Wind driven generator load spectrum data processing method

Abstract

The invention provides a wind driven generator load spectrum data processing method. The method comprises the following steps: acquiring a torque sequence; in the torque sequence, obtaining different scales of rated torque value neighborhoods by adjusting a left expansion coefficient alpha 1 and a right expansion coefficient alpha 2; equally dividing the rated torque value neighborhood and the left section interval and the right section interval outside the rated torque value neighborhood according to the rated torque Tr, the load interval number m and the concentration coefficient beta; and when cross-interval equipartition exists, re-dividing the left section interval and the right section interval by adopting a local linear interpolation method. According to the method, the dynamic torque neighborhood can be obtained by inputting the rated torque and the two expansion coefficients; the granularity of interval division can be changed in any target domain by adjusting the concentration coefficient; local linear interpolation is carried out on the left section interval and the right section interval outside the neighborhood of the rated torque value, so that the total error rate during equivalent calculation after division can be reduced.

Description

Technical Field

[0001] The invention relates to the technical field of load spectrum data processing, and in particular to a method for processing load spectrum data of a wind turbine generator. Background Art

[0002] In wind power generation systems, wind turbine gearboxes play a vital role. It is not only a key component that increases the mechanical energy captured by the wind rotor blades to the speed required by the generator through the gear system, but also needs to withstand various loads generated under complex and changeable working conditions. These loads include but are not limited to torque, horizontal bending moment, vertical bending moment, thrust, gravity, and horizontal force. In particular, for the first-stage planetary carrier in the wind turbine gearbox, as the first link in the movement and load input of the entire gearbox, its bearings must not only transmit torque, but also withstand multi-directional loads transmitted from the wind turbine main shaft. Therefore, accurately simulating and analyzing these complex force conditions is crucial to designing a gearbox that is both safe, reliable, and cost-effective.

[0003] To achieve this goal, researchers developed a method based on load spectrum (Load Duration Distribution, LDD) to predict structural fatigue life and reliability assessment. Load spectrum is a graphical data compiled by collecting or predicting random load data that the wind turbine may encounter under different operating conditions and based on certain statistical laws. This method can fully reflect the various stress states and their occurrence frequencies experienced by the equipment during actual use, providing a solid foundation for subsequent design optimization.

[0004] Specifically, it is necessary to first record the load conditions under different working conditions in the wind farm in detail or use advanced simulation technology to estimate the maximum stress value that may occur in the future; then use Miner cumulative damage theory to calculate the total damage accumulated by each component due to long-term loading; finally, combine material properties and design requirements to formulate a reasonable safety factor to ensure that the final product can meet the expected service life while having sufficient strength reserves. In addition, professional software tools such as the Romax model can be used to directly apply the generated load spectrum to the virtual environment for more accurate testing and verification, thereby further improving the rationality and operability of the design solution.

[0005] With the continuous deepening of technical capabilities, the wind turbine power generation, frame load capacity, local structural strength, and effective load range have all undergone certain changes. The original calculation program used for LDD data processing can no longer meet the current design requirements. Specifically, the original calculation program uses an average division algorithm when dividing the input load. The advantage of this algorithm is that it is fast and convenient. You only need to enter the number of intervals to be divided to complete the division; but the disadvantage is that the division granularity of the main power generation load interval and the invalid load interval is the same, resulting in a waste of more computing resources in the invalid load interval, and the estimation of the main power generation load interval is too conservative, which ultimately reduces the design margin, limits the improvement of power density to a certain extent, and cannot give full play to the advantages of the product.

[0006] In summary, the prior art has at least the following defects:

[0007] 1. When calculating the load spectrum, the load interval division method is not flexible enough, and the granularity of the interval division cannot be adjusted;

[0008] 2. The final calculation result has a large error. Summary of the invention

[0009] The purpose of the present invention is to overcome the defects of the prior art and provide a method for processing load spectrum data of a wind turbine generator, which can solve the problems existing in the prior art that the load interval division is not flexible enough and the calculation result has large errors.

[0010] In order to achieve the above and other purposes, the present invention is implemented by including the following technical solutions: The present invention proposes a method for processing load spectrum data of a wind turbine generator, comprising the steps of:

[0011] Get the torque sequence;

[0012] In the torque sequence, by adjusting the left expansion coefficient α1 and the right expansion coefficient α2, the rated torque value neighborhoods of different sizes are obtained;

[0013] According to the rated torque Tr, the number of load intervals m, and the concentration factor β, the rated torque value neighborhood and the left section interval and the right section interval outside the rated torque value neighborhood are equally divided;

[0014] When there is cross-interval equalization, the left segment interval and the right segment interval are re-divided using local linear interpolation.

[0015] In one embodiment, m is a positive integer, 0<β<1, 0≤α1, α2<1 and α1+α2>0.

[0016] In one embodiment, the left endpoint b1 of the neighborhood of the rated torque value is b1 = Tr - (Tr - T1) × α1, and the right endpoint b2 is b2 = Tr + (Tn - Tr) × α2, where T1 is the minimum value of the torque sequence and Tn is the maximum value of the torque sequence.

[0017] In one embodiment, the step of "equally dividing the neighborhood of the rated torque value and the left and right section intervals outside the neighborhood of the rated torque value" specifically includes the steps of: equally dividing the neighborhood of the rated torque value into m1 = round(m × β) parts according to m and β; equally dividing the left and right section intervals into m2 = m - m1 parts.

[0018] In one embodiment, y = round(x) represents the rounding function; after equally dividing the neighborhood of the rated torque value, m1 + 1 endpoints including b1 and b2 are obtained.

[0019] In one embodiment, the step of "re - dividing the left and right section intervals by using the local linear interpolation method" specifically includes the steps of: shifting the section from b2 to Tn left by a distance of (b2 - b1) so that b1 and b2 overlap, and equally dividing the section from T1 to [Tn - (b2 - b1)] after translation into m2 parts; finding the point Pk closest to b1 among the interval endpoints, where 1 ≤ k ≤ m2 + 1; when k = 1, the number of equal divisions of the left section interval degenerates to 0, and the right section interval is divided into m2 parts; when k = m2 + 1, the number of equal divisions of the right section interval degenerates to 0, and the left section interval is divided into m2 parts; when 1 < k < m2 + 1, the section from T1 to b1 is re - equally divided into k - 1 parts; the section from b1 to [Tn - (b2 - b1)] is equally divided into m2 - k + 1 parts, and the section from T1 to [Tn - (b2 - b1)] is shifted right by a distance of (b2 - b1) to restore the original position.

[0020] In one embodiment, the torque sequence is obtained by traversing the time - series data under all wind conditions and sorting the torque data from smallest to largest.

[0021] In one embodiment, the process of obtaining the time - series data includes the steps of: obtaining the original load data; pre - processing the original load data to obtain the time - series data under each wind condition.

[0022] In one embodiment, the method further includes the steps of: splicing the equally - divided left section interval, the neighborhood of the rated torque value, and the right section interval to obtain a complete torque division; traversing the time - series data under all wind conditions again, and synchronously putting the torque sequence and the corresponding point loads and rotational speeds in the other five directions into the corresponding load intervals; performing equivalent calculations on each load interval and outputting the results.

[0023] In one embodiment, the output result is one or more of a load duration distribution table, a load spectrum, a torque-rotation number bar graph, or a load spectrum report.

[0024] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention can obtain the rated torque value neighborhood by inputting the rated torque and two expansion coefficients, and any torque neighborhood within the upper and lower limits of the torque can be obtained by this algorithm; by adjusting the concentration coefficient, the granularity of the interval division can be changed in any target domain, so that the final divided intervals are denser in the main working condition intervals and sparser in the extreme working condition intervals; by performing local linear interpolation on the left and right intervals outside the rated torque value neighborhood, the division scheme is made as uniform as possible under the premise of giving priority to meeting the design requirements, so that the total error rate when performing equivalent calculations after division is reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 Shown is a flow chart of a method for processing load spectrum data of a wind turbine generator according to the present invention. DETAILED DESCRIPTION

[0026] The embodiment of the present invention solves the problems existing in the prior art such as inflexible load interval division and large calculation result errors by providing a method for processing load spectrum data of a wind turbine.

[0027] Below, the technical solutions in the embodiments of the present invention will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments of the present invention. It should be understood that the present invention is not limited to the example embodiments described herein. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. It should also be noted that, for the convenience of description, only the parts related to the present invention are shown in the accompanying drawings, rather than all of them.

[0028] The overall idea of ​​the technical solution provided by the present invention is as follows: A method for processing load spectrum data of a wind turbine generator comprises the following steps:

[0029] Get the torque sequence;

[0030] In the torque sequence, by adjusting the left expansion coefficient α1 and the right expansion coefficient α2, the rated torque value neighborhoods of different sizes are obtained;

[0031] According to the rated torque Tr, the number of load intervals m, and the concentration factor β, the rated torque value neighborhood and the left section interval and the right section interval outside the rated torque value neighborhood are equally divided;

[0032] When there is cross-interval equalization, the left segment interval and the right segment interval are re-divided using local linear interpolation.

[0033] The main idea is: by inputting the rated torque and two expansion coefficients, the rated torque value neighborhood can be obtained, and within the upper and lower limits of the torque, any torque neighborhood can be obtained through this algorithm; by adjusting the concentration coefficient, the granularity of the interval division can be changed in any target domain, so that the final divided intervals are denser in the main working condition interval and sparser in the extreme working condition interval; by performing local linear interpolation on the left and right intervals outside the rated torque value neighborhood, the division scheme is made as uniform as possible while giving priority to meeting the design requirements, so as to reduce the total error rate when performing equivalent calculations after division.

[0034] After introducing the basic principles of the present invention, various non-limiting embodiments of the present invention are described in detail below.

[0035] like Figure 1 As shown, the present invention provides a method for processing load spectrum data of a wind turbine generator, comprising the steps of:

[0036] S1, input time series data, rated torque Tr, load interval number m, concentration factor β, left expansion coefficient α1 and right expansion coefficient α2; where m is a positive integer, 0<β<1, 0≤α1, α2<1 and α1+α2>0;

[0037] Specifically, the process of acquiring the time series data includes the following steps:

[0038] (1) Obtaining original load data, including: on-site testing; generating and adjusting load spectra according to certain rules; considering the dynamic effects of the structure and obtaining through simulation; timing and design life provided by Party A;

[0039] (2) Preprocessing the raw load data to obtain time series data under various wind conditions, so as to facilitate subsequent data processing using data processing software (such as Matlab); the preprocessing includes data cleaning, standardization and outlier processing. The time series data includes 8 columns of data, including time, rotation speed, torque, horizontal bending moment, vertical bending moment, thrust, gravity, and horizontal force.

[0040] S2, traverse the time series data under all wind conditions, sort the torque data in ascending order, obtain the torque sequence T1, T2, ..., Tn, and calculate the range TR = Tn-T1;

[0041] S3, by dynamically adjusting the left expansion coefficient α1 and the right expansion coefficient α2, obtaining rated torque value neighborhoods of different sizes; equally dividing the rated torque value neighborhood, the left segment interval and the right segment interval outside the rated torque value neighborhood according to the rated torque Tr, the number of load intervals m and the concentration coefficient β;

[0042] Specifically, the rated torque value neighborhood is a dynamic neighborhood, and the range can be determined based on the rated torque Tr, the minimum value T1 and the maximum value Tn of the torque sequence, and the left expansion coefficient α1 and the right expansion coefficient α2, wherein the left endpoint b1 of the rated torque value neighborhood is Tr-(Tr-T1)×α1, and the right endpoint b2 is Tr+(Tn-Tr)×α2.

[0043] The method of equally dividing the torque intervals into three sections specifically includes the following steps:

[0044] S31, according to the load interval number m and the concentration factor β, the rated torque value neighborhood is divided into m1=round(m×β) parts, and after the equal division, m1+1 endpoints R1, R2, ..., R(m1+1) are obtained; wherein y=round(x) represents a rounding function, R1=b1, R(m1+1)=b2;

[0045] S32, dividing the left section and the right section outside the neighborhood of the rated torque value into m2=m-m1 parts;

[0046] S4, judging whether there is cross-interval division in step S3, if yes, proceeding to step S5; if no, proceeding to step S6;

[0047] S5, using local linear interpolation method to re-divide the left segment interval and the right segment interval so that the division intervals of the left segment interval and the right segment interval are as close as possible; after completion, proceed to step S6;

[0048] The specific process of local linear interpolation is as follows:

[0049] S51, shift the b2~Tn segment to the left by a distance of (b2-b1) so that b1 and b2 overlap, and divide the shifted T1~[Tn-(b2-b1)] segment into m2 equal parts, and obtain m2+1 endpoints P1, P2, ..., P(m2+1) after equal division;

[0050] S52, find the point Pk closest to b1 among the endpoints of the interval, 1≤k≤m2+1;

[0051] When k=1, the left endpoint of the area outside the neighborhood of the rated torque value approaches T1, the number of equal divisions of the left segment interval degenerates to 0, and the right segment interval is divided into m2 parts, and after equal division, m2+1 endpoints Q1, Q2, ..., Q(m2+1) are obtained, where Q1=R(m1+1);

[0052] When k=m2+1, the right endpoint of the area outside the neighborhood of the rated torque value approaches Tn, the number of equal divisions of the right segment interval degenerates to 0, and the left end interval is divided into m2 parts, and after equal division, m2+1 endpoints Q1, Q2, ..., Q(m2+1) are obtained, where Q(m2+1)=R1;

[0053] When 1 < k < m2 + 1, re - divide the T1 - b1 segment into k - 1 equal parts. After division, k endpoints Q1, Q2,..., Qk are obtained, where Q1 = T1 and Qk = R1; divide the b1 - [Tn-(b2 - b1)] segment into m2 - k + 1 equal parts. After division, m2 - k + 2 endpoints Qk, Q(k + 1),..., Q(m2 + 1) are obtained. Translate them to the right by a distance of (b2 - b1) to restore the original position, and get Q’k, Q’(k + 1),..., Q’(m2 + 1), where Q’k = R(m1 + 1) and Q’(m2 + 1) = Tn.

[0054] S6. Stitch the evenly - divided left - hand interval, the rated torque value neighborhood, and the right - hand interval to obtain a complete torque division;

[0055] Specifically, after the step S5 re - divides the left - hand interval and the right - hand interval, the obtained torque division results are as follows:

[0056] When k = 1, the m + 1 load interval endpoints finally obtained are T1, R2, R3,..., Rm1, R(m1 + 1) / Q1, Q2,..., Q(m2 + 1);

[0057] When k = m2 + 1, the m + 1 load interval endpoints finally obtained are Q1, Q2,..., Q(m2 + 1) / R1,..., R(m1 - 1), Rm1, Tn;

[0058] When 1 < k < m2 + 1, the m + 1 load interval endpoints finally obtained are Q1 / T1, Q2,..., Q(k - 1), Qk / R1, R2,..., Rm1, R(m1 + 1) / Q’k, Q’(k + 1),..., Q’(m2 + 1) / Tn.

[0059] S7. Traverse the time - series data under all wind conditions again, and synchronously put T1, T2,..., Tn and the corresponding point loads and rotational speeds in the other five directions into the corresponding load intervals;

[0060] S8. Perform equivalent calculations on each load interval and output the results.

[0061] Specifically, calculate the total time t, the average rotational speed rpm, and the equivalent value of the load M within each load interval range.

[0062] The total time t is the sum of the time t of all data points within the torque range corresponding to the load interval i, and the calculation formula is i the sum, and the calculation formula is

[0063]

[0064] The average rotation speed rpm is the average rotation speed of all data points within the torque range corresponding to load interval i, and the calculation formula is: Among them, c i is the speed corresponding to each load in load interval i (rad / s);

[0065] The load M is the load M of all data points within the torque range corresponding to the load interval i. i The equivalent value of is calculated as Among them, p is a parameter related to material, stress ratio, loading method, etc. Different manufacturers and different parts use different p values, which can be empirical values.

[0066] The output result can be a load duration distribution table obtained by summarizing the corresponding results of m load intervals; a load spectrum and a torque-rotation number bar graph can also be drawn based on the load duration distribution table to check whether the Romax requirements are met. If the requirements are met, a load spectrum report can be further generated.

[0067] Therefore, the present invention effectively overcomes various shortcomings in the prior art and has a high industrial utilization value. The above embodiments are only illustrative of the principles and effects of the present invention, and are not intended to limit the present invention. Anyone familiar with the technology can modify or change the above embodiments without violating the spirit and scope of the present invention. Therefore, all equivalent modifications or changes made by those with ordinary knowledge in the technical field without departing from the spirit and technical ideas disclosed by the present invention should still be covered by the claims of the present invention.

Claims

1. A method for processing load spectrum data of a wind turbine generator, characterized in that: Including the steps: Obtain the torque sequence; Within the torque sequence, by adjusting the left expansion coefficient α1 and the right expansion coefficient α2, obtain neighborhoods of rated torque values of different scales; According to the rated torque Tr, the number of load intervals m, and the concentration coefficient β, evenly divide the neighborhood of the rated torque value and the left and right section intervals outside the neighborhood of the rated torque value; When there is cross-interval equal division, use the local linear interpolation method to re-divide the left and right section intervals.

2. The method for processing wind turbine load spectrum data according to claim 1, characterized in that: m is a positive integer, 0 < β < 1, 0 ≤ α1, α2 < 1 and α1 + α2 > 0.

3. The method for processing wind turbine load spectrum data according to claim 2, characterized in that: The left endpoint b1 of the neighborhood of the rated torque value = Tr - (Tr - T1) × α1, and the right endpoint b2 = Tr + (Tn - Tr) × α2, where T1 is the minimum value of the torque sequence and Tn is the maximum value of the torque sequence.

4. The method for processing wind turbine load spectrum data according to claim 3, characterized in that: The "evenly divide the neighborhood of the rated torque value and the left and right section intervals outside the neighborhood of the rated torque value" specifically includes the steps: According to m and β, evenly divide the neighborhood of the rated torque value into m1 = round(m × β) parts; Evenly divide the left and right section intervals into m2 = m - m1 parts.

5. The method for processing wind turbine load spectrum data according to claim 4, characterized in that: y = round(x) represents the rounding function; after the neighborhood of the rated torque value is evenly divided, m1 + 1 endpoints including b1 and b2 are obtained.

6. The method for processing wind turbine load spectrum data according to claim 5, characterized in that: The "use the local linear interpolation method to re-divide the left and right section intervals" specifically includes the steps: Shift the section from b2 to Tn left by a distance of (b2 - b1) so that b1 and b2 overlap, and evenly divide the shifted section from T1 to [Tn - (b2 - b1)] into m2 parts; Find the point Pk closest to b1 among the interval endpoints, 1 ≤ k ≤ m2 + 1; When k = 1, the number of equal divisions of the left section interval degenerates to 0, and the right section interval is divided into m2 parts; When k = m2 + 1, the number of equal divisions of the right section interval degenerates to 0, and the left section interval is divided into m2 parts; When 1 < k < m2 + 1, re-evenly divide the section from T1 to b1 into k - 1 parts; evenly divide the section from b1 to [Tn - (b2 - b1)] into m2 - k + 1 parts, and shift the section from T1 to [Tn - (b2 - b1)] right by a distance of (b2 - b1) to restore the original position.

7. The method for processing wind turbine load spectrum data according to claim 1, characterized in that: The torque sequence is obtained by traversing the time series data under all wind conditions and sorting the torque data from smallest to largest.

8. The method for processing wind turbine load spectrum data according to claim 7, characterized in that: The process of obtaining the time series data includes the steps: obtain the original load data; preprocess the original load data to obtain the time series data under each wind condition.

9. The method for processing wind turbine load spectrum data according to claim 1, characterized in that: Also including the steps: Stitch the evenly divided left section interval, the neighborhood of the rated torque value, and the right section interval to obtain a complete torque division; Traverse the time series data under all wind conditions again, and synchronously put the torque sequence and the corresponding point loads and rotational speeds in the other five directions into the corresponding load intervals; Perform equivalent calculations on each load interval and output the results.

10. The method for processing load spectrum data of a wind turbine generator according to claim 9, characterized in that: The output result is one or more of a load duration distribution table, a load spectrum, a torque-rotation times histogram, or a load spectrum report.

Images