Drive-by-wire wheel edge steering angle module with locking mechanism and measuring method of drive-by-wire wheel edge steering angle module

Abstract

The invention discloses a drive-by-wire wheel edge steering angle module with a locking mechanism and a measuring method thereof.The angle module comprises a steering driving and executing unit, a locking unit, a sensing unit and a power source unit, the locking unit integrates an electric driving and spring self-locking mechanism and has the active and passive dual-mode locking capacity, and the sensing unit is connected with the power source unit; automatic locking is achieved under the power-off or out-of-control working condition, and the vehicle posture retentivity is enhanced; the sensing unit realizes multi-dimensional detection of a rotation angle, an angular velocity and a locking state. The measurement method is based on dynamic modeling in a locking state and a physical information neural network, angle drift can be corrected in real time, a credible angular velocity estimation result is output, and a confidence interval is constructed. The stability and reliability of the drive-by-wire system in an extreme or non-working state are improved, and the drive-by-wire system is suitable for intelligent driving and commercial vehicle chassis systems.

Description

Technical Field

[0001] The present invention relates to the technical field of automotive power steering, and particularly to a steer-by-wire wheel-side steering angle module with a locking mechanism and a measurement method thereof. Background Art

[0002] With the rapid development of intelligent driving and vehicle electronic control technology, the steer-by-wire wheel-side steering system, with its characteristics of fast response speed, high control accuracy, and strong scalability, gradually replaces the traditional mechanical steering structure and becomes an important technical direction for high-performance chassis systems of commercial vehicles. However, the steer-by-wire system has a reduced structural stiffness due to the cancellation of mechanical linkages, and is prone to shimmy under road disturbances in the non-steering state, affecting handling stability. Existing locking technologies have significant defects:

[0003] 1. Passive mechanical locking (such as wedge blocks, gear locks) has a fast response, but lacks the ability of flexible adjustment;

[0004] 2. Active locking (such as motor closed-loop control) has high energy consumption and insufficient accuracy under strong disturbances, and is difficult to adapt to complex working conditions;

[0005] 3. In addition, due to mechanical wear or environmental disturbances in the locked state, the zero-point offset problem is difficult for traditional static sensors to correct in real time, resulting in a deviation between the control command and the actual state of the vehicle;

[0006] Therefore, there is an urgent need for a locking and measurement solution with both high stiffness and dynamic zero-point correction ability to improve the robustness of the system. Summary of the Invention

[0007] The purpose of this part is to outline some aspects of the embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this part, as well as in the abstract and title of the present application, to avoid obscuring the purpose of this part, the abstract, and the title, but such simplifications or omissions shall not be used to limit the scope of the present invention.

[0008] Therefore, to solve the above technical problems, the present invention provides the following technical solution: A steer-by-wire wheel-side steering angle module with a locking mechanism, comprising:

[0009] A steering drive and execution unit, including a steering motor, a steering motor controller integrated with the steering motor, a worm and worm gear reduction mechanism, and a steering arm; the steering motor drives the steering arm to turn around the kingpin axis through the worm and worm gear reduction mechanism; the steering arm is connected to the steering column through an upper fork arm, a lower fork arm, and a shock absorber, and the steering column is connected to the tire and the in-wheel motor.

[0010] The locking unit includes a locking module disposed below the steering arm column. The locking module includes a locking execution motor, a planetary gear set with a high reduction ratio, and a mechanical wedge locking structure. The locking execution motor drives the wedge to insert into or withdraw from the corresponding locking tooth groove through the planetary gear set with a high reduction ratio, and realizes passive locking through a spring pre-tightening component when power is off.

[0011] The sensing unit includes a steering angle and angular velocity sensor disposed on the steering arm and a locking state sensor disposed on the locking module. The steering angle and angular velocity sensor is used to collect the steering angle and angular velocity, and the locking state sensor is used to detect whether the locking wedge is fully inserted into the slot and feedback the state signal to the upper controller.

[0012] The power supply unit includes a high-voltage power supply and a low-voltage power supply. The high-voltage power supply supplies power to the hub motor, and the low-voltage power supply supplies power to the steering motor controller and the locking motor controller.

[0013] As a preferred solution of the steer-by-wire wheel-side steering angle module with a locking mechanism of the present invention, wherein: the steering arm column is rigidly connected to the worm and worm gear reduction mechanism, and the output end of the worm and worm gear reduction mechanism is rigidly connected to the steering arm to transmit torque.

[0014] As a preferred solution of the steer-by-wire wheel-side steering angle module with a locking mechanism of the present invention, wherein: the locking module further includes:

[0015] A barrier ring, a locking disc and a self-locking housing;

[0016] The self-locking housing is fixed on the output shaft of the planetary gear set with a high reduction ratio, and the locking disc is provided with a plurality of slots matching the mechanical wedge locking structure;

[0017] The barrier ring is rotatably disposed between the self-locking housing and the locking disc, and a baffle corresponding to the slot is provided on the inner side thereof, and the baffle covers the slot opening in the locking state.

[0018] As a preferred solution of the steer-by-wire wheel-side steering angle module with a locking mechanism of the present invention, wherein: the shock absorber is integrated with a suspension air spring, a suspension adjustable damping shock absorber actuator and a suspension air spring actuator, and the suspension actuator controller dynamically adjusts the suspension stiffness and damping according to the vehicle state.

[0019] As a preferred solution of the steer-by-wire wheel-side steering angle module with a locking mechanism of the present invention, wherein: the high-voltage power supply is connected to the hub motor through a high-voltage wire harness, and the low-voltage power supply is connected to the steering motor controller and the locking motor controller through a low-voltage wire harness.

[0020] As a preferred embodiment of the steer-by-wire wheel-side steering angle module with a locking mechanism according to the present invention, wherein: the mechanical wedge locking structure includes two symmetrically arranged locking wedges, which are installed in the radial guiding chutes between the steering arm column and its base; a spiral groove is provided inside the self-locking housing, and a protrusion matching the spiral groove is provided on the locking wedge, and the wedge is driven to move linearly along the radial chute by rotating the self-locking housing.

[0021] The present invention also provides a measurement method for the steer-by-wire wheel-side steering angle module with a locking mechanism as described above, which is characterized in that: it includes the following specific steps:

[0022] Step 1: Establish a dynamic model of the wheel-side angle module in the locked state, collect the original rotation angle and angular velocity signals in the locked state, and perform filtering, noise reduction, and time synchronization processing.

[0023] Step 2: Based on the data processed in Step 1, construct a physics-informed neural network model to dynamically calculate the current zero-position offset of the rotation angle, and correct the original rotation angle signal.

[0024] Step 3: According to the corrected rotation angle signal and angular velocity signal, construct an angular velocity estimator to output the angular velocity value after zero-position correction, and evaluate the credibility of the angular velocity estimation based on the confidence interval.

[0025] As a preferred embodiment of the measurement method for the steer-by-wire wheel-side steering angle module with a locking mechanism according to the present invention, wherein: in Step 2, the physics-informed neural network model transforms the angle drift estimation problem into a minimization problem in the function space through the residual function and functional optimization.

[0026] As a preferred embodiment of the measurement method for the steer-by-wire wheel-side steering angle module with a locking mechanism according to the present invention, wherein: in Step 3, an angular velocity estimation sample sequence is generated through multiple samplings, the confidence mean and confidence standard deviation are calculated, and a bilateral confidence interval is constructed to quantify the uncertainty of the angular velocity estimation result.

[0027] As a preferred embodiment of the measurement method for the steer-by-wire wheel-side steering angle module with a locking mechanism according to the present invention, wherein: in Step 1, the trigger moment of the locked state is determined by the jump of the switch signal of the locked state sensor, and the angle change trend and angular velocity micro-drift data are extracted based on a preset time window.

[0028] The beneficial effects of the present invention:

[0029] 1. By constructing a dual-mode locking structure combining active driving and passive locking, the present invention introduces a flexible transition characteristic while ensuring the locking rigidity, effectively alleviates the attitude offset problem caused by mechanical clearance and slight perturbations, and significantly improves the static stability of the system in the non-steering state.

[0030] 2. By introducing a neural network model based on physical modeling constraints, the present invention dynamically estimates and real-time corrects the angular drift in the locked state, breaks through the dependence on traditional static encoder calibration, improves the real-time performance and accuracy of angle perception, and reduces the "control-state" deviation.

[0031] 3. The present invention uses the mechanisms of multiple sampling and Dropout perturbation induction to construct a sample sequence for angular velocity estimation, and outputs bilateral confidence intervals and confidence scores, realizing quantitative evaluation of the credibility of sensing data, providing auxiliary criteria for system decision-making, and enhancing control robustness.

[0032] 4. The passive locking unit of the present invention can be automatically triggered under abnormal conditions such as power failure and control failure, realizing attitude maintenance without continuous power supply, significantly reducing the standby power consumption of the whole vehicle and improving the redundant safety ability of the system.

[0033] 5. The locking module of the present invention is decoupled from the steering drive unit, which is convenient for modular integration and vehicle-end layout design, and can meet the differentiated requirements of various vehicle platforms for the structural layout and space utilization of the wheel-end system.

[0034] 6. The present invention uses multiple mechanisms such as integrated angle trend extraction, angular velocity micro-drift perception, and weak derivative regular estimation to construct an integrated dynamic perception scheme, effectively coping with the challenges of angle measurement accuracy and control stability under different road conditions, load changes, and extreme temperature and humidity environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts. Among them:

[0036] Figure 1 is a schematic diagram of the overall structure of the present invention.

[0037] Figure 2 is a schematic diagram of the specific structure of the locking module of the present invention.

[0038] Figure 3 is a schematic diagram of the working principle of the present invention.

[0039] In the figure: 100, steering drive and execution unit; 101, steering motor; 102, steering motor controller; 103, worm and worm gear reduction mechanism; 104, steering arm; 105, steering column; 106, upper fork arm; 107, lower fork arm; 108, shock absorber; 108a, suspension air spring; 108b, suspension adjustable damping shock absorber actuator; 108c, suspension air spring actuator; 109, tire; 110, in-wheel motor; 111, steering arm pipe column; 112, suspension actuator controller; 113, braking system;

[0040] 200, locking unit; 201, locking module; 201a, barrier ring; 201b, locking disc; 201c, self-locking housing; 202, locking execution motor; 203, high reduction ratio planetary gear set; 203a, high reduction ratio planetary gear set output shaft; 204, mechanical wedge locking structure; 204a, locking wedge; 205, spring pre-tightening assembly; 206, locking motor controller;

[0041] 300, sensing unit; 301, steering angle and angular velocity sensor; 302, locking state sensor; 303, vehicle frame connection part; 304, mounting bracket;

[0042] 400, power supply unit; 401, high-voltage power supply; 401a, high-voltage wire harness; 402, low-voltage power supply; 402a, low-voltage wire harness. Detailed implementation manners

[0043] To make the above objects, features and advantages of the present invention more obvious and understandable, the following detailed description of the specific implementation manners of the present invention will be made in conjunction with the accompanying drawings of the specification.

[0044] In the following description, many specific details are set forth in order to fully understand the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0045] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure or characteristic that may be included in at least one implementation manner of the present invention. The "in one embodiment" appearing in different places in this specification does not all refer to the same embodiment, nor is it a separate or alternative embodiment that is mutually exclusive with other embodiments.

[0046] Next, the present invention will be described in detail with reference to the schematic diagrams. When describing the embodiments of the present invention in detail, for the convenience of explanation, the cross-sectional views showing the device structure will be enlarged locally not in accordance with the general scale, and the schematic diagrams are only examples and should not limit the scope of protection of the present invention herein. In addition, in actual production, three-dimensional spatial dimensions including length, width, and depth should be included.

[0047] Referring to Figures 1 to 3 , an embodiment of the present invention provides a wire-controlled wheel-side steering angle module with a locking mechanism, which includes structural modules such as a steering drive and execution unit 100, a locking unit 200, a sensing unit 300, a power supply unit 400, etc., as follows:

[0048] The steering drive and execution unit 100 includes structures such as a steering motor 101, a steering motor controller 102, a worm and gear reduction mechanism 103, a steering arm 104, a steering arm column 111, an upper fork arm 106, a lower fork arm 107, a shock absorber 108, a steering column 105, a tire 109, and a hub motor 110. Specifically:

[0049] The steering motor controller 102 is integrally designed with the housing of the steering motor 101. The steering motor controller 102 outputs a signal to control the rotation of the steering motor 101. The output shaft of the steering motor is connected to the input end of the worm and gear reduction mechanism 103. The worm and gear reduction mechanism 103 is rigidly connected to the steering arm column 111 and transmits the torque of the steering motor 101 to the steering arm 104. The lower end of the steering arm column 111 is rigidly connected to the short side of the steering arm 104. The output end of the worm and gear reduction mechanism 103 is rigidly connected to the steering arm 104, which is used to drive the angle module to turn around the kingpin axis.

[0050] The steering arm 104 is connected to the steering column 105 through the shock absorber 108, the upper fork arm 106, and the lower fork arm 107. Specifically, the upper end of the steering column 105 is respectively connected to the lower ends of the upper fork arm 106 and the shock absorber 108. The lower end of the steering column 105 is connected to the upper end of the lower fork arm 107. The upper ends of the upper fork arm 106, the shock absorber 108, and the lower ends of the lower fork arm 107 are respectively connected to the steering arm 104.

[0051] The shock absorber 108 integrates a suspension air spring 108a, a suspension adjustable damping shock absorber actuator 108b, and a suspension air spring actuator 108c. The suspension actuator controller 112 dynamically adjusts the suspension stiffness and damping according to the vehicle state. Among them, the suspension adjustable damping shock absorber actuator 108b and the suspension air spring actuator 108c are installed at the upper end of the shock absorber 108, and the lower end of the shock absorber 108 is installed with the suspension actuator controller 112.

[0052] The central part of the steering column 105 is successively installed with a braking system 113, a hub motor 110 and a tire 109, which are used to realize the drive and braking of the wheel end.

[0053] The locking unit 200 includes a locking module 201 integrally arranged below the steering arm column 111, which is used to lock the angle of the steering arm column 111 in the non-working state to prevent the attitude drift of the wheel end module; the locking module 201 is structurally separated from the worm and gear reduction mechanism 103, does not affect the normal steering control process, and realizes the attitude holding function of the steering module in the non-working state of the system. It is integrally arranged below the steering arm column 111, and a locking state sensor 302 is installed on the locking module 201 to detect the locking state switch signal.

[0054] The locking module 201 includes a locking execution motor 202, a locking motor controller 206, a high reduction ratio planetary gear set 203, a mechanical wedge locking structure 204 and a spring preloading assembly 205; the mechanical wedge locking structure 204 includes two symmetrically arranged locking wedges 204a, which are installed in the radial guide chute between the steering arm column 111 and its base.

[0055] The output shaft of the locking execution motor 202 is connected to the input shaft of the high reduction ratio planetary gear set 203. When the locking execution motor 202 receives the locking or unlocking instruction from the locking motor controller 206, its output shaft starts to rotate, and the torque is amplified and the speed is reduced through the high reduction ratio planetary gear set 203, so as to drive the output shaft 203a of the planetary gear set to rotate; this output shaft 203a is fixedly connected to the self-locking shell 201c, so the self-locking shell 201c rotates accordingly. The locking disc 201b is fixed at the center of the bottom plane of the self-locking shell 201c, and the self-locking shell 201c drives the locking disc 201b to rotate.

[0056] Through the mutual rotation and cooperation of the thread structure inside the self-locking shell 201c and the thread structure on the wedge 204a, the linear motion of the wedge 204a can be realized; specifically, a spiral groove is designed inside the self-locking shell 201c, and a matching protrusion is provided on the wedge 204a; when the self-locking shell 201c rotates, the cooperation between the spiral groove and the protrusion makes the wedge 204a move linearly along the radial guide chute; in the locked state, the wedge 204a is affected by the spring preloading assembly 205 and always maintains a preloading force in the radial direction towards the locking disc 201b to ensure the close contact between the wedge 204a and the locking disc 201b.

[0057] The locking disk 201b is provided with a plurality of slots for cooperating with the wedge block 204a to be inserted therein to achieve locking; the blocking ring 201a is rotatably arranged between the self-locking housing 201c and the locking disk 201b and can rotate relative to each other. A plurality of baffles corresponding to and matching the slots of the locking disk 201b are arranged on the inner side thereof; during the locking process, when the wedge block 204a is completely inserted into the slot of the locking disk 201b, the blocking ring 201a rotates under the action of the spring preloading assembly 205, so that the baffles on the inner side thereof cover the opening of the slot, further blocking the withdrawal of the wedge block 204a and enhancing the reliability of locking;

[0058] The locking execution motor 202 drives the locking wedge block 204a to insert into or withdraw from the slot through the high reduction ratio planetary gear set 203 to achieve steering locking and unlocking; under abnormal conditions such as power failure or controller failure, the spring preloading assembly 205 enables the locking wedge block 204a to automatically insert into the slot to achieve mechanical passive locking; the locking state sensor 302 is used to detect the locking state of the locking wedge block 204a and output a state signal to the upper controller to confirm whether the locking state takes effect.

[0059] The sensing unit 300 includes a steering angle and angular velocity sensor 301 and a locking state sensor 302. The steering arm 104 is vertically installed on the vehicle frame connecting piece 303 and the mounting bracket 304. The steering angle and angular velocity sensor 301 is installed on the steering arm 104 and is used to obtain the steering angle and angular velocity signals in real time; the locking state sensor 302 is installed on the locking module 201 and is used to detect whether the locking wedge block 204a is completely inserted into the slot to confirm the locking state.

[0060] The power supply unit 400 includes a high-voltage power supply 401 and a low-voltage power supply 402. The high-voltage power supply 401 supplies power to the in-wheel motor 110 through the high-voltage wire harness 401a; the low-voltage power supply 402 is respectively connected to the steering motor controller 102 and the locking motor controller 206 through the low-voltage wire harness 402a to provide a stable low-voltage power supply;

[0061] The mounting bracket 304 is used to integrally mount the power supply assembly and the controller assembly and is connected to the steering structure through the vehicle frame connecting piece 303 to achieve an integrated layout, which is convenient for the vehicle layout and maintenance.

[0062] This embodiment also provides an angular velocity dynamic measurement method for the above-mentioned steer-by-wire wheel-side steering angle module with a locking mechanism. The angular velocity dynamic measurement method is applied to the above-mentioned steer-by-wire wheel-side steering angle module with a locking mechanism. The method steps are specifically as follows:

[0063] 1) Establish a dynamic model of the wheel corner module in the locked state. In the wheel corner module, collect the original rotation angle signal and angular velocity signal inside the kingpin column in the locked state through an angle sensor and a lock state sensor. Filter, denoise, and synchronize the collected data in terms of time, and extract the angle change trend, angular velocity drift data, and lock trigger moment as the input for subsequent zero point offset modeling, which specifically includes the following sub-steps:

[0064] 11) Dynamic modeling of the corner module in the locked state; Although the locking mechanism structurally constrains the rotation angle, due to factors such as mechanical clearances, material elastic deformations, and structural looseness, the system actually still has a small ability for free movement, manifested as a slow and low-amplitude angle offset process; This process cannot be described by an ideal static model, and a weak constraint dynamic modeling framework needs to be introduced; In this solution, the angle response process of the corner module in the locked state is simplified to a single-degree-of-freedom equivalent torsional system, and its dynamic equation is expressed as follows:

[0065]

[0066] In the above formula, θ(t) is the actual rotation angle of the wheel in the locked state, θ0 is the theoretical reference angle set by the locking mechanism, that is, the "zero point" read by the sensor; τ d (t) represents the disturbing torque caused by external disturbances such as terrain excitation; J, c, and k respectively represent the equivalent moment of inertia, structural damping, and locking stiffness;

[0067] 12) Acquisition of the original signal; When the steering angle module enters the locked state, the system first obtains the original rotation angle signal θ(t) and angular velocity signal through the angle sensor installed inside the kingpin column At the same time, the switch signal Slock(t) provided by the lock state sensor is used to identify whether the locking mechanism is in the locked state; By detecting the moment when Slock(t) jumps from 0 to 1, the trigger moment t of the locked state can be determined lock , and a local time window for analysis is constructed accordingly:

[0068] [t lock ,t lock +Δt]; (2)

[0069] In the formula, t lock is the trigger moment of the locked state, and Δt is the duration of the time window, which is a preset time increment used to define the length of the time interval for the system to perform analysis after the locked state is triggered;

[0070] 13) Denoise and filter the original signal; Since the original signal of the angle sensor is usually accompanied by high-frequency noise, it is necessary to perform smoothing and filtering first; In this solution, a sliding weighted average filter is used to reduce the noise of the rotation angle and angular velocity signals. The recommended range of the sliding window length N is 10 to 50, and the specific value can be adjusted according to the sampling frequency and signal characteristics; When the sampling frequency is relatively high (such as higher than 100 Hz), a larger sliding window length (such as 30 to 50) can be selected to achieve a better smoothing effect; When the sampling frequency is relatively low (such as lower than 50 Hz), a smaller sliding window length (such as 10 to 20) can be selected to avoid signal delay caused by excessive smoothing. The expression is as follows:

[0071]

[0072] In the formula, N is the sliding window length, and w i is the weighting coefficient, which satisfies the normalization condition ∑w i = 1;

[0073] 14) Extract the drift trend and characteristics; In the locked state, for the possible slow trend change of the angle curve output by the sensor, in this solution, a first-order linear fitting is performed on the smoothed angle sequence within the lock time window:

[0074]

[0075] In the formula, δ θ is the slope coefficient of the drift trend, reflecting the zero-point offset speed in the locked state; t lock is the lock time window.

[0076] 2) Based on the extracted rotation angle change trend, angular velocity micro-drift data, and lock trigger moment inside the kingpin column in step 1, combined with the rotation constraint conditions of the kingpin structure, construct a zero-offset estimation model based on a physics-informed neural network in the locked state, dynamically calculate the offset of the current rotation angle zero position, and compensate and correct the original rotation angle signal in step 1 to obtain the rotation angle signal in the locked state, which is used to calibrate the reference benchmark for subsequent angular velocity estimation. Specifically, it includes the following sub-steps:

[0077] 21) Based on the angular module dynamic equation in step 11), establish the structural modeling layer of the physics-informed neural network, and explicitly transform the physical structure into constraint conditions;

[0078] According to the smoothed angle sequence obtained from the preprocessed sensor data define the angle drift function as:

[0079]

[0080] Substituting it into the angular module dynamic equation in step 11), the residual function of Δθ0(t) can be obtained as shown in Equation (6); this residual is used to measure whether the system conforms to the preset physical laws under the drift hypothesis Δθ0(t).

[0081]

[0082] In the formula, is the smoothed angular velocity sequence;

[0083] 22) Build the physical information neural network state estimation layer and transform the state recovery task into an optimal weak solution problem

[0084] Assume that Δθ0(t) is a smooth variation function, and estimation modeling can be carried out in the space of differentiable functions; for this purpose, this solution transforms the angular drift modeling problem into a functional minimization problem, and introduces the following objective functional in the smooth function space H = H 2 on [0, T]:

[0085]

[0086] In the formula, R(t; f) represents the dynamic residual term after replacing Δθ0(t) with the function f(t), λ1 and λ2 are weight coefficients for balancing different constraints in the objective functional, and T is the integration time;

[0087]

[0088] The three terms in the above functional J[f] respectively correspond to the structural consistency constraint, the derivative response consistency constraint and the time smoothing regularization term;

[0089] 23) Build the physical information neural network neural representation layer and establish the structural embedding and weak derivative solving mechanism for network function approximation;

[0090] Since the aforementioned functional minimization problem is defined on the second-order Sobolev space and its analytical solution is usually not available, a numerical approximation expression of the differentiable function family needs to be found in the function space; considering that the neural network has the universal function approximation ability and has an automatic differentiable structure, this embodiment uses a 3-layer fully connected neural network to construct the function mapping f φ : R 2N →R, each layer contains 64 nodes, and the activation function is ReLU; it is used to approximate the drift function Δθ0(t); let the input be a vector sequence composed of the observation time window:

[0091]

[0092] Then the network output is:

[0093]

[0094] The neural network is defined in the temporal latent space, encoding historical state information through a sliding window to represent the current estimate; since the activation functions of each layer of the network are smooth and differentiable functions, the overall network output belongs to C ∞ , and in the numerical sense, it can be used as an approximation element of integrable and differentiable functions in the functional space;

[0095] Furthermore, with the help of the automatic differentiation mechanism, any-order temporal derivatives of the network output (explicitly constructed with respect to the input x t ) can be gradually constructed on the computational graph through the chain rule of differentiation and written as:

[0096]

[0097]

[0098] In the formula, are the first-order and second-order derivatives of the angular drift estimation function obtained by solving the physics-informed neural network represents the first-order gradient of the network with respect to the input vector, represents the first-order derivative term of the original observation, obtained by finite difference approximation; is the Hessian tensor; the above derivatives are directly substituted into the residual term, angular velocity alignment term, and temporal regularization term in the functional definition to construct the training objective;

[0099] Therefore, the network training process is actually equivalent to searching for a function approximator that satisfies the physical structure and constraint conditions in the function space, and the optimization process can be regarded as the minimization of the projection of the functional on the neural function subspace , that is:

[0100]

[0101] In the formula, R φ (t) is the physical residual automatically constructed based on the network derivatives, E1(t) represents the derivative consistency term, and E2(t) represents the drift change regularization term;

[0102] The training of the neural network is carried out in the way of "offline training first, then online inference deployment"; in the offline training stage, the network is fully trained using historical data, and the training convergence criterion is that the loss function (Loss) is less than 1e-4 or the number of training epochs (Epoch) exceeds 1000; after training is completed, the trained model is deployed to the actual in-wheel motor steering angle module for online inference, to estimate the angular drift in real time and correct the steering angle signal; during the network inference process, Dropout activation is maintained, and a family of functions is constructed through multiple independent samplings, thereby inducing the randomness structure of the drift estimation output to quantify the uncertainty of the estimation results;

[0103] After obtaining the output angle drift of the physical information neural network, the original rotation angle signal measured by the angle sensor in step 12) can be corrected to obtain the rotation angle correction value in the locked state:

[0104]

[0105] 3) Based on the preprocessed angular velocity signal in step 1) and the first derivative of the angle drift estimation function output by the physical information neural network in step 2), construct an angular velocity estimator in the locked state, output the angular velocity estimation value after zero position correction, and construct a bilateral confidence interval based on the confidence mean and confidence standard deviation to obtain the credibility evaluation flag of the angular velocity estimation result, which specifically includes the following sub-steps:

[0106] 31) Establish a function family distribution model induced by physical information neural network parameter perturbation; to express the uncertainty of the output of the neural network induced by parameter perturbation, keep the Dropout activation during training and perform multiple independent samplings during the inference process to construct a set of function families Induce the random characteristics of the angle drift estimation output; the angular velocity estimation sample sequence in the locked state can be expressed as:

[0107]

[0108] where is the angular velocity estimation sample sequence;

[0109] This process defines the posterior distribution of the angular velocity estimation at each moment t Its mean and variance can be given in the following form:

[0110]

[0111] where μ(t) represents the confidence mean of the angular velocity in the locked state under this working condition, and ρ 2 (t) quantifies the uncertainty range of this value, that is, the confidence variance, which is the basic quantity for the subsequent system to judge the trust boundary;

[0112] 32) Confidence expression based on the functional expectation structure; based on the structural residual functional established in step 22):

[0113]

[0114] Under the posterior distribution condition, this minimization objective no longer acts on a single function, but is an expectation functional defined on the function distribution family q φ (f):

[0115]

[0116] where is the desired functional operator, and the functional definition represents the average structural consistency of all function weak solution paths under perturbation activation, q φ (f) is the function distribution of the neural network under parameter perturbation;

[0117] To provide a quantitative confidence assessment of the angular velocity, a bilateral confidence interval based on the confidence mean and the confidence standard deviation σ(t) is defined:

[0118] I CI (t) = [μ(t) - zσ(t), μ(t) + zσ(t)]; (19)

[0119] where z is the confidence coefficient under the standard normal distribution;

[0120] In addition, a continuous confidence scoring function is constructed as shown in Equation (19); the scoring value Γ(t) can be directly used in the fuzzy controller criterion to represent the credibility of the current angular velocity estimation result;

[0121]

[0122] where α is a parameter used to adjust the sensitivity of the confidence scoring function Γ(t) to uncertainty;

[0123] 33) Construct a variance regularization mechanism driven by structural residuals; on the basis of the original structural residual R(t; f), a structure-driven distribution regularization term is introduced to construct the following function:

[0124]

[0125] At the same time, to suppress the random high-frequency instability caused by activation perturbation, a smoothing constraint on the estimated variance is further introduced:

[0126]

[0127] The final training objective integrates the structural functional, the posterior expectation, and the distribution regularization to form the following optimization problem:

[0128]

[0129] where λ3 and λ4 are the weight coefficients of different constraints in the function after introducing the structure drive.

[0130] In this embodiment, by adopting a dual-mode locking mechanism, that is, under the working state, the motor drives the wedge block to be accurately inserted through the planetary reduction system to achieve rigid locking; in the passive state such as abnormal power failure or electrical control failure, the spring pre-tightening assembly 205 is used to achieve adaptive locking; compared with the traditional passive wedge block, this design provides an angular flexible transition ability while retaining the advantage of response speed, can cope with slight terrain perturbations, and improves the wheel attitude stability in the parked state.

[0131] Traditional angle measurement methods rely on static encoders and are vulnerable to factors such as temperature drift and mechanical looseness, resulting in the deviation of "control instructions - vehicle state". To address this problem, this embodiment introduces a physical information neural network, which integrates an angle dynamic response model and angular velocity micro-drift data to achieve real-time dynamic calibration of the zero point. This method is more robust than the static calibration mechanism and is particularly suitable for accurate measurement requirements in long-term operation or extreme temperature and humidity environments.

[0132] In terms of signal measurement and fusion, this embodiment uses multiple samplings to construct an angular velocity estimation sample sequence. Combining with the random perturbation mechanism in neural network reasoning, a confidence interval and scoring mechanism are introduced to output a quantified credibility label. This mechanism enables the system to not only have the ability to "output values" but also provide a "credible boundary", providing an auxiliary decision-making basis for the vehicle-end controller to judge whether the current state is available and effectively addressing the redundant failure problem caused by multi-source heterogeneous data.

[0133] Compared with the traditional scheme of keeping the continuous motor closed-loop locked, this embodiment realizes "position self-locking" by combining a mechanical wedge block and a high reduction ratio mechanism. After the locking is completed, continuous power supply is not required, and zero power consumption can be maintained to keep the locked posture in states such as system sleep and standby, significantly reducing the power consumption pressure of the steer-by-wire system during the vehicle's standby phase.

[0134] The locking module in this embodiment is decoupled from the steering actuator in terms of structure and is located entirely below the steering arm column, without interfering with the normal steering transmission link. This design facilitates system modular integration and maintenance operations, is conducive to rapid deployment among different vehicle models, and is particularly suitable for commercial platforms and special vehicles with multi-wheel independent steering.

[0135] Facing complex working conditions such as potholed terrain and road surface excitation, this embodiment combines kingpin zero drift modeling and angular velocity uncertainty assessment to construct a state perception subsystem with adaptive recognition and self-correction capabilities. Even under strong disturbances and long-term loads, it can dynamically correct the angle reference to ensure the handling and stability reliability of the chassis system.

[0136] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. A steer-by-wire wheel-end steering angle module with a locking mechanism, characterized in that: including, a steering drive and execution unit (100), comprising a steering motor (101), a steering motor controller (102) integrated with the steering motor (101), a worm and worm gear reduction mechanism (103) and a steering arm (104); the steering motor (101) drives the steering arm (104) to steer around the kingpin axis through the worm and worm gear reduction mechanism (103); the steering arm (104) is connected to the steering column (105) through an upper fork arm (106), a lower fork arm (107) and a shock absorber (108), and the steering column (105) connects the tire (109) and the in-wheel motor (110); a locking unit (200), comprising a locking module (201) arranged below the steering arm column (111), and the locking module (201) comprises a locking execution motor (202), a high reduction ratio planetary gear set (203) and a mechanical wedge locking structure (204); the locking execution motor (202) drives the wedge to insert into or withdraw from the corresponding locking tooth groove through the high reduction ratio planetary gear set (203), and realizes passive locking through a spring preloading assembly (205) when power is off; a sensing unit (300), comprising a steering angle and angular velocity sensor (301) arranged on the steering arm (104) and a locking state sensor (302) arranged on the locking module (201), the steering angle and angular velocity sensor (301) is used for collecting the steering angle and angular velocity, and the locking state sensor (302) is used for detecting whether the locking wedge (204a) is completely inserted into the slot and feeding back the state signal to the upper controller; a power supply unit (400), comprising a high-voltage power supply (401) and a low-voltage power supply (402), the high-voltage power supply (401) supplies power to the in-wheel motor (110), and the low-voltage power supply (402) supplies power to the steering motor controller (102) and the locking motor controller (206).

2. The steer-by-wire wheel-end steering angle module with a locking mechanism according to claim 1, characterized in that: The steering arm column (111) is rigidly connected to the worm and worm gear reduction mechanism (103), and the output end of the worm and worm gear reduction mechanism (103) is rigidly connected to the steering arm (104) to transmit torque.

3. The steer-by-wire wheel-end steering angle module with a locking mechanism according to claim 2, characterized in that: The locking module (201) further comprises: a barrier ring (201a), a locking disc (201b) and a self-locking housing (201c); the self-locking housing (201c) is fixed on the output shaft (203a) of the high reduction ratio planetary gear set (203), and the locking disc (201b) is provided with a plurality of slots matching the mechanical wedge locking structure (204); the barrier ring (201a) is rotatably arranged between the self-locking housing (201c) and the locking disc (201b), and a baffle corresponding to the slot is arranged on the inner side thereof, and the baffle covers the slot opening in the locking state.

4. The steer-by-wire wheel-end steering angle module with a locking mechanism according to claim 3, wherein: The shock absorber (108) is integrated with a suspension air spring (108a), a suspension adjustable damping shock absorber actuator (108b) and a suspension air spring actuator (108c), and the suspension actuator controller (112) dynamically adjusts the suspension stiffness and damping according to the vehicle state.

5. The steer-by-wire wheel-end steering angle module with a locking mechanism according to claim 4, characterized in that: The high-voltage power supply (401) is connected to the in-wheel motor (110) through a high-voltage harness (401a), and the low-voltage power supply (402) is connected to the steering motor controller (102) and the locking motor controller (206) through a low-voltage harness (402a).

6. The steer-by-wire wheel-end steering angle module with a locking mechanism according to claim 5, characterized in that: The mechanical wedge locking structure (204) includes two symmetrically arranged locking wedges (204a), which are installed in the radial guiding chutes between the steering arm column (111) and its base; a spiral groove is provided inside the self-locking housing (201c), and a protrusion matching the spiral groove is provided on the locking wedge (204a), and the wedge (204a) is driven to move linearly along the radial chute by rotating the self-locking housing (201c).

7. The measuring method of the steer-by-wire wheel side steering angle module with a locking mechanism according to any one of claims 1-6, characterized in that: It includes the following specific steps: Step 1: Establish a dynamic model of the wheel corner module in the locked state, collect the original rotation angle and angular velocity signals in the locked state, and perform filtering, noise reduction, and time synchronization processing; Step 2: Based on the data processed in Step 1, construct a physics-informed neural network model, dynamically calculate the current zero position offset of the rotation angle, and correct the original rotation angle signal; Step 3: According to the corrected rotation angle signal and angular velocity signal, construct an angular velocity estimator, output the angular velocity value after zero position correction, and evaluate the credibility of the angular velocity estimation based on the confidence interval.

8. The measuring method of the steer-by-wire wheel-end steering angle module with a locking mechanism according to claim 7, characterized in that: In Step 2, the physics-informed neural network model transforms the angle drift estimation problem into a minimization problem in the function space through the residual function and functional optimization.

9. The measuring method of the steer-by-wire wheel side steering angle module with a locking mechanism as described in claim 8, characterized in that: In Step 3, an angular velocity estimation sample sequence is generated through multiple samplings, the confidence mean and confidence standard deviation are calculated, and a bilateral confidence interval is constructed to quantify the uncertainty of the angular velocity estimation result.

10. The measuring method of the steer-by-wire wheel-end steering angle module with a locking mechanism according to claim 9, characterized in that: In Step 1, the trigger moment of the locked state is determined by the jump of the switch signal of the locked state sensor (302), and the angle change trend and angular velocity micro-drift data are extracted based on a preset time window.

Images