Method and system for predicting hysteretic performance of coupling beam damper

Abstract

The present application relates to the technical field of performance monitoring, in particular to a coupled beam damper hysteresis performance prediction method and system, comprising the following steps: obtaining monitoring data to generate a prediction input sequence, discretizing and fitting an equivalent stiffness prediction primitive based on a multi-cycle extreme value range to generate a path, using cross-cycle comparison and residual displacement evolution calculation to generate stable prediction data, screening stable cycle sections and combining evolution direction to establish a performance prediction constraint set, combining the to-be-predicted path and the constraint set to calculate force increment and correct the state to generate hysteresis performance prediction results. In the present application, the cycle recognition is used to cut the displacement and force time history data, the prediction input sequence is generated, the displacement extreme value proportion is discretized, the interval force displacement ratio is calculated, the cross-cycle comparison and residual displacement smoothing judgment are used to realize the stable prediction of hysteresis performance, improve the performance evaluation accuracy and reliability, and maintain the consistency of the prediction results.

Description

Technical Field

[0001] This invention relates to the field of performance monitoring technology, and in particular to a method and system for predicting the hysteresis performance of a connecting beam damper. Background Technology

[0002] The field of performance monitoring technology refers to a technical system that continuously acquires and analyzes the mechanical response and working status of engineering structures and key components during service. Its core aspects include the acquisition of load data, displacement data, stress-strain data, and dynamic response data; the time-series processing of sensor signals; the calculation and comparison of key performance indicators; and the evaluation and judgment of the working characteristics of structures or components based on historical and real-time monitoring data. This technical field is typically applied to building structures, bridge structures, and vibration damping components to track and record the performance changes of components.

[0003] Among them, the traditional method and system for predicting the hysteresis performance of a coupling beam damper refers to the technical solution for predicting and characterizing the force-displacement hysteresis relationship formed by the coupling beam damper under cyclic loading conditions. It usually involves arranging displacement sensors and force sensors at both ends of the coupling beam damper to obtain displacement and damping force data during the cyclic loading process, recording the loading cycle in segments, calculating the slope and yield point of multiple loading stages according to the preset skeleton curve fitting rules, and combining the hysteresis loop area and residual deformation data extracted from historical loading cycles to extrapolate the hysteresis curve parameters under subsequent loading conditions, thereby forming a prediction and characterization system for the hysteresis performance of the coupling beam damper.

[0004] Current technologies for predicting the hysteretic performance of coupling beam dampers often rely solely on pre-defined skeleton curve fitting rules. This approach, when faced with complex loading conditions and varying operating environments, easily overlooks extreme displacement values ​​and force differences during cyclic loading, leading to significant errors in the predicted hysteresis curve and an inability to promptly adjust the accuracy of hysteretic performance predictions. Furthermore, existing methods fail to effectively capture the changing trends of each loading cycle when dealing with actual operating conditions, resulting in potentially unstable hysteretic performance assessments that cannot fully reflect the damper's true performance under different loads, thus affecting the accuracy of structural health monitoring and assessment. Summary of the Invention

[0005] To address the technical problems existing in the prior art, embodiments of the present invention provide a method for predicting the hysteresis performance of a coupling beam damper, comprising the following steps: S1: Obtain displacement time history and force time history by monitoring the loading process of the coupling beam damper, determine the direction reversal position of the displacement time history, and perform cyclic identification and segmentation of continuous monitoring data to generate a predictive input sequence; S2: Based on the predicted input sequence, obtain the extreme value range of multi-cycle displacement, proportionally discretize the cyclic displacement propulsion path into multiple displacement intervals, calculate and fit the force difference to displacement difference ratio of adjacent displacement intervals, and generate equivalent stiffness prediction primitives. S3: Based on the equivalent stiffness prediction primitive, cross-cycle comparison and stiffness difference judgment are performed on the same displacement interval, the predicted residual displacement evolution sequence is calculated and exponential smoothing and direction consistency judgment are performed to generate stable prediction data. S4: Based on the stability prediction data, filter the stable cyclic segment, calculate the interval equivalent stiffness parameter corresponding to the equivalent stiffness prediction element, and combine the residual displacement evolution sequence to determine the evolution direction, and generate a performance prediction constraint set; S5: Obtain the predicted cyclic displacement propagation path and locate the corresponding displacement interval. Combine the performance prediction constraint set to calculate the predicted force increment and analyze the force displacement path. Combine the residual displacement evolution direction correction state to generate hysteresis performance prediction results.

[0006] As a further embodiment of the present invention, the predicted input sequence includes a displacement cycle segment index, a force-displacement corresponding frame, and a cycle direction marker; the equivalent stiffness prediction primitive includes a displacement interval number set, interval stiffness fitting, and force-displacement difference ratio; the stable prediction data includes a residual displacement evolution sequence, a smoothing weight coefficient, and a direction consistency identifier; the performance prediction constraint set includes an interval equivalent stiffness parameter table, an evolution direction criterion, and a stable cycle segment index; and the hysteresis performance prediction result includes a predicted force-displacement trajectory, a state correction marker, and a cycle termination feature point.

[0007] As a further aspect of the present invention, the specific steps of S1 are as follows: S101: Obtain the displacement time history and force time history obtained by synchronous monitoring during the loading process of the coupling beam damper, judge the change of the sign of the displacement increment of adjacent sampling points in the displacement time history, record the sampling index position corresponding to the displacement increment turning from positive to negative or from negative to positive, mark the corresponding displacement value, and generate a set of direction reversal inflection point indexes. S102: Based on the direction reversal inflection point index set, call the corresponding sampling sequence of displacement time history, sequentially search the data segments between adjacent inflection point indices, determine the loop start index and loop end index according to the continuity of displacement change before and after the inflection point index, and obtain the loop boundary index sequence. S103: Based on the cyclic boundary index sequence, call the data frames corresponding to the boundary indices in the displacement time history and force time history, perform time-series alignment and length alignment on the displacement vectors and force vectors in the multi-cyclic intervals, and sequentially concatenate the rearranged vectors to generate a prediction input sequence.

[0008] As a further aspect of the present invention, the specific steps of S2 are as follows: S201: Based on the predicted input sequence, retrieve multi-cycle segment displacement time history data frames, perform extreme value judgment on the displacement sample value in each cycle, determine the maximum and minimum displacement values ​​based on the sign change of the displacement difference between adjacent sampling points, and aggregate the multi-cycle extreme value pairs to generate a multi-cycle displacement extreme value range. S202: Based on the multi-cycle displacement extreme value range, call the single-cycle displacement advancement path sequence, construct a proportional coefficient according to the upper and lower limits of the extreme value range, perform proportional discretization on the displacement sample values ​​within the path, use them as the sequentially arranged displacement interval boundaries, and index and encode the interval boundaries to establish a displacement interval sequence. S203: For the displacement interval sequence, call the stress time history and displacement time history data frames in the prediction input sequence, calculate the ratio of the force difference and displacement difference at the boundary points of adjacent displacement intervals, and perform parameter fitting operation on the multi-interval ratios in the order of interval index to generate equivalent stiffness prediction primitives.

[0009] As a further aspect of the present invention, the specific steps of S3 are as follows: S301: Based on the equivalent stiffness prediction primitive corresponding to the continuous cycle, retrieve the stiffness scalar sequence under the same displacement interval index of multiple cycles, calculate the difference between the stiffness scalars of adjacent cycles and compare it with the preset stiffness difference judgment threshold, determine the sign state of the difference and mark it, and generate a cross-cycle stiffness difference sequence. S302: Based on the cross-cycle stiffness difference sequence, collect displacement time history data frames corresponding to the end time of multiple cycles, index and align the displacement values ​​at the end of multiple cycles and converge the values, and map the displacement values ​​sequentially according to the stiffness difference judgment sequence index and analyze the displacement vector to obtain the predicted residual displacement evolution sequence. S303: For the predicted residual displacement evolution sequence, a preset exponential weight coefficient sequence and displacement vector are called to perform recursive smoothing calculation, and the consistency of the sign direction of adjacent sequence items is judged. The sequence items that pass the judgment are indexed, updated and aggregated to generate stable prediction data.

[0010] As a further aspect of the present invention, the stiffness difference determination threshold is determined by acquiring the initial loading cycle monitoring data of the connecting beam damper in the elastic linear deformation stage, calculating the difference sequence of equivalent stiffness between adjacent cycles in the initial loading cycle, calculating the stiffness dispersion by standard deviation of the difference sequence, and multiplying the stiffness dispersion with a preset confidence interval factor.

[0011] As a further aspect of the present invention, the specific steps of S4 are as follows: S401: Based on the stable prediction data, filter the continuous cyclic segments of stable prediction states, determine the consistency of the stable prediction data sequence, index and concatenate the cyclic numbers that meet the preset stable prediction state determination threshold, mark the segment boundaries, and perform segment length verification to generate a stable cyclic segment index set. S402: Based on the stable cyclic segment index set, call the equivalent stiffness prediction primitive under the corresponding cyclic number, perform numerical aggregation of stiffness scalars in multiple segments according to segment index, calculate the mean within the segment, arrange the mean results of multiple segments according to index order, and obtain the interval equivalent stiffness parameter. S403: For the interval equivalent stiffness parameters, call the predicted residual displacement evolution sequence, perform sign and direction discrimination on the displacement sequence corresponding to the multi-cycle segment, and perform joint mapping with the segment equivalent stiffness parameters. Construct an index set for parameter entries that meet the direction consistency condition, and generate a performance prediction constraint set.

[0012] As a further aspect of the present invention, the stability prediction state determination threshold is determined by obtaining the residual displacement evolution sequence of the connecting beam damper, performing differential operations on the displacement values ​​of consecutive adjacent cycles in the sequence to obtain the displacement fluctuation sequence, calculating the standard deviation statistical value of the displacement fluctuation sequence, and weighting the standard deviation statistical value with a preset tolerance coefficient.

[0013] As a further aspect of the present invention, the specific steps of S5 are as follows: S501: Obtain the displacement propagation path of the cycle to be predicted, arrange multiple discrete displacement points in the displacement propagation path, determine the continuity of adjacent displacement points according to the numerical increment relationship, complete the path segmentation identification, map the displacement point sequence to the corresponding displacement value range, and generate the predicted cycle displacement interval sequence. S502: Based on the predicted cyclic displacement interval sequence, call the equivalent stiffness parameter of the corresponding interval in the performance prediction constraint set, multiply the adjacent displacement increments and equivalent stiffness parameters in the multiple displacement intervals to calculate the force increment sequence and perform cumulative calculation to generate predicted cyclic force displacement path data. S503: For the predicted cyclic force-displacement path data, obtain the displacement state at the predicted cycle termination position, call the residual displacement evolution direction in the performance prediction constraint set, perform direction consistency judgment on the force-displacement state at the termination position, correct the sign of inconsistent terms, and generate hysteresis performance prediction results.

[0014] The hysteresis performance prediction system for coupling beam dampers includes: The data processing module obtains displacement time history and force time history by monitoring the loading process of the coupling beam damper, determines the direction reversal position of the displacement time history, and performs cyclic identification and segmentation of continuous monitoring data to generate a predictive input sequence and transmit it to the stiffness modeling module. The stiffness modeling module obtains the extreme value range of multi-cycle displacement based on the predicted input sequence, proportionally discretizes the cyclic displacement propagation path into multiple displacement intervals, calculates and fits the ratio of force difference to displacement difference between adjacent displacement intervals, generates equivalent stiffness prediction primitives, and transmits them to the evolution prediction module. The evolution prediction module, based on the equivalent stiffness prediction primitive, performs cross-cycle comparison and stiffness difference judgment on the same displacement interval, calculates the predicted residual displacement evolution sequence and performs exponential smoothing and direction consistency judgment, generates stable prediction data and transmits it to the constraint construction module. The constraint construction module filters stable cyclic segments based on the stability prediction data, calculates the interval equivalent stiffness parameters corresponding to the equivalent stiffness prediction primitives, and determines the evolution direction by combining the residual displacement evolution sequence, generating a performance prediction constraint set and passing it to the performance prediction module. The performance prediction module obtains the predicted cyclic displacement propagation path and locates the corresponding displacement interval. It calculates the predicted force increment and analyzes the force-displacement path in combination with the performance prediction constraint set. It also generates the hysteresis performance prediction result by combining the residual displacement evolution direction correction state.

[0015] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, by cyclically identifying and segmenting the displacement and force time history data during the loading process of a coupling beam damper, a predictive input sequence is generated in real time. Combined with proportional discretization of the displacement extreme range, the ratio of force difference to displacement difference between adjacent displacement intervals is accurately calculated, effectively revealing the variation law of force and displacement in different loading cycles. Based on this, through cross-cycle comparative analysis and exponential smoothing and directional consistency judgment of the residual displacement evolution sequence, stable prediction of hysteretic performance can be achieved, and stable loading cycle segments can be screened, ensuring accurate prediction of hysteretic performance. This solves the problems of unstable hysteretic performance prediction and large errors in existing technologies, improving the accuracy and reliability of performance evaluation. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying 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.

[0017] Figure 1 This is a schematic diagram of the steps of the present invention; Figure 2 This is a detailed schematic diagram of S1 of the present invention; Figure 3 This is a detailed schematic diagram of S2 of the present invention; Figure 4 This is a detailed schematic diagram of S3 of the present invention; Figure 5 This is a detailed schematic diagram of S4 of the present invention; Figure 6 This is a detailed schematic diagram of S5 of the present invention; Figure 7 This is a system module diagram of the present invention. Detailed Implementation

[0018] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0019] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0020] Please see Figure 1 This invention provides a method for predicting the hysteresis performance of a coupling beam damper, comprising the following steps: S1: Obtain displacement time history and force time history by monitoring the loading process of the coupling beam damper, determine the direction reversal position of the displacement time history, and perform cyclic identification and segmentation of continuous monitoring data to generate a predictive input sequence; S2: Based on the predicted input sequence, obtain the extreme value range of multi-cycle displacement, proportionally discretize the cyclic displacement advancement path into multiple displacement intervals, calculate and fit the force difference to displacement difference ratio of adjacent displacement intervals, and generate equivalent stiffness prediction primitives. S3: Based on the equivalent stiffness prediction primitive, cross-cycle comparison and stiffness difference judgment are performed on the same displacement interval, the predicted residual displacement evolution sequence is calculated and exponential smoothing and direction consistency judgment are performed to generate stable prediction data. S4: Based on the stable prediction data, select stable cyclic segments, calculate the interval equivalent stiffness parameters of the corresponding equivalent stiffness prediction elements, and combine the residual displacement evolution sequence to determine the evolution direction and generate a performance prediction constraint set. S5: Obtain the predicted cyclic displacement propagation path and locate the corresponding displacement interval. Combine the performance prediction constraint set to calculate the predicted force increment and analyze the force displacement path. Combine the residual displacement evolution direction to correct the state and generate the hysteresis performance prediction result.

[0021] The predicted input sequence includes displacement cycle segment index, force-displacement corresponding frame, and cycle direction marker. The equivalent stiffness prediction primitives include displacement interval number set, interval stiffness fitting, and force-displacement difference ratio. The stable prediction data includes residual displacement evolution sequence, smoothing weight coefficient, and direction consistency marker. The performance prediction constraint set includes interval equivalent stiffness parameter table, evolution direction criterion, and stable cycle segment index. The hysteresis performance prediction results include predicted force-displacement trajectory, state correction marker, and cycle termination feature point.

[0022] Please see Figure 2 The specific steps of S1 are as follows: S101: Obtain the displacement time history and force time history obtained by synchronous monitoring during the loading process of the coupling beam damper, judge the change of the sign of the displacement increment of adjacent sampling points in the displacement time history, record the sampling index position corresponding to the displacement increment turning from positive to negative or from negative to positive, mark the corresponding displacement value, and generate a set of direction reversal inflection point indexes. First, high-precision magnetostrictive displacement sensors and strain gauge force sensors deployed at key nodes of the coupling beam damper are activated, with a sampling frequency set to 1024 Hz, to acquire the simulated voltage signals of the coupling beam damper in real time during the reciprocating loading experiment. The simulated voltage signals are converted into digital signals via a multi-channel data acquisition card, transmitted to the processing terminal, and stored sequentially as displacement time history arrays and force time history arrays. Subsequently, a fourth-order Butterworth low-pass filtering algorithm is used to denoise the raw time history data, with the cutoff frequency parameter set to 50 Hz to filter out high-frequency electromagnetic interference and equipment vibration noise from the environment. After obtaining clean displacement time history data, a traversal search operation is initiated to check the sign changes of displacement increments at adjacent sampling points. Specifically, the current sampling point value and the previous sampling point value are extracted from the displacement time history, and a subtraction operation is performed to obtain the current displacement increment. Simultaneously, the values ​​of the previous sampling point and the sampling point before that are subtracted to obtain the preceding displacement increment. Next, the current displacement increment is multiplied by the previous displacement increment. If the product is negative, it indicates that the sign of the displacement increment has reversed, and this position is determined to be an inflection point where the direction of motion changes. The sampling index position corresponding to this inflection point is recorded in real time, and the displacement amplitude corresponding to this index is extracted as an extreme value marker. All identified inflection point indices are stored in the direction reversal inflection point index set in chronological order. For example, in actual monitoring, a segment of displacement data is extracted, which are 12.0 mm, 12.5 mm, 12.8 mm, and 12.6 mm. For the sampling point of 12.8 mm, its previous increment is 12.8 - 12.5 = 0.3 mm, and its current increment is 12.6 - 12.8 = -0.2 mm. Substituting 0.3 and -0.2 into the multiplication logic, the result is -0.06. Since the result is less than zero, the index position corresponding to 12.8 mm is determined to be a direction reversal inflection point, and its index number is recorded in the index set.

[0023] S102: Based on the direction reversal inflection point index set, call the corresponding sampling sequence of displacement time history, sequentially search the data segments between adjacent inflection point indices, determine the loop start index and loop end index based on the continuity of displacement change before and after the inflection point index, and obtain the loop boundary index sequence. Based on the key node information provided by the direction reversal inflection point index set, the long-sequence data of the original displacement time history is invoked to initiate sequential retrieval and morphological determination of data segments between adjacent inflection point indices. The i-th, (i+1)-th, and (i+2)-th inflection point indices in the index set are read sequentially. First, the absolute displacement difference between the i-th and (i+2)-th inflection points is calculated, and it is verified whether it satisfies the closed-loop condition. If the i-th inflection point is a trough, the (i+1)-th inflection point is a peak, and the (i+2)-th inflection point returns to the trough region, and the aforementioned absolute displacement difference is less than a preset minimum half-amplitude threshold, then the data segment from the i-th inflection point index to the (i+2)-th inflection point index is determined to be a complete hysteresis loop. This threshold is set to 2% of the maximum displacement amplitude. The i-th index is marked as the loop start index, and the (i+2)-th index is marked as the loop end index. These two index values ​​are combined into a boundary tuple and stored in the loop boundary index sequence. For each extracted potential cycle boundary, the cumulative energy dissipation index within that interval is further calculated, i.e., the force and displacement are integrated to ensure that the extracted cycle has effective energy dissipation characteristics. Table 1 shows some of the identified cycle boundary indices and their corresponding physical feature data. Taking the first set of data in Table 1 as an example, the starting index is identified as 500, the intermediate peak index as 750, and the ending index as 1000. First, the displacement value of -20 mm at index 500 and the displacement value of -19.8 mm at index 1000 are verified, and the absolute difference between the two is calculated to be 0.2 mm. The maximum displacement amplitude is set to 25 mm, and the threshold is 0.5 mm. Since 0.2 mm is less than 0.5 mm, it is confirmed as a closed cycle.

[0024] Table 1 Hysteresis Cycle Boundary Identification Table for Coupling Beam Dampers

[0025] S103: Based on the cyclic boundary index sequence, call the data frames corresponding to the boundary indices in the displacement time history and force time history, perform time-series alignment and length alignment on the displacement vectors and force vectors in the multi-cyclic intervals, and sequentially concatenate the rearranged vectors to generate the prediction input sequence. Based on the interval range determined by the loop boundary index sequence, corresponding local data frames are extracted from the global displacement time history and force time history, respectively. First, the maximum number of sampling points in all extraction loops is calculated and used as the baseline length. Then, for each displacement and force vector whose length is less than the baseline length, a cubic spline interpolation algorithm is called for resampling. This algorithm constructs a homogeneous polynomial between adjacent data points to calculate the corresponding new interpolation points on the normalized time axis, thereby stretching data sequences of different lengths to a unified baseline length, for example, setting the baseline length to 256 sampling points. After length alignment, Z-score standardization is performed on each vector, specifically calculating the arithmetic mean and standard deviation of the vector. The mean is subtracted from each element of the vector, and then the result is divided by the standard deviation to eliminate the influence of dimensions. Finally, the processed displacement and force vectors are concatenated along their feature dimensions, that is, the displacement vector and force vector of length 256 are merged column-wise to generate a standardized feature matrix of size 256 rows and 2 columns. The feature matrices of each loop are then concatenated in chronological order. For example, the original length of an extracted cyclic displacement vector is set to 200 points, and the baseline length is set to 256 points. First, an original time axis from 0 to 199 is constructed and mapped to a normalized interval from 0 to 1. Then, 256 equally spaced time nodes are generated within the 0-1 interval, and the displacement values ​​corresponding to these nodes are calculated using a cubic spline function. If the mean of the cyclic displacement is 10 mm and the standard deviation is 5 mm, for one interpolated displacement value of 15 mm, 15-10=5, 5 / 5=1, resulting in a normalized value of 1.0. This normalized displacement sequence is then concatenated with the corresponding normalized force sequence.

[0026] Please see Figure 3 The specific steps of S2 are as follows: S201: Based on the predicted input sequence, retrieve multi-cycle segment displacement time history data frames, perform extreme value judgment on the displacement sample value in each cycle, determine the maximum and minimum displacement values ​​based on the sign change of the displacement difference between adjacent sampling points, and aggregate the multi-cycle extreme value pairs to generate the multi-cycle displacement extreme value range. Based on the generated prediction input sequence, multi-cycle displacement time history data frames are retrieved in batches from memory, and a point-by-point scanning operation is performed on the continuous data stream containing multiple hysteresis cycles. A temporary extremum storage array is established to record the local maximum and local minimum displacement values ​​within each independent cycle. The specific execution logic is as follows: the displacement value of the current sampling point, the displacement value of the previous time step, and the displacement value of the next time step are read sequentially. First, a forward difference operation is performed, i.e., the forward increment is obtained by subtracting the previous time step displacement value from the current sampling point displacement value. Then, a backward difference operation is performed, i.e., the backward increment is obtained by subtracting the current sampling point displacement value from the next time step displacement value. A logical AND operation is performed on the signs of the forward and backward increments. If the forward increment is positive and the backward increment is negative, the current point is determined to be a local peak, i.e., the maximum displacement value; if the forward increment is negative and the backward increment is positive, the current point is determined to be a local trough, i.e., the minimum displacement value. The entire prediction input sequence is traversed, and each pair of identified extrema is stored in an extremum pair list in cyclic order. Subsequently, a convergence analysis was performed on the list, extracting the absolute maximum value among all local maxima as the global displacement upper limit and the absolute minimum value among all local minima as the global displacement lower limit, thus defining the extreme value range of the multi-cycle displacement. For example, in the monitoring data, the first cycle identified a peak of 20.0 mm and a trough of -20.0 mm; the second cycle had a peak of 20.2 mm and a trough of -19.8 mm; and the third cycle had a peak of 25.0 mm and a trough of -19.5 mm. By aggregating these values ​​and comparing them, the global displacement upper limit was determined to be 25.0 mm, and the global displacement lower limit to be -20.0 mm, thus determining the extreme value range of the multi-cycle displacement to be from -20.0 mm to 25.0 mm.

[0027] S202: Based on the extreme value range of multi-cycle displacement, call the single-cycle displacement advancement path sequence, construct the proportional coefficient according to the upper and lower limits of the extreme value range, perform proportional discretization on the displacement sampled values ​​within the path, use them as the boundary of the displacement interval to arrange them in sequence, and index and encode the interval boundary to establish the displacement interval sequence. Based on the determined extreme range of multi-cycle displacement, a standard single-cycle displacement propulsion path sequence is invoked. This sequence is preset to a normalized sine wave or triangular wave shape, aiming to provide a standardized loading path reference. First, the global span value of the extreme range is calculated, i.e., the global displacement upper limit minus the global displacement lower limit. Then, the discretization resolution parameter is set, for example, to 50 intervals, and a linear scaling factor sequence is constructed accordingly. The global span value is divided by the number of intervals to obtain the basic step size value. Next, starting from the global displacement lower limit, a series of sequentially arranged displacement numerical nodes are generated by accumulating the basic step size value. These nodes constitute the physical boundary of the displacement interval. For each generated interval, a unique integer index code is assigned according to the numerical order. Taking linear discretization as an example, the global displacement lower limit is set to -24.2 mm, the global displacement upper limit is set to 24.5 mm, and the number of intervals is set to 50. First, the global span value is calculated to be 24.5 - (-24.2) = 48.7 mm. Next, the base step size is calculated to be 48.7 / 50 = 0.974 mm. The starting boundary of the first interval is -24.2 mm, and the ending boundary is -24.2 + 0.974 = -23.226 mm, and so on, until it covers 24.5 mm. The interval from -24.2 mm to -23.226 mm is labeled as index 1, and the interval from -23.226 mm to -22.252 mm is labeled as index 2, ultimately forming a sequence containing 50 intervals.

[0028] S203: For the displacement interval sequence, call the stress time history and displacement time history data frames in the prediction input sequence, calculate the ratio of the force difference and displacement difference at the boundary points of adjacent displacement intervals, and perform parameter fitting operation on the multi-interval ratios in the interval index order to generate equivalent stiffness prediction primitives. For the generated displacement interval sequence, the corresponding force time history data frames and displacement time history data frames are called in parallel from the prediction input sequence. First, each displacement interval is traversed, identifying all sampling points falling within the interval boundary range, and extracting the force and displacement values ​​corresponding to the start and end positions of the interval. The difference between the end and start force values ​​is calculated as the interval force increment, and the difference between the end and start displacement values ​​is calculated as the interval displacement increment. Then, a division operation is performed, dividing the interval force increment by the interval displacement increment to obtain the secant stiffness ratio corresponding to that interval. The stiffness ratios of all intervals are arranged in index order to construct a stiffness distribution sequence. To generate continuous prediction primitives, a least-squares polynomial fitting algorithm is called to parametrically model the stiffness distribution sequence. This algorithm first constructs an error sum-of-squares function, that is, calculates the sum of squares of the differences between the fitted curve values ​​and the actual stiffness ratios. By taking the derivative and setting the derivative to zero, the normal equations are solved to obtain the polynomial coefficients. As shown in Table 2, a portion of the interval data is selected for calculation. For the interval with index 15, the displacement increment is 0.974 mm, corresponding to an initial force of 120.5 kN and an ending force of 135.2 kN, with a force increment of 14.7 kN. Dividing 14.7 by 0.974 yields an equivalent stiffness of 15.09 kN / mm for this interval. Similarly, for index 16, the force increment is 148.8 - 135.2 = 13.6 kN, and the stiffness is calculated as 13.6 / 0.974 = 13.96 kN / mm. For index 17, the force increment is 161.5 - 148.8 = 12.7 kN, and the stiffness is calculated as 12.7 / 0.974 = 13.04 kN / mm. These discrete stiffness values ​​are then fed into the fitting logic to generate equivalent stiffness prediction primitives.

[0029] Table 2 Reference Table for Segment Stiffness Calculation and Fitting of Coupling Beam Dampers

[0030] Please see Figure 4 The specific steps of S3 are as follows: S301: Based on the equivalent stiffness prediction primitive corresponding to the continuous cycle, retrieve the stiffness scalar sequence under the same displacement interval index of multiple cycles, calculate the difference between the stiffness scalars of adjacent cycles and compare it with the preset stiffness difference judgment threshold, determine the sign state of the difference and mark it, and generate a cross-cycle stiffness difference sequence. Based on the generated equivalent stiffness prediction primitives, a specific displacement interval index is first locked, such as index 15 in the aforementioned embodiment. Then, all consecutive loading cycles stored in the database are traversed, and the secant stiffness value corresponding to this specific index in each cycle is extracted to construct a stiffness scalar sequence. The stiffness value of the current cycle and the stiffness value of the previous cycle are selected sequentially, and a subtraction operation is performed, i.e., the stiffness value of the current cycle is subtracted from the stiffness value of the previous cycle to obtain the stiffness difference between adjacent cycles. To effectively distinguish between normal measurement noise and substantial stiffness degradation, a dynamic stiffness difference judgment threshold is constructed. The logic for setting this threshold is as follows: first, the standard deviation of all interval stiffness values ​​within the elastic segment range of the first cycle is calculated, and three times this standard deviation is used as the judgment threshold, thereby covering 99.7% of the random error interval. The calculated stiffness difference is compared with the threshold: if the stiffness difference is less than zero and its absolute value is greater than the threshold, it is determined to be a stiffness degradation state and marked as status code -1; if the absolute value of the stiffness difference is less than or equal to the threshold, it is determined to be a stable state and marked as status code 0; if the stiffness difference is greater than zero and its absolute value is greater than the threshold, it is determined to be an abnormal stiffness hardening state and marked as status code 1. These status codes are arranged in cyclic order. As shown in Table 3, for index 15, the first cycle stiffness is 15.09 kN / mm, the second cycle stiffness is 14.35 kN / mm, and the third cycle stiffness is 13.80 kN / mm. The standard deviation of the first cycle elastic segment stiffness is set to 0.05 kN / mm, so the threshold is set to 0.05 * 3 = 0.15 kN / mm. The difference between the second cycle and the first cycle is calculated as 14.35 - 15.09 = -0.74 kN / mm. Since the absolute value of -0.74 is greater than 0.15 and the sign is negative, it is determined to be stiffness degradation. Similarly, the difference between the third cycle and the second cycle is -0.55 kN / mm, which is also determined to be stiffness degradation. The advantage of this logical operation is that, through dynamic threshold determination based on statistical principles, it can eliminate experimental noise interference and accurately extract the stiffness evolution characteristics of the coupling beam damper under cumulative damage.

[0031] S302: Based on the cross-cycle stiffness difference sequence, collect displacement time history data frames corresponding to the end time of multiple cycles, index and align the displacement values ​​at the end of multiple cycles and aggregate the values, and map the displacement values ​​sequentially according to the stiffness difference judgment sequence index and analyze the displacement vector to obtain the predicted residual displacement evolution sequence. Based on the generated cross-cycle stiffness difference sequence, the original displacement time history database is accessed in parallel. According to the cycle termination index defined in step S102, the displacement sampling value corresponding to the end of each cycle is precisely extracted. These displacement values ​​represent the unrecoverable deformation of the coupling beam damper after experiencing complete hysteresis energy dissipation, i.e., residual displacement. The extracted residual displacement values ​​are indexed and aligned according to the cycle number, and correspond one-to-one with the stiffness difference sequence. A filtering logic is executed, retaining only the residual displacement data marked as "stiffness degradation" (i.e., status code -1) and "stable" (i.e., status code 0) in the stiffness difference judgment sequence, removing noise data corresponding to "hardening anomaly," and recombining the retained displacement values ​​in chronological order to form a predicted residual displacement evolution sequence. For example, continuing with the aforementioned data, the residual displacement at the end of the first cycle is 0.45 mm, at the end of the second cycle is 0.92 mm, and at the end of the third cycle is 1.45 mm. Since the second and third cycles both exhibit stiffness degradation in the aforementioned analysis, these three data points are completely retained and arranged in order to construct the initial displacement vector. This process also requires zero-point drift correction, where the residual displacement of the first cycle is used as the baseline offset. The residual displacement values ​​of all subsequent cycles are then subtracted from the algebraic sum of this baseline offset and the initial preload deformation, yielding the net residual displacement purely caused by accumulated plastic damage. Setting the initial preload deformation to 0 mm ensures the corrected sequence remains unchanged. The advantage of this logical operation is that it establishes a direct mapping relationship between the physical phenomenon of stiffness degradation and macroscopic residual deformation, providing a physically consistent data foundation for subsequent life prediction.

[0032] S303: For the predicted residual displacement evolution sequence, the preset exponential weight coefficient sequence and displacement vector are called to perform recursive smoothing calculation, and the consistency of the sign direction of adjacent sequence items is judged. The sequence items that pass the judgment are indexed, updated and aggregated to generate stable prediction data. For the generated predicted residual displacement evolution sequence, a preset exponential weight coefficient sequence is invoked to initiate recursive smoothing calculations to eliminate random fluctuations and extract the evolution trend. A smoothing coefficient is set, typically between 0.6 and 0.8, to balance the weights of current observations and historical trends. The calculation follows the logic of an exponentially weighted moving average: the first smoothed value equals the first observation; from the second point onwards, the current smoothed value equals the smoothing coefficient multiplied by the current observation, plus (1 minus the smoothing coefficient) multiplied by the "smoothed value of the previous moment." After smoothing calculations, a monotonicity consistency check is performed on the smoothed sequence. Specifically, the difference sign of adjacent elements in the smoothed sequence is calculated. If the difference signs of three consecutive points remain consistent (e.g., all positive, representing continuous cumulative deformation), the trend is considered valid; if a sign reversal occurs, it is considered a local oscillation, triggering a backtracking mechanism to replace the values ​​in that oscillation interval using linear interpolation. Finally, the sequence that passes consistency verification is indexed, rearranged, and aggregated to generate the final stable prediction data. For example, a smoothing coefficient of 0.7 is set. Substituting the aforementioned sequence values ​​of 0.45 mm, 0.92 mm, and 1.45 mm into the calculation, the first smoothed value is 0.45 mm. The second smoothed value is 0.7 * 0.92 + 0.3 * 0.45 = 0.644 + 0.135 = 0.779 mm. The third smoothed value is 0.7 * 1.45 + 0.3 * 0.779 = 1.015 + 0.2337 = 1.2487 mm. A trend assessment is then performed: the second value (0.779 - 0.45) = 0.329 (positive), and the third value (1.2487 - 0.779) = 0.4697 (positive). Since the signs are consistent and all are positive, the sequence passes the consistency test and is confirmed as stable prediction data. Table 3 summarizes the correlation data between stiffness degradation and residual displacement evolution.

[0033] Table 3. Data on stiffness degradation and residual displacement evolution of coupling beam dampers

[0034] Please see Figure 5 The specific steps of S4 are as follows: S401: Based on stable prediction data, filter continuous cyclic segments of stable prediction states, determine the consistency of the stable prediction data sequence, index and concatenate the cyclic numbers that meet the preset stable prediction state determination threshold, mark the segment boundaries, and perform segment length verification to generate a stable cyclic segment index set. Based on the generated stable prediction data sequence, i.e., the residual displacement evolution sequence after exponential smoothing, a sliding window scanning algorithm based on the rate of change was constructed to extract continuous segments representing the structural stability damage evolution characteristics from the long sequence data of the entire life cycle. First, the length of the sliding window is set, for example, to 5 cycles, and the first-order difference sequence of residual displacement values ​​is calculated within the window. Next, a stable prediction state determination threshold is defined. This threshold is based on the statistical characteristics of the displacement growth rate of the coupling beam damper in the stable section of the fatigue test. Specifically, it is obtained by calculating the root mean square value of the displacement increment in the stable section of the test and multiplying it by a safety factor of 1.5, for example, set to 0.02 mm per cycle. The absolute value of the difference between adjacent data points within the window is calculated one by one. If the absolute value of all differences within the window is strictly less than the stable prediction state determination threshold, the area covered by the current window is determined to be in a stable prediction state, and the starting and ending cycle numbers corresponding to the window are extracted. Adjacent and overlapping stable windows are indexed and concatenated to form independent continuous cyclic segments. Subsequently, the segment length verification logic is executed, setting the minimum effective length threshold to 10 cycles. Transient fluctuation segments with insufficient length are eliminated, and the start and end indices of segments that meet the conditions are finally stored in the stable cyclic segment index set. For example, for a sequence of data: 1.249 mm, 1.255 mm, 1.260 mm, 1.264 mm, 1.268 mm, the adjacent differences are calculated to be 0.006 mm, 0.005 mm, 0.004 mm, and 0.004 mm, respectively. The judgment threshold is set to 0.02 mm. Since all differences are less than 0.02 mm, the data segment is determined to be in a stable state. If the segment continues to the 50th cycle and the total length exceeds 10 cycles, it is recorded as a valid segment (e.g., index interval 10 to 50) in the index set. The advantage of this logic operation is that, through strict threshold constraints and length verification, high-confidence data intervals for subsequent stiffness-displacement coupling analysis are accurately locked.

[0035] S402: Based on the stable cyclic segment index set, call the equivalent stiffness prediction primitive under the corresponding cyclic number, perform numerical aggregation of stiffness scalars in multiple segments according to segment index, calculate the mean within the segment, arrange the mean results of multiple segments according to index order, and obtain the interval equivalent stiffness parameter. Based on the generated stable cyclic segment index set, for each cyclic segment marked as stable, the generated equivalent stiffness prediction primitive data is back-called. First, a specific displacement interval index is locked (such as index 15 mentioned above). Then, according to the start and end cycle numbers of the stable segment, the secant stiffness scalar values ​​corresponding to each cycle within that time span are extracted in batches from the database. Numerical aggregation is performed on the extracted multi-segment stiffness scalars. Specifically, an accumulator variable and a counter variable are constructed, and all stiffness values ​​within the segment are accumulated and summed. Finally, the sum is divided by the total number of cycles to obtain the average stiffness value within the segment, i.e., the interval equivalent stiffness parameter. The above calculation is repeated for all identified stable segments, and the calculated mean results are arranged according to the chronological order of the segments on the time axis. As shown in Table 4, the stiffness aggregation calculation results of some stable segments are displayed. Taking segment number 1 in Table 4 as an example, this segment covers the cyclic range from cycle 50 to 80, with a total of 31 data points. The stiffness values ​​at displacement index 15 of these 31 cycles were extracted, with a maximum of 12.80 kN / mm and a minimum of 12.40 kN / mm. These 31 values ​​were summed to obtain a total of 390.6 kN / mm. Then, 390.6 / 31 = 12.60, yielding the average equivalent stiffness of this section as 12.60 kN / mm. Similarly, for section number 2 (cycles 120 to 150), the average stiffness was calculated to be 11.85 kN / mm. These statistically averaged stiffness parameters effectively eliminate random fluctuations in single-cycle measurements, providing robust physical parameters for establishing a high-precision damage constitutive model.

[0036] Table 4. Calculation Table of Equivalent Stiffness Parameters for Stable Cyclic Section

[0037] S403: For the equivalent stiffness parameters of the interval, call the predicted residual displacement evolution sequence, perform sign and direction discrimination on the displacement sequence corresponding to the multi-cycle segment, and perform joint mapping with the equivalent stiffness parameters of the segment. Construct an index set for the parameter entries that meet the direction consistency condition, and generate a performance prediction constraint set. For the calculated interval equivalent stiffness parameters, the generated predicted residual displacement evolution sequence is called again to jointly verify and encapsulate the evolution law between physical parameters. First, the residual displacement subsequence corresponding to each stable cycle segment is extracted, and the trend direction discrimination logic is executed. Specifically, the difference between the displacement value at the end of the segment and the displacement value at the beginning of the segment is calculated. If the difference is positive, it is determined to be in the "cumulative growth" direction; if the difference is negative, it is determined to be in the "recovery and decrease" direction. Subsequently, the displacement direction determination result is jointly mapped and analyzed with the corresponding interval equivalent stiffness parameters. According to the damage mechanics principle of the coupling beam damper, the degradation of stiffness (i.e., the stiffness value decreases with time) should be accompanied by the accumulation of residual displacement (i.e., the displacement value increases with time). Therefore, a consistency verification rule is set: if the average stiffness value of the current segment decreases relative to the previous segment, and the residual displacement in the segment shows a cumulative growth trend, then the direction consistency condition is satisfied. The average stiffness value, displacement change rate, and segment index range that meet this condition are packaged to generate a performance prediction constraint set. Taking the data in Table 4 as an example, comparing segment 1 and segment 2, the average stiffness decreased from 12.60 kN / mm to 11.85 kN / mm, showing stiffness degradation. Simultaneously, examining the residual displacement sequence corresponding to segment 2, with an initial displacement of 1.8 mm and an ending displacement of 2.1 mm, the difference is 0.3 mm (a positive value), indicating cumulative growth. Since the decrease in stiffness and the increase in displacement are highly consistent in physical logic, "stiffness 11.85 kN / mm" and "displacement growth rate 0.01 mm per cycle" are stored as a related entry in the constraint set.

[0038] Please see Figure 6 The specific steps of S5 are as follows: S501: Obtain the displacement propagation path of the cycle to be predicted, arrange multiple discrete displacement points in the displacement propagation path, determine the continuity of adjacent displacement points according to the numerical increment relationship, complete the path segmentation identification, map the displacement point sequence to the corresponding displacement value range, and generate the predicted cycle displacement interval sequence. First, the displacement propagation path data of the coupling beam damper to be predicted is read through a high-frequency data interface. This data usually comes from a preset seismic performance loading regime or seismic response time history analysis results, and includes a series of discrete displacement coordinate points arranged according to time steps. A maximum displacement step judgment threshold is set, which is based on the minimum response resolution of the servo loading, for example, set to 0.5 mm. The numerical difference between two adjacent displacement points in the displacement propagation path is calculated sequentially. If the absolute value of this difference is less than or equal to the maximum displacement step judgment threshold, the path continuity is deemed valid; if it is greater than the threshold, an interpolation completion mechanism is triggered, inserting linear transition points at the breakpoints to ensure path smoothness. Subsequently, the sequence of displacement points that pass the continuity check is segmented and labeled, identifying the extreme points (i.e., peaks and troughs) in the path. The segment from trough to peak is marked as the "forward loading segment," and the segment from peak to trough is marked as the "reverse unloading segment." Each discrete displacement point after segmentation is traversed, and its value is matched and mapped with the displacement value range divided in step S102. For example, if the displacement point value is 12.5 mm, and the preset displacement interval 3 ranges from 10 mm to 15 mm, this displacement point is marked as interval index 3. The interval index values ​​corresponding to all displacement points are rearranged in chronological order to generate a predicted cyclic displacement interval sequence. Setting the path to be predicted to a monotonically loaded range of 0 mm to 20 mm, with a sampling interval displacement of 1 mm and an interval span of 5 mm, an ordered sequence containing index 1 (0 to 5 mm), index 2 (5 to 10 mm), index 3 (10 to 15 mm), and index 4 (15 to 20 mm) is generated. The advantage of this logical operation is that, through refined discretization and interval mapping, continuous macroscopic physics is transformed into a digital state flow that can be recognized by a discrete stiffness model.

[0039] S502: Based on the predicted cyclic displacement interval sequence, call the equivalent stiffness parameters of the corresponding interval in the performance prediction constraint set, multiply the adjacent displacement increments and equivalent stiffness parameters in the multi-displacement intervals to calculate the force increment sequence and perform cumulative calculation to generate predicted cyclic force displacement path data. Based on the generated sequence of predicted cyclic displacement intervals, the equivalent stiffness parameter corresponding to each displacement interval index is retrieved from the generated performance prediction constraint set. The current cumulative force value variable is initialized to zero (or the end force value of the previous moment), and the point-by-point recursive calculation process is started. First, the difference between the current displacement point and the previous displacement point is calculated to obtain the adjacent displacement increment; then, the equivalent stiffness parameter corresponding to the interval to which the current displacement point belongs is called, and the adjacent displacement increment is multiplied by the equivalent stiffness parameter to obtain the force increment value of the current step. This force increment value is summed with the current cumulative force value variable to update the predicted force value of the current moment, and the force value and the corresponding displacement value are combined to form a coordinate pair and stored in the predicted cyclic force-displacement path data. For example, when running to the 3rd displacement interval (10 mm to 15 mm), the equivalent stiffness parameter of this interval is found to be 12.0 kN / mm. If the displacement at the previous moment was 10.5 mm, the displacement at the current moment is 11.0 mm, and the cumulative force value at the previous moment was 120 kN. The displacement increment is calculated as 11.0 - 10.5 = 0.5 mm. Multiplying this displacement increment of 0.5 mm by the stiffness parameter of 12.0 kN per millimeter yields an internal force increment of 6.0 kN. Adding this force increment of 6.0 kN to the cumulative force value of the foundation of 120 kN gives the predicted force value at the current moment as 126 kN. Table 5 shows the detailed data flow of this recursion. Through the cumulative mapping of the entire path, the complete hysteresis curve shape can be reconstructed. This calculation logic effectively transforms the discrete stiffness degradation characteristics into a continuous nonlinear mechanical response.

[0040] Table 5 Recursive Calculation Table for Predicted Cyclic Force Displacement Path

[0041] S503: For the predicted cyclic force-displacement path data, obtain the displacement state at the predicted cycle termination position, call the residual displacement evolution direction in the performance prediction constraint set, perform direction consistency judgment on the force-displacement state at the termination position, correct the sign of inconsistent terms, and generate hysteresis performance prediction results. For the generated predicted cyclic force-displacement path data, the focus is on the boundary state at the end of the cycle (i.e., when the external force is unloaded to zero or the displacement reverses). The displacement value of the last point in the predicted path is extracted, i.e., the predicted residual displacement value, and the "residual displacement evolution direction" marker (e.g., "cumulative growth" or "recovery decrease") recorded in the performance prediction constraint set of step S403 is invoked. The direction consistency judgment logic is executed: if the constraint set is marked as "cumulative growth," the absolute value of the current predicted residual displacement value must be greater than or equal to the absolute value of the residual displacement value of the previous cycle. If the judgment result is inconsistent (e.g., the predicted value decreases abnormally), a sign correction mechanism is triggered. This mechanism first calculates the sum of the residual displacement value of the previous cycle and the preset minimum growth step size as the correction reference value; then, all displacement points in the unloading segment of the current predicted path are comprehensively translated and corrected, so that the final residual displacement value is forcibly aligned to this correction reference value, thereby eliminating prediction deviations that violate physical laws due to local fluctuations in stiffness parameters. For example, if the residual displacement of the previous cycle is 1.5 mm, the constraint set indicates the direction as "growth." If the original predicted residual displacement calculated in step S502 is 1.4 mm (less than 1.5 mm, inconsistent), and the preset minimum growth step size is 0.01 mm, the corrected baseline value is calculated as 1.5 + 0.01 = 1.51 mm. The offset is calculated as 1.51 - 1.4 = 0.11 mm, and the displacement values ​​of all points in the unloading section are increased by 0.11 mm to generate the final hysteresis performance prediction result. The advantage of this logical operation is that it introduces post-processing constraints based on the physical damage evolution law, ensuring the thermodynamic consistency of the prediction results in the long-period sequence and avoiding the physical consistency deviation that may be generated by the pure data-driven method.

[0042] Please see Figure 7 A hysteresis performance prediction system for coupling beam dampers includes: The data processing module obtains displacement time history and force time history by monitoring the loading process of the coupling beam damper, determines the direction reversal position of the displacement time history, and performs cyclic identification and segmentation of continuous monitoring data to generate a predictive input sequence and transmit it to the stiffness modeling module. The stiffness modeling module obtains the extreme value range of multi-cycle displacement based on the predicted input sequence, proportionally discretizes the cyclic displacement propagation path into multiple displacement intervals, calculates and fits the ratio of force difference to displacement difference between adjacent displacement intervals, generates equivalent stiffness prediction primitives, and passes them to the evolution prediction module. The evolution prediction module, based on the equivalent stiffness prediction primitive, performs cross-cycle comparison and stiffness difference judgment on the same displacement interval, calculates the predicted residual displacement evolution sequence and performs exponential smoothing and direction consistency judgment, generates stable prediction data and transmits it to the constraint construction module. The constraint construction module filters stable cyclic segments based on stability prediction data, calculates the interval equivalent stiffness parameters of the corresponding equivalent stiffness prediction primitives, and determines the evolution direction by combining the residual displacement evolution sequence, generating a performance prediction constraint set and passing it to the performance prediction module. The performance prediction module obtains the propagation path of the cyclic displacement to be predicted and locates the corresponding displacement interval. It calculates the predicted force increment and analyzes the force-displacement path in combination with the performance prediction constraint set. It also corrects the state in combination with the evolution direction of the residual displacement and generates the hysteresis performance prediction result.

[0043] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for predicting the hysteresis performance of a coupling beam damper, characterized in that, Includes the following steps: S1: Obtain displacement time history and force time history by monitoring the loading process of the coupling beam damper, determine the direction reversal position of the displacement time history, and perform cyclic identification and segmentation of continuous monitoring data to generate a predictive input sequence; S2: Based on the predicted input sequence, obtain the extreme value range of multi-cycle displacement, proportionally discretize the cyclic displacement propulsion path into multiple displacement intervals, calculate and fit the force difference to displacement difference ratio of adjacent displacement intervals, and generate equivalent stiffness prediction primitives. S3: Based on the equivalent stiffness prediction primitive, cross-cycle comparison and stiffness difference judgment are performed on the same displacement interval, the predicted residual displacement evolution sequence is calculated and exponential smoothing and direction consistency judgment are performed to generate stable prediction data. S4: Based on the stability prediction data, filter the stable cyclic segment, calculate the interval equivalent stiffness parameter corresponding to the equivalent stiffness prediction element, and combine the residual displacement evolution sequence to determine the evolution direction, and generate a performance prediction constraint set; S5: Obtain the predicted cyclic displacement propagation path and locate the corresponding displacement interval. Combine the performance prediction constraint set to calculate the predicted force increment and analyze the force displacement path. Combine the residual displacement evolution direction correction state to generate hysteresis performance prediction results.

2. The method for predicting the hysteresis performance of a coupling beam damper according to claim 1, characterized in that, The predicted input sequence includes a displacement cycle segment index, force-displacement corresponding frames, and cycle direction markers. The equivalent stiffness prediction primitives include a displacement interval number set, interval stiffness fitting, and force-displacement difference ratio. The stable prediction data includes a residual displacement evolution sequence, smoothing weight coefficients, and direction consistency identifiers. The performance prediction constraint set includes an interval equivalent stiffness parameter table, evolution direction criterion, and stable cycle segment index. The hysteresis performance prediction results include predicted force-displacement trajectories, state correction markers, and cycle termination feature points.

3. The method for predicting the hysteresis performance of a coupling beam damper according to claim 1, characterized in that, The specific steps of S1 are as follows: S101: Obtain the displacement time history and force time history obtained by synchronous monitoring during the loading process of the coupling beam damper, judge the change of the sign of the displacement increment of adjacent sampling points in the displacement time history, record the sampling index position corresponding to the displacement increment turning from positive to negative or from negative to positive, mark the corresponding displacement value, and generate a set of direction reversal inflection point indexes. S102: Based on the direction reversal inflection point index set, call the corresponding sampling sequence of displacement time history, sequentially search the data segments between adjacent inflection point indices, determine the loop start index and loop end index according to the continuity of displacement change before and after the inflection point index, and obtain the loop boundary index sequence. S103: Based on the cyclic boundary index sequence, call the data frames corresponding to the boundary indices in the displacement time history and force time history, perform time-series alignment and length alignment on the displacement vectors and force vectors in the multi-cyclic intervals, and sequentially concatenate the rearranged vectors to generate a prediction input sequence.

4. The method for predicting the hysteresis performance of a coupling beam damper according to claim 1, characterized in that, The specific steps of S2 are as follows: S201: Based on the predicted input sequence, retrieve multi-cycle segment displacement time history data frames, perform extreme value judgment on the displacement sample value in each cycle, determine the maximum and minimum displacement values ​​based on the sign change of the displacement difference between adjacent sampling points, and aggregate the multi-cycle extreme value pairs to generate a multi-cycle displacement extreme value range. S202: Based on the multi-cycle displacement extreme value range, call the single-cycle displacement advancement path sequence, construct a proportional coefficient according to the upper and lower limits of the extreme value range, perform proportional discretization on the displacement sample values ​​within the path, use them as the sequentially arranged displacement interval boundaries, and index and encode the interval boundaries to establish a displacement interval sequence. S203: For the displacement interval sequence, call the stress time history and displacement time history data frames in the prediction input sequence, calculate the ratio of the force difference and displacement difference at the boundary points of adjacent displacement intervals, and perform parameter fitting operation on the multi-interval ratios in the order of interval index to generate equivalent stiffness prediction primitives.

5. The method for predicting the hysteresis performance of a coupling beam damper according to claim 1, characterized in that, The specific steps for S3 are as follows: S301: Based on the equivalent stiffness prediction primitive corresponding to the continuous cycle, retrieve the stiffness scalar sequence under the same displacement interval index of multiple cycles, calculate the difference between the stiffness scalars of adjacent cycles and compare it with the preset stiffness difference judgment threshold, determine the sign state of the difference and mark it, and generate a cross-cycle stiffness difference sequence. S302: Based on the cross-cycle stiffness difference sequence, collect displacement time history data frames corresponding to the end time of multiple cycles, index and align the displacement values ​​at the end of multiple cycles and converge the values, and map the displacement values ​​sequentially according to the stiffness difference judgment sequence index and analyze the displacement vector to obtain the predicted residual displacement evolution sequence. S303: For the predicted residual displacement evolution sequence, a preset exponential weight coefficient sequence and displacement vector are called to perform recursive smoothing calculation, and the consistency of the sign direction of adjacent sequence items is judged. The sequence items that pass the judgment are indexed, updated and aggregated to generate stable prediction data.

6. The method for predicting the hysteresis performance of a coupling beam damper according to claim 5, characterized in that, The stiffness difference determination threshold is determined by acquiring the initial loading cycle monitoring data of the coupling beam damper in the elastic linear deformation stage, calculating the difference sequence of equivalent stiffness between adjacent cycles in the initial loading cycle, calculating the stiffness dispersion by standard deviation of the difference sequence, and multiplying the stiffness dispersion with a preset confidence interval factor.

7. The method for predicting the hysteresis performance of a coupling beam damper according to claim 1, characterized in that, The specific steps of S4 are as follows: S401: Based on the stable prediction data, filter the continuous cyclic segments of stable prediction states, determine the consistency of the stable prediction data sequence, index and concatenate the cyclic numbers that meet the preset stable prediction state determination threshold, mark the segment boundaries, and perform segment length verification to generate a stable cyclic segment index set. S402: Based on the stable cyclic segment index set, call the equivalent stiffness prediction primitive under the corresponding cyclic number, perform numerical aggregation of stiffness scalars in multiple segments according to segment index, calculate the mean within the segment, arrange the mean results of multiple segments according to index order, and obtain the interval equivalent stiffness parameter. S403: For the interval equivalent stiffness parameters, call the predicted residual displacement evolution sequence, perform sign and direction discrimination on the displacement sequence corresponding to the multi-cycle segment, and perform joint mapping with the segment equivalent stiffness parameters. Construct an index set for parameter entries that meet the direction consistency condition, and generate a performance prediction constraint set.

8. The method for predicting the hysteresis performance of a coupling beam damper according to claim 7, characterized in that, The stability prediction state determination threshold is determined by obtaining the residual displacement evolution sequence of the connecting beam damper, performing differential operations on the displacement values ​​of consecutive adjacent cycles in the sequence to obtain the displacement fluctuation sequence, calculating the standard deviation statistical value of the displacement fluctuation sequence, and weighting the standard deviation statistical value with a preset tolerance coefficient.

9. The method for predicting the hysteresis performance of a coupling beam damper according to claim 1, characterized in that, The specific steps of S5 are as follows: S501: Obtain the displacement propagation path of the cycle to be predicted, arrange multiple discrete displacement points in the displacement propagation path, determine the continuity of adjacent displacement points according to the numerical increment relationship, complete the path segmentation identification, map the displacement point sequence to the corresponding displacement value range, and generate the predicted cycle displacement interval sequence. S502: Based on the predicted cyclic displacement interval sequence, call the equivalent stiffness parameter of the corresponding interval in the performance prediction constraint set, multiply the adjacent displacement increments and equivalent stiffness parameters in the multiple displacement intervals to calculate the force increment sequence and perform cumulative calculation to generate predicted cyclic force displacement path data. S503: For the predicted cyclic force-displacement path data, obtain the displacement state at the predicted cycle termination position, call the residual displacement evolution direction in the performance prediction constraint set, perform direction consistency judgment on the force-displacement state at the termination position, correct the sign of inconsistent terms, and generate hysteresis performance prediction results.

10. A hysteresis performance prediction system for a coupling beam damper, characterized in that, The system is used to implement the hysteresis performance prediction method for a coupling beam damper according to any one of claims 1-9, and the system comprises: The data processing module obtains displacement time history and force time history by monitoring the loading process of the coupling beam damper, determines the direction reversal position of the displacement time history, and performs cyclic identification and segmentation of continuous monitoring data to generate a predictive input sequence and transmit it to the stiffness modeling module. The stiffness modeling module obtains the extreme value range of multi-cycle displacement based on the predicted input sequence, proportionally discretizes the cyclic displacement propagation path into multiple displacement intervals, calculates and fits the ratio of force difference to displacement difference between adjacent displacement intervals, generates equivalent stiffness prediction primitives, and transmits them to the evolution prediction module. The evolution prediction module, based on the equivalent stiffness prediction primitive, performs cross-cycle comparison and stiffness difference judgment on the same displacement interval, calculates the predicted residual displacement evolution sequence and performs exponential smoothing and direction consistency judgment, generates stable prediction data and transmits it to the constraint construction module. The constraint construction module filters stable cyclic segments based on the stability prediction data, calculates the interval equivalent stiffness parameters corresponding to the equivalent stiffness prediction primitives, and determines the evolution direction by combining the residual displacement evolution sequence, generating a performance prediction constraint set and passing it to the performance prediction module. The performance prediction module obtains the predicted cyclic displacement propagation path and locates the corresponding displacement interval. It calculates the predicted force increment and analyzes the force-displacement path in combination with the performance prediction constraint set. It also generates the hysteresis performance prediction result by combining the residual displacement evolution direction correction state.

Images