Disclosed herein are methods, systems, and devices for utilizing distributed reinforcement learning and consensus control to most effectively generate and utilize energy. In some embodiments, individual turbines within a wind farm may communicate to reach a consensus as to the desired yaw angle based on the wind conditions.

Provided is a method and system of positive and negative sequence rotor currents control for DFIGs under a single dq-PI control structure, comprising: adjusting the negative sequence rotor current reference

[00001]$\stackrel{.fwdarw.}{I}{\overline{{}_r}}_{\_ref-\ast }$

according to the negative sequence stator terminal voltage to obtain a reference adjustment value

[00002]$\stackrel{.fwdarw.}{I}{\overline{{}_r}}_{\_com-\ast }$

; converting the reference adjustment value to the forward-rotating dq coordinate system and superimposing it with the positive sequence rotor current reference as the input of a PI-regulator-based current feedback controller to uniformly control the positive and negative sequence rotor current; determining the output voltage reference of the rotor-side converter by the PI-regulator-based current feedback controller, according to which, the switching signal of the rotor-side converter can be determined through the SPWM function, controlling the turn-on and turn-off of the bridge arms of the rotor-side converter to form the output voltage. It realizes the control of the positive and negative rotor currents under the single dq-PI control structure, retaining the good control performance of the control structure for the positive sequence rotor current, and enabling the control of the negative sequence rotor current under unbalancing conditions, with a control structure with lower order and lower complexity compared to the existing control structures for realizing the DFIG positive and negative sequence current control.

The invention provides a method for controlling a wind turbine. The method predicts behaviour of the wind turbine components for the time stages over a prediction horizon using a wind turbine model describing dynamics of the wind turbine, where the time stages include a first set of time stages from an initial time stage and a second set of time stages subsequent to the first set. The method determines control outputs, e.g. individual blade pitch, for time stages based on the predicted behaviour. The method then transmits a control signal to implement only the control outputs for each of the second set of time stages so as to control the wind turbine. Advantageously, the invention reduces both average and peak computational loads relative to standard predictive control algorithms.

The present disclosure is directed to a method for responding to a friction coefficient signal of a pitch bearing of a pitch drive mechanism of a wind turbine and/or for controlling the pitch drive mechanism(s) and/or a bank of ultracapacitors. The method and system include: accessing high-frequency measurement data of the at least one pitch bearing; estimating, via a torque balance model implemented by a controller, a frictional torque of the at least one pitch bearing based, at least in part, on the high-frequency measurement data; estimating, via the controller, a friction coefficient signal of the at least one pitch bearing based, at least in part, on the frictional torque; comparing the friction coefficient signal with a friction threshold; determining whether the friction coefficient signal deviates from the friction threshold based, at least in part, on the comparison; and, if so, acting.

A method for providing backup control for a supervisory controller of at least one wind turbine includes observing, via a learning-based backup controller of the at least one wind turbine, at least one operating parameter of the supervisory controller under normal operation. The method also includes learning, via the learning-based backup controller, one or more control actions of the at least one wind turbine based on the operating parameter(s). Further, the method includes receiving, via the learning-based backup controller, an indication that the supervisory controller is unavailable to continue the normal operation. Upon receipt of the indication, the method includes controlling, via the learning-based backup controller, the wind turbine(s) using the learned one or more control actions until the supervisory controller becomes available again. Moreover, the control action(s) defines a delta that one or more setpoints of the wind turbine(s) should be adjusted by to achieve a desired outcome.

A method to operate a wind turbine is provided, the method including determining a correction model associated with the wind turbine, determining a corrected turbulence intensity parameter associated with the wind turbine based on the correction model, and operating the wind turbine based on the corrected turbulence intensity parameter.

A system and method are provided for controlling a wind turbine of a wind farm. Accordingly, a controller determines a performance differential for the wind turbine at multiple sampling intervals of a yaw event. The controller determines a trendline for the wind turbine correlating the performance differential to a deviation of a wind direction at each of the multiple sampling intervals from an first yaw angle. A difference between an angle associated with the vertex of the trendline and the first yaw angle are utilized by the controller to determine a yaw angle offset. The yaw angle offset is used to adjust a second yaw angle of the wind turbine.

A method for determining a correction function for a wind turbine, a method and system for determining an alignment correction function for a nacelle of a wind turbine, and a method for operating a wind turbine are provided. Measurement values of the power measure of the wind turbine and of the leeward wind direction are assigned to measurement values of the leeward wind speed, corrected by a correction function, and are grouped into at least one wind-speed bin on the basis of instants at which the measurement values were recorded. A model function is determined and outputted for a relationship between the power measure and the leeward wind direction for the wind-speed bin, and an alignment correction function is determined for a target alignment of the nacelle relative to the measured leeward wind direction on the basis of the model function.

Disclosed are a method and apparatus for detecting a fault, and a method and apparatus for training a model. The method includes: acquiring characteristic data and actual temperature of a first wind turbine among n wind turbines, wherein the characteristic data of the first wind turbine is intended to characterize a working state of the first wind turbine, and n is an integer greater than 1; acquiring a prediction temperature set by inputting the characteristic data of the first wind turbine into a temperature prediction model corresponding to each of the n wind turbines; and detecting, based on the predicted temperature set and the actual temperature of the first wind turbine, whether the first wind turbine encounters a fault. Compared with the related art which depends on the working experience of the staff, the technical solution according to the embodiments of the present disclosure can more accurately detect whether a wind turbine encounters a fault, and provide early warning in time, so as to reduce the failure rate of the wind turbine.

A control system for a power generation system includes a generator coupled to a turbine via a shaft. The control system includes a memory storing instructions. The control system also includes a processor coupled to the memory and configured to execute the instructions. When the instructions are executed it causes the processor to receive a direct current (DC)-link voltage from an automatic voltage regulator (AVR), wherein the AVR is configured to control voltage characteristics of the generator, and to determine a speed of the generator based on the DC-link voltage.