Improving convergence properties of autonomous demand side management algorithms

For implementing Autonomous Demand Side Management (ADSM) in the smart grid, when there is a large number of consumers, there are some problems like low convergence speed and large volume of calculation for finding the optimal amount of energy consumption in any time slot. In this paper, we propose a new method to speed up the convergence of ADSM algorithms and to reduce the volume of computations. When the consumers try to calculate their energy consumption individually in parallel, a problem of peak shifting will be appeared, which causes divergence or slow convergence rate in the ADSM algorithm. Therefore, we modify the ADSM algorithm such that the consumers can avoid peak shifting while making decisions in parallel. The proposed algorithm increases the convergence rate via solving a primal problem instead of the main problem, and decomposing it for each consumer to solve it in parallel. The proposed algorithm is compared with the ADSM algorithms suggested for the same system model in the literature. Our results show a significant improvement in convergence rate and computation volume. In summary, the proposed algorithm is more than 7.5 times faster than the existing algorithms and reduces the number of required operations by more than 2.8 times.

Automated summary

This paper proposes a new method to speed up the convergence of Autonomous Demand Side Management (ADSM) algorithms and reduce the volume of computations by solving a primal problem and decomposing it for each consumer to solve in parallel. The algorithm avoids peak shifting and shows significant improvement in convergence rate and computation volume compared to existing algorithms.