Provided is a method for computer-implemented monitoring of a component of a wind turbine, having access to a trained machine learning model which has been trained for one or more components of the same type of wind turbines. The trained machine learning model is configured to provide an output referring to a predetermined fault occurring at a component of a wind turbine by processing vibration signals in a predetermined domain which are measured in the vicinity of the component during the operation of the wind turbine. Vibration signals are mapped to corresponding vibration signals valid for the component based on one or more given kinematic parameters of the component and one or more given kinematic parameters of another component. The machine learning model is applied to the vibration signals valid for the component, resulting in an output referring to the predetermined fault occurring at the another component.

A method of controlling a wind turbine, the method comprising: determining an initial thermal model representing thermal characteristics of a plurality of components of a first wind turbine; receiving operational data relating to thermal characteristics of components of a plurality of wind turbines; processing the initial thermal model and the operational data using an optimisation algorithm to determine a modified thermal model for the plurality of components of the first wind turbine; and controlling the first wind turbine in accordance with the modified thermal model.

The present disclosure provides a method and an apparatus for self-adaption of a cut-out strategy. The method may include: predicting, using a wind speed prediction model, a wind resource parameter of a wind turbine at each machine location; predicting, using a load prediction model, a fatigue load and a limit load of the wind turbine based on the predicted wind resource parameter and an air density; comparing the predicted fatigue load and limit load with a reference load; and determining the cut-out strategy based on a result of the comparison, wherein determining the cut-out strategy includes determining a cut-out wind speed and an output power.

Disclosed are a system and a method for reinforcement learning-based real time robust variable pitch control of a wind turbine system. The system includes: a wind speed collecting module to collect wind speed values of a wind farm; a wind turbine information collecting module to collect a rotor angular speed; a reinforcement signal generating module to generate a reinforcement signal based on the collected rotor angular speed and the rated rotor angular speed; a variable pitch robust control module including an action network and a critic network, wherein the action network is configured to generate an action value based on the wind speed of the wind farm and the rotor angular speed and output the action value to the critic network; the critic network is configured to perform learning training based on the reinforcement signal and the action value, generate a cumulative return value and output the cumulative return value to the action network; and the action network performs learning training based on the cumulative return value to update the action value and output the updated action value; and a control signal generating module connected to the action network, configured to generate a corresponding control signal based on the received action value. The wind power generator adjusts the pitch angle based on the control signal, which realizes adjustment of the rotor angle speed and guarantees smooth and stable power output of the wind turbine.

Systems and methods for estimating a direction of wind incident on a wind turbine, the wind turbine comprising a tower; a rotor-nacelle-assembly (RNA) carried by the tower; a deflection sensor configured to sense a position of the RNA or a deflection of the tower; and a wind direction sensor. One approach includes: obtaining deflection training data from the deflection sensor; obtaining wind direction training data from the wind direction sensor; training a machine learning model on the basis of the deflection training data and the wind direction training data in order to obtain a trained machine learning model; obtaining further deflection data from the deflection sensor; inputting the further deflection data into the trained machine learning model; and operating the machine learning model to output a wind direction estimate on the basis of the further deflection data.

A method of operating a chiller includes applying chiller operating data associated with the chiller as an input to one or more machine learning models, generating a boundary for a controllable chiller variable based on an output of the one or more machine learning models, and affecting operation of the chiller based on the boundary.

A method for computer-implemented prediction of power production of a wind farm includes: obtaining first weather forecast data for a first time period, obtaining first power production data for the first time period, obtaining second weather forecast data for a second time period; determining second power production data for the second time period by processing the first weather forecast data, the first power production data and the second weather forecast data by a trained recurrent neural network, where the first weather forecast data, the first power production data and the second weather forecast data are fed as a digital input to the trained recurrent neural network and the recurrent neural network provides the second power production data as a digital output, the second power production data being a prediction of power production for the second time period.

An adaptive dynamic planning control method and system for a large-scale energy storage station. The method comprises: setting a structure and control target parameters of an adaptive dynamic planning control system; initializing the parameters and importing an initial state of a controlled object; calculating an original wind electricity power fluctuation rate at a current moment t and smoothing the original wind electricity power according to a change rate control strategy; calculating a smoothed wind storage power fluctuation rate, a power of an energy storage system, and a state of charge (SOC) of the energy storage system; initializing and training an evaluation module and an execution module; calculating and saving a control strategy, a smoothed wind storage power fluctuation rate, an energy storage power and a (SOC) at each moment; and outputting the control strategy at each moment, the smoothed wind storage power fluctuation rate, the energy storage power and the (SOC).

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.

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.