A method for predicting the remaining life of metal components based on magnetic coercivity

Abstract

This invention provides a method for predicting the remaining service life of metallic components based on magnetic coercivity, relating to the fields of materials testing and service life prediction. The method first determines the coercivity values ​​in the original and failed states of the material to establish a coercivity reference range. During the component's service life, multi-point in-situ measurements are performed using pulse magnetization to calculate the current service coercivity value, thereby obtaining a continuous damage severity factor. Then, utilizing the nonlinear relationship between the damage severity factor and the cumulative damage life fraction established through accelerated damage testing, and combining it with component geometric features, stress distribution, and environmental parameters, a service life prediction model is constructed. Finally, the remaining safe service life is calculated based on the model and service history, and a confidence interval is provided. Simultaneously, the test data is stored in a database for model updates. This achieves continuous quantitative characterization of damage states and reliable prediction of remaining service life.

Description

Technical Field

[0001] This invention relates to the field of materials testing and life prediction technology, specifically a method for predicting the remaining life of metal components based on magnetic coercivity. Background Technology

[0002] In industries such as petrochemicals, power, and energy, numerous metal structural components operate under complex conditions including high pressure, high temperature, and corrosive media for extended periods. As operating time increases, microscopic damage accumulates within the materials, leading to degradation of their mechanical properties and potentially causing sudden failures that seriously threaten personnel and facility safety. Currently, damage assessment and remaining service prediction for such in-service components primarily rely on macroscopic inspections after periodic shutdowns, empirical formula calculations, or localized destructive sampling for metallographic and mechanical property testing.

[0003] Although various non-destructive testing (NDT) techniques have been applied to industrial inspection, most existing methods focus on identifying macroscopic defects or measuring specific physical parameters, making it difficult to directly, sensitively, and quantitatively characterize the continuous evolution of microscopic damage throughout the entire process of a material's lifespan, from its initial state to failure. This results in the inability to obtain a universally applicable indicator that is strictly synchronized with the material's damage process and can be used to quantitatively describe its health. Precisely because of this lack of continuous damage quantification throughout the entire lifespan, existing remaining life prediction methods generally suffer from poor adaptability, large result dispersion, and low confidence levels, failing to provide accurate and reliable scientific support for differentiated operation and maintenance and safety decisions for in-service components. Summary of the Invention

[0004] To achieve the above objectives, this invention proposes a method for predicting the remaining life of metal components based on magnetic coercivity, comprising:

[0005] S1. Based on the material type and service status of the metal component, obtain the initial coercivity value of the material in the original undamaged state and the fracture coercivity value in the failure state, so as to determine the coercivity reference range for damage quantification.

[0006] S2. During the service of the metal component, a detection area is selected, and multi-point in-situ measurement of the detection area is performed using pulse magnetization. The obtained set of current coercivity measurement values ​​is processed, and after removing outliers, the average value is calculated as the current service coercivity value.

[0007] S3. Based on the coercivity reference range and the current service coercivity value, calculate the current damage degree factor of the metal component, wherein the current damage degree factor changes continuously between 0 and 1, corresponding to the undamaged state and the complete failure state, respectively.

[0008] S4. Input the calculated damage degree factor into the constructed life prediction model. The life prediction model establishes a nonlinear relationship between the damage degree factor and the cumulative damage life score by simulating accelerated damage tests with the same materials and working conditions as the component under test, and introduces component geometric features, stress distribution and environmental parameters for model correction.

[0009] S5. Based on the cumulative damage life score output by the life prediction model, combined with the service time and load history of the metal component, calculate its remaining safe service life and provide a confidence interval assessment. At the same time, store the current service coercivity value measured this time and the calculated data into the database for updating and optimizing the life prediction model.

[0010] As a further technical solution, the process of determining the coercivity reference range is as follows: By acquiring samples with the same material origin and processing technology as the metal component under test, a non-destructive reference sample set and a failure reference sample set are constructed respectively. For the non-destructive reference sample set, based on the material's magnetization curve, multiple discrete magnetization intensity levels are selected within the range from the initial magnetization region to the saturation region. For each sample, repeated magnetization and coercivity measurement operations are performed at each magnetization intensity level to obtain multiple sets of raw data. Subsequently, statistical methods are used to identify and remove outliers in each set of data, and the average value of the remaining valid data is calculated. The overall mean of the average values ​​obtained at all magnetization intensity levels is determined as the initial coercivity value H used to characterize the original state of the material. c0 For the failure baseline sample set, accelerated testing simulating the actual failure mechanism of the component is conducted to induce fracture. During the test, continuous tracking and measurement are performed on the critical damage stages before fracture to obtain a series of coercive force values. Data from this series of measurements that are in a stable plateau period are selected, and their average value is calculated to determine the fracture coercive force value under the failure state. Based on the obtained initial coercivity value H c0 and fracture coercivity value Furthermore, a statistical distribution relationship reflecting the inherent performance dispersion of the material batch is introduced for calculation. Specifically, the standard deviations of the coercivity measurements of the non-destructive sample set and the failure sample set are calculated separately, and the combined standard deviation is calculated based on this. Then, the distribution coefficient k is determined according to the preset confidence level. Finally, the initial coercivity value minus k times the combined standard deviation is used as the lower limit of the reference interval, and the fracture coercivity value plus k times the combined standard deviation is used as the upper limit of the reference interval, thereby determining the coercivity benchmark reference interval [H]. c0 -kσ, +kσ], where σ is the combined standard deviation of the coercivity measurement values, reflecting the dispersion of material properties.

[0011] As a further technical solution, the process of performing multi-point in-situ measurements and calculating the current service coercivity value after selecting the detection area is as follows: After selecting the detection area, based on the geometric characteristics of the metal component and the known distribution of service stress concentration areas, a measurement point array covering the detection area is planned; the planning principle of this array is to ensure that the point distribution can simultaneously reflect the overall damage state of the component and the damage state of known high-stress local areas; pulse magnetization is used to repeatedly excite and collect coercivity data for each measurement point in the measurement point array with a constant magnetization intensity and pulse width; for each measurement point, a raw data sequence is obtained, and the range of multiple measurements at the same measurement point is compared to see if it exceeds the set value based on the first measurement at that point. Anomalies are identified using a dynamic threshold: this threshold is set as a fixed percentage of the initial measurement value; if a measurement value deviates from the median of the sequence by more than this threshold, it is considered an abnormal measurement value and is removed; after removing abnormal values, the arithmetic mean of the remaining valid measurements at that point is calculated as the representative coercivity value for that measurement point; after obtaining the representative coercivity values ​​of all measurement points, anomalies are further identified based on the numerical distribution of all points: the mean and standard deviation of all representative coercivity values ​​are calculated, and points falling outside the range of plus or minus three times the standard deviation of the mean are identified as anomalies and removed; finally, the arithmetic mean of the representative coercivity values ​​of all remaining measurement points is calculated to obtain the current service coercivity value used to characterize the average damage state of the entire detection area. .

[0012] As a further technical solution, the specific process of measuring coercivity using pulse magnetization is as follows: a positive magnetization pulse with sufficient intensity to bring the material at the measurement point to a magnetic saturation state is applied; the intensity of this magnetization pulse is determined through prior experiments on similar material samples to ensure sufficient orientation of the magnetic domains within the material; after the positive pulse is removed and a period of time is waited to ensure sufficient magnetic stability, a sequence of reverse magnetization pulses is applied; the intensity of this reverse pulse sequence increases sequentially according to a preset fixed difference ΔH, and after each reverse pulse is applied and the same stabilization time is waited, the residual magnetization intensity M at the measurement point is measured and recorded. rem When a change in the sign of the remanence intensity values ​​obtained from two consecutive measurements is detected (i.e., from positive to negative or from negative to positive), it indicates that the reverse field strength value that would bring the remanence intensity to zero lies between these two pulse intensities. At this point, a set of reverse pulses with a smaller step size (e.g., ΔH / 10) is inserted between the previous and current reverse pulse intensities for supplementary measurement to capture the sign change point. Finally, the remanence intensity values ​​(M) at the two measurement points before and after the sign change are calculated. rem 1,M rem 2) and its corresponding pulse intensity (H1, H2), using the linear interpolation formula H c =H1-M rem1(H2-H1) / (M rem 2-M rem 1) Calculate the reverse magnetization field strength required to bring the residual magnetization intensity to zero precisely. This value is then determined as the current coercivity measurement value under this excitation. The duration of each magnetization pulse, i.e., the pulse width, and the waiting time for stabilization after the pulse is removed are all determined through previous experiments to be the shortest time required to ensure that the magnetic state of the material is fully stable after the pulse is removed.

[0013] As a further technical solution, the process of calculating the current damage severity factor D is as follows: First, the corresponding initial coercivity value H is extracted from the determined coercivity benchmark reference range. c0 and fracture coercivity value Then, the current service coercivity value obtained from on-site measurement and calculation will be... Substitute into the formula Perform calculations; the formula has a clear physical meaning: molecule ( -H c0 The denominator () represents the increase in coercivity due to injury. -H c0 The ratio D represents the total change in coercivity from no damage to complete failure; the ratio D directly quantifies the proportion of current damage to the total failure potential, thus varying continuously between 0 and 1; when =H c0 When D=0, it indicates a lossless state; when = When D=1, it indicates that the system is in a state of complete failure.

[0014] As a further technical solution, the construction and modification process of the life prediction model is as follows: First, by simulating accelerated damage tests on the material and working conditions of the component under test, discrete data points showing the current damage severity factor D increasing with time t are obtained; piecewise function fitting is performed on these discrete data points to establish a continuous nonlinear functional relationship between the current damage severity factor D and the theoretical cumulative damage life fraction L(t); this relationship is then personalized for the specific component under test by performing the following steps: First, the key geometric dimensions of the component under test are extracted from its design drawings, including the ratio R between the minimum wall thickness and the adjacent wall thickness. t The weld transition angle θ and two other geometric parameters are input into a multi-parameter geometric influence empirical formula F, which has been pre-calibrated through numerous standard specimen tests. geo (R t The geometric correction coefficient K is calculated from θ). g The empirical formula for the multi-parameter geometric influence was established through multiple regression analysis, reflecting the correlation between the stress concentration factor and the rate of change of coercivity. Secondly, the maximum principal stress value σ was obtained based on the service load spectrum of the component under test. maxand the area ratio of stress concentration region A ratio Based on the sensitivity curves of coercivity versus stress obtained from laboratory uniaxial tensile tests, the stress correction factor K was determined by bilinear interpolation. s The sensitivity curve describes the relative rate of change of material coercivity relative to the stress-free state under different static stress levels. Finally, the temperature fluctuation range ΔT and the historical average concentration C of the main corrosive components of the medium in the service environment of the component under test are obtained. Using these two parameters as an index, an environmental damage rate comparison table established through multiple sets of accelerated corrosion tests with different combinations of temperature and medium concentration is consulted to obtain the environmental correction factor K. e The environmental damage rate comparison table establishes a quantitative relationship between environmental erosion factors and the growth rate of additional resistance to magnetic domain wall movement; finally, the theoretical cumulative damage lifetime fraction L is... theory Multiply by the geometric correction factor K in sequence g Stress correction factor K s and environmental correction factor K e L c =L theory ×K g ×K s ×K e The cumulative damage life score of the component under the actual service conditions is obtained after correction.

[0015] As a further technical solution, the process of establishing a continuous nonlinear functional relationship between the current damage severity factor D and the theoretical cumulative damage life fraction is as follows: The discrete data points obtained from accelerated damage testing, i.e., the current damage severity factor D and the corresponding normalized test time t, are divided into stages and fitted piecewise. First, based on the rate of change of the current damage severity factor D, the abrupt change points of the D growth rate in the data sequence are identified by calculating the first-order difference of D with respect to t between adjacent data points. These abrupt change points mark the transformation of the damage mechanism. Using the identified abrupt change points as boundaries, the entire damage process is divided... The damage is divided into three stages: initial damage, stable damage, and accelerated damage. Then, data points within each stage are modeled using functions: In the initial damage stage, damage is mainly characterized by micro-defect nucleation with a gradually slowing growth rate, and a logarithmic function L1(D) = αln(βD+1) is used for fitting; in the stable damage stage, damage is mainly characterized by stable microcrack propagation with a relatively constant growth rate, and a linear function L2(D) = γD + δ is used for fitting; in the accelerated damage stage, damage is mainly characterized by unstable propagation of the main crack with a rapidly accelerating growth rate, and a power function L3(D) = εD is used. ζ A fitting process is performed; where α, β, γ, δ, ε, and ζ are real coefficients determined by least-squares fitting of data points at each stage; ultimately, a piecewise function L with D as the independent variable is formed. theory (D): When D is in the interval [0, D1), Ltheory =L1(D); when D is in the interval [D1, D2), L theory =L2(D); when D is in the interval [D2,1], L theory =L3(D).

[0016] As a further technical solution, the process of calculating the remaining safe service life and performing confidence interval assessment is as follows: First, define the remaining service life factor R, whose calculation formula is... , where H c0 This is the initial coercive force value; This represents the fracture coercivity value. The current service coercivity value; this remaining service life factor is complementary to the damage severity factor D, i.e., R = 1 - D, which physically represents the proportion of remaining damage capacity to total capacity; based on R and the service calendar time T extracted from the component service record. passed Through formula T pred = (R / (1-R))T passed Preliminary forecast of remaining safe service life T pred The formula is derived based on the equivalent time assumption of linear accumulation of damage. Subsequently, a confidence interval assessment is performed to quantify the uncertainty of the prediction: based on the original data sequence obtained from multiple repeated measurements at each valid measurement point during the measurement phase, the standard deviation σ of the representative coercive force values ​​at all valid measurement points is calculated. Hc Based on the error propagation principle, the calculation is performed using σ. Hc Uncertainty U of the resulting current damage level factor D D The calculation formula is: Then calculate the uncertainty U of the remaining lifetime factor R. R Its value is related to U D Equal; ultimately, the remaining safe service life T pred The upper and lower limits of the confidence interval are respectively given by the upper limit formula T. predup =(R+U R ) / (DU D ))T pred And the lower limit formula T predlow =((RU R ) / (D+U D ))T pred Confirm; the current service coercivity value measured in this study, and the calculated D, R, and T values. pred The confidence interval, along with the unique identifier of the component corresponding to this test, the test time, and the test area information, are stored in the historical database. The accumulated data in this database is used for periodic regression updates and optimization of the coefficients of similar component life prediction models.

[0017] This invention provides a method for predicting the remaining life of metal components based on magnetic coercivity, which has the following beneficial effects:

[0018] 1. This invention solves the problem that magnetic detection methods cannot continuously and stably quantify material damage by constructing a coercivity reference range based on the original and failure states of materials, and by combining multi-point in-situ measurements with strict outlier removal, thereby significantly improving the accuracy and repeatability of damage quantification evaluation.

[0019] 2. This invention proposes a damage degree factor mapping method that includes piecewise nonlinear fitting and multi-dimensional correction based on geometric, stress, and environmental parameters, thereby achieving an effective and reliable conversion from laboratory test data to actual engineering life assessment.

[0020] 3. This invention feeds back on field measurement data, calculated damage factors, and prediction results to a database in real time and uses them to dynamically optimize the life prediction model, thus constructing a detection method with self-learning and adaptive capabilities, which continuously improves the intelligence level of the entire life prediction process for metal components. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the process of the present invention;

[0022] Figure 2 This is a sample drawing of a Q345R welding test plate.

[0023] Figure 3 This is a graph showing the relationship between creep damage and coercivity.

[0024] Figure 4 A diagram illustrating the intrinsic relationship between fatigue damage and coercivity;

[0025] Figure 5 This is a diagram showing the effect of wall thickness on coercivity.

[0026] Figure 6 This is a graph showing the relationship between the degree of degradation and coercivity. Detailed Implementation

[0027] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] refer to Figure 1The present invention discloses a method for predicting the remaining service life of metal components based on magnetic coercivity, the specific embodiment of which is as follows: This embodiment takes an old petrochemical hydrogenation reactor shell in service as the test object; the main body of the reactor is made of 12Cr1MoVG pearlitic heat-resistant steel, some of its connecting pipes are made of Q345R steel, the design pressure is 8.5MPa, the design temperature is 480℃, and it has accumulated about 150,000 hours of service; this method quantitatively evaluates the damage state of its main body material and key welded joints without damaging the component, and predicts its remaining safe service life.

[0029] First, the coercivity reference range for damage quantification was determined. Since the original undamaged state sample of the reactor under test could not be directly obtained, a homologous substitution method was used. For the main material, 12Cr1MoVG steel, standard samples were cut from steel plates of the same batch and heat treatment state archived by the equipment manufacturer to construct a non-destructive reference sample set. The sample set was measured using a ferromagnetic material safety analyzer. Referring to the relationship between the mechanical properties and coercivity of common carbon steel materials shown in Table 1 of the experimental data, the initial coercivity value H of 12Cr1MoVG steel in its original state was determined. c0 4.1 A·cm -1 The initial coercivity value H was determined through repeated measurements and statistics. c0 4.1 A·cm -1 And calculate its standard deviation; for the welded joint portion, a welding test plate with the same process as the actual joint needs to be prepared to obtain a benchmark; such as Figure 2 The coercivity measurement results of the Q345R welding test plate shown are shown in Table 2.

[0030] Table 1 Relationship between mechanical properties and coercivity

[0031]

[0032] Note: Materials from different batches and manufacturers may vary.

[0033] Table 2. Coercivity Measurement Results in the Original State

[0034]

[0035] By analyzing multiple sets of measurement data of the weld, upper heat-affected zone, and lower heat-affected zone in their original state in Table 2, the average value and distribution range of the original coercivity in each region can be statistically obtained. For example, the average value in the weld region is approximately 5.8 A·cm. -1 This serves as the initial coercivity benchmark for the region; the failure benchmark needs to be obtained through accelerated testing simulating the actual failure mechanism; for the 12Cr1MoVG base material, accelerated creep tests are conducted at 580℃ and 110MPa using materials from the same batch until fracture; Figure 3The creep damage test shown indicates that the coercivity increases sharply when the damage is significant; as shown in Table 1, the fracture coercivity values ​​are in the fracture state. 8.4 A·cm -1 Based on the stability measurement platform before fracture in this experiment, the fracture coercivity value was determined. 8.4 A·cm -1 Based on the combined initial and fracture values, and considering the inherent dispersion of material properties, the coercivity reference range for the 12Cr1MoVG base material was ultimately determined to be 3.9 A·cm. -1 Up to 8.6 A·cm -1 The failure coercivity benchmark for welded joints needs to be calibrated separately based on the corresponding fatigue or brittle fracture test data.

[0036] After determining the reference range, in-situ measurements were performed under the service conditions of the reactor shell. The lower circumferential zone of the shell and one longitudinal weld were selected as the inspection areas. Based on the geometric characteristics of the components and the known distribution of stress concentration areas, an array of measurement points was planned. In the base material area, points were evenly distributed circumferentially; in the longitudinal weld area, reference points were used. Figure 2 For example, a measurement point array is arranged along the weld centerline and the heat-affected zones on both sides; pulse magnetization is used to excite and acquire data at each measurement point; the specific process is as follows: a positive magnetization pulse with sufficient intensity to achieve magnetic saturation is applied to the measurement point, for example, an intensity greater than 20 A·cm. -1 Then, a sequence of reverse magnetization pulses with intensities increasing sequentially by a fixed difference is applied, and the remanent magnetization intensity after each pulse is recorded. When a change in the sign of the remanent magnetization intensity between two adjacent measurements is detected, a small-step reverse pulse is inserted between these two pulse intensities for supplementary measurement. Finally, the reverse field strength required to bring the remanent magnetization intensity to zero is calculated through linear interpolation, which is the single coercivity measurement value at that point. This process is repeated 5 times for each measurement point to obtain the original data sequence. Subsequently, the data is processed: First, for the 5 data points at each measurement point, outliers at that point are identified and removed by comparing whether their range exceeds a dynamic threshold set based on the initial measurement value. This threshold is set to 10% of the initial measurement value or 0.5 A·cm. -1 Calculate the average of the effective measurements at that point as its representative coercivity value; after obtaining the representative values ​​of all measurement points, use statistical methods to identify and remove overall outliers based on the numerical distribution of all points; finally, calculate the arithmetic mean of the representative values ​​of all remaining measurement points to obtain the current service coercivity value. Assuming that the effective average value obtained after processing the parent material area is 6.6 A·cm -1 After processing the weld area, the average value obtained was 7.8 A·cm. -1 .

[0037] After obtaining the current coercivity measurement value, calculate the current damage severity factor D; for the parent material region, extract the initial coercivity value H from the reference interval. c0 The values ​​are 4.1 A·cm⁻¹ and fracture coercivity. It is 8.4 A·cm⁻¹; according to the formula Perform calculations; substitute the parent material region =6.6A·cm-1, so D≈0.581; For the weld area, it is necessary to use its own original benchmark and failure benchmark for calculation; The current damage degree factor D changes continuously between 0 and 1, which quantitatively characterizes the damage state of the material from no damage to complete failure. In this example, the base material area is already in the moderate damage stage.

[0038] After quantifying the current damage level, it needs to be linked to the lifespan. The damage level factor D is input into the constructed lifespan prediction model. The core of this lifespan prediction model is to establish a nonlinear relationship between the damage level and the cumulative damage lifespan fraction by simulating accelerated damage tests with the same material and operating conditions as the component under test. For example, accelerated creep tests are conducted using 12Cr1MoVG steel samples from the same batch, with periodic interruptions and coercivity measurements to obtain a series of results. Figure 3 The data points for D and normalized test time t are shown. Piecewise function fitting is performed on these discrete data points: first, the abrupt change points in the growth rate of D are identified, dividing the process into initial, stable, and accelerated damage stages; then, as shown... Figure 4 The fatigue damage data shown also reveals similar piecewise characteristics. Logarithmic, linear, and power functions were used to fit the data for each stage to determine the function coefficients for each stage. (Where (a) is the parallel detection method, i.e., the coercive force detection direction is parallel to the fatigue load application direction; (b) is the perpendicular detection method, i.e., the coercive force detection direction is perpendicular to the fatigue load application direction. The comparison results show that the detection direction has a certain influence on the coercive force response value. In actual testing, the consistency of the detection direction should be maintained to ensure the comparability of the measurement results). Thus, a theoretical cumulative damage life fraction L is formed with D as input. theory The continuous nonlinear functional relationship; however, the theoretical model is based on standard specimens and needs to be modified for specific components; the modification process introduces geometric, stress, and environmental parameters: geometric parameters such as minimum wall thickness ratio and weld transition angle are extracted from the design drawings, and input from... Figure 5 The empirical formula for calibrating data such as the wall thickness influence relationship is shown, and the geometric correction factor K is calculated. g(where (a) is a schematic diagram of the sample shape type with different wall thicknesses; (b), (c), and (d) correspond to the variation law of coercivity of Q345R steel, 14Cr1MoR steel, and 20# steel under different wall thicknesses, respectively, revealing the material correlation influence of wall thickness parameters on the measured coercivity value); Based on the service load spectrum, the maximum principal stress and stress concentration are obtained, combined with the coercivity-stress sensitivity curve obtained from the laboratory tensile test, and with reference to the stress influence reflected by the change in coercivity before and after annealing of the welded test plate, the stress correction coefficient K is determined. s ; Obtain the service environment temperature and medium concentration, and query the data generated by... Figure 6 The environmental damage rate comparison table established from accelerated corrosion test data, showing the relationship between degradation degree and coercivity, yields the environmental correction factor K. e Finally, the theoretical cumulative damage life fraction L is calculated. theory Multiply by these three correction factors in sequence to obtain the corrected cumulative damage life score applicable to the specific component.

[0039] Finally, the remaining useful life is calculated and assessed; first, the remaining useful life factor R is calculated using the formula: For the parent material region, R≈0.419; combined with the service time T of the component. passed For 150,000 hours, using formula T pred = (R / (1-R))T passed Preliminary forecast of remaining safe service life T pred ≈108,000 hours; To assess the reliability of the prediction, confidence interval calculation is required: Based on multiple measurement data, calculate the standard deviation σ of the representative coercive force values ​​at all valid measurement points. Hc Assuming a value of 0.25 Acm⁻¹, the uncertainties of the damage factor D and the remaining lifetime factor R are estimated based on the error propagation principle. Then, the upper and lower limits of the remaining lifetime confidence interval are calculated using the formula, yielding a predicted remaining lifetime range of approximately 62,000 to 102,000 hours at a 95% confidence level. All key data from this test, including... D, R, T pred The data, along with its confidence interval, is stored in the database along with information such as component identification and inspection time. This historical data will be used for subsequent periodic inspections and comparisons of the same component. Through the continuous accumulation of damage evolution data, the fitting and correction parameters in the life prediction model will be fed back and optimized, thereby achieving dynamic updates and self-evolution of the model and continuously improving prediction accuracy.

[0040] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for predicting the remaining life of metal components based on magnetic coercivity, characterized in that, Includes the following steps: S1. Based on the material type and service status of the metal component, obtain the initial coercivity value of the material in the original undamaged state and the fracture coercivity value in the failure state, so as to determine the coercivity reference range for damage quantification. S2. During the service of the metal component, a detection area is selected, and multi-point in-situ measurement of the detection area is performed using pulse magnetization. The obtained set of current coercivity measurement values ​​is processed, and after removing outliers, the average value is calculated as the current service coercivity value. S3. Based on the coercivity reference range and the current service coercivity value, calculate the current damage degree factor of the metal component, wherein the current damage degree factor changes continuously between 0 and 1, corresponding to the undamaged state and the complete failure state, respectively. S4. Input the calculated damage degree factor into the constructed life prediction model. The life prediction model establishes a nonlinear relationship between the damage degree factor and the cumulative damage life score by simulating accelerated damage tests with the same materials and working conditions as the component under test, and introduces component geometric features, stress distribution and environmental parameters for model correction. S5. Based on the cumulative damage life score output by the life prediction model, combined with the service time and load history of the metal component, calculate its remaining safe service life and provide a confidence interval assessment. At the same time, store the current service coercivity value measured this time and the calculated data into the database for updating and optimizing the life prediction model. S5 includes: defining and calculating the remaining lifetime factor R, the calculation formula of which is as follows: ,in, This is the initial coercive force value; This represents the fracture coercivity value. The current service coercivity value; based on R and the service calendar time extracted from the component service record. Through formula Preliminary forecast of remaining safe service life Subsequently, confidence interval assessment was performed, specifically: based on the original data sequence obtained from repeated excitation and acquisition at each measurement point, the standard deviation σ of the representative coercivity values ​​of all valid measurement points was calculated. Hc Based on the error propagation principle, the calculation is performed using σ. Hc Uncertainty U of the resulting current damage level factor D D and the uncertainty U of the remaining lifetime factor R R Ultimately, the remaining safe service life The upper and lower limits of the confidence interval are respectively given by the upper limit formula. and lower limit formula Confirm; determine the current service coercivity value measured in this study. The calculated D, R, The confidence interval, along with the component identification, testing time, and testing area information corresponding to this test, are stored in the database.

2. The method for predicting the remaining life of metal components based on magnetic coercivity according to claim 1, characterized in that: S1 includes: constructing a non-destructive benchmark sample set and a failure benchmark sample set by acquiring samples with the same material origin and processing technology as the metal component to be tested; for the non-destructive benchmark sample set, selecting multiple discrete magnetization intensity levels within the range from the initial magnetization region to the saturation region based on the magnetization curve of the material, repeatedly magnetizing and measuring them, and calculating the initial coercivity value used to characterize the original state of the material after removing outlier data; for the failure benchmark sample set, accelerating the component to fracture through a simulated actual failure mechanism test, and continuously tracking and measuring before fracture, determining the average value of the obtained stable measurement results as the fracture coercivity value under the failure state; and calculating the coercivity benchmark reference interval by introducing a statistical distribution relationship reflecting the inherent performance dispersion of the material batch based on the initial coercivity value and the fracture coercivity value.

3. The method for predicting the remaining life of metal components based on magnetic coercivity according to claim 1, characterized in that: S2 includes: after selecting the detection area, planning a measurement point array covering the detection area to reflect its overall and local damage state based on the geometric characteristics of the metal component and the known distribution of service stress concentration areas; using pulse magnetization, repeatedly exciting and acquiring coercive force data for each measurement point in the measurement point array with constant magnetization intensity and pulse width; for the original data sequence obtained for each measurement point, identifying and removing abnormal measurement values ​​of the point by comparing whether the range of multiple measurements of the same measurement point exceeds a dynamic threshold set based on the first measurement value of the point, and then calculating the average of the effective measurement values ​​of the point as the representative coercive force value of the point; after obtaining the representative coercive force values ​​of all measurement points, identifying and removing abnormal points that have statistical differences from the data of other points based on the numerical distribution of all points, and finally calculating the current service coercive force value by arithmetically averaging the representative coercive force values ​​of all remaining measurement points.

4. The method for predicting the remaining life of metal components based on magnetic coercivity according to claim 3, characterized in that: The pulse magnetization method includes: applying a positive magnetization pulse with sufficient intensity to magnetically saturate the material at the measurement point; then applying a sequence of reverse magnetization pulses with intensities increasing sequentially by a fixed difference, and measuring and recording the residual magnetization intensity at the measurement point after each reverse pulse; when a change in the sign of the residual magnetization intensity values ​​obtained from two adjacent measurements is detected, a set of reverse pulses with a step size smaller than the fixed difference is inserted between the previous and current reverse pulse intensities for supplementary measurement; finally, based on the residual magnetization intensity values ​​of the two measurement points before and after the sign change and their corresponding pulse intensities, the reverse magnetization field intensity value required to bring the residual magnetization intensity to zero is calculated by linear interpolation, and this value is determined as the current coercivity measurement value.

5. The method for predicting the remaining life of metal components based on magnetic coercivity according to claim 1, characterized in that: The process of calculating the current damage severity factor in S3 is as follows: extract the corresponding initial coercivity value and fracture coercivity value from the coercivity reference interval; then, calculate the current damage severity factor of the current service coercivity value relative to the initial coercivity value. ;in, It refers to the coercivity value of the object under its original state, that is, the initial coercivity value; It refers to the coercivity value of the tested object under failure conditions, i.e., the fracture coercivity value. It refers to the coercivity value of the object under its current service conditions.

6. The method for predicting the remaining life of metal components based on magnetic coercivity according to claim 1, characterized in that: The life prediction model includes: obtaining discrete data points of the current damage severity factor increasing over time by simulating accelerated damage tests on the material and working conditions of the component under test; fitting the discrete data points to establish a continuous nonlinear function relationship between the current damage severity factor and the theoretical cumulative damage life fraction; and correcting this continuous nonlinear function relationship for a specific component under test by performing the following steps: extracting key geometric dimensions of the component under test from its design drawings, including the ratio of minimum wall thickness to adjacent wall thickness and weld transition angle; inputting the geometric parameters into a multi-parameter geometric influence empirical formula to calculate the geometric correction coefficient; and calculating the service load spectrum of the component under test. The maximum principal stress value and the area ratio of the stress concentration region are obtained. Combined with the coercivity and stress sensitivity curve of the material obtained from the laboratory uniaxial tensile test, the stress correction coefficient is determined by linear interpolation. The temperature fluctuation range of the service environment of the component under test and the historical average concentration of the main corrosive components of the medium are obtained. Using these two parameters as an index, an environmental damage rate comparison table established by multiple sets of accelerated corrosion tests with different combinations of temperature and medium concentration is consulted to obtain the environmental correction coefficient. Finally, the theoretical cumulative damage life score is multiplied by the geometric correction coefficient, the stress correction coefficient and the environmental correction coefficient in sequence to obtain the corrected cumulative damage life score.

7. The method for predicting the remaining life of a metal component based on magnetic coercivity according to claim 6, characterized in that: The process of establishing a continuous nonlinear functional relationship is as follows: Discrete data points obtained from accelerated damage testing, i.e., the current damage severity factor D and the corresponding normalized test time t, are used to establish the relationship between the current damage severity factor D and the theoretical cumulative damage lifespan fraction L(t). Based on the rate of change of the current damage severity factor D, the abrupt change points in the growth rate of D in the data sequence are identified by calculating the first-order difference of D with respect to t between adjacent data points. Using these abrupt change points as boundaries, the entire damage process is divided into an initial damage stage, a stable damage stage, and an accelerated damage stage. Then, function modeling is performed on the data points within each stage: in the initial damage stage, a logarithmic function L1(D) = αln(βD+1) is used for fitting; in the stable damage stage, a linear function L2(D) = γD + δ is used for fitting; and in the accelerated damage stage, a power function L3(D) = εD is used. ζ Fitting is performed; where α, β, γ, δ, ε and ζ are all real coefficients determined by least squares fitting of data points at each stage; finally, D is used as the discrimination condition, and when D is in different numerical ranges, the corresponding stage function is automatically selected to calculate the theoretical cumulative damage life score, thereby forming a continuous nonlinear function relationship.

Images