A method for predicting fatigue life of a radiator structure based on vibration test bench data

Abstract

The present application belongs to the technical field of data processing, and particularly relates to a heat radiator structure fatigue life prediction method based on vibration test bench data, comprising: dividing the vibration acceleration signal of the heat radiator into multiple time windows, identifying the modal parameters of each time window to construct a time-varying frequency sequence; processing the time-varying frequency sequence by using an asymmetric weighted moving average filtering method, and extracting a long-term damage trend item reflecting the cumulative structure fatigue damage; separating a short-term thermal disturbance item according to the difference between the natural frequency and the long-term damage trend item; evaluating the cumulative thermal fatigue damage according to the short-term thermal disturbance item; determining the cumulative mechanical fatigue damage according to the long-term damage trend item; and comprehensively predicting the life by combining the cumulative thermal fatigue damage and the cumulative mechanical fatigue damage. The present application realizes the accurate prediction of the fatigue life of the heat radiator under complex working conditions by decoupling the two damage sources of mechanical vibration and thermal cycle, and improves the reliability of the prediction.

Description

Technical Field

[0001] This invention relates to the field of data processing technology. More specifically, this invention relates to a method for predicting the fatigue life of a radiator structure based on vibration test bench data. Background Technology

[0002] In the automotive and industrial sectors, radiators are critical heat exchange components, and the reliability of their structure is of paramount importance. Vibration test benches are essential tools for evaluating the structural durability of radiators, predicting their fatigue life by simulating their vibration environment under actual working conditions.

[0003] In existing technologies, one analytical method employs modal analysis techniques, such as random subspace identification, to determine the health status of a heat sink structure by continuously monitoring changes in modal parameters such as natural frequencies and damping ratios during vibration. However, this method has significant technical limitations when applied to assessing real-world operating conditions involving thermal cycling. During operation, heat sinks not only experience mechanical vibration but also drastic temperature fluctuations. Temperature changes alter the material's elastic modulus and generate thermal stress, which in turn lead to changes in modal parameters. Simultaneously, mechanical vibration itself causes microscopic fatigue damage accumulation, also resulting in alterations in modal parameters.

[0004] Therefore, the changes in modal parameters directly monitored by existing algorithms are actually a mixed result of the combined effects of thermal effects and fatigue damage effects. Existing technologies cannot effectively separate these two effects, making it impossible to independently and accurately measure the potential fatigue damage introduced by temperature changes. This results in a large deviation between the final life prediction results and the actual working conditions, and the reliability of the assessment is insufficient.

[0005] Therefore, there is an urgent need for a method to predict the fatigue life of radiator structures based on vibration test bench data. Summary of the Invention

[0006] To address the technical problem of inaccurate life prediction caused by the inability of existing technologies to effectively decouple mechanical and thermal fatigue damage, this invention provides a method for predicting the fatigue life of radiator structures based on vibration test bench data, comprising:

[0007] The vibration acceleration signal of the radiator on a vibration test bench is acquired in real time and segmented into multiple time windows. Modal parameters for each time window are identified, and a time-varying frequency sequence containing natural frequencies is constructed. An asymmetric weighted moving average filtering method is used to process the time-varying frequency sequence. By applying weighted suppression to frequencies higher than the long-term damage trend term of the previous time window, a long-term damage trend term reflecting the accumulation of structural fatigue damage is extracted. For each time window, a short-term thermal disturbance term is separated based on the difference between the natural frequency and the long-term damage trend term. Accumulated thermal fatigue damage is assessed based on the frequency change amplitude of the frequency fluctuation cycle contained in the short-term thermal disturbance term. Accumulated mechanical fatigue damage is determined based on the long-term damage trend term. Life prediction is performed by combining the accumulated thermal fatigue damage and the accumulated mechanical fatigue damage.

[0008] This invention decomposes continuous vibration signals into multiple time windows for modal analysis, enabling the capture of high-frequency mechanical characteristic changes driven by slow thermodynamic processes. Utilizing signal processing techniques, it separates the long-term trend representing irreversible mechanical damage and the short-term fluctuations representing reversible thermal effects from the coupled frequency variations, providing a clear physical basis for the independent analysis of these two types of damage. By combining the evaluation results of the long-term damage trend term and the short-term thermal disturbance term for comprehensive life prediction, it can fully reflect the damage accumulation process of the radiator under real operating conditions, improving the accuracy and reliability of fatigue life prediction.

[0009] Preferably, the weighting expression in the asymmetric weighted moving average filtering method is: In the formula, For the first The weight of each time window, For the first Long-term damage trend term for each time window; For the first The inherent frequency corresponding to each time window; It is a function for maximizing the value; It is a natural exponential function; This is the preset penalty coefficient.

[0010] This invention employs an asymmetric weighted moving average filtering method to process time-varying frequency sequences. Based on the irreversible nature of fatigue damage, when the natural frequency of a certain time window is higher than the long-term damage trend term of the previous time window, it is determined to be a reversible disturbance caused by temperature reduction. An exponential weight is applied to suppress this disturbance, ensuring that the calculated long-term damage trend term strictly tracks the decreasing envelope of the time-varying frequency sequence. This allows for the precise separation of the unidirectional degradation process that reflects the accumulation of structural fatigue damage.

[0011] Preferably, the step of evaluating cumulative thermal fatigue damage based on the frequency change amplitude of the frequency fluctuation cycles contained in the short-term thermal disturbance items includes: constructing a short-term thermal disturbance sequence from all time windows of short-term thermal disturbance items; applying the rainflow counting method to the short-term thermal disturbance sequence to identify the frequency fluctuation cycles contained therein; treating each frequency fluctuation cycle as a thermal cycle and obtaining the frequency change amplitude of each thermal cycle; obtaining the fatigue damage degree caused by each thermal cycle based on the frequency change amplitude of each thermal cycle; and accumulating the fatigue damage degrees caused by all thermal cycles to obtain the cumulative thermal fatigue damage.

[0012] Preferably, the fatigue damage degree satisfies the expression: In the formula, For the first Fatigue damage caused by a thermal cycle; For the first The amplitude of frequency variation in each thermal cycle; These are the equivalent conversion factors; The coefficient of thermal fatigue performance of the material; It is a dimensionless constant characterizing the fatigue performance of materials.

[0013] This invention identifies complete thermal cycles in a short-term thermal disturbance sequence using the rainflow counting method. It uses the amplitude of frequency change as a proxy index of the thermal stress cycle amplitude and establishes a power function relationship between the damage caused by a single thermal cycle and the amplitude of frequency change in that cycle. This can reflect the nonlinear cumulative characteristics of fatigue damage in metallic materials. By linearly accumulating the damage caused by all thermal cycles, it can accurately measure the thermal fatigue effect of temperature changes on the radiator.

[0014] Preferably, determining the cumulative mechanical fatigue damage based on the long-term damage trend term includes: for each time window, determining the cumulative mechanical fatigue damage of that time window based on the change in the long-term damage trend term of that time window relative to the initial time window.

[0015] Preferably, the cumulative mechanical fatigue damage satisfies the expression: In the formula, For the first Cumulative mechanical fatigue damage over a time window; This represents the long-term damage trend term for the first time window; For the first Long-term damage trend term for each time window; This is the preset frequency failure threshold.

[0016] This invention obtains cumulative mechanical fatigue damage by calculating the relative magnitude of the change in the long-term damage trend term, accurately measures the unidirectional degradation of structural stiffness caused by vibration load, avoids the influence of thermal disturbance, and the damage value changes linearly in the range of zero to one, providing a reliable basis for life prediction.

[0017] Preferably, the life prediction based on the combined cumulative thermal fatigue damage and cumulative mechanical fatigue damage includes: determining the total cumulative damage based on the cumulative mechanical fatigue damage and the cumulative thermal fatigue damage; and predicting the total fatigue life of the radiator based on the total cumulative damage using a linear extrapolation method.

[0018] This invention predicts the lifespan of radiators by comprehensively considering both cumulative thermal fatigue damage and cumulative mechanical fatigue damage. It not only takes into account the influence of a single damage source but also measures the interaction between the two damage mechanisms. This enables accurate assessment of the fatigue life of radiators under complex operating conditions and provides strong technical support for online health monitoring and early fault warning of radiators.

[0019] Preferably, the total cumulative damage satisfies the expression: In the formula, For the first Total cumulative damage over a time window; For the first Cumulative mechanical fatigue damage over a time window; For the first Cumulative thermal fatigue damage over a time window; is the damage coupling coefficient.

[0020] Preferably, the total fatigue life satisfies the expression: In the formula, The predicted total fatigue life; This represents the time elapsed since the current experiment began. This represents the total cumulative damage within the current time window.

[0021] Preferably, it also includes: dynamically updating the predicted total fatigue life based on real-time total cumulative damage.

[0022] The beneficial effects of this invention are as follows: This invention constructs a time-varying frequency sequence through segmented vibration signal processing and modal parameter identification. It employs an asymmetric weighted moving average filtering method based on the irreversible nature of fatigue damage to accurately separate the long-term damage trend term from the short-term thermal disturbance term, effectively decoupling the two physical processes of thermal effect and fatigue damage. This invention cyclically quantifies thermally induced fatigue damage by analyzing frequency fluctuations in the short-term thermal disturbance sequence, while simultaneously assessing mechanical fatigue damage based on changes in the long-term damage trend term. Finally, it integrates the two damage mechanisms for life prediction, achieving independent and accurate assessment of thermal fatigue and mechanical fatigue, thus improving the reliability of life prediction. This invention requires only a standard vibration acceleration sensor to complete the assessment, eliminating the need for expensive equipment such as thermocouples or strain gauges, significantly reducing testing costs. This invention supports dynamic tracking of the damage evolution process, providing key technical support for online health monitoring and early fault warning of radiators. Attached Figure Description

[0023] Figure 1 This is a flowchart illustrating a method for predicting the fatigue life of a radiator structure based on vibration test bench data, as described in this invention.

[0024] Figure 2 This is a schematic diagram of vibration acceleration signals;

[0025] Figure 3 This is a schematic diagram of a time-varying frequency sequence curve;

[0026] Figure 4 This is a schematic diagram of the long-term damage trend curve;

[0027] Figure 5 This is a schematic diagram of the short-term thermal disturbance term curve;

[0028] Figure 6 This is a schematic diagram of the damage accumulation process;

[0029] Figure 7 This is a schematic diagram of the total fatigue life prediction results. Detailed Implementation

[0030] 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, not all, of the embodiments of the present invention. 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.

[0031] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0032] This invention discloses a method for predicting the fatigue life of a radiator structure based on vibration test bench data, referring to... Figure 1This includes steps S1-S6:

[0033] S1: Real-time acquisition of vibration acceleration signals of the radiator on the vibration test bench, and segmentation of the vibration acceleration signals into multiple time windows.

[0034] It should be noted that the working environment of a radiator is extremely complex, and changes in its structural state are driven by two physical processes with different time scales: one is the millisecond-level high-frequency mechanical response caused by vibration loads, and the other is the minute-level slow thermodynamic response caused by changes in coolant temperature. In order to capture the influence of the slow thermodynamic process on the high-frequency mechanical characteristics within a unified analytical framework, this invention divides the continuous vibration acceleration signal into multiple time windows, and analyzes each time window as a snapshot of a relatively stable thermodynamic state.

[0035] Specifically, during the test, the radiator was fixed on the vibration test bench according to the actual installation method of the radiator. At the same time, in order to simulate the thermal cycle experienced by the radiator under real working conditions, the temperature of the fluid inside the radiator was periodically controlled by a temperature-controlled circulating water tank, so that it experienced a preset temperature change process while being subjected to mechanical vibration.

[0036] Furthermore, by using accelerometers positioned at key locations on the radiator structure, the vibration acceleration signals of the radiator under the combined effects of vibration and thermal cycling are acquired in real time, resulting in a one-dimensional time-series signal. For example, Figure 2 This is a schematic diagram of vibration acceleration signals.

[0037] Furthermore, the time-series signal is segmented according to a preset time window length and overlap rate to obtain multiple consecutive time windows. The time window length should be long enough to include a sufficient number of signal periods to ensure accurate identification of low-frequency modes, while being shorter than the significant change period of the thermal effect; an empirical value is 60 seconds. The overlap rate should be high enough to improve time resolution and reduce spectral leakage; an empirical value is 50%. In other embodiments, the implementer can set the time window length and overlap rate according to the specific modal frequency and thermal cycling rate of the heat sink. It is particularly important to note that to ensure the subsequent rainflow counting method can effectively identify thermal cycles, the thermal cycling period should be no less than twice the time window length.

[0038] S2: Identify the modal parameters of each time window and construct a time-varying frequency sequence containing the inherent frequency.

[0039] It should be noted that the heat sink is composed of numerous thin-walled pipes, heat sink fins, and brazed or welded joints. Its overall stiffness is highly sensitive to the initiation and propagation of microcracks and changes in the material's elastic modulus. The stiffness characteristics of the structure are macroscopically reflected in its modal parameters, such as its natural frequencies. Therefore, this invention deduces changes in the internal physical state of the structure by tracking the evolution of its modal parameters over time, providing a physical basis for subsequent damage decoupling and assessment.

[0040] Specifically, in order to eliminate the interference of external excitation signals from the vibration test bench, notch filtering is applied to the vibration acceleration signals of each time window to obtain a clean signal. The center frequency of the notch filter is set to the vibration frequency of the vibration test bench, which can be obtained through the parameter setting interface of the vibration test bench.

[0041] Furthermore, using the pure signal corresponding to each time window as input, a random subspace identification algorithm is independently executed to identify the modal parameters corresponding to the lowest-order mode of the heat sink structure within that time window. These modal parameters include natural frequencies. The natural frequencies identified in all time windows are arranged in chronological order to form a time-varying frequency sequence. For example, Figure 3 This is a schematic diagram of a time-varying frequency sequence curve.

[0042] S3: The time-varying frequency sequence is processed using an asymmetric weighted moving average filtering method. By applying weighted suppression to frequencies higher than the previous time window, the long-term damage trend term reflecting the accumulation of structural fatigue damage is extracted.

[0043] It should be noted that the obtained time-varying frequency sequence necessarily couples changes from two sources: First, the continuous initiation and propagation of microcracks within the metal material caused by vibration loads is an irreversible physical process, inevitably leading to a unidirectional and slow degradation of structural stiffness, reflected as a long-term decreasing trend in frequency; second, the thermal expansion and contraction caused by temperature fluctuations in the coolant inside the radiator is a reversible physical process, causing the structural stiffness to fluctuate rapidly in both directions near the baseline. To obtain fatigue damage data, this invention extracts the unidirectional and slow long-term trend from the time-varying frequency sequence based on the irreversible nature of fatigue damage.

[0044] Specifically, to smooth short-term thermal disturbances in time-varying frequency sequences and capture the unidirectional decreasing profile of damage accumulation, this invention employs an asymmetric weighted moving average filtering method to calculate the long-term damage trend term:

[0045]

[0046] In the formula, For the first Long-term damage trend term for each time window; For the first The inherent frequency corresponding to each time window; For the first The weight of each time window; The length of the review window used for the moving average, the length of the review window The filtering smoothness and trend response speed should be considered simultaneously. An empirical value of 10 is recommended. In other embodiments, implementers can set the value according to the actual implementation situation. It should be noted that when the first... The number of time windows before the current time window is insufficient. At that time, the number of actual time windows is used to determine the first... The long-term damage trend term for a time window, for example, when At that time, the first If the number of time windows before a given time window is 0, then... ,in The weight of the first time window, This is the natural frequency corresponding to the first time window.

[0047] The method for obtaining the weights is as follows:

[0048]

[0049] In the formula, For the first The weight of each time window, For the first Long-term damage trend term for each time window; For the first The inherent frequency corresponding to each time window; It is a function for maximizing the value; It is a natural exponential function; The preset penalty coefficient is a positive real number used to adjust the strength of suppression against frequency revival; its empirical value is 0.5. The implementers can set the penalty coefficient based on the actual implementation situation. For the initial state, set... That is, the natural frequency of the first time window is used as the initial long-term damage trend term to ensure that the filtering process can be started.

[0050] It should be noted that weight The calculation simulated the irreversibility of the damage, when the natural frequency of a certain time window... Long-term damage trend term higher than the previous time window At this time, it indicates that the structural stiffness has temporarily recovered, which contradicts the unidirectional cumulative law of fatigue damage. This phenomenon is attributed to reversible disturbances such as temperature decrease. When the value is positive, the penalty term takes effect, increasing the weight. It decreases exponentially, thereby significantly reducing the inherent frequency of this time window. In calculating the new long-term damage trend term Contribution at the time; conversely, when Not higher than When calculating the new long-term damage trend term At that time, the inherent frequency of this time window The information was fully adopted. Therefore, the long-term damage trend term was calculated. It can track the decreasing envelope of time-varying frequency sequences, thus enabling the effective extraction of a unidirectional degradation process that represents real fatigue damage.

[0051] For example, Figure 4 This is a schematic diagram of the long-term damage trend curve.

[0052] S4: For each time window, the short-term thermal disturbance term is separated based on the difference between the inherent frequency and the long-term damage trend term.

[0053] It should be noted that the total frequency change is composed of both long-term damage and short-term thermal disturbance. Therefore, this invention removes the extracted long-term damage trend term from the total frequency change to obtain the short-term thermal disturbance term that reflects the thermal cycling effect.

[0054] Specifically, for each time window, the short-term thermal disturbance term for that time window is obtained by subtracting the long-term damage trend term from the natural frequency corresponding to that time window:

[0055]

[0056] In the formula, For the first Short-term thermal disturbance term within a time window; For the first The inherent frequency corresponding to each time window; For the first Long-term damage trend term for each time window.

[0057] Furthermore, the short-term thermal perturbation terms of all time windows are constructed into a short-term thermal perturbation sequence. For example, Figure 5 This is a schematic diagram of the short-term thermal disturbance term curve.

[0058] It should be noted that when the temperature of the coolant inside the radiator increases, the elastic modulus of the material decreases, resulting in a change in the actual natural frequency. Below the long-term damage trend term Then the short-term thermal disturbance term It is a negative value; when the temperature of the coolant inside the radiator decreases, the material regains its stiffness, causing the natural frequency to... Higher than the long-term damage trend term Then the short-term thermal disturbance term It is a positive value. Therefore, the short-term thermal disturbance sequence is a sequence that fluctuates around the zero axis, and the shape and amplitude of the fluctuation reflect the cycle of structural stiffness change experienced by the radiator due to changes in operating temperature.

[0059] S5: Assess cumulative thermal fatigue damage based on the frequency variation amplitude of the frequency fluctuation cycle included in the short-term thermal disturbance term.

[0060] It should be noted that each fluctuation in a short-term thermal disturbance sequence corresponds to a thermal expansion and contraction caused by temperature changes. This creates a thermal cycle in constrained areas such as brazed and welded joints of the radiator, leading to thermal fatigue damage. This invention argues that the amplitude of frequency variation directly reflects changes in structural stiffness, which are highly correlated with changes in thermal stress. According to thermodynamic relationships, thermal stress... With frequency change satisfy ,in For material constants, Since it is the elastic modulus, the amplitude of frequency change can be used as an effective proxy for the amplitude of thermal stress in damage assessment.

[0061] Specifically, using short-term thermal disturbance sequences as input, the rainflow counting method is applied to identify the elements contained within them. A complete frequency fluctuation cycle is obtained, and each frequency fluctuation cycle is treated as a separate thermal cycle. The frequency change amplitude of each thermal cycle is then obtained.

[0062] Furthermore, for each thermal cycle, the degree of fatigue damage caused by the thermal cycle is obtained based on the amplitude of the frequency change of the thermal cycle:

[0063]

[0064] In the formula, For the first Fatigue damage caused by a thermal cycle; For the first The amplitude of frequency variation in each thermal cycle; It is a dimensionless equivalent conversion coefficient; This is the thermal fatigue performance coefficient of a material, which physically represents the material's performance under typical temperature cycles (e.g., ...). The corresponding reference frequency change is expressed in Hz. It is a dimensionless constant characterizing the fatigue performance of a material. The equivalent conversion coefficient... Material thermal fatigue performance coefficient and dimensionless constants characterizing the fatigue properties of materials. These are all material property parameters, obtained through standard thermal fatigue calibration tests on the materials used in the heat sink. For commonly used aluminum alloy materials, The empirical value is 1. The empirical value is 0.5Hz. The empirical value is 2.5. In other embodiments, implementers can set the parameters through testing and calibration based on the specific material grade and structural form. , and .

[0065] It should be noted that this invention establishes the damage caused by a single thermal cycle. The amplitude of the frequency change of this thermal cycle The power function relationship between them means that, under the same total temperature change, a drastic temperature change leads to a large [various effects]. The damage it causes is far greater than the damage caused by several gradual temperature changes. The sum of these values ​​is highly consistent with the fatigue characteristics of metallic materials.

[0066] Furthermore, the cumulative thermal fatigue damage is obtained by measuring the fatigue damage caused by all thermal cycles.

[0067] S6: Determine the cumulative mechanical fatigue damage based on the long-term damage trend term, and perform life prediction by combining the cumulative thermal fatigue damage and the cumulative mechanical fatigue damage.

[0068] It should be noted that the failure of a heat sink under real-world operating conditions is the result of the coupled and combined effects of continuous mechanical vibration and repeated thermal cycling until the structure can no longer withstand the load. Mechanical vibration damage accelerates the propagation of thermally induced cracks, while the residual stress generated by thermal cycling also reduces the structure's mechanical vibration tolerance. Therefore, this invention establishes a coupled damage model that can integrate the results of these two damage assessments to achieve accurate prediction of the fatigue life of heat sinks under complex operating conditions.

[0069] Specifically, for each time window, the cumulative mechanical fatigue damage caused by mechanical vibration in that time window is determined based on the change in the long-term damage trend term relative to the initial time window:

[0070]

[0071] In the formula, For the first Cumulative mechanical fatigue damage over a time window; For the first Long-term damage trend term for each time window; This is a preset frequency failure threshold. The frequency failure threshold... Based on industry standards or testing experience, the failure criterion is usually set as a specific percentage decrease in the initial frequency. For example, it can be set to the initial frequency. In 98% of cases, in other embodiments, the implementer may set the failure rate according to the definition of radiator failure. .

[0072] Furthermore, the total cumulative damage is determined based on the cumulative mechanical fatigue damage and the cumulative thermal fatigue damage:

[0073]

[0074] In the formula, For the first Total cumulative damage over a time window; For the first Cumulative mechanical fatigue damage over a time window; For the first Cumulative thermal fatigue damage over a time window; The damage coupling coefficient reflects the contribution weight of thermal fatigue damage relative to mechanical fatigue damage. Its value is calibrated through standard thermomechanical fatigue tests on the materials used in the radiator; for aluminum alloy radiators, the empirical value is 0.8. In other embodiments, the implementer can set the coefficient according to the actual implementation situation. For example, Figure 6 It is a schematic diagram of the damage accumulation process, including the curves showing the changes in accumulated thermal fatigue damage, accumulated mechanical fatigue damage, and total accumulated damage.

[0075] Furthermore, based on the total cumulative damage within the current time window, the total fatigue life of the heat sink is predicted using a linear extrapolation method:

[0076]

[0077] In the formula, The predicted total fatigue life; This represents the time elapsed since the current experiment began. This represents the total cumulative damage within the current time window. Assuming the damage accumulation rate remains macroscopically stable, this invention linearly extrapolates the current damage state to the total time until the damage reaches the failure threshold 1, thereby obtaining the total fatigue life of the heat sink.

[0078] Furthermore, based on real-time total cumulative damage Dynamically update total fatigue life This enables continuous correction of the total fatigue life prediction results.

[0079] For example, Figure 7 This is a schematic diagram of the total fatigue life prediction results.

Claims

1. A method for predicting the fatigue life of a radiator structure based on vibration test bench data, characterized in that, include: The vibration acceleration signal of the heat sink on the vibration test bench is acquired in real time, and the vibration acceleration signal is segmented into multiple time windows; the modal parameters of each time window are identified, and a time-varying frequency sequence containing the natural frequency is constructed. An asymmetric weighted moving average filtering method is used to process the time-varying frequency sequence. By applying weighted suppression to the frequency of the long-term damage trend term that is higher than the previous time window, the long-term damage trend term reflecting the accumulation of structural fatigue damage is extracted. For each time window, the short-term thermal disturbance term is separated based on the difference between the natural frequency and the long-term damage trend term; the cumulative thermal fatigue damage is assessed based on the frequency change amplitude of the frequency fluctuation cycle contained in the short-term thermal disturbance term. The cumulative mechanical fatigue damage is determined based on the long-term damage trend term; the lifespan is predicted by combining the cumulative thermal fatigue damage and the cumulative mechanical fatigue damage.

2. The method for predicting the fatigue life of a radiator structure based on vibration test bench data according to claim 1, characterized in that, The weights in the asymmetric weighted moving average filtering method satisfy the expression: ； In the formula, For the first The weight of each time window, For the first Long-term damage trend term for each time window; For the first The inherent frequency corresponding to each time window; It is a function for maximizing the value; It is a natural exponential function; This is the preset penalty coefficient.

3. The method for predicting the fatigue life of a radiator structure based on vibration test bench data according to claim 1, characterized in that, The assessment of cumulative thermal fatigue damage based on the frequency variation amplitude of the frequency fluctuation cycle included in the short-term thermal disturbance term includes: The short-term thermal disturbances of all time windows are used to construct a short-term thermal disturbance sequence. Rainflow counting is applied to the short-term thermal disturbance sequence to identify the frequency fluctuation cycles contained therein. Each frequency fluctuation cycle is treated as a thermal cycle, and the frequency change amplitude of each thermal cycle is obtained. The fatigue damage caused by the thermal cycle is obtained based on the frequency change amplitude of each thermal cycle, and the fatigue damage caused by all thermal cycles is accumulated to obtain the cumulative thermal fatigue damage.

4. The method for predicting the fatigue life of a radiator structure based on vibration test bench data according to claim 3, characterized in that, The fatigue damage degree satisfies the expression: ； In the formula, For the first The degree of fatigue damage caused by each thermal cycle; For the first The amplitude of frequency variation in each thermal cycle; These are the equivalent conversion factors; The coefficient of thermal fatigue performance of the material; It is a dimensionless constant characterizing the fatigue performance of materials.

5. The method for predicting the fatigue life of a radiator structure based on vibration test bench data according to claim 1, characterized in that, The determination of cumulative mechanical fatigue damage based on the long-term damage trend term includes: For each time window, the cumulative mechanical fatigue damage of that time window is determined based on the change in the long-term damage trend term relative to the initial time window.

6. The method for predicting the fatigue life of a radiator structure based on vibration test bench data according to claim 1, characterized in that, The cumulative mechanical fatigue damage satisfies the expression: ； In the formula, For the first Cumulative mechanical fatigue damage over a time window; This represents the long-term damage trend term for the first time window; For the first Long-term damage trend term for each time window; This is the preset frequency failure threshold.

7. A method for predicting the fatigue life of a radiator structure based on vibration test bench data according to any one of claims 1 to 6, characterized in that, The comprehensive cumulative thermal fatigue damage and cumulative mechanical fatigue damage are used to predict lifespan, including: The total cumulative damage is determined based on the cumulative mechanical fatigue damage and the cumulative thermal fatigue damage; based on the total cumulative damage, the total fatigue life of the radiator is predicted by linear extrapolation.

8. The method for predicting the fatigue life of a radiator structure based on vibration test bench data according to claim 7, characterized in that, The total cumulative damage satisfies the expression: ； In the formula, For the first Total cumulative damage over a time window; For the first Cumulative mechanical fatigue damage over a time window; For the first Cumulative thermal fatigue damage over a time window; is the damage coupling coefficient.

9. The method for predicting the fatigue life of a radiator structure based on vibration test bench data according to claim 7, characterized in that, The total fatigue life satisfies the expression: ； In the formula, The predicted total fatigue life; This represents the time elapsed since the current experiment began. This represents the total cumulative damage within the current time window.

10. The method for predicting the fatigue life of a radiator structure based on vibration test bench data according to claim 7, characterized in that, Also includes: The prediction of total fatigue life is updated dynamically based on real-time total cumulative damage.

Images