A motorcycle front fork shock absorber offline detection system and method

Abstract

This invention discloses a system and method for off-line inspection of motorcycle front fork shock absorbers, relating to the field of motorcycle shock absorber inspection technology. It collects real-vehicle operating condition data to establish a continuous weaving field for operating condition segments; establishes a behavioral evolution network and extracts behavioral migration features and force flow memory maps to characterize the legacy effects of historical loads; identifies stress-bearing nodes, stress accumulation areas, and stress fracture areas, as well as inertial stagnation chains, and predicts future expansion trends of historical behavior; identifies risk areas such as motion lag, damping drift, lateral instability, and abnormal rebound to form a performance deviation risk map; and establishes an off-line judgment index by combining structural risk occupancy, spatial distortion index, inertial continuity, and the comprehensive value of the risk map, outputting the risk map, defect tracing results, and off-line inspection conclusions. This invention identifies legacy effects of historical operating conditions and potential performance deviation risks, enabling risk prediction of front fork shock absorber factory quality and improving the accuracy and reliability of off-line inspection.

Description

Technical Field

[0001] This invention relates to the field of motorcycle shock absorber testing technology, specifically to a system and method for detecting the off-line performance of motorcycle front fork shock absorbers. Background Technology

[0002] Motorcycle front fork shock absorbers are key components affecting vehicle handling stability, ride comfort, and riding safety. Their performance directly determines the contact capability between the front wheel and the road surface, as well as the overall dynamic response characteristics of the vehicle. After manufacturing, front fork shock absorbers typically undergo off-line testing to confirm whether the products meet factory requirements.

[0003] Current methods for inspecting front fork shock absorbers before they leave the factory primarily employ stroke testing, damping force testing, rebound testing, sealing testing, and bench durability testing. These methods assess product quality by testing static parameters or dynamic parameters under a single operating condition. While these methods can detect obvious structural defects and performance anomalies, the evaluation process is typically based on independent test items, making it difficult to reflect the cumulative effects of historical loads on the front fork shock absorber under complex real-world riding conditions and continuous loads. Furthermore, they lack the ability to effectively analyze the transmission relationships between different operating conditions, the residual effects of historical loads, and the potential performance deviation risks that may arise during long-term operation. This can easily lead to some products with potential failure risks passing factory inspection, thus affecting the long-term reliability of the vehicle.

[0004] Therefore, how to establish a post-construction detection system and method that can identify and predict the potential performance deviation risk of front fork shock absorbers by combining the actual working condition evolution process, historical load transfer relationship and long-term structural influence accumulation characteristics has become a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing front fork shock absorber off-line testing processes, which only focus on single test results, cannot identify the effects of historical loads, and are difficult to predict potential performance deviation risks. Therefore, this invention proposes an off-line testing system and method for motorcycle front fork shock absorbers.

[0006] The objective of this invention can be achieved through the following technical solution: a vehicle front fork shock absorber off-line testing system, comprising a real vehicle operating condition segment establishment module, a motion behavior acquisition module, a historical impact analysis module, a stress bearing analysis module, and an inertial retention prediction module;

[0007] The real vehicle working condition segment establishment module collects multi-dimensional dynamic data and segments the working condition segment, calculates the normalized residual amount, inheritance strength and winding coefficient, and establishes a continuous weaving field for the working condition segment.

[0008] The motion behavior acquisition module receives the weaving field, collects four types of dynamic response combinations into behavior vectors, establishes behavior activation potential and competition intensity, divides the receiving area and competition area through no-load calibration, and statistically analyzes migration flux and return index, integrating them into a behavior evolution network;

[0009] The historical impact analysis module receives the behavioral evolution network, calculates the behavioral residual potential, convergence intensity and propagation gain, calculates the conservation degree, filters the force flow memory chains through path search and memory intensity, calculates the cross-link intensity between chains, and establishes the force flow memory map.

[0010] The stress bearing analysis module receives the force flow memory spectrum, calculates the connection potential to establish a global connection matrix, calculates the bearing contribution, diffusion index and stress bearing strength to determine the bearing node, determine the accumulation area and fracture area, establishes the structural space torsion field and calculates the torsion index.

[0011] The inertial retention prediction module receives data from the above modules, calculates the inertial residual factor, retention potential, circulation index and accumulation potential, determines the inertial retention chain, calculates the expansion potential and risk occupancy, maps the risk area and classifies the level, establishes a performance deviation risk map, traces back to the source, calculates the weighted sum of the four indicators to obtain the lower limit judgment index, compares it with the threshold and outputs the qualified or unqualified conclusion, risk map and source tracing results.

[0012] In a preferred embodiment of the present invention, the real vehicle operating condition segment establishment module collects multi-dimensional dynamic data by means of brake pressure sensors, wheel speed sensors, steering angle sensors, steering torque sensors, fork displacement sensors, vehicle acceleration sensors, vehicle posture sensors, and fork end acceleration sensors arranged on the target vehicle model; synchronously records the data according to a unified time reference to form a continuous riding data stream; and performs operating condition identification and time window segmentation on the continuous riding data stream based on braking deceleration threshold, steering angle change rate threshold, and road excitation amplitude threshold to obtain a set of operating condition segments and establish the motion state space corresponding to each operating condition segment.

[0013] In a preferred embodiment of the present invention, the motion state space includes the fork compression stroke, compression speed, compression acceleration, damping force, spring force, internal oil pressure, and fork internal temperature; the actual vehicle working condition segment establishment module establishes the working condition segment residual amount based on the actual motion state vector and static equilibrium state vector at the end of the working condition segment; the normalized residual amount is obtained by normalizing the full-scale values ​​of the sensors corresponding to each physical dimension; the inheritance strength is calculated based on the overlap between the normalized residual amount and the starting motion state vector of the subsequent working condition segment, and the working condition segment winding coefficient is established based on each inheritance strength.

[0014] The real vehicle working condition segment establishment module establishes a continuous weaving field for working condition segments based on the winding coefficients of all working condition segments. The continuous weaving field for working condition segments uses working condition segments as nodes and the winding coefficients of working condition segments as edge weights to record the cross-working condition inheritance relationship of historical loads between different working condition segments and sends it to the motion behavior acquisition module.

[0015] In a preferred embodiment of the present invention, the motion behavior acquisition module acquires compression trajectory response, rebound trajectory response, lateral drift response, and structural micro-vibration response, and establishes behavior vectors; establishes behavior activation potential based on historical working condition residuals and corresponding attenuation periods; establishes behavior competition matrix based on behavior activation potential and distance relationship between behavior vectors; determines acceptance threshold and competition threshold based on no-load calibration results, and divides behavior acceptance zone and behavior competition zone according to competition intensity.

[0016] In a preferred embodiment of the present invention, the motion behavior acquisition module establishes a behavior migration flux based on behavior activation potential, behavior change amount, and competition intensity; establishes a behavior return flow index based on the total migration flux entering the behavior region and the total migration flux leaving the behavior region; establishes a behavior evolution network based on the behavior receiving area, behavior competition area, behavior migration flux, and behavior return flow index; when the behavior return flow index is continuously greater than one and the duration exceeds the corresponding decay period, the corresponding behavior region is marked as a stagnation area.

[0017] In a preferred embodiment of the present invention, the historical impact analysis module establishes behavioral residual potential based on behavioral migration flux and behavioral backflow index; establishes behavioral convergence intensity based on behavioral residual potential and migration flux; establishes behavioral propagation gain based on behavioral convergence intensity and behavioral backflow index; establishes behavioral impact conservation degree based on the total amount of all behavioral residual potential and the total amount of all behavioral propagation gain; and determines the release state, conservation state or abnormal amplification state of historical behavior based on the behavioral impact conservation degree.

[0018] The historical impact analysis module uses a depth-first search approach to search for paths in the behavioral evolution network; it establishes path memory strength based on behavioral migration flux, behavioral residual potential, and path length; it selects force flow memory chains based on path memory strength; it establishes memory chain cross strength based on the number of common behavioral state nodes between different force flow memory chains and path memory strength; and it establishes a force flow memory graph based on force flow memory chains, path memory strength, and memory chain cross strength.

[0019] In a preferred embodiment of the present invention, the stress bearing analysis module establishes a force flow continuity potential based on behavioral residual potential, memory chain cross strength, and memory chain average time center; establishes a global continuity matrix based on all force flow continuity potentials; establishes a bearing contribution based on behavioral residual potential, force flow continuity potential, and path memory strength; establishes a diffusion index based on bearing contribution and behavioral migration flux; establishes stress bearing strength based on bearing contribution, diffusion index, and behavioral backflow index; and identifies stress bearing nodes based on stress bearing strength.

[0020] In a preferred embodiment of the present invention, the stress bearing analysis module establishes an accumulation index based on the number of force flow memory chains passing through the behavioral region, the average stress bearing strength of the region, and the average return flow index of the region, and identifies stress accumulation zones based on the accumulation index; identifies stress fracture zones based on the memory decay rate and the continuous decay length; establishes a structural spatial torsion field based on the stress bearing nodes, stress accumulation zones, and stress fracture zones, and establishes a structural spatial torsion index based on the accumulation index, stress bearing strength, and continuous decay length.

[0021] In a preferred embodiment of the present invention, the inertial retention prediction module establishes an inertial retention factor based on behavioral residual potential, behavioral backflow index, and stress bearing strength; establishes an inertial retention potential based on the inertial retention factor, memory chain cross strength, and path memory strength; establishes a behavioral circulation index based on cyclic migration flux and total outward diffusion flux; establishes an inertial accumulation potential based on the inertial retention potential, behavioral circulation index, and accumulation index; identifies inertial retention chains based on the inertial accumulation potential; establishes structural risk occupancy based on future expansion potential, forming a performance deviation risk map; establishes a dropout judgment index based on structural risk occupancy, structural space distortion index, inertial continuity, and risk map, and outputs a risk map, defect tracing results, and dropout detection conclusions.

[0022] Another aspect of the present invention provides a method for detecting the off-line condition of a motorcycle front fork shock absorber, comprising the following steps:

[0023] Step 1: Collect braking, steering, road surface undulation and continuous impact data, segment them to form a set of working condition segments; calculate the normalized residual amount, inherited strength and winding coefficient, and establish a continuous weaving field for the working condition segments;

[0024] Step 2: Collect compression trajectory, rebound trajectory, lateral drift, and structural micro-vibration data to form behavior vectors; calculate behavior activation potential and competition intensity, and divide the behavior acceptance zone and behavior competition zone; statistically analyze migration flux and backflow index to establish a behavior evolution network;

[0025] Step 3: Calculate the behavioral residual potential, convergence strength, propagation gain, and conservation degree based on the behavioral evolution network; perform path search and calculate path memory strength to filter force flow memory chains; calculate the memory chain cross strength and establish a force flow memory map;

[0026] Step 4: Calculate the force flow continuity potential based on the force flow memory map and establish a global continuity matrix; calculate the bearing contribution, diffusion index and stress bearing strength, and identify stress bearing nodes; identify stress accumulation areas and stress fracture areas, establish a structural spatial torsion field and calculate the spatial torsion index;

[0027] Step 5: Calculate the inertial residual factor, inertial retention potential, behavioral cycle index, and inertial accumulation potential; identify inertial retention chains and deduce future expansion potential; calculate the structural risk occupancy, identify motion jamming, damping drift, lateral instability, and abnormal rebound risks, establish a performance deviation risk map, and complete defect source tracing.

[0028] Step Six: Weight and fuse the structural risk occupancy, spatial distortion index, inertial continuity, and risk map comprehensive value to obtain the delisting judgment index; output the risk map, source tracing results, and delisting detection conclusion based on the dynamically updated threshold.

[0029] Compared with the prior art, the beneficial effects of the present invention are:

[0030] 1. This invention establishes a continuous weaving field for working condition segments, a behavior evolution network, and a force flow memory map to perform correlation analysis on the load legacy effects formed by the front fork shock absorber under different historical working conditions. This enables continuous tracking of the inheritance relationship, behavior migration relationship, and historical influence propagation path between working conditions. It breaks through the limitation of traditional offline testing that only focuses on the results of a single test. It can identify potential anomalies that have accumulated in the front fork for a long time but have not yet appeared, and improve the ability of offline testing results to characterize the actual use state.

[0031] 2. This invention analyzes the process of historical influences being received, accumulated, fractured, and diffused within the fork by establishing a structural spatial torsion field and inertial retention prediction mechanism. It can identify risk areas of motion jamming, damping drift, lateral instability, and abnormal rebound, and achieve reverse tracing of the defect formation path. This allows potential performance deviation problems to be discovered before the product leaves the factory, thereby improving the reliability, safety, and quality control level of the fork shock absorber.

[0032] 3. This invention integrates structural risk occupancy, spatial distortion index, inertial continuity, and performance deviation risk map for analysis, and establishes a dynamically updated offline determination mechanism. This enables the test results to not only reflect the current performance status but also predict future performance evolution trends. It elevates the test from a single-parameter qualification determination to a comprehensive testing mode oriented towards full life cycle risk assessment, thereby improving the accuracy and engineering applicability of offline determination. Attached Figure Description

[0033] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0034] Figure 1 This is a schematic diagram of the principle of the present invention;

[0035] Figure 2 This is a flowchart illustrating the steps of the method of the present invention. Detailed Implementation

[0036] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. 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.

[0037] Please see Figure 1 As shown, the present invention provides a system for detecting the off-line operation of a motorcycle front fork shock absorber, comprising: a real vehicle operating condition segment establishment module, a motion behavior acquisition module, a historical impact analysis module, a stress bearing analysis module, and an inertia retention prediction module.

[0038] The real-vehicle condition segment creation module collects multi-dimensional dynamic data during actual riding using various types of sensors deployed on the target vehicle model. This includes braking behavior data, steering behavior data, road surface undulation data, vehicle roll data, and continuous impact data. Specifically, braking behavior data is acquired through brake pressure and wheel speed sensors; steering behavior data is acquired through steering angle and steering torque sensors; road surface undulation data is acquired through fork displacement and vehicle acceleration sensors; vehicle roll data is acquired through an IMU (Induction Unit); and continuous impact data is acquired through a fork-end acceleration sensor. All sensor data is synchronously collected and recorded using a unified time base, forming a complete continuous riding data stream.

[0039] Next, following the time sequence, based on preset operating condition segment identification rules such as braking deceleration threshold, steering angle change rate threshold, and road excitation amplitude threshold, individual operating condition segments are identified and segmented from the continuous riding data, forming a set of operating condition segments: G = {g1, g2, ..., g i ,…,g n}, where G is the set of operating condition segments; g iThis is the i-th operating condition segment. Each operating condition segment contains complete time window data for the start time, duration, and end time, as well as the recorded values ​​of all sensors within that time period. Specifically, the braking behavior segment is marked as starting when the brake master cylinder pressure is ≥0.5MPa and the wheel speed deceleration is ≥1.5m / s², and ending when the brake pressure drops to <0.3MPa and the deceleration is <0.5m / s² for 50ms; the steering behavior segment is marked as starting when the steering angle change rate is ≥20° / s and the steering torque is ≥2N·m, and ending when the angle change rate is <5° / s and the torque is <1N·m for 100ms; the road undulation segment is marked as starting when the fork displacement amplitude exceeds ±5mm or the vehicle vertical acceleration exceeds ±0.3g, and ending when the displacement amplitude drops to within ±2mm and the acceleration is <±0.1g for 80ms; the continuous impact segment is marked as a continuous impact segment when ≥3 pulses with peak values ​​exceeding ±5g occur within a 100ms window; if multiple rules are triggered, they are merged into a composite segment according to time overlap and the composite type is marked.

[0040] For the working condition segment g i Establish the corresponding motion state space: M i = {m1,m2,…,m k}, where M i For any working condition segment g i The corresponding set of motion states; m k This represents the end-effector motion state at the end of the working condition segment. The motion state space specifically includes the following physical quantities: fork compression stroke, compression velocity, compression acceleration, damping force, spring force, internal hydraulic pressure, and fork internal temperature. Each motion state is a complete numerical vector of the above multidimensional physical quantities at the corresponding moment.

[0041] Due to the nonlinear mechanical characteristics of the fork, such as internal frictional damping, secondary spring effects (e.g., air spring assistance), and delayed structural deformation recovery, not all motion states immediately return to their initial equilibrium positions at the end of a work segment. A work segment residual quantity L is defined. i Before calculation, dimensionless processing is performed: for each dimension d, its sensor full-scale value R is used. d Normalization: L i = M i end - M i eq Among them, L i For the working condition segment g i The legacy vector; M i end M is the actual motion state vector at the end of this work segment; i eqThis represents the static equilibrium state vector of the front fork structure under no external force. Before calculation, each physical dimension is dimensionless, and each dimension d uses the full-scale value R of the sensor. d Normalization, i.e., L id = (M i end d - M i eq d ) / R d The full-scale values ​​for each dimension are as follows: compression stroke ±100mm, compression speed ±2000mm / s, compression acceleration ±50g, damping force ±5000N, spring force ±3000N, internal oil pressure ±10MPa, and internal temperature ±80℃; all subsequent values ​​related to L... i Summation and intersection operations are both based on normalized values.

[0042] Among them, M i end During real-vehicle data acquisition, each sensor operates continuously at a synchronous sampling frequency of no less than 1kHz. At the trigger moment, such as when braking deceleration falls below a preset threshold or steering angular velocity returns to zero, the instantaneous values ​​of all sensor channels at that moment are immediately frozen, forming a synchronous snapshot of multidimensional physical quantities. This synchronous snapshot includes the following components arranged by physical dimensions: the compression stroke values ​​measured by the relative displacement sensors at the upper and lower ends of the fork; the compression velocity and acceleration values ​​obtained by differentiating displacement with respect to time; the real-time damping force measured by the strain gauge force sensor mounted on the damper piston rod; the spring support force measured by the force sensor mounted on the fork spring seat; and the internal oil pressure and internal temperature measured by the miniature pressure transmitter and thermocouple embedded inside the damping cavity. It should be noted that all components have positive and negative signs to represent physical meanings such as the direction of compression stroke and the direction of force application.

[0043] M i eq A fixed reference vector was pre-established through a static calibration experiment on a bench. The calibration process was conducted under standard environmental conditions: an ambient temperature of 20±2℃, with the fork assembly placed vertically and freely without external load. Under these conditions, the fork spring was at its designed free length, the damper piston was at its mid-stroke static equilibrium position, there was no relative motion tendency between the inner and outer tubes, and the frictional force was at its static equilibrium minimum. Sensor readings for all the aforementioned dimensions were recorded, and these readings were used as their respective zero-point reference values ​​to form the static equilibrium state vector M. i eqTo further prevent zero-point drift of the sensor from causing calculation errors during long-term use, the system automatically repeats this zero-point calibration process every fixed cycling distance (e.g., every 100 kilometers) or at fixed time intervals (e.g., during the self-test phase before each ride), updating the M... i eq The reference values ​​for each component.

[0044] Next, to quantify the impact of the previous operating condition segment on subsequent operating condition segments, the module establishes the operating condition segment inheritance strength: T ij = L i ∩ M j start / L i , among which, T ij For the working condition segment g i For the working condition segment g j Inheritance strength (value range 0~1); M j start For the working condition segment g j The actual motion state vector at the initial moment; L i ∩ M j start This represents the component of the remaining quantity from the previous working condition segment that is inherited by the starting state of the next working condition segment—the specific calculation method is as follows: for the remaining quantity L i With initial state M j start The corresponding physical dimensions are compared, and the smaller value is taken as the inheritance amount for that dimension. Then, the sum is applied to all dimensions to obtain the total inherited legacy state. Specifically, for each dimension d, if L... id With M j start d If the signs are the same, then the inherited quantity is min(|L). id |, |M j start d |) and retain the sign; if the signs are opposite, the inherited value is 0; finally L i ∩M j start = Σ d Inheritance d , denominator | L i | represents the sum of the absolute values ​​of each dimension of the legacy vector (L1 norm).

[0045] Furthermore, we define the winding coefficient for a work condition segment and calculate the ability of a work condition segment's impact to persist across multiple subsequent work condition segments after its completion: E is obtained through statistical summation. i E i For the working condition segment g iThe resulting winding coefficient of the working condition segment; T ij This represents the inheritance strength between this working condition segment and the subsequent j-th working condition segment.

[0046] Establish a continuous weaving field for each working condition segment based on the winding coefficient of all working condition segments: W ={(g i E i )|i=1,2,…,n}, where W is the continuous weaving field of the working condition segment; Eᵢ is the winding coefficient of the working condition segment, representing the winding capability of the corresponding working condition segment.

[0047] The continuous weaving field of the working condition segment records the non-local, cross-linking coupling relationship between working condition segments.

[0048] For example, after a high-intensity braking segment ends, the residual compression stroke, increased oil temperature, and pressure within it persist across multiple subsequent steering and impact segments, continuously influencing the fork's initial response characteristics in those segments. The essence of a continuous weaving field for braking segments is a directed weighted graph with braking segments as behavioral state nodes and winding coefficients as weights, completely preserving all traces of load history transmission between different braking segments.

[0049] After completing the above weaving, the module sends the continuous weaving field W of the working condition segment as structured data to the motion behavior acquisition module.

[0050] After receiving the continuous weaving field W, the motion behavior acquisition module determines the winding coefficient E corresponding to each working condition in the continuous weaving field. i Establish a behavioral inheritance space so that behaviors left over from historical working conditions can continue to participate in and influence the entire subsequent movement process.

[0051] The four types of dynamic response sequences generated by the fork during continuous motion are uniformly mapped to the same behavior space. Compression trajectory response C t The displacement is acquired by a wire-type displacement sensor or magnetostrictive displacement sensor installed between the upper end of the outer tube and the lower end of the inner tube of the fork, and is characterized by the real-time relative displacement between the upper and lower ends of the fork, in millimeters. Positive values ​​indicate the compression direction, and negative values ​​indicate the extension direction; the rebound trajectory response R t By responding to the compressed trajectory C t The time derivative is used to obtain the real-time value of the relative recovery velocity between the inner and outer tubes after the fork compression stroke is released, represented by millimeters per second. Positive values ​​indicate the direction of rebound extension, and negative values ​​indicate the direction of continued compression. Lateral drift response D... t The lateral displacement is acquired via a bidirectional accelerometer or Hall effect sensor mounted on the fork bushing, and is characterized by the fork's offset in a lateral plane perpendicular to the vehicle's direction of travel, in millimeters; structural micro-vibration response V tAccelerometers are acquired using a piezoelectric accelerometer attached to the middle of the fork outer tube surface. The effective acceleration value, expressed in g (gravitational acceleration), is represented by the high-frequency, small-amplitude vibrations of the fork tube surface. The sampling frequency is no less than 10 kHz to capture micro-vibration characteristics such as damping valve flutter and high-frequency hydraulic pulsation within the fork. All four types of responses are acquired at a synchronous sampling rate of no less than 1 kHz and combined at time t to form a behavior vector B. t =(C t , R t D t V t Each component in this vector is a signed real-time value, which together constitute a complete description of the fork's motion behavior at time t.

[0052] Based on the historical operating condition legacy L output by the previous module i Establish behavioral activation potential A t The activation potential at time t is calculated by multiplying the legacy of each historical condition by its exponential decay factor over time, and then summing the contributions of all historical conditions. The specific expression is A. t =ΣL i ·exp(-|tt i | / τ i ), where t i τ represents the end time of the i-th historical working condition. i This refers to the decay period corresponding to the legacy behavior. The decay period τ i The values ​​are determined based on different physical dimensions: the decay period of residual compression stroke is determined by the natural frequency and damping ratio of the fork spring through free vibration decay experiments, with a typical value of 0.5 to 2.0 seconds; the decay period of residual oil pressure is determined by the flow characteristics of the damping orifice and the bulk modulus of the oil through pressure release experiments, with a typical value of 1.0 to 5.0 seconds; the decay period of residual temperature rise is determined by the heat capacity and heat dissipation conditions of the fork assembly through thermal balance cooling experiments, with a typical value of 30 to 300 seconds. For each output residual vector L... i The weighted average of the attenuation periods in each dimension is taken as the comprehensive attenuation period τ for this operating condition. i The weights are determined by the contribution coefficients of each dimension to the motion behavior; among them, the weight coefficients of each dimension are determined by principal component analysis (PCA) of the B values ​​in the real vehicle data. t The variance explained by the vector is determined, with typical weights as follows: compression stroke 0.30, compression speed 0.20, damping force 0.20, spring force 0.10, internal oil pressure 0.10, and internal temperature 0.10; the overall decay period τ is also considered. i =Σ(w d ·τ d ) / Σw dThe behavioral activation potential quantitatively characterizes the combined influence of all historical operating conditions still retained at the current time t. The closer the end time of the historical operating condition, the larger the residual amount, and the longer the decay period, the greater its contribution to the activation potential at the current time.

[0053] In actual riding, multiple historical conditions often influence the current movement of the fork. These historical behaviors compete for dominance over the current movement. A behavioral competition matrix Q is then established, where each element q... i1j1 This represents the competition intensity between the legacy behaviors of the i1th historical condition and the j1th historical condition. The specific calculation method for the competition intensity is to add a stability coefficient to the distance between the behavior vectors corresponding to the two historical behaviors as the denominator, and take the product of their respective activation potentials as the numerator. Dividing these results in the competition intensity, q. i1j1 =(A i1 ·A j1 ) / (|B i1 -B j1 |+ε), where A i1 and A j1 Bi and Bj are the activation potentials of the i1th and j1st historical conditions at the current moment, respectively, and Bi and Bj are the corresponding action vectors, respectively. i1 -B j1 | represents the Euclidean distance between the two, and ε is the stability coefficient. The stability coefficient ε is set to 0.5% of the diagonal length in the behavior space, where the range of each component is the full-scale value of the sensor. Its physical function is to prevent the denominator from becoming zero and causing computational overflow when the two behavior vectors completely overlap, in which case the competition intensity degenerates into the product of activation potentials. The diagonal length is calculated based on the full-scale range of each component. ε=0.005×L diag .

[0054] Based on the behavior competition matrix, two key regions in the behavior space are further identified: the behavior acceptance region and the behavior competition region; where: the behavior acceptance region Z... k ={B t |q i1j1 <θ1} represents the set of regions among all behavior vectors where the competition intensity between them is lower than a preset acceptance threshold θ1.

[0055] Behavioral competition zone F k ={B t |q i1j1 >θ2} represents the set of regions in all behavior vectors where the competition intensity between them is higher than the preset competition threshold θ2, and it follows the threshold θ1 and the competition threshold θ2; it satisfies 0<θ1<θ2, where the value of θ1 is 30% of the median of the competition intensity distribution among all historical working condition pairs, and the value of θ2 is 70% of the median of the distribution;

[0056] Under a pure equilibrium state with no historical legacy influences, a segment of unloaded fork motion data was collected. The competition intensity between points in the behavioral space under this state was calculated. The 95th quantile of this competition intensity distribution was used as θ1, and the 99.5th quantile as θ2, to adapt the threshold to specific vehicle models. The unloaded data collection conditions were: the fork was detached from the motorcycle, vertically mounted on a test bench, with no external load, driven by a 1Hz sinusoidal displacement excitation (amplitude ±5mm) for 10 cycles, recording all sensor data. The 95th quantile of the competition intensity distribution under this state was taken as θ1, and the 99.5th quantile as θ2, allowing for fine-tuning in subsequent real-world testing with a coefficient of 0.9 to 1.1 based on the density of the regional divisions. After identifying these two types of regions, the transfer relationship between the behavioral acceptance zone and the behavioral competition zone was analyzed, i.e., how the behavior state of the fork flows and evolves between these different regions. For this purpose, a behavioral migration flux Φ was defined. ab =Σ[ A t ·ΔB t / (1+q t ] represents the total intensity of the migration of behavioral states from region a to region b, where ΔB t =||B t -B t-1 || represents the Euclidean distance between the current action vector and the previous action vector, characterizing the magnitude of change in the action state within a unit time step, q t This represents the competition intensity at the corresponding time step. The migration flux is calculated by dividing the current activation potential by the current competition intensity at each state transition time, multiplying by the change in behavior, and summing these products over all time steps. The migration direction is determined as follows: if B... t Falling into region b and B t-1 If the value falls into region a, then the contribution value at that moment is At·||ΔB t || / (1+q t )Include Φ ab Cross-regional migration flux is only counted when a≠b. Migration within the same region is not counted as inter-regional flux but is used for internal retention analysis. Migration flux comprehensively reflects the ability of historical behavior to diffuse to a new behavioral state and cross different regions: the higher the activation potential, the lower the competition intensity, and the greater the behavioral change, the greater the migration flux.

[0057] After obtaining the migration flux between all regions, the module establishes the behavior-based network G. B=(Z,F,Φ), this network is composed of all behavior receiving regions Z, all behavior competition regions F, and the set of behavior migration fluxes between regions Φ. Specifically, the process of establishing the behavior receiving network is as follows: each behavior vector in the behavior space is labeled according to its region. Then, for any two regions a and b, all migration events that start from the vector labeled a and transition to the vector labeled b at the next time step are counted. The migration fluxes corresponding to these events are accumulated to obtain the total migration flux from region a to region b. Finally, a complete weighted directed graph is formed with regions as behavior state nodes and migration fluxes as edge weights.

[0058] To further analyze whether there is a cyclic migration phenomenon in the behavior transmission network where historical behaviors repeatedly return to a certain region, a behavior backflow index R is defined. b =ΣΦ in / ΣΦ out Where Φ in Φ represents the sum of migration fluxes entering the region. out The return flow index represents the sum of migration fluxes leaving the region. For any behavioral region, its return flow index is calculated by dividing the sum of all migration fluxes entering the region by the sum of all migration fluxes leaving the region. When the behavioral return flow index R of a region... b When the value is greater than 1, it indicates that the total amount of historical behavior entering the area is consistently greater than the total amount leaving the area. Historical behavior continues to accumulate in the area and is difficult to dissipate, forming a phenomenon of behavioral stagnation.

[0059] Finally, all behavior receiving areas, all behavior competition areas, all behavior migration fluxes, and the corresponding behavior return flow indices for each area are integrated to establish a complete behavior evolution network G. E =(Z,F,Φ,R b The core information recorded by the behavioral evolution network goes far beyond a simple record of the fork movement sequence: by tracing the source of the activation potential contribution of the current moment's behavior vector region, identifying the top three historical working condition numbers with the largest contributions and their residual components, it is found that the residual influence of historical working conditions is still playing a role at the current moment;

[0060] By identifying the top N q with the highest competition intensity at the current moment i1j1 The corresponding historical working conditions are marked as competitive active working conditions, and the influence of the corresponding historical behavior is competing for the dominance of the current movement. When the activation potential of a certain historical working condition decays to less than 5% of its initial value, it is marked as an inherited working condition and no longer participates in the subsequent competition matrix calculation. The corresponding historical behavior has been smoothly inherited by the subsequent working conditions and no longer produces independent influence.

[0061] The behavioral return index R bRegions that are consistently greater than 1 and whose duration exceeds twice the decay period are marked as stagnation zones, and the dominant historical operating condition number within the stagnation zone is output. This indicates that the corresponding historical behavior has formed a long-term stagnation in a specific region that is difficult to be replaced by subsequent movements.

[0062] After the above setup is completed, the behavioral evolution network is sent as structured data to the historical impact analysis module.

[0063] The historical impact analysis module receives the fork behavior evolution network G E =(Z, F, Φ, R) b Afterwards, the analysis examines whether historical behaviors that have exited the current motion process still continue to affect subsequent motion processes through other behavioral areas, identifying historical load transfer traces within the fork. The core task is to trace legacy behaviors that have ended in time but still exert influence in the behavioral space through the load-bearing zone, competition zone, and migration flux.

[0064] Read all structured data from the behavioral evolution network, including the set of behavioral receiving regions Z, the set of behavioral competition regions F, the set of behavioral migration fluxes Φ, and the set of behavioral backflow indices R. b .

[0065] In the behavioral evolution network, each behavioral state node N i2 All belong to one and only one behavior region, namely a certain receiving region Z. k Or a certain competitive zone F k The affiliation between a behavior state node and a region is determined by the location coordinates of the behavior state node and the boundary range of the region: if behavior state node N i2 The behavior vector coordinates fall into the behavior receiving region Z. k Within the boundary range, the region type to which the behavior state node belongs is marked as the receiving region, and the region index is k; if it falls into the behavior competition region F k Within the boundary range, the region type is marked as a competition zone, and the region index is k. For any behavior state node N i2 Its corresponding regional reflux index R i2 Obtain directly from the backflow index value of the region to which the behavior state node belongs: If the behavior state node N i2 Belongs to the receiving area Z k Then R i =R b (Z k If the behavior state node N i2 Belongs to competitive zone F k Then R i2 =R b (F kFor intermediate zone behavior state nodes that do not belong to any receiving zone or competing zone, their return flow index is uniformly assigned a value of 1, indicating that the zone neither generates additional retention nor accelerates release.

[0066] For any behavioral state node N in the behavioral evolution network i2 The module establishes the residual potential U of behavior. i2 The residual potential of the behavior is calculated as follows: for behavior state node N... i2 All subsequent behavior state nodes N k Where k ranges from i2+1 to n2, representing the behavior state node N. i To behavioral state node N k Behavioral migration flux Φ i2k Divide by the time span between the behavior state nodes |k-i2|, then multiply by the number of behavior state nodes N. k The corresponding region's behavioral return flow index R k The calculation results of all subsequent behavior state nodes are summed, i.e., U i2 =Σ[Φ i2k / (1+|k-i2|)×R k Among them, the behavioral migration flux Φ i2k For the behavior state from behavior state node N i2 Migrate to behavior state node N k The intensity value, |k-i2| is the behavior state node N. i2 With N k The time step difference between nodes. The physical meaning of the time step difference is: if the behavioral evolution network collects data at a fixed sampling interval Δt, the actual time span between behavioral state nodes is |k-i²|×Δt, in seconds. Dividing by (1+|k-i²|) attenuates the impact of distant behavioral state nodes, causing the contribution of residual potential to decrease as the span increases. Behavioral residual potential characterizes the ability of a behavior to retain influence after exiting a dominant state. In simpler terms, even if a behavioral state node is no longer in a dominant position, as long as there is migration flux between it and subsequent behavioral state nodes, and behavioral backflow exists in the region where the subsequent behavioral state nodes are located, then that behavioral state node still has considerable residual influence, which decays as the time span increases.

[0067] After obtaining the residual potential of each behavioral state node, the behavioral legacy field Y = {U1, U2, ..., U} is established. n2 The behavioral legacy field describes the distribution of the residual effects of all historical behaviors throughout the entire movement process. It is not a single value, but rather a set of residual potentials across all behavioral state nodes in the entire behavioral evolution network, reflecting the intensity of the continuous influence of historical behaviors at various points in the network.

[0068] Further calculation of behavioral convergence intensity C j2 For any behavior state node N j2 The convergence strength is calculated as follows: for all nodes N in the behavior state... j2 Previous historical behavior state node N i2 Where i2 ranges from 1 to j2-1, representing the behavior state node N. i2 behavioral residual potential U i2 Multiplied by a weighting factor, which is the behavioral state node N. i2 To behavioral state node N j2 migration flux Φ i2j2 Divide by all preceding behavior state nodes to behavior state node N j2 The sum of the migration fluxes from all preceding behavioral state nodes to behavioral state node Nj is denoted as ΣΦkj2 (k ranges from 1 to j2-1). Then, the calculation results for all preceding behavioral state nodes are summed, i.e., C. j2 =Σ[U i2 ×Φ i2j2 / (ΣΦ kj2 )]. Wherein, ΣΦ kj2 For all preceding action state nodes (k=1 to j2-1) to the current action state node N j2 The sum of migration fluxes is used as the normalized denominator to ensure that the sum of all weight coefficients is 1. For an initial behavioral state node without any preceding migration fluxes (i.e., j2=1), its convergence strength is defined as 0. Behavioral convergence strength is used to characterize how much influence of historical behaviors the current behavioral state node has received. The greater the convergence strength, the more the current behavioral state node's motion state is jointly shaped by historical conditions.

[0069] Subsequently, the diffusion of behavioral convergence intensity in the behavioral evolution network is analyzed. For any behavioral state node Nj2, a behavioral propagation gain G is established. j2 Its calculation method is based on the behavior state node N. j2 Behavioral convergence strength C j2 Multiply by the behavior backflow index R of the region corresponding to the behavior state node. j2 G j2 =C j2 ×R j2 Among them, R j2 For behavior state node N j2 The backflow index of the behavioral region. Behavioral propagation gain characterizes the ability of historical behavior to be amplified again at the current behavioral state node. When a behavioral state node has high convergence intensity and a high backflow index, it indicates that historical behavior not only converges in large quantities at this node but also continuously circulates and accumulates in this region, forming a significant amplification effect. For R...j2 The intermediate region behavior state node with =1 has a propagation gain equal to the convergence strength, which neither amplifies nor attenuates it.

[0070] To further assess the overall persistence of historical behaviors in the network, the module establishes a behavior impact conservation degree K. This conservation degree is calculated by dividing the total residual potential ΣUi of all historical behaviors by the total propagation gain ΣG. j2 That is, K=ΣU i2 / ΣG j2 ΣU i2 ΣG represents the total residual potential of all historical behaviors, i.e., the sum of the residual potentials over all behavioral state nodes; j2 The total propagation gain is calculated by summing the propagation gains of all behavioral state nodes. When the behavior impact conservation K consistently approaches 1 (in engineering, K ∈ [0.85, 1.15] is defined as close to 1), it indicates that historical behaviors have not disappeared but are continuously migrating and transforming within the network, manifesting as a dynamic balance between the total input and output. When the behavior impact conservation K < 0.5, it is considered a rapid decrease, indicating that the residual impact of historical behaviors is being released or dissipated, and the transmission of historical loads gradually decays until it disappears. When K > 1.5, it is considered an abnormal amplification, indicating that there may be a self-reinforcing phenomenon of historical behaviors caused by loop backflow in the network, requiring further inspection of whether there are abnormalities in the network data or damage to the fork structure. This judgment threshold serves as a self-checking indicator of the system's operating status. The output K value is updated with each operating cycle, and a warning signal is output when the K value remains abnormal.

[0071] Subsequently, specific propagation paths for the persistence of historical behavior are identified. For any possible behavior propagation path P in the network... q This path is composed of multiple behavioral state nodes connected in chronological order, denoted as P. q =(N a →N b →N c →…). Specifically, the path set is generated using a depth-first search strategy: starting from each non-last behavior state node, tracing backward along directed edges with migration flux Φ > 0, stopping when the path length reaches the preset maximum search step size (usually 3 to 6 steps, determined by balancing computational resources and real-time requirements) or when the last behavior state node is reached, and recording all possible paths. To prevent path combination explosion, the migration flux at each step is filtered during the search process: only Φ is retained. i2j2 >0.1×Φ max Edges participate in path establishment, where Φ max This represents the largest behavioral migration throughput across the entire network.

[0072] Calculate the memory strength M for each pathq Specifically, the memory strength is calculated as follows: for each migration edge in the path, calculate its behavioral migration flux Φ. e With the largest behavioral migration throughput Φ in the entire network max The ratio is calculated by multiplying the ratios of all edges together, and then multiplying by the total residual potential ΣU of the behavior of each state node in the path. e With path length |P q The ratio of | to M q =[Π(Φ e / Φ max )]×[ΣU e / |P q |]. Among them, Π(Φ e / Φ max ΣU is the product of the normalized fluxes of all edges in the path. e The sum of the behavioral residual potentials of all behavioral state nodes in the path, |P q | represents the number of migration edges contained in the path, i.e., the path length.

[0073] When the memory strength M of a certain path q Exceeding the preset threshold M th When this happens, it is determined that the path forms a force flow memory chain L. q This is recorded as a sequence of behavioral state nodes arranged in chronological order. A preset threshold M is used. th The method for determining M is as follows: calculate the memory intensity distribution of all searched paths, and take 3 times the median of the distribution as M. th Set the lower limit protection value M th =0.01, ensuring the threshold is not lower than this value. Specifically, path search only considers paths with a length of no more than 6 steps and a residual potential of the starting node greater than 0.01, and calculates the M of all valid paths. q Distribution, M th = max(3×Median, 0.01); If the median Median = 0, then take M. th =0.02.

[0074] Since multiple force flow memory chains may share common behavioral state nodes, the memory chain cross strength J is further established. uv For any two force flow memory chains L u and L v The cross strength is calculated as follows: calculate the number of common behavior state nodes |L u ∩L v |Divided by the total number of behavioral state nodes across both chains|L u ∪L v | Then multiply by the sum of the path memory strengths of the two chains, i.e., J uv =[|Lu ∩L v | / |L u ∪L v |]×(M u +M v Among them, |L u ∩L v | is the number of behavioral state nodes shared by the two memory chains, calculated using the intersection of sets; |L u ∪L v The union size is the sum of the number of individual behavioral state nodes in each of the two memory chains minus the number of nodes sharing a common behavioral state node; M is the size of the union. u and M v They are memory chains L u and L v The cross-link strength is determined by the path memory strength of each chain. A higher proportion of shared behavioral state nodes and greater individual memory strength between the two chains result in greater cross-link strength. Cross-link strength reflects the degree of coupling effect of different historical loads within the fork. High cross-link strength indicates that two different types of historical loads interact and couple within the fork through shared behavioral state nodes.

[0075] Finally, based on all force flow memory chains L, the path memory strength M corresponding to each memory chain, and the cross strength J between each pair of memory chains, a complete force flow memory map T is constructed. M =(L, M, J). The specific data structure of the force flow memory graph is as follows: L is the set of force flow memory chains, and each chain records a sequence of behavior state node numbers [Node ID1 Node ID2 […], along with the timestamps of the starting and ending behavior state nodes and the chain number; M is the set of path memory strengths, corresponding one-to-one with each chain in L, stored in key-value pairs (chain number → memory strength value); J is the set of memory chain cross strengths, stored in the form of an adjacency matrix, with the matrix dimension being the total number of memory chains m×m, where J[u][v]=J uv (u≠v), diagonal element J[u][u]=0.

[0076] The force flow memory map records not just a simple force process, but a long-term influence propagation structure formed by historical loads within the fork through multiple mechanisms such as behavioral reception, behavioral competition, behavioral backflow, and behavioral amplification. Presented as a topological network, the force flow memory map comprehensively records how the load history within the fork starts from its initial position, propagates along specific paths, aggregates and amplifies at key behavioral state nodes, and how multiple paths intersect and couple, ultimately forming a complex influence network covering the entire behavioral space.

[0077] After the above steps are completed, the force flow memory map is packaged in binary serialization format and sent to the stress acceptance parsing module via the agreed interface protocol. The data packet contains complete information such as time stamps, total number of memory chains, total number of behavioral state nodes, chain list, intensity list, and cross matrix.

[0078] After receiving the force flow memory map output by the aforementioned historical impact analysis module, the stress bearing analysis module analyzes how the legacy effects of historical operating conditions are inherited, transferred, superimposed, and interrupted between different force flow memory chains. This allows it to identify the long-term bearing structure formed inside the fork by historical loads, ultimately establishing the structural spatial torsion field. The specific process is as follows:

[0079] The force flow memory map consists of three core components: the set of force flow memory chains, the set of path memory strengths, and the set of memory chain cross-strengths.

[0080] The force flow memory chain set L in the force flow memory map contains all load propagation paths identified as having stable transmission capabilities. Each force flow memory chain consists of a sequence of behavioral state node numbers, along with timestamps of the starting and ending behavioral state nodes, used to determine the chain's position on the timeline. The path memory intensity set M corresponds one-to-one with each memory chain, storing the memory intensity value for each chain using its chain number as an index. The memory chain cross-strength set J uses an upper triangular matrix to store the cross-strength values ​​between any two different memory chains. The matrix dimension is m times the total number of memory chains, where the element in the u-th row and v-th column is the memory chain L. u With L v The cross strength between elements is zero when u equals v, and the diagonal elements are zero. When reading the force flow memory map, the total number m of all force flow memory chains is obtained from the set L, and then each memory chain is traversed sequentially to extract the sequence of behavioral state nodes and the associated time information contained therein.

[0081] Each force flow memory chain consists of multiple behavioral state nodes connected in chronological order. The number of behavioral state nodes in the chain is denoted as p. The number of behavioral state nodes can be different for different memory chains. For any two memory chains L u With L v The module establishes the force flow continuity potential Q. uv The calculation method for this succession potential is as follows: traverse the first memory chain L u Each behavioral state node N i2 With the second memory chain L v Each behavioral state node N j2 , will the behavior state node N i2 behavioral residual potential U i2 With behavioral state node N j2behavioral residual potential U j2 Multiply by, then multiply by the cross strength J between the two memory chains. uv The product of all behavior state node pairs is summed and then divided by the absolute value of the difference between the average time centers of the two memory chains.

[0082] Among them, the behavioral residual potential U of the behavioral state node i2 and U j2 The behavioral legacy field Y, output by the historical impact analysis module, records the residual potential value of each behavioral state node in the behavioral evolution network. The average time center t... u and t v The definition is: for memory chain L u The average time center of a memory chain is the sum of the timestamps of all behavioral state nodes it contains, representing its centroid position in the time dimension. Specifically, it is calculated by adding the timestamps of all behavioral state nodes in the chain and dividing by the number of behavioral state nodes, p. This summates the average time center of the chain. Dividing this sum by the absolute value of the difference between the average time centers of two memory chains results in a higher continuity potential for chains that are closer in time, while the continuity potential between chains that are farther apart in time is attenuated, reflecting the continuity and timeliness of historical influence in the time dimension. All timestamps are in seconds, and the absolute value of the difference in average time centers is in seconds. If the difference in the average time centers of two memory chains exceeds 100 seconds, they are considered too far apart to be continuous, and their continuity potential is truncated to 0.

[0083] The continuity potential characterizes the ability to transfer loads between two historical load propagation paths. The larger the continuity potential, the easier it is for the historical influence of the previous memory chain to be transmitted to the next memory chain, indicating a stable load transfer relationship between the two.

[0084] After calculating the continuation potential for all memory chain pairs, a global continuation matrix Q is established. This matrix is ​​an m x m square matrix, and the element in the u-th row and v-th column records the memory chain L. u For memory chain L v The continuation potential is such that when u equals v, the diagonal elements are zero, indicating that the memoe chain does not generate a continuation for itself.

[0085] Since the memory strength of different memory chains varies, the contribution of behavioral state nodes is further calculated. For any behavioral state node N in the behavioral evolution network... i2 It is necessary to determine the memory chain to which the behavior state node belongs. The affiliation between the behavior state node and the memory chain is determined by matching the behavior state node number with the sequence of behavior state node numbers recorded in the memory chain: if behavior state node N i2 The number appears in memory chain L uIn the sequence of behavior state nodes, the behavior state node N is determined. i2 Belongs to memory chain L u If a behavior state node appears in multiple memory chains, then that behavior state node belongs to all memory chains that contain it.

[0086] Behavioral state node contribution A i2 The calculation method is as follows: multiply the behavioral residual potential Ui2 of the behavioral state node by the sum of the continuity potentials of the memory chain to which the behavioral state node belongs to all other memory chains, and then multiply by the ratio of the memory strength of the memory chain to which the behavioral state node belongs to the maximum memory strength of the entire network. Specifically, if the behavioral state node N i2 Belongs to memory chain L u First, calculate the relationship between memory chain Lu and all other memory chains L. v The continuation potential Q (v from 1 to m, v not equal to u) uv The sum is then multiplied by the residual potential U of the behavioral state node. i2 Then multiply by the memory chain L u Memory strength M u Divide by the maximum memory strength M across the entire network max If a behavioral state node belongs to multiple memory chains, the contribution of each chain is calculated separately, and the maximum value is taken as the final contribution of the behavioral state node. The contribution of a behavioral state node reflects its ability to assume and absorb influence during the historical impact transmission process. A higher contribution indicates a more crucial intermediary role the behavioral state node plays in the historical payload transmission network.

[0087] The module then establishes the diffusion index D of the behavioral state nodes. i2 The index is calculated as follows: the contribution A of the behavior state node is taken as the contribution. i2 Multiply by 1 and add the ratio of the total output migration flux to the total input migration flux of this behavior state node. The total output migration flux is the sum of the behavior migration flux output by this behavior state node to all subsequent behavior state nodes in the network, i.e., the summation Φ over all behavior state nodes with an index greater than the current behavior state node j. i2j2 The total input migration throughput is the sum of the migration throughput received by this behavior state node from all its predecessor behavior state nodes in the network, which is calculated by summing Φ over all behavior state nodes whose index is less than k. ki2 The migration flux Φ originates from the migration flux values ​​between pairs of behavioral state nodes recorded in the edge table of the behavioral evolution network.

[0088] The diffusion index of a behavioral state node describes its ability to spread historical influences from that node to subsequent states. A diffusion index greater than one indicates that the node's outward output influence is greater than its received influence, making it a source of influence. A diffusion index less than one indicates that the node receives more influence than it outputs, making it a sink of influence. Behavioral state nodes with larger diffusion indices tend to have their historical influences propagate to more distant subsequent nodes.

[0089] Furthermore, the module establishes stress bearing strength S i2 This strength is equal to the contribution A of the behavior state node. i2 Multiply by the diffusion index D of the behavioral state nodes i2 Multiply by the backflow index R of the region to which the behavior state node belongs. i2 Stress bearing capacity comprehensively reflects the bearing capacity of a behavioral state node in three dimensions: bearing contribution characterizes the transit importance of the behavioral state node, diffusion index characterizes the activity of outward propagation of influence, and backflow index characterizes the tendency of influence to accumulate cyclically in the region. When the stress bearing capacity S of a certain behavioral state node... i2 Greater than the maximum stress bearing strength S of the entire network max When the value reaches 70%, the module determines the behavior state node as a stress-bearing behavior state node, where S max The maximum stress bearing strength is taken after traversing all behavioral state nodes. The stress bearing behavioral state nodes represent the key positions that bear the main bearing role in the historical load propagation process inside the fork. These positions are the core hubs in the load history transfer network.

[0090] Subsequently, the diffusion of stress-bearing behavior state nodes in subsequent working conditions is analyzed. For any behavior state node N that has been identified as a stress-bearing behavior state node... i2 Starting from this behavior state node, trace backwards along the time sequence to establish the diffusion path P. i2 The diffusion path is established as follows: starting from the current receiving behavior state node, following the direction of migration flux in the behavior evolution network, the next behavior state node with the largest migration flux is selected as the next behavior state node in the path, until the last behavior state node in the network is reached or the migration flux drops to less than one percent of the maximum migration flux in the entire network, at which point the diffusion path extension stops. Diffusion path P i2 It includes the receiving behavior state node and a series of subsequent behavior state nodes, and the path length is determined by the number of behavior state nodes actually tracked.

[0091] Further calculation of the diffusion energy E of this diffusion path i2The diffusion energy is calculated as follows: for each migration edge and each behavioral state node in the diffusion path, the migration flux Φ of the migration edge is calculated. e With the largest migration throughput in the entire network Φ max The ratio multiplied by the acceptance strength S of the behavior state node e With the highest bearing capacity S in the entire network max The ratio is calculated, and then the calculation results of all edges and behavioral state nodes in the path are summed. Path diffusion energy reflects the ability of historical influences to continue in subsequent working conditions. The greater the diffusion energy, the more likely the historical influences originating from the receiving behavioral state node can maintain their effect in subsequent long-term movements.

[0092] Further analysis is needed to determine whether the lingering effects of multiple historical construction conditions continue to converge in the same spatial region. The region R referred to here... g The analysis is based on the set of behavioral receiving areas Z and behavioral competing areas F output by the historical impact analysis module; all identified receiving and competing areas are used as the analysis objects. For any behavioral region R... g Count the number N memory chains that pass through this region. g The statistical method is to perform a statistical analysis on each force flow memory chain L. u The system makes a determination: if any behavior state node in the memory chain falls into the region, the count is one; otherwise, the count is zero. The sum of all determination results gives the number of memory chains that have passed through the region.

[0093] The module then calculates the regional aggregation index C. g The index is equal to the number of memory chains Ng passing through the region multiplied by the average bearing strength S of the region. g Multiply by the average reflux index R of the region g The regional average bearing capacity is calculated by summing the stress bearing capacity of all behavioral state nodes within the region and dividing by the total number of behavioral state nodes in the region. The regional average return flow index is calculated by summing the return flow indices of all behavioral state nodes within the region and dividing by the total number of behavioral state nodes in the region. The regional aggregation index comprehensively reflects the degree of convergence and superposition of the legacy effects of multiple historical working conditions in the same region. When the aggregation index C of a certain region... g Greater than the largest clustering index C in the entire network max When the stress level reaches 80%, the area is identified as a stress accumulation zone. A stress accumulation zone indicates a location where the effects of multiple historical operating conditions are continuously superimposed and amplified in the same area. This type of area is the most dangerous area in the fork structure where fatigue accumulation and performance degradation are most likely to occur.

[0094] On the other hand, we analyze the interruptions that occurred during the historical transmission of influence. For each force flow memory chain L... uEstablish the memory decay rate B between adjacent behavioral state nodes in the chain, based on their sequential order. k The decay rate is equal to the memory strength M of the previous row's state node. k In addition to the difference in memory strength between the previous row and the next row of state nodes (M) k -M k+1 The greater the decay rate, the more severe the decay of historical influence between adjacent behavioral state nodes.

[0095] Then the continuous attenuation length T is calculated. u This refers to the number of consecutive occurrences of adjacent behavior-state node pairs whose attenuation rate exceeds a preset attenuation threshold β. The attenuation threshold β is determined by calculating the memory attenuation rate distribution of all adjacent behavior-state node pairs in the entire network, taking twice the median of this distribution as the β value, and setting a lower limit protection value β. min =2, ensuring the threshold is not lower than this value, so that only behavior-state node pairs above the average decay level are included in the continuous decay statistics. Specifically, decay rate B k =M k / (M k -M k+1 ), where M k The instantaneous memory contribution value of this node (defined as the memory strength of the chain to which this node belongs multiplied by the behavioral residual potential of this node); if M k ≤M k+1 Then B k Treating it as infinite (neither decaying nor increasing) and not including it in the continuous decay statistics; if B is encountered during the continuous decay statistics process. k If the value is ≤β, then the count is reset, T u Find the longest consecutive number of occurrences among all consecutive segments within the chain.

[0096] When the continuous decay length T on a certain memory chain u Reaching three or more, and the memory strength M of the third subsequent behavioral state node. k+3 It has dropped to the highest memory strength M across the entire network. max When the stress concentration drops below 10%, the corresponding area is considered to have formed a stress fracture zone. The spatial location of the stress fracture zone is determined by the physical spatial location of the behavioral state node through which the continuous attenuation segment passes. The stress fracture zone represents the location where the historical influence suddenly stops during its propagation. It should be noted that these areas usually correspond to geometrical abrupt changes, material interfaces, or damage initiation points in the fork structure, and are key areas to focus on in structural integrity assessments.

[0097] After identifying all stress-bearing behavior state nodes, stress accumulation zones, and stress fracture zones, the module establishes the structural spatial torsion field W. sThe torsion field contains four core pieces of information: A is the set of all stress-bearing behavior state nodes and their spatial coordinates; S is the set of stress-bearing strength values ​​corresponding to each stress-bearing behavior state node, corresponding one-to-one with each behavior state node in A; C is the set of spatial boundaries and region numbers of all stress accumulation zones; and T is the set of spatial locations and ranges of all stress fracture zones. The data structure of the structural spatial torsion field is stored in key-value pair format for easy retrieval by subsequent modules.

[0098] Further establish the structural space distortion index K s Structural space distortion index K s The calculation method is as follows: The numerator is the sum of the accumulation indices of all stress accumulation zones plus the sum of the bearing strengths of all stress bearing behavior state nodes; the denominator is the sum of the continuous attenuation lengths of all stress fracture zones. The result is obtained by dividing the sum by the sum of the indices of all stress accumulation zones, i.e., K. s =(ΣC g +ΣS i2 ) / (1+ΣT u ), where K s To balance the order of magnitude, ΣC g and ΣS i2 Normalize by dividing each by its respective maximum value across the entire network, i.e., K. s = [(ΣC g / C max )+(ΣS i2 / S max )] / (1+ΣT u If ΣT u =0, then the denominator is 1. The larger the structural spatial distortion index, the more pronounced the load-bearing and accumulation effects of historical loads within the fork, and the weaker the stress release capacity, making the structure more prone to long-term displacement and hidden damage. Based on engineering experience, when K... s When K is greater than 2, it is considered a high-risk distortion state. s When K is between 1 and 2, it is considered a moderate distortion state; when K... s A value less than 1 indicates a low-twist state.

[0099] After receiving the structural space torsion field, force flow memory map, and fork behavior evolution network, the inertial retention prediction module analyzes historical behaviors that have exited the current operating condition, identifies inertial retention phenomena in historical behaviors, and infers their future expansion trends. The structural space torsion field includes a set of stress-bearing nodes, a set of stress-bearing strengths, a set of stress accumulation zones, and a set of stress fracture zones. The force flow memory map includes a set of force flow memory chains, a set of path memory strengths, and a set of memory chain intersection strengths. The fork behavior evolution network includes a set of behavior-bearing zones, a set of behavior-competing zones, a set of behavior migration fluxes, and a set of behavior backflow indices. The specific process is as follows:

[0100] The node sequence of each force flow memory chain is read from the force flow memory map. Each memory chain consists of multiple behavioral state nodes connected in chronological order, and the node sequence is extracted sequentially according to the storage order of the node numbers in the memory chain record. During the reading process, the following information associated with each node is extracted: the node's index number in the behavioral evolution network, the behavioral region type and region number to which the node belongs, the timestamp corresponding to the node, the node's behavioral residual potential value, the backflow index value of the region to which the node belongs, the node's stress bearing strength value, and the memory strength value of the memory chain to which the node belongs. After completing the reading of the node sequences of all memory chains, it is analyzed whether historical behaviors continue to appear in subsequent behavioral chains after exiting the dominant working condition. The specific process is as follows:

[0101] For any behavioral state node, an inertial residual factor is established. The inertial residual factor is calculated by multiplying the node's behavioral residual potential by the backflow index of the node's region, and then multiplying by the ratio of the node's stress bearing capacity to the maximum stress bearing capacity of the entire network. Here, the behavioral residual potential originates from the behavioral legacy field output by the historical influence analysis module, reflecting the node's remaining influence capacity after exiting the dominant state; the backflow index originates from the backflow index value of the node's region in the behavioral evolution network, reflecting the tendency of behavior to accumulate cyclically in that region; and the stress bearing capacity originates from the output of the stress bearing analysis module, reflecting the node's bearing capacity in the historical load transfer network. A larger inertial residual factor indicates that historical behavior is less likely to disappear from the node, and historical loads have higher persistence inertia at that node location.

[0102] After obtaining the inertial retention factor for each node, an inertial retention potential is established. The inertial retention potential is calculated as follows: multiply the node's inertial retention factor by the sum of the cross-strengths of that node with all other memory chains, and then multiply by the ratio of the memory strength of the memory chain to the maximum memory strength of the entire network. For nodes belonging to multiple memory chains, traverse all memory chains containing that node, calculate the sum of the cross-strengths of each memory chain with all other memory chains, and take the maximum value among the calculated results of all chains as the comprehensive cross-strength value of that node for calculation. The inertial retention potential describes the long-term persistence of historical behavior in the network; the larger the retention potential of a node, the more likely historical behavior is to persist in that node and its surrounding area.

[0103] Since historical behaviors may cyclically occur through multiple propagation paths, a behavior cycle index is established: the sum of the cyclic migration fluxes of all nodes divided by the sum of the total outward diffusion fluxes of all nodes. Here, cyclic migration flux refers to the flux value along the migration path from a given node, tracing the sequence of subsequent nodes along the migration flux direction, and if the node returns to its original behavioral region within no more than five time steps, the flux value along that migration path is included in the cyclic migration flux. The total outward diffusion flux is the sum of all migration fluxes output by the given node to all subsequent nodes in the network. Both cyclic migration flux and total outward diffusion flux originate from the edge table in the behavior evolution network, which records the migration flux values ​​between any two behavioral state nodes. The time step is the fixed sampling interval Δt (typically 1 ms) of the behavior evolution network; five time steps equal 5 ms. From node N... i2 Starting from the beginning, trace subsequent nodes hop by hop along the direction Φ>0. If the region to which any node belongs within 5 hops is related to N... i2 If the paths are identical, the sum of the fluxes of all edges along that path is included in the cyclic migration flux; if the same node is visited multiple times, only the first cycle is counted. The behavior cycle index is used to characterize the ability of historical behavior to repeatedly return to the same region. The larger the index, the stronger the tendency of historical behavior to form loops in the network, and the easier it is to accumulate in local regions. When the cycle index is zero, it means that there is no cyclic migration phenomenon in the network; when the cycle index is greater than one, it means that the total amount of cyclic migration has exceeded the total amount of outward diffusion, and historical behavior exhibits significant cyclic retention characteristics.

[0104] Further, the inertial accumulation potential is established: the inertial retention potential of a node is multiplied by the behavioral cycle index, and then multiplied by the ratio of the accumulation index of the region where the node is located to the maximum accumulation index of the entire network. The accumulation index originates from the structural spatial torsion field output by the stress-bearing analysis module, reflecting the comprehensive intensity of historical loads accumulating in that region. For nodes located in the intermediate region, i.e., not belonging to any receiving or competing zone, their regional accumulation index is taken as the median of the accumulation index distribution of the entire network. The inertial accumulation potential characterizes the ability of historical behavior to form a long-term accumulation in a local region. The larger the accumulation potential, the easier it is for historical behavior to form an accumulation effect in that region, continuously accumulating until it exceeds the structure's bearing capacity.

[0105] Subsequently, inertial persistence chains are identified: For each force flow memory chain, an intra-chain inertial continuity is established. This is calculated by dividing the sum of the inertial accumulation potentials of all nodes in the chain by the chain length, yielding the average inertial accumulation potential of that memory chain. If a force flow memory chain contains shared nodes belonging to multiple memory chains, the inertial accumulation potential of that node is calculated using the actual value for each chain it belongs to, without additional weighting or allocation. Intra-chain inertial continuity reflects the average strength of historical behavior persisting along the entire memory chain. When the intra-chain inertial continuity of a memory chain exceeds 75% of the maximum inertial continuity of the entire network, the memory chain is considered to form an inertial persistence chain. The maximum inertial continuity of the entire network is the maximum value of the intra-chain inertial continuity after traversing all force flow memory chains. Inertial persistence chains represent propagation paths where historical behavior has departed from the original operating condition but still survives in subsequent operating conditions for a long period. These paths are the survival channels of historical behavior in the behavior network.

[0106] Then, future expansion projections are performed on the identified inertial loitering chains. For the terminal node in each inertial loitering chain, that is, the last behavioral state node in the chain in time sequence, a future expansion potential is established: the inertial accumulation potential of the terminal node is multiplied by the total output migration flux of that node, and when the input flux is zero, the output / input ratio degenerates into the output flux (because the denominator is 1). In this way, the future expansion potential can still reflect the pure output capability.

[0107] If the output flux is zero, the future expansion potential equals the inertial accumulation potential, indicating that the node no longer diffuses but remains in place. The ratio of the output migration flux to the inertial accumulation potential is used. Here, the total output migration flux is the sum of the migration flux output by this node to all subsequent nodes, and the total input migration flux is the sum of the migration flux received by this node from all preceding nodes. Specifically, future expansion potential = inertial accumulation potential × (1 + total output migration flux) / (total input migration flux + 1), with one added to the denominator to prevent division by zero; when the total output migration flux is zero, the future expansion potential degenerates into the inertial accumulation potential. The future expansion potential predicts the ability of historical behavior to continue spreading to areas where it has not yet occurred; if the total output flux is zero, the future expansion potential equals the inertial accumulation potential, meaning that historical behavior stops spreading but still persists at the current node.

[0108] Further, a structural risk occupancy degree is established: the sum of all future expansion potentials is divided by the sum of all future expansion potentials, plus the sum of the influence values ​​of all stress fracture zones. The influence value of the stress fracture zone is the average stress bearing strength of all nodes within each fracture zone in the structural spatial torsion field output by the stress bearing analysis module. The total influence value is obtained by summing the influence values ​​of all fracture zones. The structural risk occupancy degree evaluates the balance between the historical behavior diffusion capacity and the structural release capacity. The closer the occupancy degree is to one, the greater the historical behavior diffusion capacity is than the structural release capacity, indicating a higher risk; the closer the occupancy degree is to zero, the more sufficient the structural release capacity is to mitigate the diffusion of historical behavior. A structural risk occupancy degree greater than 0.7 is considered a high-risk occupancy state.

[0109] Subsequently, fault type mapping is performed based on the distribution of inertial stagnation chains in the spatial torsion field. For any inertial stagnation chain, the proportion of nodes passing through stress accumulation zones to the total number of nodes in the chain is counted. If this proportion exceeds 60%, a motion jamming risk zone is identified, which physically means that historical behavior has accumulated in local stress accumulation zones for a long time, leading to a decrease in the fork's degrees of freedom. The ratio of the sum of stress bearing strength of all stress-bearing nodes to the sum of all inertial accumulation potentials is calculated. If this ratio is greater than 1.3, a damping drift risk zone is identified, which physically means that the amplification effect of historical behavior exceeds the structure's bearing capacity, causing the damping response to gradually deviate from the design state. The ratio of the sum of all path memory strengths to the sum of all memory chain intersection strengths is calculated. If this ratio is greater than 0.8, a lateral instability risk zone is identified, which physically means that strong coupling effects occur between multiple historical load propagation paths, leading to abnormal amplification of the lateral motion response. The detection method checks whether the behavioral cycle index is greater than 1.2 and whether the structural risk occupancy is greater than 0.7. If both conditions are met, a rebound anomaly risk zone is identified. Physically, this means that historical behaviors continue to flow back and occupy a large amount of structural space, preventing the rebound process from quickly returning to a normal state. Risk levels are classified based on the extent to which each indicator exceeds its threshold: exceeding the threshold by less than 10% is considered low risk; exceeding by 10% to 30% is considered medium risk; and exceeding by more than 30% is considered high risk.

[0110] After completing all risk identifications, a performance offset risk map is built. This map contains three core pieces of information: the complete set of inertial loiter chains, the complete set of fault risk regions, and the risk level corresponding to each risk region. The risk map's data structure uses key-value pairs for easy access by subsequent modules.

[0111] Subsequently, reverse tracing is performed based on the inertial stagnation chain, force flow memory chain, and stress bearing nodes corresponding to the risk area. The specific implementation of reverse tracing is as follows:

[0112] Starting from the identified risk areas, determine the set of inertial lingering chain numbers that pass through the areas;

[0113] For each inertial retention chain, obtain its corresponding original force flow memory chain number;

[0114] Traverse the node sequence of the force flow memory chain from the end to the beginning in reverse order to find the node with the greatest stress bearing strength in the chain, which is the main bearing hub in the chain.

[0115] Based on the location of the stress-bearing node in the behavioral evolution network, the source working condition segment that initially caused the node's historical impact is traced back by querying the working condition start timestamp and working condition type label associated with the node. The entire reverse tracing path can be represented as a complete link from the risk area to the inertial retention chain, then to the force flow memory chain, and finally to the stress-bearing node. Through the above reverse tracing, the number, working condition type, and occurrence time of the historical working condition segment that initially caused the impact, as well as the key behavioral node number, spatial location, and bearing strength value, and the sequence of regions traversed by the key propagation path are finally determined, forming a complete defect source tracing result, clearly indicating which historical working condition's legacy impact ultimately led to the formation of the currently identified risk area.

[0116] Finally, a lower limit judgment index is established based on structural risk occupancy, spatial distortion index, inertial continuity, and risk map. The index is calculated as follows: the lower limit judgment index equals structural risk occupancy multiplied by the first weighting coefficient, plus spatial distortion index multiplied by the second weighting coefficient, plus inertial continuity multiplied by the third weighting coefficient, plus the comprehensive value of the risk map multiplied by the fourth weighting coefficient; the sum of these four weighting coefficients is one. The specific values ​​of each weight are calibrated using the analytic hierarchy process (AHP) based on fork bench durability test data. The typical initial value is 0.25 for each weight, and adjustments can be made according to different fork models and operating conditions. Specifically, the calibration uses 100,000-cycle bench durability test data from at least 50 sets of the same model of front fork. Multiple linear regression is performed using performance degradation indicators such as damping force attenuation rate, rebound time offset, and lateral stiffness change as dependent variables, and four intermediate indicators of this module as independent variables. The normalized regression coefficients are then used as weights. In the initial stage when there is no historical sample data for new models, each of the four weights is set to 0.25. Subsequently, after accumulating 20 sets of real vehicle data with complete road test results, the regression is updated again. The comprehensive value of the risk map is calculated by quantifying and scoring the risk level of all identified risk areas: high risk is scored as three points, medium risk as two points, and low risk as one point. Then, the scores of all risk areas are summed and divided by the total number of risk areas. When the failure rate judgment index is greater than the preset failure rate judgment threshold, the front fork is deemed unqualified for failure testing and should not be installed in the vehicle. When the failure rate judgment index is less than or equal to the threshold, the front fork is deemed qualified for failure testing and can be installed in the vehicle normally. The determination of the judgment threshold is based on the 95th quantile of the judgment index distribution of a large number of qualified fork samples. For new models without historical sample data, simulation data can be used as a substitute, or an initial threshold can be set based on the design target value. Subsequent iterations and calibrations are performed using data accumulated from actual vehicles. Specifically, the initial threshold for new models is set to 0.6 (dimensionless). After every 100 sets of samples that pass 5000km of actual road testing, the 95th quantile of their judgment index is calculated as a candidate threshold, and an exponentially weighted moving average (EWMA, smoothing coefficient α=0.3) is used to smooth and update the threshold. If defective products appear continuously, the original threshold is temporarily maintained and production anomalies are immediately investigated.

[0117] The final output consists of three items: the first is a performance offset risk map, which fully records the distribution location, risk type, and risk level of the risk areas;

[0118] The second item is the defect tracing results, which clearly indicates the historical operating condition segment number, the location of key behavioral nodes, and the key propagation path that caused the current risk;

[0119] The third item is the fork failure determination result, which includes a qualitative conclusion of whether it is qualified or unqualified, as well as a quantitative value of the failure determination index.

[0120] Please see Figure 2 As shown, another aspect of the present invention provides a method for detecting the off-line condition of a motorcycle front fork shock absorber, comprising the following steps:

[0121] Step 1: Sensors are deployed on the target vehicle model to synchronously collect braking, steering, road surface undulation, and continuous impact data at a frequency of at least 1 kHz. These data are overlaid and merged into composite segments, resulting in a set of operating conditions. At the end of each segment, stroke, speed, acceleration, damping force, spring force, oil pressure, and temperature are extracted. The residual amount is calculated using a static equilibrium benchmark, and each dimension is normalized to full scale. For subsequent segments, the inherited strength is calculated: the minimum absolute value is taken for segments in the same direction, and zero is taken for segments in opposite directions. The sum is then divided by the residual amount norm. The winding coefficient is then calculated: all subsequent inherited strengths are multiplied by the time-exponential decay factor and summed to construct a weaving field.

[0122] Step 2: Real-time acquisition of compression trajectory, rebound trajectory, lateral drift, and structural micro-vibration combinations to form a behavior vector. The comprehensive period is obtained by weighting the decay periods of each historical legacy quantity according to its dimension and weight, and an activation potential is constructed: the legacy quantity is multiplied by the time exponential decay accumulation. The competition intensity is calculated: the product of the two activation potentials is divided by the Euclidean distance of the behavior vector plus a stability coefficient. On an unloaded test bench, the 95th percentile of the competition intensity distribution is taken as the acceptance threshold, and the 99.5th percentile as the competition threshold, dividing the acceptance zone and the competition zone. Cross-regional migration flux is statistically analyzed, the return flow index is calculated, and the data are integrated into a behavior evolution network.

[0123] Step 3: Calculate residual potential for each node: Divide the migration flux to all subsequent nodes by the time step difference, multiply by the return flow index of the subsequent nodes, and sum them up. Calculate convergence strength: Normalize and weight the preceding residual potentials according to migration fluxes; propagation gain equals convergence strength multiplied by return flow index; conservation equals the sum of residual potentials divided by the sum of propagation gains, less than 0.5 indicates dissipation, greater than 1.5 triggers an alarm. Search for paths with depth priority along edges where migration flux is greater than one-tenth of the maximum value, and calculate path memory strength: Multiply the normalized fluxes of each edge by the sum of residual potentials and divide by the path length; take three times the median of all path memory strengths as the threshold to filter force flow memory chains. Calculate inter-chain cross strength: Divide the number of common nodes by the number of nodes in the union of the two chains and multiply by the sum of the memory strengths of the two chains to construct a memory graph.

[0124] Step 4: Calculate the continuity potential for any two memory chains: Multiply the product of the residual potentials of the two chain nodes by the cross strength, sum them, divide by the absolute value of the average time center difference, and construct a global continuity matrix. Node continuity contribution: Residual potential multiplied by the sum of the continuity potentials of the chain to all other chains, then multiplied by the ratio of the chain's memory strength to its maximum strength; Diffusion index: Contribution multiplied by the output flux divided by the input flux; Stress continuity strength: Contribution multiplied by the diffusion index multiplied by the backflow index; nodes greater than 70% of the maximum strength are considered stress continuity nodes. Count the number of memory chains traversed in each region, multiply by the region's average continuity strength and average backflow index to obtain the clustering index; nodes greater than 80% of the maximum clustering index are considered stress clustering regions. Intra-chain adjacent node decay rate: Divide the memory value of the previous node by the difference between the previous and subsequent memory values, take twice the median decay rate of the entire network as the threshold, and calculate the continuous decay length. When three consecutive decay steps are reached and the memory strength of the third subsequent node is less than 10% of the maximum strength, it is considered a stress break zone. The above set is considered as a structural space torsion field, and the torsion index is equal to the sum of the accumulation index and the bearing strength, divided by the sum of the fracture zone attenuation length.

[0125] Step 5: Inertial Residual Factor for Each Node: Residual potential multiplied by the backflow index multiplied by one, plus the ratio of the receiving strength to the maximum strength; Inertial Retention Potential: Residual factor multiplied by one, plus the sum of the cross-crossing strengths with other chains, multiplied by the ratio of the memory strength of the chain to the maximum strength. Circulating flux is calculated for nodes that return to the same region within five steps along the migration edge. The circulating index equals the sum of circulating flux divided by the sum of outward diffusion flux. Inertial Accumulation Potential: Retention potential multiplied by the circulating index multiplied by one, plus the ratio of the regional accumulation index to the maximum accumulation index. The average inertial accumulation potential of each memory chain node is the inertial continuity within the chain; chains with an inertial retention potential greater than 75% of the maximum continuity are classified as inertial retention chains. The future expansion potential of the terminal node equals the inertial accumulation potential multiplied by one, plus the output flux divided by the input flux plus one; Structural risk occupancy equals the sum of all future expansion potentials divided by the sum of future expansion potentials plus the sum of the fracture zone influence values; a value greater than 0.7 indicates high risk. Based on the following criteria: Motion stagnation is judged when the proportion of the stagnant chain exceeding the accumulation zone exceeds 60%; damping drift is judged when the ratio of the total bearing strength to the total accumulation potential is greater than 1.3; lateral instability is judged when the ratio of the total memory strength to the total cross strength is greater than 0.8; and abnormal rebound is judged when the cycle index is greater than 1.2 and the occupancy degree is greater than 0.7. Risk maps are constructed by classifying these into low, medium, and high levels according to the magnitude of the exceedance. Reverse tracing: Identify the inertial stagnant chain from the risk zone; for the stress flow memory chain, reverse the process to find the node with the maximum stress bearing strength; associate it with the initial working condition stamp; and trace back to the source segment.

[0126] Step Six: The four indicators—structural risk occupancy, spatial distortion index, inertial continuity, and risk map composite value—are weighted and summed according to the weights determined by the multiple regression calibration of the bench durability test to obtain the failure criterion index. The initial threshold is 0.6. For every 100 sets of parts that have passed 5,000 kilometers of road testing, the 95th percentile of the criterion index is used, and the threshold is updated using an index-weighted moving average. If the criterion index is greater than the threshold, the part is considered unqualified; otherwise, it is considered qualified. Finally, the risk map, source tracing results, and failure conclusion are output, completing the testing.

[0127] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A system for detecting the off-line operation of a motorcycle front fork shock absorber, characterized in that: It includes a real vehicle operating condition segment creation module, a motion behavior acquisition module, a historical impact analysis module, a stress bearing analysis module, and an inertial retention prediction module; The real vehicle working condition segment establishment module collects multi-dimensional dynamic data and segments the working condition segment, calculates the normalized residual amount, inheritance strength and winding coefficient, and establishes a continuous weaving field for the working condition segment. The motion behavior acquisition module receives the weaving field and collects four types of dynamic responses: compression trajectory response, rebound trajectory response, lateral drift response, and structural micro-vibration response. These responses are combined into a behavior vector, and the behavior activation potential and competition intensity are established. The receiving area and competition area are divided through no-load calibration, and the migration flux and return flow index are statistically analyzed and integrated into a behavior evolution network. The historical impact analysis module receives the behavioral evolution network, calculates the behavioral residual potential, convergence intensity and propagation gain, calculates the conservation degree, filters the force flow memory chains through path search and memory intensity, calculates the cross-link intensity between chains, and establishes the force flow memory map. The stress bearing analysis module receives the force flow memory spectrum, calculates the connection potential to establish a global connection matrix, calculates the bearing contribution, diffusion index and stress bearing strength to determine the bearing node, determine the accumulation area and fracture area, establishes the structural space torsion field and calculates the torsion index. The inertial retention prediction module calculates the inertial residual factor, retention potential, circulation index, and accumulation potential, determines the inertial retention chain, calculates the expansion potential and risk occupancy, maps risk areas and classifies them, establishes a performance deviation risk map, traces back to the source, and obtains the lower limit judgment index by weighted summation of four indicators: structural risk occupancy, spatial distortion index, inertial continuity, and risk map. It compares the index with the threshold and outputs the conclusion of whether the test is qualified, as well as the risk map and source tracing results.

2. The off-line inspection system for a motorcycle front fork shock absorber according to claim 1, characterized in that, The real-vehicle operating condition segment establishment module collects multi-dimensional dynamic data by deploying brake pressure sensors, wheel speed sensors, steering angle sensors, steering torque sensors, fork displacement sensors, vehicle acceleration sensors, vehicle attitude sensors, and fork end acceleration sensors on the target vehicle model; synchronously records the data according to a unified time base to form a continuous riding data stream; and performs operating condition identification and time window segmentation on the continuous riding data stream based on braking deceleration threshold, steering angle change rate threshold, and road excitation amplitude threshold to obtain a set of operating condition segments and establish the motion state space corresponding to each operating condition segment.

3. The off-line inspection system for a motorcycle front fork shock absorber according to claim 2, characterized in that, The motion state space includes the fork compression stroke, compression speed, compression acceleration, damping force, spring force, internal oil pressure, and fork internal temperature. The real vehicle working condition segment establishment module establishes the working condition segment residual amount based on the actual motion state vector and static equilibrium state vector at the end of the working condition segment. The normalized residual amount is obtained by normalizing the full-scale values ​​of the sensors corresponding to each physical dimension. The inheritance strength is calculated based on the overlap between the normalized residual amount and the starting motion state vector of the subsequent working condition segment, and the working condition segment winding coefficient is established based on each inheritance strength. The real vehicle working condition segment establishment module establishes a continuous weaving field for working condition segments based on the winding coefficients of all working condition segments. The continuous weaving field for working condition segments uses working condition segments as nodes and the winding coefficients of working condition segments as edge weights to record the cross-working condition inheritance relationship of historical loads between different working condition segments and sends it to the motion behavior acquisition module.

4. The off-line inspection system for a motorcycle front fork shock absorber according to claim 1, characterized in that, The motion behavior acquisition module collects compression trajectory response, rebound trajectory response, lateral drift response, and structural micro-vibration response, and establishes behavior vectors; it establishes behavior activation potential based on historical working condition residuals and corresponding decay periods; it establishes behavior competition matrix based on behavior activation potential and distance relationship between behavior vectors; it determines acceptance threshold and competition threshold based on no-load calibration results, and divides behavior acceptance zone and behavior competition zone according to competition intensity.

5. The off-line inspection system for a motorcycle front fork shock absorber according to claim 4, characterized in that, The motion behavior acquisition module establishes the behavior migration flux based on the behavior activation potential, behavior change, and competition intensity; it establishes the behavior return flow index based on the total migration flux entering the behavior area and the total migration flux leaving the behavior area; it establishes the behavior evolution network based on the behavior receiving area, behavior competition area, behavior migration flux, and behavior return flow index; when the behavior return flow index is continuously greater than one and the duration exceeds the corresponding decay period, the corresponding behavior area is marked as a stagnation area.

6. The off-line inspection system for a motorcycle front fork shock absorber according to claim 1, characterized in that, The historical impact analysis module establishes behavioral residual potential based on behavioral migration flux and behavioral backflow index; establishes behavioral convergence intensity based on behavioral residual potential and migration flux; establishes behavioral propagation gain based on behavioral convergence intensity and behavioral backflow index; establishes behavioral impact conservation degree based on the total amount of all behavioral residual potential and the total amount of all behavioral propagation gain; and determines the release state, conservation state, or abnormal amplification state of historical behavior based on the behavioral impact conservation degree. The historical impact analysis module uses a depth-first search approach to search for paths in the behavioral evolution network; it establishes path memory strength based on behavioral migration flux, behavioral residual potential, and path length; it selects force flow memory chains based on path memory strength; it establishes memory chain cross strength based on the number of common behavioral state nodes between different force flow memory chains and path memory strength; and it establishes a force flow memory graph based on force flow memory chains, path memory strength, and memory chain cross strength.

7. The off-line inspection system for a motorcycle front fork shock absorber according to claim 1, characterized in that, The stress-bearing analysis module establishes the force flow continuity potential based on behavioral residual potential, memory chain cross strength, and memory chain average time center; establishes a global continuity matrix based on all force flow continuity potentials; establishes the continuity contribution based on behavioral residual potential, force flow continuity potential, and path memory strength; and establishes a diffusion index based on the continuity contribution and behavioral migration flux. The stress bearing strength is established based on the bearing contribution, diffusion index, and behavioral backflow index, and the stress bearing nodes are identified based on the stress bearing strength.

8. The off-line inspection system for a motorcycle front fork shock absorber according to claim 7, characterized in that, The stress bearing analysis module establishes an accumulation index based on the number of force flow memory chains passing through the behavior area, the average stress bearing intensity of the area, and the average return flow index of the area, and identifies stress accumulation areas based on the accumulation index. Stress fracture zones are identified based on memory decay rate and continuous decay length. The structural spatial torsion field is established based on the stress bearing nodes, stress accumulation zones, and stress fracture zones, and the structural spatial torsion index is established based on the accumulation index, stress bearing strength, and continuous attenuation length.

9. The off-line inspection system for a motorcycle front fork shock absorber according to claim 1, characterized in that, The inertial retention prediction module establishes an inertial retention factor based on behavioral residual potential, behavioral backflow index, and stress bearing strength; establishes an inertial retention potential based on the inertial retention factor, memory chain cross strength, and path memory strength; establishes a behavioral circulation index based on cyclic migration flux and total outward diffusion flux; establishes an inertial accumulation potential based on the inertial retention potential, behavioral circulation index, and accumulation index; identifies inertial retention chains based on the inertial accumulation potential; establishes structural risk occupancy based on future expansion potential, forming a performance deviation risk map; and establishes a dropout judgment index based on structural risk occupancy, structural space distortion index, inertial continuity, and risk map, and outputs a risk map, defect tracing results, and dropout detection conclusions.

10. A method for detecting the off-line condition of a motorcycle front fork shock absorber, characterized in that... The specific steps of implementing the off-line detection system for a motorcycle front fork shock absorber according to any one of claims 1-9 are as follows: Step 1: Collect braking, steering, road surface undulation and continuous impact data, segment them to form a set of working condition segments; calculate the normalized residual amount, inherited strength and winding coefficient, and establish a continuous weaving field for the working condition segments; Step 2: Collect compression trajectory, rebound trajectory, lateral drift, and structural micro-vibration data to form behavior vectors; calculate behavior activation potential and competition intensity, and divide the behavior acceptance zone and behavior competition zone; statistically analyze migration flux and backflow index to establish a behavior evolution network; Step 3: Calculate the behavioral residual potential, convergence strength, propagation gain, and conservation degree based on the behavioral evolution network; perform path search and calculate path memory strength to filter force flow memory chains; Calculate the cross strength of memory chains and establish a force flow memory map; Step 4: Calculate the force flow continuity potential based on the force flow memory map and establish a global continuity matrix; Calculate the bearing contribution, diffusion index, and stress bearing strength to identify stress bearing nodes; Identify stress concentration zones and stress fracture zones, establish the structural spatial torsion field, and calculate the spatial torsion index; Step 5: Calculate the inertial residual factor, inertial retention potential, behavioral cycle index, and inertial accumulation potential; identify inertial retention chains and deduce future expansion potential; calculate the structural risk occupancy, identify motion jamming, damping drift, lateral instability, and abnormal rebound risks, establish a performance deviation risk map, and complete defect source tracing. Step Six: Weight and fuse the structural risk occupancy, spatial distortion index, inertial continuity, and risk map comprehensive value to obtain the lower limit judgment index; Based on dynamically updated thresholds, the system outputs risk maps, source tracing results, and offline testing conclusions.

Images