Forklift axle fatigue safety life prediction method and system based on load spectrum

By acquiring impact response data based on load conditions and impact energy decay characteristics, the cumulative fatigue damage degree is corrected, solving the problem of difficulty in distinguishing the sources of damage contribution in existing methods, and improving the accuracy and reliability of forklift axle fatigue life prediction.

CN122263275BActive Publication Date: 2026-11-10ZHEJIANG LINDE AXLE CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202610728475.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2026-05-26
Publication Date
2026-11-10
Estimated Expiration
2046-05-26

AI Technical Summary

Technical Problem

Existing methods for predicting the fatigue life of forklift axles based on load spectra are unable to effectively distinguish the sources of damage under complex working conditions, resulting in insufficient accuracy of prediction results and difficulty in supporting refined maintenance decisions and differentiated risk control.

Method used

By acquiring impact response data on vehicle load status, impact amplitude, and impact energy attenuation characteristics, the impact event is characterized in multiple dimensions. Based on the attenuation characteristics and load status, hardware state change characteristics are extracted, cumulative fatigue damage is corrected, and remaining life is predicted.

Benefits of technology

It significantly improves the accuracy and reliability of fatigue life prediction, can more accurately reflect changes in hardware condition, and supports refined maintenance decisions and differentiated risk control.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122263275B_ABST
    Figure CN122263275B_ABST
Patent Text Reader

Abstract

The present application relates to the technical field of engineering machinery, in particular to a forklift axle fatigue safety life prediction method and system based on a load spectrum. The technical problems of difficulty in effectively distinguishing the damage contribution of different sources and insufficient accuracy of life prediction results are solved. The method comprises: obtaining impact response data of a plurality of impact events occurring in the operation process of the vehicle; determining a cumulative fatigue damage degree of the plurality of impact events based on the impact amplitude of each impact event and the material fatigue characteristics of the vehicle structure; determining a first feature for characterizing the change of the vehicle hardware state based on the impact energy attenuation characteristics of each impact event and the vehicle load state when each impact event occurs; correcting the cumulative fatigue damage degree based on the first feature to determine the corrected cumulative fatigue damage degree; and predicting the remaining life of the vehicle structure based on the corrected cumulative fatigue damage degree. The present application is used in the forklift axle life prediction scene.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of engineering machinery technology, specifically to a method and system for predicting the fatigue safety life of forklift axles based on load spectrum. Background Technology

[0002] In the field of construction machinery, especially in the lifecycle management of industrial vehicles such as forklifts, the axle, as a critical load-bearing component, directly affects the operational safety and economic efficiency of the entire vehicle. Accurately predicting the remaining life of the axle, and thus developing predictive maintenance plans, is a core technical requirement for ensuring reliable equipment operation and avoiding unexpected downtime and major safety accidents. Currently, this field commonly employs fatigue life prediction methods based on load spectra. These methods collect vibration and stress data during operation by placing sensors on the axle, and then use models such as linear cumulative damage theory to statistically analyze the collected load time history, thereby estimating the structural damage accumulation and remaining life. However, existing prediction technologies face significant challenges in the complex real-world forklift operating scenarios. Forklift operating conditions are diverse, and their loads are simultaneously affected by multiple factors, including the performance degradation of hardware components such as tires and the driver's operating habits. Existing methods struggle to effectively distinguish the contributions of these different damage sources, and the constructed prediction models are overly idealistic. Insufficient decoupling capability for mixed excitation sources in actual operating conditions leads to insufficient accuracy in the final life prediction results, making it difficult to support refined maintenance decisions and differentiated risk control. Summary of the Invention

[0003] To address the shortcomings of existing methods in effectively distinguishing damage contributions from different sources and their insufficient decoupling capability for mixed excitation sources under actual working conditions, which leads to inaccurate life prediction results and difficulty in supporting refined maintenance decisions and differentiated risk control, this invention aims to provide a method and system for predicting the fatigue safety life of forklift axles based on load spectrum. The specific technical solution adopted is as follows: In a first aspect, the present invention provides a method for predicting the fatigue safety life of a forklift axle based on a load spectrum. The method includes: acquiring impact response data of multiple impact events occurring during vehicle operation; the impact response data includes: the vehicle's load state at the time of each impact event, the impact amplitude of each impact event, and the impact energy decay characteristics of each impact event; determining the cumulative fatigue damage degree of the multiple impact events based on the impact amplitude of each impact event and the material fatigue characteristics of the vehicle structural components; determining a first feature to characterize changes in the vehicle's hardware state based on the impact energy decay characteristics of each impact event and the vehicle's load state at the time of each impact event; correcting the cumulative fatigue damage degree based on the first feature to determine the corrected cumulative fatigue damage degree; and predicting the remaining life of the vehicle structural components based on the corrected cumulative fatigue damage degree.

[0004] In conjunction with the first aspect mentioned above, in one possible implementation, the method specifically includes: real-time monitoring of the vehicle's vertical vibration signal; determining an impact event when the amplitude of the vertical vibration signal exceeds a preset impact threshold; acquiring the vehicle's load status at the time of the impact event in response to its occurrence; determining the impact amplitude of the impact event based on waveform segments of the vertical vibration signal within a preset time period after the impact event; verifying the attenuation effectiveness of the waveform segments; the attenuation effectiveness verification is used to determine whether the waveform segments conform to a preset free attenuation model; and extracting impact energy attenuation characteristics from the waveform segments that have passed the attenuation effectiveness verification.

[0005] In conjunction with the first aspect mentioned above, in one possible implementation, the method specifically includes: determining whether the ratio of the first peak to the second peak is greater than a preset first threshold; the first peak and the second peak are two temporally adjacent peaks in the waveform segment; determining whether the time interval between the first peak and the second peak is within a preset inherent period range; and determining that the waveform segment passes the attenuation validity verification when the ratio is greater than the first threshold and the time interval is within the inherent period range.

[0006] In conjunction with the first aspect mentioned above, in one possible implementation, the impact response data further includes: the vehicle operating state at the time of the impact; the method further includes: determining a second feature to characterize the driver's operating style based on the vehicle operating state, load state, and the first feature.

[0007] In conjunction with the first aspect mentioned above, in one possible implementation, the method specifically includes: calculating the expected impact amplitude based on the operating state, load state, and a first feature using a preset dynamic model; determining the deviation value between the measured impact amplitude and the expected impact amplitude for each impact event; and determining the deviation value among the deviation values ​​for each impact event that can characterize the distribution characteristics of the deviation value as a second feature.

[0008] In conjunction with the first aspect mentioned above, in one possible implementation, the method specifically includes: determining the total vehicle mass based on the vehicle's load state at the time of each impact event; constructing a load-damping relationship coordinate system based on the total vehicle mass and the impact energy attenuation characteristics of each impact event; determining multiple location points of each impact event in the load-damping relationship coordinate system; determining a fitted straight line matching the multiple location points using a preset fixed slope value as the slope constraint of the fitted straight line; the intercept of the fitted straight line being the current damping reference value; and determining a first feature based on the difference between the current damping reference value and the preset reference damping reference value.

[0009] In conjunction with the first aspect mentioned above, in one possible implementation, the method specifically includes: randomly selecting multiple candidate position points from multiple position points; determining multiple candidate fitting lines using a preset fixed slope value as the slope and passing through each candidate position point as a condition; counting the number of position points among the multiple position points whose distance to each candidate fitting line is less than a preset tolerance; and determining the candidate fitting line with the largest number of position points as the fitting line that matches the multiple position points.

[0010] In conjunction with the first aspect mentioned above, in one possible implementation, the method specifically includes: sorting the deviation values ​​corresponding to multiple impact events according to their numerical values; and selecting the deviation value at a preset percentage position in the sorting result as the second feature.

[0011] In conjunction with the first aspect described above, in one possible implementation, the method further includes: executing a closed-loop control strategy for the vehicle based on the first feature and the second feature.

[0012] Secondly, the present invention provides a forklift axle fatigue safety life prediction system based on load spectrum, used to implement the forklift axle fatigue safety life prediction method based on load spectrum mentioned in the first aspect above; the system includes: a data acquisition module, used to acquire impact response data of multiple impact events occurring during vehicle operation; the impact response data includes: the vehicle load state at the time of each impact event, the impact amplitude of each impact event, and the impact energy decay characteristics of each impact event; a cumulative fatigue damage determination module, used to determine the cumulative fatigue damage degree of multiple impact events based on the impact amplitude of each impact event and the material fatigue characteristics of the vehicle structural components; a feature extraction module, used to determine a first feature for characterizing changes in the vehicle hardware state based on the impact energy decay characteristics of each impact event and the vehicle load state at the time of each impact event; a cumulative fatigue damage correction module, used to correct the cumulative fatigue damage degree based on the first feature and determine the corrected cumulative fatigue damage degree; and a life prediction module, used to predict the remaining life of the vehicle structural components based on the corrected cumulative fatigue damage degree.

[0013] The present invention has the following beneficial effects: This invention achieves multi-dimensional characterization of impact events by acquiring impact response data including vehicle load status, impact amplitude, and impact energy attenuation characteristics. Based on this, two data lines are processed in parallel: one calculates fatigue damage based on cumulative impact amplitude, and the other extracts a first feature representing changes in hardware condition based on attenuation characteristics and load status. Finally, the first feature is used to correct the cumulative fatigue damage and predict remaining life. This overall framework solves the technical problem of traditional methods ignoring hardware condition changes, leading to overly optimistic predictions. By introducing hardware condition features to correct the basic damage, the life prediction results are made more consistent with the laws of physical evolution, significantly improving the accuracy and reliability of the predictions. This addresses the technical problem of existing methods' inability to effectively distinguish damage contributions from different sources and their insufficient ability to decouple mixed excitation sources under actual working conditions, resulting in insufficient accuracy in the final life prediction results and difficulty in supporting refined maintenance decisions and differentiated risk control. Attached Figure Description

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

[0015] Figure 1 This is a flowchart illustrating a method for predicting the fatigue safety life of forklift axles based on load spectrum, provided in one embodiment of the present invention. Figure 2 This is a schematic diagram of the system architecture of a forklift axle fatigue safety life prediction system based on load spectrum, provided in an embodiment of the present invention. Detailed Implementation

[0016] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of the forklift axle fatigue safety life prediction method and system based on load spectrum proposed by the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0017] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0018] The specific scheme of the fatigue safety life prediction method and system for forklift axles based on load spectrum provided by the present invention will be described in detail below with reference to the accompanying drawings.

[0019] Please see Figure 1 The diagram illustrates a flowchart of a method for predicting the fatigue safety life of a forklift axle based on a load spectrum, according to an embodiment of the present invention. The method includes the following steps S101-S105, which will be described in detail below.

[0020] S101. Obtain impact response data for multiple impact events that occur during vehicle operation.

[0021] The impact response data includes: the vehicle load status at the time of each impact event, the impact amplitude of each impact event, and the impact energy decay characteristics of each impact event.

[0022] In one possible implementation, the vertical vibration signals generated by the vehicle during operation are monitored in real time. When the vibration amplitude exceeds a preset impact threshold, an impact event is determined. In response to the impact event, the vehicle's load status at that moment is acquired, and vibration waveform segments are extracted from a preset time period after the impact to determine the impact amplitude. The attenuation effectiveness of the extracted waveform segments is verified, and impact energy attenuation features are extracted from the verified waveform segments. The load status, impact amplitude, and impact energy attenuation features together constitute the impact response data corresponding to the impact event. The impact response data from multiple impact events are accumulated to form a historical dataset, which is used for subsequent fatigue damage accumulation and hardware condition feature extraction.

[0023] Understandably, this step is fundamental in transforming raw vibration signals into structured data. Event triggering is achieved through impact thresholding, data correlation is ensured by synchronously acquiring load status, and waveform segments conforming to physical attenuation laws are selected through attenuation effectiveness verification, providing a high-quality data foundation for subsequent analysis.

[0024] S102. Based on the impact amplitude of each impact event and the material fatigue characteristics of the vehicle structural components, determine the cumulative fatigue damage degree of multiple impact events.

[0025] In one possible implementation, for multiple acquired impact events, based on the impact amplitude of each impact event and the material fatigue characteristics of the vehicle structural components, the single-instance fatigue damage caused by each impact event to the structure is determined. The single-instance fatigue damage corresponding to multiple impact events is then accumulated to obtain the cumulative fatigue damage degree under the combined effect of multiple impact events. This cumulative fatigue damage degree reflects the total fatigue damage degree caused to the axle structure by all impact events from the time the vehicle was put into service until the current moment.

[0026] For example, a mapping relationship between impact amplitude and single-instance fatigue damage can be established based on the fatigue characteristics of vehicle structural component materials. For each impact event, the corresponding single-instance fatigue damage value is determined based on its impact amplitude using this mapping relationship. The single-instance fatigue damage values ​​of all impact events are accumulated to obtain the cumulative fatigue damage degree. It should be noted that regardless of whether the impact event passes the attenuation effectiveness verification, its impact amplitude is included in the determination process of the cumulative fatigue damage degree to ensure the completeness of fatigue damage statistics.

[0027] S103. Based on the impact energy attenuation characteristics of each impact event and the vehicle load state at the time of each impact event, determine the first feature used to characterize the changes in vehicle hardware state.

[0028] In one possible implementation, for multiple acquired impact events, a correlation is constructed between the impact energy attenuation characteristics of each impact event and the vehicle's load state at the time of occurrence. Within this correlation, preset physical laws are used as constraints to extract benchmark parameters reflecting the current state of the hardware from the noise-containing impact event data. Based on the difference between this benchmark parameter and preset reference parameters, a first feature characterizing changes in the vehicle hardware state is determined. This first feature quantifies the degree of performance degradation of vehicle hardware (such as tires) due to factors such as aging and wear.

[0029] For example, a mapping model between load state and impact energy attenuation characteristics can be established based on the inherent physical law of damping characteristics changing with load. In this model, the impact energy attenuation characteristics under different load states should theoretically follow a specific distribution law. By fitting the baseline parameters corresponding to the current hardware state from actual data of multiple impact events and comparing them with the reference parameters corresponding to the initial hardware state, a first feature reflecting the degree of change in hardware state can be obtained. The larger the value of this first feature, the more severe the hardware performance degradation.

[0030] S104. Based on the first feature, the cumulative fatigue damage degree is corrected, and the corrected cumulative fatigue damage degree is determined.

[0031] In one possible implementation, the determined cumulative fatigue damage is corrected based on a first feature characterizing changes in the vehicle's hardware state, resulting in a corrected cumulative fatigue damage. This correction reflects the accelerating effect of hardware performance degradation on the future damage accumulation process, ensuring that the corrected cumulative fatigue damage reflects the nonlinear effect of hardware state changes on fatigue damage evolution.

[0032] For example, a correction coefficient can be determined based on the value of the first feature. This correction coefficient is positively correlated with the first feature; that is, the more severe the hardware aging, the larger the correction coefficient. The cumulative fatigue damage is then correlated with this correction coefficient to obtain the corrected cumulative fatigue damage. The corrected cumulative fatigue damage is larger than the uncorrected value, reflecting the accelerated damage effect caused by hardware aging.

[0033] For example, a correction coefficient can be determined based on the numerical value of the first feature. This correction coefficient is positively correlated with the first feature. Specifically, the correction coefficient is equal to the sum of a preset base and a product term, which is obtained by multiplying the first feature by a preset evolution coefficient. Multiplying the cumulative fatigue damage degree by this correction coefficient yields the corrected cumulative fatigue damage degree. Assuming the preset evolution coefficient is 0.5 and the preset base is 1, if the first feature is 0.5, then the correction coefficient is 1 plus the product of 0.5 and 0.5, which is 1.25. If the current cumulative fatigue damage degree is 0.4, then the corrected cumulative fatigue damage degree is 0.4 multiplied by 1.25, which equals 0.5. The corrected cumulative fatigue damage degree is 0.1 higher than the original 0.4, reflecting the accelerating effect of hardware aging on future damage. This corrected cumulative fatigue damage degree will be used for subsequent remaining life prediction.

[0034] S105. Based on the corrected cumulative fatigue damage, predict the remaining life of vehicle structural components.

[0035] In one possible implementation, based on the determined corrected cumulative fatigue damage level and a preset life assessment criterion, the remaining life of the vehicle structural component is predicted. This remaining life characterizes the expected duration or number of operating cycles that the vehicle structural component can continue to serve in its current hardware condition, taking into account historical damage accumulation and future hardware deterioration trends.

[0036] For example, the remaining life can be determined based on the relationship between the corrected cumulative fatigue damage and the structural failure threshold. The closer the corrected cumulative fatigue damage is to the failure threshold, the shorter the remaining life; conversely, the further away it is, the longer the remaining life. The predicted remaining life can be expressed as a percentage.

[0037] For example, a failure threshold of 1.0 can be set for a structural component, indicating that when the cumulative fatigue damage reaches 1.0, the component reaches its service life limit and can no longer be used. The remaining service life percentage is determined by the ratio of the difference between the corrected cumulative fatigue damage and the failure threshold to the failure threshold. The specific calculation formula is: Remaining service life percentage = (1.0 - Corrected cumulative fatigue damage) / 1.0 × 100%. If the corrected cumulative fatigue damage is 0.5, then the remaining service life percentage = (1.0 - 0.5) / 1.0 × 100% = 50%, indicating that the structural component is expected to have 50% of its service life remaining. If the corrected cumulative fatigue damage is 0.8, then the remaining service life percentage = (1.0 - 0.8) / 1.0 × 100% = 20%. The closer the corrected cumulative fatigue damage is to 1.0, the smaller the remaining service life percentage, indicating that the structural component is closer to failure.

[0038] The technical solution provided by the above embodiments can bring at least the following beneficial effects: This embodiment achieves multi-dimensional characterization of impact events by acquiring impact response data including vehicle load status, impact amplitude, and impact energy attenuation characteristics. Based on this, two data lines are processed in parallel: one is to calculate fatigue damage degree based on the cumulative impact amplitude, and the other is to extract the first feature representing hardware state changes based on attenuation characteristics and load status. Finally, the first feature is used to correct the cumulative fatigue damage degree and predict the remaining life. This overall framework solves the technical problem of traditional methods ignoring hardware state changes, leading to overly optimistic prediction results. By introducing hardware state features to correct the basic damage, the life prediction results are made more consistent with the laws of physical evolution, significantly improving the accuracy and reliability of the prediction. This solves the technical problem that existing methods struggle to effectively distinguish the contributions of damage from different sources, lack the ability to decouple mixed excitation sources under actual working conditions, resulting in insufficient accuracy of the final life prediction results and difficulty in supporting refined maintenance decisions and differentiated risk control.

[0039] In one possible implementation, the process of acquiring impact response data of multiple impact events that occur during vehicle operation can be specifically implemented through the following steps S201-S205, which will be described in detail below.

[0040] S201. Monitor the vertical vibration signal of the vehicle in real time. When the amplitude of the vertical vibration signal exceeds the preset impact threshold, determine that an impact event has occurred.

[0041] In one possible implementation, the vertical vibration signal generated by the vehicle during operation is monitored in real time. This vertical vibration signal is acquired by an accelerometer mounted on the drive axle, reflecting the impact response of road excitation transmitted from the tires to the axle during vehicle movement. The amplitude of the vertical vibration signal is continuously monitored and compared with a preset impact threshold. When the amplitude of the vertical vibration signal at a certain moment exceeds the preset impact threshold, an impact event is determined to have occurred at that moment, triggering subsequent data acquisition and processing procedures.

[0042] For example, the preset impact threshold can be set to 3g (g is the acceleration due to gravity). The preset impact threshold can be calibrated according to vehicle type, operating scenario, and structural fatigue sensitivity. Setting the threshold too low may lead to frequent triggering and data redundancy; setting it too high may miss impact events that substantially contribute to fatigue damage. By reasonably calibrating the impact threshold, a balance can be achieved between sensitivity and accuracy, ensuring effective capture of impact loads that have a significant impact on the fatigue life of the axle.

[0043] S202. In response to the occurrence of an impact event, obtain the vehicle load status at the moment of occurrence.

[0044] In one possible implementation, in response to an impact event, the vehicle's load status at the moment the impact event occurs is acquired. This load status characterizes the mass of cargo carried by the vehicle at the time of the impact and is a crucial parameter for correlating the impact characteristics with the load in subsequent analysis. Since the hydraulic system pressure fluctuates drastically during forklift operation, real-time oil pressure cannot be directly used to calculate the load. This embodiment employs a combination of static measurement and dynamic holding to maintain a globally valid load variable, and reads the current load status from this variable when an impact event is triggered.

[0045] For example, the vehicle status is continuously monitored during operation. When the vehicle is stationary and stable and meets preset weighing conditions, the current actual load is calculated based on the hydraulic pressure and updated to the load holding variable. If an impact event occurs during vehicle operation, the current load status is directly read from this load holding variable as the vehicle load status at the time of the impact event. When the hydraulic pressure is detected to be lower than a preset no-load threshold, the load holding variable is forcibly reset to an no-load state to avoid residual load data errors due to the vehicle not meeting the stationary condition after unloading.

[0046] S203. Determine the impact amplitude of the impact event based on waveform segments of vertical vibration signals within a preset time period after the impact event occurs.

[0047] In one possible implementation, in response to the triggering of an impact event, a waveform segment within a preset time period following the occurrence of the impact event is extracted from a high-frequency vibration signal queue, and the impact amplitude of the impact event is determined based on this waveform segment. This impact amplitude is used to characterize the transient load intensity generated by the impact on the axle structure and serves as the basic input for subsequent fatigue damage accumulation calculations.

[0048] For example, a high-frequency rolling data queue is maintained in memory to store vertical vibration acceleration data for the most recent few seconds. When an impact event is triggered, the moment when the acceleration reaches its peak is locked, and the acceleration amplitude corresponding to that moment is read from the high-frequency queue as the impact amplitude of the impact event. Simultaneously, a waveform segment of a preset duration (e.g., 0.5 seconds) following the impact peak is extracted for subsequent attenuation effectiveness verification and impact energy attenuation feature extraction. This preset duration must ensure that the free attenuation process after the impact is fully captured; it should not be too short, resulting in the loss of attenuation information, nor too long, introducing subsequent irrelevant vibration interference.

[0049] S204. Verify the effectiveness of attenuation on the waveform segment.

[0050] In one possible implementation, the attenuation effectiveness of waveform segments within a preset time period following the impact event is verified. This verification determines whether the waveform segments conform to a preset free attenuation model, thereby filtering out effective impact waveforms that truly reflect the system's damping characteristics and eliminating invalid data affected by random excitation from the continuous road surface.

[0051] For example, a waveform segment is analyzed to identify the first and second temporally adjacent peaks. The degree of amplitude attenuation and the time interval between the two peaks are assessed to determine whether the waveform conforms to the free attenuation law. When a waveform segment meets the preset attenuation conditions, it is deemed to have passed the attenuation validity verification and can be used for subsequent impact energy attenuation feature extraction. When a waveform segment does not meet the preset attenuation conditions, it is deemed to have failed the attenuation validity verification, and no attenuation features are extracted from it, but its impact amplitude is still retained for fatigue damage accumulation calculation.

[0052] S205. Extract the impact energy attenuation characteristics from the waveform segments that have passed the attenuation effectiveness verification.

[0053] In one possible implementation, impact energy attenuation characteristics are extracted from waveform segments that have passed attenuation effectiveness verification. These impact energy attenuation characteristics characterize the rate of energy dissipation after an impact event and are key parameters reflecting the damping characteristics of vehicle hardware (such as tires), which will be used for subsequent hardware state characteristic decoupling analysis.

[0054] For example, waveform segments that have passed attenuation effectiveness verification are analyzed, and the impact energy attenuation characteristics are determined based on the attenuation relationship between the amplitudes of adjacent peaks in the waveform. For waveforms that conform to the free attenuation law, the amplitudes of adjacent peaks exhibit a monotonically decreasing trend, and the attenuation rate is directly related to the damping characteristics of the system. By quantifying this attenuation rate, characteristic values ​​reflecting the current hardware state can be obtained. The larger the characteristic value, the faster the system dissipates energy and the better the damping performance; the smaller the characteristic value, the slower the system dissipates energy, the worse the damping performance, and the more severe the hardware aging.

[0055] The technical solution provided in the above embodiments can bring at least the following beneficial effects: This embodiment specifically defines the process of acquiring impact response data. This process achieves accurate triggering and identification of impact events by real-time monitoring of vertical vibration signals and setting impact thresholds. After the impact occurs, the method simultaneously acquires the load state and extracts waveform segments, ensuring the spatiotemporal correlation of the data. The method introduces an attenuation validity verification mechanism, extracting attenuation features only from waveform segments that conform to the free attenuation model. This mechanism effectively eliminates invalid data disturbed by random excitation from continuous road surfaces, solving the problem of algorithm collapse due to poor data quality in engineering applications, providing a high-quality data foundation for subsequent feature extraction, and ensuring the engineering feasibility of the entire technical solution.

[0056] In one possible implementation, the process of verifying the attenuation effectiveness of the waveform segment described above can be specifically implemented through the following S301-S303, which will be explained in detail below.

[0057] S301. Determine whether the ratio of the first peak to the second peak is greater than the preset first threshold.

[0058] Among them, the first peak and the second peak are two peaks that are adjacent in time in the waveform segment.

[0059] In one possible implementation, for a waveform segment that has passed the attenuation effectiveness verification, the first and second temporally adjacent peaks are identified. The ratio of the first peak to the second peak is determined, and it is judged whether this ratio is greater than a preset first threshold. This judgment is used to verify whether the waveform exhibits a significant energy attenuation trend, ensuring that the waveform conforms to the basic characteristics of a free attenuation model.

[0060] For example, the first peak p1 and the second peak p2 are two temporally adjacent peaks in the same direction within the waveform segment, determined chronologically. A preset first threshold can be set to 1.05, requiring p1 > 1.05p2. This threshold ensures that the first peak is significantly larger than the second peak, indicating that the system experienced substantial energy dissipation after the impact, rather than being subjected to continuous external excitation or noise interference. If p2 ≥ p1 or both are close, it indicates that the vehicle is under continuous road surface excitation, which does not conform to the free decay model, and the waveform segment will be deemed invalid.

[0061] S302. Determine whether the time interval between the first peak and the second peak is within the preset inherent period range.

[0062] In one possible implementation, for a waveform segment that has passed the attenuation effectiveness verification, the time interval between the first and second peaks is determined, and it is judged whether this time interval is within a preset inherent period range. This judgment is used to verify whether the oscillation frequency of the waveform conforms to the inherent dynamic characteristics of the system itself, so as to eliminate interference from high-frequency noise spikes or low-frequency unrelated vibrations.

[0063] For example, the time interval Δt between the first peak p1 and the second peak p2 reflects the time required for the system to complete half an oscillation cycle after the impact. The preset natural period range can be pre-calibrated based on the natural frequency of the vibration system composed of the vehicle's drive axle and tires. For a typical solid-tire forklift, the system's natural period is typically between 0.1 and 0.3 seconds. If Δt is within this range, it indicates that the waveform's oscillation frequency matches the system's inherent dynamic characteristics; if Δt is too short (e.g., less than 0.1 seconds), it may be due to high-frequency noise interference; if Δt is too long (e.g., greater than 0.3 seconds), it may be due to long-wave excitation from the road surface or other unrelated vibrations.

[0064] S303. When the ratio is greater than the first threshold and the time interval is within the inherent period range, the waveform segment is determined to have passed the attenuation validity verification.

[0065] In one possible implementation, when the ratio of the first peak to the second peak is greater than a preset first threshold, and the time interval between the first peak and the second peak is within a preset inherent period range, the current waveform segment is determined to have passed the attenuation validity verification based on the satisfaction of these two conditions. This determination result indicates that the waveform segment conforms to the basic characteristics of the free attenuation model, can truly reflect the energy dissipation process of the system after the impact, and can be used for subsequent impact energy attenuation feature extraction.

[0066] Understandably, the verification of the attenuation effectiveness of waveform segments employs a dual-condition joint judgment approach. A crest-to-peak ratio greater than the first threshold ensures that the waveform exhibits a significant energy attenuation trend, excluding invalid waveforms affected by continuous excitation interference; the time interval being within the natural period range ensures that the waveform's oscillation frequency matches the system's inherent dynamic characteristics, excluding interference from high-frequency noise and low-frequency unrelated vibrations. Only when both conditions are simultaneously met is it confirmed that the waveform segment indeed originates from a free attenuation process following an effective impact, rather than from random road surface excitation or sensor noise.

[0067] The technical solution provided in the above embodiments can bring at least the following beneficial effects: This embodiment defines the specific judgment criteria for attenuation effectiveness verification. By identifying the ratio of adjacent peaks and the time interval in a waveform segment, a dual judgment criterion is constructed. A peak ratio greater than a preset threshold ensures that the waveform exhibits a clear energy attenuation trend, which conforms to the basic characteristics of a free attenuation model; a time interval within the natural period effectively eliminates interference from high-frequency noise spikes. This judgment criterion based on physical laws is simple, clear, and easy to implement in engineering. It can accurately screen out effective impact waveforms that meet the analysis requirements under complex random road surface excitation backgrounds, providing a reliable data source for subsequent damping feature extraction.

[0068] In one possible implementation, the impact response data also includes: the vehicle's operating state at the time of the impact; it is also necessary to determine a second feature to characterize the driver's operating style, which can be specifically implemented through the following S401, which will be described in detail below.

[0069] S401. Based on the vehicle's operating status, load status, and the first feature, determine the second feature used to characterize the driver's operating style.

[0070] In one possible implementation, for multiple acquired impact events, a second feature is determined based on the vehicle's operating state, load state, and a predetermined first feature at the time of each impact event to characterize the driver's operating style. This second feature is used to quantify the driver's aggressiveness during braking, bump crossing, and other maneuvers, reflecting the additional impact intensity beyond normal road surface response and hardware aging effects.

[0071] For example, for each impact event, based on its operating and load conditions, and combined with the hardware aging degree reflected by the first feature, the expected impact amplitude of the impact event under normal operation can be calculated using a preset dynamic model. The actual impact amplitude of the impact event is compared with the expected impact amplitude to obtain a deviation value reflecting the degree of operational abnormality. Statistical analysis is performed on the deviation values ​​corresponding to multiple impact events to extract a feature value that characterizes the driver's operating style as a second feature. The larger the value of this second feature, the more aggressive the driver's operating style, and the stronger the additional destructive impact on the axle structure.

[0072] Understandably, this step is a crucial step in quantifying operational behavior, building upon the established hardware state characteristics. By introducing operational and load states, and combining the impact of the first characteristic on transmission characteristics, the system can isolate abnormal impact components caused by operational behavior from the measured impact amplitude, achieving decoupling attribution between hardware aging and operational behavior. The resulting second characteristic will be used for subsequent differentiated control strategies, providing a basis for decision-making regarding intervention measures for operational problems, and also improving the system's comprehensive understanding of the sources of axle damage.

[0073] The technical solution provided by the above embodiments can bring at least the following beneficial effects: This embodiment introduces vehicle operating status and a second feature on the basic solution, realizing attribution analysis of damage sources. An operating status dimension is added to the impact response data, and a second feature characterizing the driver's operating style is extracted based on the operating status, load status, and the determined first feature. This can distinguish whether the impact load borne by the axle originates from hardware aging or operational behavior, solving the technical problem that traditional monitoring systems cannot identify the source of damage. Through the decoupling of the two orthogonal features, a decision-making basis is provided for subsequent differentiated control strategies, achieving accurate differentiation between hardware problems and operational problems.

[0074] In one possible implementation, the process of determining the second feature to characterize the driver's operating style based on the vehicle's operating state, load state, and the first feature can be specifically implemented through the following S501-S503, which will be described in detail below.

[0075] S501. Based on the operating state, load state, and first characteristic, calculate the expected impact amplitude using a preset dynamic model.

[0076] In one possible implementation, for each impact event, based on the vehicle's operating state, load state, and a predetermined first characteristic at the time of the impact event, the expected impact amplitude corresponding to the impact event is calculated using a preset dynamic model. This expected impact amplitude is used to characterize the theoretically expected impact response intensity under the current operating intensity, vehicle load, and hardware aging level, serving as a benchmark for subsequent assessment of the degree of abnormal operating behavior.

[0077] For example, vehicle operating status may include parameters reflecting the driver's intention, such as braking deceleration and driving speed. Load status reflects the actual load on the vehicle at the time of the impact. The first feature quantifies the degree of hardware aging, reflecting the influence of components such as tires on impact transmission characteristics. A preset dynamic model comprehensively considers the above factors and establishes a mapping relationship from operating input to impact output. In this model, the greater the operating intensity, the heavier the load, and the more severe the hardware aging, the greater the expected impact amplitude. Through this model, the expected impact value that conforms to the current vehicle state and hardware conditions can be calculated for each impact event.

[0078] For example, the expected impact amplitude Satisfy the following formula 1: Formula 1 in, The total mass of the vehicle; The braking-shock transmission gain coefficient (calibrated experimentally, for example, a value of 0.02). This is the hardening influence coefficient; The aging index is used; the expected impact amplitude is proportional to the braking intensity and mass, and increases with aging. This reflects the amplification of transfer rate due to aging; It is the theoretical impact amplitude, used to compare with the actual value to extract operational deviations.

[0079] S502. Determine the deviation between the measured impact amplitude and the expected impact amplitude for each impact event.

[0080] In one possible implementation, for each impact event, a deviation value is determined based on the measured impact amplitude of the event and the expected impact amplitude calculated through a dynamic model. This deviation value is used to quantify the degree of difference between the actual impact response and the theoretical expectation, reflecting the additional impact component caused by abnormal driver behavior, in addition to normal operation and hardware aging factors.

[0081] For example, for each impact event, the difference between the measured impact amplitude and the corresponding expected impact amplitude is calculated to obtain the deviation value for that impact event. A positive deviation value indicates that the measured impact amplitude is higher than the theoretical expectation, meaning the driver's operation generated additional impact load, indicating an aggressive driving style. A deviation value close to zero indicates that the measured impact amplitude is basically in line with the theoretical expectation, and the driving behavior is within the normal range. A negative deviation value indicates that the measured impact amplitude is lower than the theoretical expectation, and the driving style is more moderate. The deviation values ​​of multiple impact events constitute a deviation value sequence for subsequent statistical analysis.

[0082] S503. Among the deviation values ​​of each impact event, the deviation value that can characterize the distribution characteristics of the deviation value is the second characteristic.

[0083] In one possible implementation, a second feature is determined from a sequence of deviation values ​​corresponding to multiple impact events, which characterizes the overall distribution of the deviation values. This second feature is used to quantify the overall aggressiveness of the driver's operating style, reflecting the statistical characteristics of operating behavior during long-term operations.

[0084] For example, statistical analysis is performed on the deviation values ​​corresponding to multiple impact events. Since aggressive maneuvers are typically intermittent events, using average values ​​for characterization is easily diluted by a large amount of normal driving data, failing to effectively capture the characteristics of abnormal maneuvers that are infrequent but extremely intense. Therefore, deviation values ​​at specific statistical positions in the deviation value sequence can be selected as a second feature. The selection of this statistical position must ensure that the extracted feature value represents the tail characteristics of the deviation value distribution, i.e., it can reflect extreme situations of maneuvering. The larger the value of the second feature, the more severe the aggressive maneuvers the driver exhibited during long-term operations.

[0085] Understandably, this step is the final output of the quantification of driving style. By statistically analyzing the deviation values ​​of multiple impact events, it is possible to extract overall characteristics reflecting the driver's long-term driving habits from the discrete deviations of a single impact. This second characteristic, together with the first characteristic, constitutes a complete understanding of the source of axle damage: the first characteristic reflects the load amplification effect caused by hardware aging, and the second characteristic reflects the additional impact load caused by driving behavior. The combination of these two characteristics provides a comprehensive basis for subsequent differentiated control strategies and also achieves a scientific quantitative assessment of the driver's driving style.

[0086] The technical solution provided by the above embodiments can bring at least the following beneficial effects: This embodiment defines the specific determination process of the second feature. The expected impact amplitude is calculated using a preset dynamic model, which simultaneously considers the influence of operating conditions, load conditions, and hardware aging characteristics on impact transmission, making the expected value more consistent with the actual physical process. Based on this, by calculating the deviation between the measured amplitude and the expected amplitude, the impact caused by normal braking logic and hardware aging factors is effectively separated, extracting the abnormal impact component purely caused by operating behavior. Finally, a statistical quantity that can characterize the distribution characteristics is extracted from multiple deviation values ​​as the second feature, realizing a quantitative characterization of the driver's operating style and providing an objective basis for the evaluation and improvement of operating behavior.

[0087] In one possible implementation, the process of determining the first feature to characterize the change in vehicle hardware state based on the impact energy attenuation characteristics of each impact event and the vehicle load state at the time of each impact event can be specifically implemented through the following S601-S605, which will be described in detail below.

[0088] S601. Determine the total mass of the vehicle based on the vehicle's load condition at the time of each impact event.

[0089] In one possible implementation, for each impact event, the total vehicle mass corresponding to that impact event is determined based on the vehicle's load state at the time of the impact. This total vehicle mass is used to characterize the overall mass of the vehicle at the time of the impact and is a fundamental parameter for subsequently constructing the relationship between load and damping.

[0090] For example, the total vehicle mass consists of two parts: the vehicle's curb weight and the load mass at the time of the impact. The vehicle's curb weight is its own weight in its unloaded state, and is a preset fixed value. The load mass at the time of the impact is the weight of the cargo carried by the vehicle at the time of the impact event, obtained through the load state obtained in the preceding steps. Adding the vehicle's curb weight and the load mass gives the total vehicle mass corresponding to the impact event. For each impact event, the corresponding total vehicle mass is determined in the above manner, forming a sequence of total vehicle masses corresponding to multiple impact events.

[0091] S602. Based on the total mass of the vehicle and the impact energy attenuation characteristics of each impact event, a load-damping relationship coordinate system is constructed.

[0092] In one possible implementation, for each impact event, a load-damping relationship coordinate system is constructed based on the vehicle's total mass and impact energy attenuation characteristics corresponding to that impact event. This coordinate system is used to establish the mapping relationship between the vehicle's total mass and impact energy attenuation characteristics, revealing the physical laws existing between the two, and providing an analytical framework for subsequent extraction of hardware state characteristics.

[0093] For example, the load-damping relationship coordinate system uses the vehicle's total mass as the horizontal axis variable and the impact energy decay characteristic as the vertical axis variable. Each impact event corresponds to a unique location point in this coordinate system based on its vehicle's total mass and impact energy decay characteristic. Multiple location points corresponding to multiple impact events constitute the data point set in this coordinate system. According to the inherent physical law of damping characteristics changing with load, when the hardware state is stable, these data points should theoretically exhibit a specific distribution pattern; when the hardware state changes, the overall distribution of data points will shift accordingly.

[0094] S603. Determine multiple location points of each impact event in the load-damping relationship coordinate system.

[0095] In one possible implementation, for each impact event, the corresponding location point is determined in a pre-established load-damping relationship coordinate system based on the vehicle's total mass and impact energy attenuation characteristics. Multiple location points corresponding to multiple impact events constitute a data point set in this coordinate system, which is used for subsequent linear fitting analysis.

[0096] For example, the load-damping relationship coordinate system uses the vehicle's total mass as the horizontal axis and the impact energy attenuation characteristic as the vertical axis. For each impact event, a unique point is determined in the coordinate system using the vehicle's total mass as the horizontal axis and the impact energy attenuation characteristic as the vertical axis. This point represents the system's damping response characteristics under a specific load condition at the time of the impact. Performing the above operation on all impact events that have passed the attenuation effectiveness verification yields a set of points in the coordinate system consisting of multiple location points. The distribution of this point set implicitly reveals the law of system damping variation with load and the current characteristics of the hardware state.

[0097] S604. Using a preset fixed slope value as the slope constraint of the fitted line, determine a fitted line that matches multiple location points.

[0098] In one possible implementation, for multiple location points in the load-damping relationship coordinate system, a pre-set fixed slope value is used as the slope constraint for the fitted straight line to determine a fitted straight line that matches the multiple location points. The intercept of this fitted straight line on the vertical axis of the coordinate system is determined as the current damping reference value. This current damping reference value is used to characterize the system damping reference level after deducting the load effect under the current hardware conditions.

[0099] For example, the preset fixed slope value is determined based on the inherent physical law of damping variation with load. In the load-damping relationship coordinate system, when the hardware is stable, multiple location points should theoretically be distributed near a straight line with a slope of this fixed value. Using this fixed slope value as a constraint, a fitting algorithm determines an optimal fitted straight line from multiple location points, such that this line matches the majority of location points to the greatest extent. The intercept of this fitted straight line reflects the baseline level of the system's damping characteristics under the current hardware conditions. As the hardware ages, this intercept value will change accordingly.

[0100] S605. Based on the difference between the current damping reference value and the preset reference damping reference value, determine the first feature.

[0101] In one possible implementation, a determined current damping reference value is compared with a preset reference damping reference value, and a first feature is determined based on the degree of difference between the two to characterize changes in the vehicle's hardware state. This first feature is used to quantify the degree of performance degradation of hardware (such as tires) due to factors such as aging and wear.

[0102] For example, the reference damping baseline value is a benchmark value calibrated when the vehicle is in a healthy hardware condition (such as with new tires), representing the initial level of hardware performance. The current damping baseline value is a real-time value fitted from recent impact event data through the aforementioned steps, reflecting the current performance state of the hardware. As the hardware ages, its damping performance gradually declines, causing the current damping baseline value to deviate from the reference damping baseline value. By calculating the difference between the two, the first characteristic reflecting the degree of hardware aging can be obtained. The larger the difference, the more severe the hardware performance degradation and the higher the degree of aging; the smaller the difference, the closer the hardware state is to its initial healthy state.

[0103] For example, the difference between the reference damping baseline value and the current damping baseline value can be determined as the first feature. Assuming the reference damping baseline value (calibrated under new tire conditions) is 2.5, and the current damping baseline value fitted from impact event data is 2.0, then the difference is 0.5. This value is the first feature, indicating that hardware aging has caused a 0.5 decrease in damping performance. The larger the difference, the more severe the aging. If the difference is negative or zero, it indicates that the hardware condition has not significantly deteriorated.

[0104] The technical solution provided in the above embodiments can bring at least the following beneficial effects: This embodiment defines the specific determination process of the first feature. A load-damping relationship coordinate system is constructed based on the vehicle's total mass and impact energy attenuation characteristics, mapping each impact event to a position point in this coordinate system. A fitted straight line is determined by constraining it with a preset fixed slope, and the intercept is extracted as the current damping benchmark value, thus achieving a quantitative characterization of the hardware state. This process fully utilizes the inherent physical law of damping variation with load, transforming the nonlinear relationship into a linear relationship through the construction of a logarithmic coordinate system, making the extraction of the hardware state more accurate and reliable. Finally, the first feature is calculated based on the difference between the current damping benchmark value and the reference value, intuitively reflecting the degree of tire rubber hardening and providing a quantitative indicator for hardware state assessment.

[0105] In one possible implementation, the process of determining a fitted line that matches multiple location points by using a preset fixed slope value as the slope constraint of the fitted line can be specifically implemented through the following S701-S704, which will be explained in detail below.

[0106] S701. Randomly select multiple candidate location points from multiple location points.

[0107] In one possible implementation, multiple points are randomly selected from a set of points in the load-damping relationship coordinate system as candidate points. Each candidate point is used to construct a candidate fitted line, and multiple candidate points correspond to multiple different candidate fitted lines to cover the possible linear parameter space.

[0108] For example, a preset number of iterations is set. In each iteration, a location point is randomly selected from multiple location points as a candidate location point for that iteration. Since each selection is performed independently and randomly, different iterations may select the same location point or different location points. After multiple iterations, multiple candidate location points are obtained, each representing a baseline point for a possible straight-line hypothesis. This random sampling mechanism ensures that the algorithm can fully explore different straight-line possibilities and avoid getting trapped in local optima.

[0109] S702. Using a preset fixed slope value as the slope and passing through each candidate position point as a condition, determine multiple candidate fitted lines.

[0110] In one possible implementation, for each of a randomly selected plurality of candidate locations, a corresponding candidate fitting line is determined by using a preset fixed slope value as the slope of the line and passing through that candidate location as a constraint. Multiple candidate locations correspond to multiple candidate fitting lines, each line satisfying the conditions of a fixed slope and passing through the corresponding candidate location.

[0111] For example, each candidate location point has specific coordinate values ​​in the load-damping relationship coordinate system. Using a preset fixed slope value as the slope of a straight line, combined with the coordinates of the candidate location point, a straight line passing through that point can be uniquely determined. The intercept of this straight line on the vertical axis of the coordinate system is determined by both the coordinates of the candidate location point and the fixed slope. By traversing all candidate location points, a set of candidate fitted straight lines can be obtained, each corresponding to a candidate location point, and all having the same slope but different intercepts. These candidate fitted straight lines constitute a set of assumptions about the possible distribution trends of the location point set.

[0112] For example, the preset fixed slope value can be determined based on the inherent physical law of damping variation with load. Taking a solid-tire forklift as an example, based on the theory of single-degree-of-freedom damped systems, the damping ratio is inversely proportional to the square root of the mass. In the load-damping relationship coordinate system (horizontal axis is mass, vertical axis is damping), the theoretical slope value corresponding to this physical relationship is -0.5. Therefore, the preset fixed slope value can be set to -0.5.

[0113] S703. Count the number of locations among multiple locations whose distance to each candidate fitted line is less than the preset tolerance.

[0114] In one possible implementation, for each of the identified candidate fitted lines, the distances from all points in the load-damping relationship coordinate system to that candidate fitted line are calculated, and the number of points whose distances are less than a preset tolerance is counted. This number is used to evaluate the degree of matching between each candidate fitted line and the set of points, serving as the basis for subsequently selecting the optimal fitted line.

[0115] For example, for each candidate fitted line, iterate through each of the multiple location points and calculate the longitudinal distance from that point to the current candidate fitted line. Compare this distance with a preset tolerance value. If the distance is less than the preset tolerance, the point is considered to match the current line well and is counted as an interior point; if the distance is greater than or equal to the preset tolerance, the point is considered to deviate too far and is not counted as an interior point. After performing the above judgment on all location points, the number of interior points corresponding to the current candidate fitted line can be obtained. The value of the preset tolerance can be calibrated according to the measurement error range and engineering experience; for example, it can be set to 0.1, representing the allowable range of deviation in the distance from the point to the line.

[0116] For example, the vertical distance from the k-th data point to the i-th candidate fitted line. The following formula 2 is satisfied: Formula 2 in, and Let K be the coordinates of the k-th data point. The slope of the current candidate line; The intercept of the current candidate line; Calculate the absolute residual from the point to the line to determine whether it is an interior point. The smaller the distance, the more the point fits the line. Count the number of interior points and select the best line.

[0117] S704. The candidate fitted line with the largest number of location points is determined as the fitted line that matches multiple location points.

[0118] In one possible implementation, for multiple candidate fitted lines, the number of inliers corresponding to each candidate fitted line is compared, and the candidate fitted line with the largest number of inliers is determined as the fitted line that matches multiple location points. This fitted line represents the line that can best explain the distribution trend of most location points in the current data point set, and its intercept will be used as the current damping benchmark value for subsequently determining the first feature.

[0119] For example, suppose 100 candidate fitting lines are generated through random sampling and fixed slope constraints, each line corresponding to a different intercept. For each candidate line, the number of inliers whose distance from all points to the line is less than a preset tolerance is counted. The number of inliers among these 100 candidate lines is compared, and the line with the largest number of inliers is identified. Since the number of inliers reflects the degree of matching between the line and the majority of points in the data set, the line with the largest number of inliers means that it best represents the overall distribution trend of the entire data set and is least affected by outliers. This line is then determined as the final fitting line that matches the set of points.

[0120] The technical solution provided in the above embodiments can bring at least the following beneficial effects: This embodiment specifically defines the process of determining the fitted straight line. Using the idea of ​​random sampling consistency, candidate position points are randomly selected from multiple position points, and multiple candidate fitted straight lines are generated with a fixed slope as a constraint. The optimal straight line is then selected by statistically counting the number of interior points. This technical feature effectively overcomes the influence of outliers caused by interference factors such as uneven road surfaces and cargo movement in actual working conditions, avoiding the defect of traditional least squares methods being easily biased by noise. Through iterative search and tolerance judgment, this method can robustly extract benchmark parameters representing the true state of the tire in environments with poor data quality, significantly improving the robustness and accuracy of feature extraction.

[0121] In one possible implementation, the process of determining the deviation value of each impact event as a second feature, which can characterize the distribution characteristics of the deviation value, can be specifically implemented through the following S801-S802, which will be described in detail below.

[0122] S801. Sort the deviation values ​​corresponding to multiple impact events according to their numerical values.

[0123] In one possible implementation, a sequence of deviation values ​​corresponding to multiple impact events is sorted according to their numerical values. The sorted sequence of deviation values ​​is then used to extract statistics that characterize the distribution of the deviation values, in order to determine a second feature.

[0124] For example, suppose we accumulate deviation values ​​corresponding to 100 impact events, such as +2.5, -0.3, +5.1, +1.2, -0.8, +3.7, etc. These deviation values ​​are arranged in ascending order to obtain a sorted sequence of deviation values. In the sorted sequence, the minimum value is located at the beginning of the sequence, the maximum value is located at the end of the sequence, and the deviation values ​​in the middle reflect the values ​​at different quantiles. The sorting process transforms the disordered set of deviation values ​​into an ordered sequence, providing a foundation for subsequently extracting deviation values ​​at specific statistical positions.

[0125] S802. Select the deviation value at the preset percentage position in the sorting result as the second feature.

[0126] In one possible implementation, for a sorted sequence of deviation values, the deviation value at a preset percentage position is selected as a second feature to characterize the driver's operating style. This second feature is used to quantify the driver's aggressiveness during long-term operations, reflecting representative statistical characteristics in the distribution of deviation values.

[0127] For example, the preset percentage position can be set according to the requirements of sensitivity to operating style, such as 90%. For the sorted sequence of deviation values, the deviation value at the 90% position means that among all the deviation values ​​of impact events, 90% of the deviation values ​​are less than or equal to this value, and only 10% of the deviation values ​​are greater than this value. Selecting the deviation value at this position as the second feature can effectively capture abnormal operating characteristics that are infrequent but extremely intense, avoiding the problem of the average value being diluted by a large amount of normal driving data. The larger the value of the second feature, the more severe the aggressive operation by the driver during long-term operation, and the greater the additional impact load on the axle structure.

[0128] The technical solution provided in the above embodiments can bring at least the following beneficial effects: This embodiment specifically defines the extraction method of the deviation value distribution feature in the second feature. By sorting multiple deviation values ​​and selecting the value at a preset percentage position as the second feature, a high quantile statistical strategy is adopted. This method differs from traditional average statistics and can effectively capture abnormal operation features that are infrequent but extremely intense, avoiding the problem of the average value being diluted by a large amount of normal driving data. The selection of the preset percentage position can be flexibly adjusted according to actual needs, enabling the second feature to more sensitively reflect the driver's aggressive operating behavior and providing a more effective quantitative indicator for operational risk assessment.

[0129] In one possible implementation, a closed-loop control strategy for the vehicle also needs to be executed. This process can be specifically implemented through the following S901, which will be described in detail below.

[0130] S901. Based on the first feature and the second feature, execute a closed-loop control strategy for the vehicle.

[0131] In one possible implementation, a closed-loop control strategy for the vehicle is executed based on the determined first and second characteristics, according to the degree of hardware state change reflected by these characteristics and the aggressiveness of the driver's operating style. This closed-loop control strategy is used to take differentiated intervention measures for different sources of damage, thereby reducing the rate of damage accumulation in the axle structure and extending its service life.

[0132] For example, the first feature reflects the degree of performance degradation of vehicle hardware (such as tires) due to aging, and the second feature reflects the aggressiveness of the driver's operating behavior. Different types of control strategies can be executed based on the values ​​of the first and second features and their comparison with preset thresholds. When the first feature indicates severe hardware aging, the control strategy may focus on limiting the vehicle's speed to reduce road impact energy input and compensate for the load amplification effect caused by increased transmission rate. When the second feature indicates an aggressive driving style and normal hardware condition, the control strategy may focus on smoothing the driver's input commands and reducing transient impact loads caused by human error. When both the first and second features are within the normal range, the conventional control mode can be maintained without additional intervention.

[0133] The technical solution provided by the above embodiments can bring at least the following beneficial effects: This embodiment introduces a closed-loop control strategy based on the first and second features. Applying the decoupling results of the features to vehicle control enables differentiated intervention for damage sources. When the hardware condition deteriorates, the speed is actively limited to directly reduce the impact energy input from the road surface; when the driving style is aggressive, driving commands are smoothed to reduce transient impacts caused by human error. This technical feature solves the drawback of the one-size-fits-all approach of traditional control strategies, enabling targeted control measures based on different damage sources. While protecting the axle structure, it maximizes operational efficiency, achieving a complete closed loop from condition monitoring to active control, and improving the vehicle's full lifecycle management capabilities.

[0134] Please see Figure 2This document illustrates a schematic diagram of the system architecture of a forklift axle fatigue safety life prediction system 200 based on load spectrum, provided in an embodiment of the present invention. This system is used to implement the aforementioned forklift axle fatigue safety life prediction method based on load spectrum. The system includes: a data acquisition module 201, used to acquire impact response data of multiple impact events occurring during vehicle operation; the impact response data includes: the vehicle load state at the time of each impact event, the impact amplitude of each impact event, and the impact energy decay characteristics of each impact event; a cumulative fatigue damage determination module 202, used to determine the cumulative fatigue damage degree of multiple impact events based on the impact amplitude of each impact event and the material fatigue characteristics of the vehicle structural components; a feature extraction module 203, used to determine a first feature characterizing changes in the vehicle hardware state based on the impact energy decay characteristics of each impact event and the vehicle load state at the time of each impact event; a cumulative fatigue damage correction module 204, used to correct the cumulative fatigue damage degree based on the first feature and determine the corrected cumulative fatigue damage degree; and a life prediction module 205, used to predict the remaining life of the vehicle structural components based on the corrected cumulative fatigue damage degree.

[0135] The technical solutions provided in the above embodiments offer at least the following beneficial effects: This embodiment provides a forklift axle fatigue safety life prediction system based on load spectrum, corresponding to the above method. This system, through the organic coordination of a data acquisition module, a cumulative fatigue damage determination module, a feature extraction module, a cumulative fatigue damage correction module, and a life prediction module, transforms the method in the above embodiments into an engineering-implementable physical system. The functional division between each module is clear, and the data flow is well-defined, enabling automated completion of the entire process from data acquisition, feature extraction, damage correction to life prediction. The system has a complete architecture and comprehensive functions, and can be directly deployed in the forklift controller, providing real-time life assessment and decision support for vehicle operation and maintenance, demonstrating significant engineering application value.

[0136] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0137] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for predicting the fatigue safety life of forklift axles based on load spectrum, characterized in that, The method includes: Acquire impact response data for multiple impact events that occur during vehicle operation; the impact response data includes: vehicle load status at the time of each impact event, impact amplitude of each impact event, and impact energy attenuation characteristics of each impact event. Based on the impact amplitude of each impact event and the material fatigue characteristics of the vehicle structural components, the cumulative fatigue damage of the multiple impact events is determined. Based on the impact energy attenuation characteristics of each impact event and the vehicle load state at the time of each impact event, a first feature is determined to characterize the changes in vehicle hardware state. Based on the first feature, the cumulative fatigue damage is corrected to determine the corrected cumulative fatigue damage. Based on the corrected cumulative fatigue damage, the remaining life of the vehicle structural components is predicted. The first feature for characterizing changes in vehicle hardware state is determined based on the impact energy attenuation characteristics of each impact event and the vehicle load state at the time of each impact event, including: The total mass of the vehicle is determined based on the vehicle's load status at the time of each impact event. Based on the total mass of the vehicle and the impact energy attenuation characteristics of each impact event, a load-damping relationship coordinate system is constructed. Determine multiple location points of each impact event in the load-damping relationship coordinate system; Using a preset fixed slope value as the slope constraint for the fitted line, a fitted line matching the plurality of position points is determined; the intercept of the fitted line is the current damping reference value. The first feature is determined based on the difference between the current damping reference value and the preset reference damping reference value.

2. The method for predicting the fatigue safety life of forklift axles based on load spectrum according to claim 1, characterized in that, The acquisition of impact response data for multiple impact events occurring during vehicle operation includes: The vertical vibration signal of the vehicle is monitored in real time. When the amplitude of the vertical vibration signal exceeds a preset impact threshold, an impact event is determined to have occurred. In response to the occurrence of the impact event, the vehicle load status at the moment of occurrence is obtained; The impact amplitude of the impact event is determined based on the waveform segments of the vertical vibration signal within a preset time period after the impact event occurs. The attenuation effectiveness is verified on the waveform segment; the attenuation effectiveness verification is used to determine whether the waveform segment conforms to a preset free attenuation model; Impact energy attenuation features are extracted from waveform segments that have passed attenuation effectiveness verification.

3. The method for predicting the fatigue safety life of forklift axles based on load spectrum according to claim 2, characterized in that, The verification of the attenuation effectiveness of the waveform segment includes: Determine whether the ratio of the first peak to the second peak is greater than a preset first threshold; the first peak and the second peak are two temporally adjacent peaks in the waveform segment; Determine whether the time interval between the first peak and the second peak is within a preset inherent period range; When the ratio is greater than the first threshold and the time interval is within the inherent period range, the waveform segment is determined to have passed the attenuation effectiveness verification.

4. The method for predicting the fatigue safety life of forklift axles based on load spectrum according to claim 1, characterized in that, The impact response data further includes: the vehicle's operating status at the time of the impact; the method further includes: Based on the vehicle operating state, the load state, and the first feature, a second feature is determined to characterize the driver's operating style.

5. The method for predicting the fatigue safety life of forklift axles based on load spectrum according to claim 4, characterized in that, The step of determining the second feature characterizing the driver's operating style based on the vehicle operating state, the load state, and the first feature includes: Based on the operating state, the load state, and the first feature, the expected impact amplitude is calculated using a preset dynamic model. Determine the deviation between the impact amplitude of each impact event and the expected impact amplitude; Among the deviation values ​​of each impact event, the deviation value that can characterize the distribution characteristics of the deviation value is determined as the second feature.

6. The method for predicting the fatigue safety life of forklift axles based on load spectrum according to claim 1, characterized in that, The step of determining a fitting line that matches the plurality of position points by using a preset fixed slope value as the slope constraint for the fitting line includes: Randomly select multiple candidate location points from the plurality of location points; Using the preset fixed slope value as the slope and passing through each of the candidate position points as a condition, multiple candidate fitting lines are determined; The number of location points whose distance to each candidate fitted line is less than a preset tolerance is counted among the multiple location points. The candidate fitted line with the largest number of location points is determined as the fitted line that matches multiple of the location points.

7. The method for predicting the fatigue safety life of forklift axles based on load spectrum according to claim 5, characterized in that, In determining the deviation value for each impact event, the deviation value that characterizes the distribution characteristics of the deviation value is the second characteristic, including: The deviation values ​​corresponding to multiple impact events are sorted according to their numerical values; The deviation value at a preset percentage position in the sorting results is selected as the second feature.

8. The method for predicting the fatigue safety life of forklift axles based on load spectrum according to claim 4, characterized in that, The method further includes: Based on the first feature and the second feature, a closed-loop control strategy for the vehicle is executed.

9. A forklift axle fatigue safety life prediction system based on load spectrum, characterized in that, The system is used to implement the forklift axle fatigue safety life prediction method based on load spectrum as described in any one of claims 1 to 8; the system includes: The data acquisition module is used to acquire impact response data of multiple impact events that occur to the vehicle during operation; the impact response data includes: the vehicle load status at the time of each impact event, the impact amplitude of each impact event, and the impact energy decay characteristics of each impact event. The cumulative fatigue damage determination module is used to determine the cumulative fatigue damage degree of the multiple impact events based on the impact amplitude of each impact event and the material fatigue characteristics of the vehicle structural components. The feature extraction module is used to determine a first feature to characterize the change in vehicle hardware state based on the impact energy decay characteristics of each impact event and the vehicle load state at the time of each impact event. The cumulative fatigue damage correction module is used to correct the cumulative fatigue damage degree based on the first feature and determine the corrected cumulative fatigue damage degree. The life prediction module is used to predict the remaining life of vehicle structural components based on the corrected cumulative fatigue damage.

Citation Information

Patent Citations

  • Method for identifying fatigue damage of metal component based on inherent damping characteristics of cross-point frequency response

    CN114996914A

  • Automobile ridge crossing control method, device and equipment and storage medium

    CN117644750A