Gearbox noise suppression method and system

Abstract

The present application relates to the technical field of vehicle control, and particularly discloses a gear box noise suppression method and system. The method comprises the following steps: collecting image information of a road surface in front of a vehicle, determining whether there is an obstacle based on the image information of the road surface in front of the vehicle, and obtaining obstacle characteristics in front of the vehicle if there is an obstacle; obtaining a predicted time for the vehicle to pass through the obstacle and a predicted impact energy of the gear box under impact; comparing the predicted impact energy with a set impact threshold value and comparing the predicted time with a set time threshold value to determine whether a trigger condition is met; if the trigger condition is met, controlling the front drive torque to increase to a preset torque; and if the trigger condition is not met, not adjusting the front drive torque. The present application predicts the vehicle driving, adjusts the front drive torque in advance when the gear box may be impacted to generate noise, pre-tightens the gear box to eliminate the tooth side gap, and avoids generating noise.

Description

Technical Field

[0001] This invention relates to the field of vehicle control technology, specifically to a gearbox noise suppression method and system. Background Technology

[0002] To reduce energy consumption and extend driving range, new energy four-wheel drive vehicles generally adopt a "front-wheel drive without load" control strategy in economy mode—that is, the front-wheel drive motor does not output torque and relies solely on the rear-wheel drive system for propulsion. However, when the vehicle drives over obstacles such as potholes and bumps on the road, the front wheels will passively generate non-driven speed fluctuations due to the linkage between the suspension and the wheels. Since the front reducer gears are in a non-meshing state (with tooth backlash), the speed fluctuations will cause the active and passive gears to collide with each other, generating low-frequency impact noise of 100~300Hz, which seriously affects driving comfort and has become one of the core pain points of user complaints.

[0003] Currently, the industry's solutions for gearbox impact noise are mainly divided into four categories, but all of them have insurmountable defects as verified by actual vehicles, as follows: 1. Gear optimization Using high-precision gear grinding, crowning, or reducing tooth backlash has the following drawbacks: high cost, requiring investment in high-precision machining equipment, increasing the manufacturing cost of gears per vehicle by 30-50 yuan; limited effect, only reducing impact noise by 3-5dB, unable to eliminate meshing impact under non-load conditions; poor mass production consistency, with tooth backlash accuracy fluctuating by ±0.02mm in mass production, resulting in unstable noise reduction effect; 2. Soundproofing Wrapping the gearbox housing with high-density sound insulation materials such as butyl rubber pads and sound-absorbing cotton has the following drawbacks: increased energy consumption, with the vehicle weight increasing by 2-5 kg, indirectly leading to a 1-3% increase in energy consumption in economy mode; ineffective low-frequency noise reduction, with only a 2-3 dB reduction in 100-300 Hz low-frequency impact noise, resulting in extremely low cost-effectiveness; and space occupation, as the sound insulation materials require additional installation space, conflicting with the "lightweight and compact" design of new energy vehicles. 3. Continuously apply a small torque Applying a continuous 5-10 Nm of toothed torque to the front-drive gearbox to maintain gear meshing has the following drawbacks: it violates the original intention of the economy mode, the continuous torque causes the motor to consume additional energy, and the driving range in the economy mode is reduced by 3-8%. 4. Vibration feedback control Deploying piezoelectric vibration sensors in the gearbox to detect impacts and adjust torque has the following drawbacks: severe response delay, requiring a "signal acquisition (50ms) - data processing (200ms) - command issuance (150ms)" process, with a total delay of 300-500ms; inability to predict impacts, with control only being executed after the impact has occurred; and unavoidable noise, which can only be "remedied" rather than "suppressed".

[0004] Furthermore, existing vehicle vision systems (such as ADAS cameras) are only used for collision warning of macroscopic obstacles such as pedestrians and vehicles, without involving the quantitative identification of the size of microscopic obstacles on the road surface (potholes, bumps), and there is no correlation control logic between "road obstacle - impact energy of transmission system" - lacking an integrated closed-loop control architecture of "prediction - suppression", which cannot solve the impact noise problem in the economic mode from the root. Summary of the Invention

[0005] The purpose of this application is to address the shortcomings of the aforementioned background technology and provide a gearbox noise suppression method and system.

[0006] The technical solution of this application is: a gearbox noise suppression method, comprising: Collect road surface image information in front of the vehicle, determine whether there is an obstacle based on the road surface image information, and if there is an obstacle, obtain the obstacle features in front of the vehicle. Based on the characteristics of obstacles in front of the vehicle, the current vehicle speed, and preset gearbox parameters, the predicted time for the vehicle to pass through the obstacle and the predicted impact energy that the gearbox will withstand are obtained. The predicted impact energy is compared with the set impact threshold, and the predicted time is compared with the set time threshold. Based on the comparison results, it is determined whether the triggering conditions are met. If the triggering conditions are met, the front-wheel drive torque is increased to the preset torque based on the predicted time until the vehicle passes the obstacle. If the triggering conditions are not met, the front-drive torque will not be adjusted.

[0007] According to a gearbox noise suppression method provided in this application, the method of controlling the front drive torque to increase to a preset torque based on prediction time includes: calculating the minimum preload torque based on the vehicle gearbox parameters, using the minimum preload torque as the preset torque, and increasing the front drive torque to the preset torque at the control starting point when the difference between the prediction time and the set interval is the control starting point.

[0008] According to a gearbox noise suppression method provided in this application, the method for calculating the minimum preload torque based on automotive gearbox parameters includes: calculating the minimum preload torque according to the following formula. in: T min —Minimum preload torque; k —Meshing stiffness of automotive gearboxes; Dd —Gear backlash in automotive gearboxes; r —Car wheel radius; i —The gear ratio of a car's gearbox.

[0009] According to the gearbox noise suppression method provided in this application, the method for obtaining the characteristics of obstacles in front of the vehicle includes: acquiring image information at a set distance in front of the vehicle, performing data processing on the image information, and obtaining the width and depth of the obstacle.

[0010] According to a gearbox noise suppression method provided in this application, the method for determining whether an obstacle exists based on the image information of the road surface ahead includes: performing Gaussian filtering on the acquired image information to remove noise; using adaptive histogram equalization to enhance contrast; fitting the ground plane based on the random sampling consensus algorithm, excluding non-road areas, and extracting the region of interest of the road surface; using a lightweight target detection network model and based on a training dataset constructed from standard scene images to identify obstacles in the region of interest of the road surface; and determining whether an obstacle exists based on the identification results.

[0011] According to a gearbox noise suppression method provided in this application, the method for obtaining predicted impact energy includes: constructing a three-dimensional mapping table between vehicle speed, obstacle width, obstacle depth and predicted impact energy based on a standard test; querying the three-dimensional mapping table based on the acquired real-time vehicle speed, obstacle width and obstacle depth when the vehicle is in motion to obtain the corresponding predicted impact energy.

[0012] According to the gearbox noise suppression method provided in this application, the method for determining whether the triggering condition is met based on the comparison result includes: if the predicted impact energy is greater than the set impact threshold and the prediction time is not less than the set time threshold, it is determined that the triggering condition is met; otherwise, the triggering condition is not met.

[0013] According to a gearbox noise suppression method provided in this application, the front-drive torque is reduced to 0 after a set delay time following the vehicle's passage over an obstacle.

[0014] According to the gearbox noise suppression method provided in this application, if the front drive torque of the vehicle increases to a preset torque and the time for maintaining the preset torque exceeds a set maintenance time, it is determined that a fault has occurred and the front drive torque is reduced to 0.

[0015] This application also relates to a gearbox noise suppression system, which operates according to the gearbox noise suppression method described above, including: The data acquisition module is used to acquire image information of the road surface in front of the vehicle and the current vehicle speed; The recognition module determines whether an obstacle exists based on the image information of the road surface ahead; The feature extraction module is used to obtain the features of the obstacle in front of the vehicle based on road image information when there is an obstacle in front of the vehicle; The prediction data calculation module obtains the predicted time for the vehicle to pass through the obstacle and the predicted impact energy of the gearbox based on the characteristics of the obstacle in front of the vehicle, the current vehicle speed, and preset gearbox parameters. The trigger judgment module is used to compare the predicted impact energy with a set impact threshold and the predicted time with a set time threshold, and to determine whether the triggering conditions are met based on the comparison results. An adjustment module is used to control the front drive torque to increase to a preset torque based on a predicted time when the triggering condition is met, until the vehicle passes the obstacle, and to not adjust the front drive torque when the triggering condition is not met.

[0016] The advantages of this application are: 1. This application obtains road surface image information in front of the vehicle, judges and identifies obstacles in front of the vehicle, predicts the possible impact on the gearbox during vehicle driving, adjusts the front drive torque in advance, and eliminates tooth backlash by pre-tightening the gearbox teeth, which not only suppresses noise, but also maximizes the range advantage of the economic mode. The whole suppression method is simple and the suppression control process is fully automated, which improves the experience of drivers and passengers. 2. This application calculates the minimum preload torque using automotive gearbox parameters, and determines the target torque based on the minimum preload torque. On the one hand, this can maximize the energy savings of front-wheel drive torque enhancement, and on the other hand, it can meet the purpose of suppressing noise by preloading the gearbox. 3. The method for calculating the minimum preload torque in this application is simple and can be calculated based on the selected gearbox structure parameters in the early stages of vehicle design. It can then be stored in the vehicle's overall controller for easy retrieval. 4. This application obtains the width and depth of the obstacle ahead through image information. The width and depth are proportional to the impact energy of the gearbox when the vehicle passes through the obstacle. By obtaining the width and depth, the predicted impact energy when the vehicle passes through the obstacle can be predicted. The extraction method is simple and greatly simplifies the calculation of the predicted impact energy. 5. The method of this application for determining whether there is an obstacle in front of the vehicle is very simple. By processing the collected image information and comparing the processed image information with the training dataset, it is possible to determine whether there is a certain obstacle in the image information as shown in the training dataset. The method of judgment and recognition is simple and has a high accuracy. 6. This application uses a three-dimensional mapping table that calibrates vehicle speed, obstacle width, obstacle depth and predicted impact energy, and stores the three-dimensional mapping table in the vehicle's whole vehicle controller. When needed, the required predicted impact energy can be quickly obtained by directly calling the table. The calculation method is simple, greatly reducing the computing power requirement and the calculation results are accurate. 7. The method for determining whether the triggering conditions are met in this application is very simple. The triggering conditions are only determined when both the predicted impact energy and the prediction time meet the relevant conditions, which greatly reduces the possibility of misjudgment and makes the control of the entire front-drive torque more accurate. 8. After the vehicle passes an obstacle, this application will maintain the front-drive torque for a period of time and then quickly reduce it to 0, avoiding the waste of energy caused by maintaining the front-drive torque for a long time, improving the energy utilization efficiency of the vehicle, and saving energy. 9. The purpose of maintaining the front drive torque in this application is to suppress noise by passing obstacles. The maintenance of the front drive torque is not a normal state. Once the front drive torque is maintained for a long time, it proves that a fault has occurred. Therefore, the front drive torque should be reduced quickly to avoid wasting energy. 10. This application also relates to an automotive gearbox noise suppression system based on road obstacle recognition. The suppression system of this application can be integrated into the vehicle controller and automatically adjust the front drive torque of the vehicle. By collecting image information of the road ahead, it predicts whether the vehicle will pass through an obstacle. If there is an obstacle that may cause gearbox noise, it adjusts the front drive torque in advance and pre-tightens the gearbox teeth to eliminate possible noise and improve the driving experience.

[0017] This application predicts vehicle movement using road surface image information in front of the vehicle. In cases where obstacles on the road ahead may cause excessive impact on the vehicle's gearbox and generate noise, the front-drive torque is adjusted in advance, and the gearbox is pre-tightened to eliminate backlash, thus avoiding noise. This improves the driving experience and saves energy consumption to the greatest extent, and has great potential for widespread application. Attached Figure Description

[0018] Figure 1 : A schematic diagram of the automotive gearbox noise suppression method of this application. Detailed Implementation

[0019] The embodiments of this application are described in detail below, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0020] In the description of this application, it should be understood that the terms "longitudinal", "lateral", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application.

[0021] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0022] The present application will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0023] This application relates to a gearbox noise suppression method, primarily for use in new energy four-wheel drive vehicles. In economy mode, these vehicles employ a front-wheel drive control strategy with no load (i.e., front-wheel drive torque is 0), relying solely on rear-wheel drive for propulsion. This application uses a visual module to analyze the road surface ahead, detecting any obstacles and predicting the vehicle's driving state. When it predicts an obstacle and the potential impact on the gearbox, the front-wheel drive torque is adjusted in advance. This pre-tensioning torque eliminates gear backlash, preventing noise from the gearbox upon impact. This control strategy, implemented in economy mode, ensures low energy consumption while preventing gearbox noise during impacts, improving both vehicle range and passenger experience.

[0024] Specifically, such as Figure 1 As shown, a gearbox noise suppression method according to this application includes the following steps: S1. Collect road surface image information in front of the vehicle, determine whether there is an obstacle based on the road surface image information, and if there is an obstacle, obtain the obstacle features in front of the vehicle. This application involves collecting road surface image information in front of the vehicle, and therefore is equipped with a corresponding vision module, primarily a binocular camera. In practical applications, this device structure is not limited to this; devices such as LiDAR can be used to acquire road surface information in front of the vehicle and determine the presence of obstacles. Obstacles in this application refer to raised or recessed structures on the road surface. These can be defined as parts exceeding or falling below the road surface height by 0.05m. When the vehicle passes over these obstacles, it will experience bumps, and the gearbox will be impacted. If the vehicle is in economy mode at this time, i.e., the front-wheel drive torque is 0, the gears in the gearbox are not fully engaged, resulting in backlash. This impact will generate noise. This application anticipates this potential noise situation and then develops corresponding strategies to avoid noise affecting the driving experience. This application assumes that the current vehicle is in the front-wheel drive unloaded economy mode, i.e., the front-wheel drive torque is 0. S2. Based on the characteristics of obstacles in front of the vehicle, the current vehicle speed, and preset gearbox parameters, obtain the predicted time for the vehicle to pass through the obstacle and the predicted impact energy that the gearbox will withstand. This application makes further judgments by calculating the predicted time for the vehicle to pass through an obstacle and the predicted impact energy that the gearbox will experience. Obtaining the predicted time for the vehicle to pass through an obstacle is to determine when the gearbox may emit noise. Control can be made in advance based on this predicted time to avoid energy waste due to control being too early and to avoid the gearbox emitting noise after the vehicle has passed through the obstacle due to control being too late. Calculating the predicted impact energy is to determine whether this impact will cause the gearbox to emit noise. In some cases, the predicted impact energy is too small, and the gearbox will not emit noise, or the noise emitted will not be felt by the driver and passengers. Therefore, no adjustment to the control is required. Calculating the predicted impact energy is to eliminate this situation. S3. Compare the predicted impact energy with the set impact threshold and the predicted time with the set time threshold, and determine whether the triggering conditions are met based on the comparison results. In essence, it means predicting that the impact energy is greater than the set impact threshold and the prediction time is not less than the set time threshold. The setting that the predicted impact energy is greater than the set impact threshold is easy to understand. As mentioned above, by setting the impact threshold, impact situations that will not cause noise can be eliminated, avoiding excessive control that would lead to energy waste and complicated operation. The prediction time is greater than the set time threshold because there needs to be enough time for adjustment. If the prediction time is less than the set time threshold, it means that the time it takes for the vehicle to reach the obstacle is too short, and there is not enough time to adjust the front drive, so there is no need to trigger the control of the front drive. If the triggering conditions are met, the front-wheel drive torque is increased to the preset torque based on the predicted time until the vehicle passes the obstacle. If the triggering conditions are not met, the front-drive torque will not be adjusted.

[0025] In some embodiments of this application, step S1 described above has been optimized. Specifically, this embodiment uses a vision module to acquire image information at a set distance in front of the vehicle. The hardware combination used is a binocular camera and an AI processor, which supports reuse with existing ADAS systems. The hardware parameters and selection criteria are shown in the table below: Table 1: Key Hardware Parameters The system acquires road surface image information at a set distance in front of the vehicle using a binocular camera (the set distance in this embodiment is 20m, but it is not limited to this value in actual applications and can be set according to needs). This results in a time-series image of the set road surface in front. Noise is removed by 3×3 Gaussian filtering, and adaptive histogram equalization (CLAHE) is used to enhance the contrast between the road surface and obstacles. The system fits the ground plane based on the random sampling consistency (RANSAC) algorithm, excludes non-road areas such as the sky and roadside green belts, and extracts the region of interest (ROI) of the road surface.

[0026] Then, obstacle feature recognition and contour extraction were performed: a lightweight object detection network YOLOv5-Lite (model size ≤ 5 MB, inference speed ≤ 100 ms / frame) was used to identify obstacles in the ROI region; a training dataset was constructed, containing 100,000 images of different scenes (covering dry / rainy / nighttime lighting, asphalt / cement / gravel road surfaces, potholes with depths of 5~20cm / protrusions with heights of 3~15cm / widths of 0.1~3m), labeled with three categories: "potholes", "protrusions", and "smooth road surfaces", with a label accuracy of ≥ 98%; the Canny edge detection algorithm (with adaptive threshold adjustment) was used to extract obstacle contours, with a contour extraction accuracy of ≤ 1 pixel, ensuring that subsequent size calculation errors are controllable.

[0027] By extracting obstacle features as described above, it's possible to determine whether an obstacle exists at a set distance in front of the vehicle. Specifically, if the acquired image information contains obstacle features corresponding to the training dataset, and the image information matches the training dataset, then an obstacle from the training dataset is present in front of the vehicle. If there's no match, then the obstacle from the training dataset is not present. The comparison method is essentially image-by-image analysis. If the similarity exceeds a set similarity threshold (e.g., 80%, or another value), a match is considered successful; otherwise, no match is found. Image data comparison is a qualitative analysis; as long as the real-time acquired image matches a standard image in the training dataset, an obstacle is determined to exist, and further comparisons are unnecessary.

[0028] After the comparison is complete, assuming there is an obstacle at a set distance in front of the vehicle, further identification and judgment of the obstacle is required. The width and depth of the obstacle need to be obtained, as these are indicators directly related to the predicted impact energy; obtaining the width and depth of the obstacle allows for the calculation of the predicted impact energy.

[0029] Specifically, the method for calculating the obstacle width in this embodiment is as follows: Calculate the horizontal distance between the center point of the obstacle and the vehicle. S Based on the principle of binocular parallax, the horizontal distance between the center point of the obstacle and the vehicle is calculated using the following formula. S , in: S — The horizontal distance between the center of the obstacle and the vehicle, in meters; f —Focal length of the binocular camera, in meters; B —Binocular camera baseline distance, m; Δd — Parallax of the center point of the obstacle in the left and right camera images of a binocular camera, m; Then calculate the horizontal pixel equivalent using the following formula. k w : in: k w —Horizontal pixel equivalent; S — The horizontal distance between the center of the obstacle and the vehicle, in meters; i h —This refers to the horizontal field of view of the camera, typically 60°~80°, determined by the camera selection; β h —Image horizontal pixels, such as 1920px; Δd — Parallax of the center point of the obstacle in the left and right camera images of a binocular camera, m.

[0030] Horizontal pixel equivalent k w The coefficient value is calculated according to the above formula, that is, it is calculated according to the above parameter value; Obtain horizontal pixel equivalent k w Then, the width W of the obstacle can be calculated using the following formula: in: W—Barrier width, in meters; k w —Horizontal pixel equivalent; x 2 -x 1 —Barrier pixel width.

[0031] The method for calculating obstacle depth in this embodiment is as follows: Obstacle depth is the vertical dimension of the obstacle; a protrusion is a positive value, and a depression is a negative value. Extract the bottom pixels of the obstacle y bottom road surface reference pixels y road Vertical pixel difference d pix =y bottom -y road (The road surface reference is determined by the ground plane fitted by RANSA); Calculate the vertical pixel equivalent using the following formula. k d : in: k d —Vertical pixel equivalent; H —Installation height of the binocular camera, in meters; β v —Image vertical pixels, such as 1080px; α —Top-down view of the binocular camera, °; Vertical pixel equivalent k d The coefficients are calculated according to the above formula and obtained based on the above parameter values; Obtain vertical pixel equivalent k d Then, the width D of the obstacle can be calculated using the following formula: in: D —Barrier depth, m; k d —Vertical pixel equivalent; d pix — Vertical pixel difference.

[0032] After obtaining the width and depth of the obstacle, size verification is performed. A sliding window filter (window size of 3 frames) is used to smooth the width and depth values ​​identified in three consecutive frames. If the filtered error is less than a set error (10% in this embodiment, but not limited to this value in actual applications), the calculated width and depth values ​​are considered reliable. If the identification result of a certain frame deviates from the set deviation of the two adjacent frames (15% in this embodiment, but not limited to this value in actual applications), it is judged as an outlier and discarded. Linear interpolation is used to supplement the outlier to ensure output stability.

[0033] In other embodiments of this application, step S2 described above has been optimized. Specifically, this embodiment uses the obstacle width W and depth D output by the vision module, combined with the vehicle status and speed V obtained from the CAN bus and preset gearbox parameters (obtained based on the selected gearbox design parameters), to calculate the gearbox impact energy within 50 ms using an "offline calibration of a three-dimensional mapping table + online table lookup" method. E impact This provides a basis for control decisions (avoiding the high computational cost of solving dynamic equations in real time).

[0034] The parameters involved are shown in the table below: Table 2: Key Parameter Definition Table Step 1: Establish a physical model: i. Wheel Vertical Excitation Model: Simplifying road obstacles into "height" D ,width W The convex function of the wheel at the vehicle speed V When the vehicle passes over an obstacle, the excitation is converted into a vertical displacement excitation at the center of the wheel. Z(t) , Z(t) The steepness is due to D / W Decide( D / W The larger the value, the steeper the excitation and the more significant the speed fluctuation.

[0035] ii. Suspension-Wheel System Dynamics Model: The vehicle is simplified to a "half-suspension model" (including sprung mass, unsprung mass, and spring stiffness). K s Shock absorber damping C s and tire stiffness K t ),enter Z(t) Then, the vertical acceleration of the wheel is solved using the dynamic equations. az(t) and instantaneous speed fluctuation Give(t) ( Give(t) (The core influencing factor of impact energy).

[0036] iii. Impact energy model of transmission system: speed fluctuation Give(t) The signal is transmitted to the input end of the reducer via the half-shaft, due to tooth backlash. Dd When the driving gear idles, it strikes the driven gear, and the impact energy... E impact ≈(1 / 2)×J×(Dω) 2 ( J The moment of inertia of the transmission system equivalent to that of the driving gear is given. Give (This refers to the peak value of the speed fluctuation).

[0037] Step 2: Offline calibration of the 3D mapping table i. Calibration tools: ADAMS multibody dynamics software was used to simulate and calibrate the vehicle suspension and transmission system, and the results were verified by real vehicle testing (10 new energy four-wheel drive models, covering SUVs and sedans). ii. Calibration scenario: Covering vehicle speed V =0~80km / h (step length 5km / h), obstacle width W =0.1~3m (step length 0.1m), obstacle depth D =0.01~0.2m (protrusion, step size 0.01m) / D =~0.01~~0.2m (dimple, step length ~0.01m), total 17 ( V )×30 ( W )×20 ( D ) = 10200 sets of data; iii. Mapping table generation: For each group of data... E impact The values ​​(simulation calculations + real vehicle testing corrections) are organized into a 3D lookup table model. E impact =f(V,D,W) It is stored in the Flash memory of the vehicle controller (VCU), and the table lookup response time is ≤50ms.

[0038] Step 3: Online Real-Time Calculation i. Data input: The vision module outputs data once every 100ms. W、D The CAN bus synchronously uploads the vehicle speed every 100ms. V ; ii. Table lookup calculation: After receiving W, V, and D, the VCU calls the three-dimensional mapping table and calculates the current value through linear interpolation (if the parameters are not on the calibration node). E impact The calculation error is ≤5%; iii. Anomaly Handling: If W or D recognition fails (e.g., due to occlusion causing inability to recognize), then the default setting will be used. E impact=0 (do not perform torque adjustment) to avoid accidental triggering.

[0039] In a further embodiment of this application, step S3 described above has been optimized. Specifically, if the impact energy is predicted... E impact Greater than the set impact threshold E threshold If the prediction time is not less than the set time threshold (the set time threshold in this embodiment is 2s, but in actual applications it is not limited to the above number), the trigger condition is determined to be met; otherwise, the trigger condition is not met.

[0040] If the triggering conditions are met, the front-wheel drive torque is increased to a preset torque based on the predicted time, until the vehicle passes the obstacle. The minimum preload torque is calculated based on the vehicle's gearbox parameters and used as the preset torque. The minimum preload torque is calculated using the following formula. in: T min —Minimum preload torque, Nm; k —Meshing stiffness of automotive gearbox, N / mm; Dd —Gear backlash of an automotive gearbox, in mm; r —Car wheel radius, meters; i —The gear ratio of a car's gearbox.

[0041] The minimum preload torque is a fixed value that corresponds to the design parameters of the automotive gearbox.

[0042] When it is necessary to increase the front-drive torque, the adjustment time of the front-drive torque is obtained based on the prediction time. In this embodiment, the adjustment of the front-drive torque is completed 1.0s before the prediction time (in actual applications, other times can also be selected). This can avoid energy waste caused by adjusting the front-drive torque too early, and can also avoid gearbox noise caused by the inability to adjust the front-drive torque to the preset torque when the vehicle reaches an obstacle.

[0043] After the vehicle passes an obstacle, the front-drive torque is reduced to 0 after a set delay time. In this embodiment, the set delay time is 1.0s, but it is not limited to this value. That is, the front-drive torque is only adjusted back after the obstacle has been completely passed, to avoid adjusting the front-drive torque before the obstacle has been completely passed. If the vehicle's front-drive torque increases to a preset torque and maintains that preset torque for a longer than a set maintenance time (in this embodiment, the set maintenance time is 10s, but it is not limited to this value), then a fault is determined to have occurred, and the front-drive torque is reduced to 0.

[0044] The above-described impact threshold setting in this embodiment E threshold Yes, it's designed to obtain the testers' subjective feelings through a subjective evaluation test method. The subjective evaluation test method selects 50 subjects (covering the age range of 20-60 years and 1-20 years of driving experience), and tests different impact energy values ​​in a semi-anechoic chamber simulating the in-vehicle environment (background noise ≤35dB). E s Noise level from the lower gearbox (10 points, 1 point = severe complaint, 10 points = no noticeable noise). Upon impact, a piezoelectric vibration sensor (range ±100g, accuracy ±0.01g) is used to measure the gearbox impact acceleration, and then the impact energy value is calculated based on the impact acceleration. E s The impact energy value corresponding to an average subject score of 7 (other values ​​can be selected in practical applications) is calculated. E s The set impact threshold for this embodiment E threshold .

[0045] The specific control method is as follows: the prediction time is calculated based on the obstacle horizontal distance output by the vision module. S and current vehicle speed V The obtained result is the predicted time, which is the time it takes for the vehicle to reach the obstacle. t arrive =S / V ; t arrive ≥2.0s, reserving control response time to avoid time deviation caused by vehicle speed fluctuations; if t arrive If the time is less than 2.0 seconds, it is determined as "unable to control in time" and torque adjustment is not performed; at the same time, it meets the following conditions: E impact >E threshold and t arrive ≥2.0s, the control decision subsystem sends a "torque preload command" (including target torque value and execution time) to the electric drive controller; E impact ≤ E threshold (Noise level acceptable) t arrive If the obstacle disappears within 2.0 seconds (unable to be controlled in time) or the vision module detects that the obstacle has passed (e.g., the vehicle has passed), send a "restore economy mode command"; based on gearbox meshing stiffness. k and tooth backlash Dd Calculate the minimum preload torque T min =(k×Δδ×r) / i (likek =8×10 5 N / m, Dd =0.1mm, r =0.38m, i =10, Tmin ≈3.04Nm), considering redundancy, the setting is 5~10Nm (fine-tuned according to the vehicle's gearbox parameters); after receiving the "torque preload command", the electric drive controller... t arrive The torque adjustment is initiated in the first 1.0s, using a PI control algorithm to smoothly increase the front-wheel drive torque from 0Nm to the preset torque (overshoot ≤5%, adjustment time ≤0.2s). After the vehicle passes an obstacle (the vision module detects the obstacle disappearing), there is a 1.0s delay (to ensure the wheels are completely out of the obstacle area and avoid secondary impact) before the front-wheel drive torque is reduced to 0Nm, restoring the "front-wheel drive without load" economy mode. If the torque adjustment timeout (>1.0s without reaching the target value), a fault alarm is triggered (a fault code is sent via the CAN bus), and the current torque is maintained for 10s before resuming the economy mode to avoid gearbox overload.

[0046] Conduct real-world scenario tests: 1. Implementation Scenarios and Basic Parameters Test vehicle: A mass-produced new energy four-wheel drive SUV, with a total vehicle weight of 2000kg and a front-wheel drive axle load. m 1 =1100kg (55%), wheel radius r =0.38m, gearbox transmission ratio i =10.55, meshing stiffness k =8×10 5 N / m, input shaft moment of inertia J =0.05kg･m 2 Gear backlash Dd =0.1mm; Vision module: Binocular camera (baseline distance) B =0.4m, focal length f =8mm, horizontal field of view i h =60°, vertical installation height H =1.5m, pitch angle α =45°, sampling frame rate 30fps), AI processor is NVIDIA Orin-Nano; Control parameters: E threshold =10J The torque adjustment response time is 0.2s, the recovery delay is 1.0s, and the target tooth preload torque is 5Nm.

[0047] 2. Implementation Steps Step 1: The vision module identifies the obstacle size. Image acquisition: Vehicles with V =Driving at 10km / h (2.78m / s), the binocular camera is in S Road surface images were acquired at a distance of 30m, with a frame synchronization error of ≤0.5ms; Preprocessing: Noise was removed by 3×3 Gaussian filtering, the CLAHE algorithm improved the contrast between the road surface and potholes by 35%, and the RANSAC algorithm extracted the ROI (accounting for 88%) to eliminate interference from the sky and roadside green belts. Recognition and contour extraction: The YOLOv5-Lite network identifies pits within the ROI region, and Canny edge detection is used to obtain... w pix =35px, d pix =25px (the vertical pixel difference between the bottom of the pit and the road surface reference). Size conversion: a) Horizontal distance S calculation: Calculated based on the above formula, where... f =8*10 -3 m, B =0.4m, Δd =0.092px (calculated using the SGBM algorithm), substituting it in gives... S ≈30m; b) Width W calculation: Calculate the horizontal pixel equivalent. k w ,in i h =60°, β h =1920px, the actual width is 0.018m / px according to the above calculation formula, and the width of the pit is 0.54m (the actual width of the pit is 0.5m, and the recognition error is 8%≤10%). c) Depth D calculation: Calculate the vertical pixel equivalent, where α =45°, H =1.5m, β v =1080px, substituting into the above calculation formula, the vertical loudness equivalent is calculated to be 0.00196m / px; the actual depth is 0.049m (the actual depth of the pit is 0.05m, and the recognition error is 2%≤10%).

[0048] Size verification: After smoothing by sliding window filtering (3 frames), W=0.52m, D=0.05m, and the error was further reduced to 4%.

[0049] Step 2: Impact Energy Calculation Subsystem Calculation E impactData input: W=0.52m, D=0.05m, V=10km / h; Table lookup calculation: call the three-dimensional mapping table. E impact =f(V=60km / h,D=0.05m,W=0.52m) Since the parameters are not on the calibration nodes, linear interpolation is used to calculate the results. E impact =11.4J(>E threshold =10J） ; Predicted time: t arrive =10.8s (≥2.0s, meets control window requirements).

[0050] Step 3: The control decision and execution subsystem performs torque control. Decision trigger: E impact =11.4J>10J and t arrive =10.8s≥2.0s, send "torque preload command" (preset torque 5Nm, execution time 10.8s-1.0s=9.8s). Torque adjustment: The electric drive controller starts PI control at 9.8s, and increases the front drive torque from 0Nm to 5Nm within 0.2s (overshoot 3%, no oscillation). Mode recovery: When the vehicle passes the obstacle at time 10.8s, the vision module detects that the obstacle has disappeared, and after a delay of 1.0s (i.e. 11.8s), the front-wheel drive torque is restored to 0Nm, and the economy mode is restored.

[0051] 3. Verification of test results Impact acceleration: decreased from 9.07g before implementation to 6.20g, a reduction of 32%, which is an improvement of approximately dB; In-vehicle noise: No complaints in subjective rating; Energy consumption verification: During the entire control process (from 9.8s to 11.8s, a total of 2s), the additional energy consumption of the front drive motor is approximately 5Nm × 7.31rad / s × 2s ≈ 73.1J (equivalent to a range loss of approximately 0.002km), which can be ignored.

[0052] In addition, this application also relates to a gearbox noise suppression system, including a data acquisition module, an identification module, a feature extraction module, a prediction data calculation module, a trigger judgment module, and an adjustment module. The data acquisition module is used to acquire road surface image information in front of the vehicle and the current vehicle speed; it determines whether there is an obstacle based on the road surface image information; the feature extraction module is used to obtain the obstacle features in front of the vehicle based on the road surface image information when there is an obstacle in front; the prediction data calculation module obtains the predicted time for the vehicle to pass through the obstacle and the predicted impact energy of the gearbox based on the obstacle features in front of the vehicle, the current vehicle speed, and preset gearbox parameters; the trigger judgment module is used to compare the predicted impact energy with a set impact threshold and the predicted time with a set time threshold, and determine whether the trigger condition is met based on the comparison result; the adjustment module is used to control the front drive torque to increase to a preset torque based on the predicted time when the trigger condition is met, until the vehicle passes through the obstacle, and to not adjust the front drive torque when the trigger condition is not met.

[0053] The foregoing has shown and described the basic principles, main features, and advantages of this application. Those skilled in the art should understand that this application is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of this application. Various changes and modifications can be made to this application without departing from the spirit and scope thereof, and all such changes and modifications fall within the scope of this application as claimed. The scope of protection of this application is defined by the appended claims and their equivalents.

Claims

1. A method for suppressing gearbox noise, characterized in that: include: Collect road surface image information in front of the vehicle, determine whether there is an obstacle based on the road surface image information, and if there is an obstacle, obtain the obstacle features in front of the vehicle. Based on the characteristics of obstacles in front of the vehicle, the current vehicle speed, and preset gearbox parameters, the predicted time for the vehicle to pass through the obstacle and the predicted impact energy that the gearbox will withstand are obtained. The predicted impact energy is compared with the set impact threshold, and the predicted time is compared with the set time threshold. Based on the comparison results, it is determined whether the triggering conditions are met. If the triggering conditions are met, the front-wheel drive torque is increased to the preset torque based on the predicted time until the vehicle passes the obstacle. If the triggering conditions are not met, the front-drive torque will not be adjusted. The method for controlling the front drive torque to increase to a preset torque based on predicted time includes: calculating the minimum preload torque based on the vehicle gearbox parameters, using the minimum preload torque as the preset torque, and increasing the front drive torque to the preset torque at the control starting point when the difference between the predicted time and the set interval is the control starting point. The method for obtaining the predicted impact energy includes: constructing a three-dimensional mapping table between vehicle speed, obstacle width, obstacle depth and predicted impact energy based on a standard test; querying the three-dimensional mapping table based on the real-time vehicle speed, obstacle width and obstacle depth obtained when the vehicle is in motion to obtain the corresponding predicted impact energy. The method for determining whether the triggering condition is met based on the comparison result includes: if the predicted impact energy is greater than the set impact threshold and the prediction time is not less than the set time threshold, the triggering condition is met; otherwise, the triggering condition is not met.

2. The gearbox noise suppression method as described in claim 1, characterized in that: The method for calculating the minimum preload torque based on automotive gearbox parameters includes: calculating the minimum preload torque according to the following formula. in: T min —Minimum preload torque; k —Meshing stiffness of automotive gearboxes; Δδ —Gear backlash in automotive gearboxes; r —Car wheel radius; i —The gear ratio of a car's gearbox.

3. The gearbox noise suppression method as described in claim 1, characterized in that: The method for obtaining the features of obstacles in front of the vehicle includes: acquiring image information at a set distance in front of the vehicle, processing the image information, and obtaining the width and depth of the obstacle.

4. The gearbox noise suppression method as described in claim 3, characterized in that: The method for determining whether an obstacle exists based on the image information of the road surface ahead includes: performing Gaussian filtering on the acquired image information to remove noise; using adaptive histogram equalization to enhance contrast; fitting the ground plane based on the random sampling consensus algorithm, excluding non-road areas, and extracting the region of interest of the road surface; using a lightweight target detection network model and a training dataset constructed from standard scene images to identify obstacles in the region of interest of the road surface; and determining whether an obstacle exists based on the identification results.

5. The gearbox noise suppression method as described in claim 1, characterized in that: A set delay time is set after the vehicle passes an obstacle to reduce the front-wheel drive torque to 0.

6. The gearbox noise suppression method as described in claim 1, characterized in that: If the front-wheel drive torque of the vehicle increases to the preset torque and maintains the preset torque for a longer period than the set maintenance time, a fault is determined to have occurred, and the front-wheel drive torque is reduced to 0.

7. A gearbox noise suppression system, characterized in that: The suppression system operates according to a gearbox noise suppression method as described in any one of claims 1 to 6, including: The data acquisition module is used to acquire image information of the road surface in front of the vehicle and the current vehicle speed; The recognition module determines whether an obstacle exists based on the image information of the road surface ahead; The feature extraction module is used to obtain the features of the obstacle in front of the vehicle based on road image information when there is an obstacle in front of the vehicle; The prediction data calculation module obtains the predicted time for the vehicle to pass through the obstacle and the predicted impact energy of the gearbox based on the characteristics of the obstacle in front of the vehicle, the current vehicle speed, and preset gearbox parameters. The trigger judgment module is used to compare the predicted impact energy with a set impact threshold and the predicted time with a set time threshold, and to determine whether the triggering conditions are met based on the comparison results. An adjustment module is used to control the front drive torque to increase to a preset torque based on a predicted time when the triggering condition is met, until the vehicle passes the obstacle, and to not adjust the front drive torque when the triggering condition is not met.

Images