A wind turbine video clearance automatic calibration method

Abstract

The application provides a wind turbine video clearance automatic calibration method, which obtains a pixel scale through a chord length of a blade during a stable rotation of an impeller and a corresponding pixel number, and then calculates a tower cylinder projection point. The wind turbine video clearance automatic calibration method provided by the application mainly calculates a pixel scale through a chord length of a blade during a clearance calculation of a wind turbine video and a corresponding pixel number, and calculates a tower cylinder projection point through design parameters of the wind turbine and a simulation clearance. The method does not require a construction personnel to have rich experience and special calibration tools, and can conveniently and accurately and effectively complete required calibration work through parameter setting and image processing, so as to finally calculate and obtain required pixel scale and tower cylinder projection point results.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of wind turbine, in particular to a wind turbine video clearance automatic calibration method. BACKGROUND

[0002] The video clearance monitoring system is a system for monitoring the tower clearance distance of the wind turbine. The system can monitor the tower clearance distance of the wind turbine in real time, and the change trend is transmitted to the main control system of the wind turbine for control and protection.

[0003] The basic principle of the image ranging recognition algorithm of the video clearance monitoring system is generally as follows: the tip vane equivalent point is captured and recognized through image processing technology, the pixel number of the tip vane distance from the tower projection point is calculated according to the tower projection point obtained in the calibration activity, and the actual spatial distance of the tip vane from the tower projection point is calculated according to the pixel scale of the tip vane height.

[0004] At present, the pixel scale and the tower projection point are usually obtained by manual calibration on the ground, and special tools and skilled experience are needed to complete the calibration. This brings many inconveniences to the site construction, and a simple and easy automatic calibration process and method need to be developed. SUMMARY

[0005] In order to solve the problems of the prior art, the present application provides a wind turbine video clearance automatic calibration method, which mainly calculates the pixel scale by calculating the chord length of the blade passing through and the corresponding pixel point number of the video clearance calculation of the wind turbine, and calculates the tower projection point by the design parameters of the wind turbine and the simulation clearance. Without the need for construction personnel to have rich experience and special calibration tools, the required calibration work can be conveniently and accurately completed through parameter setting and image processing, and finally the required pixel scale and tower projection point results are calculated.

[0006] The technical scheme adopted by the present application to solve the technical problems is: a wind turbine video clearance automatic calibration method is provided, comprising the following steps:

[0007] S1, reading the wind turbine configuration;

[0008] S2, loading the clearance monitoring video segment to obtain the tip vane distance time sequence;

[0009] S3, calculating the impeller rotating speed according to the tip vane distance time sequence;

[0010] S4, calculating the process record matrix according to the tip vane distance time sequence and the wind turbine configuration, each row of data of the process record matrix including time, temporary pixel scale, confidence, rotating speed, rotating speed error, pixel scale error and slope;

[0011] S5, calculating the pixel scale by using the process record matrix;

[0012] S6, calculating the tower projection point in the image space according to the geometric relationship by using the wind turbine design parameters and the simulation clearance data.

[0013] The wind turbine configuration described in step S1 includes blade size parameters.

[0014] Step S2 obtains the blade tip distance time sequence through the following process:

[0015] S2.1, loading the clearance monitoring video clip;

[0016] S2.2, for each frame in the video data clip, taking the middle vertex A in the picture as the starting point and the blade tip trajectory point P as the midpoint to obtain the pixel distance value of the line segment AP;

[0017] The pixel distance value obtained by each frame is the blade tip distance time sequence.

[0018] Step S3 calculates the rotor speed through the following process:

[0019] S3.1, taking the serial number of each frame in the clearance monitoring video clip as the horizontal axis and the inverse of the pixel distance value of the line segment AP calculated for the frame as the vertical axis to make a pixel distance value inverse-number waveform graph;

[0020] S3.2, identifying the wave peaks in the pixel distance value inverse-number waveform graph and saving the pixel distance value and time data at each group of wave peaks;

[0021] S3.3, for the pixel distance value and time data at each group of wave peaks, the rotor speed rs is calculated through the following formula respectively:

[0022] rs = 1 / (num x period x 3) x 60

[0023] Where num represents the sum of the number of the blade tip trajectory point represented by the current wave peak and the consecutive blade tip trajectory points before and after the wave peak, and period represents the time interval of adjacent blade tip trajectory points.

[0024] After the rotor speed is calculated in step S3, the effectiveness of the rotor speed is judged. If the following conditions are met at the same time, it is judged that the result of the rotor speed has effectiveness:

[0025] Condition a1, the maximum error of the speed is less than 5%;

[0026] Condition a2, the speed is greater than 1.2 times the grid-connected speed and less than 0.95 times the rated speed;

[0027] Condition a3, the wave peaks in the pixel distance value inverse-number waveform graph are not less than 6;

[0028] Otherwise, return to step S2, load a new empty monitoring video clip, and obtain the blade tip distance time sequence.

[0029] In step S3.1, the effectiveness of the video acquisition point is judged. If the following conditions are met at the same time, the video acquisition point is judged to be effective:

[0030] Condition b1: among the blade tip track points used for single calculation, the positions of the consecutive blade tip track points before the wave crest increase on the Y axis, and the positions of the consecutive blade tip track points after the wave crest decrease on the Y axis;

[0031] Condition b2: the distance between adjacent blade tip track points is no more than 150%;

[0032] Condition b3: the wave crest satisfying conditions b1 and b2 is no less than 3;

[0033] Otherwise, return to step S2, load a new empty monitoring video clip, and obtain the blade tip distance time sequence.

[0034] Step S4 obtains the process record matrix through the following process:

[0035] S4.1: for each group of wave crests, consecutive blade tip track points before and after the wave crest are obtained, and the pixel distance c crossed by the blade tip is calculated;

[0036] S4.2: using the same time interval as step S4.1, the actual length C crossed by the blade tip in the empty space before and after each group of wave crests is obtained by using the cosine theorem:

[0037]

[0038] Wherein, a is the length of the blade, and rs is the rotating speed;

[0039] S4.3: the corresponding temporary pixel scale scale of each group of wave crests is calculated by the following formula:

[0040] scale=C / c

[0041] S4.4: the confidence Ci at each group of wave crests is calculated by the following formula:

[0042] Ci=1-rse*w1-p2le*w2

[0043] Wherein, rse represents the error of the impeller rotating speed, which is obtained by taking the root mean square of all impeller rotating speeds, w1 represents the error weight of the impeller rotating speed, p2le represents the error of the pixel scale, which is obtained by taking the root mean square of each group of pixel scales, and w2 represents the error weight of the pixel scale;

[0044] S4.5: the trajectory slope of the blade tip track point P is obtained by using the least square method of linear fitting;

[0045] S4.6, record the time, temporary pixel scale, confidence, rotation speed, rotation speed error, pixel scale error and slope at each group of wave peaks, and constitute a process record matrix.

[0046] Step S5 specifically comprises the following process: when the length of the process record matrix is less than a threshold value, the state word is set to 1; otherwise, the state word is set to 2; when the state word is 1, the temporary pixel scale of the process record matrix is taken as the average value as the coarse calibration pixel scale result, otherwise, the temporary pixel scale of the top 50% confidence is taken as the average value as the final calibration pixel scale result.

[0047] Step S6 calculates the tower projection point specifically comprises the following process: set p point as the tip point identified according to the image recognition technology in the locked impeller state, move the equivalent pixel point number of the pre-bending distance in the blade trajectory slope direction according to the blade design pre-bending value to find q point, the slope of straight line pq is k1, move the equivalent pixel number of the idle clearance value in the k1 vertical direction k2 according to the clearance value cl of the wind turbine when the wind turbine is idling to find o point, which is the tower projection point, and the tower projection point calibration is completed.

[0048] The present application has the beneficial effects of the technical scheme in that:

[0049] The present application uses the video clearance monitoring system to identify and calculate the image crossed by the monitored blade, uses the image recognition means in the cabin to calculate the pixel scale through the ratio of the blade rotation arc length and the pixel number, calculates the tower projection point through the geometric relationship of the image using the design and simulation data of the wind turbine, and automatically calculates the pixel scale and the tower projection point required for calibration using the design and simulation parameters of the wind turbine, which is fast and convenient, does not require special tools, and does not require personnel to have rich construction experience. BRIEF DESCRIPTION OF DRAWINGS

[0050] Figure 1 is the calibration process schematic diagram of the present application.

[0051] Figure 2 is the principle schematic diagram of the video clearance monitoring system.

[0052] Figure 3 is the tip distance schematic diagram.

[0053] Figure 4 is the pixel distance value reciprocal-time waveform diagram.

[0054] Figure 5 is the triangle side angle definition schematic diagram.

[0055] Figure 6 is the tower projection point coordinate calculation principle schematic diagram.

[0056] Figure 7is a schematic diagram of tower projection point coordinates. DETAILED DESCRIPTION

[0057] The application will be further described below in conjunction with the accompanying drawings and examples.

[0058] Referring to Figure 2 , the principle of the application is that the tower projection point is the imaging point of the horizontal direction of the blade tip in the tower in the camera, that is, the starting point of the image calculation of the clearance distance. The pixel scale is the actual spatial distance value represented by the unit pixel of the blade tip height in the image. The calculation method of the pixel scale scale is:

[0059] scale = CL / cl

[0060] wherein CL is the actual clearance value, and cl is the imaging clearance value.

[0061] Based on the above principle, the application provides a wind turbine video clearance automatic calibration method, referring to Figure 1 , comprising the following steps:

[0062] S1, read the wind turbine configuration, including blade length, blade design pre-bending value, and idling clearance value and other size parameters.

[0063] S2, load the clearance monitoring video segment, and obtain the blade tip distance time sequence. Specifically, the following process is included:

[0064] S2.1, load the clearance monitoring video segment.

[0065] S2.2, referring to Figure 3 , for each frame in the video data segment, take the middle vertex A in the picture as the starting point, and take the blade tip track point P as the midpoint to obtain the pixel distance value of the line segment AP.

[0066] The pixel distance value obtained by calculating each frame is the blade tip distance time sequence.

[0067] S3, calculate the impeller speed according to the blade tip distance time sequence. Specifically, the following process is included:

[0068] S3.1, referring to Figure 4 , take the serial number of each frame in the clearance monitoring video segment as the horizontal axis, and take the inverse number of the pixel distance value of the line segment AP calculated in the frame as the vertical axis to make a pixel distance value inverse number-serial number waveform graph.

[0069] S3.2, identify the wave crest in the pixel distance value inverse number-serial number waveform graph, and save the pixel distance value and time data at each group of wave crests.

[0070] S3.3, for the pixel distance value and time data at each group of wave crests, calculate the impeller speed rs respectively through the following formula:

[0071] rs = 1 / (num x period x 3) x 60

[0072] wherein:

[0073] rs represents the impeller speed, in rpm (revolutions per minute);

[0074] num represents the sum of the number of the blade tip trajectory point represented by the current peak and the consecutive blade tip trajectory points before and after the peak, in a dimensionless unit;

[0075] period represents the time interval of adjacent blade tip trajectory points, in s. That is, if the video is 25 frames of images per second, period is 0.04 s.

[0076] After the impeller speed is calculated, the effectiveness of the impeller speed can be judged. If the following conditions are met at the same time, it is judged that the result of the impeller speed has effectiveness:

[0077] Condition a1, the maximum error of the speed is less than 5%;

[0078] Condition a2, the speed is greater than 1.2 times the grid-connected speed and less than 0.95 times the rated speed;

[0079] Condition a3, the number of peaks in the pixel distance value-serial number waveform graph is not less than 6;

[0080] Otherwise, return to step S2, load a new empty monitoring video segment, and obtain the blade tip distance time sequence.

[0081] After the result of the impeller speed is judged to have effectiveness, the effectiveness of the video acquisition point is judged. If the following conditions are met at the same time, it is judged that the video acquisition point has effectiveness:

[0082] Condition b1, among the blade tip trajectory points used for single calculation, the positions of the consecutive blade tip trajectory points before the peak on the Y axis increase, and the positions of the consecutive blade tip trajectory points after the peak on the Y axis decrease;

[0083] Condition b2, the distance ratio between adjacent blade tip trajectory points is not more than 150%;

[0084] Condition b3, the number of peaks satisfying condition b1 and condition b2 is not less than 3;

[0085] Otherwise, return to step S2, load a new empty monitoring video segment, and obtain the blade tip distance time sequence.

[0086] S4, calculate a process record matrix according to the blade tip distance time sequence and the wind turbine configuration. Each row of data of the process record matrix includes time, temporary pixel scale, confidence, speed, speed error, pixel scale error, and slope. Specifically, the following processes are included:

[0087] S4.1, for each group of wave crest, obtain the consecutive blade tip trajectory points before and after the wave crest, and calculate the pixel distance c swept by the blade tip.

[0088] Referring to Figure 5 , given the fixed point position of a triangle, the angle γ between two edges a and b, and the lengths of the two edges, according to the cosine law:

[0089]

[0090] The length of the third edge c can be obtained. The lengths of the blades of the wind turbine are the same, a = b, so the calculation can be simplified to obtain:

[0091]

[0092] From this, step S4.2 continues to calculate.

[0093] S4.2, take the same time interval as step S4.1, and use the cosine law to obtain the actual length C swept by the blade tip in the clearance before and after each group of wave crests:

[0094]

[0095] Where a is the length of the blade, and rs is the rotational speed.

[0096] S4.3, calculate the corresponding temporary pixel scale scale at each group of wave crests by the following formula:

[0097] scale = C / c

[0098] S4.4, calculate the confidence Ci at each group of wave crests by the following formula:

[0099] Ci = 1-rse*w1-p2le*w2

[0100] Where rse represents the error of the impeller rotational speed, which is obtained by taking the root mean square of all impeller rotational speeds, w1 represents the weight of the impeller rotational speed error, which is 0.7, p2le represents the pixel scale error, which is obtained by combining the number of effective wave crests and the slope of the straight line fitting of the blade tip trajectory points, and w2 represents the weight of the pixel scale error, which is 0.3.

[0101] S4.5, obtain the trajectory slope of the blade tip trajectory point P using the least squares method of straight line fitting;

[0102] S4.6, record the time, temporary pixel scale, confidence, rotational speed, rotational speed error, pixel scale error and slope at each group of wave crests to form a process record matrix, as shown in the following table:

[0103]

[0104] Table 1 process record matrix

[0105] When the length of the process record matrix is less than a threshold value, the state word is set to 1, otherwise the state word is set to 2, when the state word is 1, the temporary pixel scale of the process record matrix takes the average as the coarse calibration pixel scale result, otherwise the temporary pixel scale of the top 50% confidence takes the average as the final calibration pixel scale result.

[0106] S5, calculate the pixel scale by using the process record matrix, specifically including the following process: when the length of the process record matrix is less than a threshold value, the state word is set to 1, otherwise the state word is set to 2, when the state word is 1, the temporary pixel scale of the process record matrix takes the average as the coarse calibration pixel scale result, otherwise the temporary pixel scale of the top 50% confidence takes the average as the final calibration pixel scale result.

[0107] S6, calculate the tower projection point in the image space according to the geometric relationship by using the wind turbine design parameters and the simulation clearance data, refer to Figure 6 , specifically including the following process: set p point as the tip point identified according to the image recognition technology in the locked impeller state, move the equivalent pixel point number of the pre-bending distance in the blade trajectory slope direction according to the blade design pre-bending value to find q point, the slope of the straight line pq is k1, move the equivalent pixel number of the idle clearance value in the k1 vertical direction k2 according to the idle clearance value of the wind turbine when the unit is idling to find o point, which is the tower projection point, complete the tower projection point calibration. The actual application is shown in Figure 7 .

[0108] The wind turbine video clearance automatic calibration method provided by the application mainly calculates the pixel scale by the chord length of the blade drawn by the unit video clearance calculation and the corresponding pixel point number, calculates the tower projection point by the unit design parameters and the simulation clearance, does not need the construction personnel to have rich experience and special calibration tools, and through parameter setting and image processing, the required calibration work is conveniently, accurately and effectively completed, and finally the required pixel scale and tower projection point results are calculated and obtained.

Claims

1. A wind turbine video clearance automatic calibration method, characterized in that The method comprises the following steps: S1, reading the wind turbine configuration; S2, loading the clearance monitoring video segment, and obtaining the blade tip distance time sequence; S3, calculating the impeller rotating speed according to the blade tip distance time sequence; S4, calculating the process record matrix according to the blade tip distance time sequence and the wind turbine configuration, wherein each row of data of the process record matrix comprises time, temporary pixel scale, confidence, rotating speed, rotating speed error, pixel scale error and slope; S5, calculating the pixel scale by using the process record matrix; S6, calculating the tower tube projection point in the image space according to the geometric relationship by using the wind turbine design parameters and the simulated clearance data.

2. The wind turbine video yaw automatic calibration method of claim 1, wherein: The wind turbine configuration in step S1 comprises blade size parameters.

3. The wind turbine video yaw automatic calibration method of claim 1, wherein: Step S2 obtains the blade tip distance time sequence through the following process: S2.1, loading the clearance monitoring video segment; S2.2, for each frame in the video data segment, taking the middle vertex A in the picture as the starting point and the blade tip track point P as the midpoint to obtain the pixel distance value of the line segment AP; S2.3, the pixel distance value obtained in each frame is the blade tip distance time sequence.

4. The wind turbine video yaw automatic calibration method of claim 3, wherein: Step S3 calculates the impeller rotating speed through the following process: S3.1, taking the serial number of each frame in the clearance monitoring video segment as the horizontal axis and the inverse of the pixel distance value of the line segment AP calculated in the frame as the vertical axis to make a pixel distance value inverse-number waveform graph; S3.2, identifying the wave peaks in the pixel distance value inverse-number waveform graph, and saving the pixel distance value and time data at each group of wave peaks; S3.3, for the pixel distance value and time data at each group of wave peaks, the impeller rotating speed rs is calculated through the following formula respectively: rs= 1 / ( num×period×3) ×60, wherein num represents the sum of the number of the blade tip track point represented by the current wave peak and the continuous blade tip track points before and after the wave peak, and period represents the time interval of adjacent blade tip track points.

5. The wind turbine video yaw automatic calibration method of claim 4, wherein: After the impeller rotating speed is calculated in step S3, the effectiveness of the impeller rotating speed is judged, and if the following conditions are met simultaneously, it is judged that the result of the impeller rotating speed has effectiveness: Condition a1, the maximum error of the rotating speed is less than 5%; Condition a2, the rotating speed is greater than 1.2 times the grid-connected rotating speed and less than 0.95 times the rated rotating speed; Condition a3, the wave peaks in the pixel distance value inverse-number waveform graph are not less than 6; Otherwise, return to step S2, load a new clearance monitoring video segment, and obtain the blade tip distance time sequence.

6. The wind turbine video yaw automatic calibration method of claim 5, wherein: After it is judged that the result of the impeller rotating speed has effectiveness, the effectiveness of the video acquisition point is judged, and if the following conditions are met simultaneously, it is judged that the video acquisition point has effectiveness: Condition b1, among the blade tip track points used for single calculation, the continuous blade tip track points before the wave peak increase in the position of the Y axis, and the continuous blade tip track points after the wave peak decrease in the position of the Y axis; Condition b2, the distance ratio between adjacent blade tip track points is not more than 150%; Condition b3, the wave peaks satisfying condition b1 and condition b2 are not less than 3; Otherwise, return to step S2, load a new clearance monitoring video segment, and obtain the blade tip distance time sequence.

7. The wind turbine video yaw automatic calibration method of claim 1, wherein: Step S4 calculates the process record matrix through the following process: S4.1, for each group of peaks, obtain the consecutive blade tip trajectory points before and after the peak, and calculate the pixel distance c swept by the blade tip; S4.2, take the same time interval as step S4.1, and use the cosine theorem to obtain the actual length C swept by the blade tip in the clearance before and after each group of peaks: ， Wherein, a is the length of the blade, rs is the rotating speed, num represents the sum of the number of the blade tip trajectory points and the consecutive blade tip trajectory points before and after the current peak, period represents the time interval of the adjacent blade tip trajectory points; S4.3, calculate the corresponding temporary pixel scale scale at each group of peaks by the following formula: ， S4.4, calculate the confidence Ci at each group of peaks by the following formula: Ci= 1- rse*w1-p2le*w2, Wherein, rse represents the error of the rotating speed of the impeller, which is obtained by taking the root mean square of all the rotating speeds of the impeller, w1 represents the weight of the error of the rotating speed of the impeller, which is 0.7, p2le represents the error of the pixel scale, which is obtained according to the number of effective peaks and the slope of the straight line fitting of the blade tip trajectory points, w2 represents the weight of the error of the pixel scale, which is 0.3; S4.5, obtain the trajectory slope of the blade tip trajectory points P by using the least square method of straight line fitting; S4.6, record the time, temporary pixel scale, confidence, rotating speed, error of the rotating speed, error of the pixel scale and slope at each group of peaks to form a process record matrix.

8. The wind turbine video yaw automatic calibration method of claim 7, wherein: Step S5 specifically Comprises the following process: when the length of the process record matrix is less than a threshold value, the state word is set to 1; otherwise, the state word is set to 2; when the state word is 1, the temporary pixel scale of the process record matrix is taken as the average value as the coarse calibration pixel scale result, otherwise, the temporary pixel scale of the top 50% of the confidence is taken as the average value as the final calibration pixel scale result.

9. The wind turbine video yaw automatic calibration method of claim 1, wherein: Step S6 of calculating the tower projection point specifically comprises the following process: set p point as the blade tip point identified according to the image recognition technology in the locked impeller state, move the equivalent pixel point number of the pre-bending distance in the blade trajectory slope direction according to the pre-bending value of the blade to find q point, the slope of the straight line pq is k1, move the equivalent pixel number of the idling clearance value in the vertical direction k2 of k1 according to the clearance value cl of the wind turbine when idling to find o point, which is the tower projection point, and complete the tower projection point calibration.

Images