Deployment of distributed model training services based on edge computing in transmission lines

In recent years, driven by the concept of edge intelligence, how to effectively use the geographic distribution of edge data for model training has become a key issue. For this reason, considering the revenue issue that is critical to service deployment from the perspective of the period time, a service deployment decision-making method that maximizes system revenue is proposed. First, we establish a layered network architecture centered on the edge service platform in the transmission lines and simplify the service deployment problem to a nonlinear 0-1 integer programming problem. Then, the SDAM model training service deployment algorithm was proposed. Finally, we verify the performance of this algorithm from the perspective of three indicators: comprehensive income, system average load rate, and load rate variance. Experimental results show that compared with the benchmark algorithm, the SDAM algorithm can increase system revenue by 25% while ensuring a balanced load.