed Acyclic Graph (DAG) structure, with the Heterogeneous Earliest Finish Time (HEFT) algorithm being implemented to derive execution priority sequences. Subsequently, a multi-objective utility function integrating both temporal latency and energy expenditure is formulated as the optimization criterion. Building upon this framework, we establish a Markov Decision Process (MDP) model enhanced with fault handing mechanisms, enabling the autonomous derivation of optimal offloading policies through DDPG algorithm. Experimental results reveal that compared to existing offloading schemes, the proposed FD-DDPG exhibits superior capabilities of latency, energy, and fault handing.