An Operator Splitting Scheme for Distributed Optimal Load-Side Frequency Control With Nonsmooth Cost Functions

The increasing penetration of renewable energy resources and utilization of energy storage systems pose new challenges in maintaining power system's stability. Specifically, the cost function of regulation no longer remains smooth, which complicates the task of ensuring nominal frequency and power balance, particularly in a distributed manner. This article proposes a distributed proximal primal-dual (DPPD) algorithm, based on a modified primal-dual dynamics equipped with operator splitting technique, to address this nonsmooth frequency regulation problem on load side. By Lyapunov stability and invariance theory, we prove that DPPD algorithm achieves global asymptotic convergence to the optimal solution that minimizes the nonsmooth regulation cost, while restoring the frequency under constraints, such as capacity limits of load and power flow. Finally, we demonstrate the effectiveness and robustness of DPPD algorithm by simulations on the IEEE 39-bus system.