A rain-flow counting based time series load reconstruction method

Abstract

The application discloses a time sequence load reconstruction method based on rain flow counting and belongs to the field of engineering technology load spectrum compilation. The steps of the method comprise the following steps: preprocessing of load data; extraction of load peak and valley values; resampling of load data; rain flow counting of load data; division of rain flow matrix according to large and small load amplitudes; elimination of small-amplitude load area; reconstruction of residual load; and evaluation of time sequence load. The application proposes a new time sequence load signal reconstruction method, eliminates load cycles with small damage contribution of original load signals, reconstructs into new time sequence load cycles, shortens the length of load data, and provides a new method for rig accelerated test.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of engineering load spectrum compilation, and more particularly to a time sequence load reconstruction method based on rainflow counting. BACKGROUND

[0002] At present, load signal editing technology has been widely used in aerospace, vehicles, engineering machinery, railways, new energy and other fields. Load spectrum is the basis of fatigue durability analysis and life prediction in engineering. Since mechanical parts bear complex random loads in actual working environment, the size of these loads directly affects the reliability and service life of mechanical structures. The ultimate goal of load signal editing is to truly reproduce the actual loading conditions of mechanical structures on the test bench, quickly find the defects and deficiencies of product design through accelerated fatigue testing, and timely improve them to shorten the research and development cycle and reduce the research and development cost.

[0003] When conducting indoor test bench accelerated fatigue test, test load spectrum for loading is needed as excitation signal. In the existing technology, load spectrum compilation methods include load extrapolation method, strengthened load amplitude method, increased test load frequency method, equivalent multi-level load method and time domain signal editing method. Strengthened load amplitude is directly multiplied by an amplification factor to improve the fatigue damage of single loading to the structure. The amplified load cannot change the failure mode of the structure. Increasing the frequency of the test load is to compress the original signal in the time scale, ensuring that the total energy of the signal remains unchanged under the premise of improving the frequency of the load. The equivalent multi-level load method is based on the damage equivalence principle to convert the measured time domain data into a multi-level constant amplitude load block spectrum. The time domain signal editing method mainly extracts the load peak and valley values and filters small amplitude load cycles.

[0004] In the above methods, the time length of the load signal cannot be effectively shortened. Therefore, how to effectively shorten the time length of the load signal and realize the test bench accelerated test has become a problem to be solved. SUMMARY

[0005] Therefore, the present application provides a time sequence load reconstruction method based on rainflow counting, which aims to shorten the time length of the load by eliminating the cycles in the load signal that cause less damage to the structure and reconstructing the time sequence load, so as to realize the test bench accelerated test.

[0006] In order to achieve the above purpose, the present application adopts the following technical solutions:

[0007] The present application provides a time sequence load reconstruction method based on rainflow counting, comprising the following steps:

[0008] S1, obtaining the measured load signal of the target mechanical structure through a data acquisition instrument and pre-processing;

[0009] S2, extracting load peak and valley values from the pretreated load signal;

[0010] S3, resampling the signal of the extracted load peak and valley values;

[0011] S4, performing rain flow counting according to the resampled extracted load peak and valley values to obtain a two-dimensional rain flow matrix;

[0012] S5, partitioning the two-dimensional rain flow matrix according to load amplitude to obtain a small-amplitude load area and

[0013] a large-amplitude load area;

[0014] ;

[0015] S6, eliminating the small-amplitude load area according to a preset condition;

[0016] S7, reconstructing the remaining load in the two-dimensional rain flow matrix.

[0017] Further, the load signal preprocessing in step S1 includes:

[0018] S101, differentiating the load signal to identify a high slope area; the high slope area is a burr segment, and the burr signal is eliminated by smoothing processing;

[0019] S102, using a Butterworth high-pass filtering method to set a preset frequency threshold value to eliminate the drift of the load signal.

[0020] Further, the step S3 includes:

[0021] S301, increasing the resampling frequency of the extracted load peak and valley values to an integer multiple of the original sampling frequency;

[0022] S302, when there is no data point at any time resampling, using a difference method to complete the sample point with no data.

[0023] Further, the difference method for completing data in step S302 is as follows:

[0024] At this moment, the adjacent two data points before and after are interpolated once, and the formula is:

[0025]

[0026] Where, x i is the time when the data point is empty, y i is the data point value after linear interpolation at this moment; x i+1 is the next moment, y i+1 is the data point value at the next moment; x i-1 is the previous moment, and y i-1The data point value of the previous time is set as the initial value of the current time.

[0027] Further, the step S4 comprises:

[0028] S401, according to the extracted load peak and valley values after resampling, four consecutive points S1, S2, S3 and S4 are used to extract a cycle, and the four consecutive stress points define the inner stress interval ΔS1 = |S2-S3| and the outer stress interval ΔS0 = |S1-S4|;

[0029] S402, if the outer stress interval is greater than or equal to the inner stress interval: ΔS0≥ΔS1, and the points constituting the inner stress interval are contained in the outer stress interval, then S2 and S3 are considered to constitute a cycle; if not, no cycle counting is performed; discard the two inner stress points S2 and S3, and connect the two outer stress points S1 and S4;

[0030] S403, the same comparison method in step S42 is used for the subsequent four consecutive stress points, until all data points are counted, and a two-dimensional rainflow matrix in the form of From-To is obtained.

[0031] Further, the step S5 specifically comprises:

[0032] Set 10% of the maximum load cycle amplitude as a threshold value, if the load amplitude in the rainflow matrix is less than the threshold value, it is divided into a small amplitude load area; the remaining load cycles constitute a large amplitude load area.

[0033] Further, the step S6 comprises:

[0034] S601, calculate the pseudo-damage using the Basquin equation:

[0035]

[0036] Wherein, S i is the load amplitude, β is the damage index, and d is the pseudo-damage value;

[0037] S602, calculate the pseudo-damage of all cycles in the rainflow matrix as d1, determine the small amplitude load area to be removed; and calculate the pseudo-damage of the remaining load cycles in the rainflow matrix after removing the small amplitude load area as d2; if 0.95≤d2 / d1≤1 is satisfied, the small amplitude load area to be removed is removed.

[0038] Further, the step S7 comprises:

[0039] Extract all the remaining loads in the rainflow matrix, and connect all the loads in series according to the complete combination mode; based on the nearest principle, the turning points closest to the load cycle are connected first;

[0040] If the previous load cycle point and the next load cycle point cannot be directly coincided, half a load cycle is automatically added to form a cycle; the new load sequence contains all the number of times the cycle is repeated.

[0041] Furthermore, the method also includes:

[0042] S8. Compare the cumulative cycle count, power spectral density, and cumulative damage ratio of the load signal in step S1 with the reconstructed load in step S7 to evaluate the reconstructed time-series load.

[0043] As can be seen from the above technical solution, compared with the prior art, the present invention discloses a time-domain load reconstruction method based on rainflow counting, which has the following advantages:

[0044] (1) This invention proposes a time-series load reconstruction method based on rainflow counting. The reconstructed load cumulative cycle count is reduced, shortening the time length of the load signal and achieving the purpose of accelerating the test.

[0045] (2) The present invention eliminates small-amplitude loads by rainflow counting method, ensuring that the pseudo-damage of the reconstructed load sequence is between 0.95 and 1 of the original load, without changing the original load's damage mode on the structure.

[0046] (3) The fully combined load reconstruction proposed in this invention takes into account the "proximity principle", retains the number of repetitions of all load cycles, and considers the potential cycles between load cycles. It is accurate, efficient, suitable for engineering practice, and has a wide range of application value. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0048] Figure 1 A flowchart of the time-series load reconstruction method based on rainflow counting provided by the present invention;

[0049] Figure 2 The original load data diagram provided for this invention;

[0050] Figure 3a A schematic diagram showing the load inflection point in the original load provided by the present invention;

[0051] Figure 3b A schematic diagram illustrating the extraction of peak and valley values ​​from the original load provided by this invention;

[0052] Figure 4a A schematic diagram illustrating the extraction of a loop from four consecutive points provided by this invention;

[0053] Figure 4b This is a schematic diagram of the four-point rainflow counting process provided by the present invention;

[0054] Figure 5 This invention provides a rainflow matrix partitioning diagram based on large and small load amplitudes.

[0055] Figure 6 This invention provides a method for eliminating small-amplitude loads on the diagonal.

[0056] Figure 7 A schematic diagram of load cycling is provided for the present invention.

[0057] Figure 8 A schematic diagram of the load reconfiguration method of the complete combination provided by the present invention;

[0058] Figure 9 The reconstructed timing load diagram provided by this invention;

[0059] Figure 10 A comparison chart of transcendental cycle numbers provided by this invention;

[0060] Figure 11 Damage percentage comparison chart provided for this invention;

[0061] Figure 12 A power spectral density comparison chart provided by the present invention. Detailed Implementation

[0062] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0063] This invention provides a time-series load reconstruction method based on rainflow counting. For example, a measured strain load signal of a mechanical structure is selected as the research object. After rainflow counting, small-amplitude load cycles are eliminated, and finally, load reconstruction is performed. The strain signal duration is 300 seconds, and the sampling frequency is 50 Hz. The flow of the time-series load reconstruction method based on rainflow counting is as follows: Figure 1 As shown, it includes the following steps:

[0064] Step 1: Preprocessing of load data

[0065] The load signal obtained by the data acquisition instrument can be interfered with by external factors (noise, wind load, temperature, humidity, etc.); for example Figure 2 As shown, the horizontal axis represents time, and the vertical axis represents the micro-strain amplitude. Assuming that the measured load signal contains burr signals and drifts, the load signal is preprocessed to eliminate the influence of external factors.

[0066] Step 1.1 Removal of burr signals

[0067] Differentiate the load signal to identify high-slope regions, which are considered to be burr segments. Smooth the signal to remove the burr.

[0068] Step 1.2 Eliminate signal drift

[0069] Compared to the test load signal, the drift signal has a long period and a low corresponding frequency. A Butterworth high-pass filter is used, with a low frequency threshold value set, to eliminate the drift in the load signal.

[0070] Step 2: Extraction of load peak and valley values

[0071] Small fluctuations in load data caused by changes in the test environment (temperature, noise, etc.) will also be recorded. Figure 3a To extract local signals, peak and valley points are the turning points of the load signal, containing important information about the load. The peak and valley values ​​of the load signal are identified and extracted. Figure 3b This is for extracting load data after peak and valley values.

[0072] Step 3: Resampling of load data

[0073] Step 3.1 To facilitate subsequent synchronous counting and improve the accuracy of cyclic counting, the time step between two data points should be as short as possible and equally spaced. Therefore, the signal from which the load peak and valley values ​​are extracted is resampled. The resampling frequency is increased to twice the original sampling frequency, so the time step Δt between the two data points after resampling is 0.01s.

[0074]

[0075] Where f = 50Hz is the original sampling frequency, and m = 2 is the ratio of the resampling frequency to the original sampling frequency;

[0076] Step 3.2 Due to the increased sampling frequency, some data points may be missing during resampling at a certain moment. The sampling interpolation method is used to fill in the missing sample points. Interpolation is performed between two adjacent data points at that moment, and the formula is as follows:

[0077]

[0078] Where, x iFor the time when the data point is empty, y i x represents the linearly interpolated data point value at that moment; i+1 For the next moment, y i+1 The value of the data point at the next time step; x i-1 For the previous moment, y i-1 This represents the value of the data point from the previous time step.

[0079] Step 4: Rainflow counting of load data

[0080] Based on the load peak and valley values ​​extracted after resampling, a four-point cyclic rainflow counting method was used, combined with... Figures 4a-4b Explanation follows. Step 3 resamples the extracted load peak and valley values, so rainflow counting can be performed directly on the load data. Four consecutive points from the load data are selected and denoted as S1, S2, S3, and S4 to form a loop, as follows: Figure 4a As shown, four consecutive load data points define the internal stress interval ΔS1 = |S2-S3| and the external stress interval ΔS0 = |S1-S4|. If the external stress interval is greater than or equal to the internal stress interval (ΔS0 ≥ ΔS1), and the points constituting the internal stress interval are included in the external stress interval, then the second and third points are considered to form a cycle. If this condition is not met, there is no cycle count. Two internal stress points are discarded, and the two external stress points (the first and fourth points) are connected, as shown below. Figure 4b As shown, the same comparison method is used for the next four consecutive stress points until all data points are counted, resulting in a two-dimensional rainflow matrix in the form of From-To.

[0081] Step 5: Partitioning the rainflow matrix according to large and small load amplitudes.

[0082] The obtained From-To form rainflow matrix is ​​partitioned according to the magnitude of the load amplitude. A threshold Δ is set at 10% of the maximum load cycle amplitude. If the load amplitude in the rainflow matrix is ​​less than the threshold Δ, the region formed by this portion of the load cycle is considered a small load amplitude region; the region formed by the remaining load cycles in the rainflow matrix is ​​considered a large load amplitude region. After counting, the quadrilateral region represents small-amplitude loads, and the two triangular regions on either side represent large-amplitude loads, such as... Figure 5 As shown.

[0083] Step 6: Removal of small-amplitude load regions

[0084] Step 6.1 Calculate pseudo-damage using the Basquin equation:

[0085]

[0086] Among them, S i β is the load amplitude, β is the damage index, and d is the pseudo-damage value.

[0087] Step 6.2 First, calculate the spurious damage d1 for all load cycles of the rainflow matrix, then remove the quadrilateral regions of the rainflow matrix, such as... Figure 6 As shown, the spurious damage of the remaining load cycle is d2. After counting, the ratio of the two damages satisfies 0.95≤d2 / d1≤1, so it is considered reasonable to remove the small-amplitude load in the quadrilateral region.

[0088] Step 7 Reconstruction of Residual Load

[0089] Step 7.1 Extract all remaining loads from the rainflow matrix and connect all loads in series in a fully combined manner.

[0090] For example from Figure 6 Two load cycles were selected for the large-amplitude load region of the moderate rain flow matrix: load cycle 1 and load cycle 2. Load cycle 1 was repeated twice, and load cycle 2 was repeated three times. If considered individually, the load cycles recorded by the two load cycles include 2a, 2b, 3c, 3d, and 3e, as follows. Figure 7 As shown. Based on the "proximity principle," the closest turning points of load cycles are connected first. If the previous load cycle point and the next load cycle point cannot directly coincide, half a load cycle is automatically added to form a complete cycle, ensuring minimal error. All loads are connected in series in a fully combined manner. The recorded load cycles include 1a, 2b, 3c, 2d, 1e, 1f, and 1g, as shown. Figure 8 As shown. The novel load sequence includes all the number of repetitions of the cycles and takes into account potential cycles between load cycles. Figure 8 In this approach, the fully combined approach takes into account the potential cycles 1f and 1g between load cycles, compared to considering them individually.

[0091] Step 7.2 Following the method in Step 7.1, Figure 6 The load cycles in the medium-to-large magnitude load region are reconstructed into entirely new time-series loads. These reconstructed time-series loads include all the cycles repeated, such as... Figure 9 As shown, the time length of the load signal is significantly shortened, thus achieving the goal of accelerating the test.

[0092] Step 8: Evaluation of Timing Loads

[0093] Step 8.1 Comparison Figure 9 and Figure 2 The reconstructed timing load duration is significantly shorter than the original load, thus achieving the goal of accelerating the test.

[0094] Step 8.2 uses the cumulative cycle counting method to calculate the number of exceedance cycles and the cumulative damage ratio of the original load and the reconstructed time-series load, and compares them, such as... Figure 10 and Figure 11As shown, after reconstruction, the number of overrun cycles of the time-series load decreases at small amplitudes, and the spurious damage is 96.5% of the original load, satisfying the damage equivalence principle.

[0095] Step 8.3 uses the frequency domain Fourier transform method to calculate the power spectral density of the original load and the reconstructed time-series load, as follows: Figure 12 As shown, the horizontal axis represents frequency, and the vertical axis represents power spectral density (PSD). Compared with the original load, the reconstructed time-series load ignores the loading order, phase information, and spectral information, exhibiting the frequency domain characteristics of a Gaussian signal. This method is more suitable for mechanical structures where the influence of the loading frequency is not considered.

[0096] This invention proposes a time-series load reconstruction method based on rainflow counting. The advantage of using rainflow counting to eliminate small-amplitude diagonal loads is that it retains most of the effective information of the original load, satisfies the damage equivalence principle, and significantly reduces the cumulative cycle count, greatly shortening the load signal duration and achieving the goal of accelerating testing. It is simple and reliable, suitable for engineering practice, and has broad engineering application value.

[0097] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0098] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A time-series load reconstruction method based on rainflow counting, characterized in that, Includes the following steps: S1. Obtain the measured load signal of the target mechanical structure through a data acquisition instrument and perform preprocessing; S2. Extract the load peak and valley values ​​from the preprocessed load signal; S3. Resample the signal from which the load peak and valley values ​​are extracted; S4. Count the rainflow based on the peak and valley values ​​of the load extracted after resampling to obtain a two-dimensional rainflow matrix; S5. Divide the two-dimensional rainflow matrix into regions according to the load amplitude to obtain small-amplitude load regions and large-amplitude load regions; S6. Eliminate areas with small-amplitude loads based on preset conditions; S7. Reconstruct the remaining loads in the two-dimensional rainflow matrix; Step S6 includes: S601. Calculate pseudo-damage using the Basquin equation: Among them, S i denoted as load amplitude, β as damage index, and d as pseudo-damage value; S602. Calculate the pseudo-damage of all cycles in the rainflow matrix as d1, and determine the small-amplitude load area to be removed; and calculate the pseudo-damage of the remaining load cycles in the rainflow matrix after excluding the small-amplitude load area to be removed as d2; if 0.95≤d2 / d1≤1 is satisfied, then the small-amplitude load area to be removed is removed.

2. The time-series load reconstruction method based on rainflow counting according to claim 1, characterized in that, The preprocessing of the load signal in step S1 includes: S101. Differentiate the load signal to identify high-slope regions; the high-slope regions are burr segments, and the burr signals are eliminated through smoothing processing; S102. Using the Butterworth high-pass filtering method, a preset frequency threshold value is set to eliminate the drift of the load signal.

3. The time-series load reconstruction method based on rainflow counting according to claim 1, characterized in that, Step S3 includes: S301. Increase the resampling frequency of the extracted load peak and valley values ​​to an integer multiple of the original sampling frequency; S302. When there are empty data points during resampling at any time, the difference method is used to fill in the empty sample points.

4. The time-series load reconstruction method based on rainflow counting according to claim 3, characterized in that, The data completion method using the difference method in step S302 is as follows: Interpolation is performed between two adjacent data points at that moment, using the following formula: Where, x i For the time when the data point is empty, y i x represents the linearly interpolated data point value at that moment; i+1 For the next moment, y i+1 The value of the data point at the next time step; x i-1 For the previous moment, y i-1 This represents the value of the data point from the previous time step.

5. The time-series load reconstruction method based on rainflow counting according to claim 1, characterized in that, Step S4 includes: S401. Based on the load peak and valley values ​​extracted after resampling, a cycle is extracted using four consecutive points S1, S2, S3, and S4 of the load quantity. The four consecutive stress points define the internal stress interval ΔS1 = |S2-S3| and the external stress interval ΔS0 = |S1-S4|. S402. If the external stress interval is greater than or equal to the internal stress interval: ΔS0≥ΔS1, and the points that make up the internal stress interval are contained in the external stress interval, then S2 and S3 are considered to form a cycle; if not satisfied, there is no cycle count; discard the two internal stress points S2 and S3, and connect the two external stress points S1 and S4. S403. For the next four consecutive stress points, use the same comparison method as in step S42 until all data points have been counted, and obtain a two-dimensional rainflow matrix in the form of From-To.

6. The time-series load reconstruction method based on rainflow counting according to claim 1, characterized in that, Step S5 specifically includes: Set 10% of the maximum load cycle amplitude as the threshold. If the load amplitude in the rainflow matrix is ​​less than the threshold, it is divided into a small-amplitude load area; the area formed by the remaining load cycles is divided into a large-amplitude load area.

7. The time-series load reconstruction method based on rainflow counting according to claim 1, characterized in that, Step S7 includes: Extract all remaining loads from the rainflow matrix and connect them in series in a fully combined manner; based on the principle of proximity, connect the inflection points of the load cycles that are closest to each other first. If the previous load cycle point and the next load cycle point cannot directly coincide, half a load cycle is automatically added to form a cycle; the new load sequence contains all the number of times the cycle is repeated.

8. The time-series load reconstruction method based on rainflow counting according to claim 1, characterized in that, The method also includes: S8. Compare the cumulative cycle count, power spectral density, and cumulative damage ratio of the load signal in step S1 with the reconstructed load in step S7 to evaluate the reconstructed time-series load.

Images