By considering the steady wind regime and the obvious influence on wind turbine power due to wake effect on the offshore farm

the yaw angle and active power among wind turbines are comprehensively coordinated

which improves aerodynamic coupling among turbines and total active power. A wake model is proposed which takes the yaw angle into account

and it is pointed out that as the boundary in the classical wake model is discontinuous

it is hard to solve the plant power optimization. By setting yaw angle and induction factor as regulation means

a wind farm active power optimization model is developed. Then

grouping on turbines based on the wake propagation path

the whole wind farm optimization problem is transformed into internal optimization problem in each group

which decreases the number of optimization objects and reduces problem scale. Focus is on combining online simulation with machine learning

and a solving method for internal power optimization problem for each group is proposed. Finally

optimization results are set as reference active power and reference yaw angle for each turbine to go by. Each turbine operates according to reference active power and reference yaw angle set by the optimization results. The scheme requires less computation cost

does not need to increase computing resources of the wind farm control system and has no special requirement for communication environment. In addition

numerical simulations show that the proposed scheme effectively increases the total active power and improve economic benefits in an offshore wind farm.