The invention discloses a
wind power plant real-time
yaw cooperative
control system and method fused with
machine learning, and belongs to the technical field of
wind power generation control. The invention aims to solve the problems of low precision, long calculation time and sparse samples of a
wind power plant wake flow control model under complex terrains. According to the technical scheme, the method comprises the following steps: constructing a prior
knowledge base by using OpenFOAM
simulation and a virtual
laser radar technology; establishing a
hybrid deep neural network comprising a main prediction network and a residual error correction network, and realizing accurate flow field prediction of virtual and real data fusion; and based on a multi-agent
reinforcement learning algorithm, a collaborative
yaw strategy is generated by taking maximization of the whole-field
generating capacity as a target. According to the invention,
millisecond-level real-
time response can be realized, the overall power generation benefit of the wind power
plant under the complex
terrain is effectively improved, and the equipment load is reduced.