A method for predicting multi-axis fatigue life of high-pressure internal gear pump

Abstract

This invention relates to the field of high-pressure internal gear pump technology, and discloses a method for predicting the multi-axis fatigue life of a high-pressure internal gear pump. The method involves performing typical full-cycle operations on the target gear pump within its expected lifespan, and acquiring the flow parameters of the target gear pump's discharge port at fixed time intervals. The flow parameters are then segmented based on the discharge tooth cavity to form M flow characteristic sequences. For the m-th flow characteristic sequence, the first peak, valley, and second peak are extracted to calculate the m-th bimodal index, where 1 ≤ m ≤ M, and m is a positive integer. The M bimodal indices are used to determine the characteristic time of the target gear pump through change point detection, and the cumulative number of cycles corresponding to the characteristic time is taken as the first cycle life. The characteristic time represents the transition time between the stable period and the period of significant change of the target gear pump. The cumulative number of cycles of the target gear pump within its expected lifespan is taken as the second cycle life.

Description

Technical Field

[0001] This invention relates to the field of high-pressure internal gear pump technology, and more specifically, to a method for predicting the multi-axis fatigue life of a high-pressure internal gear pump. Background Technology

[0002] High-pressure internal gear pumps rely on the precise meshing of internal and external gears to pump oil: when the internal gear drives the external gear to rotate, the gear teeth in the suction zone disengage to create a vacuum for oil suction, while the gear teeth in the pressure zone mesh to squeeze the oil and achieve high-pressure oil discharge. Its life cycle includes a stable period and a period of rapid change. During the stable period, the gears mesh precisely, and the pump operates smoothly. During the period of rapid change, when the mechanical wear of the gears exceeds a critical threshold, the wear accelerates exponentially, leading to rapid pump failure. Once the period of rapid change begins, the decreased gear meshing precision causes unstable pressure in the pump chamber, resulting in a change in the flow rate curve of the discharge chamber from a single peak to a double or multi-peak pattern, indicating that the pump has lost its reliable operating capability.

[0003] A method for predicting the multiaxial fatigue life of a high-pressure internal gear pump, disclosed in CN110287546B, includes: a) creating a 3D model of the high-pressure internal gear pump under test using 3D software based on its actual operating conditions; b) performing fluid simulation on the established 3D model using CAE software; c) establishing a Kriging surrogate model between pressure, speed, temperature, and multiaxial fatigue equivalent stress under actual operating conditions in the fluid simulation of step b; d) conducting a simulation test on step c; acquiring and recording the time history of the multiaxial fatigue equivalent stress after one revolution of the high-pressure internal gear pump; e) calculating the full life cycle of the high-pressure internal gear pump based on the records from step d. Beneficial effects: Using the Kriging surrogate model to evaluate the life of a high-pressure internal gear pump saves simulation time in fluid simulation, while providing a quick and convenient way to obtain the required time history of equivalent stress, offering forward-looking guidance for product development.

[0004] In actual use, this solution can only obtain the theoretical lifespan of the high-pressure internal gear pump. However, the theoretical lifespan ignores the actual failure characteristics during the period of drastic change, which leads to the pump being judged as "usable" even when its accuracy is inaccurate at the end of its life, ultimately causing economic losses to the user. Summary of the Invention

[0005] This invention provides a method for predicting the multi-axis fatigue life of a high-pressure internal meshing gear pump, solving the technical problems mentioned in the background art.

[0006] This invention provides a method for predicting the multi-axis fatigue life of a high-pressure internal meshing gear pump, comprising:

[0007] S1, Perform typical operation of the target gear pump throughout its expected life cycle and obtain the flow parameters of the target gear pump's drain port at fixed time intervals;

[0008] S2, based on the oil discharge cavity, the flow parameters of the time series are divided to form M flow characteristic sequences; for the m-th flow characteristic sequence, the first peak, the valley and the second peak are extracted to calculate the m-th bimodal index; where 1≤m≤M, and m is a positive integer;

[0009] S3, by detecting the change points of M bimodal indices, the characteristic moment of the target gear pump is determined, and the cumulative number of cycles corresponding to the characteristic moment is taken as the first cycle life; where the characteristic moment represents the transition moment between the stable period and the period of rapid change of the target gear pump.

[0010] S4, the cumulative number of cycles of the target gear pump within its expected lifespan is taken as the second cycle lifespan;

[0011] S5, take the minimum of the first cycle life and the second cycle life as the actual life of the target gear pump.

[0012] Furthermore, the flow parameters for the time series are segmented based on the oil discharge cavity to form M flow characteristic sequences, including:

[0013] A first tag is set at the same position on N teeth of the external gear of the target gear pump, and a second tag is set at the oil outlet of the target gear pump; the start time and end time of each oil outlet tooth cavity are obtained by near-field communication based on the first tag and the second tag; the flow parameters of the time series are cyclically divided according to the start time and end time to obtain M flow characteristic sequences.

[0014] Furthermore, for the m-th flow characteristic sequence, the first peak, trough, and second peak are extracted, including:

[0015] Map the m-th flow characteristic sequence to a flow curve;

[0016] Identify the number K waveforms in the flow rate curve;

[0017] If the number of waveforms K = 1, then the m-th flow curve is divided into a single-peak curve;

[0018] If the number of waveforms K≥2, then the m-th flow curve is divided into a bimodal curve;

[0019] For a single-peak curve, the maximum flow rate is taken as the first peak, and the trough and the second peak are equal.

[0020] For a bimodal curve, K local maxima are extracted in chronological order, and adjacent local maxima are grouped together to obtain K-1 sets of peak-valley features. The i-th set of peak-valley features includes the i-th local maxima and the (i+1)-th local maxima, as well as the local minimum value located between the i-th local maxima and the (i+1)-th local maxima. The i-th local maxima, the local minimum value, and the (i+1)-th local maxima are respectively taken as the first peak, the valley bottom, and the second peak, where 1≤i≤K-1, and i is a positive integer.

[0021] Furthermore, the m-th bimodal index is calculated, including:

[0022] If the flow curve of the m-th flow characteristic sequence is a single-peaked curve, then:

[0023]

[0024] in, This represents the m-th bimodal index;

[0025] If the flow curve of the m-th flow characteristic sequence is a bimodal curve, then:

[0026]

[0027] in, This represents the m-th bimodal index. This represents the first peak in the i-th group of peak-valley features. This represents the valley bottom in the i-th group of peak-valley features. It represents the second peak in the i-th group of peak-valley characteristics.

[0028] Furthermore, by detecting change points, the characteristic moments of the target gear pump are determined, including:

[0029] Divide the M bimodal indices into periods according to time sequence to obtain The bimodal index within each cycle is averaged to obtain the corresponding cycle variation parameters.

[0030] Initialize and generate R clustering analysis schemes; where the r-th clustering analysis scheme is based on time order. The period is divided into several clusters, and each cluster includes the same number of periods; where 1≤r≤R, and r is a positive integer;

[0031] Calculate the segmentation quality of the r-th clustering scheme, including:

[0032] Calculate the mean of the variation parameters of each cluster in the r-th cluster analysis scheme to obtain the error parameters of the corresponding clusters;

[0033] Calculate the sum of variances of the error parameters of the r-th clustering analysis scheme, and apply the first weighting to the sum of variances to obtain the first segment evaluation factor;

[0034] The second segment evaluation factor is obtained by applying a second weighting to the number of clusters in the r-th cluster analysis scheme.

[0035] The reciprocal of the sum of the first and second segment evaluation factors is used as the segmentation quality of the r-th clustering analysis scheme;

[0036] The target clustering analysis scheme is determined by maximizing the quality of segmentation.

[0037] The characteristic moments in the target clustering scheme are determined by maximum likelihood estimation.

[0038] Furthermore, the characteristic moments in the target clustering scheme are determined through maximum likelihood estimation, including:

[0039] The time interval between each adjacent cluster in the target clustering scheme is used as a candidate cut point;

[0040] Likelihood estimation is performed for the u-th candidate cut point, including:

[0041] Calculate the mean error parameter for the first to the uth clusters;

[0042] Based on the mean error parameter, the sum of the variances of the error parameters of the 1st to sth clusters is calculated as the first estimated loss;

[0043] The error parameters of the u-th to U-th clusters are logarithmically transformed to obtain the logarithm; where U represents the number of clusters in the target clustering scheme.

[0044] Perform a least-squares linear regression on the logarithm to obtain the intercept and slope;

[0045] The difference between the logarithm and the sum of the intercept and slope is used as the second estimated loss;

[0046] Calculate the sum of the first estimated loss and the second estimated loss as the estimated loss for the u-th candidate tangent point;

[0047] Traverse U-1 candidate cut points and determine the time corresponding to the candidate cut point with the minimum estimated loss as the feature time.

[0048] Furthermore, the cumulative number of iterations corresponding to the characteristic time point is used as the first cycle lifetime, including:

[0049] Determine the number of flow characteristic sequences between the start time and the characteristic time of a typical operation. The first cycle life is .

[0050] Furthermore, the minimum of the first and second cycle lives is taken as the actual lifespan of the target gear pump, including:

[0051] Calculate the difference between the first cycle life and the second cycle life; if the difference is greater than or equal to a preset threshold, the first cycle life is taken as the actual life of the target gear pump; otherwise, the second cycle life is taken as the actual life of the target gear pump.

[0052] The beneficial effects of this invention are as follows: by deploying near-field communication tags at the external gear and oil outlet of the gear pump, precise segmentation of the flow signal of each oil outlet cavity is achieved. Combined with the degree of abrupt change from a single peak to a double peak in the bimodal exponential quantification flow curve, and the inflection point detection based on multi-scheme clustering quality maximization and maximum likelihood estimation, the critical failure moment of the pump body from the stable period to the period of rapid change can be accurately captured. This significantly improves the accuracy and reliability of fatigue life prediction, avoids overuse of equipment due to overestimation of life or premature replacement due to underestimation of life, and greatly reduces the economic losses of users. Attached Figure Description

[0053] Figure 1 This is a flowchart of a method for predicting the multi-axis fatigue life of a high-pressure internal meshing gear pump according to the present invention. Detailed Implementation

[0054] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, features described in some examples may be combined in other examples.

[0055] like Figure 1 As shown, a method for predicting the multi-axis fatigue life of a high-pressure internal gear pump includes:

[0056] S1, Perform typical operation of the target gear pump throughout its expected life cycle and obtain the flow parameters of the target gear pump's drain port at fixed time intervals;

[0057] S2, based on the oil discharge cavity, the flow parameters of the time series are divided to form M flow characteristic sequences; for the m-th flow characteristic sequence, the first peak, the valley and the second peak are extracted to calculate the m-th bimodal index; where 1≤m≤M, and m is a positive integer;

[0058] S3, by detecting the change points of M bimodal indices, the characteristic moment of the target gear pump is determined, and the cumulative number of cycles corresponding to the characteristic moment is taken as the first cycle life; where the characteristic moment represents the transition moment between the stable period and the period of rapid change of the target gear pump.

[0059] S4, the cumulative number of cycles of the target gear pump within its expected lifespan is taken as the second cycle lifespan;

[0060] S5, take the minimum of the first cycle life and the second cycle life as the actual life of the target gear pump.

[0061] It should be noted that "typical operation throughout the entire lifecycle" refers to simulating the most representative standardized working process in actual application within the expected lifespan of the target gear pump, covering the complete lifecycle from initial operation to potential failure. This process requires strictly defining and fixing operating conditions (such as load, speed, oil medium characteristics, and other key parameters) to ensure consistency and repeatability of each test. By collecting data under a unified typical operating scenario, the life prediction results of different gear pumps can be based on the same benchmark, forming a comparable reference.

[0062] Assuming that during the stable period of the target gear pump, the wear rate of the gear meshing surface is 0.01 mm / thousand cycles, the gear meshing accuracy is high, and the oil discharge chamber flow curve exhibits a stable single-peak shape. When the wear accumulates to a critical threshold (e.g., 0.1 mm), the gear meshing clearance exceeds the limit, leading to increased oil leakage, a sharp increase in meshing impact force, and the wear rate entering an exponential growth phase.

[0063] Phase 1 (Stable period, 0-1000 cycles): Wear increases linearly to 0.01mm×1000=0.1mm (critical threshold), and the flow curve shows no obvious abnormalities;

[0064] Phase 2 (the period of great change, after 1001 cycles): Due to excessive meshing clearance, stress concentration occurs on the tooth surface, and the wear rate accelerates exponentially at a rate of 50% per day.

[0065] 1001 cycles: Wear amount 0.015mm / thousand cycles (+50%), the flow rate curve begins to show slight bimodal fluctuations;

[0066] 1002 cycles: Wear amount 0.0225mm / thousand cycles (plus 50%), significant bimodal characteristics, and a sharp drop in pump efficiency;

[0067] 1050 cycles: The wear has reached 0.1mm + (0.015 + 0.0225 + ...), far exceeding the critical threshold. The pump body has lost its high-pressure output capability due to severe leakage.

[0068] Among these factors, exponentially accelerated wear results in an extremely short transition period from "usable" to "failed" for the pump body.

[0069] The expected lifespan is based on theoretical calculations, but for users with high flow accuracy requirements, the standard is limited to the stabilization period of the target gear pump. Therefore, for such users, it is necessary to determine the actual lifespan relative to their usage standard.

[0070] In one embodiment of the present invention, the flow parameters for the time series are segmented based on the oil discharge cavity to form M flow characteristic sequences, including:

[0071] A first tag is set at the same position on N teeth of the external gear of the target gear pump, and a second tag is set at the oil outlet of the target gear pump; the start time and end time of each oil outlet tooth cavity are obtained by near-field communication based on the first tag and the second tag; the flow parameters of the time series are cyclically divided according to the start time and end time to obtain M flow characteristic sequences.

[0072] Specifically, by using hardware tagging and near-field communication technology, the relative motion sequence between the external gear teeth and the oil outlet is precisely located, thereby dividing the continuous flow parameters into a physically meaningful periodic feature sequence. A first tag, typically a detectable physical or electronic identifier (e.g., an RFID chip), is placed at the same position on N teeth of the target gear pump's external gear. This ensures that each tooth can be uniformly identified when it passes the same spatial position during rotation, thus establishing a periodic reference benchmark for tooth movement. A second tag, typically a fixed-position detection device (proximity sensor), is placed in the oil outlet area of ​​the target gear pump. This tag defines the effective working range of the oil outlet cavity, i.e., the starting point when the cavity enters the outlet to begin pressurizing and discharging oil, and the ending point when it leaves the outlet to end the discharging process. When the external gear rotates, the first tag moves with the teeth to the near-field detection range of the second tag. The two tags interact non-contactly to generate pulse signals, recording the start time (teeth entering the outlet, beginning to squeeze oil) and the end time (teeth leaving the outlet, the discharging process ending) of the oil outlet cavity. The time it takes for each tooth to go from entering the oil drain port to leaving corresponds to a complete working cycle of the oil draining tooth cavity. During this period, the changes in flow parameters directly reflect the changes in the volume and meshing state of the tooth cavity.

[0073] In one embodiment of the present invention, for the m-th flow characteristic sequence, extracting the first peak, the trough, and the second peak includes:

[0074] Map the m-th flow characteristic sequence to a flow curve;

[0075] Identify the number K waveforms in the flow rate curve;

[0076] If the number of waveforms K = 1, then the m-th flow curve is divided into a single-peak curve;

[0077] If the number of waveforms K≥2, then the m-th flow curve is divided into a bimodal curve;

[0078] For a single-peak curve, the maximum flow rate is taken as the first peak, and the trough and the second peak are equal.

[0079] For a bimodal curve, K local maxima are extracted in chronological order, and adjacent local maxima are grouped together to obtain K-1 sets of peak-valley features. The i-th set of peak-valley features includes the i-th local maxima and the (i+1)-th local maxima, as well as the local minimum value located between the i-th local maxima and the (i+1)-th local maxima. The i-th local maxima, the local minimum value, and the (i+1)-th local maxima are respectively taken as the first peak, the valley bottom, and the second peak, where 1≤i≤K-1, and i is a positive integer.

[0080] It should be noted that the mapping of the m-th flow feature sequence to a flow curve is implemented using Excel.

[0081] It should be noted that the number of waveforms in the flow rate curve includes:

[0082] First, the flow rate curve is smoothed, for example, using a moving average filter. A sliding time window is selected to average the data near each sampling point, reducing noise interference and making the curve smoother. Next, the first derivative of the smoothed curve is calculated, and the rate of increase or decrease is determined by observing the changes at adjacent points. Then, in the derivative sequence, zero-crossing points where the derivative changes from positive to negative are searched. When the derivative is positive at one point and negative at the next, a peak point is identified between these two points, and the positions of these peak points and their corresponding flow rates are recorded. Finally, the total number of detected peak points is counted; this total number is the waveform count K of the flow rate curve.

[0083] In one embodiment of the present invention, the calculation of the m-th bimodal index includes:

[0084] If the flow curve of the m-th flow characteristic sequence is a single-peaked curve, then:

[0085]

[0086] in, This represents the m-th bimodal index;

[0087] If the flow curve of the m-th flow characteristic sequence is a bimodal curve, then:

[0088]

[0089] in, This represents the m-th bimodal index. This represents the first peak in the i-th group of peak-valley features. This represents the valley bottom in the i-th group of peak-valley features. It represents the second peak in the i-th group of peak-valley characteristics.

[0090] It should be noted that a single-peak curve indicates that the gear pump is in a stable period, with precise gear meshing and a uniform change in the oil discharge cavity volume, without any bimodal characteristics. (Bimodal index) This clearly defines the characteristic boundaries of the stable period.

[0091] Single-group peak-valley characteristic ratio Physical meaning

[0092] It reflects the decrease in flow rate from the first peak to the bottom of the valley, and reflects the flow rate change caused by wear or meshing clearance in the initial stage of oil discharge from the tooth cavity;

[0093] It reflects the extent of flow recovery from the trough to the second peak, indicating the degree of flow recovery during subsequent meshing in the gear cavity.

[0094] The ratio of these two values ​​quantifies the relative change between adjacent peaks and valleys. If gear wear is slight and meshing accuracy is high, and The ratio should be close to 1; if wear is severe and the meshing clearance is uneven, It will be significantly lower than The ratio decreases, thus effectively reflecting the wear condition and meshing abnormalities inside the gear pump.

[0095] When K ≥ 2 (bimodal or multimodal), there are K-1 sets of peak-valley characteristics. Since the flow curve may be affected by instantaneous disturbances or uneven local wear, the ratio of a single set of peak-valley characteristics may be biased. By averaging the K-1 sets of ratios, the bimodal characteristics of the entire flow curve can be comprehensively considered, weakening the influence of local disturbances, and thus... It provides a more comprehensive and stable reflection of the overall wear level and flow distortion of the gear pump.

[0096] A quantitative model is constructed based on the working principle of gear pumps and the physical characteristics of their flow curves. It uses mathematical methods to capture the bimodal flow characteristics caused by gear wear, transforming the abstract degree of wear into calculable and comparable values, thus providing an objective basis for determining whether a gear pump has entered a period of significant change (wear exceeding limits).

[0097] In one embodiment of the present invention, determining the characteristic moment of the target gear pump by detecting change points includes:

[0098] Divide the M bimodal indices into periods according to time sequence to obtain The bimodal index within each cycle is averaged to obtain the corresponding cycle variation parameters.

[0099] Initialize and generate R clustering analysis schemes; where the r-th clustering analysis scheme is based on time order. The period is divided into several clusters, and each cluster includes the same number of periods; where 1≤r≤R, and r is a positive integer;

[0100] Calculate the segmentation quality of the r-th clustering scheme, including:

[0101] Calculate the mean of the variation parameters of each cluster in the r-th cluster analysis scheme to obtain the error parameters of the corresponding clusters;

[0102] Calculate the sum of variances of the error parameters of the r-th clustering analysis scheme, and apply the first weighting to the sum of variances to obtain the first segment evaluation factor;

[0103] The second segment evaluation factor is obtained by applying a second weighting to the number of clusters in the r-th cluster analysis scheme.

[0104] The reciprocal of the sum of the first and second segment evaluation factors is used as the segmentation quality of the r-th clustering analysis scheme;

[0105] The target clustering analysis scheme is determined by maximizing the quality of segmentation.

[0106] The characteristic moments in the target clustering scheme are determined by maximum likelihood estimation.

[0107] It should be noted that, firstly, the M bimodal indices are divided into periods according to time sequence, resulting in... The algorithm iterates through several cycles (N being the number of teeth on the external gear) and averages the bimodal index within each cycle to obtain the variation parameters for the corresponding cycle.

[0108] Next, R clustering schemes are generated. Each clustering scheme is based on time order, and... Each period is divided into several clusters, and each cluster contains the same number of periods.

[0109] When calculating the segmentation quality of the r-th clustering analysis scheme, the core is to balance "intra-segment volatility" and "number of segments." The error parameter (mean of the varying parameters) for each cluster is calculated, and then the sum of variances of the error parameters is calculated. The smaller the sum of variances, the more concentrated the varying parameters within the same cluster, and the smaller the intra-segment volatility. A first weighting is applied to this sum of variances to obtain the first segmentation evaluation factor, which quantifies the impact of intra-segment volatility on segmentation quality. A second weighting is applied to the number of clusters to obtain the second segmentation evaluation factor. Too many segments will make the model too complex, while too few may ignore key changes; this factor is used to measure the reasonableness of the number of segments. The reciprocal of the sum of the first and second segmentation evaluation factors is used as the segmentation quality. High segmentation quality is achieved when intra-segment volatility is small (small first factor) and the number of segments is reasonable (small second factor). By maximizing segmentation quality, target clustering analysis schemes with small intra-segment volatility and an appropriate number of segments can be selected.

[0110] In one embodiment of the present invention, both the first weighted coefficient and the second weighted coefficient are not 0, and their sum is 1. The larger the first weighted coefficient, the more emphasis is placed on controlling intra-segment fluctuations (schemes with smaller variances are more advantageous); the larger the second weighted coefficient, the more attention is paid to the rationality of the number of clusters (avoiding too many or too few clusters).

[0111] In one embodiment of the present invention, determining the characteristic moments in the target clustering scheme through maximum likelihood estimation includes:

[0112] The time interval between each adjacent cluster in the target clustering scheme is used as a candidate cut point;

[0113] Likelihood estimation is performed for the u-th candidate cut point, including:

[0114] Calculate the mean error parameter for the first to the uth clusters;

[0115] Based on the mean error parameter, the sum of the variances of the error parameters of the 1st to sth clusters is calculated as the first estimated loss;

[0116] The error parameters of the u-th to U-th clusters are logarithmically transformed to obtain the logarithm; where U represents the number of clusters in the target clustering scheme.

[0117] Perform a least-squares linear regression on the logarithm to obtain the intercept and slope;

[0118] The difference between the logarithm and the sum of the intercept and slope is used as the second estimated loss;

[0119] Calculate the sum of the first estimated loss and the second estimated loss as the estimated loss for the u-th candidate tangent point;

[0120] Traverse U-1 candidate cut points and determine the time corresponding to the candidate cut point with the minimum estimated loss as the feature time.

[0121] It should be noted that during the stable period, the gear pump operates stably, with the bimodal index fluctuating slightly around a relatively fixed level, resulting in low data dispersion. The mean error parameter and variance sum (i.e., the first estimated loss) of the 1st to uth clusters are calculated. A smaller variance sum indicates more concentrated data within these clusters, better reflecting the stable period's characteristics of "small fluctuations and stable state." Entering the fluctuating period, gear wear intensifies, and the bimodal index exhibits an unstable trend resembling exponential changes. A logarithmic transformation is performed on the error parameters of the uth to Uth clusters (the original data exhibits exponential changes, which can be approximated as linearized after the logarithmic transformation), and then the intercept is obtained through least-squares linear regression. and slope .calculate As the second estimated loss, if the data exhibits exponential characteristics, this loss value will be smaller, indicating that the subsequent data matches the "exponential unstable fluctuation" characteristic of the fluctuation period. By iterating through all candidate cut points, the first estimated loss (measuring the matching degree of the stationary period characteristics) and the second estimated loss (measuring the matching degree of the fluctuation period characteristics) are summed. The candidate cut point with the smallest total estimated loss corresponds to the characteristic moment—at this point, the preceding data matches the stable characteristics of the stationary period (small variance sum), and the subsequent data matches the exponential change characteristics of the fluctuation period (small deviation in linear regression after logarithmic transformation). This accurately captures the key time point of the gear pump's transition from the stationary period to the fluctuation period, providing crucial evidence for lifespan prediction.

[0122] In one embodiment of the present invention, the cumulative number of cycles corresponding to the characteristic moment is used as the first cycle lifetime, including:

[0123] Determine the number of flow characteristic sequences between the start time and the characteristic time of a typical operation. The first cycle life is .

[0124] Specifically, determine the number of flow characteristic sequences between the start time of a typical operation and the characteristic time (the turning point when the gear pump transitions from the stable period to the period of rapid change). Since each rotation of the external gear (completing one cycle) generates N flow characteristic sequences sequentially from its N teeth, therefore... It contains several complete gear rotation cycles. Divide by N, that is This yields the actual number of rotations of the external gear from the initial moment to the characteristic moment, a value defined as the first cycle life. It quantifies the actual number of working cycles of the pump body before significant failure characteristics (such as bimodalization of the flow curve, accelerated wear, etc.) appear, providing a key parameter based on actual operating conditions for evaluating pump body life, and reflecting the working cycle limit of the pump body under actual failure characteristics.

[0125] In one embodiment of the present invention, the minimum value between the first cycle life and the second cycle life is taken as the actual life of the target gear pump, including:

[0126] Calculate the difference between the first cycle life and the second cycle life; if the difference is greater than or equal to a preset threshold, the first cycle life is taken as the actual life of the target gear pump; otherwise, the second cycle life is taken as the actual life of the target gear pump.

[0127] It should be noted that the first cycle life is determined by the characteristic moment (the turning point where the pump body transitions from the stable period to the period of rapid change), reflecting the lifespan during which the pump body experiences a significant performance degradation due to actual failure characteristics such as wear. The second cycle life is the lifespan value based on theoretical models or design expectations.

[0128] If the difference is greater than or equal to the preset threshold, it indicates that the first cycle life is significantly less than the second cycle life, meaning that the pump body has shown obvious failure characteristics (such as bimodalization of the flow curve, accelerated wear, etc.) before the theoretical expected life. In this case, the first cycle life is taken as the actual life, and the actual failure state of the pump body is followed first to avoid safety risks or equipment failures caused by overestimating the theoretical life.

[0129] If the difference is less than the preset threshold, it means that the first cycle life and the second cycle life are relatively close, and the theoretical life can still reflect the overall performance of the pump body well. In this case, the second cycle life is taken as the actual life, while taking into account the reference value of the theoretical evaluation.

[0130] It should be noted that the interval and threshold sizes are set for ease of comparison. The size of the threshold depends on the amount of sample data and the base number set by those skilled in the art for each set of sample data, as long as it does not affect the proportional relationship between the parameter and the quantized value. Furthermore, the above formulas are all dimensionless calculations, and the formulas are derived from software simulations using a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0131] The embodiments of this example have been described above. However, this example is not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms based on the guidance of this example, and all of them are within the protection scope of this example.

Claims

1. A method for predicting the multi-axis fatigue life of a high-pressure internal gear pump, characterized in that, include: S1, perform a full life cycle operation of the target gear pump under typical working conditions within its expected lifespan, and obtain the flow parameters of the target gear pump's drain port at fixed time intervals; S2, based on the oil discharge cavity, the flow parameters of the time series are divided to form M flow characteristic sequences; For the m-th flow characteristic sequence, the first peak, the valley, and the second peak are extracted to calculate the m-th bimodal index. The bimodal index represents the relative degree of change between adjacent peaks and valleys in the flow characteristic sequence; where 1≤m≤M, and m is a positive integer. S3, by detecting the change points of M bimodal indices, the characteristic moment of the target gear pump is determined, and the cumulative number of cycles corresponding to the characteristic moment is taken as the first cycle life; where the characteristic moment represents the moment when the wear on the tooth surface of the target gear pump accumulates to the critical threshold. S4, the cumulative number of cycles of the target gear pump within its expected lifespan is taken as the second cycle lifespan; S5. Compare the first cycle life and the second cycle life to determine the actual life of the target gear pump, including: calculating the difference between the first cycle life and the second cycle life; if the difference is greater than or equal to a preset threshold, then the first cycle life is taken as the actual life of the target gear pump; otherwise, the second cycle life is taken as the actual life of the target gear pump.

2. The method for predicting the multi-axis fatigue life of a high-pressure internal meshing gear pump according to claim 1, characterized in that, The flow parameters for the time series are segmented based on the oil discharge cavity to form M flow characteristic sequences, including: A first tag is set at the same position on N teeth of the external gear of the target gear pump, and a second tag is set at the oil outlet of the target gear pump; the start time and end time of each oil outlet tooth cavity are obtained by near-field communication based on the first tag and the second tag; the flow parameters of the time series are cyclically divided according to the start time and end time to obtain M flow characteristic sequences.

3. The method for predicting the multi-axis fatigue life of a high-pressure internal meshing gear pump according to claim 2, characterized in that, For the m-th flow characteristic sequence, extract the first peak, the trough, and the second peak, including: Map the m-th flow characteristic sequence to a flow curve; Identify the number K waveforms in the flow rate curve; If the number of waveforms K = 1, then the m-th flow curve is divided into a single-peak curve; If the number of waveforms K≥2, then the m-th flow curve is divided into a bimodal curve; For a single-peak curve, the maximum flow rate is taken as the first peak, and the trough and the second peak are equal. For a bimodal curve, K local maxima are extracted in chronological order, and adjacent local maxima are grouped together to obtain K-1 sets of peak-valley features. The i-th set of peak-valley features includes the i-th local maxima and the (i+1)-th local maxima, as well as the local minimum value located between the i-th local maxima and the (i+1)-th local maxima. The i-th local maxima, the local minimum value, and the (i+1)-th local maxima are respectively taken as the first peak, the valley bottom, and the second peak, where 1≤i≤K-1, and i is a positive integer.

4. The method for predicting the multi-axis fatigue life of a high-pressure internal meshing gear pump according to claim 3, characterized in that, The m-th bimodal index is calculated, including: If the flow curve of the m-th flow characteristic sequence is a single-peaked curve, then: in, This represents the m-th bimodal index; If the flow curve of the m-th flow characteristic sequence is a bimodal curve, then: in, This represents the m-th bimodal index. This represents the first peak in the i-th group of peak-valley features. This represents the valley bottom in the i-th group of peak-valley features. It represents the second peak in the i-th group of peak-valley characteristics.

5. The method for predicting the multi-axis fatigue life of a high-pressure internal meshing gear pump according to claim 4, characterized in that, By detecting change points, the characteristic moments of the target gear pump are determined, including: Divide the M bimodal indices into periods according to time sequence to obtain The bimodal index within each cycle is averaged to obtain the corresponding cycle variation parameters. Initialize and generate R clustering analysis schemes; where the r-th clustering analysis scheme is based on time order. The period is divided into several clusters, and each cluster includes the same number of periods; where 1≤r≤R, and r is a positive integer; Calculate the segmentation quality of the r-th clustering scheme, including: Calculate the mean of the variation parameters of each cluster in the r-th cluster analysis scheme to obtain the error parameters of the corresponding clusters; Calculate the sum of variances of the error parameters of the r-th clustering analysis scheme, and apply the first weighting to the sum of variances to obtain the first segment evaluation factor; The second segment evaluation factor is obtained by applying a second weighting to the number of clusters in the r-th cluster analysis scheme. The reciprocal of the sum of the first and second segment evaluation factors is used as the segmentation quality of the r-th clustering analysis scheme; The target clustering analysis scheme is determined by maximizing the quality of segmentation. The characteristic moments in the target clustering scheme are determined by maximum likelihood estimation.

6. The method for predicting the multi-axis fatigue life of a high-pressure internal meshing gear pump according to claim 5, characterized in that, The characteristic moments in the target clustering scheme are determined by maximum likelihood estimation, including: The time interval between each adjacent cluster in the target clustering scheme is used as a candidate cut point; Likelihood estimation is performed for the u-th candidate cut point, including: Calculate the mean error parameter for the first to the uth clusters; Based on the mean error parameter, the sum of the variances of the error parameters of the 1st to sth clusters is calculated as the first estimated loss; The error parameters of the u-th to U-th clusters are logarithmically transformed to obtain the logarithm; where U represents the number of clusters in the target clustering scheme. Perform a least-squares linear regression on the logarithm to obtain the intercept and slope; The difference between the logarithm and the sum of the intercept and slope is used as the second estimated loss; Calculate the sum of the first estimated loss and the second estimated loss as the estimated loss for the u-th candidate tangent point; Traverse U-1 candidate cut points and determine the time corresponding to the candidate cut point with the minimum estimated loss as the feature time.

7. The method for predicting the multi-axis fatigue life of a high-pressure internal meshing gear pump according to claim 6, characterized in that, The cumulative number of iterations corresponding to the characteristic time point is taken as the first cycle lifetime, including: Determine the number of flow characteristic sequences between the start time and the characteristic time of a typical operation. The first cycle life is .

Images