A drive axle wheel end load distribution adaptive calibration method and system

Abstract

The application relates to a drive axle wheel end load distribution adaptive calibration method and system, and relates to the technical field of drive axle load calibration, and comprises the following steps: analyzing drive axle wheel end load distribution related elements of a target vehicle, obtaining inner control elements, outer control elements and road condition elements; performing road condition zoning on a target area to obtain a plurality of zones; taking the inner control elements and the outer control elements of the target vehicle as inputs, and taking drive axle wheel end load distribution parameters as inputs, training a drive axle wheel end load distribution calibration model; based on a working area of the target vehicle, scheduling the drive axle wheel end load distribution calibration model to perform drive axle wheel end load distribution adaptive calibration. The application solves the problem that the traditional drive axle wheel end load distribution calibration does not fully consider the differences among vehicles, road conditions and goods, thereby leading to a complex load-related processing process and a lack of universal solution.

Description

Technical Field

[0001] This application relates to the field of drive axle load calibration, and in particular to an adaptive calibration method and system for drive axle wheel end load distribution. Background Technology

[0002] With increasing demands for vehicle safety and stability in fields such as freight transportation and engineering operations, accurate calibration of the load distribution at the drive axle wheel ends has become a key technical requirement to ensure reliable vehicle operation.

[0003] Currently, the traditional drive axle wheel end load distribution calibration method does not fully consider the differences between different vehicles, road conditions and goods, and cannot be adapted to various complex operating scenarios. This not only makes the load prediction process complicated, but also makes it difficult to form a universal calibration scheme, increasing the possibility of safety risks caused by improper load distribution of vehicles. Summary of the Invention

[0004] This application provides an adaptive calibration method and system for drive axle wheel end load distribution, which improves the applicability and effectiveness of drive axle wheel end load distribution calibration when load prediction is complex and a general solution is lacking due to differences in different vehicles, road conditions and goods.

[0005] The embodiments of this application disclose the following technical solutions:

[0006] In a first aspect, embodiments of this application provide an adaptive calibration method for the load distribution at the wheel ends of a drive axle, the method comprising:

[0007] Analysis of the load distribution correlation elements at the drive axle wheel ends of the target vehicle is performed to obtain the inner control elements, outer control elements and road condition elements. Among them, the inner control elements represent the vehicle control parameters and the outer control elements represent the cargo parameters.

[0008] Based on the aforementioned road condition elements, the target area is divided into several road condition zones;

[0009] Traverse the aforementioned partitions, using the inner and outer control elements of the target vehicle as inputs, and the drive axle wheel end load distribution parameters as inputs, to train several drive axle wheel end load distribution calibration models;

[0010] Based on the target vehicle's operating area, the aforementioned drive axle wheel-end load distribution calibration models are scheduled to perform adaptive calibration of the drive axle wheel-end load distribution.

[0011] Secondly, embodiments of this application provide an adaptive calibration system for drive axle wheel end load distribution, the system comprising:

[0012] The load correlation element analysis module is used to perform correlation element analysis on the drive axle wheel end load distribution of the target vehicle to obtain inner control elements, outer control elements and road condition elements. Among them, the inner control elements represent vehicle control parameters and the outer control elements represent cargo parameters.

[0013] The road condition element partitioning module is used to partition the target area based on the road condition elements to obtain several partitions.

[0014] The partition calibration model training module is used to traverse the several partitions, taking the inner and outer control elements of the target vehicle as inputs and the drive axle wheel end load distribution parameters as inputs, to train several drive axle wheel end load distribution calibration models.

[0015] The work area model scheduling and calibration module is used to schedule the several drive axle wheel end load distribution calibration models to perform adaptive calibration of the drive axle wheel end load distribution based on the target vehicle's work area.

[0016] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0017] This application proposes an adaptive calibration method and system for the load distribution at the wheel ends of a drive axle. Through multi-dimensional correlation element analysis, road condition zoning, model training and dynamic scheduling, the adaptive calibration of the load distribution at the wheel ends of the drive axle of the target vehicle is achieved. First, a correlation analysis of the drive axle wheel-end load distribution of the target vehicle is conducted. An initial set of internal control elements, external control elements, and road condition elements are loaded and classified into qualitative and quantitative sets. Core internal control elements, external control elements, and road condition elements are selected through type switching fluctuation sorting and correlation analysis. Then, based on the selected road condition elements, the target area is partitioned in multiple dimensions. Each road condition element is extracted and partitioned sequentially according to the deviation threshold. The results of cross-multidimensional partitioning are used to obtain several partitions with unified road condition characteristics. Subsequently, each partition is traversed, and the recorded values ​​of internal control elements, external control elements, and load distribution parameters are collected, using the partition road condition parameters as constants. A dedicated drive axle wheel-end load distribution calibration model for each partition is trained and stored in association. Finally, based on the real-time operating area of ​​the target vehicle, the partition is located and the corresponding dedicated calibration model is scheduled. Real-time internal control and external control element data are collected and input into the model. Calibration parameters are output and their effectiveness is verified. Load distribution adjustment is performed to achieve adaptive calibration.

[0018] The technical solution of this application first screens core related elements to ensure the validity of calibration data, then uses road condition zoning and distributed models to solve the problem of insufficient accuracy of a single model in adapting to multiple road conditions, and finally achieves accurate matching between the operation scenario and the calibration model through dynamic scheduling model. This solves the problems in traditional drive axle wheel end load calibration caused by inaccurate element screening, large calibration deviation caused by road condition differences, and poor adaptability caused by fixed models, thus realizing the adaptive and accurate calibration of drive axle wheel end load distribution. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 A flowchart illustrating an adaptive calibration method for drive axle wheel end load distribution provided in an embodiment of this application;

[0021] Figure 2 This is a schematic diagram of the structure of an adaptive calibration system for drive axle wheel end load distribution provided in an embodiment of this application.

[0022] The components represented by each number in the attached diagram are explained below:

[0023] Load correlation element analysis module 01, road condition element zoning module 02, zoning calibration model training module 03, and work area model scheduling calibration module 04. Detailed Implementation

[0024] This application provides an adaptive calibration method and system for drive axle wheel end load distribution, which solves the technical problems in the prior art that the differences between different vehicles, road conditions and goods are not fully considered, resulting in a complex drive axle wheel end load prediction process and a lack of a universal drive axle wheel end load distribution calibration solution.

[0025] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0026] In the description of this application, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0027] In the description of this application, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this application is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed in this application.

[0028] Example 1, as shown in the appendix Figure 1 As shown, this application provides an adaptive calibration method for drive axle wheel end load distribution, the method comprising the following steps:

[0029] S110: Perform a correlation analysis of the load distribution at the drive axle wheel ends of the target vehicle to obtain the inner control elements, outer control elements, and road condition elements. Among them, the inner control elements represent the vehicle control parameters, and the outer control elements represent the cargo parameters.

[0030] In this embodiment of the application, in order to accurately screen out the key elements that have a significant impact on load distribution, it is necessary to conduct a systematic analysis of the load distribution related elements of the drive axle wheel end of the target vehicle, determine the core inner control elements, outer control elements and road condition elements, so as to ensure that the subsequent model can accurately capture the relationship between the elements and the load distribution, and at the same time improve the adaptability and accuracy of the calibration method based on the screened elements.

[0031] Specifically, the first step is to focus on the target vehicle and load the initial set of internal control elements, the initial set of external control elements, and the initial set of road condition elements related to the load distribution at the wheel ends of the vehicle's drive axle, providing initial data sources for subsequent element classification and screening.

[0032] Furthermore, the initial internal control element set, initial external control element set, and initial road condition element set are uniformly classified, and divided into categorical element set and quantitative element set according to the attribute characteristics of the elements.

[0033] Furthermore, the categorized element set obtained by the division is traversed, and the type switching drive axle wheel end load distribution fluctuation sorting is performed on each categorized element to screen out the categorized elements that cause significant fluctuations in the drive axle wheel end load distribution due to type switching, and they are respectively classified into the first inner control element, the first outer control element and the first road condition element.

[0034] Meanwhile, the quantitative element set is traversed, and a correlation analysis with the drive axle wheel end load distribution fluctuation value is performed on each quantitative element to screen out quantitative elements that are strongly correlated with the drive axle wheel end load distribution fluctuation, and these elements are respectively classified into the second inner control element, the second outer control element, and the second road condition element.

[0035] Finally, the first and second inner control elements are integrated and added to the final inner control elements, the first and second outer control elements are integrated and added to the final outer control elements, and the first and second road condition elements are integrated and added to the final road condition elements.

[0036] This step first loads an initial set of elements to build a data foundation, and then, through classification, sorting of categorized elements based on fluctuations, quantitative analysis of element correlation, and element integration, forms a precise and comprehensive set of core elements. This provides crucial data support for subsequent zoning based on road condition elements and training and calibrating models based on internal and external control elements.

[0037] Step S110 in the method provided in this application embodiment includes:

[0038] Load the initial internal control element set, initial external control element set, and initial road condition element set of the target vehicle;

[0039] Classification is performed on the initial internal control element set, the initial external control element set, and the initial road condition element set to obtain a categorized element set and a quantitative element set;

[0040] Traverse the defined element set, perform type switching drive axle wheel end load distribution fluctuation sorting, and obtain the first inner control element, the first outer control element, and the first road condition element;

[0041] By traversing the quantitative element set and performing correlation analysis with the drive axle wheel end load distribution fluctuation value, the second internal control element, the second external control element, and the second road condition element are obtained.

[0042] Add the first internal control element and the second internal control element to the internal control element; add the first external control element and the second external control element to the external control element; add the first road condition element and the second road condition element to the road condition element.

[0043] In this embodiment of the application, in order to screen out the core elements that have a significant impact on the load distribution of the drive axle wheel ends from the load distribution correlation data of the target vehicle, it is necessary to load the initial inner control, outer control and road condition element sets, classify them into fixed and quantitative element sets according to attributes, perform type switching fluctuation sorting on the fixed elements, perform correlation analysis on the quantitative elements, and finally integrate the screening results to ensure that the finally determined inner control elements, outer control elements and road condition elements can truly reflect the correlation between vehicle control, cargo status and road conditions and load distribution.

[0044] Specifically, the initial set of elements for the target vehicle is first loaded, including the initial set of internal control elements, the initial set of external control elements, and the initial set of road condition elements.

[0045] The initial internal control element set includes basic data related to vehicle control such as steering, speed, and gantry position. The initial external control element set includes information related to cargo status such as load weight and load center distance (the distance from the cargo center of gravity to the gantry is the load center distance). The initial road condition element set involves parameters related to the driving environment such as road surface unevenness, longitudinal slope, lateral slope, and road surface adhesion coefficient.

[0046] Furthermore, after the initial element set is loaded, a classification operation is performed on the initial internal control element set, the initial external control element set, and the initial road condition element set to distinguish between the categorical element set and the quantitative element set.

[0047] Among them, the categorized feature set includes features with non-numerical and discrete attributes, such as the "high adhesion", "medium adhesion" and "low adhesion" level classification of road surface adhesion coefficient.

[0048] In addition, the quantitative feature set includes features with numerical and continuous attributes, such as steering angle, speed, load weight, load center distance, road surface unevenness, longitudinal slope, and lateral slope. This classification lays the foundation for subsequent differentiated screening methods for different types of features.

[0049] Subsequently, the categorized element set obtained by the division is traversed, and the type switching drive axle wheel end load distribution fluctuation sorting is performed on each categorized element.

[0050] The method provided in this application embodiment, which involves "traversing the categorized element set, performing type-switching drive axle wheel-end load distribution fluctuation sorting, and obtaining the first internal control element, the first external control element, and the first road condition element," includes:

[0051] From the defined category element set, extract the first defined category element, configure it as a unique variable, and collect the drive axle wheel end load distribution deviation coefficient set before and after a preset number of type switching of the target vehicle model;

[0052] Calculate the mean value of the set of load distribution deviation coefficients at the drive axle wheel ends, and set it as the fluctuation value of the first categorized element;

[0053] When the fluctuation value of the first categorized element is greater than or equal to the fluctuation value threshold, the first categorized element is added to the first internal control element, the first external control element, and the first road condition element.

[0054] Otherwise, delete the first categorized element.

[0055] In this embodiment of the application, in order to accurately select the elements that have a significant impact on the load distribution at the wheel end of the drive axle from the set of categorized elements, it is necessary to observe the load distribution fluctuation by switching types, taking into account the discrete attribute characteristics of the categorized elements, so as to ensure that the selected first internal control element, first external control element and first road condition element can truly reflect the correlation between the changes in categorized attributes and the load distribution.

[0056] Specifically, firstly, a first-classified element is extracted from the predefined set of categorized elements and configured as a unique variable. That is, the values ​​of all other elements in the initial inner control element set, the initial outer control element set, and the initial road condition element set are kept fixed, and only the type of the first-classified element is changed. This isolates the influence of the first-classified element and avoids analytical bias caused by interference from other elements.

[0057] Furthermore, the set of drive axle wheel end load distribution deviation coefficients for the target vehicle model before and after switching between different types of the first classification element is collected, and the number of collections must meet the preset number requirement, such as a preset collection of 30 times.

[0058] Multiple data collections can reduce the impact of accidental factors on load distribution data during a single data collection process, ensuring that the set of load distribution deviation coefficients at the drive axle wheel ends is statistically representative and can objectively reflect the true impact of the first classification element type switching on load distribution.

[0059] During the data collection process, it is necessary to strictly control the status of other elements besides the first classification element. For example, if the first classification element is "road surface adhesion coefficient level", when switching between "high adhesion" and "low adhesion" types, it is necessary to keep the internal control elements such as vehicle steering angle, speed, and gantry position, as well as the external control elements such as load weight and load center distance unchanged, and only change the road surface adhesion coefficient level condition.

[0060] Simultaneously, the drive axle wheel end load distribution data before and after each type switch are recorded synchronously, and the drive axle wheel end load distribution deviation coefficient is obtained according to the calculation rules of the drive axle wheel end load distribution deviation coefficient, thus forming a set of drive axle wheel end load distribution deviation coefficients.

[0061] In the method provided in this application embodiment, the calculation rules for the "drive axle wheel end load distribution deviation coefficient" include:

[0062] Load the first load distribution and the second load distribution at the drive axle wheel end, perform the same position deviation calculation, and obtain the load deviation set at the drive axle wheel end;

[0063] The proportion of positions where the load deviation at the drive axle wheel end is greater than or equal to the load deviation threshold is calculated and set as the load distribution deviation coefficient at the drive axle wheel end.

[0064] In this embodiment of the application, in order to objectively quantify the degree of difference in the distribution of drive axle wheel-end load under different conditions, it is necessary to calculate the drive axle wheel-end load distribution deviation coefficient to ensure the accuracy and reliability of the subsequent screening results of internal control elements, external control elements and road condition elements.

[0065] Specifically, a first load distribution and a second load distribution are first applied to the drive axle wheel ends. These two load distributions must correspond to load data of the same target vehicle model under specific comparison conditions.

[0066] In the scenario of switching between categorical feature types, the first load distribution can be the load distribution before the feature type switch, and the second load distribution can be the load distribution after the type switch; in the scenario of changing the gradient of quantitative features, the first load distribution can be the load distribution under a certain gradient value of the feature, and the second load distribution can be the load distribution under another gradient value.

[0067] Meanwhile, during the loading process, it is necessary to ensure that the data collection environment and vehicle status of the two load distributions are consistent, such as the same road surface conditions and the same initial placement of goods, so as to eliminate the interference of irrelevant variables on the differences in load distribution.

[0068] Further, perform positional deviation calculations to obtain the drive axle wheel end load deviation set.

[0069] The drive axle wheel end contains multiple load locations that need to be monitored, such as key stress points on the left, right, and middle sides of the wheel end. The deviation calculation at the same location involves subtracting the load value of the second load distributed at the same location from the load value of the first load distributed at that location for each monitoring location, thus obtaining the load deviation value at that location.

[0070] Therefore, by calculating the deviation at each location, the differences in the distribution of the two sets of loads at different stress points can be fully captured, avoiding incomplete judgment of the overall load distribution differences due to focusing only on some locations.

[0071] For example, if the drive axle wheel end contains 5 monitoring positions, after calculating the load deviation value of each position, the 5 deviation values ​​together constitute the drive axle wheel end load deviation set.

[0072] Furthermore, the proportion of locations where the concentrated deviation value of the drive axle wheel end load deviation is greater than or equal to the load deviation threshold is calculated, and this proportion is the drive axle wheel end load distribution deviation coefficient.

[0073] The load deviation threshold is a critical value set based on the actual conditions such as the load-bearing capacity of the drive axle wheel end and the vehicle's operational safety requirements. For example, it is preset to 500N. This means that when the load deviation at a certain position reaches or exceeds this value, the load change at that position is sufficient to have a real impact on the force balance of the drive axle wheel end and the vehicle's stability.

[0074] When calculating the proportion, first count the number of locations where the load deviation concentration meets the condition of "deviation value ≥ load deviation threshold", then divide this number by the total number of monitoring locations at the drive axle wheel ends. The resulting proportion is the drive axle wheel end load distribution deviation coefficient.

[0075] For example, if the deviation value of 3 out of 5 monitoring positions is ≥500N, then the load distribution deviation coefficient of the drive axle wheel end is 3 / 5=0.6. This value intuitively reflects the proportion of positions with significant influence in the two sets of load distributions, providing a quantitative reference for subsequent element screening.

[0076] Throughout the calculation process, it is essential to ensure the accuracy and consistency of the data. When applying the load distribution, load monitoring equipment with the required accuracy must be used, such as pressure sensors with an error of no more than ±1%, to avoid data distortion due to insufficient equipment accuracy.

[0077] Meanwhile, when performing the same position deviation calculation, it is necessary to strictly correspond to the same monitoring position to avoid errors in deviation value calculation due to incorrect position matching; when setting the load deviation threshold, it is necessary to make personalized adjustments based on the specific parameters of the target vehicle model to ensure that the threshold can truly reflect the critical impact degree of load changes of that vehicle model.

[0078] For example, for the drive axle wheel end of a certain type of freight forklift, a load deviation threshold of 500N is set, and the total number of monitoring positions is 6. When analyzing the impact of switching the "road surface adhesion coefficient level" (classification element) on the load distribution, the first load distribution before the switch (load values ​​at each position: 2800N, 2750N, 2900N, 2850N, 2700N, 2650N) and the second load distribution after the switch (load values ​​at each position: 3200N, 3100N, 2880N, 2820N, 3050N, 2600N) are loaded, and the deviation at the same position is calculated to obtain the deviation set: 400N, 350N, -20N, -30N, 350N, -50N.

[0079] Furthermore, if the number of positions ≥500N in the statistical deviation set is 0, then the load distribution deviation coefficient for this scenario is 0. If, when analyzing the influence of the gradient change of "load weight" (quantitative element), a first load distribution under a certain gradient and a second load distribution under another gradient are applied, the calculated deviation set is 600N, 550N, 480N, 520N, 580N, and 450N, where the number of positions ≥500N is 4, then the load distribution deviation coefficient is 4 / 6≈0.67.

[0080] By using the standardized calculation rules above, a quantitative index that can objectively reflect the differences in load distribution at the drive axle wheel ends can be obtained. This index not only covers the magnitude of the load deviation but also reflects the coverage of the deviation, providing a unified data basis for subsequent classification and sorting of fluctuation factors.

[0081] Furthermore, the mean value of the load distribution deviation coefficient set at the drive axle wheel ends is calculated and set as the first categorical element fluctuation value.

[0082] The mean calculation comprehensively reflects the overall level of load distribution deviation under multiple type switching, thus avoiding the misleading effect of extreme values ​​of a single deviation coefficient on the judgment result. For example, if the deviation coefficients obtained from 30 collections fluctuate between 0.15 and 0.25, the calculated mean is 0.20. This value represents the fluctuation value of the first-classified element that can represent the degree of influence of the type switching of the first-classified element on the load distribution fluctuation.

[0083] Furthermore, the fluctuation value of the first classification element is compared with a preset fluctuation value threshold to determine whether the type switching of the first classification element will have a significant impact on the load distribution at the drive axle wheel end.

[0084] Among them, the fluctuation value threshold is a judgment standard set based on a large amount of experimental data and actual application needs. For example, the preset threshold is 0.18, which means that when the average fluctuation of the load distribution caused by the switching of the category element type reaches or exceeds the fluctuation value threshold, it indicates that the influence of the element on the load distribution is significant enough and needs to be included in the subsequent core element system.

[0085] Specifically, if the fluctuation value of the first categorized element is lower than the preset fluctuation value threshold, it indicates that the element has a weak impact on the load distribution. Even if it is included, it will not have an effective effect on the calibration of the load distribution at the drive axle wheel end, but will instead increase data redundancy.

[0086] Conversely, when the fluctuation value of the first categorized element is greater than or equal to the fluctuation value threshold, it is added to the first inner control element, the first outer control element, or the first road condition element according to the attribute category of the first categorized element. For example, "road surface adhesion coefficient level" belongs to the road condition related categorized element, so it is classified as the first road condition element. If the fluctuation value of the first categorized element is less than the fluctuation value threshold, it is directly deleted from the element set and will not participate in the subsequent process.

[0087] For example, if the target vehicle is a certain type of forklift, the first category element in the category element set is "road surface adhesion coefficient level", the preset number of collections is 30, the fluctuation value threshold is 0.18, and the load deviation threshold is 500N.

[0088] During the screening process, the forklift steering angle of 15°, the moving speed of 5km / h, the mast position of 1500mm, the load weight of 2000kg, and the load center distance of 800mm were kept constant. Only the road surface adhesion coefficient level was switched between "high adhesion" and "low adhesion". After each switch, the load distribution data of each position at the drive axle wheel end was recorded. 30 deviation coefficients were calculated according to the rules. If the average value (i.e. the fluctuation value of the first classification element) was 0.21 (≥0.18), then the "road surface adhesion coefficient level" was added to the first road condition element.

[0089] If another first-classifying element is "cargo packaging type", and the fluctuation value of the first-classifying element obtained by the same process is 0.12 (<0.18), then "cargo packaging type" will be deleted from the element set.

[0090] Through the above steps, we can accurately identify the categorized elements that have a significant impact on the distribution of loads at the drive axle wheel ends, providing reliable data for the subsequent construction of the first internal control element, the first external control element, and the first road condition element. We can also effectively eliminate irrelevant categorized elements and reduce data redundancy.

[0091] The method provided in this application embodiment, which involves "traversing the quantitative element set, performing correlation analysis with the drive axle wheel end load distribution fluctuation value, and obtaining the second internal control element, the second external control element, and the second road condition element", includes:

[0092] From the set of quantitative elements, extract the first quantitative element, configure it as a unique variable, and collect the drive axle wheel end load distribution deviation coefficient sequence of the preset variable gradient sequence of the target vehicle model.

[0093] Correlation analysis was performed on the preset variable gradient sequence and the drive axle wheel end load distribution deviation coefficient sequence to obtain the Pearson correlation coefficient;

[0094] When the absolute value of the Pearson correlation coefficient is greater than or equal to the correlation coefficient threshold, the first quantitative element is added to the second internal control element, the second external control element, and the second road condition element.

[0095] Otherwise, delete the first quantitative element.

[0096] In this embodiment of the application, in order to accurately select elements that are strongly correlated with the load distribution fluctuation of the drive axle wheel end from the set of quantitative elements, it is necessary to analyze the continuous numerical attribute characteristics of the quantitative elements and the variable gradient change and correlation analysis to ensure that the selected second inner control element, second outer control element and second road condition element can accurately reflect the mapping relationship between the quantitative attributes and the load distribution.

[0097] Specifically, a first quantitative element is extracted from the predefined set of quantitative elements and configured as a unique variable. That is, the values ​​of all other elements in the initial inner control element set, the initial outer control element set, and the initial road condition element set are kept fixed, and only the value of the first quantitative element is changed, so as to avoid the correlation analysis bias caused by interference from other elements.

[0098] Furthermore, the drive axle wheel end load distribution deviation coefficient sequence of the target vehicle model under the preset variable gradient sequence of the first quantitative element is collected.

[0099] The preset variable gradient sequence needs to be set in conjunction with the actual application range and change pattern of the first quantitative element to ensure that it can fully cover the commonly used value range of the element. For example, for "load weight", the variable gradient sequence can be set to 1000kg, 2000kg, 3000kg, 4000kg, 5000kg; for "longitudinal slope", the variable gradient sequence can be set to 0%, 2%, 4%, 6%, 8%.

[0100] During the data collection process, it is also necessary to strictly control the status of other elements besides the first quantitative element. For example, if the first quantitative element is "moving speed", when adjusting from 3km / h to 15km / h according to the gradient sequence, it is necessary to keep the internal control elements such as the vehicle steering angle of 10° and the gantry position of 1200mm, as well as the external control elements such as the load weight of 2500kg and the load center distance of 700mm unchanged, and only change the moving speed value.

[0101] Simultaneously, the load distribution data of the drive axle wheel end under each gradient is recorded synchronously, and the deviation coefficient of the corresponding gradient is obtained according to the calculation rule of the drive axle wheel end load distribution deviation coefficient, and finally a drive axle wheel end load distribution deviation coefficient sequence corresponding one-to-one with the variable gradient sequence is formed.

[0102] Furthermore, a correlation analysis was performed on the preset variable gradient sequence and the corresponding drive axle wheel end load distribution deviation coefficient sequence, and the Pearson correlation coefficient was obtained by calculation.

[0103] The Pearson correlation coefficient quantifies the degree of linear correlation between two variables. Its value ranges from -1 to 1. The closer the absolute value is to 1, the stronger the linear correlation between the two variables; the closer the absolute value is to 0, the weaker the linear correlation between the two variables.

[0104] Furthermore, the absolute value of the Pearson correlation coefficient is compared with a preset correlation coefficient threshold to determine whether there is a strong correlation between the numerical change of the first quantitative element and the fluctuation of the drive axle wheel end load distribution.

[0105] The correlation coefficient threshold is a judgment standard set based on a large amount of experimental data and the calibration requirements of drive axle wheel end loads. For example, the preset correlation coefficient threshold is 0.65, which means that when the absolute value of the Pearson correlation coefficient reaches or exceeds this value, it indicates that the correlation strength between the first quantitative element and the load distribution fluctuation can have a significant impact on the subsequent calibration results and needs to be included in the core element system; if it is lower than this value, it indicates that the correlation strength is insufficient, and even if it is included, it cannot provide effective support for calibration, but will instead increase data redundancy.

[0106] Specifically, when the absolute value of the Pearson correlation coefficient is greater than or equal to the correlation coefficient threshold, it is added to the second inner control element, the second outer control element, or the second road condition element according to the attribute category of the first quantitative element. For example, "gantry position" belongs to the vehicle control related quantitative element, so it is classified as the second inner control element. If the absolute value of the Pearson correlation coefficient is less than the correlation coefficient threshold, it is directly deleted from the element set and will no longer participate in the subsequent process.

[0107] For example, if the target vehicle is a certain type of industrial forklift, the first quantitative element in the quantitative element set is "load weight," and the preset variable gradient sequence is 1000kg, 2000kg, 3000kg, 4000kg, and 5000kg, with a correlation coefficient threshold of 0.65 and a load deviation threshold of 500N. During the analysis, the forklift steering angle of 15°, the moving speed of 6km / h, the mast position of 1400mm, the road surface unevenness of 3mm / m, and the longitudinal slope of 2% are kept unchanged. Only the load weight is adjusted according to the gradient. Under each gradient, the load distribution data of the drive axle wheel end is collected and the deviation coefficient is calculated, resulting in a deviation coefficient sequence of 0.12, 0.28, 0.45, 0.63, and 0.78.

[0108] Furthermore, a correlation analysis was performed on the load weight gradient sequence and the deviation coefficient sequence. If the absolute value of the Pearson correlation coefficient is 0.92 (≥0.65), then "load weight" is added to the second external control element. If the other first quantitative element is "ambient temperature" (assuming it is included in the initial quantitative element set), and the absolute value of the correlation coefficient obtained by collecting and calculating according to the gradient sequence is 0.35 (<0.65), then "ambient temperature" is deleted from the quantitative element set.

[0109] Through the above steps, we can accurately identify quantitative elements that are strongly correlated with the fluctuation of the drive axle wheel end load distribution, providing reliable data for the subsequent construction of the second internal control element, the second external control element, and the second road condition element. We can also effectively eliminate irrelevant quantitative elements, simplify the data dimensions, and lay the element foundation for the adaptive calibration of the drive axle wheel end load distribution.

[0110] S120: Divide the target area into several zones based on the road condition elements;

[0111] In this embodiment of the application, in the scenario of adaptive calibration of load distribution at the drive axle wheel end, in order to improve the accuracy of load distribution calibration under different scenarios, it is necessary to first perform a hierarchical partitioning operation on the target area based on the selected road condition elements to ensure that the road condition features in each partition are consistent, and to provide a clear scenario division basis for subsequent regional model training.

[0112] Specifically, the first road condition element, the second road condition element, and so on up to the Nth road condition element are extracted sequentially from the road condition elements obtained from the analysis. Here, N represents the total number of road condition elements. These road condition elements cover the key road condition characteristics that affect the distribution of drive axle wheel end loads, and each element reflects the differences in road conditions in the target area from different dimensions.

[0113] Furthermore, for the extracted first road condition element, a pre-set first road condition element deviation threshold is first loaded, and then the first road condition element monitoring values ​​of all locations in the target area are collected. The monitoring values ​​are compared with the deviation threshold, and the first zoning operation is performed on the target area to obtain the first zoning result.

[0114] Following the same logic, the second road condition element deviation threshold, the third road condition element deviation threshold, and so on up to the Nth road condition element deviation threshold are loaded sequentially. The monitoring values ​​of the corresponding road condition elements in the target area are collected respectively, and the partitioning operation is performed one by one until the Nth partition result is obtained.

[0115] Furthermore, the results of the first partition, the second partition, and so on up to the Nth partition are cross-fused, that is, the partition results under different road condition element dimensions are superimposed and matched, and finally several partitions with unique and clear road condition features are obtained.

[0116] This step involves first completing preliminary partitioning based on individual road condition elements, and then cross-integrating the multi-dimensional partitioning results to construct a partitioning system covering all typical road conditions in the target area. Each partition corresponds to a specific combination of road conditions, providing clear scene boundaries for subsequent training of dedicated drive axle wheel end load distribution calibration models for different partitions.

[0117] Step S120 in the method provided in this application embodiment includes:

[0118] From the road condition elements, extract the first road condition element up to the Nth road condition element, where N represents the total number of road condition elements;

[0119] Load the first road condition element deviation threshold, collect the first road condition element monitoring value of the target area, perform road condition zoning on the target area, and obtain the first zoning result;

[0120] Until the Nth road condition element deviation threshold is loaded, the Nth road condition element monitoring value of the target area is collected, and road condition zoning is performed on the target area to obtain the Nth zoning result;

[0121] The first partitioning result up to the Nth partitioning result are interleaved to obtain the plurality of partitions.

[0122] In this embodiment of the application, in order to avoid insufficient accuracy of single standard calibration due to differences in road conditions, the selected road condition elements need to be divided into multi-dimensional hierarchical partitions of the target area to ensure that the road condition features in each partition are consistent, provide clear scene boundaries for regional model training, and thus improve the accuracy of overall adaptive calibration.

[0123] Specifically, the first road condition element, the second road condition element, and so on, are extracted sequentially from the obtained road condition elements.

[0124] Wherein, N is the total number of road condition elements. These road condition elements are key indicators that have a significant impact on the distribution of drive axle wheel end loads, selected through correlation analysis. Examples include road surface unevenness, longitudinal slope, lateral slope, and road surface adhesion coefficient. Each road condition element reflects the road condition characteristics of the target area from different dimensions, and together they constitute the core basis for road condition zoning.

[0125] Furthermore, for the extracted first road condition element, a pre-set first road condition element deviation threshold is first loaded to clarify the critical standard for distinguishing different road condition levels under this road condition element.

[0126] The first road condition element deviation threshold is determined by combining the target vehicle's drive axle load-bearing capacity, operational safety standards, and actual road condition impact test results. It is used to distinguish the critical values ​​of different road condition levels under the road condition element. For example, if the first road condition element is road surface unevenness, its deviation threshold can be set to 4mm / m to define the boundary between "low unevenness" and "high unevenness".

[0127] Furthermore, by using road condition monitoring equipment, such as vehicle-mounted laser smoothness detectors and slope sensors, the first road condition element monitoring values ​​of all key locations within the target area are collected. The monitoring values ​​of each location are compared with the deviation threshold, and the target area is then divided into zones for the first time.

[0128] For example, areas with monitoring values ​​≤ 4 mm / m are classified as "low unevenness areas", and areas with monitoring values ​​> 4 mm / m are classified as "high unevenness areas", thus obtaining the first partitioning result.

[0129] Similarly, following the same partitioning steps described above, the second road condition element deviation threshold, the third road condition element deviation threshold, and so on, are loaded sequentially until the Nth road condition element deviation threshold is reached.

[0130] Meanwhile, for each road condition element, the monitoring value of the corresponding element within the target area is collected and zoning is performed: if the second road condition element is longitudinal slope and the deviation threshold is set to 3%, then the area with a monitoring value ≤ 3% is divided into "low slope area" and the area with a monitoring value > 3% is divided into "high slope area", thus obtaining the second zoning result; if the third road condition element is road surface adhesion coefficient and the deviation threshold is set to 0.6, then the area with a monitoring value ≥ 0.6 is divided into "high adhesion area" and the area with a monitoring value < 0.6 is divided into "low adhesion area", thus obtaining the third zoning result.

[0131] This process continues until the Nth road condition element is partitioned, yielding the Nth partition result. Throughout the single-element partitioning process, it is essential to ensure that the monitoring values ​​for each element cover the entire operational area of ​​the target region to avoid incomplete partitioning results due to monitoring blind spots. Simultaneously, the accuracy of the monitoring equipment must be guaranteed to prevent data errors from affecting the accuracy of the partitioning.

[0132] Furthermore, after obtaining the results of the first to Nth partitions, these partition results are cross-fused to combine the partition features under different road condition element dimensions into a unique partition identifier, ensuring that each final partition corresponds to a specific combination of road conditions.

[0133] Cross-fusion involves overlaying and matching the partitioning results under different road condition element dimensions, combining multi-dimensional road condition features into a unique partition identifier.

[0134] Finally, through the multi-dimensional cross-tracing of the above steps, several unique and well-defined partitions of road condition features are obtained. Each partition corresponds to a specific combination of road conditions to ensure that the models trained for different partitions can accurately match the road conditions of that area.

[0135] For example, if the target area is a logistics park and its surrounding work area, the total number of road condition elements N=3, namely road surface unevenness (first road condition element), longitudinal slope (second road condition element), and road surface adhesion coefficient (third road condition element).

[0136] Among them, the threshold for road surface unevenness deviation is 4 mm / m, the threshold for longitudinal slope deviation is 3%, and the threshold for road surface adhesion coefficient deviation is 0.6.

[0137] Further, after collecting road condition monitoring values ​​within the logistics park, the first zoning results were "low unevenness area (A1)" and "high unevenness area (A2)", the second zoning results were "low slope area (B1)" and "high slope area (B2)", and the third zoning results were "high adhesion area (C1)" and "low adhesion area (C2)". After cross-referencing these three results, eight zoning areas were obtained: A1B1C1, A1B1C2, A1B2C1, A1B2C2, A2B1C1, A2B1C2, A2B2C1, and A2B2C2.

[0138] The partitioning method described above ensures that the road condition characteristics within each partition are consistent, providing a clear scenario basis for subsequent regional model training. At the same time, it fully covers all possible road condition combinations within the target area, avoiding model adaptation failures due to road condition omissions, and laying a scenario foundation for the effective implementation of the adaptive calibration method for drive axle wheel end load distribution.

[0139] S130: Traverse the aforementioned partitions, using the inner and outer control elements of the target vehicle as inputs and the drive axle wheel end load distribution parameters as inputs, to train several drive axle wheel end load distribution calibration models.

[0140] In this embodiment of the application, in order to solve the problem of insufficient load analysis accuracy caused by the difficulty in real-time monitoring of road condition parameters, and to enable the calibration model to adapt to the road condition characteristics of different regions, it is necessary to build a dedicated distributed calibration model for each partition. This will improve the accuracy of load distribution analysis and calibration by fixing the road condition parameters of the partition and focusing on the influence of internal and external control elements.

[0141] Specifically, starting from the target area that has been partitioned, the first partition, the second partition, and so on, are extracted sequentially. Each partition is associated with and stored in relation to the road condition parameters of the corresponding first partition. These road condition parameters are fixed attributes of key road condition features such as road surface condition and slope within that partition.

[0142] Furthermore, for the extracted first partition, the road condition parameters of the first partition associated with the storage are set to constants, that is, they are not used as variables for model training. Only the inner and outer control elements of the target vehicle are used as model inputs, and the drive axle wheel end load distribution parameters are used as model outputs.

[0143] Furthermore, the recorded values ​​of the inner control elements, outer control elements, and drive axle wheel end load distribution parameters of the target vehicle during operation in the first zone are collected.

[0144] Meanwhile, the data collection process needs to cover common operating scenarios within the first partition to ensure that the training data can fully reflect the correlation between the internal and external control elements and the load distribution within the partition.

[0145] After data acquisition is completed, the first zone drive axle wheel end load distribution calibration model is trained based on the collected inner control element record values, outer control element record values, and drive axle wheel end load distribution parameter record values.

[0146] Following the same technical approach, the above operations are performed sequentially on the second partition, the third partition, and all partitions to train the activity-specific partition drive axle wheel end load distribution calibration model, and then associated and stored, adding it to several drive axle wheel end load distribution calibration model sets.

[0147] This step solves the problem of real-time road condition monitoring by first fixing road condition parameters that are difficult to monitor in real time by partitioning them, and then training a distributed model for each partition. This enables each model to focus on the correlation between internal and external control elements and load distribution within a specific partition, providing accurate model support for subsequent adaptive calibration based on the vehicle's actual operating area scheduling model, and effectively improving the accuracy of load analysis and calibration.

[0148] Step S130 in the method provided in this application embodiment includes:

[0149] From the plurality of partitions, the first partition is extracted, wherein the first partition is stored in association with the traffic condition parameters of the first partition;

[0150] Using the road condition parameters of the first zone as constants, the recorded values ​​of the inner control elements, outer control elements, and drive axle wheel end load distribution parameters of the target vehicle are collected. The drive wheel end load distribution calibration model of the first zone is trained, and after being associated and stored with the first zone, it is added to the plurality of drive wheel end load distribution calibration models.

[0151] In this embodiment of the application, in order to solve the problem of insufficient load analysis accuracy caused by the difficulty in real-time monitoring of road condition parameters, and to enable the drive axle wheel end load distribution calibration model to accurately adapt to the road condition characteristics of different areas of the workshop, it is necessary to first extract the corresponding partition and associated road condition parameters based on the partition results, fix the partition road condition parameters to a constant number, collect training data and build a dedicated model to improve the accuracy of load analysis and calibration.

[0152] Specifically, the first partition is extracted from several partitions obtained by dividing the road conditions in the workshop. This first partition is associated with and stored with the corresponding road condition parameters during the partitioning process. These road condition parameters are fixed values ​​of key indicators such as road surface unevenness, longitudinal slope, lateral slope, and road surface adhesion coefficient within the first partition.

[0153] For example, “first zone - road surface unevenness 3mm / m, longitudinal slope 0.8%, road surface adhesion coefficient 0.75”. Since the road conditions in each area of ​​the workshop are relatively stable in the short term, storing these road condition parameters in association with the zones can replace the difficult-to-achieve real-time monitoring and provide clear road condition constraints for subsequent model training.

[0154] Furthermore, the road condition parameters associated with the first partition are set to constant values. That is, in subsequent model training, road condition parameters will no longer be used as variables; only the influence of internal and external control elements on the load distribution at the drive axle wheel ends will be considered, thereby avoiding model learning bias caused by the lack or fluctuation of real-time monitoring of road condition parameters.

[0155] Furthermore, the recorded values ​​of the inner control elements, outer control elements, and drive axle wheel end load distribution parameters of the target vehicle during operation in the first zone are collected.

[0156] During the data collection process, it is necessary to cover common operating scenarios within the first partition, such as combinations of different steering angles and speeds, and combinations of different load weights and center distances, to ensure that the training data can fully reflect the correlation between the internal and external control elements and load distribution within the partition, and to avoid insufficient generalization ability of the model due to incomplete data coverage.

[0157] After data acquisition is completed, the collected inner control element record values ​​and outer control element record values ​​are used as model inputs, and the drive axle wheel end load distribution parameter record values ​​are used as model outputs to train the first zone drive wheel end load distribution calibration model.

[0158] During training, a deep learning neural network is used as the basic framework. This framework has strong nonlinear fitting capabilities and can effectively capture the complex mapping relationship between the inner control elements, the outer control elements and the load distribution parameters of the drive axle wheel ends.

[0159] During specific training, the recorded values ​​of the inner and outer control elements collected in the first partition are used as input layer data, and the recorded values ​​of the load distribution parameters at each position of the corresponding drive axle wheel end are used as output layer target data. The core parameters of the network, such as weights and biases, are iteratively adjusted through the backpropagation algorithm.

[0160] After each iteration, the mean square error (MSE) between the model output value and the actual detection value is calculated, and the parameters are optimized based on the error results until the error is stably controlled within a preset range, such as MSE ≤ 0.001. This ensures that the model can accurately learn the influence of changes in internal and external control elements on load distribution in the first partition, thereby improving the prediction accuracy of the model in that partition.

[0161] After training is completed, in order to achieve rapid matching and management of the model with the corresponding partition, the calibration model of the first partition drive wheel end load distribution is associated and stored with the first partition.

[0162] Specifically, this is achieved by annotating the model's metadata. For example, adding an attribute label to the model such as "Adapt to the first zone - road condition parameters: road surface unevenness ≤ 4mm / m, longitudinal slope ≤ 1%, road surface adhesion coefficient ≥ 0.75". This label must fully cover the key road condition features of the first zone to ensure that the adapted model can be quickly retrieved based on the road condition parameters of the target vehicle's operating area.

[0163] After completing the associated storage, the model is formally added to the set of several drive axle wheel end load distribution calibration models, making it a dedicated calibration model for the first partition.

[0164] Similarly, following the same technical solution, a second zone is extracted from several zones. This second zone is also associated with and stored with pre-determined road condition parameters, such as "road surface unevenness 5mm / m, longitudinal slope 1.5%, road surface adhesion coefficient 0.7".

[0165] Similarly, after setting the road condition parameters of the second zone to a constant value, the recorded values ​​of the inner control elements, the recorded values ​​of the outer control elements, and the corresponding recorded values ​​of the drive axle wheel end load distribution parameters are collected when the target vehicle is operating in the zone.

[0166] Based on this collected data, a calibration model for the load distribution at the wheel ends of the second drive axle is trained to ensure that the error between the load distribution parameters output by the model and the actual detected values ​​is controlled within a preset range.

[0167] After training is completed, add attribute labels such as "adapt to second zone - road condition parameters: road surface unevenness ≤ 6mm / m, longitudinal slope ≤ 2%" to the second zone drive axle wheel end load distribution calibration model, store it in association with the second zone, and then add it to a set of several drive axle wheel end load distribution calibration models.

[0168] This process continues until all partitions have been trained and stored, ultimately forming a set of several drive axle wheel end load distribution calibration models that cover all partitions of the target area and each partition has its own dedicated model. This lays the foundation for subsequent model adaptation based on the target vehicle's operating area.

[0169] S140: Based on the target vehicle's operating area, schedule the aforementioned drive axle wheel end load distribution calibration models to perform adaptive calibration of the drive axle wheel end load distribution.

[0170] In this embodiment of the application, in the scenario of adaptive calibration of load distribution at the drive axle wheel end, in order to ensure that the calibration process can accurately adapt to the road conditions of the target vehicle's real-time operating area, it is necessary to first locate the partition corresponding to the vehicle's current operating area, and then schedule the calibration model dedicated to that partition to perform calibration, so as to achieve the accuracy and adaptability of load distribution calibration under different operating scenarios.

[0171] Specifically, the first step is to locate the current working area of ​​the target vehicle in real time and determine the zone to which the working area belongs.

[0172] During the positioning process, vehicle positioning technology is combined with geographical boundary data of road condition zones to establish a mapping relationship between the work area and the zones. For example, if the workshop has been divided into three zones, A, B, and C, according to road conditions, and the geographical coordinate range of each zone has been stored, such as the zone coordinates of zone A being X1-Y1 to X2-Y2, when the vehicle positioning device detects that the vehicle's current coordinates are (X3, Y3) and are within the coordinate range of zone A, it can be determined that the vehicle's current work area belongs to zone A.

[0173] Meanwhile, to ensure the accuracy of positioning, the positioning data will be verified in real time. For example, positioning data will be collected every 10 seconds, and after 5 consecutive collections, the average coordinates will be taken and compared with the partition boundary to avoid incorrect partition determination due to instantaneous positioning deviation.

[0174] Furthermore, based on the determined zone to which the work area belongs, the dedicated calibration model corresponding to that zone is retrieved from a set of several drive axle wheel end load distribution calibration models.

[0175] Since the "adaptation zone - road condition parameters" attribute was added to each model during the early training of the model, the corresponding model can be quickly matched by searching the zone name. For example, the calibration model of the first zone drive axle wheel end load distribution corresponding to zone A can be retrieved.

[0176] After the retrieval is completed, the adaptability of the model needs to be confirmed a second time. That is, extract the key features of real-time traffic conditions in the current work area and compare them with the range of traffic condition parameters in the model identifier. If the real-time traffic condition features are within the range of model adaptability, the model is confirmed to be schedulable; if they are outside the range, the partitioning judgment results are rechecked, and if necessary, the work area is relocated and an adaptable model is retrieved.

[0177] Furthermore, real-time data on the target vehicle's internal and external control elements within the current work area are collected and used as input parameters for the post-scheduling model.

[0178] Similarly, during the data acquisition process, it is necessary to ensure the real-time nature and accuracy of the data. For example, the steering angle data can be sampled three times consecutively and the average value can be taken to reduce the impact of instantaneous sensor errors on the data. At the same time, outliers can be removed by using data verification algorithms in existing technologies.

[0179] Furthermore, the collected real-time internal and external control element data are input into the dedicated calibration model after scheduling. Based on the correlation between the internal and external control elements and the load distribution in this partition learned in the previous training, the model outputs the corresponding drive axle wheel end load distribution calibration parameters.

[0180] For example, after inputting "steering angle 15°, speed 3km / h, gantry position 1500mm, load weight 2000kg, load center distance 1000mm" into the A-zone exclusive calibration model, the output calibration parameters include the load value of the left wheel end 18000N, the load value of the right wheel end 17800N, and the load value of the middle wheel end 17500N.

[0181] Furthermore, the calibration parameters output by the model need to be validated to ensure that the calibration results meet the actual operational requirements.

[0182] During verification, the actual load distribution data of the vehicle is collected in real time by the on-board wheel end load detection equipment, and compared with the calibration parameters output by the model to calculate the deviation rate (deviation rate = |calibration parameter - actual detection parameter| / actual detection parameter × 100%).

[0183] If the deviation rate of all wheel end positions is less than the preset threshold (e.g., 5%), the calibration is deemed valid, and the vehicle performs load distribution adjustment according to the calibration parameters. If the deviation rate exceeds the threshold, for example, the load calibration deviation rate on the left side of the wheel end is 7%, the inner and outer control element data of the current working area are re-collected and input into the model again for calibration calculation. If necessary, the model adaptability or the positioning result of the working area is checked until the deviation rate between the calibration parameters and the actual detection data meets the requirements.

[0184] For example, a target vehicle is operating in area A of the workshop. The location confirms that the vehicle belongs to area A, and the calibration model for the load distribution at the wheel ends of the drive axle in the first zone dedicated to area A is dispatched. The real-time internal control element data collected are "steering angle 15°, speed 3km / h, gantry position 1500mm", and the external control element data are "load weight 2000kg, load center distance 1000mm". After inputting these data into the model, the output calibration parameters are "left wheel end 18000N, right wheel end 17800N, center wheel end 17500N".

[0185] Furthermore, the actual load distribution data obtained by the on-board testing equipment is "18500N on the left side of the wheel end, 18000N on the right side of the wheel end, and 17800N in the middle of the wheel end". The calculated deviation rates for each position are 2.7%, 1.1%, and 1.7%, respectively, all of which are less than the preset threshold of 5%. Therefore, the calibration is deemed effective, and the vehicle adjusts the load distribution at the wheel end of the drive axle according to the calibration parameters.

[0186] This step achieves the goal of dynamically scheduling and adapting models according to the vehicle's operating area, enabling the drive axle wheel end load distribution calibration to accurately match road conditions in different areas. It avoids calibration deviations of a single model in multiple road condition scenarios, providing accurate load distribution assurance for the safe and stable operation of the target vehicle in complex operating environments.

[0187] The embodiments of this application, through the specific implementation methods described above, achieve the following technical effects:

[0188] This application proposes an adaptive calibration method for drive axle wheel-end load distribution. First, it performs correlation element analysis on the target vehicle, loading initial sets of internal control, external control, and road condition elements, and classifying them into categorical and quantitative element sets. It then iterates through the categorical elements, performing type-switching fluctuation sorting, and iterates through the quantitative elements, performing correlation analysis to integrate and filter out the core internal control, external control, and road condition elements. Next, based on road condition element partitioning, it extracts multi-dimensional road condition elements, sequentially partitions them according to deviation thresholds, and cross-merges them to obtain several unified road condition partitions. Subsequently, it iterates through the road condition partitions to train the drive axle wheel-end load distribution calibration model. Using the partitioned road conditions as a constant, it collects element record values ​​and load parameters, trains a dedicated calibration model, and stores the data in association. Finally, based on vehicle positioning technology, the vehicle's work zone is located. By searching the correspondence between "adaptive zone - road condition parameters" in the model attribute identifier, the appropriate drive axle wheel end load distribution calibration model is retrieved. By inputting the real-time internal control element data and external control element data of the vehicle's current work area, the calibration parameters of the load values ​​at each position of the drive axle wheel end are output. After verifying that the deviation rate between the calibration parameters and the measured load data meets the standard, the load distribution is adjusted to achieve adaptive calibration of the target vehicle's drive axle wheel end load distribution.

[0189] The method provided in this application, through the technical solution of "precise element selection - multi-dimensional road condition zoning - distributed model training - dynamic model scheduling", solves the problems of inaccurate element selection, large deviation of single model in adapting to multiple road conditions, and inability of calibration parameters to dynamically match the working scenario in traditional drive axle wheel end load calibration. It effectively improves the accuracy and adaptability of drive axle wheel end load distribution calibration, avoids vehicle safety risks or drive axle wear caused by unreasonable calibration, and provides reliable support for the stable operation of vehicles in complex working environments.

[0190] Example 2, as shown in the appendix Figure 2 As shown, based on the inventive concept of the adaptive calibration method for drive axle wheel end load distribution provided in Embodiment 1, this application also provides an adaptive calibration system for drive axle wheel end load distribution, specifically including:

[0191] The load correlation element analysis module 01 is used to perform load distribution correlation element analysis on the drive axle wheel end of the target vehicle to obtain inner control elements, outer control elements and road condition elements. Among them, the inner control elements represent vehicle control parameters and the outer control elements represent cargo parameters.

[0192] The road condition element partitioning module 02 is used to partition the target area based on the road condition elements to obtain several partitions.

[0193] The partition calibration model training module 03 is used to traverse the several partitions, taking the inner control elements and outer control elements of the target vehicle as inputs and the drive axle wheel end load distribution parameters as inputs, to train several drive axle wheel end load distribution calibration models.

[0194] The work area model scheduling and calibration module 04 is used to schedule the several drive axle wheel end load distribution calibration models to perform adaptive calibration of drive axle wheel end load distribution based on the target vehicle's work area.

[0195] In one embodiment, the load association element analysis module 01 is further configured to:

[0196] Load the initial internal control element set, initial external control element set, and initial road condition element set of the target vehicle; perform classification on the initial internal control element set, the initial external control element set, and the initial road condition element set to obtain a categorized element set and a quantitative element set; traverse the categorized element set and perform type-switching drive axle wheel-end load distribution fluctuation sorting to obtain a first internal control element, a first external control element, and a first road condition element; traverse the quantitative element set and perform correlation analysis with the drive axle wheel-end load distribution fluctuation value to obtain a second internal control element, a second external control element, and a second road condition element; add the first internal control element and the second internal control element to the internal control element set, add the first external control element and the second external control element to the external control element set, and add the first road condition element and the second road condition element to the road condition element set.

[0197] Furthermore, the load correlation element analysis module 01 also includes:

[0198] From the defined category element set, extract the first defined category element, configure it as a unique variable, and collect the drive axle wheel end load distribution deviation coefficient set before and after a preset number of type switching for the target vehicle model; calculate the mean of the drive axle wheel end load distribution deviation coefficient set, and set it as the first defined category element fluctuation value; when the first defined category element fluctuation value is greater than or equal to the fluctuation value threshold, add the first defined category element to the first inner control element, the first outer control element, and the first road condition element; otherwise, delete the first defined category element.

[0199] Furthermore, the load correlation element analysis module 01 also includes:

[0200] From the set of quantitative elements, a first quantitative element is extracted and configured as a unique variable. The drive axle wheel end load distribution deviation coefficient sequence of the preset variable gradient sequence of the target vehicle model is collected. Correlation analysis is performed on the preset variable gradient sequence and the drive axle wheel end load distribution deviation coefficient sequence to obtain the Pearson correlation coefficient. When the absolute value of the Pearson correlation coefficient is greater than or equal to the correlation coefficient threshold, the first quantitative element is added to the second inner control element, the second outer control element, and the second road condition element; otherwise, the first quantitative element is deleted.

[0201] Furthermore, the load correlation element analysis module 01 also includes:

[0202] Load the first load distribution and the second load distribution at the drive axle wheel end, perform positional deviation calculation, and obtain the drive axle wheel end load deviation set; calculate the proportion of positions in the drive axle wheel end load deviation set where the drive axle wheel end load deviation is greater than or equal to the load deviation threshold, and set it as the drive axle wheel end load distribution deviation coefficient.

[0203] In one embodiment, the road condition element partitioning module 02 is further configured to:

[0204] From the road condition elements, extract the first road condition element up to the Nth road condition element, where N represents the total number of road condition elements; load the first road condition element deviation threshold, collect the monitoring values ​​of the first road condition element in the target area, and perform road condition zoning on the target area to obtain the first zoning result; until the Nth road condition element deviation threshold is loaded, collect the monitoring values ​​of the Nth road condition element in the target area, and perform road condition zoning on the target area to obtain the Nth zoning result; cross the first zoning result up to the Nth zoning result to obtain the plurality of zoning.

[0205] In one embodiment, the partition calibration model training module 03 is further used for:

[0206] From the plurality of partitions, the first partition is extracted, wherein the first partition is associated with and stored in relation to the road condition parameters of the first partition; using the road condition parameters of the first partition as a constant, the recorded values ​​of the inner control elements, the recorded values ​​of the outer control elements, and the recorded values ​​of the drive axle wheel end load distribution parameters of the target vehicle are collected, the drive wheel end load distribution calibration model of the first partition is trained, and after being associated with and stored in relation to the first partition, it is added to the plurality of drive wheel end load distribution calibration models.

[0207] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0208] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0209] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application intends to include such modifications and variations.

Claims

1. A drive axle wheel end load distribution self-adaptive calibration method, characterized in that, The method comprises the following steps: Performing driving axle wheel end load distribution correlation element analysis on a target vehicle to obtain inner control elements, outer control elements and road condition elements, wherein the inner control elements represent vehicle control parameters including steering and speed, and the outer control elements represent cargo parameters including load weight and load center distance; Based on the road condition elements, the target area is divided into several subareas, and several driving axle wheel end load distribution calibration models are trained by taking the inner control elements and the outer control elements of the target vehicle as inputs and taking driving axle wheel end load distribution parameters as outputs, including: Extracting a first subarea from the several subareas, wherein the first subarea is stored in association with first subarea road condition parameters; Taking the first subarea road condition parameters as constant quantities, collecting the inner control element record values, the outer control element record values and the driving axle wheel end load distribution parameter record values of the target vehicle in different combinations of steering angles and speeds, different combinations of load weights and center distances, using a deep learning neural network as a basic framework, iteratively adjusting the weights and biases of the network through a back propagation algorithm, training a first subarea driving wheel end load distribution calibration model, calculating the mean square error between the model output value and the actual detection value, and optimizing the parameters according to the error result until the error is stably controlled within a preset range, and after being stored in association with the first subarea, other subarea driving wheel end load distribution calibration models are trained by the same training method and added to the several driving wheel end load distribution calibration models; Based on the target vehicle operation area, the several driving axle wheel end load distribution calibration models are dispatched to perform driving axle wheel end load distribution adaptive calibration, including positioning the subarea corresponding to the current operation area of the vehicle first, and then dispatching the calibration model exclusive to the subarea to perform calibration. Performing driving axle wheel end load distribution correlation element analysis on a target vehicle to obtain inner control elements, outer control elements and road condition elements, including:

2. The method of claim 1, wherein, Loading an initial inner control element set, an initial outer control element set and an initial road condition element set of the target vehicle; Performing classification on the initial inner control element set, the initial outer control element set and the initial road condition element set to obtain a classified element set and a quantitative element set; Iterating through the classified element set to perform type switching driving axle wheel end load distribution fluctuation sorting to obtain first inner control elements, first outer control elements and first road condition elements; Iterating through the quantitative element set to perform correlation analysis with driving axle wheel end load distribution fluctuation values to obtain second inner control elements, second outer control elements and second road condition elements; Adding the first inner control elements and the second inner control elements to the inner control elements, adding the first outer control elements and the second outer control elements to the outer control elements, and adding the first road condition elements and the second road condition elements to the road condition elements. Iterating through the classified element set to perform type switching driving axle wheel end load distribution fluctuation sorting to obtain first inner control elements, first outer control elements and first road condition elements, including:

3. The method of claim 2, wherein, ​ From the set of fixed-class elements, extract a first fixed-class element, configure it as a unique variable, and collect a preset number of type switching before and after the drive axle wheel end load distribution deviation coefficient set of the target vehicle model; Calculate the mean of the drive axle wheel end load distribution deviation coefficient set, and set it as the first fixed-class element fluctuation value; When the first fixed-class element fluctuation value is greater than or equal to the fluctuation value threshold, add the first fixed-class element to the first inner control element, the first outer control element, and the first road condition element; Otherwise, delete the first fixed-class element.

4. The method of claim 2, wherein, Traverse the set of quantitative elements, and perform correlation analysis with the drive axle wheel end load distribution fluctuation value to obtain a second inner control element, a second outer control element, and a second road condition element, including: From the set of quantitative elements, extract a first quantitative element, configure it as a unique variable, and collect a preset variable gradient sequence of the drive axle wheel end load distribution deviation coefficient sequence of the target vehicle model; Perform correlation analysis on the preset variable gradient sequence and the drive axle wheel end load distribution deviation coefficient sequence to obtain a Pearson correlation coefficient; When the absolute value of the Pearson correlation coefficient is greater than or equal to the correlation coefficient threshold, add the first quantitative element to the second inner control element, the second outer control element, and the second road condition element; Otherwise, delete the first quantitative element.

5. The method of claim 3 or 4, wherein, The drive axle wheel end load distribution deviation coefficient calculation rule is as follows: Load the drive axle wheel end first load distribution and the drive axle wheel end second load distribution, perform same position deviation calculation to obtain a drive axle wheel end load deviation set; Calculate the proportion of positions in the drive axle wheel end load deviation set where the drive axle wheel end load deviation is greater than or equal to the load deviation threshold, and set it as the drive axle wheel end load distribution deviation coefficient.

6. The method of claim 1, wherein, Based on the road condition element, perform road condition partitioning on the target area to obtain a plurality of partitions, including: From the road condition element, extract a first road condition element to an Nth road condition element, where N represents the total number of road condition elements; Load the first road condition element deviation threshold, collect the first road condition element monitoring value of the target area, and perform road condition partitioning on the target area to obtain a first partition result; Until the Nth road condition element deviation threshold is loaded, collect the Nth road condition element monitoring value of the target area, and perform road condition partitioning on the target area to obtain an Nth partition result; Cross the first partition result to the Nth partition result to obtain the plurality of partitions, where cross fusion means that the partition results under different road condition element dimensions are superimposed and matched to form a unique partition identifier, and each partition corresponds to a specific road condition combination.

7. A drive axle wheel end load distribution adaptive calibration system, characterized by, The system is used to perform the drive axle wheel end load distribution self-adaptive calibration method of any one of claims 1-6, and the system includes: A load-related element analysis module for performing drive axle wheel end load distribution-related element analysis on a target vehicle to obtain an inner control element, an outer control element, and a road condition element, wherein the inner control element represents a vehicle control parameter, the vehicle control parameter includes steering and speed, the outer control element represents a cargo parameter, and the cargo parameter includes load weight and load center distance; The road condition element partition module is configured to perform road condition partition on the target region based on the road condition elements to obtain a plurality of partitions; The partition calibration model training module is configured to traverse the plurality of partitions, take the inner control elements and the outer control elements of the target vehicle as inputs, and take the drive axle wheel end load distribution parameters as outputs to train a plurality of drive axle wheel end load distribution calibration models; The work area model scheduling calibration module is configured to schedule the plurality of drive axle wheel end load distribution calibration models to perform drive axle wheel end load distribution adaptive calibration based on the work area of the target vehicle, including first locating a partition corresponding to a current work area of the vehicle, and then scheduling a calibration model dedicated to the partition to perform calibration.

Images