To address the challenges posed by the complex and variable operation conditions

as well as the coexistence of multi-mode oscillation in offshore wind farms connected with modular multilevel converter based high-voltage DC (MMC-HVDC) transmission systems

the mechanism and the key influencing factors of the wideband oscillations of the permanent magnet synchronous generator (PMSG)-based wind farm connected with MMC-HVDC were first revealed using the impedance analysis method. On this basis

an artificial intelligence-based adaptive method for suppressing wideband oscillations was then proposed

which integrates a virtual impedance loop into the basic controller of the sending-end MMC. By offline training a neural network that takes the damping coefficient of virtual impedance and system operating conditions as inputs and outputs the system’s phase margin

this network is then applied for online adjustment of damping coefficients. Combined with rapid detection of wideband oscillations

adaptive suppression of wideband oscillations is achieved. Furthermore

the method is extended to mitigate multi-mode oscillation by utilizing a neural network trained with multi-channel damping parameters

allowing for the online adjustment of multi-channel damping coefficients. Finally

the effectiveness of the proposed adaptive wideband oscillation suppression method is verified by simulations.