Load detection method and system for multi-terrain damping suspension system

Abstract

The invention relates to the technical field of dynamic load measurement, in particular to a load detection method and system for a multi-terrain damping suspension system, and the method comprises the steps: building a geometric correlation model; chassis attitude angle data and real-time suspension displacement data of the vehicle are obtained and input into the geometric correlation model, and a three-dimensional pose change vector is solved; an original strain signal is collected, dynamic signal separation is carried out in combination with the three-dimensional pose change vector, and a static load strain value is generated; performing geometric correction on the real-time suspension displacement data according to the three-dimensional pose change vector to generate a displacement characterization load value; detecting the current terrain environment type, and weighting the displacement characterization load value and the static load strain value to obtain an uncorrected real-time load value; and by taking the displacement representation load value and the static load strain value as references, learning to generate a nonlinear compensation parameter, and compensating and outputting a real-time load value in real time. According to the invention, the problem of dynamic load measurement distortion caused by multi-degree-of-freedom motion of the suspension system in a complex driving environment is effectively solved.

Description

Technical Field

[0001] This invention relates to the field of dynamic load measurement technology, and in particular to a load detection method and system for multi-terrain damping suspension systems. Background Technology

[0002] In the field of vehicle engineering, real-time and accurate measurement of dynamic loads is a core technical requirement for optimizing suspension control and improving driving safety. Especially for engineering vehicles, off-road equipment, and other multi-terrain operation machinery, the dynamic loads borne by their suspension systems directly affect structural fatigue assessment, center of gravity stability warning, and power distribution decisions.

[0003] Traditional load detection mainly relies on two technical approaches: one is the direct measurement method using physical sensors, which involves placing strain gauges or force sensors on suspension links or shock absorbers and directly outputting measured values ​​based on the calibrated voltage and load relationship; the other is the kinematic indirect deduction method, which uses vehicle attitude sensors or displacement gauges to obtain suspension geometric parameters and then uses a vehicle dynamics model to infer the load distribution. However, existing technologies often suffer from inherent defects in motion and load coupling distortion, exhibiting the following characteristics: the six-dimensional spatial motion of the suspension system under complex road conditions causes sensors to simultaneously output signals of real load strain and mechanical deformation interference. Traditional strain gauge methods cannot separate these two types of physical quantities, resulting in measured values ​​containing noise components not caused by the load; existing filtering algorithms, such as frequency domain bandpass filtering, may mistakenly filter out the real load signal on bumpy roads, while linear compensation models constructed under uniform speed conditions cannot respond to nonlinear distortions caused by random impacts; and statically calibrated sensors accumulate errors under continuous varying conditions due to factors such as temperature drift and mechanical friction. These defects collectively lead to a systematic deviation between dynamic load measurements and actual physical loads. Especially when driving at high speeds on unpaved roads, measurement distortion can cause significant engineering risks such as inaccurate vehicle safety assessments and delayed structural overload warnings.

[0004] The information disclosed in this background section is intended only to enhance the understanding of the general background of this disclosure and should not be construed as an admission or in any way implying that the information constitutes prior art known to those skilled in the art. Summary of the Invention

[0005] This invention provides a load detection method and system for multi-terrain shock absorption suspension systems, which can effectively solve the problems in the background art.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A load detection method for a multi-terrain damping suspension system, the method comprising: Establish a geometric correlation model describing the three-dimensional kinematic relationship of the vehicle suspension mechanism; Obtaining chassis attitude angle data and real-time suspension displacement data of each suspension point of the vehicle, inputting the chassis attitude angle data and the real-time suspension displacement data into the geometric correlation model, and solving a three-dimensional pose change vector of each suspension point relative to the vehicle frame; Collecting original strain signals of the shock absorber, performing dynamic signal separation on the three-dimensional pose change vector of each suspension point, filtering out strain interference components caused by the movement of the vehicle suspension mechanism, and generating static load strain values; According to the three-dimensional pose change vector of each suspension point, the real-time suspension displacement data is geometrically corrected to generate displacement representation load values; Detecting the terrain environment type in which the vehicle is currently driving, weighting the displacement representation load values and the static load strain values based on the terrain environment type, and fusing and calculating the weighted results to obtain uncorrected real-time load values of each suspension point; When it is detected that the vehicle is in a low disturbance driving state in which the centroid vertical vibration frequency energy value continuously falls below a preset threshold and exceeds a set time length, the displacement representation load values and the static load strain values are used as a reference to learn and generate nonlinear compensation parameters and to compensate the uncorrected real-time load values in real time to output corrected real-time load values of each suspension point.

[0007] Further, solving the three-dimensional pose change vector of each suspension point relative to the vehicle frame comprises: Loading the spatial constraint conditions of the vehicle suspension mechanism into the geometric correlation model to construct a motion chain topological relationship; Decomposing the real-time suspension displacement data into suspension stroke components and flexible deformation components and inputting them into the motion chain topological relationship; Performing a knuckle rotation transformation in the geometric correlation model on the chassis attitude angle data and the suspension stroke components to derive a wheel center position offset; Based on the flexible deformation components, performing an elastic correction compensation on the wheel center position offset to generate hard point coordinate update values; Establishing a conversion relationship between the vehicle body coordinate system and the ground reference coordinate system in the geometric correlation model and calculating a coordinate transformation matrix; According to the hard point coordinate update values, inversely solving the spatial attitude quaternions of each suspension point based on the coordinate transformation matrix, and performing Lie group differential operation on the spatial attitude quaternions to continuously derive and evolve the three-dimensional pose change vector.

[0008] Further, generating static load strain values comprises: According to the three-dimensional pose change vector of each suspension point, constructing an institution motion coupling template signal; Performing time-domain decoupling operation of complex exponential modulation on the original strain signal separates a dynamic strain base related to vehicle posture; Generating phase-synchronous interference wave by cross-correlation interference calculation of the mechanism motion coupling template signal and the dynamic strain base; Performing coherent suppression on the original strain signal with the phase-synchronous interference wave as reference signal to generate load strain initial value; Cascading derivative terms of the three-dimensional pose change vector to perform dynamic leakage compensation on the load strain initial value to output the static load strain value.

[0009] Further, geometric correction of the real-time suspension displacement data according to the three-dimensional pose change vector of each suspension point includes: Analyzing the three-dimensional pose change vector into a spatial displacement projection component of dynamic deformation of the vehicle suspension mechanism; Decomposing the real-time suspension displacement data into a load-sensitive vector and a geometric interference vector according to a real-time construction motion coordinate system conversion matrix; Separating a deformation interference quantity corresponding to the spatial displacement projection component from the geometric interference vector, and eliminating the deformation interference quantity in the load-sensitive vector based on a dynamic inverse compensation mechanism to generate a geometric correction load-sensitive vector; Converting the physical dimension of the geometric correction load-sensitive vector to generate the displacement representation load value.

[0010] Further, decomposing the real-time suspension displacement data into a load-sensitive vector and a geometric interference vector according to a real-time construction motion coordinate system conversion matrix includes: Dynamically solving displacement parameters of the instantaneous rotation center of the vehicle suspension mechanism from the three-dimensional pose change vector and taking them as the coordinate system reference origin, and determining the pointing directions of three mutually orthogonal coordinate axes according to the spatial constraint conditions of the geometric correlation model; Constructing a dynamic reference coordinate system with the coordinate system reference origin and the coordinate axis pointing directions, mapping the real-time suspension displacement data into an orthogonal projection space of the dynamic reference coordinate system to generate original projection vector components; Identifying a vertical load-bearing axis sensitive to suspension load in the orthogonal projection space, and obtaining the projection vector component in the vertical load-bearing axis as the load-sensitive vector; Synthesizing the remaining projection vector components in the orthogonal projection space except the vertical load-bearing axis to constitute the geometric interference vector representing mechanism deformation interference.

[0011] Further, weighting the displacement representation load value and the static load strain value based on the terrain environment type includes: analyzing the current characteristics of the terrain environment type to generate a dynamic confidence level of the displacement representation load value and the static load strain value; adjusting the contribution intensity ratio of the displacement representation load value and the static load strain value based on the dynamic confidence level; performing a displacement weighting operation on the displacement representation load value combined with the contribution intensity ratio to obtain a spatial dimension weighted load value; performing a strain weighting operation on the static load strain value combined with the contribution intensity ratio to obtain a time dimension weighted strain value; integrating the spatial dimension weighted load value and the time dimension weighted strain value to generate a synthesized load intermediate value; performing a load representation conversion calculation on the synthesized load intermediate value to generate the uncorrected real-time load value.

[0012] Further, learning to generate a nonlinear compensation parameter, comprising: When the vehicle is determined to be in a low disturbance driving state, synchronously collecting the displacement representation load value sequence and the static load strain value sequence to generate a load representation sample pair; calculating the difference value of the load representation sample pair at the same time for each suspension point to generate a dynamic error sequence; Based on the change trend of the dynamic error sequence, a compensation factor increment distribution map is constructed, and the nonlinear variation law of displacement and strain coupling deviation is identified according to the compensation factor increment distribution map; Combined with the nonlinear variation law, the compensation learning step value is dynamically updated, and the relative correction coefficient between the displacement representation load value and the static load strain value is iteratively generated based on the compensation learning step value; Mapping the relative correction coefficient to the polynomial fitting parameters of the suspension system load transfer curve forms a nonlinear parameter table; Integrating the nonlinear parameter table of all the suspension points at the current time outputs the nonlinear compensation parameter.

[0013] Further, real-time compensation is performed on the uncorrected real-time load value to output the real-time load value of each suspension point after correction, comprising: Invoking the nonlinear compensation parameter combined with the uncorrected real-time load value at the current time to perform nonlinear deviation compensation calculation to generate an instantaneous correction amount corresponding to each suspension point; Superimposing the instantaneous correction amount on the uncorrected real-time load value to generate a compensated linear prediction load value; Based on the suspension motion compensation ability verification mechanism of the geometric correlation model, real-time inversion checking is performed on the linear prediction load value, and the compensation residual error is adjusted during the real-time inversion checking process to output a stable load prediction sequence; The stable load prediction sequence is converted into a dimensionally normalized value consistent with the physical unit of the vehicle load to generate a vehicle load numerical stream; The vehicle load numerical stream is applied to update the load state data of each suspension point in real time, and real-time load values of each suspension point after correction are output.

[0014] The load detection system of the multi-terrain shock-absorbing suspension system comprises: A motion correlation module is configured to establish a geometric correlation model describing the three-dimensional kinematic relationship of the vehicle suspension mechanism; A pose solving module is configured to obtain chassis attitude angle data and real-time suspension displacement data of each suspension point, and input the chassis attitude angle data and the real-time suspension displacement data into the geometric correlation model to solve a three-dimensional pose change vector of each suspension point relative to the vehicle frame; A signal separation module is configured to collect original strain signals of the shock absorber, and perform dynamic signal separation on the three-dimensional pose change vector of each suspension point to filter out strain interference components caused by the motion of the vehicle suspension mechanism, and generate static load strain values; A geometric correction module is configured to perform geometric correction on the real-time suspension displacement data according to the three-dimensional pose change vector of each suspension point to generate displacement-represented load values; A terrain weighting module is configured to detect the terrain environment type in which the vehicle is currently driving, weight the displacement-represented load values and the static load strain values based on the terrain environment type, and fuse and calculate the weighting results to obtain uncorrected real-time load values of each suspension point; A parameter compensation module is configured to, when detecting that the vehicle is in a low disturbance driving state in which the vertical vibration frequency energy value of the center of mass continuously falls below a preset threshold value and exceeds a set time length, learn to generate nonlinear compensation parameters based on the displacement-represented load values and the static load strain values, and compensate the uncorrected real-time load values in real time to output real-time load values of each suspension point after correction.

[0015] Further, the geometric correction module comprises: A displacement projection unit is configured to analyze the three-dimensional pose change vector into a spatial displacement projection component of the dynamic deformation of the vehicle suspension mechanism; A geometric decomposition unit is configured to decompose the real-time suspension displacement data into a load-sensitive vector and a geometric interference vector according to a real-time motion coordinate system conversion matrix; A load correction unit is configured to separate a deformation interference quantity corresponding to the spatial displacement projection component from the geometric interference vector, and eliminate the deformation interference quantity in the load-sensitive vector based on a dynamic inverse compensation mechanism to generate a geometric correction load-sensitive vector; A load representation unit is configured to convert the physical dimension of the geometric correction load-sensitive vector to generate displacement-represented load values.

[0016] The technical scheme of the present application can achieve the following technical effects: By establishing a three-dimensional pose change dynamic model of the vehicle suspension mechanism, separating the motion interference components in the displacement and strain signals, combining the terrain adaptive weighted fusion and the nonlinear real-time learning compensation mechanism in the low disturbance state, decoupling the coupling effect of mechanical motion and real load, the dynamic load measurement distortion problem caused by the multi-degree-of-freedom motion of the suspension system in the complex driving environment is solved, and the high-precision continuous detection of the suspension load in the multi-terrain scene is realized.

[0017] The above description is only a summary of the technical solutions of the present application. In order to more clearly understand the technical means of the present application, the specific embodiments of the present application can be implemented according to the content of the specification, and in order to make the above and other purposes, characteristics and advantages of the present application more obvious and easy to understand, the following specific embodiments of the present application are described. BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are only some embodiments described in the present application, and those skilled in the art can obtain other drawings according to these drawings without creating any creative labor.

[0019] Figure 1 Flowchart for load detection method of multi-terrain shock absorption suspension system; Figure 2 Flowchart for solving three-dimensional pose change vector; Figure 3 Flowchart based on terrain environment type weighting; Figure 4 Flowchart for learning to generate nonlinear compensation coefficients. DETAILED DESCRIPTION

[0020] The technical solutions in the embodiments of the present application will be described clearly and completely in the following with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in the description of the present application is only for the purpose of describing specific embodiments and is not intended to limit the present application. The term "and / or" used herein includes any and all combinations of one or more related listed items.

[0022] Embodiment one; As Figure 1As shown, the present application provides a load detection method for a multi-terrain shock-absorbing suspension system, the method comprising: establishing a geometric correlation model describing the three-dimensional kinematic relationship of the vehicle suspension mechanism; obtaining chassis attitude angle data and real-time suspension displacement data of each suspension point, inputting the chassis attitude angle data and real-time suspension displacement data into the geometric correlation model, and solving the three-dimensional pose change vector of each suspension point relative to the vehicle frame; collecting the original strain signal of the shock absorber, combining the three-dimensional pose change vector of each suspension point to perform dynamic signal separation, filtering out the strain interference components caused by the movement of the vehicle suspension mechanism, and generating static load strain values; geometrically correcting the real-time suspension displacement data according to the three-dimensional pose change vector of each suspension point to generate displacement-represented load values; detecting the terrain environment type in which the vehicle is currently driving, weighting the displacement-represented load values and the static load strain values based on the terrain environment type, and fusing and calculating the weighted results to obtain uncorrected real-time load values of each suspension point; when it is detected that the vehicle is in a low disturbance driving state in which the centroid vertical vibration frequency energy value continuously falls below a preset threshold and exceeds a set time length, taking the displacement-represented load values and the static load strain values as the reference, learning to generate nonlinear compensation parameters and real-time compensating the uncorrected real-time load values to output the real-time load values of each suspension point after correction.

[0023] In particular, first, according to the specific type of vehicle suspension structure, such as double wishbone suspension, Macpherson suspension, five-link suspension, etc., the geometric structure of the suspension mechanism is completely modeled, and in the preferred case, the kinematics of each component in the suspension mechanism, such as the swing arm, connecting rod, wheel, etc., is accurately described by using coordinate transformation, and specifically, the vehicle suspension system is simplified as a multi-link structure, each joint connection is modeled as a hinge relationship, and the correlation matrix between each moving part is established; secondly, for the suspension system, the vehicle frame or body is selected as the spatial reference frame, and a fixed reference coordinate system is defined, for example, the longitudinal axis of the vehicle body is defined as the X axis, the transverse axis is defined as the Y axis, and the vertical direction is defined as the Z axis, by analyzing the geometric constraint relationship between the suspension point and the vehicle frame, combining the length of the bar, the motion limiting angle, etc. as known parameters, a mathematical model capable of describing the motion trajectory of the suspension point relative to the vehicle frame in three-dimensional space is established, and in the preferred case, the linear displacement and angular displacement characteristics of all suspension points in three-dimensional space can be refined according to the kinematics of the wheel motion; next, in order to build the correlation model, the influence of wheel and suspension mechanism motion on vehicle attitude needs to be considered, and in the preferred implementation, high-precision displacement sensors, tension and compression sensors can be installed at the suspension points, and combined with the inertial measurement unit (IMU) in the vehicle chassis, the three-dimensional attitude angle data of the chassis is collected in real time, including pitch angle, roll angle and yaw angle. These attitude angle data can dynamically reflect the attitude change of the vehicle during driving, and are used to correct the actual position of the suspension point in the three-dimensional space of the vehicle body; then, the sensor measured data is input into the geometric correlation model, and the calculation method based on coordinate transformation and rigid body kinematics description is used to solve the three-dimensional pose change vector of each suspension point relative to the vehicle frame, including the displacement change and angle change of the suspension point in three directions. Specifically, the three-dimensional kinematics equation can be solved by using the step-by-step iterative convergence method, and the preferred data solving scheme needs to be combined with the real-time data stream to ensure that the model solving result is consistent with the actual motion state of the vehicle; secondly, in the dynamic signal separation step, the original strain signal needs to be collected by high-precision strain sensors, and the signal contains strain interference components caused by the motion of the suspension mechanism. In the preferred implementation process, the three-dimensional pose change vector can be used to separate the dynamic signal by signal filtering algorithm, and the signal noise component processing mechanism can be refined to form accurate static load strain value, and the preferred implementation method can also include dynamically adjusting the key parameters of the filtering and signal separation algorithm according to the vehicle working state, load condition and terrain complexity; in terms of displacement representing load value generation, the real-time suspension displacement data is geometrically corrected according to the calculated three-dimensional pose change vector, and this process preferably includes dynamically adjusting the weight parameters of the displacement representing load correction calculation by combining the suspension arm geometric characteristics and the vehicle body mass distribution parameters, and finally obtaining high-precision load value closely matching the actual motion state of the vehicle;Further, to improve the adaptability of the load detection method, a terrain environment type-based detection mechanism is preferably adopted, which can realize real-time sensing of the terrain environment type in which the vehicle is located, such as flat road, low-disturbance hilly terrain or high-disturbance rugged mountainous terrain, etc., through a terrain sensor or a terrain classification algorithm, and then perform terrain type-based weighted processing on the displacement representing load value and the static load strain value, preferably, a dynamic weight adjustment mechanism is adopted to ensure that the weighted result can accurately reflect the comprehensive influence of the terrain environment on the load data, thereby generating the uncorrected real-time load value of each suspension point; finally, preferably applied to the low-disturbance driving state, the vehicle collects the vertical vibration signal through the inertial sensor unit fixedly installed at the center of the vehicle body structure, performs digital filtering processing on the original vibration signal, and retains the specific low-frequency band reflecting the inherent vibration of the suspension system, which excludes engine high-frequency interference and road high-frequency noise, divides the filtered signal into isochronous analysis segments, performs spectrum conversion processing on each segment, extracts the energy contribution value of all frequency components in the target frequency band, generates the frequency domain energy quantization index of the vertical vibration in this period through integral operation, and the preset energy determination threshold can be dynamically adjusted according to the actual load state of the vehicle, that is, when the vehicle load increases or the suspension system rigidity increases, the determination threshold is correspondingly increased; otherwise, it is lowered, and the foregoing frequency domain energy quantization index is continuously monitored, when the index is continuously low for multiple analysis segments, the total time covers several inherent vibration periods of the vehicle suspension system, and if it is lower than the current dynamic threshold, it is determined that the vehicle is in a low-disturbance driving state, at this time, the vehicle usually drives on a flat paved road such as a highway, a city asphalt road, or maintains a uniform speed or a slow acceleration state, or does not encounter sudden road undulations or obstacle impacts, after detecting this state, a deep learning model or a nonlinear feedback control mechanism is established based on the displacement representing load value and the static load strain value, nonlinear compensation parameters for real-time compensation are generated, and an algorithm iteration optimization technology is preferably adopted to correct the parameter setting of the nonlinear compensation mechanism in real time, so as to ensure that the compensation effect has high adaptive ability to the change of the vehicle suspension load in different terrain environments, and finally the corrected real-time load value of each suspension point is output.

[0024] Through the technical scheme of the present application, a three-dimensional pose change dynamic model of the vehicle suspension mechanism is established, the motion interference components in the displacement and strain signals are separated, the terrain adaptive weighted fusion and the nonlinear real-time learning compensation mechanism in the low-disturbance state are combined, the coupling effect of mechanical motion and real load is decoupled, the dynamic load measurement distortion problem caused by the multi-degree-of-freedom motion of the suspension system in the complex driving environment is solved, and high-precision continuous detection of the suspension load in multiple terrain scenes is realized.

[0025] Further, as shown in Figure 2 , the three-dimensional pose change vector of each suspension point relative to the vehicle frame is calculated, including: loading the spatial constraint conditions of the vehicle suspension mechanism into the geometric correlation model to construct the kinematic chain topology relationship; decomposing the real-time suspension displacement data into suspension stroke components and flexible deformation components and inputting the kinematic chain topology relationship; performing knuckle rotation transformation on the chassis attitude angle data and the suspension stroke components in the geometric correlation model to derive the wheel center position offset; performing elastic correction compensation on the wheel center position offset based on the flexible deformation components to generate hard point coordinate update values; establishing the conversion relationship between the vehicle body coordinate system and the ground reference coordinate system in the geometric correlation model and calculating the coordinate transformation matrix; according to the hard point coordinate update values, inversely solving the spatial posture quaternions of each suspension point based on the coordinate transformation matrix, and performing Lie group differential operation on the spatial posture quaternions to continuously derive and generate three-dimensional pose change vectors.

[0026] As a preferred embodiment of the above embodiment, the spatial constraint conditions of the vehicle suspension mechanism need to be loaded into the geometric correlation model first to construct the kinematic chain topological relationship. The preferred implementation of this process requires first analyzing the structural characteristics of the vehicle suspension mechanism. The spatial constraint conditions of different suspension structures include the movement limits of the suspension points, the swing range, and the length and movement constraints of the suspension rods. Specifically, the kinematic chain topological relationship can be established based on the spatial hinge relationship in the suspension geometry, and a topological framework can be constructed to describe the correlation between each key movement point in the suspension mechanism. Next, the real-time suspension displacement data is decomposed. In the preferred embodiment, the suspension displacement signals collected in real time by the suspension system can be refined and decomposed into suspension stroke components and flexible deformation components. In actual scenarios, the suspension stroke component is usually determined by the extension state changes of the suspension spring and shock absorber, while the flexible deformation component is mainly derived from the slight elastic deformation of the suspension arm or swing arm during vehicle movement. In the preferred case, the data collected by multiple sensors should be separated, and the two components should be accurately decomposed through filtering mechanisms and flexible mechanics analysis to ensure that the data input into the kinematic chain topological relationship has high precision and accuracy. In the geometric correlation model, the steering knuckle rotation variable is preferably transformed according to the suspension stroke component and the chassis attitude angle data to derive the wheel center position offset. In this process, the movement angle of the steering knuckle is first calculated based on the input chassis pitch angle, roll angle, and yaw angle data. The suspension stroke component is combined to form the offset trajectory of the wheel center position. Dynamic coordinate transformation is preferably used to dynamically predict the position changes of the tire center and the suspension geometry in three-dimensional space relative to the vehicle frame. Subsequently, the flexible deformation component is used to elastically correct and compensate the wheel center position offset to generate updated values of the hard point coordinates. The dynamic wheel center position is preferably analyzed through a flexible mechanics model to compensate for the interference of the elastic deformation of the suspension rods on the spatial movement of the wheel center. In the preferred case, the flexible deformation can be calculated by fitting the material properties of the vehicle, such as the material strength and elastic modulus of the suspension arm, to correct the actual wheel center offset and make the updated hard point coordinates more consistent with the actual vehicle movement structure. Subsequently, to establish the conversion relationship between the vehicle body coordinate system and the ground reference coordinate system and calculate the coordinate transformation matrix, a unified reference frame conversion framework needs to be defined based on the movement characteristics of the chassis and the vehicle body. In the preferred case, the vehicle body coordinate system is defined as the center reference system of the suspension system, and the original ground coordinate system is used as the reference to establish the coordinate transformation rules based on the position and angle relationship. The chassis attitude angle data is used as the input to real-time solve the position and attitude change trajectory of each suspension point in the ground reference system. The rotation angle specification change rule in three-dimensional space is used in the conversion process to generate a dynamic coordinate matrix relationship that matches the vehicle movement.Finally, according to the hard point coordinate update value and the coordinate transformation matrix, the spatial posture quaternion of each suspension point is obtained by the inverse solution method. In the preferred embodiment, in order to improve the accuracy of the quaternion solution, dynamic sequence data is collected for compensation and correction, and the effectiveness of the result is improved based on the joint constraint condition of the suspension mechanism. Based on the obtained spatial posture quaternion, in order to maintain the continuity of the posture change, a Lie group differential operation is preferably performed. The Lie group describes the characteristics of continuous transformation, and through differential calculation, the continuous pose change trajectory of the suspension point in the three-dimensional space can be obtained. This preferred processing method can ensure that the calculated three-dimensional pose change vector has time correlation and space linkage, providing necessary support for dynamic real-time monitoring of the suspension system.

[0027] Further, the static load strain value is generated, including: According to the three-dimensional pose change vector of each suspension point, a mechanism motion coupling template signal is constructed; Performing a time domain decoupling operation of complex exponential modulation on the original strain signal to separate the dynamic strain base related to the vehicle posture; Performing mutual correlation interference calculation on the mechanism motion coupling template signal and the dynamic strain base to generate a phase-synchronous interference wave; Performing coherent suppression on the original strain signal with the phase-synchronous interference wave as a reference signal to generate a load strain initial value; Cascading the derivative term of the three-dimensional pose change vector to the load strain initial value for dynamic leakage compensation to output the static load strain value.

[0028] As a preferred embodiment of the above embodiment, first, a mechanism motion coupling template signal needs to be constructed according to the three-dimensional pose change vector of each suspension point, in a preferred embodiment, the three-dimensional pose change vector includes displacement and attitude rotation information of the suspension point in each direction, and the dynamics directly reflects the motion characteristics of the suspension system, based on which, the spatiotemporal distribution characteristics of the key nodes are obtained through the geometric correlation model, the suspension system architecture of the vehicle and the road disturbance response are combined to generate a coupling template signal of the overall motion trajectory of the suspension system, the key point of the coupling template signal is to describe the overall motion state of the vehicle by using the constraint relationship and correlation characteristics between the suspension points, and the non-continuity is reduced through a high-precision interpolation method to ensure that the template signal can dynamically reflect the synchronization relationship between the suspension mechanism and the actual motion of the vehicle in the form of a time series; next, a time-domain decoupling operation of complex exponential modulation is performed on the original strain signal to separate the dynamic strain base related to the vehicle attitude from the strain signal, preferably, the original strain signal is collected by a strain sensor installed on the suspension mechanism, and the signal contains the composite characteristics generated by the superposition of multiple factors such as external load, vehicle attitude change and road disturbance, in order to distinguish the dynamic strain component caused by the vehicle attitude change, the complex exponential modulation method is usually used, and the original strain signal is usually analyzed in the time domain based on the signal time domain mode and frequency response relationship, a preferred embodiment is to combine real-time chassis attitude angle data including roll angle, pitch angle and other synchronous adjustment calculation parameters, gradually decompose the signal through a dynamic filtering algorithm, and finally retain the dynamic strain base coupled with the vehicle motion; further, the mechanism motion coupling template signal and the dynamic strain base are subjected to cross-correlation interference calculation to generate a phase-synchronous interference wave, in a preferred embodiment, the synchronization relationship between the two signals is calculated by using the cross-correlation function, the motion coupling degree is reflected from the phase difference characteristics of the signal, and the cross-correlation result reflects the time and amplitude relationship between the vehicle attitude change vibration and the original strain response, which can effectively identify the anti-interference characteristics of the system, and further construct the fluctuation profile of the dynamic interference, the phase-synchronous interference wave obtained through this process can be used as a reference signal for subsequent coherent suppression; subsequently, the original strain signal is subjected to coherent suppression based on the phase-synchronous interference wave to further generate a load strain initial value, in a preferred embodiment, the coherent suppression process compares the amplitude and phase of the original strain signal and the dynamic interference signal, eliminates the strain component related to the coupling of the vehicle attitude and the suspension motion, and only retains the strain part related to the external load, in order to improve the detection accuracy, a successive iteration filtering adjustment method is preferably used to further reduce noise and suppress additional information such as noise components and frequency domain interference of the original signal, so that the generated load strain initial value is closer to the ideal state;Finally, through the cascade processing mode, the derivative term of the three-dimensional pose change vector is combined to perform dynamic leakage compensation on the initial load strain value, and finally the static load strain value is generated. The strain leakage error caused by the high-frequency dynamic characteristics of the vehicle motion state is corrected. Under the optimal condition, by fitting and analyzing the time derivative characteristics of the pose change vector, a typical dynamic leakage model is established to correct the dynamic error component in the initial load strain value. In addition, multiple sets of compensation rules can be configured for different road conditions such as rugged terrain and flat road, and the dynamic compensation ratio of the strain signal is continuously adjusted, so that the output static load strain value can cover more complex application scenarios.

[0029] Further, the real-time suspension displacement data is geometrically corrected according to the three-dimensional pose change vector of each suspension point, including: analyzing the three-dimensional pose change vector into a spatial displacement projection component of the dynamic deformation of the vehicle suspension mechanism; decomposing the real-time suspension displacement data into a load-sensitive vector and a geometric interference vector according to the real-time construction motion coordinate system conversion matrix; separating the deformation interference quantity corresponding to the spatial displacement projection component from the geometric interference vector, and eliminating the deformation interference quantity in the load-sensitive vector based on a dynamic inverse compensation mechanism to generate a geometrically corrected load-sensitive vector; converting the physical dimension of the geometrically corrected load-sensitive vector to generate a displacement-represented load value.

[0030] As a preferred embodiment of the above-mentioned embodiment, firstly, the spatial displacement projection component of the dynamic deformation of the vehicle suspension mechanism is analyzed by using the three-dimensional pose change vector, and in the preferred embodiment, the three-dimensional pose change vector includes linear displacement and rotational displacement information of the suspension point. By analyzing the information, the motion trajectory of the suspension system in each direction in the three-dimensional space and the dynamic deformation of the suspension system relative to the vehicle frame are determined. To achieve this analysis process, the spatial vector decomposition technology is preferably combined with the geometric characteristics and dynamic response law of the vehicle suspension system to identify each deformation component in a specific motion coordinate system, thereby embodying the spatial distribution characteristics of vehicle motion. Then, according to the real-time construction of the motion coordinate system conversion matrix, the real-time suspension displacement data is decomposed into a load-sensitive vector and a geometric interference vector. Specifically, in the preferred embodiment, the real-time suspension displacement data includes the spatial position and angle change of each suspension point during vehicle motion. By constructing a coordinate system conversion matrix based on vehicle motion parameters in real time, the original suspension displacement data can be transformed and decomposed to eliminate non-load-related components in suspension motion. The load-sensitive vector reflects the static and dynamic load actually borne by the suspension point, while the geometric interference vector is mainly caused by instantaneous influencing factors such as irregular terrain and vehicle jolt. Then, the deformation interference quantity corresponding to the spatial displacement projection component is separated from the geometric interference vector, and these deformation interference quantities are eliminated in the load-sensitive vector based on a dynamic inverse compensation mechanism to generate a geometric correction load-sensitive vector. In this preferred implementation step, the dynamic behavior pattern of the interference source is determined by combining the suspension motion trajectory with the spatial displacement projection. The dynamic inverse compensation mechanism is used to correct in real time by solving the inverse problem model, iteratively reduce the geometric interference, and realize the correction of the load-sensitive vector. This process involves mathematical processing and algorithm optimization, aiming to ensure the dynamic matching of the displacement transformation relationship and improve the accuracy of load representation. Finally, the physical dimensions of the geometric correction load-sensitive vector are converted to generate displacement representation load values. Preferably, this step needs to convert the load-sensitive vector processed by the geometric correction into the physical dimension of the load signal, such as force or pressure. For this purpose, the characteristics of the vehicle load and the material performance parameters of the suspension system can be combined to convert by applying the corresponding physical laws, ensuring that the obtained load values can reflect the real vehicle working state and the load characteristics of the interaction with the ground. The physical dimension conversion process especially needs to consider the dynamic response of the vehicle under various working conditions to ensure the effective adaptability of the obtained load data under different operating conditions.

[0031] Further, the real-time suspension displacement data is decomposed into a load-sensitive vector and a geometric interference vector according to the real-time construction of the motion coordinate system conversion matrix, comprising: The displacement parameters of the instantaneous center of rotation of the vehicle suspension mechanism are dynamically solved from the three-dimensional pose change vector and taken as the coordinate system reference origin, and the directions of three mutually orthogonal coordinate axes are determined according to the spatial constraint conditions predicted by the geometric correlation model. A dynamic reference coordinate system is constructed with the coordinate system reference origin and the coordinate axis pointing direction, and real-time suspension displacement data is mapped into the orthogonal projection space of the dynamic reference coordinate system to generate original projection vector components; A vertical load-bearing axial direction sensitive to the suspension load is identified in the orthogonal projection space, and the projection vector component in the vertical load-bearing axial direction is obtained as a load-sensitive vector; The remaining projection vector components in the orthogonal projection space except the vertical load-bearing axial direction are synthesized to constitute a geometric interference vector representing the mechanism deformation interference.

[0032] As a preferred embodiment of the above-mentioned embodiment, first, the displacement parameter of the instantaneous rotation center of the vehicle suspension mechanism needs to be dynamically solved from the three-dimensional pose change vector, and it is taken as the coordinate system reference origin. The definition of the instantaneous rotation center is based on the dynamic motion characteristics of the suspension system, for example, when the vehicle is driving on an irregular ground, the suspension point will produce instantaneous rotation relative to the ground. By calculating the displacement parameter of the rotation center, the spatial reference origin of the suspension system can be obtained. In the preferred implementation, this position can be obtained by analyzing the dynamic response of the suspension arm and combining real-time dynamic data analysis. Then, the pointing directions of the three mutually orthogonal coordinate axes are determined according to the spatial constraint conditions preset by the geometric correlation model. These coordinate axes usually include the front-rear direction, left-right direction and up-down direction of the vehicle. In this way, a dynamic reference coordinate system is established, which can closely reflect various dynamic change factors of the vehicle suspension during the movement process. Next, the dynamic reference coordinate system is constructed with the coordinate system reference origin and the coordinate axis pointing direction, and the real-time suspension displacement data is mapped into the orthogonal projection space of the dynamic reference coordinate system to generate the original projection vector component. In the preferred implementation, the reference point and the axial direction of the reference coordinate system are dynamically adjusted using real-time vehicle motion data, so that the projection data obtained by mapping through the coordinate system can accurately reflect the instantaneous state of the suspension system. The original projection vector component is a corresponding three-dimensional space vector group generated by the projection mechanism of the coordinate system, which represents the motion trajectory of the suspension point in the reference space. In the orthogonal projection space, the preferred implementation needs to identify the vertical load-bearing axis sensitive to the suspension load to obtain the projection vector component in the vertical load-bearing axis as the load-sensitive vector. The selection of the vertical load-bearing axis is based on the load-bearing characteristics of the vehicle and the dynamic response of the suspension structure. Usually, it is perpendicular to the ground direction and connected with the load to achieve the preferred identification. The most influential axis can be identified by combining the terrain sensor and the mechanical property data of the suspension system. This axis is the axis that bears the load and directly affects the suspension movement. The vector component projected on it can most accurately reflect the real-time change characteristics of the load. Finally, the remaining projection vector components in the orthogonal projection space except the vertical load-bearing axis are synthesized to form a geometric interference vector representing the mechanism deformation interference. In the preferred implementation process, the geometric interference vector usually contains the projections of various factors such as terrain complexity, vehicle dynamic movement and suspension mechanism deformation. By real-time calculation and synthesis of the remaining projection vectors, a geometric interference vector representing the dynamic deformation influence of the suspension system can be formed. These interference factors are mainly non-inertial factors that cause deviations in load detection. In the preferred implementation, real-time correction is needed through dynamic algorithms to ensure that the analysis of the interference vector can accurately reflect the vehicle movement and road conditions, providing necessary adjustment reference for subsequent load calculation.

[0033] Further, as shown in Figure 3 the displacement representing load value and the static load strain value are weighted based on the terrain environment type, including: analyzing the current characteristics of the terrain environment type to generate a dynamic confidence level for the displacement representation load values and the static load strain values; adjusting the contribution intensity ratio of the displacement representation load values and the static load strain values based on the dynamic confidence level; performing a displacement weighting operation on the displacement representation load values in combination with the contribution intensity ratio to obtain spatial dimension weighted load values; and performing a strain weighting operation on the static load strain values in combination with the contribution intensity ratio to obtain time dimension weighted strain values; integrating the spatial dimension weighted load values and the time dimension weighted strain values to generate a resultant load intermediate value; performing a load representation conversion calculation on the resultant load intermediate value to generate an uncorrected real-time load value.

[0034] As a preferred embodiment of the above-mentioned embodiment, first, the current characteristics of the terrain environment are analyzed to generate a dynamic confidence level of displacement representation load values and static load strain values, preferably, the terrain environment characteristics can be obtained in real time through terrain classification sensors or terrain recognition algorithms, for example, distinguishing whether the vehicle is driving on a flat road, rough ground or mountain path, these terrain types directly affect the suspension dynamics and strain characteristics of the vehicle, according to this information, a dynamic confidence model can be constructed, so that different confidence levels are given to the output results of different parameters, to ensure that the detection results are more reliable; then, the contribution intensity ratio of displacement representation load values and static load strain values is adjusted based on the dynamic confidence level, in the preferred implementation process, in order to ensure the accuracy and balance of the results, a weight distribution mechanism can be established, and the contribution ratio of load values and strain values in comprehensive detection is dynamically adjusted according to the actual terrain complexity and dynamic response, for example, if the terrain is complex and unstable, the weight of the static load strain value should be increased to reduce the error caused by terrain disturbance, through reasonable proportion adjustment, the real load state of the vehicle can be better reflected; then, the displacement representation load value is combined with the contribution intensity ratio to perform displacement weighting operation to obtain a spatial dimension weighted load value, in the preferred implementation, the displacement representation load value is weighted to make it have a higher spatial detection weight in a specific terrain environment, combined with the contribution intensity ratio, the dynamic response of the terrain environment is directly related to the spatial displacement change of the suspension system to output the most representative spatial load value, this process ensures that the direct influence of the terrain on the load change of the suspension is always accurately reflected; at the same time, the static load strain value is combined with the contribution intensity ratio to perform strain weighting operation to obtain a time dimension weighted strain value, the time dimension weighting should be more focused on stable terrain, because stable terrain reduces dynamic disturbance, at this time, the strain value can more accurately reflect the load influence, preferably, through dynamic strain weighting, the short-term local disturbance influence is filtered out, so that the strain signal can represent the long-term load effect; further, the spatial dimension weighted load value and the time dimension weighted strain value are integrated to generate a synthesized load intermediate value, in the preferred implementation process, this integration process integrates the feature data of different loading dimensions by fusing the feature data of different loading dimensions and combining the comprehensive weight of the input signal, to integrate the feature information of spatial and time dynamic response together to provide an accurate load profile; finally, the synthesized load intermediate value is subjected to load representation conversion calculation to generate an uncorrected real-time load value, in the preferred implementation, the conversion calculation involves further normalization processing of the integrated result, so that it can be used for multi-dimensional load detection, through real-time correction and conversion of the synthesized load, the uncorrected real-time load value output can intuitively reflect the comprehensive load state of the vehicle suspension system in the current terrain environment.

[0035] Further, as shown in Figure 4 the learning generates non-linear compensation parameters, including: When the vehicle is determined to be in a low-disturbance driving state, a load representation sample pair is generated by synchronously collecting a displacement representation load value sequence and a static load strain value sequence; A dynamic error sequence is generated by calculating the difference value of the load representation sample pair at the same time for each suspension point; A compensation factor increment distribution map is constructed based on the change trend of the dynamic error sequence, and a nonlinear change law of displacement and strain coupling deviation is identified according to the compensation factor increment distribution map; The compensation learning step value is dynamically updated in combination with the nonlinear change law, and a relative correction coefficient between the displacement representation load value and the static load strain value is iteratively generated based on the compensation learning step value; The relative correction coefficient is mapped to the polynomial fitting parameters of the suspension system load transmission curve to form a nonlinear parameter table; The nonlinear compensation parameters are output by integrating the nonlinear parameter tables of all suspension points at the current time.

[0036] As a preferred embodiment of the above-mentioned embodiment, firstly, when the vehicle is determined to be in a low-disturbance driving state, it is necessary to synchronously collect the displacement representation load value sequence and the static load strain value sequence to generate a load representation sample pair. In the preferred implementation process, the state of the vehicle in a smooth road or similar low-disturbance driving environment is determined through environmental monitoring and vehicle dynamic judgment. This state recognition is determined through the comprehensive judgment of suspension dynamic change amplitude and road information, ensuring that the disturbance factors are minimized when collecting samples. The displacement representation load value sequence reflects the change of the vehicle suspension displacement, while the static load strain value sequence shows the strain condition borne by the vehicle frame. The combination of the two constitutes a complete load representation sample pair. Next, the difference value of each suspension point is calculated to generate a dynamic error sequence at the same time for each load representation sample pair. This step involves comparative analysis of sample data. By calculating the instantaneous difference between displacement and strain, the dynamic deviation between load and strain representation is identified. In the preferred case, a precise time synchronization algorithm is needed to ensure that the generated dynamic error sequence can truly reflect the influence of the current environment on the change of suspension characteristics. The difference value should be able to show the microscopic error characteristics of different suspension points in a low-disturbance state. Then, according to the trend of the dynamic error sequence, a compensation factor increment distribution map is constructed. Through the information in the map, the nonlinear variation law of the displacement and strain coupling deviation is identified. In the preferred implementation, the combination of dynamic error sequence analysis and node trend extraction enables the compensation factor increment to reflect complex coupling deviations, such as how the terrain characteristics cause static and dynamic deviations in displacement and strain at certain times. However, the coupling changes caused by different types of terrain need to be adjusted through curve fitting to achieve the best compensation effect. Next, the compensation learning step value is dynamically updated based on the above nonlinear variation law. Through the compensation learning step value, the relative correction coefficient between the displacement representation load value and the static load strain value is iteratively generated. In the preferred case, based on adaptive learning and timely response to dynamic strain effects with terrain changes, the step value can be flexibly adjusted in different states. The relative correction coefficient is calculated through multiple iterations, with each iteration aiming to reduce error deviation to make the reflection between load value and strain value more accurate. After that, the relative correction coefficient is mapped to the polynomial fitting parameters of the suspension system load transfer curve to form a nonlinear parameter table. In the preferred implementation, the nonlinear parameter table reflects the load transfer relationship between suspension points in the form of a polynomial, including important geometric and mechanical characteristics. The polynomial fitting parameters ensure that the behavior of the suspension system under different load and strain conditions can be accurately predicted. Integrating the relative correction coefficient as a key feature parameter ensures that the suspension dynamic response can be predicted and adjusted within the nonlinear range.Finally, the nonlinear parameter table of all suspension points at the current time is integrated to output the nonlinear compensation parameters. In the preferred embodiment, the compensation parameter table provides a basis for vehicle control and suspension tuning by comprehensively analyzing the load transfer characteristics of the suspension system. The nonlinear compensation information of each suspension point is integrated to form an overall compensation strategy to ensure that the load detection of the suspension system adapts to various dynamic strain effects in different environments.

[0037] Further, the real-time compensation outputs the real-time load values of each suspension point after correction, including: The nonlinear compensation parameters are called to perform nonlinear deviation compensation calculation in combination with the uncorrected real-time load values at the current time to generate the instantaneous correction amount corresponding to each suspension point. The instantaneous correction amount is superimposed on the uncorrected real-time load values to generate the compensated linear predicted load values. The suspension motion compensation capability verification mechanism based on the geometric correlation model performs real-time inversion verification on the linear predicted load values. In the real-time inversion verification process, the compensation residual error is adjusted to output a stable load prediction sequence. The stable load prediction sequence is converted and processed into dimensionally normalized values consistent with the physical units of the vehicle load to generate a vehicle load numerical stream. The vehicle load numerical stream is applied to real-time update the load state data of each suspension point to output the real-time load values of each suspension point after correction.

[0038] As a preferred embodiment of the above embodiment, first, when performing the nonlinear deviation compensation calculation, the generated nonlinear compensation parameters need to be called, which are derived from the previously constructed nonlinear parameter table and combined with the uncorrected real-time load value measured at the current time. The purpose of this step is to identify and adjust the inherent deviation in the load calculation of each suspension point. Through compensation calculation, a corresponding instantaneous correction amount is generated for each suspension point. In the preferred implementation process, a quick response to various terrains and suspension dynamic changes is required to ensure accurate real-time correction. Then, the above instantaneous correction amount is superimposed on the uncorrected real-time load value, thereby generating a compensated linear predicted load value. The compensated load value reflects a more accurate vehicle load state after compensation correction. It is required to quickly perform superimposed calculation under dynamic conditions. In the preferred case, the influence of suspension dynamics and nonlinear characteristics needs to be considered during calculation to ensure that the corrected load value remains consistent under different loads. Subsequently, the linear predicted load value is executed by the suspension motion compensation capability verification mechanism of the geometric correlation model. In this verification process, the model is used for real-time feedback adjustment to determine whether there are any uncorrected residual errors. If there are, continue to adjust to output a stable load prediction sequence. It not only relies on a single compensation result, but also dynamically adjusts the feedback mechanism to minimize errors and achieve stability and accuracy of the load value. Further, the stable load prediction sequence is converted into a dimensionally normalized value consistent with the physical units of the vehicle load to generate a vehicle load numerical stream. This conversion process ensures the comparability and stability of the load data stream delivered under long time domain and different environments through multidimensional data constraints and physical unit consistency correction, making the load information easy to analyze and apply subsequently. Finally, the vehicle load numerical stream is used to update the load state data of each suspension point in real time, outputting the corrected real-time load value of each suspension point. This update mechanism continuously accesses real-time data information to adjust the load state of the current suspension point, so that the load value output at any time during dynamic operation has high accuracy and site adaptability.

[0039] Embodiment two; Based on the same inventive concept as the load detection method of the multi-terrain shock-absorbing suspension system in the foregoing embodiments, the present application also provides a load detection system for a multi-terrain shock-absorbing suspension system, which comprises: a motion correlation module for establishing a geometric correlation model describing the three-dimensional kinematic relationship of the vehicle suspension mechanism; a pose solving module for obtaining chassis attitude angle data of the vehicle and real-time suspension displacement data of each suspension point, inputting the chassis attitude angle data and real-time suspension displacement data into the geometric correlation model, and solving a three-dimensional pose change vector of each suspension point relative to the vehicle frame; The signal separation module collects the original strain signal of the shock absorber, combines the three-dimensional pose change vector of each suspension point to perform dynamic signal separation, filters out the strain interference components caused by the movement of the vehicle suspension mechanism, and generates a static load strain value. The geometry correction module performs geometry correction on the real-time suspension displacement data according to the three-dimensional pose change vector of each suspension point, and generates a displacement representing load value. The terrain weighting module detects the terrain environment type in which the vehicle is currently driving, weights the displacement representing load value and the static load strain value based on the terrain environment type, and fuses and calculates the weighted results to obtain the uncorrected real-time load value of each suspension point. The parameter compensation module, when detecting that the vehicle is in a low disturbance driving state in which the centroid vertical vibration frequency energy value continuously falls below a preset threshold and exceeds a set time length, learns to generate nonlinear compensation parameters based on the displacement representing load value and the static load strain value, and real-time compensates the uncorrected real-time load value to output the real-time load value of each suspension point after correction.

[0040] The above adjustment system in the present application can effectively realize the load detection method of the multi-terrain shock-absorbing suspension system, and the technical effects are as described in the above embodiments, which will not be repeated here.

[0041] Further, the geometry correction module comprises: The displacement projection unit analyzes the three-dimensional pose change vector into a spatial displacement projection component of the dynamic deformation of the vehicle suspension mechanism; The geometry decomposition unit decomposes the real-time suspension displacement data into a load sensitive vector and a geometric interference vector according to a real-time construction motion coordinate system conversion matrix; The load correction unit separates the deformation interference quantity corresponding to the spatial displacement projection component from the geometric interference vector, and eliminates the deformation interference quantity in the load sensitive vector based on a dynamic inverse compensation mechanism to generate a geometry corrected load sensitive vector; The load representation unit converts the physical dimension of the geometry corrected load sensitive vector to generate a displacement representing load value.

[0042] Similarly, the above optimization scheme of the system can also correspondingly realize the optimization effect of the method in Embodiment 1, which will not be repeated here.

[0043] Although the present application has been described in connection with certain specific features and embodiments thereof, it is to be understood that it is provided as an example to the best of the applicant's knowledge and that various modifications and combinations will occur to those skilled in the art. Accordingly, all modifications, combinations and equivalents that fall within the scope of the application are intended to be included herein. It is evident that those skilled in the art can, without departing from the scope of the application, make various changes and modifications of the application to adapt it to various usages and conditions. Thus, such changes and modifications are intended to be included within the scope of the application as defined in the appended claims.

Claims

1. A load detection method for a multi-terrain damping suspension system, characterized in that, The method includes: Establish a geometric correlation model describing the three-dimensional kinematic relationship of the vehicle suspension mechanism; Acquire the chassis attitude angle data and real-time suspension displacement data of each suspension point of the vehicle, input the chassis attitude angle data and the real-time suspension displacement data into the geometric association model, and calculate the three-dimensional pose change vector of each suspension point relative to the frame. The original strain signal of the shock absorber is collected, and dynamic signal separation is performed by combining the three-dimensional pose change vector of each suspension point to filter out the strain interference component caused by the movement of the vehicle suspension mechanism and generate static load strain value. Geometric correction is performed on the real-time suspension displacement data based on the three-dimensional pose change vector of each suspension point to generate a displacement characterization load value. The terrain environment type where the vehicle is currently driving is detected. The displacement characterization load value and the static load strain value are weighted based on the terrain environment type. The weighted results are then fused to calculate the uncorrected real-time load value of each suspension point. When the vehicle is detected to be in a low-disturbance driving state where the vertical vibration frequency domain energy value of the center of mass is continuously lower than a preset threshold and exceeds a set time, nonlinear compensation parameters are learned and generated based on the displacement characterization load value and the static load strain value, and the uncorrected real-time load value is compensated in real time to output the corrected real-time load value of each suspension point.

2. The load detection method for a multi-terrain damping suspension system according to claim 1, characterized in that, Solving for the three-dimensional pose change vectors of each suspension point relative to the vehicle frame includes: The spatial constraints of the vehicle suspension mechanism are loaded into the geometric association model to construct the kinematic chain topology; The real-time suspension displacement data is decomposed into suspension travel components and flexible deformation components, which are then input into the kinematic chain topology. The chassis attitude angle data and the suspension travel component are subjected to a steering knuckle rotation transformation in the geometric association model to derive the wheel center position offset. Based on the flexible deformation component, elastic correction compensation is performed on the wheel center position offset to generate hard point coordinate update values; In the geometric association model, the transformation relationship between the vehicle coordinate system and the ground reference coordinate system is established and the coordinate transformation matrix is ​​calculated; Based on the updated hard point coordinates, the spatial attitude quaternions of each suspension point are solved by inversely solving the coordinate transformation matrix, and the three-dimensional pose change vector is continuously generated by performing Lie group differential operations on the spatial attitude quaternions.

3. The load detection method for a multi-terrain damping suspension system according to claim 1, characterized in that, Generate static load strain values, including: Construct a mechanism motion coupling template signal based on the three-dimensional pose change vector of each suspension point; A time-domain decoupling operation involving complex exponential modulation is performed on the original strain signal to separate the dynamic strain substrate related to the vehicle attitude. The phase-synchronous interference wave is generated by cross-correlation interference calculation between the motion coupling template signal of the mechanism and the dynamic strain substrate. Using the phase synchronization interference wave as a reference signal, coherent suppression is performed on the original strain signal to generate an initial value of the load strain. The derivative terms of the cascaded three-dimensional pose change vector are used to dynamically compensate for leakage in the initial load strain value, and the static load strain value is output.

4. The load detection method for a multi-terrain damping suspension system according to claim 1, characterized in that, Geometric correction is performed on the real-time suspension displacement data based on the three-dimensional pose change vector of each suspension point, including: The three-dimensional pose change vector is analyzed into the spatial displacement projection components of the dynamic deformation of the vehicle suspension mechanism. Based on the real-time constructed motion coordinate system transformation matrix, the real-time suspension displacement data is decomposed into load-sensitive vectors and geometric disturbance vectors. The deformation disturbance amount corresponding to the spatial displacement projection component is separated from the geometric disturbance vector, and the deformation disturbance amount is eliminated in the load sensitive vector based on the dynamic inverse compensation mechanism to generate a geometrically corrected load sensitive vector. The displacement characterization load value is generated by transforming the physical dimension of the geometrically corrected load sensitivity vector.

5. The load detection method for a multi-terrain damping suspension system according to claim 4, characterized in that, Based on the real-time constructed motion coordinate system transformation matrix, the real-time suspension displacement data is decomposed into load-sensitive vectors and geometric disturbance vectors, including: The displacement parameters of the instantaneous rotation center of the vehicle suspension mechanism are dynamically solved from the three-dimensional pose change vector and used as the reference origin of the coordinate system. At the same time, the orientation of three mutually orthogonal coordinate axes is determined according to the spatial constraints of the geometric association model. A dynamic reference coordinate system is constructed using the origin of the coordinate system and the direction of the coordinate axes. The real-time suspension displacement data is then mapped to the orthogonal projection space of the dynamic reference coordinate system to generate original projection vector components. In the orthogonal projection space, identify the vertical bearing axis that is sensitive to the suspended load, and obtain the projection vector component on the vertical bearing axis as the load-sensitive vector; The remaining projection vector components in the orthogonal projection space, excluding those perpendicular to the bearing axis, are synthesized to form the geometric disturbance vector characterizing the deformation disturbance of the mechanism.

6. The load detection method for a multi-terrain damping suspension system according to claim 1, characterized in that, The displacement characterization load value and the static load strain value are weighted based on the terrain environment type, including: Analyze the current characteristics of the terrain environment type to generate dynamic confidence levels for the displacement characterization load value and the static load strain value; The contribution intensity ratio of the displacement characterization load value and the static load strain value is adjusted based on the dynamic confidence level. The displacement-characterized load value is combined with the contribution intensity ratio to perform a displacement-weighted operation to obtain a spatial dimension-weighted load value; the static load strain value is combined with the contribution intensity ratio to perform a strain-weighted operation to obtain a time-dimensional-weighted strain value; The spatially weighted load value and the time-weighted strain value are integrated to generate a synthetic intermediate load value; The uncorrected real-time load value is generated by performing a load characterization transformation calculation on the intermediate value of the synthetic load.

7. The load detection method for a multi-terrain damping suspension system according to claim 1, characterized in that, Learn to generate nonlinear compensation parameters, including: When the vehicle is determined to be in a low-disturbance driving state, the displacement characterization load value sequence and the static load strain value sequence are simultaneously collected to generate load characterization sample pairs; For each suspension point, the difference values ​​of the load characterization sample pairs at the same time are calculated to generate a dynamic error sequence; Based on the changing trend of the dynamic error sequence, a compensation factor increment distribution map is constructed, and the nonlinear changing law of displacement and strain coupling deviation is identified according to the compensation factor increment distribution map; The compensation learning step value is dynamically updated based on the nonlinear change law, and the relative correction coefficient between the displacement characterization load value and the static load strain value is iteratively generated based on the compensation learning step value. The relative correction coefficients are mapped to polynomial fitting parameters of the load transfer curve of the suspension system to form a nonlinear parameter table; The nonlinear compensation parameters are output by integrating the nonlinear parameter table of all the suspension points at the current time.

8. The load detection method for a multi-terrain damping suspension system according to claim 7, characterized in that, Real-time compensation of the uncorrected real-time load value outputs corrected real-time load values ​​for each of the suspending points, including: The nonlinear compensation parameters are called in combination with the uncorrected real-time load value at the current moment to perform nonlinear deviation compensation calculation, generating instantaneous correction amounts corresponding to each suspension point; The instantaneous correction is superimposed on the uncorrected real-time load value to generate a compensated linear predicted load value; The suspension motion compensation capability verification mechanism based on the geometric correlation model performs real-time inversion verification of the linear predicted load value, and adjusts the compensation residual error to output a stable load prediction sequence during the real-time inversion verification process. The stable load prediction sequence is converted into a dimensionally normalized value consistent with the physical unit of vehicle load to generate a vehicle load numerical stream. The load status data of each suspension point is updated in real time using the vehicle load value stream, and the corrected real-time load value of each suspension point is output.

9. A load detection system for a multi-terrain damping suspension system, characterized in that, The system includes: The motion association module establishes a geometric association model that describes the three-dimensional kinematic relationships of the vehicle's suspension mechanism; The pose calculation module acquires the vehicle's chassis attitude angle data and real-time suspension displacement data of each suspension point. It inputs the chassis attitude angle data and real-time suspension displacement data into the geometric association model to calculate the three-dimensional pose change vector of each suspension point relative to the frame. The signal separation module collects the original strain signal of the shock absorber, combines it with the three-dimensional pose change vector of each suspension point to perform dynamic signal separation, filters out strain interference components caused by the movement of the vehicle suspension mechanism, and generates static load strain values. The geometric correction module performs geometric correction on the real-time suspension displacement data based on the three-dimensional pose change vector of each suspension point, and generates displacement characterization load values. The terrain weighting module detects the terrain environment type in which the vehicle is currently driving, weights the displacement characterization load value and static load strain value based on the terrain environment type, and merges the weighting results to calculate the uncorrected real-time load value of each suspension point. The parameter compensation module, when it detects that the vehicle is in a low-disturbance driving state where the vertical vibration frequency energy value of the center of gravity is continuously lower than the preset threshold and exceeds the set time, learns to generate nonlinear compensation parameters based on the displacement characterization load value and the static load strain value, and outputs the corrected real-time load value of each suspension point after real-time compensation of the uncorrected real-time load value.

10. The load detection system for a multi-terrain damping suspension system according to claim 9, characterized in that, The geometry correction module includes: The displacement projection unit resolves the three-dimensional pose change vector into spatial displacement projection components of the dynamic deformation of the vehicle suspension mechanism. The geometric decomposition unit decomposes real-time suspension displacement data into load-sensitive vectors and geometric disturbance vectors based on the real-time constructed motion coordinate system transformation matrix. The load correction unit separates the deformation disturbance corresponding to the spatial displacement projection component from the geometric disturbance vector, and eliminates the deformation disturbance in the load sensitive vector based on the dynamic inverse compensation mechanism to generate a geometrically corrected load sensitive vector. The load characterization unit transforms the physical dimension of the geometrically corrected load sensitivity vector to generate displacement characterization load values.

Images