This article introduces a novel hybrid adaptation algorithm comprising both continuous mixed p-norm (CMPN) strategy and the block-sparse Bayesian (BSB) technology to online adapt all the proportional plus integral (PI) controller gains of the power converters of flywheel energy storage units (FESUs). The FESU is based on the doubly fed induction machine (DFIM). The principal target is to improve the transient stability of grid-connected wind farms. To obtain a realistic study, the two-drive train model is utilized in the wind farm modeling affecting the transient analyses. The cascaded control scheme is implemented on controlling both the generator side converter and the grid side inverter of the FESU using the proposed adaptive CMPN-BSB PI controller. The FESU is tied to the point of common coupling of the wind farm, which is tied to the standard IEEE 39 bus system. The performance of the proposed technology is tested under subject the system to different severe faults. The validity of adaptive CMPN-BSB PI controlled FESU is tested by comparing the numerical results with that achieved by least mean square adaptive controlled FESU. The numerical results are performed by PSCAD software. The proposed controlled FESUs will help in improving the transient stability of wind farms and their power quality concerns will be enhanced.