Wind turbine replacement schedule
Provided is a method and arrangement of estimating replacement schedule of a, in particular mechanical, component of a wind turbine, the method comprising: estimating remaining producible energy until this component is to be replaced.
1. A method of estimating replacement schedule of a mechanical component of a wind turbine, the method comprising:
detecting a value of a physical quantity indicating a malfunction of the mechanical component of the wind turbine;
estimating remaining producible energy until the mechanical component is to be replaced, wherein estimating is performed when the value of the physical quantity exceeds a threshold; and
operating the wind turbine for optimizing power output based on the estimated remaining producible energy;
wherein estimating the remaining producible energy uses machine learning and includes using a neural network which has been trained by training data from plural training wind turbines, the training data comprising training values of input parameters and a training value of at least one output parameter, the input parameters comprising at least one wind turbine operational parameter and/or at least one ambient condition parameter, the at least one output parameter being indicative for the remaining producible energy until the mechanical component is to be replaced.
2. The method according to claim 1 ,
wherein the physical quantity indicating the malfunction of the component comprises a temperature, a noise, an electrical quantity related to power output, rotational speed, and/or friction.
3. The method according to claim 1 , wherein the physical quantity indicating the malfunction of the component comprises a vibration.
4. The method according to claim 3 , wherein the vibration is measured by a sensor close to the component.
5. The method according to claim 3 , wherein the component is a bearing.
6. The method according to claim 5 , wherein the bearing is a bearing of a main rotation shaft.
7. The method according to claim 1 , wherein estimating the remaining producible energy is performed during power production of the wind turbine.
8. The method according to claim 1 , wherein
the training data from plural training wind turbines includes training data from at least 50 training wind turbines.
9. The method according to claim 1 , the training wind turbines comprising a same or similar component having same or similar malfunction.
10. The method according to claim 1 , wherein estimating the remaining producible energy comprises using the neural network comprising:
supplying test values of the input parameters to the trained neural network, the test values pertaining to the wind turbine;
outputting, by the trained neural network, a probability distribution of the remaining producible energy pertaining to the wind turbine.
11. The method according to claim 10 , wherein the probability distribution is given for plural points in time in the future.
12. The method according to claim 1 , wherein, as estimation of the remaining producible energy at a desired point in time, one of:
a mean;
a median;
a mode; and/or
a set density
of the probability distribution at the desired point in time is used.
13. The method according to claim 1 , wherein training the neural network comprises:
weighting the input parameters depending on the training data regarding their importance for arriving at the respective training value of the output parameter by applying a cost function.
14. The method according to claim 1 , wherein the at least one wind turbine operational parameter comprises at least one of:
a power output;
a rotational speed of a rotation shaft at which plural rotor blades are mounted;
a torque of a generator;
energy diffusion;
construction characteristics of the component.
15. The method according to claim 1 , wherein the at least one ambient condition parameter comprises at least one of:
a wind speed;
a wind turbulence;
a humidity;
a temperature.
16. The method according to claim 1 , further comprising:
indicating to replace the component, if the remaining producible energy is smaller than an energy threshold.
17. An arrangement for estimating a replacement schedule of a mechanical component of a wind turbine the arrangement being adapted to control or perform a method according to claim 1 .
18. The method according to claim 1 , wherein the neural network is a recurrent neural network.