Digital Modeling Method for the Operation of Watch Escapement Gear Shaft

By constructing a low-dimensional state update function and a recursive convolutional state recursion relationship, the elastic deformation displacement and contact force of the escapement gear shaft are updated in real time. This solves the problem of microsecond-level deformation and contact force coupling in the drop impact of the escapement gear shaft, and improves the prediction accuracy and the reliability of fracture risk assessment.

CN122389606BActive Publication Date: 2026-11-17QINGDAO ZHISHENGDA INSTRUMENT CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202610528796.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2026-04-21
Publication Date
2026-11-17
Estimated Expiration
2046-04-21

AI Technical Summary

Technical Problem

Existing technologies cannot effectively characterize the instantaneous strong coupling dynamic process of microsecond-level elastic deformation and contact force in digital modeling of escapement gear shaft drop impact, resulting in inaccurate prediction results.

Method used

A low-dimensional state update function is constructed. By collecting time-domain response data of the escapement gear shaft, intrinsic orthogonal decomposition and impulse response analysis are performed to establish a recursive convolution state recursive relationship. The elastic deformation displacement increment and contact force are updated in real time. Combined with the Hertzian contact stiffness expression, iterative calculation is performed to set a fracture early warning signal.

Benefits of technology

This method enables the prediction of transient stress distribution of the escapement gear shaft under drop impact to be closer to physical reality, reduces the number of physical tests, and improves the reliability of fracture risk assessment.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122389606B_ABST
    Figure CN122389606B_ABST
Patent Text Reader

Abstract

This invention relates to the field of watchmaking mechanical design and simulation analysis technology, specifically disclosing a digital modeling method for the operating state of a watch escapement gear shaft. The method involves collecting the displacement and stress history of the tooth tip and adjacent shoulder areas; extracting the dominant deformation basis vector based on intrinsic orthogonal decomposition; establishing a low-dimensional state update function through Laplace domain continuous fraction expansion and bilinear transformation; and using historical state vectors and current contact force as inputs and outputs the elastic deformation displacement increment. The contact geometry is iteratively updated and the contact force is recalculated until convergence during time progression. Simultaneously, a fracture warning signal is generated when the stress exceeds a threshold for three consecutive time steps using a first-in, first-out buffer. This invention achieves strong online coupling between elastic deformation and contact force, improving the accuracy of stress distribution prediction under drop impact.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of clock and watch mechanical design and simulation analysis technology, specifically to a digital modeling method for the operating state of a clock and watch escapement gear shaft. Background Technology

[0002] In the field of watchmaking mechanical design, the escapement mechanism, as a core component of timekeeping, directly affects the reliability and structural integrity of the movement when its gear shaft is subjected to a drop impact. Existing digital modeling methods for drop impact on escapement gear shafts employ an offline sequential coupling strategy combining multibody dynamics and the finite element method: first, the global multibody dynamics model calculates the overall contact force time history during the impact process; then, the force at the moment of maximum load or the entire force time history is used as boundary conditions and mapped onto the local gear shaft finite element model for stress and deformation analysis. This type of method can obtain relatively reasonable prediction results under static or quasi-static conditions.

[0003] This invention solves the problem that existing offline sequential coupling methods cannot characterize the instantaneous strong coupling dynamic process of microsecond-level elastic deformation and contact force of escapement gear shaft in drop impact due to the misalignment of load transmission timing. This is achieved by constructing a low-dimensional state update function with historical state vector and current contact force as input, and iteratively updating the elastic deformation displacement increment and contact geometry in each time step. Summary of the Invention

[0004] The purpose of this invention is to provide a digital modeling method for the operating state of the escapement gear shaft in a clock, so as to solve the problems mentioned above.

[0005] The objective of this invention can be achieved through the following technical solutions: A digital modeling method for the operating state of a watch escapement gear shaft includes the following steps: S1, collect time-domain response data of the escapement gear shaft under a typical impact load spectrum. The time-domain response data includes the displacement history and stress history of the tooth tip and the adjacent part of the shaft shoulder. S2, based on the displacement history, the dominant deformation basis vector is extracted by eigenorthogonal decomposition to construct a reduced-order subspace, and the impulse response analysis of each basis vector is performed according to the stress history to establish a recursive relationship of the recursive convolution state; S3 compiles the recursive relationship of the recursive convolution state into a low-dimensional state update function. The low-dimensional state update function takes the contact force input at the current time step and the historical state vector as input and outputs the elastic deformation displacement increment at the current time step. S4, during the drop impact time advance, after obtaining the contact force at each time step, the low-dimensional state update function is called to obtain the elastic deformation displacement increment. Then, the instantaneous contact geometry of the escapement tooth shaft is corrected based on the elastic deformation displacement increment, and the contact force is recalculated based on the corrected geometry until convergence. The converged contact force is used as the input for the next time step. S5. In each convergence time step, the stress output stored in the low-dimensional state update function is compared with the preset transient fracture criterion threshold. When the stress output exceeds the transient fracture criterion threshold for three consecutive time steps, a fracture warning signal is generated.

[0006] As a further aspect of the present invention: S2 specifically includes: The snapshot matrix formed by the displacement history is subjected to eigenorthogonal decomposition to obtain the initial basis vector set; The stress history is projected onto each initial basis vector, the stress contribution factor corresponding to each basis vector is calculated, and the stress contribution factors are sorted from high to low. The first few basis vectors whose cumulative stress contribution factors reach a set threshold are selected as the dominant deformation basis vectors, and the impulse response function is fitted in the Laplace domain based on the dominant deformation basis vectors to construct the recursive relationship of the recursive convolution state.

[0007] As a further aspect of the present invention: the step of fitting the impulse response function to the Laplace domain based on the dominant deformable basis vector to construct a recursive relationship for the recursive convolution state specifically includes: Transform the impulse response functions corresponding to each dominant deformation basis vector to the Laplace domain to obtain a set of complex frequency domain transfer functions. We use a continued fraction expansion to perform a reduced-order rational approximation of each complex frequency domain transfer function to obtain the set of low-order poles and residues. By inversely transforming the poles and residues back to the time domain, discrete recursive convolution coefficients are formed with the historical state vector as input and the current elastic deformation displacement increment as output.

[0008] As a further aspect of the present invention: the calculation process of the discrete recursive convolution coefficients specifically includes: Perform a bilinear transformation on each pair of poles and residues to map the poles and residues in the continuous time domain to the discrete poles and residues in the discrete time domain. Using discrete residues as weighting coefficients, the components of the historical state vector are multiplied and accumulated one by one with the weighting coefficients in reverse chronological order, and then multiplied by the sampling interval of the unit impulse response. By summing the product of all poles, we obtain the discrete recursive convolution coefficient between the current elastic deformation displacement increment and the historical state vector.

[0009] As a further aspect of the present invention: S3 specifically includes: The dimension of the historical state vector is determined by the number of poles in the recursive relationship of the recursive convolution state, and an operational array consisting of serial delay units and parallel weighting units is constructed based on the dimension. The contact force input at the current time step is used as the excitation value input to the first delay unit of the operation array, while each delay unit sequentially transmits the historical state components of the previous moment. Each parallel weighted unit reads the state component output by the corresponding delay unit and multiplies and accumulates it with the pre-stored recursive convolution coefficients, and directly assigns the accumulated result as the elastic deformation displacement increment of the current time step.

[0010] As a further aspect of the present invention: the construction process of the computational array is as follows: The dimension of the historical state vector is used as the total number of stages of the series delay unit. The output of each stage of the delay unit is sequentially connected to the input of the next stage of the delay unit, and the contact force input of the current time step is introduced into the input of the first stage of the delay unit. Assign an independent parallel weighting unit to each pole, and connect the input of each parallel weighting unit to the output of all delay units at each stage. The outputs of all parallel weighted units are connected to a summing node, and the output value of the summing node is directly used as the elastic deformation displacement increment of the current time step.

[0011] As a further aspect of the present invention: S4 specifically includes: The elastic deformation displacement increment is superimposed on the initial contact geometry coordinates of the escapement tooth shaft to obtain the corrected tooth tip spatial pose, and the penetration depth and normal vector in the contact force calculation are updated. Substitute the corrected penetration depth and normal vector back into the contact force constitutive relation, recalculate the contact force, and compare the relative difference between the two contact forces. If the relative difference exceeds the preset force convergence threshold, the newly calculated contact force is used as the contact force input for the current time step. The low-dimensional state update function is called repeatedly to update the geometry until the relative difference is less than or equal to the threshold. At this point, the converged contact force is output as the input for the next time step.

[0012] As a further aspect of the present invention: the recalculation of the contact force specifically includes: Based on the corrected penetration depth and normal vector, the pre-stored curvature radius data of the tooth tip and fork contact pair is read, and the curvature radius is dynamically updated according to the elastic deformation displacement increment. Substitute the updated radius of curvature into the Hertzian contact stiffness expression to calculate the instantaneous contact stiffness related to deformation; Multiply the instantaneous contact stiffness by the corrected penetration depth to obtain the recalculated contact force.

[0013] As a further aspect of the present invention: S5 specifically includes: The stress output values ​​corresponding to each convergence time step are read from the low-dimensional state update function and stored in a fixed-length first-in-first-out buffer in chronological order. Each time a new convergence time step is entered, the current stress output value is compared with the preset transient fracture criterion threshold, and the stress output values ​​of the previous two time steps in the buffer are traversed at the same time. If the current stress output value and the stress output values ​​of the first two time steps in the buffer are both greater than the transient fracture criterion threshold, a fracture warning signal is output, and the stress output value stored first in the buffer is removed.

[0014] The beneficial effects of this invention are: (1) By constructing a low-dimensional state update function that takes the historical state vector and the current contact force as input and directly outputs the elastic deformation displacement increment, and iteratively updates the contact geometry and contact force in each time step, real-time strong coupling of elastic deformation to the subsequent collision process is realized. This avoids the problem of stress concentration after deformation being ignored or excessive redundancy design caused by the misalignment of load transfer timing in traditional offline sequential coupling, making the prediction of transient stress distribution at the tooth tip and shoulder adjacent parts under drop impact closer to the physical reality, thereby reducing the number of physical drop tests required to verify fracture risk.

[0015] (2) A fixed-length first-in-first-out buffer is set in the stress output judgment stage, and a fracture warning signal is triggered only when the stress exceeds a preset threshold for three consecutive time steps. This effectively distinguishes between single impact noise and real continuous overload conditions, and improves the reliability of fracture risk judgment. Attached Figure Description

[0016] The invention will now be further described with reference to the accompanying drawings.

[0017] Figure 1 This is a flowchart of the digital modeling method for the operating state of the watch escapement gear shaft according to the present invention; Figure 2 This is a flowchart illustrating the calculation process of discrete recursive convolution coefficients in this invention. Detailed Implementation

[0018] 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.

[0019] Please see Figure 1 As shown, this invention provides a digital modeling method for the operating state of a watch escapement gear shaft, comprising the following steps: S1, collect time-domain response data of the escapement gear shaft under a typical impact load spectrum. The time-domain response data includes the displacement history and stress history of the tooth tip and the adjacent part of the shaft shoulder. S2, based on the displacement history, the dominant deformation basis vector is extracted by eigenorthogonal decomposition to construct a reduced-order subspace, and the impulse response analysis of each basis vector is performed according to the stress history to establish a recursive relationship of the recursive convolution state; S3 compiles the recursive relationship of the recursive convolution state into a low-dimensional state update function. The low-dimensional state update function takes the contact force input at the current time step and the historical state vector as input and outputs the elastic deformation displacement increment at the current time step. S4, during the drop impact time advance, after obtaining the contact force at each time step, the low-dimensional state update function is called to obtain the elastic deformation displacement increment. Then, the instantaneous contact geometry of the escapement tooth shaft is corrected based on the elastic deformation displacement increment, and the contact force is recalculated based on the corrected geometry until convergence. The converged contact force is used as the input for the next time step. S5. In each convergence time step, the stress output stored in the low-dimensional state update function is compared with the preset transient fracture criterion threshold. When the stress output exceeds the transient fracture criterion threshold for three consecutive time steps, a fracture warning signal is generated.

[0020] In S1, time-domain response data of the escapement gear shaft under a typical impact load spectrum is collected. The time-domain response data includes the displacement history and stress history of the tooth tip and the adjacent part of the shaft shoulder, specifically including: The escapement gear shaft to be tested was selected and installed on a miniature drop hammer impact test bench. The impact head of the test bench used a carbide contact with the same radius of curvature as the ruby ​​fork pad to simulate actual drop impact conditions.

[0021] A typical impact load spectrum is defined as a half-sine wave sequence with amplitudes ranging from 10g to 5000g and pulse widths ranging from 10 microseconds to 500 microseconds, with a total of 20 different amplitude-pulse width combinations. At each load point, the drop hammer is controlled to release from different heights, and a predetermined load is applied to the tooth tip and the adjacent part of the shaft shoulder through the impact head.

[0022] Displacement history was acquired using dual-channel laser displacement sensors: one channel aligned with the outer edge of the tooth tip, and the other with the outer cylindrical surface of the shoulder adjoint region, recording displacement values ​​at each sampling moment. Stress history was acquired using miniature resistance strain gauges attached to the surfaces of the tooth tip and the shoulder adjoint region. The strain gauge grid length was no greater than 0.2 mm. After being amplified by a dynamic strain gauge, strain values ​​were recorded at the same sampling frequency as the displacement, and then converted into stress values ​​according to Hooke's law. Each impact test was repeated three times, and the arithmetic mean was taken as the displacement and stress history at that load point, ultimately forming a raw data set containing 20 load points, each containing time-domain sequences of displacement and stress.

[0023] Please see Figure 2 As shown, in S2, the dominant deformation basis vectors are extracted through eigenorthogonal decomposition based on the displacement history to construct a reduced-order subspace. Furthermore, impulse response analysis is performed on each basis vector according to the stress history to establish a recursive convolution state relationship, specifically including: The displacement history is arranged chronologically, with the displacement value at each time sampling point forming a column vector. All column vectors from all time points are arranged side-by-side to form a snapshot matrix. This snapshot matrix is ​​then subjected to eigenorthogonal decomposition. Specifically, the product of the snapshot matrix and its transpose is calculated to obtain the covariance matrix. Then, all eigenvalues ​​and eigenvectors of this covariance matrix are solved. The eigenvectors are arranged in descending order of their corresponding eigenvalues, and the first few eigenvectors are selected as the initial basis vector set. Each initial basis vector represents a spatial distribution pattern of displacement.

[0024] The stress history is also arranged in chronological order to obtain a stress snapshot matrix. Each column vector of the stress snapshot matrix is ​​projected onto each initial basis vector. The projection is calculated by taking the dot product of the stress column vector and the initial basis vector to obtain the stress contribution factor corresponding to that initial basis vector. After traversing all initial basis vectors, a set of stress contribution factor values ​​is obtained. The initial basis vectors are then reordered in descending order of value, so that the basis vectors with larger stress contributions are placed first.

[0025] A cumulative stress contribution threshold of 95% is set. Starting from the first sorted basis vector, its stress contribution factor is accumulated sequentially. When the accumulated sum first reaches or exceeds 95%, the accumulation stops, and all accumulated basis vectors are selected as dominant deformation basis vectors. Unselected basis vectors and their corresponding displacement modes are discarded, thus realizing the construction of the reduced-order subspace.

[0026] For each dominant deformation basis vector, its corresponding impulse response function is constructed. The construction method is as follows: the half-sine pulse in the typical impact load spectrum from step one is used as the input excitation, and the projected component in the displacement history along the corresponding basis vector direction is used as the output response. The unit impulse response sequence of the basis vector is calculated through numerical deconvolution. Then, based on the impulse response functions of all dominant deformation basis vectors, a recursive convolution state relationship is prepared.

[0027] Perform a Laplace transform on the impulse response function corresponding to each dominant deformable basis vector using the standard numerical Laplace integral method. Set the lower bound to zero and the upper bound to the total duration of the impulse response sequence. This yields a transfer function in the complex frequency domain, which is a rational fraction with complex frequencies as independent variables. Perform this transform on each dominant deformable basis vector to obtain a set of complex frequency domain transfer functions.

[0028] For each complex frequency domain transfer function, a reduced-order rational approximation is performed using a continued fraction expansion. Specifically, the transfer function is expressed as a continued fraction, and constant terms and first-order fractions are extracted layer by layer, truncated to a predetermined number of expansion layers (8 to 12 layers). This yields a set of low-order poles and corresponding residues. The poles are the complex frequencies when the denominators of the fractions in the continued fraction expansion are zero, and the residues are the constant coefficients in the numerators of the fractions. Each transfer function corresponds to a set of poles and residues, and the sets of poles and residues of all dominant deformable basis vectors together constitute the parameters of the reduced-order model.

[0029] Each set of poles and residues is inversely transformed from the continuous time domain back to the discrete time domain, forming discrete recursive convolution coefficients. Before the inverse transformation, the discrete time step is determined, which is consistent with the time step of the drop impact simulation and is set to 1 microsecond. The purpose of the inverse transformation is to establish a recursive relationship from the historical state vector to the current elastic deformation displacement increment, where the historical state vector consists of the contact force input values ​​of several past time steps, and its dimension is determined by the number of poles.

[0030] For each pair of poles and residues, a bilinear transformation is performed to map the continuous-time poles and residues to discrete-time poles and residues. The specific operation of the bilinear transformation is as follows: multiply the continuous pole by half the discrete-time step, and then perform an algebraic transformation to obtain the discrete pole; the discrete residue is obtained by multiplying the continuous residue by a scaling factor related to the discrete pole. This operation is repeated for all poles and residues to obtain the corresponding discrete poles and residues.

[0031] Using each discrete residue as a weighting coefficient, the calculation process for the discrete recursive convolution coefficient is constructed as follows: First, the components of the historical state vector are arranged in reverse chronological order from the current step forward, with the most recent component at the front and the earliest component at the back. Then, each component in the reverse order is multiplied by its corresponding discrete residue, and all multiplications are summed up. The sum is then multiplied by the sampling interval of the unit impulse response, which is 1 microsecond. The sums of the above accumulations corresponding to all discrete poles are then superimposed to obtain the discrete recursive convolution coefficient between the elastic deformation displacement increment of the current time step and the historical state vector.

[0032] In S3, the recursive convolutional state recursion is compiled into a low-dimensional state update function. This function takes the contact force input at the current time step and the historical state vector as input, and outputs the elastic deformation displacement increment at the current time step, specifically including: The recursive convolutional state recurrence relation is compiled into a low-dimensional state update function. Before compilation, the number of poles in the recurrence relation is determined, which directly determines the dimension of the historical state vector, i.e., the number of contact force input values ​​from past time steps that need to be stored in the historical state vector. For example, if the number of poles is 20, the dimension of the historical state vector is set to 20. Based on this dimension, a computational array composed of serial delay units and parallel weighted units is constructed. This array is used to quickly calculate the elastic deformation displacement increment in each time step.

[0033] The specific process of constructing the computation array is as follows: The dimension of the historical state vector is used as the total number of stages of the serial delay units, that is, the number of delay units is equal to the dimension of the historical state vector. The output of each stage of delay units is sequentially connected to the input of the next stage of delay units, forming a delay chain. At the same time, the contact force input value of the current time step is introduced into the input of the first stage of delay units, and this value serves as the excitation signal for the entire computation array.

[0034] Each pole is assigned an independent parallel weighting unit, meaning the number of parallel weighting units is equal to the number of poles. The input of each parallel weighting unit is connected to the output of all delay units at each stage, meaning each parallel weighting unit can simultaneously read the currently stored state component values ​​of all delay units from the first stage to the last stage.

[0035] The outputs of each parallel weighted unit are connected to a summing node. This summing node is responsible for accumulating the output values ​​of all parallel weighted units, and the accumulated result is directly output as the elastic deformation displacement increment for the current time step. At this point, the computation array is complete, which realizes the mapping from the contact force input and historical state vector of the current time step to the elastic deformation displacement increment.

[0036] In each time step of the drop impact simulation, the contact force input of the current time step is used as the excitation value and fed into the input of the first-level delay unit of the computation array. Simultaneously, each delay unit sequentially passes the historical state components from the previous moment: the first-level delay unit passes its original value to the second-level delay unit, the second-level delay unit passes its original value to the third-level delay unit, and so on, with the last-level delay unit discarding its original value. After this transmission, the first-level delay unit stores the contact force input of the current time step, while the remaining delay units store the contact force inputs from previous time steps, i.e., the historical state vector.

[0037] Each parallel weighted unit reads the state components output by all delay units connected to its input and multiplies and accumulates these state components with pre-stored recursive convolution coefficients. Each parallel weighted unit corresponds to a set of coefficients, the number of which equals the total number of delay units. The specific operation of multiplication and accumulation is as follows: the state component output by each delay unit is multiplied by the coefficient corresponding to that level of the parallel weighted unit, and then the product results of all levels are added together. After each parallel weighted unit completes multiplication and accumulation, it outputs its result to the summation node. The summation node accumulates the output values ​​of all parallel weighted units, and the accumulated result is directly assigned to the elastic deformation displacement increment of the current time step.

[0038] In S4, during the drop impact time progression, after obtaining the contact force at each time step, the low-dimensional state update function is called to obtain the elastic deformation displacement increment. Then, the instantaneous contact geometry of the escapement gear shaft is corrected based on the elastic deformation displacement increment, and the contact force is recalculated based on the corrected geometry until convergence. The converged contact force is used as the input for the next time step, specifically including: In each time step of the drop impact time progression, after obtaining the contact force at the current time step, the low-dimensional state update function is immediately invoked. This function outputs the elastic deformation displacement increment at the current time step. This displacement increment is superimposed onto the initial contact geometry coordinates of the escapement tooth shaft to obtain the corrected tooth tip spatial pose. The initial contact geometry coordinates are obtained in advance through three-dimensional measurement, including the spatial coordinates of the tooth tip sphere center and the radius of the tooth tip sphere. Based on the corrected tooth tip spatial pose, the penetration depth between the tooth tip and the pallet is recalculated. The penetration depth is calculated as the difference between the normal direction of the pallet plane and the directed distance from the tooth tip sphere center to that plane, minus the radius of the tooth tip sphere. If the difference is positive, it indicates that penetration has occurred. Simultaneously, the normal vector is updated. This vector is taken as the unit normal of the pallet plane at the contact point, with its direction pointing from the pallet to the tooth tip.

[0039] Based on the corrected penetration depth and normal vector, the pre-stored radius of curvature data of the tooth tip and pallet contact pair is read. The initial radius of curvature of the tooth tip is denoted as... The contact surface of the fork bearing is considered a plane with an infinite radius of curvature. The radius of curvature of the tooth tip is dynamically updated based on the elastic deformation displacement increment, and the update rule adopts a linear relationship: ,in This represents the elastic deformation displacement increment (in millimeters). This is the radius of curvature variation coefficient, which is obtained through prior micro-indentation tests on the same batch of gear shaft materials, with a typical value range of 0.2 to 0.4. Updated equivalent radius of curvature. Calculated by the following formula: Due to the radius of curvature of the fork tile If it is infinity, then equal In the formula, The unit is millimeters, representing the combined radius of curvature of the contact pair.

[0040] The updated equivalent radius of curvature Substituting the Hertzian contact stiffness expression, the instantaneous contact force related to deformation is calculated. For point contact between the tooth tip and the pallet, the contact force is... With penetration depth The relationship between them satisfies Hertzian elastic contact theory, and is as follows: In the formula, Contact force, measured in Newtons; The equivalent elastic modulus, in Pascals, is calculated using the following formula: ,in and These are the Young's moduli of the gear shaft material (such as silicon or steel) and the fork pad material (ruby), respectively. and This represents the Poisson's ratio of the corresponding material. The first calculated penetration depth is in millimeters; it is calculated by first taking the square root of the penetration depth and then multiplying it by the penetration depth itself. This formula directly outputs the recalculated contact force.

[0041] After obtaining the newly calculated contact force Then, it is compared with the contact force used in the previous iteration. Compare the two. When comparing, calculate the relative difference between the two, which is defined as: [The value of the relative difference is missing from the original text]. and The absolute value of the difference, divided by and The larger value in the range.

[0042] like and If both are zero, the relative difference is set to zero. A force convergence threshold is preset, which is 0.01, or 1%. If the relative difference is less than or equal to 0.01, the contact force is considered to have converged; if the relative difference is greater than 0.01, the iterative step is executed.

[0043] When the relative difference exceeds the force convergence threshold, the newly calculated contact force will be... As the contact force input for the current time step, the low-dimensional state update function is invoked again. This function, based on the new contact force input and the currently saved historical state vector, outputs a new elastic deformation displacement increment. Subsequently, the tooth tip spatial pose is corrected again based on this new displacement increment, the penetration depth and normal vector are updated, and the contact force is recalculated according to the methods in the second and third sections, resulting in an updated... .

[0044] The iterative process is repeated, with each iteration using the latest calculated contact force as input and calling a low-dimensional state update function to update the elastic deformation displacement increment, thereby updating the geometry and contact force. Iteration stops when the relative difference calculated after a certain iteration is less than or equal to 0.01, and the converged contact force at this point is output as the input for the next time step. Simultaneously, the elastic deformation displacement increment, the corrected penetration depth, and the tooth tip spatial pose at this convergence point are used as the final state for the current time step, for fracture early warning analysis. The entire iterative process converges within 2 to 4 iterations.

[0045] In S5, at each convergence time step, the stress output stored within the low-dimensional state update function is compared with a preset transient fracture criterion threshold. When the stress output exceeds the transient fracture criterion threshold for three consecutive time steps, a fracture warning signal is generated, specifically including: Stress output value acquisition and buffer construction: At the end of each convergence time step, the stress output value corresponding to that time step is read from the low-dimensional state update function. The specific method for acquiring this stress output value is as follows: In the computation array, each parallel weighted unit, in addition to outputting the elastic deformation displacement increment component to the summing node, also generates an additional stress contribution component. This stress contribution component is obtained by multiplying and summing the components of the historical state vector read by the parallel weighted unit with its corresponding other set of stress recursive convolution coefficients. The stress recursive convolution coefficients are calculated in parallel by the impulse response analysis of the stress history.

[0046] The stress contribution components of all parallel weighted elements are summed at another summing node to obtain the stress output value at the current time step, in megapascals.

[0047] A transient fracture criterion threshold is preset. This threshold is determined by taking the fracture strength value of the gear shaft material (such as silicon or hardened steel) under dynamic impact. This value is measured through a split Hopkinson bar test at a strain rate of 1000 cycles per second, and then multiplied by a safety factor of 0.8 to obtain the final threshold value. For silicon, this threshold is typically 960 MPa; for hardened steel, it is 1200 MPa.

[0048] Simultaneously, a fixed-length first-in-first-out (FIFO) buffer is constructed to store the stress output values ​​for three consecutive convergence time steps. The buffer has a fixed capacity of three storage units, with each unit storing one stress output value in chronological order. Initially, all three units of the buffer are filled with zero values.

[0049] The stress output value comparison operation is as follows: Upon entering a new convergence time step and acquiring the current stress output value, the current stress output value is first compared with a preset transient fracture criterion threshold to determine if it exceeds the threshold. Subsequently, the stress output values ​​from the previous two time steps stored in the buffer are traversed—that is, the values ​​in the earliest and second-earliest stored cells. During traversal, these two values ​​are compared sequentially with the same transient fracture criterion threshold to determine if they exceed it. All the above comparison operations are performed sequentially within the same time step and do not involve parallel computation. If the buffer is not full in the first two time steps (e.g., the first two time steps at the beginning of the simulation), the stress output value corresponding to the missing time step is considered 0, failing to meet the condition of exceeding the threshold.

[0050] Output of fracture warning signal and update of buffer: A fracture warning signal is output only when the following three conditions are met simultaneously: the stress output value of the current time step is greater than the preset transient fracture criterion threshold, the stress output value of the first time step stored in the buffer is greater than the threshold, and the stress output value of the second-first time step stored in the buffer is also greater than the threshold. The fracture warning signal is in the form of a rising edge pulse with a duration of one time step, which can be captured by an external recording device or simulation control program. Regardless of whether a warning signal is output, after the processing of the current time step is completed, the stress output value stored first in the buffer is removed from the buffer, and the stress output value of the current time step is stored at the end of the buffer, completing the rolling update of the buffer.

[0051] Thus, when the next convergence time step arrives, the buffer always retains the stress output values ​​of the most recent three time steps, thereby supporting accurate judgment when the threshold is exceeded for three consecutive time steps. This mechanism can effectively avoid false alarms caused by single stress spike noise, while ensuring timely fracture warning under real continuous overload conditions.

[0052] The working principle of this invention is as follows: First, the displacement and stress histories of the escapement gear shaft under a typical impact load spectrum are acquired. Then, based on the displacement histories, intrinsic orthogonal decomposition is performed to extract the dominant deformation basis vectors. Based on the stress histories, impulse response analysis is performed on each basis vector. Discrete recursive convolution coefficients are established through Laplace domain continuous fraction expansion and bilinear transformation to form a low-dimensional state update function that takes the historical state vector and the current contact force as inputs and outputs the elastic deformation displacement increment. During the drop impact time progression, the current contact force is obtained at each time step. The function is called to obtain the displacement increment and correct the tooth tip spatial pose. The penetration depth and normal vector are updated. Then, the contact force is recalculated by substituting the Hertzian contact stiffness expression containing the dynamic curvature radius. After iteration until convergence, the contact force is used as the input for the next time step. At the same time, in each convergence time step, the stress output is read from inside the function and compared with the preset transient fracture criterion threshold. A fracture warning signal is generated when the stress in three consecutive time steps exceeds the threshold using a first-in-first-out buffer.

[0053] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. Method for digital modeling of the running state of a timepiece escapement pinion, characterized in that, Includes the following steps: S1, collect time-domain response data of the escapement gear shaft under a typical impact load spectrum. The time-domain response data includes the displacement history and stress history of the tooth tip and the adjacent part of the shaft shoulder. S2, based on the displacement history, the dominant deformation basis vector is extracted by eigenorthogonal decomposition to construct a reduced-order subspace, and the impulse response analysis of each basis vector is performed according to the stress history to establish a recursive relationship of the recursive convolution state; S3 compiles the recursive relationship of the recursive convolution state into a low-dimensional state update function. The low-dimensional state update function takes the contact force input at the current time step and the historical state vector as input and outputs the elastic deformation displacement increment at the current time step. S4, during the drop impact time advance, after obtaining the contact force at each time step, the low-dimensional state update function is called to obtain the elastic deformation displacement increment. Then, the instantaneous contact geometry of the escapement tooth shaft is corrected based on the elastic deformation displacement increment, and the contact force is recalculated based on the corrected geometry until convergence. The converged contact force is used as the input for the next time step. S5. In each convergence time step, the stress output stored in the low-dimensional state update function is compared with the preset transient fracture criterion threshold. When the stress output exceeds the transient fracture criterion threshold for three or more consecutive time steps, a fracture warning signal is generated.

2. The method of digital modeling of the running state of an escapement pinion of a timepiece according to claim 1, characterized in that, S2 specifically includes: The snapshot matrix formed by the displacement history is subjected to eigenorthogonal decomposition to obtain the initial basis vector set; The stress history is projected onto each initial basis vector, the stress contribution factor corresponding to each basis vector is calculated, and the stress contribution factors are sorted from high to low. The first few basis vectors whose cumulative stress contribution factors reach a set threshold are selected as the dominant deformation basis vectors, and the impulse response function is fitted in the Laplace domain based on the dominant deformation basis vectors to construct the recursive relationship of the recursive convolution state.

3. The method of digital modeling of the running state of an escapement pinion of a timepiece according to claim 2, characterized in that, The method of fitting the impulse response function to the dominant deformable basis vector in the Laplace domain and constructing a recursive relationship for the recursive convolution state specifically includes: Transform the impulse response functions corresponding to each dominant deformation basis vector to the Laplace domain to obtain a set of complex frequency domain transfer functions. We use a continued fraction expansion to perform a reduced-order rational approximation of each complex frequency domain transfer function to obtain the set of low-order poles and residues. By inversely transforming the poles and residues back to the time domain, discrete recursive convolution coefficients are formed with the historical state vector as input and the current elastic deformation displacement increment as output.

4. The method of claim 3, wherein, The calculation process of the discrete recursive convolution coefficients specifically includes: Perform a bilinear transformation on each pair of poles and residues to map the poles and residues in the continuous time domain to the discrete poles and residues in the discrete time domain. Using discrete residues as weighting coefficients, the components of the historical state vector are multiplied and accumulated one by one with the weighting coefficients in reverse chronological order, and then multiplied by the sampling interval of the unit impulse response. By summing the product of all poles, we obtain the discrete recursive convolution coefficient between the current elastic deformation displacement increment and the historical state vector.

5. The digital modeling method for the operating state of the escapement gear shaft in a clock as described in claim 1, characterized in that, S3 specifically includes: The dimension of the historical state vector is determined by the number of poles in the recursive relationship of the recursive convolution state, and an operational array consisting of serial delay units and parallel weighting units is constructed based on the dimension. The contact force input at the current time step is used as the excitation value input to the first delay unit of the operation array, while each delay unit sequentially transmits the historical state components of the previous moment. Each parallel weighted unit reads the state component output by the corresponding delay unit and multiplies and accumulates it with the pre-stored recursive convolution coefficients, and directly assigns the accumulated result as the elastic deformation displacement increment of the current time step.

6. The digital modeling method for the operating state of the watch escapement gear shaft according to claim 5, characterized in that, The construction process of the computing array is as follows: The dimension of the historical state vector is used as the total number of stages of the series delay unit. The output of each stage of the delay unit is sequentially connected to the input of the next stage of the delay unit, and the contact force input of the current time step is introduced into the input of the first stage of the delay unit. Assign an independent parallel weighting unit to each pole, and connect the input of each parallel weighting unit to the output of all delay units at each stage. The outputs of all parallel weighted units are connected to a summing node, and the output value of the summing node is directly used as the elastic deformation displacement increment of the current time step.

7. The digital modeling method for the operating state of the escapement gear shaft in a clock as described in claim 1, characterized in that, S4 specifically includes: The elastic deformation displacement increment is superimposed on the initial contact geometry coordinates of the escapement tooth shaft to obtain the corrected tooth tip spatial pose, and the penetration depth and normal vector in the contact force calculation are updated. Substitute the corrected penetration depth and normal vector back into the contact force constitutive relation, recalculate the contact force, and compare the relative difference between the two contact forces. If the relative difference exceeds the preset force convergence threshold, the newly calculated contact force is used as the contact force input for the current time step. The low-dimensional state update function is called repeatedly to update the geometry until the relative difference is less than or equal to the threshold. At this point, the converged contact force is output as the input for the next time step.

8. The digital modeling method for the operating state of the escapement gear shaft in a clock as described in claim 7, characterized in that, The recalculation of contact force specifically includes: Based on the corrected penetration depth and normal vector, the pre-stored curvature radius data of the tooth tip and fork contact pair is read, and the curvature radius is dynamically updated according to the elastic deformation displacement increment. Substituting the updated radius of curvature into the Hertzian contact stiffness expression, the instantaneous contact force related to deformation is calculated. The formula for calculating the instantaneous contact force is as follows: ;in, For instantaneous contact force, For the equivalent elastic modulus, The updated equivalent radius of curvature. This is the corrected penetration depth.

9. The digital modeling method for the operating state of the escapement gear shaft in a clock as described in claim 1, characterized in that, S5 specifically includes: The stress output values ​​corresponding to each convergence time step are read from the low-dimensional state update function and stored in a fixed-length first-in-first-out buffer in chronological order. Each time a new convergence time step is entered, the current stress output value is compared with the preset transient fracture criterion threshold, and the stress output values ​​of the previous two time steps in the buffer are traversed at the same time. If the current stress output value and the stress output values ​​of the first two time steps in the buffer are both greater than the transient fracture criterion threshold, a fracture warning signal is output, and the stress output value stored first in the buffer is removed.

Citation Information

Patent Citations

  • Clock movement with dead second indicator

    CH698440B1

  • Tunnel rockburst numerical simulation method based on node continuous-discontinuous deformation analysis

    CN121302785A