A hysteretic model parameter identification method based on hysteretic characteristic point mapping

Abstract

The application discloses a hysteretic model parameter identification method based on a hysteretic characteristic point mapping, and belongs to the technical field of civil engineering structure testing and structure numerical analysis. The method comprises the following steps: obtaining discrete hysteretic data and corresponding test process identification obtained by a pseudo-static test of a node of an engineering structure. The discrete hysteretic data comprises generalized displacement and generalized force used for forming a hysteretic curve, and the test process identification comprises sampling sequence identification and loading level identification. Skeleton characteristic points, pinch characteristic points, degradation and reloading tracking characteristic points are extracted from the discrete hysteretic data, different characteristic point groups are used for evaluating corresponding parameter groups, and a single-cycle energy dissipation area is used for screening candidate parameter groups in advance, so that the hysteretic model parameter identification process can constrain skeleton morphology, pinch position, degradation trend and energy dissipation characteristics, thereby solving the problem that the existing overall error fitting mode is prone to causing insufficient physical consistency of parameter identification results.

Description

Technical Field

[0001] This invention relates to the field of civil engineering structural testing and structural numerical analysis technology, and more specifically, to a method for identifying hysteresis model parameters based on hysteresis feature point mapping. Background Technology

[0002] The connection nodes of engineering structures will produce obvious hysteretic responses under the action of earthquakes, wind vibrations or cyclic loads. The accuracy of the hysteretic model parameters of the nodes directly affects the reliability of structural nonlinear analysis, seismic performance assessment and damage evolution judgment.

[0003] In existing methods for identifying parameters of nodal hysteresis models, the errors between corresponding sampling points of the experimental hysteresis curve and the simulated hysteresis curve, the errors between a few peak points, or the errors between several control points are often combined into a single objective function, and this objective function is directly used as the basis for parameter optimization.

[0004] While this method can reduce the overall coordinate error of the curve, it lacks individual constraints on the characteristics of the skeleton bearing, pinching path, unloading stiffness degradation, strength degradation, and single-cycle energy consumption during the recognition process. This can easily lead to situations where the local curve fit is good but the pinching valley point is misaligned, the unloading path is distorted, or the single-cycle energy consumption error is relatively large.

[0005] Especially when there are loading pauses, acquisition disturbances and repeated cycles of different loading levels in the quasi-static test data of engineering structural nodes, simply relying on the overall error optimization can easily cause mutual compensation between skeleton parameters, pinching parameters and degradation parameters, making it difficult for the identified hysteresis model parameters to maintain the consistency of curve shape, degradation trend and energy dissipation characteristics at the same time. Summary of the Invention

[0006] To address the problems mentioned in the background section, the present invention provides the following technical solution: A method for identifying hysteresis model parameters based on hysteresis feature point mapping includes the following steps: The discrete hysteresis data and corresponding test process identifiers obtained from the quasi-static test of the engineering structure nodes are acquired. The discrete hysteresis data includes generalized displacement and generalized force used to form the hysteresis curve, and the test process identifiers include sampling sequence identifiers and loading level identifiers. The loading direction reversal point is determined based on the generalized displacement increment between adjacent sampling points, and the data segment between two adjacent loading direction reversal points is taken as the loading branch segment. According to the loading level identifier, the order of loading direction change, and the sampling order between adjacent loading branch segments, multiple loading branch segments belonging to the same cycle process are combined into a hysteresis loop, thereby decomposing the experimental hysteresis curve into multiple hysteresis loops. Based on data points where the absolute value of the generalized force exceeds the historical maximum absolute value of the generalized force under the same loading direction, a skeleton envelope curve is formed, and hysteresis feature points are extracted from the hysteresis loop and the skeleton envelope curve. The hysteresis feature points include skeleton peak feature points, yield inflection feature points, pinching initiation points, pinching valley points, unloading stiffness degradation feature points, strength degradation feature points, and reloading path offset feature points. In the above, the yield transition feature point represents the position on the skeleton envelope curve where the stiffness decreases significantly, which is used to reflect the transition feature of the node from approximately elastic stress to post-yield stress. A feature point group is generated based on the hysteresis feature points, and an energy consumption constraint evaluation term is generated based on the closed path area formed by each hysteresis loop in the generalized force to generalized displacement plane. The feature point group includes a skeleton feature point group, a pinching feature point group, and a cyclic evolution feature point group. Candidate parameter groups are generated based on the preset value range of each parameter group in the hysteresis model to be identified. The candidate parameter groups are input into the hysteresis model to be identified to obtain the simulated hysteresis curve. The hysteresis feature points and single-turn energy consumption area of ​​the simulated hysteresis curve are extracted in the same way as the experimental hysteresis curve. First, candidate parameter groups are selected based on the single-turn energy consumption area error between the experimental hysteresis loop and the simulated hysteresis loop. Then, the selected candidate parameter groups are evaluated based on the matching error of the feature point group, and the candidate parameter groups that meet the preset error requirements are output as the parameters of the engineering structure node hysteresis model.

[0007] Furthermore, the loading direction reversal point is determined as follows: The effective increment is determined based on the generalized displacement difference between adjacent sampling points; When the sign of the current effective increment is opposite to that of the previous effective increment, and the absolute value of the generalized force of the current sampling point is greater than the preset noise threshold, the current sampling point is determined as a candidate point for reversing the loading direction. When multiple candidate points for reversing the loading direction occur due to test pauses or data acquisition disturbances, the sampling point where the absolute value of the generalized displacement reaches a local extreme value is selected as the loading direction reversal point. When the difference in the absolute value of the generalized displacement between multiple candidate sampling points is lower than the sampling resolution, the sampling point with the largest absolute value of the generalized force is selected as the loading direction reversal point.

[0008] Furthermore, the hysteresis feature points include skeleton peak feature points, yield inflection point feature points, pinching initiation point, pinching valley point, unloading stiffness degradation feature points, strength degradation feature points, and reloading path offset feature points. Among them, the skeleton peak feature point and yield inflection feature point are determined by the local peak and secant stiffness variation in the skeleton envelope curve; The pinching initiation point and pinching valley point are determined by the degree of deviation of the reloading path from the generalized force of the skeleton envelope under the same generalized displacement. The unloading stiffness degradation characteristic point and the strength degradation characteristic point are determined by the unloading secant stiffness reduction ratio and the peak generalized force reduction ratio of adjacent hysteresis loops within the same loading level; The reload path offset feature point is determined by the degree of deviation of the reload path from the target pointing line, which is determined by the zero-return residual deformation point and the next peak point.

[0009] Furthermore, the evaluation relationship between the feature point set and the parameter set is as follows: Skeleton feature point sets are used to evaluate skeleton parameter sets; The pinch feature point set is used to evaluate the pinch parameter set; The cyclic evolution feature point set is used to evaluate the stiffness degradation parameter set, strength degradation parameter set, and unloading rule parameter set; The energy consumption constraint evaluation item is used to evaluate the consistency of single-cycle energy consumption of candidate parameter groups.

[0010] Furthermore, the energy consumption constraint evaluation item is generated in the following manner: Calculate the energy consumption area of ​​a single loop in the test based on the closed path area of ​​the hysteresis loop. Calculate the simulated single-loop energy consumption area based on the closed path area of ​​the corresponding simulated hysteresis loop; Energy consumption constraint evaluation items are generated based on the relative error between the energy consumption area of ​​a single test lap and the energy consumption area of ​​a single simulated lap. When the relative error of any hysteresis loop is greater than the preset energy consumption error threshold, the corresponding candidate parameter group is removed.

[0011] Furthermore, the candidate parameter set is evaluated and output in the following manner: According to the loading direction, loading level identifier, hysteresis loop sequence and feature point type, the hysteresis feature points in the test hysteresis curve are matched with the hysteresis feature points in the simulated hysteresis curve. Calculate the matching errors for the skeleton feature point group, the pinched feature point group, and the cyclic evolution feature point group respectively; The comprehensive error is calculated based on the single-cycle energy consumption area error, the skeleton feature point group matching error, the pinching feature point group matching error, and the cyclic evolution feature point group matching error. Candidate parameter groups are selected in ascending order of comprehensive error, and the range of parameter values ​​is narrowed with the selected candidate parameter groups as the center. Candidate parameter groups are then regenerated and iterated. When a candidate parameter set meets the preset error requirement, the candidate parameter set is output as the hysteresis model parameter of the engineering structure node; when no candidate parameter set meets the preset error requirement is found after reaching the preset maximum number of iterations, the candidate parameter set with the smallest comprehensive error is output as the hysteresis model parameter of the engineering structure node.

[0012] In summary, the present invention has the following beneficial effects: By extracting skeleton peak feature points, yield inflection feature points, pinching initiation points, pinching valley points, unloading stiffness degradation feature points, strength degradation feature points, and reloading path offset feature points from the hysteresis curves of quasi-static tests of engineering structural nodes, and dividing these hysteresis feature points into skeleton feature point groups, pinching feature point groups, and degradation tracking feature point groups, the candidate parameter groups can be subject to the sub-item constraints of the corresponding features of skeleton parameters, pinching parameters, degradation parameters, and unloading rule parameters, thereby reducing the pinching misalignment, degradation distortion, and parameter mutual compensation problems caused by relying solely on single overall error optimization.

[0013] This invention also introduces single-cycle energy dissipation area error screening before the feature point group fitting evaluation, so that candidate parameter groups with small local feature point errors but large energy dissipation area deviations are eliminated first, thereby improving the consistency of the identified engineering structure node hysteresis model parameters in terms of curve shape and energy dissipation characteristics. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 This is a schematic diagram of the method flow of the present invention; Figure 2 This is a schematic diagram of the hysteresis feature point extraction method of the present invention; Figure 3 This is a schematic diagram illustrating the mapping relationship between the feature point group and the parameter group of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Example 1 The following is in conjunction with the appendix Figure 1-3 The present invention will be described in further detail below.

[0018] Please see Figure 1-3 This invention provides a technical solution: a method for identifying hysteresis model parameters based on hysteresis feature point mapping, such as... Figure 1-3As shown, it includes the following steps: The discrete hysteresis data and corresponding test process identifiers obtained from the quasi-static test of the engineering structure nodes are acquired. The discrete hysteresis data includes generalized displacement and generalized force used to form the hysteresis curve, and the test process identifiers include sampling sequence identifiers and loading level identifiers. The loading direction reversal point is determined based on the generalized displacement increment between adjacent sampling points, and the data segment between two adjacent loading direction reversal points is taken as the loading branch segment. According to the loading level identifier, the order of loading direction change, and the sampling order between adjacent loading branch segments, multiple loading branch segments belonging to the same cycle process are combined into a hysteresis loop, thereby decomposing the experimental hysteresis curve into multiple hysteresis loops. Specifically, the loading stage number is: a loading stage identifier divided according to the target rotation angle amplitude preset in the quasi-static cyclic loading test, used to distinguish the loading stage corresponding to different rotation angle amplitudes; for cyclic loading tests with progressively increasing rotation angle amplitudes, each target rotation angle amplitude corresponds to a loading stage number.

[0019] Specifically, the sampling sequence number is a consecutive number assigned to each sampling point by the data acquisition system according to the sampling order. This number is used to record the acquisition order of the angle and bending moment values, and to determine the sequential relationship in the process of loading direction change, loading branch segment division, hysteresis loop combination, and feature point extraction.

[0020] Generalized displacement is the value of rotation angle, and generalized force is the value of bending moment. Based on data points where the absolute value of the generalized force exceeds the historical maximum absolute value of the generalized force under the same loading direction, a skeleton envelope curve is formed, and hysteresis feature points are extracted from the hysteresis loop and the skeleton envelope curve. The hysteresis feature points include skeleton peak feature points, yield inflection feature points, pinching initiation points, pinching valley points, unloading stiffness degradation feature points, strength degradation feature points, and reloading path offset feature points. Among them, the pinching starting point is determined by the position where the reloading path begins, which is significantly lower than the generalized force of the skeleton envelope under the same generalized displacement. The pinch point is determined by the location in the reloading path where the degree of contraction is greatest relative to the skeleton envelope curve; A feature point group is generated based on the hysteresis feature points, and an energy consumption constraint evaluation term is generated based on the closed path area formed by each hysteresis loop in the generalized force to generalized displacement plane. The feature point group includes a skeleton feature point group, a pinching feature point group, and a cyclic evolution feature point group. The feature point group and energy consumption constraint evaluation term are respectively established with the parameter group in the hysteresis model to be identified. Among them, the skeleton feature point group is used to evaluate the skeleton parameter group, the pinching feature point group is used to evaluate the pinching parameter group, the cyclic evolution feature point group is used to evaluate the stiffness degradation parameter group, the strength degradation parameter group and the unloading rule parameter group, and the energy consumption constraint evaluation term is used to evaluate the single-cycle energy consumption consistency of the candidate parameter group. Candidate parameter groups are generated based on the preset value range of each parameter group in the hysteresis model to be identified. The candidate parameter groups are input into the hysteresis model to be identified to obtain the simulated hysteresis curve. The hysteresis feature points and single-turn energy consumption area of ​​the simulated hysteresis curve are extracted in the same way as the experimental hysteresis curve. First, candidate parameter groups are selected based on the single-loop energy consumption area error between the experimental hysteresis loop and the simulated hysteresis loop. Then, the selected candidate parameter groups are evaluated based on the matching error of the feature point group, and the candidate parameter groups that meet the preset error requirements are output as the parameters of the engineering structure node hysteresis model. This embodiment uses quasi-static cyclic loading test data of connection nodes in engineering structures as the object to identify the hysteresis model parameters of the nodes. The test object is a set of connection node specimens. The loading device performs reciprocating loading according to the target rotation angle amplitudes of ±0.005 rad, ±0.010 rad, ±0.015 rad, ±0.020 rad, and ±0.030 rad, with each loading stage repeating 2 cycles. The data acquisition device synchronously records the node rotation angle value, node bending moment value, loading stage number, and sampling sequence number at a sampling frequency of 10 Hz, obtaining a discrete hysteresis data table arranged in the sampling order.

[0021] After sampling, the discrete hysteresis data is preprocessed. Sampling points with angle or bending moment values ​​exceeding the corresponding sensor calibration range are removed. Sampling points with consecutive sampling sequence numbers whose angle difference is lower than the angle sampling resolution and whose bending moment change is lower than a preset disturbance threshold are merged. The preset disturbance threshold is 0.5% of the absolute value of the maximum bending moment in the test data. After preprocessing, the correspondence between sampling sequence number, loading stage number, angle value, and bending moment value is retained.

[0022] For the preprocessed discrete hysteresis data, the rotation angle increment between two adjacent sampling points is calculated. When the sign of the rotation angle increment changes, the corresponding position is taken as a candidate interval for the loading direction reversal position. For multiple consecutive candidate points caused by loading pauses or acquisition disturbances, the sampling point where the absolute value of the rotation angle reaches a local extremum is selected as the loading direction reversal point. When the difference in the absolute values ​​of the rotation angles of multiple candidate points is lower than the sampling resolution, the sampling point with the largest absolute value of the bending moment is selected as the loading direction reversal point. The data segment between two adjacent loading direction reversal points is considered as a hysteresis loop, thus decomposing the complete experimental hysteresis curve into multiple hysteresis loops.

[0023] After decomposing the hysteresis loop, skeleton envelope curves are extracted under both positive and negative loading. Under the same loading direction, if the absolute value of the bending moment at a sampling point exceeds the historical maximum absolute value of the bending moment previously observed in that loading direction, that sampling point is designated as a skeleton candidate point. Multiple skeleton candidate points are connected in the sampling order to form the skeleton envelope curve for the corresponding loading direction. For skeleton candidate points where there are no sampling points with the same rotation angle, linear interpolation is used to obtain the skeleton envelope bending moment value under the same rotation angle.

[0024] After obtaining the hysteresis loop and skeleton envelope curve, skeleton feature points, pinched feature points, degenerate feature points, and reload path feature points are extracted.

[0025] The sampling points on the skeleton envelope curve where the absolute value of the bending moment reaches a local maximum are taken as the skeleton peak feature points.

[0026] For yield transition feature points, first calculate the secant stiffness between adjacent skeleton candidate points, and use each skeleton candidate point as the point to be judged; when there are no less than 2 skeleton candidate points on both the front and back sides of the point to be judged, calculate the average value of the secant stiffness of the two segments on the front side and the average value of the secant stiffness of the two segments on the back side of the point to be judged; if the decrease ratio of the average secant stiffness of the back side relative to the average secant stiffness of the front side is greater than the preset stiffness change threshold, then the point to be judged is regarded as a yield transition candidate point; when there are multiple yield transition candidate points in the same loading direction, select the point to be judged with the largest decrease ratio as the yield transition feature point, and the preset stiffness change threshold can be taken as 20%.

[0027] For pinching feature points, after the hysteresis loop enters the reloading segment from the unloading segment, the interval where the bending moment sign changes or the interval where the absolute value of the bending moment is lower than the preset low bending moment threshold is first determined as the zero-return interval. In the reloading path after the zero-return interval, the difference between the absolute value of the bending moment at the sampling point and the absolute value of the frame envelope bending moment at the same rotation angle is calculated, and the ratio between the two is calculated. When the difference at three consecutive sampling points is greater than the first envelope deviation threshold, and the corresponding ratio is lower than the first pinching ratio threshold, the sampling point that first meets the condition is taken as the pinching start point; after the pinching start point, the sampling point with the lowest ratio is taken as the pinching valley point. In this embodiment, the first pinching ratio threshold is 0.60, and the first envelope deviation threshold is 10% of the absolute value of the peak bending moment of the hysteresis loop.

[0028] For degradation feature points and reloading path feature points, extraction is performed based on unloading secant stiffness, the degree of decrease in peak bending moment within the same loading level, and the degree of reloading path deviation, respectively. For the first hysteresis loop of each loading level, no degradation comparison within the same loading level is performed; unloading stiffness degradation and strength degradation judgment are performed starting from the second hysteresis loop within the same loading level. Within the same loading level, when the unloading secant stiffness of the current hysteresis loop decreases by more than 15% relative to the previous hysteresis loop, the corresponding zero bending moment corner point or the point with the lowest absolute value of bending moment is taken as the unloading stiffness degradation feature point; when the peak bending moment of the current hysteresis loop decreases by more than 10% relative to the peak bending moment of the previous hysteresis loop within the same loading level, the peak point of the current hysteresis loop is taken as the strength degradation feature point. Reloading path offset feature points are extracted after determining the target pointing line by the zero residual deformation point and the next peak point. When the vertical distance from the sampling point of the reloading segment to the target pointing line exceeds the preset reloading deviation threshold, the corresponding sampling point is taken as the reloading path offset feature point.

[0029] Based on the extracted hysteresis feature points, a skeleton feature point group, a pinching feature point group, and a cyclic evolution feature point group are formed. The skeleton feature point group includes skeleton peak feature points and yield inflection point feature points; the pinching feature point group includes pinching initiation point and pinching valley point; the cyclic evolution feature point group includes unloading stiffness degradation feature points, strength degradation feature points, and reloading path offset feature points.

[0030] This embodiment uses a hysteresis model with parameters including skeleton curve, pinching effect, stiffness degradation, strength degradation, and unloading rule as the model to be identified, specifically employing the Pinching4 hysteresis model for parameter identification. The model parameters are categorized according to their function into skeleton parameter groups, pinching parameter groups, stiffness degradation parameter groups, strength degradation parameter groups, and unloading rule parameter groups. Based on the peak bending moment, peak rotation angle, and residual deformation range in the experimental skeleton envelope curve, the initial value range for each parameter group is determined, and multiple candidate parameter groups are generated within this initial range. In this embodiment, the initial number of candidate parameter groups is 100. Subsequently, new candidate parameter groups are generated near the parameter values ​​corresponding to the top 20% of the candidate parameter groups in the comprehensive error ranking, with the neighborhood disturbance amplitude ranging from 5% to 15% of the corresponding parameter value range. After each candidate parameter group is input into the hysteresis model to be identified, the corresponding simulated hysteresis curve is obtained.

[0031] The simulated hysteresis curves are processed in the same way as the experimental hysteresis curves, sequentially performing hysteresis loop decomposition, skeleton envelope curve extraction, hysteresis feature point extraction, and single-loop energy consumption area calculation. For each experimental hysteresis loop, the experimental single-loop energy consumption area is calculated based on the area of ​​the closed path enclosed by the moment-to-turn curves; for the corresponding simulated hysteresis loop, the simulated single-loop energy consumption area is calculated using the same method. The relative error between the experimental and simulated single-loop energy consumption areas is used as an energy consumption constraint evaluation item; in this embodiment, the preset energy consumption error threshold is 15%, and if the relative error of any hysteresis loop exceeds 15%, the corresponding candidate parameter group is eliminated.

[0032] The candidate parameter groups evaluated through energy consumption constraints are further evaluated by feature point fitting. Hysteresis feature points in the experimental hysteresis curves are matched with hysteresis feature points in the simulated hysteresis curves according to loading direction, loading stage number, hysteresis loop order, and feature point type. The skeleton feature point group is used to calculate rotational and bending moment errors to evaluate the skeleton parameter group; the pinching feature point group is used to calculate rotational and bending moment errors to evaluate the pinching parameter group; and the cyclic evolution feature point group is used to calculate rotational, bending moment, and stiffness errors to evaluate the stiffness degradation parameter group, strength degradation parameter group, and unloading rule parameter group.

[0033] When a candidate parameter set simultaneously satisfies both the single-cycle energy consumption area error requirement and the matching error requirement for each feature point group, the candidate parameter set is output as the hysteresis model parameter for the engineering structure node. If no candidate parameter set simultaneously meets the requirements, a comprehensive error is calculated based on the single-cycle energy consumption area error, the skeleton feature point group matching error, the pinched feature point group matching error, and the cyclic evolution feature point group matching error. Candidate parameter sets are then selected in ascending order of comprehensive error, and new candidate parameter sets are generated within the corresponding parameter neighborhood. A maximum preset iteration count of 50 is set. If no candidate parameter set satisfies the error requirement after reaching the maximum preset iteration count, the candidate parameter set with the smallest comprehensive error is output as the hysteresis model parameter for the engineering structure node.

[0034] In one identification result of this embodiment, the simulated hysteresis curve corresponding to the output parameters is compared with the experimental hysteresis curve. The relative error of the peak bending moment of the skeleton is 3.8%, the relative error of the bending moment of the pinching valley point is 5.6%, the relative error of the unloading secant stiffness is 6.9%, the relative error of the peak bending moment of the strength degradation feature point is 4.7%, and the average relative error of the single-circle energy consumption area is 8.4%, all of which are lower than the corresponding preset error requirements.

[0035] Example 2 like Figure 1-3 As shown, the loading direction reversal point is determined as follows: Calculate the angle difference between two adjacent sampling points and use this angle difference as the angle increment between the corresponding sampling points; When the angle difference is lower than the sampling resolution, the corresponding angle increment is not used as the basis for judging the change in loading direction; The angle increment between adjacent sampling points with an angle difference not lower than the sampling resolution is taken as the effective angle increment; When the sign of the current effective rotation angle increment is opposite to that of the previous effective rotation angle increment, and the absolute value of the bending moment at the current sampling point is greater than the preset noise bending moment threshold, the current sampling point is determined as a candidate point for reversing the loading direction. If the candidate point for reversing the loading direction is a single sampling point, then that sampling point is determined as the loading direction reversal point; If the sign of the rotation increment changes repeatedly due to test pauses or data acquisition disturbances at multiple consecutive sampling points, then the sampling point where the absolute value of the rotation angle reaches a local extreme value is selected from these multiple consecutive sampling points as the candidate loading direction reversal point. When the absolute value difference of the rotation angle between multiple candidate loading direction reversal points is lower than the sampling resolution, the sampling point with the largest absolute value of bending moment is determined as the loading direction reversal point.

[0036] Hysteresis feature points are extracted as follows: On the skeleton envelope curve, the sampling point where the absolute value of the bending moment reaches a local maximum value is taken as the skeleton peak feature point; The yield inflection point is determined based on the change in secant stiffness between adjacent skeleton candidate points. After the hysteresis loop enters the reloading section from the unloading section, the difference between the absolute value of the bending moment at the sampling point in the reloading path and the absolute value of the bending moment of the skeleton envelope under the same rotation angle is calculated, and the pinching start point is determined based on the difference. After the pinching initiation point, calculate the ratio between the absolute value of the bending moment at the sampling point and the absolute value of the envelope bending moment of the skeleton under the same rotation angle, and determine the pinching valley point based on the ratio; Based on the decrease ratio of the unloading secant stiffness of the subsequent hysteresis loop relative to the previous hysteresis loop within the same loading direction and loading level, the unloading stiffness degradation characteristic point is determined. The strength degradation characteristic point is determined based on the percentage decrease in peak bending moment within the same loading level; The sampling point with the lowest absolute value of bending moment before and after the start of the reloading segment is taken as the zero-return residual deformation point, and the sampling point where the absolute value of bending moment reaches the local maximum value after the reloading segment is taken as the next peak point. The reload path offset feature point is determined based on the degree of deviation of the reload path from the target pointing line. The target pointing line is determined by the zero-return residual deformation point and the next peak point.

[0037] In this embodiment: Based on the discrete hysteresis data in Example 1, the process of hysteresis loop decomposition and hysteresis feature point extraction is described in detail. The discrete hysteresis data is arranged in the sampling order, and each row of data includes the sampling order number, loading stage number, rotation angle value, and bending moment value. To reduce the impact of acquisition disturbances on the determination of loading direction, the rotation angle increment between adjacent sampling points is first calculated; when the absolute value of the rotation angle difference between adjacent sampling points is less than the rotation angle sampling resolution, the corresponding rotation angle increment is not used as the basis for determining the change in loading direction; the rotation angle increment with an absolute value of the rotation angle difference not less than the rotation angle sampling resolution is taken as the effective rotation angle increment. In this example, the rotation angle sampling resolution is 0.00001 rad.

[0038] The sign of the effective angle increment is determined according to the sampling order. When the current effective angle increment has the opposite sign to the previous effective angle increment, and the absolute value of the bending moment at the current sampling point is greater than the preset noise bending moment threshold, the current sampling point is determined as a candidate point for loading direction reversal. The preset noise bending moment threshold can be taken as 0.5% of the maximum absolute value of the bending moment in the test data. If multiple candidate points for loading direction reversal appear within the same loading pause interval, the absolute values ​​of the angles corresponding to each candidate point are compared, and the sampling point with the largest absolute value of the angle is selected as the loading direction reversal point; if the difference in the absolute values ​​of the angles of multiple candidate points is lower than the angle sampling resolution, the sampling point with the largest absolute value of the bending moment is selected as the loading direction reversal point.

[0039] After determining the loading direction reversal point, the data segment between two adjacent loading direction reversal points is divided into a hysteresis loop. For each hysteresis loop, the loading direction, loading level number, starting sampling sequence number, ending sampling sequence number, and hysteresis loop rotation angle amplitude are recorded. The hysteresis loop rotation angle amplitude is the maximum absolute value of the rotation angle within that hysteresis loop. If the number of sampling points in a data segment is less than 5, or if the bending moment variation within that data segment is lower than a preset noise bending moment threshold, then that data segment is not considered a valid hysteresis loop for subsequent feature point extraction.

[0040] After completing the hysteresis loop partitioning, candidate skeleton points are extracted from both the positive and negative loading data. Under the same loading direction, the absolute values ​​of bending moments are compared point-by-point along the sampling sequence. When the absolute value of the bending moment at a sampling point is greater than the historical maximum absolute value of the bending moment at all previous sampling points in that loading direction, and the excess exceeds a preset skeleton update threshold, that sampling point is designated as a candidate skeleton point. The preset skeleton update threshold can be set to 0.3% of the maximum absolute value of the bending moment. Multiple candidate skeleton points are connected according to the sampling sequence to form a skeleton envelope curve. For cases where adjacent candidate skeleton points lack data at the same rotation angle, linear interpolation is used to obtain the skeleton envelope bending moment value at the same rotation angle.

[0041] Peak feature points of the skeleton are extracted according to the loading level. Within the same loading direction and the same loading level, sampling points where the absolute value of the bending moment reaches a local maximum and lies on the skeleton envelope curve are selected as peak feature points. If there are multiple local maximum points within the same loading level, the sampling point with the largest absolute value of the bending moment is selected; if the difference in the absolute values ​​of the bending moments of multiple sampling points is lower than a preset skeleton update threshold, the sampling point with the largest absolute value of the rotation angle is selected as the peak feature point of the skeleton for that loading level.

[0042] The yield transition characteristic point is determined based on the change in secant stiffness on the skeleton envelope curve. The secant stiffness is calculated for adjacent skeleton candidate points, and each candidate point is used as the point to be judged. When there are at least two skeleton candidate points both before and after the point to be judged, the average value of the two secant stiffness segments before and after the point to be judged is calculated. If the decrease in the average secant stiffness of the rear side relative to the average secant stiffness of the front side is greater than 20%, then the point to be judged is considered a yield transition candidate point. If there are multiple yield transition candidate points in the same loading direction, the point with the largest decrease in stiffness is selected as the yield transition characteristic point. If the point to be judged is located at the beginning or end of the skeleton envelope curve, and there are fewer than two skeleton candidate points before or after it, then yield transition judgment is not performed on that point.

[0043] The pinching start point and pinching valley point are extracted in the reloading path of each hysteresis loop. First, the interval near zero is determined; when the bending moment sign changes, the interval containing adjacent sampling points before and after the sign change is taken as the interval near zero; when the bending moment sign does not change but a low bending moment segment exists, the continuous sampling interval where the absolute value of the bending moment is less than 5% of the absolute value of the peak bending moment of the hysteresis loop is taken as the interval near zero. From the interval near zero, the difference between the absolute value of the bending moment at the current sampling point and the absolute value of the frame envelope bending moment at the same rotation angle is calculated point by point along the reloading direction, and the ratio of the absolute value of the bending moment at the current sampling point to the absolute value of the frame envelope bending moment at the same rotation angle is also calculated.

[0044] In the reloading path, when the difference at three consecutive sampling points is greater than 10% of the absolute value of the peak bending moment of the hysteresis loop, and the corresponding ratio is lower than 0.60, the sampling point that first meets the condition is taken as the pinching start point. After the pinching start point, the search continues along the reloading path until the ratio rises above 0.85. Within the sampling interval from the pinching start point to when the ratio rises to 0.85, the sampling point with the lowest ratio is taken as the pinching valley point. If multiple ratios are the same within this interval, or the difference between multiple ratios is lower than 0.01, the sampling point with the lowest absolute bending moment is selected as the pinching valley point.

[0045] The unloading stiffness degradation feature points are extracted based on the change in secant stiffness of the unloading segment. For each hysteresis loop, the sampling point where the absolute value of the bending moment reaches a local maximum before unloading begins is first taken as the peak point. If the bending moment sign changes during the unloading segment, linear interpolation is performed on the sampling points on both sides of the bending moment sign change interval to obtain the corner point corresponding to zero bending moment, and the unloading secant stiffness is calculated using the data segment between the peak point and the corner point corresponding to zero bending moment. If the bending moment sign does not change during the unloading segment, the sampling point with the lowest absolute value of bending moment and whose absolute difference in bending moment between its three adjacent sampling points is all less than 0.5% of the absolute value of the peak bending moment of the hysteresis loop is selected, and the unloading secant stiffness is calculated using the data segment between the peak point and this sampling point.

[0046] Within the same loading direction and loading level, unloading stiffness degradation is assessed starting from the second hysteresis loop. If the unloading secant stiffness of the current hysteresis loop decreases by more than 15% compared to the unloading secant stiffness of the previous hysteresis loop in the same loading level, the zero-moment turning point corresponding to the current hysteresis loop, or the point with the lowest absolute value of the moment when the unloading segment has not crossed zero, is taken as the characteristic point of unloading stiffness degradation. For data without a previous hysteresis loop in the same loading level, the previous hysteresis loop with a hysteresis loop turning angle amplitude difference of no more than 5% under the same loading direction and with the sampling order closest to the current hysteresis loop is selected as the comparison object.

[0047] Strength degradation characteristic points are determined based on the change in peak bending moment within the same loading level. Within the same loading direction and loading level, peak bending moments are compared starting from the second hysteresis loop. When the peak bending moment of the current hysteresis loop decreases by more than 10% relative to the peak bending moment of the previous hysteresis loop in the same loading level, the current hysteresis loop peak point is taken as the strength degradation characteristic point. If the test data for a certain loading level contains only one hysteresis loop, then the previous hysteresis loop with a hysteresis loop rotation amplitude difference of no more than 5% under the same loading direction and whose sampling order is closest to the current hysteresis loop is used as the comparison object.

[0048] The reloading path offset feature points are extracted during the reloading segment. First, the sampling points with the lowest absolute bending moment values ​​before and after the start of the reloading segment are taken as the zero-return residual deformation points. Then, the sampling points where the absolute bending moment reaches a local maximum value after the reloading segment are taken as the next peak points. The target pointing line is determined using the zero-return residual deformation points and the next peak points. To avoid the influence of differences in the dimensions of rotation angle and bending moment on distance calculation, the rotation angle amplitude and peak bending moment absolute values ​​are normalized before calculation. After normalization, the vertical distance from each sampling point in the reloading segment to the target pointing line is calculated. When the normalized vertical distance is greater than 0.05, and the secant stiffness change rate calculated using two sampling points before and after the sampling point is greater than 20%, the sampling point is taken as the reloading path offset feature point. If multiple sampling points meet the conditions, the sampling point with the largest vertical distance is selected.

[0049] After extracting feature points as described above, the skeleton peak feature points and yield inflection point feature points are grouped into the skeleton feature point group; the pinching start point and pinching valley point are grouped into the pinching feature point group; and the unloading stiffness degradation feature points, strength degradation feature points, and reloading path offset feature points are grouped into the cyclic evolution feature point group. Each feature point retains its sampling sequence number, loading direction, loading level number, hysteresis loop number, rotation angle value, and bending moment value for subsequent matching with similar feature points in the simulated hysteresis curve.

[0050] Example 3 like Figure 1-3 As shown, the correspondence between feature point groups and parameter groups is established as follows: The peak characteristic points and yield inflection point characteristic points of the skeleton are classified into the skeleton characteristic point group, and the skeleton parameter group is evaluated based on the rotational error and bending moment error between the experimental characteristic points and the simulated characteristic points in the skeleton characteristic point group. The pinching start point and pinching valley point are included in the pinching feature point group, and the pinching parameter group is evaluated based on the rotational error and bending moment error between the experimental feature points and the simulated feature points in the pinching feature point group. The unloading stiffness degradation feature points, strength degradation feature points, and reloading path offset feature points are assigned to the degradation tracking feature point group. The stiffness degradation parameter group, strength degradation parameter group, and unloading rule parameter group are evaluated based on the rotational error, bending moment error, and stiffness error between the experimental feature points and the simulated feature points in the degradation tracking feature point group.

[0051] Energy consumption constraint evaluation items are generated as follows: Calculate the energy consumption area per loop of the test based on the closed path area of ​​each test hysteresis loop; According to the loading direction, loading stage number, and hysteresis loop sequence, the experimental hysteresis loops are matched with the simulated hysteresis loops; Calculate the simulated single-loop energy consumption area based on the closed path area of ​​the corresponding simulated hysteresis loop; Energy consumption constraint evaluation items are generated based on the relative error between the energy consumption area of ​​a single test lap and the energy consumption area of ​​a single simulated lap. When the relative error of any hysteresis loop is greater than the preset energy consumption error threshold, the candidate parameter group corresponding to that relative error is removed. In this embodiment, based on the hysteresis feature points obtained in Embodiments 1 and 2, the evaluation relationship between the feature point group and the parameter group, as well as the energy consumption constraint evaluation process, are explained. Both the experimental and simulated hysteresis curves are decomposed into hysteresis loops, extracted into skeleton envelope curves, and extracted into hysteresis feature points according to the same rules, resulting in experimental and simulated feature point data tables, respectively. Each feature point data table records at least the feature point type, loading direction, loading level number, hysteresis loop number, sampling sequence number, rotation angle value, bending moment value, and corresponding secant stiffness value.

[0052] Before conducting error evaluation, the characteristic points on the experimental side and the characteristic points on the simulated side are matched accordingly.

[0053] During matching, feature points in the same direction are first selected according to the loading direction, then feature points within the same loading stage are determined according to the loading level number and hysteresis loop number, and finally, a one-to-one correspondence is made according to the feature point type.

[0054] If there are multiple feature points of the same type in a simulated hysteresis loop, the feature point whose rotation angle value is closest to that of the feature point on the test side is selected as the matching point; if the rotation angle difference is the same, the feature point whose absolute value of bending moment is closer to that of the feature point on the test side is selected as the matching point.

[0055] If no feature points of the corresponding type are extracted in a certain simulated hysteresis loop, the missing feature point error is added to the candidate parameter group. The missing feature point error is taken as the upper limit of the error of the feature point group of that type.

[0056] In this embodiment, the skeleton peak feature points and yield inflection point feature points are classified into the skeleton feature point group.

[0057] Skeleton feature point sets are used to evaluate skeleton parameter sets.

[0058] During the evaluation, the relative errors of rotation angle and bending moment between the characteristic points of the test side skeleton and the characteristic points of the simulated side skeleton are calculated respectively. The relative error of rotation angle is normalized based on the absolute value of the peak rotation angle of the corresponding loading level, and the relative error of bending moment is normalized based on the absolute value of the peak bending moment of the corresponding loading level.

[0059] The errors of skeleton feature points are calculated separately for positive and negative loading, and the average of the two is taken as the evaluation result of the skeleton parameter group.

[0060] The pinch initiation point and pinch valley point are included in the pinch feature point group. The pinch feature point group is used to evaluate the pinch parameter group.

[0061] During the evaluation, the relative errors of the rotation angle, bending moment, and pinching ratio at the pinching initiation point and pinching valley point are calculated separately. The pinching ratio is the ratio between the absolute value of the bending moment at the characteristic point and the absolute value of the bending moment envelope of the skeleton at the same rotation angle.

[0062] For the same hysteresis loop, the relative error weight of the moment at the pinching valley point is greater than the relative error weight of the moment at the pinching start point. In this embodiment, the relative error weight of the moment at the pinching valley point is 0.45, the relative error weight of the moment at the pinching start point is 0.25, the total weight of the relative error of the rotation angle of the two feature points is 0.20, and the weight of the pinching ratio error is 0.10.

[0063] The unloading stiffness degradation feature points, strength degradation feature points, and reloading path offset feature points are classified into the cyclic evolution feature point group.

[0064] This set of feature points is used to evaluate the stiffness degradation parameter set, strength degradation parameter set, and unloading rule parameter set.

[0065] When evaluating the characteristic points of unloading stiffness degradation, the relative error between the unloading secant stiffness on the test side and the unloading secant stiffness on the simulated side is calculated; when evaluating the characteristic points of strength degradation, the difference in the percentage decrease of peak bending moment within the same loading level is calculated; when evaluating the characteristic points of reloading path offset, the difference in the normalized reloading path offset distance is calculated.

[0066] The three types of errors mentioned above correspond to stiffness degradation, strength degradation, and changes in the reloading and unloading path, respectively.

[0067] To facilitate subsequent calculations, this embodiment normalizes all types of errors to the range of 0 to 1.

[0068] The rotation angle error is normalized according to the rotation angle amplitude of the corresponding hysteresis loop, the bending moment error is normalized according to the absolute value of the peak bending moment of the corresponding hysteresis loop, the stiffness error is normalized according to the secant stiffness of the first unloading cycle of the corresponding loading stage, and the energy dissipation error is normalized according to the energy dissipation area of ​​a single test cycle.

[0069] After normalization, the matching errors of skeleton feature point groups, pinched feature point groups, and cyclic evolution feature point groups are obtained respectively.

[0070] Table 1. Evaluation Relationship between Feature Point Sets and Parameter Sets Skeleton feature point group Skeleton peak feature points, yield inflection point Skeleton parameter group Angular error, bending moment error Pinch feature point group Pinching starting point, pinching valley point Pinch parameter group Angular error, bending moment error, pinching ratio error Cyclic Evolutionary Feature Point Group Unloading stiffness degradation feature points, strength degradation feature points, and reloading path offset feature points Stiffness degradation parameter group, strength degradation parameter group, unloading rule parameter group Stiffness error, peak moment reduction ratio error, reloading path deviation error Energy Constraint Evaluation Items Area of ​​the closed path of each hysteresis loop Candidate parameter set Relative error of energy consumption area per circle The energy consumption constraint evaluation item is generated according to the area of ​​the closed path of the hysteresis loop. For the test hysteresis loop, the area of ​​the bending moment to turning angle curve segment between adjacent sampling points is accumulated according to the sampling order to obtain the energy consumption area of ​​the test single loop. For the simulated hysteresis loop, the same hysteresis loop number and loading direction are used, and the same area calculation method is used to obtain the simulated single loop energy consumption area.

[0071] The relative error between the two is taken as the single-cycle energy consumption area error.

[0072] In this embodiment, the preset energy consumption error threshold is set to 15%.

[0073] During the evaluation of candidate parameter groups, energy consumption constraint evaluation is performed first. If the single-turn energy consumption area error between any simulated hysteresis loop and the experimental hysteresis loop corresponding to a candidate parameter group exceeds 15%, the candidate parameter group is eliminated, and the matching errors of the skeleton feature point group, pinching feature point group, and cyclic evolution feature point group are no longer calculated.

[0074] If the single-turn energy consumption area error of each hysteresis loop does not exceed 15%, then the candidate parameter group will be included in the feature point group matching error evaluation.

[0075] For the candidate parameter groups that pass the energy consumption constraint evaluation, the matching error of the three types of feature point groups is calculated respectively.

[0076] In this embodiment, the matching error threshold for the skeleton feature point group is set to 8%, the matching error threshold for the pinched feature point group is set to 10%, and the matching error threshold for the cyclic evolution feature point group is set to 12%.

[0077] If a candidate parameter group meets all three threshold requirements mentioned above, then the candidate parameter group is marked as a candidate parameter group that meets the feature point group fitting requirements.

[0078] If the candidate parameter group does not simultaneously meet the above three threshold requirements, then the parameter group that needs to be adjusted first is determined based on the feature point group that does not meet the threshold.

[0079] In one evaluation process of this embodiment, the maximum relative error of the single-cycle energy consumption area of ​​a certain candidate parameter group was 11.7%, which passed the energy consumption constraint evaluation; its skeleton feature point group matching error was 5.2%, the pinched feature point group matching error was 7.9%, and the cyclic evolution feature point group matching error was 9.6%, all of which were lower than the corresponding thresholds. This candidate parameter group proceeded to the subsequent output judgment.

[0080] The matching error of the skeleton feature point group of another candidate parameter group was 4.8%, but the maximum relative error of the single-cycle energy consumption area was 18.3%, which exceeded the preset energy consumption error threshold. Therefore, this candidate parameter group was eliminated.

[0081] Through the above evaluation process, the candidate parameter group must first pass the single-cycle energy consumption area error screening before entering the output judgment.

[0082] Then, the grouping error evaluation is performed on the feature points corresponding to the skeleton, pinching, degradation, and reloading paths.

[0083] The final retained candidate parameter group has corresponding energy consumption error records, feature point group error records, and parameter group evaluation records, which are used for comprehensive error sorting and parameter neighborhood regeneration in Example 4.

[0084] Example 4 like Figure 1-3 As shown, the candidate parameter set is evaluated and output in the following manner: According to the loading direction, loading level number, hysteresis loop sequence and feature point type, the hysteresis feature points in the test hysteresis curve are matched with the hysteresis feature points in the simulated hysteresis curve. Calculate the matching error of the skeleton feature point group, the pinched feature point group, and the degradation tracking feature point group respectively; When the candidate parameter set meets the preset energy consumption error requirement and the matching error requirement of each feature point group, the candidate parameter set is output as the hysteresis model parameter of the engineering structure node. When there is no candidate parameter group that simultaneously meets the energy consumption error requirement and the matching error requirement of each feature point group, the comprehensive error is calculated based on the summation value of the single-cycle energy consumption area error, the matching error of the skeleton feature point group, the matching error of the pinching feature point group, and the matching error of the degradation tracking feature point group. Candidate parameter groups are selected in ascending order of comprehensive error, and the range of parameter values ​​is narrowed with the selected candidate parameter groups as the center. Candidate parameter groups are then regenerated and iterated. If, after reaching the preset maximum number of iterations, there is still no candidate parameter group that simultaneously satisfies the energy consumption error requirement and the matching error requirement of each feature point group, then the candidate parameter group with the smallest comprehensive error is output as the parameter of the hysteresis model of the engineering structure node. Among them, the comprehensive error is the sum of the single-loop energy consumption area error, the skeleton feature point group matching error, the pinched feature point group matching error, and the degradation tracking feature point group matching error after normalization and weighted summation according to preset weights. The single-loop energy consumption area error is obtained by averaging, taking the maximum value, or weighted summation of the relative errors of the single-loop energy consumption area of ​​each hysteresis loop. This embodiment, based on embodiments 1 to 3, describes the process of candidate parameter group generation, candidate parameter group screening, parameter value range narrowing, and identification result verification. This embodiment uses quasi-static cyclic loading test data of the same set of engineering structural connection nodes as benchmark data. The hysteresis model to be identified is a hysteresis model with skeleton parameters, pinching parameters, stiffness degradation parameters, strength degradation parameters, and unloading rule parameters; specifically, the Pinching4 hysteresis model is used. The maximum number of iterations is 50, and the number of candidate parameter groups in each round is 100.

[0085] Before generating candidate parameter groups, the initial value ranges of each parameter group are determined based on the skeleton envelope curve, pinching valley point, unloading stiffness degradation characteristic point, strength degradation characteristic point, and reloading path offset characteristic point extracted from the experimental hysteresis curve. The value range of the skeleton parameter group is determined based on the positive peak bending moment, negative peak bending moment, positive peak rotation angle, and negative peak rotation angle in the experimental skeleton envelope curve; the value range of the pinching parameter group is determined based on the bending moment ratio and rotation position at the pinching valley point; the value range of the stiffness degradation parameter group is determined based on the percentage decrease in unloading secant stiffness within the same loading level; the value range of the strength degradation parameter group is determined based on the percentage decrease in peak bending moment within the same loading level; and the value range of the unloading rule parameter group is determined based on the zero residual deformation point, the rotation point corresponding to zero bending moment, and the reloading path offset characteristic point.

[0086] Table 2. Parameter sets and initial value ranges for the hysteresis model Skeleton parameter group Peak bending moment, peak rotation angle, and post-yield stiffness of the experimental skeleton envelope curve The peak bending moment parameter is taken as 0.80 to 1.20 times the experimental peak bending moment, the peak rotation angle parameter is taken as 0.85 to 1.15 times the experimental peak rotation angle, and the post-yield stiffness ratio is taken as 0.05 to 0.35. Pinch parameter group The ratio of the moment at the pinch-down valley, the position of the pinch-down valley angle, and the interval near zero return. The relative pinching displacement coefficient ranges from 0.20 to 0.80, the relative pinching force coefficient ranges from 0.10 to 0.70, and the residual force coefficient ranges from 0.00 to 0.30. Stiffness degradation parameter set Percentage decrease in unloading secant stiffness within the same loading level The stiffness degradation factor ranges from 0.00 to 0.60, and the stiffness degradation growth factor ranges from 0.00 to 0.40. Strength degradation parameter group Peak bending moment reduction ratio within the same loading level The strength degradation coefficient ranges from 0.00 to 0.50, and the strength degradation growth coefficient ranges from 0.00 to 0.40. Uninstallation rule parameter group Zero residual deformation point, zero bending moment corresponding corner point, and reloading path offset feature point The uninstallation path correction factor is set to 0.10 to 0.90, and the reload path correction factor is set to 0.10 to 0.90. The initial candidate parameter sets are generated within the value range listed in Table 2. Each parameter is selected using a stratified random method within its corresponding value range, ensuring that each parameter set has candidate parameters in the low, medium, and high value segments. In this embodiment, candidate parameters are generated for the skeleton parameter set, pinching parameter set, stiffness degradation parameter set, strength degradation parameter set, and unloading rule parameter set, and then combined to form 100 initial candidate parameter sets. Boundary processing is performed on parameter values ​​that exceed the model's allowable range, duplicate candidate parameter sets are removed, and new candidate parameter sets are generated to maintain 100 candidate parameter sets in each round.

[0087] After each set of candidate parameters is input into the hysteresis model to be identified, the corresponding simulated hysteresis curve is obtained. The simulated hysteresis curve is decomposed into hysteresis loops, extracted into skeleton envelope curves, extracted into hysteresis feature points, and the energy consumption area per loop is calculated according to the methods of Example 1 and Example 2.

[0088] Hysteresis feature points in experimental and simulated hysteresis curves are matched according to loading direction, loading stage number, hysteresis loop order, and feature point type.

[0089] Candidate parameter groups are first screened for consistency in single-cycle energy consumption. For each hysteresis loop, the experimental single-cycle energy consumption area and the simulated single-cycle energy consumption area are calculated separately. If the relative error of the single-cycle energy consumption area of ​​any hysteresis loop is greater than 15%, the corresponding candidate parameter group is eliminated.

[0090] The candidate parameter groups selected through single-cycle energy consumption consistency screening are then evaluated for feature point group matching error.

[0091] The matching error of the skeleton feature point group is calculated based on the angular error and bending moment error of the skeleton peak feature point and yield inflection point; the matching error of the pinching feature point group is calculated based on the angular error, bending moment error, and pinching ratio error of the pinching start point and pinching valley point; the matching error of the cyclic evolution feature point group is calculated based on the unloading secant stiffness error, strength degradation peak bending moment error, and reloading path deviation distance error. The single-cycle energy consumption area error, skeleton feature point group matching error, pinching feature point group matching error, and cyclic evolution feature point group matching error are all normalized and weighted according to preset weights to obtain the comprehensive error. In this embodiment, the weight of each of the above four types of errors is 25%.

[0092] After completing one round of error evaluation, the candidate parameter groups that have passed the single-cycle energy consumption consistency screening will be sorted from smallest to largest according to the comprehensive error.

[0093] If there exists a candidate parameter set that simultaneously satisfies the following conditions: the relative error of the single-cycle energy consumption area is no greater than 15%, the matching error of the skeleton feature point group is no greater than 8%, the matching error of the pinching feature point group is no greater than 10%, and the matching error of the cyclic evolution feature point group is no greater than 12%, then the candidate parameter set is output as the hysteresis model parameter of the engineering structure node.

[0094] If there is no candidate parameter group that simultaneously meets the above error requirements in the current round, the candidate parameter group that ranks in the top 20% of the comprehensive error is selected as the central parameter group, and the parameter value range is narrowed based on the central parameter group to regenerate the candidate parameter group for the next round.

[0095] In this embodiment, the parameter value range for the next round is formed by fluctuating by 5% around the current value of the skeleton parameter in the central parameter group, by 10% around the current value of the pinching parameter, and by 15% around the current values ​​of the stiffness degradation parameter, strength degradation parameter, and unloading rule parameter.

[0096] In each round, the five candidate parameter groups with the smallest comprehensive error are retained and directly enter the next round of evaluation. The remaining candidate parameter groups are generated by supplementing the candidate parameter groups with the highest comprehensive error ranking, so that the total number of candidate parameter groups in the next round remains at 100.

[0097] If the newly generated parameter value exceeds the initial value range listed in Table 2, it will be adjusted to the corresponding boundary value.

[0098] If no candidate parameter set that simultaneously meets the error requirements is found after 50 iterations, the candidate parameter set with the smallest comprehensive error is output as the parameter set for the hysteresis model of the engineering structure node.

[0099] To verify the parameter identification results, this embodiment sets up two comparative examples. Comparative example 1 uses manually selected peak points of the skeleton and intersections of coordinate axes as control points, and uses the root mean square value of the bending moment error of all control points as the evaluation index.

[0100] Comparative Example 2 uses the same feature point group matching error evaluation method as this embodiment, but does not perform single-cycle energy consumption consistency screening.

[0101] The methods in the examples, Comparative Example 1 and Comparative Example 2 all used the same set of experimental discrete hysteresis data, the same hysteresis model to be identified, the same initial range of parameter values, the same number of initial candidate parameter groups, and the same maximum number of iterations.

[0102] Table 3. Comparison of Fitting Results for Different Parameter Identification Methods Comparative Example 1 8.9 14.8 16.5 12.7 19.6 Not achieved Comparative Example 2 5.1 8.3 9.8 7.6 17.2 Not achieved Example Method 3.8 5.6 6.9 4.7 8.4 31 As shown in Table 3, Comparative Example 1, due to the use of a small number of manually selected control points for fitting, has relatively high relative errors in the bending moment at the pinched valley point, the stiffness of the unloading secant line, and the average relative error of the energy dissipation area per cycle.

[0103] After using the feature point group fitting method, the relative errors of the peak bending moment, the pinching valley bending moment, and the unloading secant stiffness of Comparative Example 2 are lower than those of Comparative Example 1. However, it did not perform single-cycle energy consumption consistency screening, and the average relative error of the single-cycle energy consumption area was 17.2%, which exceeded the preset energy consumption error threshold of 15%.

[0104] The method in the embodiment outputs a set of candidate parameters that meet the error requirements in the 31st iteration. The average relative error of the single-cycle energy consumption area is 8.4%, and the corresponding verifications are completed for the skeleton peak, pinching valley point, unloading stiffness degradation position and strength degradation position.

[0105] The simulated hysteresis curves corresponding to the model parameters output in this embodiment complete the error verification of skeleton morphology, pinching position, degradation trend and single-cycle energy consumption characteristics.

[0106] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0107] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the present invention. Various changes and modifications can be made to the present invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed.

Claims

1. A method for identifying hysteresis model parameters based on hysteresis feature point mapping, characterized in that, Includes the following steps: The discrete hysteresis data and corresponding test process identifiers obtained from the quasi-static test of the engineering structure nodes are acquired. The discrete hysteresis data includes generalized displacement and generalized force used to form the hysteresis curve, and the test process identifiers include sampling sequence identifiers and loading level identifiers. The loading direction reversal point is determined based on the generalized displacement increment between adjacent sampling points, and the data segment between two adjacent loading direction reversal points is used as the loading branch segment. According to the loading level identifier, the order of loading direction change, and the sampling order between adjacent loading branch segments, multiple loading branch segments belonging to the same cycle process are combined into a hysteresis loop, so as to decompose the experimental hysteresis curve into multiple hysteresis loops. A skeleton envelope curve is formed based on data points where the absolute value of the generalized force exceeds the historical maximum absolute value of the generalized force under the same loading direction, and hysteresis feature points are extracted from the hysteresis loop and the skeleton envelope curve. A set of feature points is generated based on the hysteresis feature points, and an energy consumption constraint evaluation term is generated based on the area of ​​the closed path formed by each hysteresis loop in the generalized force to generalized displacement plane. The feature point group includes the skeleton feature point group, the pinching feature point group, and the cyclic evolution feature point group; Establish evaluation relationships between the feature point group and the energy consumption constraint evaluation item and the parameter group in the hysteresis model to be identified, respectively. Candidate parameter groups are generated based on the preset value range of each parameter group in the hysteresis model to be identified. The candidate parameter groups are input into the hysteresis model to be identified to obtain the simulated hysteresis curve. The hysteresis feature points and single-turn energy consumption area of ​​the simulated hysteresis curve are extracted in the same way as the experimental hysteresis curve. First, candidate parameter groups are selected based on the single-turn energy consumption area error between the experimental hysteresis loop and the simulated hysteresis loop. Then, the selected candidate parameter groups are evaluated based on the matching error of the feature point group, and the candidate parameter groups that meet the preset error requirements are output as the parameters of the engineering structure node hysteresis model.

2. The hysteresis model parameter identification method based on hysteresis feature point mapping according to claim 1, characterized in that, The loading direction reversal point is determined as follows: The effective increment is determined based on the generalized displacement difference between adjacent sampling points; When the sign of the current effective increment is opposite to that of the previous effective increment, and the absolute value of the generalized force of the current sampling point is greater than the preset noise threshold, the current sampling point is determined as a candidate point for reversing the loading direction. When multiple candidate points for reversing the loading direction occur due to test pauses or data acquisition disturbances, the sampling point where the absolute value of the generalized displacement reaches a local extreme value is selected as the loading direction reversal point. When the difference in the absolute value of the generalized displacement between multiple candidate sampling points is lower than the sampling resolution, the sampling point with the largest absolute value of the generalized force is selected as the loading direction reversal point.

3. The hysteresis model parameter identification method based on hysteresis feature point mapping according to claim 1, characterized in that, The hysteresis feature points include skeleton peak feature points, yield inflection point feature points, pinching initiation point, pinching valley point, unloading stiffness degradation feature points, strength degradation feature points, and reloading path offset feature points. Among them, the skeleton peak feature point and yield inflection feature point are determined by the local peak and secant stiffness variation in the skeleton envelope curve; The pinching initiation point and pinching valley point are determined by the degree of deviation of the reloading path from the generalized force of the skeleton envelope under the same generalized displacement. The unloading stiffness degradation characteristic point and the strength degradation characteristic point are determined by the unloading secant stiffness reduction ratio and the peak generalized force reduction ratio of adjacent hysteresis loops within the same loading level; The reload path offset feature point is determined by the degree of deviation of the reload path from the target pointing line, which is determined by the zero-return residual deformation point and the next peak point.

4. The hysteresis model parameter identification method based on hysteresis feature point mapping according to claim 1, characterized in that, The evaluation relationship between the feature point set and the parameter set is as follows: Skeleton feature point sets are used to evaluate skeleton parameter sets; The pinch feature point set is used to evaluate the pinch parameter set; The cyclic evolution feature point set is used to evaluate the stiffness degradation parameter set, strength degradation parameter set, and unloading rule parameter set; The energy consumption constraint evaluation item is used to evaluate the consistency of single-cycle energy consumption of candidate parameter groups.

5. The hysteresis model parameter identification method based on hysteresis feature point mapping according to claim 1, characterized in that, The energy consumption constraint evaluation item is generated in the following manner: Calculate the energy consumption area of ​​a single loop in the test based on the closed path area of ​​the hysteresis loop. Calculate the simulated single-loop energy consumption area based on the closed path area of ​​the corresponding simulated hysteresis loop; Energy consumption constraint evaluation items are generated based on the relative error between the energy consumption area of ​​a single test lap and the energy consumption area of ​​a single simulated lap. When the relative error of any hysteresis loop is greater than the preset energy consumption error threshold, the corresponding candidate parameter group is removed.

6. The hysteresis model parameter identification method based on hysteresis feature point mapping according to claim 1, characterized in that, The candidate parameter set is evaluated and output in the following manner: According to the loading direction, loading level identifier, hysteresis loop sequence and feature point type, the hysteresis feature points in the test hysteresis curve are matched with the hysteresis feature points in the simulated hysteresis curve. Calculate the matching errors for the skeleton feature point group, the pinched feature point group, and the cyclic evolution feature point group respectively; The comprehensive error is calculated based on the single-cycle energy consumption area error, the skeleton feature point group matching error, the pinching feature point group matching error, and the cyclic evolution feature point group matching error. Candidate parameter groups are selected in ascending order of comprehensive error, and the range of parameter values ​​is narrowed with the selected candidate parameter groups as the center. Candidate parameter groups are then regenerated and iterated. When a candidate parameter set meets the preset error requirement, the candidate parameter set is output as the hysteresis model parameter of the engineering structure node; when no candidate parameter set meets the preset error requirement is found after reaching the preset maximum number of iterations, the candidate parameter set with the smallest comprehensive error is output as the hysteresis model parameter of the engineering structure node.

Images