An intelligent start-stop control method based on a wind turbine generator

Abstract

The application belongs to the technical field of wind turbine control, and relates to an intelligent start-stop control method based on a wind turbine generator set. A laser radar is used to scan and obtain a radial wind speed sequence at different heights in a rotating plane of a wind wheel. Through spatial interpolation and coordinate transformation, a vertical wind speed distribution profile matching the rotating plane of the wind wheel is generated. The profile is phase-locked with a real-time azimuth angle of the wind wheel, and instantaneous aerodynamic thrusts of each blade at the current azimuth and differences between the thrusts are accurately calculated. On this basis, a minimum real-time aerodynamic thrust difference value is taken as a target, independent variable pitch compensation amounts of each blade are calculated, and the compensation amounts are superimposed on a unified reference variable pitch curve to generate three independent execution variable pitch instructions. The three blades are actively and real-timely balanced in force from an aerodynamic root, periodic alternating loads caused by vertical wind shear are directly inhibited, blade flapwise bending moments and tower vibration are significantly reduced, and structural fatigue of key components is effectively alleviated.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of wind turbine control, and relates to an intelligent start-stop control method based on a wind turbine generator. BACKGROUND

[0002] The start and stop processes of a wind turbine generator are key links of operation control, and directly affect the power generation efficiency and structural safety of the unit. However, the natural wind field has significant vertical wind shear and complex flow direction in the wind wheel rotation plane, so that the local flow field of each blade is different when the wind wheel rotates, and a huge periodic alternating load is caused.

[0003] At present, the mainstream start-stop control generally adopts a unified symmetrical variable pitch strategy, that is, the same variable pitch instruction is applied to all blades, and there are significant defects: first, the unified symmetrical control can only be used for macroscopic adjustment, and cannot sense and respond to the instantaneous aerodynamic thrust difference between the blades, so that the asymmetric stress of the three blades cannot be balanced from the aerodynamic root, and the alternating load caused by wind shear continuously impacts the blades, the main shaft and the bearing, significantly aggravates the structural fatigue, and restricts the service life of the key components.

[0004] Second, in order to make up for the lack of load control ability, the prior art is forced to adopt a conservative operation strategy. Specifically, the start is delayed by increasing the cut-in wind speed threshold, or the shutdown is advanced by increasing the cut-out wind speed threshold. Although this method can obtain a certain safety margin, it essentially sacrifices the power generation at the cost of power generation, which limits the comprehensive power generation benefit of the unit. SUMMARY

[0005] In view of this, in order to solve the problems in the background art, an intelligent start-stop control method based on a wind turbine generator is provided.

[0006] The purpose of the application can be achieved by the following technical scheme: the application provides an intelligent start-stop control method based on a wind turbine generator, comprising: obtaining a radial wind speed sequence at different heights in the wind wheel rotation plane, generating a vertical wind speed distribution profile matched with the wind wheel rotation plane through spatial interpolation and coordinate transformation.

[0007] The vertical wind speed distribution profile is phase-locked with a real-time azimuth angle signal of the wind wheel, the instantaneous aerodynamic thrust of each blade at the current azimuth angle is calculated, and the real-time aerodynamic thrust difference value between the three blades is determined.

[0008] According to the start-stop instruction type and the cabin wind speed measurement value, a unified reference variable pitch angle curve is generated through mode control.

[0009] The independent variable pitch compensation amount of each blade is calculated and superimposed on the reference variable pitch angle curve to generate three independent execution variable pitch instructions.

[0010] When the preset process state condition is reached, the independent variable pitch compensation is stopped and the unified variable pitch control is transitioned to.

[0011] During the activation of the independent variable pitch control, the standard deviation reduction rate of the blade flap bending moment and the root mean square value of the tower front and rear vibration acceleration are continuously recorded, and the parameter optimization of the independent variable pitch compensation logic is performed based on the recorded data after the start-stop cycle is completed.

[0012] Compared with the prior art, the beneficial effects of the present application are as follows: (1) The present application constructs a vertical wind speed distribution profile that accurately matches the wind rotor rotation plane through laser radar active sensing technology. By phase locking this profile with the real-time azimuth angle of the wind rotor, the instantaneous aerodynamic thrust and its difference of each blade at the current azimuth are accurately calculated. On this basis, the independent variable pitch compensation amount of each blade is calculated and superimposed on the unified reference variable pitch curve with the goal of minimizing the real-time aerodynamic thrust difference. This technical path actively and real-time balances the forces of the three blades from the aerodynamic root, directly suppresses the periodic alternating load caused by vertical wind shear, and significantly reduces the blade flap bending moment and tower vibration, effectively alleviating the structural fatigue of key components.

[0013] (2) The present application has the ability to maintain the stability of the wind rotor force in complex wind fields. It generates a reference variable pitch angle curve that matches the start-stop instruction type and nacelle wind speed through modal control, and combines with the independent variable pitch compensation amount, so that the wind turbine can safely operate in wind conditions that are forced to avoid due to uneven load, thereby breaking the conservative strategy dilemma of sacrificing power generation for safety under unified variable pitch control.

[0014] (3) The present application introduces a parameter optimization closed loop based on data feedback. During the activation of the independent variable pitch control, the actual suppression effect of the independent variable pitch control on the blade load and tower vibration is continuously recorded, and the parameter optimization of the independent variable pitch compensation logic is automatically performed after the start-stop cycle is completed. This can continuously maintain the optimal control performance as the wind field characteristics change over time or the unit state changes, effectively improving the robustness and long-term effectiveness of the method. BRIEF DESCRIPTION OF DRAWINGS

[0015] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creating any creative labor.

[0016] Figure 1 To implement the method of the present application, a step flow chart is provided.

[0017] Figure 2 To generate the vertical wind speed profile matching the wind wheel rotation plane in the present application, a logical diagram is provided.

[0018] Figure 3 To generate the unified reference variable pitch angle curve in the present application, a logical diagram is provided. DETAILED DESCRIPTION

[0019] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0020] Please refer to Figure 1 As shown in the figure, the present application provides an intelligent start-stop control method based on a wind turbine generator, comprising: S1. obtaining a radial wind speed sequence at different heights in the wind wheel rotation plane, generating a vertical wind speed profile matching the wind wheel rotation plane through spatial interpolation and coordinate transformation.

[0021] In a preferred embodiment of the present application, the obtaining of the radial wind speed sequence at different heights in the wind wheel rotation plane comprises: emitting a laser beam in the direction of the incoming flow in front of the wind wheel by a laser radar device installed on the top of the nacelle, with the wind wheel rotation axis as the reference, to perform vertical fan-shaped scanning.

[0022] Based on the geometric characteristics of the wind wheel rotation plane, the height layers are divided, and the division of the height layers is aligned with the height of the center of the wind wheel hub and covers at least the range from the highest point to the lowest point of the wind wheel rotation plane. The specific division process is as follows: the wind wheel rotation plane is divided into three vertical sections along the blade span, which are the central core section, the top edge section and the bottom edge section. Among them, the central core section corresponds to the region of 60% in the middle of the blade span. This value is only a conventional example, and its value can be adjusted according to the specific blade design, for example, for blades with more concentrated load distribution, it can be expanded to 70%. For blades with more uniform distribution, it can be reduced to 50%.

[0023] The top and bottom edge sections correspond to the blade root and tip regions respectively.

[0024] The above section division is based on the fact that the aerodynamic thrust of the blade is not uniformly distributed along the span, and the middle part is the main area to generate aerodynamic force, with a large load change gradient. Therefore, the central core section needs more intensive measurement points to capture the wind shear details, while the blade root and tip regions have relatively low aerodynamic sensitivity, and can use less intensive measurement points.

[0025] Based on the above division basis, the central core area adopts a first preset span to set the height layer, the top and the bottom edge area adopt a second preset span to set the height layer, the first preset span is smaller than the second preset span, the sum of the spans of all height layers is smaller than the vertical height of the wind wheel rotation plane, and the span of a single height layer does not exceed half of the chord length at the corresponding position, which is an empirical value, and the implementer can fine-tune according to the wind conditions and measurement targets, for example, in the area with high turbulence intensity, the span of a single height layer is required to be less than 30% of the chord length at the corresponding position, the purpose is to avoid the problem of inaccurate measurement data due to too large span.

[0026] The laser radar stays at the corresponding elevation angle of each height layer, receives the backscattering signal of the height layer in a series of continuous pulse periods, and processes the received signal as follows: first, the backscattering signals of multiple pulse periods are subjected to spectral average processing.

[0027] Then, the power spectral density is calculated by fast Fourier transform, and the Doppler shift corresponding to the main peak frequency is extracted.

[0028] Finally, the Doppler shift and the known radar wavelength are substituted into the classical Doppler effect formula to solve the radial wind speed value of each pulse period. The Doppler effect formula calculation logic is: multiply the Doppler shift by the radar wavelength, take half of the product and apply a negative sign to obtain the radial wind speed value.

[0029] The mean value of the multiple radial wind speed values resolved for each height layer within the residence time is calculated to obtain the final radial wind speed of each height layer, and the radial wind speed sequence is formed in order from low to high height.

[0030] Referring to Figure 2 The vertical wind speed distribution profile matching the wind wheel rotation plane includes: establishing a two-dimensional rectangular coordinate system grid with the center of the wind wheel hub as the origin, and pre-setting the grid starting coordinate parameters and the grid spacing parameters.

[0031] Each data point in the radial wind speed sequence is mapped from the polar coordinate system of the laser radar to the corresponding grid cell of the two-dimensional rectangular coordinate system grid according to its corresponding scanning elevation angle and distance information, and the specific mapping process is as follows: the measurement point of the laser radar is defined as polar coordinates , wherein is the slant range of the laser radar to the center point of the height layer, is the horizontal azimuth angle, which is the angle between the projection of the laser beam emitted by the laser radar on the horizontal plane and the front of the laser radar, used to determine the direction of the laser beam in the horizontal direction, The vertical elevation angle refers to the angle between the laser beam emitted by the laser radar and the horizontal plane, which is used to determine the direction of the laser beam in the vertical direction. It should be noted that the horizontal azimuth angle and the vertical elevation angle are both scanning control parameters of the laser radar, which should be distinguished from the real-time azimuth angle of the wind wheel and the scanning elevation angle in the present application.

[0032] At the same time, a rectangular coordinate system with the center of the wind wheel hub as the origin is defined , The horizontal direction, represents the vertical direction.

[0033] For each data point in the radial wind speed sequence corresponding to the height layer in the polar coordinate, its coordinates in the rectangular coordinate system are calculated by the following standard geometric transformation formula group: .

[0034] wherein, and are the fixed horizontal offset and the fixed vertical offset of the laser radar installation position relative to the center of the wind wheel hub.

[0035] This formula group is the standard mathematical formula for converting polar coordinates to rectangular coordinates, which is a mature technology and will not be described in detail here.

[0036] By dividing the difference between the rectangular coordinate value and the grid starting coordinate value by the corresponding grid spacing, and performing a down-rounding operation on the calculation result, the grid cell indexes in the horizontal and vertical directions are obtained respectively.

[0037] According to the grid cell indexes in the horizontal and vertical directions, each data point in the radial wind speed sequence is assigned to the corresponding grid cell.

[0038] The radial wind speed data is converted into the vertical wind speed component through the sine function relationship of the scanning elevation angle.

[0039] The null grid cell which does not contain wind speed data is identified, and a bilinear interpolation algorithm is used to fill the data of the null grid cell based on the vertical wind speed components of the surrounding valid grid cells.

[0040] It should be noted that the execution process of the above bilinear interpolation algorithm is as follows: in the two-dimensional rectangular coordinate system grid, the coordinate positions of the null grid cell to be filled in the horizontal and vertical directions are determined.

[0041] Four adjacent valid grid cells are searched around the null grid cell as interpolation reference points, and the four valid grid cells are located in the upper left, upper right, lower left and lower right positions of the null grid cell respectively.

[0042] Based on the vertical wind speed components stored in the four effective grid cells, first, two linear interpolation calculations are performed in the horizontal direction to obtain the interpolation results of the upper and lower sides of the vertical direction of the empty grid cell, then one linear interpolation calculation is performed in the vertical direction on the two interpolation results to obtain the final vertical wind speed component at the empty grid cell, and the data filling is completed.

[0043] Taking the first linear interpolation in the horizontal direction as an example, the linear interpolation calculation logic is explained, and the subsequent calculation logic of the second linear interpolation in the horizontal direction and the linear interpolation in the vertical direction can be referred to this example, in addition, the calculation logic is prior art, so it is not described in detail here: by taking the ratio of the horizontal coordinate difference value of the empty grid cell and the first effective grid cell to the horizontal coordinate difference value of the two effective grid cells as the second weight coefficient, and taking the ratio of the horizontal coordinate difference value of the empty grid cell and the second effective grid cell to the horizontal coordinate difference value of the two effective grid cells as the first weight coefficient, the wind speed components of the two effective grid cells are weighted and summed using the two weight coefficients to obtain the first interpolation wind speed.

[0044] The grid data after interpolation filling is smoothed to obtain a vertical wind speed distribution profile matching the wind wheel rotation plane.

[0045] S2. The vertical wind speed distribution profile is phase-locked with the real-time azimuth angle signal of the wind wheel, the instantaneous aerodynamic thrust of each blade under the current azimuth angle is calculated, and the real-time aerodynamic thrust difference value between the three blades is determined.

[0046] In a preferred embodiment of the present application, the phase locking comprises.

[0047] A vertical wind speed distribution profile data stream provided by a laser radar system is received, and each frame of data is marked with a first timestamp.

[0048] At the same time, a real-time azimuth angle signal data stream from the wind wheel encoder is received, and each azimuth angle data is marked with a second timestamp.

[0049] The first timestamp and the second timestamp are aligned on a unified time axis.

[0050] The sum of the processing delay of the laser radar and the signal transmission delay constitutes a fixed delay, and the vertical wind speed distribution profile data is time-lag compensated to match the actual spatial position of the wind wheel at the current time, and the phase locking is completed.

[0051] It should be noted that the processing delay refers to the fixed time required for the laser radar to output radial wind speed data from the internal signal processing circuit after emitting a laser pulse, which is explicitly given by the equipment manufacturer in the product technical specification.

[0052] The signal transmission delay refers to a fixed time length introduced by a communication link for transmitting the wind speed data from the laser radar device to the main controller, which is determined by a control system hardware architecture and a communication protocol.

[0053] It should be further noted that the specific process of the time delay compensation is as follows: a fixed time interval required for the laser radar to output a complete vertical wind speed distribution profile is defined as a data refresh period.

[0054] The fixed delay value is subjected to a ratio operation with the data refresh period, and an upward rounding processing is performed on the operation result to obtain the number of data frames that need to be called in advance.

[0055] In the data buffer queue, based on the calculated number of data frames, the corresponding historical data frames are indexed as the vertical wind speed distribution profile data used in the current control period, so as to match the actual spatial position of the wind wheel at the current time.

[0056] In a preferred embodiment of the present application, the calculation of the instantaneous aerodynamic thrust of each blade at the current azimuth angle includes: determining the real-time spatial posture of each blade in the wind wheel rotation plane according to the real-time azimuth angle signal and the fixed phase difference of each blade relative to the wind wheel hub.

[0057] Each blade is divided into a plurality of micro-element segments along the spanwise direction, and the coordinate unit of the center point of each micro-element segment in the two-dimensional rectangular coordinate system network with the wind wheel hub as the origin under the real-time spatial posture is retrieved, so as to index and extract the local vertical wind speed component acting on the micro-element segment from the vertical wind speed distribution profile subjected to time delay compensation.

[0058] The local vertical wind speed component is combined with the tangential wind speed vector generated by rotation to obtain the local resultant wind speed vector of the micro-element segment and the corresponding aerodynamic angle of attack.

[0059] Based on the aerodynamic angle of attack, the thrust component of each micro-element segment in the direction of the main shaft of the wind wheel is determined.

[0060] The thrust components of all micro-element segments of each blade are subjected to an integral operation along the spanwise direction to output the instantaneous aerodynamic thrust of the blade at the current azimuth angle.

[0061] In a preferred embodiment of the present application, the determination of the real-time aerodynamic thrust difference value among the three blades includes: extracting the instantaneous aerodynamic thrust of the three blades at the current time, calculating the arithmetic mean value of the three blades as the real-time reference thrust representing the current overall force state of the wind wheel.

[0062] The instantaneous aerodynamic thrust of each blade is subjected to a difference operation with the real-time reference thrust to obtain the real-time thrust deviation value of each blade, respectively.

[0063] According to the real-time azimuth angle of the wind wheel, azimuth weight coefficients are assigned to real-time aerodynamic thrust difference values of three blades, and through linear weighted fusion, real-time aerodynamic thrust difference values between the three blades are obtained.

[0064] The basis for assigning the weight coefficients according to the real-time azimuth angle of the wind wheel is that, in the process of rotation of the wind wheel, the thrust difference of the blades at different azimuth angles has different influences on the load of the whole machine. For example, when the blades are in a horizontal position, the thrust difference is directly converted into overturning moment, which has a significant influence on the fatigue load of the tower. When the blades are in a vertical position, the thrust difference is mainly converted into yawing moment, which has a relatively small influence on the tower. Therefore, in order to more effectively reduce the load of the whole machine, the present application assigns weights to the thrust difference according to the azimuth angle, so that the thrust difference at the key position such as the horizontal position has a higher weight, so that the control strategy will preferentially reduce the thrust difference at this key position.

[0065] The assignment of the azimuth weight coefficients follows the following principles: the closer the azimuth angle of the blade in the rotation plane of the wind wheel is to the horizontal direction, the greater the weight coefficient assigned to the corresponding real-time thrust deviation value. The closer the azimuth angle is to the vertical direction, the smaller the weight coefficient assigned to the corresponding real-time thrust deviation value. The following gives an example of a design function of the azimuth weight coefficient: .

[0066] wherein is the azimuth weight coefficient assigned to the current azimuth angle of the wind wheel, is a reference weight coefficient, is a preset modulation amplitude, is the current azimuth angle of the wind wheel.

[0067] is a cosine function term for ensuring that the weight is maximum at the horizontal position and minimum at the vertical position.

[0068] S3. According to the start-stop instruction type and the cabin wind speed measurement value, a unified reference variable pitch angle curve is generated through mode control.

[0069] As shown in Figure 3 , in a preferred embodiment of the present application, generating a unified reference variable pitch angle curve includes: determining a control mode according to the start-stop instruction type, the control mode being divided into a start mode and a shutdown mode.

[0070] If it is the start-up mode: determine the pitch angle basic range according to the interval to which the cabin wind speed measurement belongs, the technical basis of which is to maintain the wind wheel as close as possible to the optimal tip speed ratio during the start-up acceleration process to optimize the aerodynamic efficiency. The specific determination process is: divide the working wind speed range into multiple continuous intervals, and based on the optimal tip speed ratio and the current wind speed interval value, calculate the corresponding wind wheel rotational linear speed target value.

[0071] For the blade preset spanwise position, vector synthesis the linear speed target value and the local wind speed from the vertical wind speed distribution profile to obtain the combined wind speed direction of the position.

[0072] The target is to make the chord direction of the blade reach the aerodynamic angle of attack corresponding to the optimal tip speed ratio with the combined wind speed direction, and the required theoretical reference pitch angle is deduced.

[0073] Based on the aerodynamic calculation error and the span factor of the wind speed interval, a margin is added to the center value of the theoretical reference pitch angle to determine the pitch angle basic range.

[0074] Exemplarily, considering the aerodynamic calculation error, a basic margin of ±2° is set, and the product of the wind speed interval span and 0.1 is taken as the adjustment margin, and the basic margin and the adjustment margin are combined to obtain the pitch angle basic range of the wind speed interval. It needs to be supplemented that the basic margin of ±2° and the adjustment coefficient of 0.1 are based on the technical experience value provided by the wind turbine manufacturer, and the inventor can customize it according to the actual specifications and application scenarios of the wind turbine.

[0075] According to the difference between the grid-connected speed and the real-time speed of the generator, the adjustment strategy is selected, and the specific process is: when the difference is greater than a first preset threshold, the aggressive adjustment strategy is adopted, and the lower limit value of the pitch angle basic range is selected as the reference pitch angle.

[0076] When the difference is less than a second preset threshold, the conservative adjustment strategy is adopted, and the upper limit value of the pitch angle basic range is selected as the reference pitch angle.

[0077] When the difference is between the first preset threshold and the second preset threshold, the balanced adjustment strategy is adopted, and the median value of the pitch angle basic range is selected as the reference pitch angle.

[0078] The first preset threshold and the second preset threshold are derived from the systematic simulation analysis and field test calibration of the start-up acceleration dynamic characteristics of the target wind turbine.

[0079] Determination of the first preset threshold: By simulating the starting process of the unit at different initial wind speeds in the simulation model, the speed-up curve is analyzed. The boundary value is set as the maximum allowable speed difference that can ensure that the unit will not cause significant impact on the transmission chain torque or power due to excessive aerodynamic thrust during the process of accelerating from the current speed to the grid-connected speed under typical wind conditions. Usually, this value is quantified as a percentage range of the instructed grid-connected speed, for example, 15% to 25%.

[0080] Determination of the second preset threshold: By analyzing the speed fluctuation data at the grid-connected moment, the boundary value is set as a small enough speed difference to ensure that after switching to the conservative strategy, the unit can smoothly and without overshoot reach the grid-connected speed within an acceptable short time. Usually, this value is quantified as a smaller percentage range of the instructed grid-connected speed, for example, 3% to 5%.

[0081] Rate limiting processing of the reference variable pitch angle: Calculate the pitch angle change rate of the current control period relative to the previous control period, compare the difference with the preset maximum allowable change rate, and when the difference exceeds the maximum allowable change rate, limit the pitch angle adjustment amount to the adjustment amount corresponding to the maximum allowable change rate, otherwise maintain the original pitch angle adjustment amount. Output the reference variable pitch angle sequence after the rate limiting processing to form a unified reference variable pitch angle curve in the starting phase.

[0082] If it is a shutdown mode: call the predefined shutdown feathering curve as the reference profile, which specifies the trajectory from the current pitch angle to the fully feathered safe angle.

[0083] It should be noted that each wind turbine has been set by the manufacturer with a shutdown feathering curve when it leaves the factory, and the calling of the predefined shutdown feathering curve is a common knowledge and a conventional design choice for those skilled in the art.

[0084] Proportionally adjust the feathering rate of the reference profile according to the current nacelle wind speed measurement. Specifically, multiply the current nacelle wind speed measurement by the preset rate adjustment coefficient to obtain the feathering rate of the reference profile, wherein the preset rate adjustment coefficient is obtained based on the simulation analysis and experimental calibration of the aerodynamic load and structural dynamic response of the target wind turbine during shutdown process under different wind speed conditions.

[0085] The adjusted feathering curve is subjected to acceleration limiting processing: a current control period relative to a previous control period is obtained, a pitch angle change amount is calculated based on a control period length, a real-time pitch angle change rate is calculated, a differential operation is performed on the change rate to obtain an acceleration value of the pitch angle, the acceleration value is compared with a preset maximum allowed acceleration, when the acceleration value exceeds the maximum allowed acceleration, the feathering curve is subjected to a smoothing filtering processing, so that the acceleration is always within the allowed range, and a unified reference pitch angle curve in the shutdown stage is formed.

[0086] Specifically, the change rate limiting processing of the starting mode and the acceleration limiting processing of the shutdown mode correspond to the application logic of the respective stages. The change rate limiting is a speed, which defines the maximum allowed change speed of the pitch angle, and the core goal of the wind turbine starting stage is to accelerate smoothly. In order to avoid power and torque impact, the change rate limiting is used to ensure smooth pitch action and prevent sudden changes in pitch angle from causing a dramatic change in the power captured by the wind wheel, thereby achieving soft start.

[0087] The acceleration limiting is the change speed of the pitch angle, which defines the upper limit of the change rate of the change speed of the pitch angle. When the wind turbine is feathered, the pitch system needs to act quickly to unload the load, at which time a high change rate is adopted. However, if the feathering starts or stops too violently, i.e., the acceleration is too large, it will cause impact on the mechanical structure of the pitch bearing, gear and the like. Therefore, the acceleration limiting is to protect the mechanical parts of the pitch mechanism and ensure that the high-speed feathering action can start and end smoothly.

[0088] S4. A pitch compensation amount for each blade is calculated based on the reference pitch angle curve, and three independent execution pitch instructions are generated.

[0089] In order to accurately suppress the periodic and asymmetric load caused by the complex flow field such as vertical wind shear in the wind wheel rotation plane, the conventional unified pitch control cannot effectively cope with the force difference of the blades in different spatial orientations. Therefore, the present application divides the wind wheel rotation plane into multiple azimuth angle intervals, establishes a mapping relationship between the spatial orientation and the compensation strategy, and thus realizes directional and closed-loop compensation of load unevenness.

[0090] Based on this, in a preferred embodiment of the present application, the pitch compensation amount calculation process for each blade includes: dividing the wind wheel rotation plane into multiple azimuth angle intervals.

[0091] During the independent pitch control activation period, any azimuth angle interval is selected, and a tentative pitch compensation disturbance is applied only to the blades in the interval.

[0092] The tentative variable-pitch compensation disturbance is a small signal disturbance, and an initial amplitude thereof is preset by a developer according to a unit type and wind condition characteristics. The amplitude is strongly related to a wind wheel diameter and rated power of the unit, and a determination principle thereof is that, for a medium or large wind turbine unit, a smaller disturbance amplitude needs to be used to avoid excessive load excitation due to high aerodynamic sensitivity of blades and large structural inertia. For a small wind turbine unit, a larger disturbance amplitude can be allowed to ensure a signal-to-noise ratio due to relatively stronger structural rigidity. Therefore, an implementer should define the amplitude based on dynamic characteristics of a specific application object under the principle.

[0093] A change amount of the aerodynamic thrust difference value before and after the disturbance is applied is measured, a ratio operation is performed on the change amount and the amplitude of the tentative variable-pitch compensation disturbance to obtain a local gradient estimation value of the interval, and a product of the local gradient estimation value and a preset learning rate coefficient is subtracted from the original compensation amount to update the independent variable-pitch compensation amount of the corresponding blade.

[0094] With rotation of the wind wheel, switching to a next azimuth interval in sequence is performed, and the tentative variable-pitch compensation disturbance application and the independent variable-pitch compensation amount updating are repeatedly executed until the aerodynamic thrust difference value is lower than a preset convergence threshold value, or after a complete wind wheel rotation cycle is completed, the optimization cycle is stopped, and the current independent variable-pitch compensation amount of each blade is output.

[0095] The current independent variable-pitch compensation amount of each blade is superimposed on the reference variable-pitch angle curve, and whether the superimposed angle exceeds a variable-pitch mechanical limit is verified.

[0096] If not, three independent execution variable-pitch instructions are generated.

[0097] If yes, the independent variable-pitch compensation amounts are scaled according to an equal proportion scaling rule under the premise of maintaining a relative proportion of the compensation amounts, until all safety constraints are met.

[0098] The equal proportion scaling rule execution content is as follows: total variable-pitch angles of the blades after the independent variable-pitch compensation amounts are superimposed are calculated, and a blade with the largest amplitude exceeding the variable-pitch mechanical limit is identified as a main scaling object.

[0099] A remaining compensation space of the main scaling blade is determined according to a difference between the variable-pitch mechanical limit and a current reference variable-pitch angle, and the remaining compensation space is divided by the current independent variable-pitch compensation amount of the main scaling blade, and a ratio value obtained is a scaling factor under a current cycle.

[0100] The independent variable-pitch compensation amounts of the three blades are scaled according to the calculated scaling factor to obtain safety compensation amounts after scaling.

[0101] It should be noted that the initial value determination of the preset learning rate coefficient and the preset convergence threshold is realized by systematic calibration in a simulation environment, and the specific process is as follows: in the simulation platform, a test sequence covering the typical operating range of the unit is constructed. The sequence contains multiple wind condition scenarios with different characteristic wind speeds, turbulence intensities and vertical wind shear profiles, each of which is sorted from high to low according to the known equivalent fatigue load caused by the wind wheel structure, forming a reference fatigue load sorting sequence.

[0102] Initialize multiple different parameter combinations containing learning rate coefficients and convergence thresholds, for each parameter combination, perform closed-loop simulation on the reference test wind condition sequence, and perform the following evaluation process: record the equivalent fatigue load of the blade root flap bending moment and the tower front and rear vibration under independent variable pitch control during the entire simulation period.

[0103] Compare the equivalent fatigue load with the reference fatigue load obtained by using the unified variable pitch control strategy under the same wind condition sequence, calculate the load reduction rate, and sort all parameter combinations under the test wind condition from high to low according to the load reduction rate. Calculate the Spearman rank correlation coefficient between the sorting and the reference fatigue load sorting sequence to obtain the load reduction performance score of each parameter combination. The higher the load reduction performance score, the stronger the ability of the parameter combination to maintain the expected load control priority under different wind conditions.

[0104] Statistically analyze the cumulative travel and rate change frequency of the variable pitch actuator during simulation, normalize them, add them first and then take the inverse, to obtain the control stability performance score of each parameter combination.

[0105] Multiply the load reduction performance score and the control stability performance score to obtain the evaluation index of each parameter combination.

[0106] After completing the simulation evaluation of all parameter combinations, traverse and select the parameter combination with the highest evaluation index as the initial parameters in the actual control application.

[0107] The embodiment of the application has the ability to maintain the stability of the wind wheel under complex wind field, which generates a reference variable pitch angle curve matched with the start and stop instruction type and the nacelle wind speed through modal control, and combines the independent variable pitch compensation amount, so that the wind turbine can safely operate under the wind condition forced to avoid due to uneven load, thereby breaking the conservative strategy dilemma of sacrificing power generation for safety under unified variable pitch control.

[0108] S5. When the preset progress state condition is reached, stop the independent variable pitch compensation and transition to the unified variable pitch control.

[0109] In a preferred embodiment of the present application, the preset process state condition comprises any of the following: I. Start-up process: the generator speed reaches the grid-connected speed, and continues to run for more than a preset time period after successful grid connection.

[0110] II. Shutdown process: the generator speed drops to a safe speed allowing mechanical braking.

[0111] III. Receiving an independent variable pitch fault alarm signal or monitoring sensor data failure.

[0112] The monitoring sensor data failure includes monitoring the physical range continuously exceeding the range or fixing a certain constant value, or cross-verification inconsistency with other associated measurement values of the unit, for example, if the tower cylinder vibration acceleration sensor shows a value of zero or very low, but at the same time the nacelle anemometer shows that it is currently in a high wind speed interval, and the generator power output is normal, it is determined that the vibration sensor data is not reliable.

[0113] In the embodiment of the present application, a vertical wind speed distribution profile that is accurately matched with the wind wheel rotation plane is constructed by laser radar active sensing technology. By phase locking the profile with the real-time azimuth angle of the wind wheel, the instantaneous aerodynamic thrust of each blade at the current azimuth and its difference are accurately calculated. On this basis, the independent variable pitch compensation amount of each blade is calculated and superimposed to a unified reference variable pitch curve, with the minimum real-time aerodynamic thrust difference value as the target. This technical path actively and real-timely balances the stress of the three blades from the aerodynamic root, directly inhibits the periodic alternating load caused by vertical wind shear, thereby significantly reducing the blade flapwise bending moment and tower cylinder vibration, and effectively alleviating the structural fatigue of key components.

[0114] S6. During the activation of the independent variable pitch control, the standard deviation reduction rate of the blade flapwise bending moment and the root mean square value of the tower cylinder front and rear vibration acceleration are continuously recorded, and the parameter optimization of the independent variable pitch compensation logic is performed based on the recorded data after the start-stop cycle ends.

[0115] In a preferred embodiment of the present application, the parameter optimization of the independent variable pitch compensation logic comprises: according to the root mean square value of the tower cylinder front and rear vibration acceleration, the root mean square value of the vibration acceleration after the independent variable pitch control is subtracted from the root mean square value of the vibration acceleration before the control, and divided by the root mean square value of the vibration acceleration before the control, to obtain the root mean square value reduction rate.

[0116] The standard deviation reduction rate and the root mean square value reduction rate are linearly weighted and fused to obtain a comprehensive performance score of the independent variable pitch compensation logic.

[0117] The comprehensive performance score is compared with the performance score of the historical start-stop cycle in terms of trend, and if the current cycle score is lower than the previous cycle score, a parameter optimization instruction is generated, and the parameters include the heuristic variable pitch compensation disturbance, the learning rate coefficient and the convergence threshold in the compensation logic.

[0118] The adjusted independent pitch compensation logic is applied to the next start-stop cycle, and relevant data is recorded repeatedly.

[0119] If the performance data is better than before adjustment, it is confirmed that the parameter optimization is effective, the current parameter configuration is locked, otherwise a parameter reset instruction is triggered and the optimization process is restarted.

[0120] The embodiment of the present application introduces a parameter optimization closed loop based on data feedback. During the activation of independent pitch control, the actual suppression effect of independent pitch control on blade load and tower vibration is recorded continuously, and the independent pitch compensation logic is automatically optimized after the end of the start-stop cycle according to the actual suppression effect, so that the optimal control performance can be continuously maintained as the long-term changes of the wind farm characteristics or the changes of the unit state, and the robustness and long-term effectiveness of the method are effectively improved.

[0121] The above is only an example and description of the concept of the present application. Those skilled in the art can make various modifications or supplements to the described specific embodiments or use similar ways to replace them, as long as they do not deviate from the concept of the present application or exceed the scope defined by the present application, which should belong to the protection scope of the present application.

Claims

1. A method for intelligent start-stop control based on a wind turbine generator, characterized in that, The method comprises the following steps: Obtaining a radial wind speed sequence at different heights in the wind wheel rotation plane, and generating a vertical wind speed distribution profile matched with the wind wheel rotation plane through spatial interpolation and coordinate transformation; Phase-locked the vertical wind speed distribution profile with the real-time azimuth angle signal of the wind wheel, calculated the instantaneous aerodynamic thrust of each blade at the current azimuth angle, and determined the real-time aerodynamic thrust difference value among the three blades; According to the start-stop instruction type and the cabin wind speed measurement value, a unified reference variable pitch angle curve is generated through mode control; Taking the minimization of the real-time aerodynamic thrust difference value as the target, the independent variable pitch compensation of each blade superimposed on the reference variable pitch angle curve is calculated, and three independent execution variable pitch instructions are generated; When the preset progress state condition is reached, stop the independent variable pitch compensation and transition to unified variable pitch control; During the activation of the independent variable pitch control, the standard deviation reduction rate of the blade flap bending moment and the root mean square value of the tower front and rear vibration acceleration are continuously recorded, and the parameter optimization of the independent variable pitch compensation logic is carried out based on the recorded data after the start-stop cycle is completed; The independent variable pitch compensation calculation process of each blade includes: dividing the wind wheel rotation plane into multiple azimuth angle intervals; during the activation of the independent variable pitch control, selecting any azimuth angle interval and only applying a tentative variable pitch compensation disturbance to the blades in the interval; measuring the change amount of the aerodynamic thrust difference value before and after the disturbance is applied, calculating the local gradient estimate value of the interval, updating the independent variable pitch compensation of the corresponding blade according to the preset learning rate coefficient; with the rotation of the wind wheel, sequentially switch to the next azimuth angle interval and repeat the execution of the tentative variable pitch compensation disturbance and the update of the independent variable pitch compensation until the aerodynamic thrust difference value is lower than the preset convergence threshold, or after completing an optimization cycle of a complete wind wheel rotation period, stop calculating and output the current independent variable pitch compensation of each blade; superimpose the current independent variable pitch compensation of each blade on the reference variable pitch angle curve, and verify whether the superimposed angle exceeds the variable pitch mechanical limit; if not, generate three independent execution variable pitch instructions; if it does, clip it according to the equal proportion scaling rule while maintaining the relative proportion of each compensation, until all safety constraints are met.

2. The intelligent start-stop control method based on a wind turbine generator according to claim 1, characterized in that, The method comprises the following steps: A laser radar device is installed on the top of the cabin, and a laser beam is emitted in the direction of the incoming flow in front of the wind wheel for vertical fan-shaped scanning with the wind wheel rotation axis as the reference; Based on the geometric characteristics of the wind wheel rotation plane, the height layers are divided, and the height layers are aligned with the height of the wind wheel hub center and cover at least the range from the highest point to the lowest point of the wind wheel rotation plane; The laser radar device stays at the corresponding elevation angle of each height layer, receives the backscattering signal of the height layer in a series of continuous pulse periods, obtains the radial wind speed of each height layer through Doppler shift analysis, and forms a radial wind speed sequence in order from low to high.

3. The intelligent start-stop control method based on a wind turbine generator according to claim 2, characterized in that, The method comprises the following steps: A two-dimensional rectangular coordinate system grid with the wind wheel hub center as the origin is established; mapping each data point in the radial wind speed sequence from a polar coordinate system of the laser radar device to a corresponding grid cell in the two-dimensional rectangular coordinate system grid according to its corresponding scanning elevation angle and distance information, and converting the radial wind speed data into a vertical direction wind speed component; identifying a null grid cell that does not contain wind speed data, and filling data in the null grid cell based on vertical direction wind speed components of surrounding valid grid cells using a bilinear interpolation algorithm; performing smoothing processing on the interpolated and filled grid data to obtain a vertical wind speed distribution profile that matches a wind wheel rotation plane.

4. The intelligent start-stop control method based on a wind turbine generator according to claim 3, characterized in that, the phase locking comprises; timestamp aligning the vertical wind speed distribution profile data with a real-time azimuth angle signal of the wind wheel; performing time lag compensation on the vertical wind speed distribution profile data based on a fixed delay of the laser radar device, so that the vertical wind speed distribution profile data matches an actual spatial position of the wind wheel at a current time, and phase locking is completed.

5. A method of intelligent start-stop control based on a wind turbine generator according to claim 4, characterized in that, the calculating of the instantaneous aerodynamic thrust of each blade at the current azimuth angle comprises: determining a real-time spatial posture of each blade in the wind wheel rotation plane according to the real-time azimuth angle signal and a fixed phase difference of each blade relative to the wind wheel hub; dividing each blade into a plurality of micro-element segments along a spanwise direction, and retrieving a coordinate cell of a center point of each micro-element segment in a two-dimensional rectangular coordinate system network with the wind wheel hub as an origin at the real-time spatial posture, so as to index and extract a local vertical wind speed component acting on the micro-element segment from the vertical wind speed distribution profile after time lag compensation; combining the local vertical wind speed component with a tangential wind speed vector generated by rotation to obtain a local resultant wind speed vector of the micro-element segment and a corresponding aerodynamic angle of attack, in combination with a real-time rotation speed vector of the wind wheel; determining a thrust component of an aerodynamic force on each micro-element segment in a main shaft direction of the wind wheel based on the aerodynamic angle of attack; performing an integral operation on the thrust components of all micro-element segments of each blade along the spanwise direction to output the instantaneous aerodynamic thrust of the blade at the current azimuth angle.

6. The intelligent start-stop control method based on a wind turbine generator according to claim 1, wherein, the determining of the real-time aerodynamic thrust difference value among the three blades comprises: extracting instantaneous aerodynamic thrusts of the three blades at a current time, calculating an arithmetic mean value of the three blades as a real-time reference thrust representing a current overall force state of the wind wheel; performing a difference operation on the instantaneous aerodynamic thrusts of the three blades and the real-time reference thrust to obtain real-time thrust deviation values of the three blades, respectively; distributing azimuth angle weight coefficients to the real-time aerodynamic thrust difference values of the three blades according to a real-time azimuth angle of the wind wheel, and obtaining the real-time aerodynamic thrust difference values among the three blades through linear weighted fusion.

7. The intelligent start-stop control method based on a wind turbine generator according to claim 1, wherein, the generating of the unified reference variable pitch angle curve comprises: determining a control mode according to a start-stop instruction type, the control mode being divided into a start mode and a shutdown mode; if it is the start mode: determining a variable pitch angle basic range according to an interval to which a nacelle wind speed measurement value belongs; selecting an adjustment strategy according to a difference between an instruction grid-connected speed and a real-time generator speed to obtain a reference variable pitch angle; performing a change rate limiting processing on the reference variable pitch angle to form a unified reference variable pitch angle curve in a start stage; if it is the shutdown mode: A predefined shutdown feathering curve is called as a reference profile, which defines a trajectory to transit from a current pitch angle to a fully feathered safety angle; The feathering rate of the reference profile is proportionally adjusted according to a current cabin airspeed measurement; An acceleration limit is applied to the adjusted feathering curve to form a unified reference pitch angle curve in the shutdown phase.

8. The intelligent start-stop control method based on a wind turbine generator according to claim 1, wherein, The preset process state condition includes any of the following: I. Start-up process: the generator speed reaches the grid-connected speed, and continues to run for more than a preset time after successful grid connection; II. Shutdown process: the generator speed drops to a safe speed allowing mechanical braking; III. Receiving an independent pitch fault alarm signal or monitoring sensor data failure.

9. The intelligent start-stop control method based on a wind turbine generator according to claim 1, wherein, The parameter optimization of the independent pitch compensation logic includes: According to the standard deviation drop rate and the root mean square value, the comprehensive performance score of the independent pitch compensation logic is calculated; The comprehensive performance score is compared with the performance score of the historical start-stop cycle in terms of trend. If the current cycle score is lower than the previous cycle score, a parameter optimization instruction is generated, and the parameters include the heuristic pitch compensation disturbance, the learning rate coefficient, and the convergence threshold in the compensation logic; The adjusted independent pitch compensation logic is applied to the next start-stop cycle, and the relevant data is repeatedly recorded; If the performance data is better than before adjustment, it is confirmed that the parameter optimization is effective, and the current parameter configuration is locked. Otherwise, a parameter reset instruction is triggered, and the optimization process is restarted.

Images