A real-time multi-axis cycle counting method

Abstract

The application discloses a real-time multi-axis cycle counting method, selects an initial reference point, receives and reads a random multi-axis strain load history in real time, calculates von Mises relative equivalent strain of each read strain data relative to the reference point, and further obtains a relative equivalent strain history of the read data in real time; each turning point in the relative equivalent strain history is identified; a remaining load interval is counted as a maximum half cycle; if there are two peak and valley value points on the last relative equivalent strain spectrum, only the remaining load interval between the starting time point and the maximum time point is counted as a maximum half cycle. Based on the concept of relative equivalent strain, on the basis of the WB cycle counting method, in combination with the idea of the Glinka counting method, the real-time cycle counting method is proposed for multi-axis variable amplitude random load, and the maximum cycle damage is supplemented, so that higher reliability is achieved.

Description

Technical Field

[0001] The application field of this invention is real-time multiaxial fatigue strength and life prediction, specifically referring to a real-time multiaxial cycle counting method under multiaxial load. Background Technology

[0002] Multiaxial fatigue damage assessment under random loading involves three key steps: multiaxial cycle counting, multiaxial damage assessment for each cycle, and cumulative fatigue damage across all cycles. Research on multiaxial cycle counting methods plays a crucial role in practical engineering applications. Generally, engineering components often endure complex multiaxial stress / strain random loads during operation. Although numerous researchers have conducted extensive studies on multiaxial fatigue over the past few decades, online prediction of multiaxial fatigue life for in-service engineering components remains a challenging problem.

[0003] In current engineering production, the counting methods used are mostly traditional offline methods, which consume a lot of memory, cannot assess damage at a specific moment, and have low reliability in production applications. Methods such as peak counting, horizontal cross counting, distance pair counting, and rainflow counting all require knowledge of the entire strain / stress load spectrum before the counting process begins. Furthermore, the load spectrum must be rearranged before counting to form the largest possible full or semi-cycle. Because the reconstruction process ignores the load sequence, it may lead to errors in assessing damage at a specific moment.

[0004] Currently, the most mature online counting methods applicable to actual production are single-axis rainflow real-time counting methods. However, most multi-axis online counting methods only count small-cycle damage within the entire load spectrum, failing to count the largest cycle damage, thus leading to an underestimation of damage. Therefore, there is an urgent need to develop a reliable real-time cycle counting method to achieve real-time assessment of current damage and lifetime prediction. Summary of the Invention

[0005] This invention aims to propose a real-time multi-axis cyclic counting method to meet the need for real-time prediction of multi-axis fatigue life under random loads. Firstly, based on the concept of relative equivalent strain, and building upon the WB cyclic counting method, a new real-time cyclic counting method is proposed for multi-axis variable amplitude random loads, supplementing the maximum cyclic damage to achieve higher reliability. The steps are as follows:

[0006] Step 1): Select the data ε at time t=0 ij (t0) is the initial reference point;

[0007] Step 2): Receive and read the random multiaxial strain load history in real time, and calculate ε for each read strain data point. ij(t) von Mises relative equivalent strain to the reference point

[0008]

[0009] in, It is the relative equivalent strain at time t, ε ij (t ref ) is t ref Data at time ε eq () is used to calculate the von Mises equivalent variation of the content inside the parentheses;

[0010] This allows for the real-time acquisition of the relative equivalent change process of the read data;

[0011] Step 3): Identify the first turning point in the relative equivalent change process. The corresponding time is t1:

[0012] if and

[0013] Step 4): Calculate ε at time t1 ij (t1) is defined as the reference point for the next stage of the process. Repeat steps 2) and 3) to find subsequent turning points;

[0014] Step 5): Given a counting range containing m peak and valley points, when the number of inflection points found satisfies the counting range, arrange the inflection points alternately with a peak point and a valley point, and connect the relative equivalent change processes between these peak and valley points end to end. Then, store the peak and valley values ​​in the process into Y. n middle;

[0015] Step 6): Count the relative equivalent variable spectrum, starting from the fourth point. The judgment rule is: if |Y n+1 -Y n |≥|Y n -Y n-1 If the condition is met, then a full loop is calculated, and Y is... n-2 Assigned to Y n , i.e. Y n =Y n-2 If Y n-1 The peak point will be used to calculate the cyclic data ε. ij (t n-1 )~ε ij (t' n-1 Remove t' from the original spectrum n-1 It is Y n ~Y n+1Relative equivalent variable spectrum and Y n-1 The time points corresponding to the points where the values ​​of Y are equal; if Y n-1 The valley point will be used to calculate the cyclic data ε. ij (t' n )~ε ij (t n Remove t' from the original spectrum n It is Y n-2 ~Y n-1 Relative equivalent variable spectrum and Y n The time corresponding to the point where the values ​​of |Y| are equal; if |Y| is not satisfied n+1 -Y n |≥|Y n -Y n-1 If the condition is not met, the full cycle cannot be extracted, and the process continues to check the next step; if no full cycle is counted within the counting range, the relative equivalent variable spectrum continues to be read until all data is counted.

[0016] Step 7): If there are more than three peak and valley points after the above steps are completed, we will discuss two cases: (1) If the last point Y n If it is the peak point, then let That is, if the value of the last point is equal to the maximum value, then start counting again; (2) if the last point is a valley point, then let That is, the value of the second to last point equals the maximum value, and then the counting starts again;

[0017] Step 8): At this point, all full-cycle counting is complete. If there are still three peak-valley points on the final relative equivalent strain spectrum, then the maximum value will be... Location at time t max At time point t end The remaining load interval between these points is counted as a half-cycle; simultaneously, the period from the starting time point t0 to the time point t... max The remaining load interval between these intervals is considered as a maximum half-cycle; if there are still two peak-valley points on the final relative equivalent strain spectrum, then only the interval from the starting time point t0 to the time point t... max The remaining load interval between these intervals is considered as a maximum half-cycle. Attached Figure Description

[0018] Figure 1 The flowchart of the real-time multi-axis cyclic counting method provided by the present invention.

[0019] Figure 2 Lifetime prediction results diagram.

[0020] Figure 3 This study aims to compile the loading history of experimental data on En15R steel cylindrical thin-walled tube specimens under tensile / torsional random loads.

[0021] Figure 4 This is the strain process of step 1.

[0022] Figure 5 This is the strain process for step 2.

[0023] Figure 6 This is the strain process for step 3.

[0024] Figure 7 This is the strain process in step 4.

[0025] Figure 8 This is the strain process in step 5.

[0026] Figure 9 This refers to the strain process in step 7.

[0027] Figure 10 This is the strain process in step 8.

[0028] Figure 11 This is the strain process for step 9.

[0029] Figure 12 This is the strain process for step 10.

[0030] Figure 13 This is the strain process for step 11. Detailed Implementation

[0031] The technical solution adopted in this invention is a new real-time multi-axis cyclic counting method, which is applied in the field of multi-axis fatigue online life prediction. The algorithm steps are as follows: (1) Select the data ε at time t=0. ij (t0) is the initial reference point; (2) the random multiaxial strain load history is received and read in real time, and the strain data ε of each read strain data is calculated. ij (t) von Mises relative equivalent strain compared to the reference point Then, the relative equivalent change process of the read data is obtained in real time; (3) the first turning point in the relative equivalent change process is identified. The corresponding time is t1; (4) ε at time t1 ij (t1) is defined as the reference point for the next stage of the process. Repeat steps 2) and 3) to find subsequent turning points; (5) Given a counting range containing m peak and valley points, when the number of turning points found satisfies the counting range, arrange the turning points alternately according to a peak point and a valley point, and connect the relative equivalent change process between these peak and valley points end to end. Then, store the peak and valley values ​​in the process into Y. n (6) Count the relative equivalent variable spectrum, starting from the fourth point, and judge according to the following rule: if |Y n+1 -Y n|≥|Y n -Y n-1 If Y is a condition, then a full loop is calculated. n-1 The peak point will be used to calculate the cyclic data ε. ij (t n-1 )~ε ij (t' n-1 Remove t' from the original spectrum n-1 It is Y n ~Y n+1 Relative equivalent variable spectrum and Y n-1 The time points corresponding to the points where the values ​​of Y are equal; if Y n-1 The valley point will be used to calculate the cyclic data ε. ij (t' n )~ε ij (t n Remove t' from the original spectrum n It is Y n-2 ~Y n-1 Relative equivalent variable spectrum and Y n The time corresponding to the point where the values ​​of |Y| are equal; if |Y| is not satisfied n+1 -Y n |≥|Y n -Y n-1 If the condition is met, the full cycle cannot be extracted, and the judgment continues; if no full cycle is counted within the counting range, the relative equivalent effect spectrum is read in until all data is counted; (7) If there are more than three peak and valley points after the above steps are completed, two cases are discussed: (1) If the last point Y n If it is the peak point, then let That is, if the value of the last point is equal to the maximum value, then start counting again; (2) if the last point is a valley point, then let That is, the value of the second to last point is equal to the maximum value, and the count is restarted; (8) all full cycle counts are completed at this point. If there are still three peak and valley points on the final relative equivalent strain spectrum, then the maximum value is set. Location at time t max At time point t end The remaining load interval between these points is counted as a half-cycle; simultaneously, the period from the starting time point t0 to the time point t... max The remaining load interval between these intervals is considered as a maximum half-cycle; if there are still two peak-valley points on the final relative equivalent strain spectrum, then only the interval from the starting time point t0 to the time point t... max The remaining load interval between these intervals is considered as a maximum half-cycle.

[0032] This invention incorporates experimental data from En15R steel cylindrical thin-walled tube specimens under tensile / torsional random loading, and its loading process is as follows: Figure 3 As shown:

[0033] The detailed execution process is explained below:

[0034] Step 1): Select the data ε at time t=0 ij (t0) is the initial reference point. Calculate the relative equivalent strain of subsequent points relative to the initial reference point, and identify the first inflection point in the relative equivalent strain process. like Figure 4 As shown;

[0035] Step 2): Calculate ε at time t1 ij (t1) is defined as the reference point for the next stage of the process. Finding the second turning point like Figure 5 As shown;

[0036] Step 3): Given a counting range containing m = 6 peak and valley points, find subsequent turning points by continuously updating the reference points, such as... Figure 6 As shown;

[0037] Step 4): When the number of inflection points found meets the given counting range, arrange the inflection points alternately with a peak point and a valley point, and connect the relative equivalent change processes between these peak and valley points end to end. Then, store the peak and valley values ​​in the process into Y. n In, such as Figure 7 As shown;

[0038] Step 5): Count the relative equivalent variable spectrum. The peak and trough points for the first counting are Y0, Y1, Y2, Y3, Y4, and Y5. Starting from Y3, if |Y3-Y2| > |Y2-Y1|, then a full cycle is calculated, where Y1 is the peak point. The data ε of the cycle is then calculated. ij (t1)~ε ij (t1') is deleted from the original spectrum, such as Figure 8 As shown;

[0039] Step 6): Move Y0 forward two places, i.e., Y2 = Y0. Then judge backward, |Y5-Y4| < |Y4-Y3|. There are no more full cycle counts within this counting range. Continue to read the relative equivalent variable spectrum. Then the peak and valley points participating in the second count are Y2, Y3, Y4, Y5, Y6, and Y7.

[0040] Step 7): |Y6-Y5|>|Y5-Y4|, then a full cycle is calculated, where Y4 is the valley point. The data ε of the cycle is calculated. ij (t'5)~ε ij (t5) Removed from the original spectrum, such as Figure 9 As shown;

[0041] Step 8): Shift Y2 and Y3 forward by two positions, so Y5 = Y3 and Y4 = Y2. Then, check if |Y7 - Y6| > |Y6 - Y5|. This calculates a complete cycle, where Y5 is the peak point. The calculated cycle data ε is then used. ij (t3)~ε ij (t3') is deleted from the original spectrum, such as Figure 10 As shown;

[0042] Step 9): Shift Y4 forward two positions, i.e., Y6 = Y4. No more full cycles are counted within this counting range. Continue reading the relative equivalent variable spectrum. The peak and trough points participating in the third counting are Y6, Y7, Y8, Y9, |Y9-Y8| < |Y8-Y7|. The full cycle cannot be extracted here. Figure 11 As shown;

[0043] Step 10): Now there are four peak-valley points on the relative equivalent strain spectrum, and the last point is the peak point. Let Y9' = Y7, and then count again. |Y9' - Y8| = |Y8 - Y7|, and calculate a full cycle. Record the data ε of the cycle. ij (t7)~ε ij (t9) Removed from the original spectrum, such as Figure 12 As shown;

[0044] Step 11): At this point, all full cycles have been counted. Finally, there are two peak-valley points, Y6 and Y7, on the relative equivalent strain spectrum. The remaining load interval between the starting time point t0 and the time point t7 corresponding to the maximum value needs to be counted as a maximum half-cycle. Figure 13 As shown.

[0045] The entire counting process is now complete, and the lifetime result is calculated using the Shang-Wang model. To verify the effectiveness of the real-time cyclic counting method under multi-axis random loading proposed in this invention, the predicted results of this method are compared with the experimentally observed lifetimes obtained from multi-axis proportional and non-proportional loading experiments. The results show that the lifetime predicted based on this counting method is within three times the error factor compared to the experimentally observed lifetimes under multi-axis proportional and non-proportional loading. Therefore, the proposed calculation method can be well applied to the real-time prediction of lifetimes under multi-axis proportional and non-proportional loading.

Claims

1. A real-time multi-axis cyclic counting method, characterized in that: The steps are as follows: Step 1): Select the data ε at time t=0 ij (t0) is the initial reference point; Step 2): Receive and read the random multiaxial strain load history in real time, and calculate ε for each read strain data point. ij (t) von Mises relative equivalent strain to the reference point in, It is the relative equivalent strain at time t, ε ij (t ref ) is t ref Data at time ε eq () is used to calculate the von Mises equivalent variation of the content inside the parentheses; This allows for the real-time acquisition of the relative equivalent change process of the read data; Step 3): Identify the first turning point in the relative equivalent change process. The corresponding time is t1: if and Step 4): Calculate ε at time t1 ij (t1) is defined as the reference point for the next stage of the process. Repeat steps 2) and 3) to find subsequent turning points; Step 5): Given a counting range containing m peak and valley points, when the number of inflection points found satisfies the counting range, arrange the inflection points alternately with a peak point and a valley point, and connect the relative equivalent change processes between these peak and valley points end to end. Then, store the peak and valley values ​​in the process into Y. n middle; Step 6): Count the relative equivalent variable spectrum, starting from the fourth point. The judgment rule is: if |Y n+1 -Y n |≥|Y n -Y n-1 If the condition is met, then a full loop is calculated, and Y is... n-2 Assigned to Y n , i.e. Y n =Y n-2 If Y n-1 The peak point will be used to calculate the cyclic data ε. ij (t n-1 )~ε ij (t' n-1 Remove t' from the original spectrum n-1 It is Y n ~Y n+1 Relative equivalent variable spectrum and Y n-1 The time points corresponding to the points where the values ​​of Y are equal; if Y n-1 The valley point will be used to calculate the cyclic data ε. ij (t' n )~ε ij (t n Remove t' from the original spectrum n It is Y n-2 ~Y n-1 Relative equivalent variable spectrum and Y n The time corresponding to the point where the values ​​of |Y| are equal; if |Y| is not satisfied n+1 -Y n |≥|Y n -Y n-1 If the condition is not met, the full cycle cannot be extracted, and the process continues to check the next step; if no full cycle is counted within the counting range, the relative equivalent variable spectrum continues to be read until all data is counted. Step 7): If there are more than three peak and valley points after the above steps are completed, we will discuss two cases: (1) If the last point Y n If it is the peak point, then let That is, if the value of the last point is equal to the maximum value, then start counting again; (2) if the last point is a valley point, then let That is, the value of the second to last point equals the maximum value, and then the counting starts again; Step 8): At this point, all full-cycle counting is complete. If there are still three peak-valley points on the final relative equivalent strain spectrum, then the maximum value will be... Location at time t max At time point t end The remaining load interval between these points is counted as a half-cycle; simultaneously, the period from the starting time point t0 to the time point t... max The remaining load interval between these intervals is considered as a maximum half-cycle; if there are two peak-valley points on the final relative equivalent strain spectrum, then only the interval from the starting time point t0 to the time point t... max The remaining load interval between these intervals is considered as a maximum half-cycle.

Images