JP2020153347A - Wind power generator - Google Patents

Abstract

To achieve a high speed, accurate, and reliable wind state prediction function by a limited data processing resource of a windmill by using a Lidar for the purpose of improving power generation efficiency and reducing fatigue loading of a wind power generator.SOLUTION: A wind power generator having a tower, a nacelle provided to the tower, and a blade connected to the nacelle is equipped with a Doppler Lidar for detecting a wind speed and a control device in the nacelle. The Doppler Lidar has a function to emit laser beams in a plurality of directions, and measure wind speeds of a plurality of measurement distances along each laser beam in a predetermined measurement time. The control device has a wind speed prediction function to input a dataset that selects a part out of wind speed data measured by the Doppler Lidar, and predict a wind speed at a position of the wind power generator.SELECTED DRAWING: Figure 1

Description

The present invention relates to a wind power generator.

The operating efficiency, reliability, life, etc. of a wind power generator are affected by wind conditions such as wind speed and direction. Therefore, wind condition forecasting is an important issue.

Patent Document 1 is a background technique in this technical field. Patent Document 1 is an operation control method of a wind power generator including a forward wind speed measuring instrument, in which a step of measuring the wind speed at a remote position in front of the wind power generator by the forward wind speed measuring instrument and at least displacement of the forward wind speed measuring instrument. Disclosed that it includes a step of specifying a front remote position where measurement by a forward wind speed measuring instrument is performed based on the above, and a step of controlling a wind power generator based on the measured wind speed and the specified front remote position. Has been done.

Japanese Unexamined Patent Publication No. 2013-148508

As described in Patent Document 1, a Doppler lidar equipped with a nacelle (hereinafter referred to as a rider), which is a forward wind speed measuring instrument, measures the wind condition in front of the wind power generator from the Doppler shift of the laser beam emitted from the Doppler lidar. .. Compared to the method of controlling the measured value at the position of the wind power generator as it is, it is controlled by assuming that the measured value of the wind speed in front by the forward wind speed measuring instrument reaches the position of the wind power generator according to the wind speed and using this for control. Improvement of accuracy can be expected. However, the actual wind conditions do not propagate as they are while reaching the wind turbine from a distance. In the vicinity of the wind power generator, the wind conditions change in a complicated manner due to the rotor itself and the wind turbine tower, and the way of change is not uniform depending on the wind speed and time. Therefore, a method of predicting the wind speed immediately before the rotor from the distant wind speed measured by the rider by using machine learning has also been proposed. However, the restrictions on implementing the forecasting system in the actual wind power generator, especially the capacity and calculation scale of the storage device, control response time, etc. are not taken into consideration, and the wind condition forecast is made with the optimum accuracy, speed, and cost. There is a need to establish a feasible prediction system.

In addition, for wind turbines installed offshore, it is important to eliminate the effect of wind and wave tower tilt on wind speed measurements. Wind speed measurements are statically and dynamically affected by tower tilt. Here, the static effect means that the direction of the laser beam emitted from the rider shifts due to the tower tilting away from the vertical direction, and the horizontal component of the wind speed cannot be measured correctly. The wind speed measurement value fluctuates because the rider attached to the nacelle moves relative to the wind as the tower inclination angle fluctuates. Patent Document 1 describes the static effect, but does not show an effective correction method for the dynamic effect, and a solution is required.

In view of the above background technology and problems, the present invention is, for example, a tower, a nacelle provided in the tower, and a wind power generator having a blade connected to the nacelle, and detects the wind speed in the nacelle. It has a Doppler lidar and a control device, and the Doppler lidar has a function of emitting a laser beam in multiple directions and measuring wind speeds of a plurality of measurement distances along each laser beam at a predetermined measurement time, and controls the wind speed. The device is equipped with a wind speed prediction function that predicts the wind speed at the position of the wind power generator by inputting a data set selected from a part of the wind speed data measured by the Doppler lidar.

According to the present invention, it is possible to provide a wind power generation device capable of realizing wind condition prediction with optimum accuracy, speed, and cost.

It is a functional block diagram of the wind speed prediction of the wind power generation apparatus in Example 1. FIG. It is a figure explaining the wind speed prediction of the wind power generation apparatus in Example 1. FIG. It is a figure which shows the example of the data set used for the wind speed prediction of the wind power generation apparatus in Example 2. FIG. It is a figure which shows the specific example of the data set used for the wind speed prediction of the wind power generation apparatus in Example 2. FIG. It is a figure which shows the example of the error and the estimated time for different input data sets in Example 2. FIG. It is a figure which shows the influence of the inclination of the wind power generation apparatus in Example 3. It is a figure which shows the influence of the inclination in the wind power generation apparatus in Example 3. It is a figure which shows the difference of the delay time by the filter composition in Example 3. FIG. It is a figure which shows the state of inclination of the wind power generation device in Example 4. It is a functional block diagram of the wind speed prediction of the wind power generation apparatus in Example 5. It is a functional block diagram of the wind speed prediction of the wind power generation apparatus in Example 6. It is a functional block diagram of the wind speed prediction of the conventional wind power generation device.

Hereinafter, examples of the present invention will be described with reference to the drawings. In the following, the wind power generator will be referred to as a wind turbine.

First, the flow of control signals based on wind speed prediction in a conventional wind turbine, which is the premise of this embodiment, will be described. FIG. 12 is a functional configuration diagram of wind speed prediction of a conventional wind turbine. In FIG. 12, the wind turbine 10 is controlled by the FB control signal θFB obtained by inputting the rotation speed ωg of the wind turbine 10 into the feedback (FB) control 8 and the wind speed signal VL measured by the rider 5. Compared to a wind turbine controlled only by FB control, the use of rider measurement signals enables highly accurate wind turbine control.

On the other hand, FIG. 1 is a functional configuration diagram of wind speed prediction of the wind turbine in this embodiment. In FIG. 1, the wind turbine 10 is controlled by an FB control signal θFB obtained by inputting the rotation speed ωg of the wind turbine 10 into the feedback (FB) control 8 and an FF control signal θFF which is an output of the feedforward (FF) control 7. Will be done. Here, the wind speed signal VL measured by the rider 5 is input to the prediction system 6, and the prediction system 6 performs learning and prediction and outputs the wind speed prediction signal VP. Then, the FF control 7 receives the wind speed prediction signal VP and outputs the FF control signal θFF. As a result, more accurate wind turbine control becomes possible by using the FF control 7 based on the rider measurement signal, as compared with the wind turbine controlled only by the FB control 8.

Here, when implementing a prediction system on an actual wind turbine for the purpose of realizing highly accurate wind turbine control, various restrictions, especially the capacity of the storage device, the calculation scale in prediction, and the control response time are taken into consideration. There is a need. Details will be described below.

FIG. 2 is a diagram for explaining the wind speed prediction of the wind turbine in this embodiment. As shown in FIG. 2A, the wind turbine in this embodiment is a wind power generator equipped with a rider 5 as a wind condition measuring device, and has at least one blade 1 and a hub 2 to which the blade 1 is attached. It is composed of a nacelle 3 to which a hub is attached and a tower 4 that supports the nacelle 3. When the blade 1 receives the wind, the rotor composed of the blade 1 and the hub 2 rotates, and power is generated by the generator installed in the nacelle 3. As a means for predicting the wind condition reaching the wind turbine, a wind condition prediction system including a learner and a predictor for which the rider 5 predicts the wind condition at the position of the wind turbine by using the nacelle 3 and the rider measurement value is installed. In FIG. 2A, a downwind type is shown in which the nacelle 3 is upwind of the blade 1 with respect to the wind direction 20, but an upwind type configuration is used in which the nacelle is downwind of the blade. It doesn't matter if there is. The rider 5 emits a laser beam along four LOS (Line of Sight) 21, 22, 23, 24, and measures the wind speed along each LOS for measurement distances R1 to R5. Here, the LOS and the number of measurement distances may be different values as the rider's specifications.

FIG. 2B shows an example of the relationship between the position and the time of the input value and the output value in the prediction. In order to improve the accuracy of wind speed prediction, it is necessary to secure a certain number of wind condition data to be input. On the other hand, the FF control, FB control, and prediction system connected to the wind turbine have a limitation on the number of data to be input to the prediction system from the viewpoint of cost and installation location. In order to cope with this limitation, an example of setting input data is illustrated in FIG. 2 (b). Laser beams for measurement are emitted from the rider 5 to LOS 21, 22, 23, and 24 in different directions. The laser beam emitted here may be any number of 1 or more as the emission direction. Further, the emission method may be continuous or pulsed. In FIG. 2A, each of the plurality of measurement distances is set to 5 points of R1, R2, R3, R4, and R5, but any setting may be used. Since the laser beam is set in four directions, four measured values can be obtained for each measurement distance, and the wind speed in the center direction of the four beams can be obtained from the average of the four beams in consideration of the laser beam emission angle.

FIG. 2B shows the timing of measurement by the rider in the lateral direction with respect to the measurement distances R1 to R5. White circles are measurement points by the rider, black circles are rider measurement points used for prediction, and black squares are prediction points. The measurement points in FIG. 2B show the values at the center of 4 beams. The measured value at the center of the four beams is output to the prediction system every time interval Δt. In FIG. 2B, wind speed prediction values at a distance R1 and a time t ₀ + t ₁ are obtained using a total of 25 measurement points at five specific distances and five times for each Δt. Here, assuming that VR5 is the wind speed at the measurement distance R5 and time t0 is the current time, t1 = (R5-R1) / VR5 is the time from t0 to the future, and therefore t0 + t1 is the time for predicting the wind speed. is there.

The rider measurement point used for prediction needs to be set by the wind turbine operation manager according to the FF control and FB control of the wind turbine, the configuration of the prediction system, the location where the wind turbine is installed, the season in which the wind turbine is operated, and the like. Therefore, as shown in FIG. 1, a parameter setting user interface 11 for setting parameters such as a measurement distance and a measurement time that defines a data set to be input to the prediction system 6 is provided and set from here. Further, if the display user interface 12 having display functions such as confirmation of setting parameters, confirmation of predicted values, and comparison of predicted values and actually measured values is provided, it is possible to confirm the validity of parameters related to prediction.

As a result, by generating a control signal based on the short-term future wind turbine position and wind speed prediction derived using the forward wind conditions measured by the rider equipped with the nacelle, the response delay to the blade angle control of the wind turbine is reduced and the control effect is reduced. Is improved. Specifically, by implementing yaw control, blade pitch control, torque control, etc. of the wind turbine based on the wind speed prediction, the power generation efficiency can be improved, the fatigue load of the wind turbine blades can be reduced, and the reliability and life of the wind turbine can be improved. ..

Each function shown in FIG. 1 is realized by a device having a control device (CPU) and a storage device, which is a general information processing device as hardware. That is, the prediction system 6, the FF control 7, and the FB control 8 in FIG. 1 execute their functions by executing the programs stored in the storage device by the CPU. The parameter setting user interface 11 is a device for inputting data such as parameters from the user, such as a keyboard and a touch panel. Further, the display user interface 12 is a display device such as a monitor. The parameter setting user interface 11 and the display user interface 12 may be integrated with a touch panel or the like.

As described above, according to the present embodiment, the control response time such as the blade angle of the wind turbine can be secured to enable optimum control, and the power generation efficiency and reliability of the wind turbine can be improved. In addition, wind power generation that can realize wind condition prediction with optimum accuracy, speed, and cost using equipment with limited wind turbines for learners, predictors, and controllers, that is, limited storage capacity, calculation speed, etc. Equipment can be provided.

This embodiment describes how to determine the rider measurement points used for wind speed prediction, that is, how to specify the input data set.

FIG. 3 is a diagram showing an example of a data set used for predicting the wind speed of the wind turbine in this embodiment. (A) is a case where one input point is set. That is, the wind speed prediction value at the distance R1 and the time t ₀ + t ₁ is obtained by using one point at the time t ₀ of the measurement distance R2. Here, assuming that VR2 is the wind speed at the measurement distance R2, t1 = (R2-R1) / VR2. In this case, since the number of input data is one point, the load on the prediction system regarding the storage capacity and the calculation is the smallest, but it is disadvantageous for improving the prediction accuracy. (B) receives the different measurement distances R1, R2, R3, R4, R5 5 single wind speed measurement distance at at time _{t 0,} the distance R1, to obtain a wind speed prediction value at time _t 0 + _{t 1.} That is, t1 = (R5-R1) / VR5. In this case, the spatial distribution information of the wind condition for only one time is sent to the prediction system, and the time change information cannot be sent to the prediction system, which is somewhat disadvantageous for improving the prediction accuracy. In (c), the wind speed at one point at R5 at a certain time and the wind speed at one point at R2 at another time are used as input values. Here, the measurement time t ₂ = (R5-R2) / VR5 of R2 is determined on the assumption that the wind condition moves in parallel with the wind speed using the wind speed of R5. In this case, since the information on the spatial change and the time change of the wind condition can be input, it is advantageous in terms of improving the prediction accuracy, but the accuracy differs depending on the setting of the time interval t ₂ between the two points. In the prediction system, the calculation load of the time interval setting increases. In (d), four points having different times and measurement distances are input. In this case, since information on spatial change and temporal change can be input, it is relatively advantageous in terms of improving prediction accuracy.

FIG. 4 is a specific example when the input data pattern of FIG. 3D is adopted. In FIG. 4A, the measurement distances of the two points are R1 = 50 m and R2 = 80 m, and the interval between the measurement times of the two points is 5 seconds (s). The difference t ₁ between the predicted time and the current time t ₀ is the same as that described in the first embodiment, and t ₁ = (R2-R1) / VR2 = (80-50) / VR2. When the wind speed VR2 is 10 m / s, t ₁ is 3 s. Within the time of t ₁ , the wind turbine position is predicted, and the control of the blade pitch angle is started based on the prediction. FIG. 4B shows an input data pattern when the measurement distances of the points are R6 = 240 m and R5 = 200 m (hereinafter, abbreviated as 240 & 200 m) and the interval between the measurement times of the two points is 5 s. When the wind speed VR6 is 10 m / s, t ₁ = (R6-R1) / VR6 = (240-50) / 10 = 19s. By increasing the measurement distance of the wind speed used for prediction, it is possible to predict the wind speed in the future, and there will be more time for controlling the wind turbine, but the prediction error will increase. This relationship is shown in FIG.

In FIG. 5, the relationship between the average absolute prediction error (MAE) (bar graph) and the prediction time (line) for five types of input data patterns (80 & 50m, 160 & 50m, 160 & 120m, 240 & 120m, 240 & 200m) is shown in FIG. 5A. .. Further, FIG. 5 (b) shows the relationship between the average percentage error (MAPE) (bar graph) and the predicted time (line) for the same input data pattern. As for the predicted time, the wind speed (HWS) of the measurement distance (far side) is shown for two cases of 10 m / s and 20 m / s. From FIG. 5, it can be seen that as the measurement distance increases, the predicted time increases and the time margin in control can be secured, while the error increases. Therefore, the user interface 11 sets parameters such as the measurement distance and the measurement time that define the data set to be input to the prediction system 6, as shown in FIG. 1, whether to prioritize the time margin or the error reduction. Make it possible to select by.

In this way, by providing a user interface for optimizing the pattern of rider measurement values (measurement distance, measurement time, number of data) input to the prediction system, it is unique to wind turbine performance, weather conditions, wind turbine installation location, etc. It is possible to flexibly input and change parameters. That is, the rider measurement point pattern (distance and time) can be arbitrarily changed according to the weather, wind turbine performance, and the like.

In this embodiment, an embodiment in which the influence of the tower inclination of the wind turbine on the wind speed measurement will be described.

FIG. 6 is a diagram showing the influence of the inclination of the wind turbine in this embodiment. As shown in FIG. 6A, the wind turbine installed in the floating structure 16 is tilted by waves in addition to the wind, and the rider 5 mounted on the nacelle moves. Therefore, in order to obtain the horizontal wind speed HWS, it is necessary to make a correction excluding the influence of the inclination. There are static correction and dynamic correction, and the static correction corresponds to the change in the angle of the laser beam emitted from the rider due to the inclination. The HWS can be calculated given the tilt angle. The dynamic correction is to correct the fluctuation of the wind speed measurement value due to the relative movement of the rider as the tower inclination angle fluctuates.

In this embodiment, a method for dynamic correction is provided. FIG. 6B shows the dynamic effect of the tower inclination (τ) on the measured wind speed. Here, τ indicates a value after low-pass filtering with respect to the measured value. The necessity and design of the low-pass filter will be shown later in this embodiment. The time change of the wind speed and the time change of the inclination by the rider measurement of different measurement distances (50m, 80m, 120m, 160m, 200m) are shown superimposed. From this graph, since the wind speed changes in phase at each measurement distance and the phase of the inclination angle change matches, it can be said that this change is not due to the spatial movement of the wind condition but due to the inclination. Recognize. The rider's peripheral speed ω17 due to the in-phase change of the wind speed, that is, the inclination (τ) can be expressed by ω = (dτ / dt) Lτ when the inclination angle is τ as shown in FIG. 6A. Here, Lτ is the length of the tilted arm when calculating the peripheral speed of the rider. That is, it is the distance from the center of gravity of the floating structure 16 and the wind turbine 10 to the nacelle 3. Dynamic correction can be performed by adding the above ω to the HWS obtained from the measured value of the rider. By introducing dynamic correction, it is possible to eliminate wind speed fluctuations due to vibration and improve the accuracy of prediction.

As described above, in FIG. 6B, the values of τ after being low-pass filtered are plotted. Although the order is reversed, FIG. 7A shows the measured value before the low-pass filter and the value after the low-pass filter processing together. A digital filter is used as the low-pass filter, the configuration is IIR (Infinite Impulse Response), and a bike-wad type filter is connected to a cascade form (the number of connections is called a dimension) to change the dimension (one dimension). And 5 dimensions) Filter characteristics were examined. Looking at the plot (without filter) before the low-pass filter processing in FIG. 7A, there is a variation due to noise of the tilt angle sensor of about ± 2.5. In order to obtain the movement speed of the rider shown in FIG. 6A for dynamic correction, it is necessary to design the filter so that the time derivative dτ / dt of τ reflects the movement speed due to the tower vibration. FIG. 7B shows the calculated value of the time derivative of the inclination angle. The filter used in FIG. 6B is five-dimensional. By dynamically correcting the wind speed measurement value by obtaining (dτ / dt) × Lτ from τ processed using the filter having such a configuration, the influence of the tower inclination is removed and the wind speed prediction error is reduced. Becomes possible.

FIG. 8 is a diagram showing a difference in delay time depending on the filter configuration in this embodiment. As shown in FIG. 8, as the dimension of the filter increases, the delay time also increases. Therefore, the accuracy of wind speed prediction can be further improved by making a prediction in consideration of the delay time.

As described above, according to the present embodiment, the deviation between the wind speed measured by the rider and the actual wind speed is reduced by the dynamic correction of the inclination angle of the wind turbine, and the accuracy of the wind speed prediction is improved.

In the dynamic correction described in the third embodiment, Lτ is an important parameter, but its derivation needs to be changed according to the operating conditions. In FIG. 6A, the distance from the center of gravity of the floating structure 16 and the wind turbine 10 to the nacelle 3 is shown as Lτ, but in reality, as shown in FIG. 9, the length of the arm including the bending of the tower is Lτ'. It becomes. Therefore, an appropriate value can be input by providing the same user interface as in Examples 1 and 2 in order to give Lτ according to the wind turbine structure and the operating environment of the wind turbine.

In this embodiment, a more specific functional configuration of wind speed prediction including a user interface that defines a data set pattern of wind speed and inclination angle input to the prediction system will be described.

FIG. 10 is a functional configuration diagram of wind speed prediction in this embodiment. In FIG. 10, the same reference numerals are given to the configurations having the same functions as those in FIG. 1, and the description thereof will be omitted. In FIG. 10, the prediction system 6 is composed of a learner 14 and a predictor 13, and the wind speed from the rider 5 and the measured values of the tilt angle from the tilt angle sensor 15 are input to each of them. Here, the prediction system 6 can adjust the inconsistency in the required time between the learner that outputs the prediction model and the predictor that outputs the real-time prediction value by separating the learner 14 and the predictor 13.

In the above embodiment, the wind speed is predicted based on the measurement result of the rider, but since the rider measures the wind speed in front from the Doppler shift of the laser beam emitted from the rider, the environment such as rain and fog Below, there is a problem that accurate measurement cannot be performed due to the influence. In this embodiment, the countermeasures will be described.

FIG. 11 is a functional configuration diagram of wind speed prediction in this embodiment. In FIG. 11, the same reference numerals are given to the configurations having the same functions as those in FIG. 10, and the description thereof will be omitted. In FIG. 11, for example, a three-cup type conventional anemometer 18 installed in the nacelle is provided so that the wind speed signal VA of the conventional anemometer 18 and the wind speed prediction signal VP of the prediction system 6 are input to the FF control 7. Constitute. Then, in the case of rain or fog, the wind turbine operation is continued by switching the input of the FF control 7 to the output of the conventional anemometer 18. As a result, the current wind speed value is used instead of the predicted wind speed of the wind turbine position, so that the response delay of the blade pitch angle cannot be dealt with, but the operation can be continued.

Although the examples have been described above, the present invention is not limited to the above-mentioned examples, and various modifications are included. For example, the above-described embodiment has been described in detail in order to explain the present invention in an easy-to-understand manner, and is not necessarily limited to those having all the described configurations. Further, it is possible to replace a part of the configuration of one embodiment with the configuration of another embodiment, and it is also possible to add the configuration of another embodiment to the configuration of one embodiment. Further, it is possible to add / delete / replace a part of the configuration of each embodiment with another configuration. Further, each of the above configurations and functions may be realized by hardware, for example, by designing a part or all of them with an integrated circuit. When each configuration and function is realized by software, information such as programs, data, and files that realize each function should be stored not only in the memory but also in a recording device such as a hard disk or a recording medium such as an IC card. It is also possible to download and install it via a wireless network or the like as needed.

1: Blade, 2: Hub, 3: Nasser, 4: Tower, 5: Rider, 6: Prediction system, 7: Feed forward (FF) control, 8: Feedback (FB) control, 10: Wind power generator (wind turbine) , 11: Parameter setting user interface, 12: Display user interface, 13: Predictor, 14: Learner, 15: Tilt angle sensor, 16: Floating structure, 17: Rider peripheral speed due to vibration, 18: Conventional anemometer , 20: Direction of wind toward the wind turbine, 21, 22, 23, 24: LOS (Line of Sight) of the laser beam emitted from the rider

Claims (6)

A wind power generator having a tower, a nacelle provided on the tower, and blades connected to the nacelle.
The nacelle has a Doppler lidar that detects the wind speed and a control device.
The Doppler lidar has a function of emitting a laser beam in a plurality of directions and measuring wind speeds of a plurality of measurement distances along each laser beam at a predetermined measurement time.
The control device is characterized by having a wind speed prediction function that predicts the wind speed at the position of the wind power generator by inputting a data set selected from a part of the wind speed data measured by the Doppler lidar. Wind power generator. The wind power generator according to claim 1.
A wind power generator comprising one or a plurality of measurement distances and a user interface for designating one or a plurality of measurement times in order to specify a data set of input values to the wind speed prediction function. A wind power generator having a tower, a nacelle provided on the tower, and blades connected to the nacelle.
The nacelle has a Doppler lidar that detects the wind speed and a control device.
The Doppler lidar has a function of emitting a laser beam in a plurality of directions and measuring wind speeds of a plurality of measurement distances along each laser beam at a predetermined measurement time.
Further, it has an inclination angle sensor that measures the inclination angle of the tower.
The control device is a wind power generation device characterized in that the wind speed measured at the same time is corrected by using the tilt angle data measured by the tilt angle sensor. The wind power generator according to claim 3.
The control device corrects the wind speed measured by the Doppler lidar based on the tilt of the laser beam due to the tilt angle and the peripheral speed of the nacelle position obtained from the tilt angular velocity to obtain the horizontal wind speed at the position of the wind power generator. A wind turbine that is characterized by being calculated and predicted. The wind power generator according to claim 4.
The control device is a wind power generation device having a function of filtering a measured value of an inclination angle of the tower and a function of predicting the inclination angle in order to correct a delay due to the filtering. The wind power generator according to claim 4.
A wind power generator comprising a user interface that specifies an effective arm length of tilt for calculating the peripheral velocity of the nacelle position from the tilt angular velocity.

Images