Tire design method for improving noise in vehicle by attaching silence cotton to longitudinal groove

Abstract

The invention relates to the technical field of tire design data processing, in particular to a tire design method for improving in-vehicle noise by attaching silence cotton to a longitudinal groove, which comprises the following steps: acquiring data through a vehicle-mounted sensing system, inputting a data set into an acoustic feature extraction model to identify a mapping relation between longitudinal groove parameters and noise resonance, the longitudinal groove geometric parameters are used for executing acoustic topological optimization to generate a resonance risk map; a resonance risk map is used as an optimization target, a material acoustic database is introduced to pre-calculate the sound attenuation effect after the mute cotton is attached, and the parameter feasibility is verified through finite element acoustic coupling simulation; driving digital design of the tire mold according to the optimized parameters, and attaching mute cotton to the inner wall of the longitudinal groove in a gradient density mode according to sound energy distribution; a simulation system is constructed on a digital twin platform to simulate a driving working condition, 1000Hz sound pressure level attenuation data is output and compared with a threshold value, the data is fed back to a data acquisition stage to form an adaptive optimization cycle, and resonance noise caused by fixed longitudinal groove parameters is effectively inhibited.

Description

Technical Field

[0001] This invention relates to the field of tire design data processing technology, and in particular to a tire design method that uses sound-absorbing cotton attached to the longitudinal grooves to improve in-vehicle noise. Background Technology

[0002] Existing tire longitudinal groove design technologies suffer from the following technical challenges: Firstly, the use of fixed longitudinal groove parameters, such as uniform groove width and number, makes the groove structure prone to acoustic resonance during tire rolling, particularly concentrated in the 1000Hz frequency band, leading to deterioration of in-vehicle noise. For instance, in high-speed passenger vehicle scenarios, fixed longitudinal grooves cannot effectively disperse vibration energy; sound waves are superimposed and reflected within the enclosed cabin, significantly reducing ride comfort. The root cause lies in the fact that fixed longitudinal grooves limit the flexibility of acoustic impedance matching, making noise control reliant on a single mold solution, which is difficult to adapt to diverse vehicle models or road conditions. This invention directly suppresses the resonance frequency band by optimizing longitudinal groove variables and attaching sound-absorbing cotton. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention provides a tire design method that uses sound-absorbing cotton attached to the longitudinal grooves to improve in-vehicle noise, thus solving the technical problem that the fixed groove width and number in existing tire longitudinal groove designs cause resonance deterioration of in-vehicle noise at a frequency of 1000Hz.

[0004] To solve the above-mentioned technical problems, the specific contents of the present invention are as follows: This invention provides a tire design method for improving in-vehicle noise by attaching sound-absorbing cotton to the longitudinal grooves, comprising: Step 1: Collect the geometric parameters of the tire longitudinal grooves and the in-vehicle noise data under driving conditions through the vehicle-mounted sensing system, and construct a noise and structure correlation dataset with the 1000Hz frequency band as the core. Step 2: Input the noise and structure association dataset into the acoustic feature extraction model to identify the mapping relationship between longitudinal groove parameters and noise resonance, and use the longitudinal groove geometric parameters in the dataset to perform acoustic topology optimization to generate a resonance risk map that marks the area of ​​concentrated acoustic energy. Step 3: Using the resonance risk map as the optimization target, a reinforcement learning algorithm is used to dynamically iterate the combination scheme of longitudinal groove width and number. During the iteration process, the acoustic database of materials is introduced to pre-calculate the sound attenuation effect after the sound-absorbing cotton is attached, and the feasibility of the parameters is verified by finite element acoustic coupling simulation, and the longitudinal groove optimization parameters are output. Step 4: Based on the longitudinal groove optimization parameters, drive the digital design of the tire mold, and attach porous composite sound-absorbing cotton to the inner wall of the longitudinal groove in a gradient density manner according to the sound energy distribution indicated by the resonance risk spectrum. Step 5: Construct a simulation system in the digital twin platform, including a tire model, a vehicle model, and an environment model after the sound-absorbing foam is applied, to simulate driving conditions and output 1000Hz sound pressure level attenuation data; compare the 1000Hz sound pressure level attenuation data with a preset threshold, and feed the comparison result back to the data acquisition stage to guide supplementary data acquisition, and start the adaptive optimization loop.

[0005] Furthermore, in the tire design method of the present invention, which utilizes longitudinal grooves to attach sound-absorbing cotton to improve in-vehicle noise, step 1 includes: A dynamic sequence of the width, depth, and distribution angle of the tire longitudinal grooves was obtained using a laser scanner. At the same time, noise signals are collected by an acoustic sensor array placed inside the carriage, and time-frequency data of sound pressure level around 1000Hz are extracted; Real-time load and speed information is read from the vehicle bus and timestamped with the road features output by the road texture recognition system. Wavelet denoising and principal component analysis are performed on the synchronized data to generate a dataset that correlates noise and structure.

[0006] Furthermore, in the tire design method of the present invention that utilizes longitudinal grooves to attach sound-absorbing cotton to improve vehicle interior noise, step 2 includes: A pre-trained convolutional neural network is used to process the dataset related to noise and structure. The weights of the 1000Hz frequency band features are strengthened through an attention mechanism, and the resonance sensitivity feature vector is output. Simultaneously, by utilizing the longitudinal groove geometric parameters in the noise-structure association dataset, the variable density method is used to solve the longitudinal groove layout topology optimization problem with the objective of minimizing acoustic energy concentration, and the topology optimization results are generated. The resonance sensitivity feature vector is compared with the topology optimization results, and regions with overlap exceeding a preset threshold are marked. A resonance risk map with weighted coefficients is generated based on the labeling results.

[0007] Furthermore, in the tire design method of the present invention that utilizes longitudinal grooves to attach sound-absorbing cotton to improve vehicle interior noise, step 3 includes: Define a reinforcement learning model with the width and number of longitudinal grooves as the state space and the parameter adjustment direction as the action space; Set a reward function that gives a positive reward for reducing the sound pressure level by 1000Hz and a negative reward for the occurrence of structural interference; Next, the parameter space is explored using a deep deterministic strategy gradient algorithm, and the complex impedance parameters of the sound-absorbing foam are obtained by calling the material acoustics database. The feasibility of the parameters will be verified by combining the exploration results with finite element acoustic coupling simulation.

[0008] Furthermore, in the tire design method of the present invention that utilizes longitudinal grooves to attach sound-absorbing cotton to improve vehicle interior noise, step 4 includes: The longitudinal groove optimization parameters are converted into a 3D CAD model to generate a digital mold file for robotic spraying. Based on the resonance risk map, the inner wall of the longitudinal trench is divided into high, medium, and low acoustic energy zones; A six-axis robot is used to spray high-density polyurethane foam sound-absorbing material onto high-sound-energy areas and low-density material onto low-sound-energy areas.

[0009] Furthermore, in the tire design method of the present invention that utilizes longitudinal grooves to attach sound-absorbing cotton to improve vehicle interior noise, step 4 further includes: Plasma activation treatment was performed on the inner wall of the longitudinal groove before attachment; The thickness of the sound-absorbing cotton is monitored using an infrared thermal imager during the application process; After the bonding is completed, a stepped temperature curing process is used, and defects are detected by X-ray flaw detection.

[0010] Furthermore, in the tire design method of the present invention that utilizes longitudinal grooves to attach sound-absorbing cotton to improve vehicle interior noise, step 5 includes: A multiphysics simulation model was established by integrating the tire model after applying sound-absorbing foam, the multibody dynamics equations of the vehicle suspension, and the acoustic boundary element conditions. Simulations were run under typical road spectrum conditions to obtain simulated attenuation curves for the 1000Hz sound pressure level.

[0011] Furthermore, in the tire design method of the present invention that utilizes longitudinal grooves to attach sound-absorbing cotton to improve vehicle interior noise, step 5 further includes: Compare the 1000Hz sound pressure level attenuation data with a preset threshold; When the attenuation data is lower than the preset threshold, a suggestion for adjusting the longitudinal groove parameters is generated; Suggestions for adjusting the longitudinal trench parameters were sent to the data acquisition system to guide supplementary data collection. The training set of the acoustic feature extraction model is updated using the supplementary collected data.

[0012] Furthermore, in the tire design method of the present invention that utilizes longitudinal grooves to attach sound-absorbing cotton to improve vehicle interior noise, step 2 includes: The confidence level of the resonance sensitivity feature vector output by the convolutional neural network is compared with that of the acoustic impedance distribution map calculated by acoustic topology optimization. The eigenvector weights of the overlapping regions are fed back into the iterative calculation of topology optimization to correct the acoustic impedance boundary conditions. The corrected topology optimization results are used as prior knowledge for feature selection in the attention mechanism of convolutional neural networks.

[0013] Furthermore, in the tire design method of the present invention that utilizes longitudinal grooves to attach sound-absorbing cotton to improve vehicle interior noise, step 2 further includes: Extract the feature vector weight values ​​corresponding to the overlapping regions, and map the weight values ​​to the adjustment coefficients of the acoustic impedance boundary conditions; Substitute the adjustment coefficient into the Helmholtz equation to correct the acoustic impedance boundary value in the equation. The modified Helmholtz equation was then solved using the finite difference method to obtain an updated acoustic impedance distribution map. The updated acoustic impedance distribution map is used as the input for the next topology optimization iteration. The weight extraction, boundary correction, and equation solving processes are repeated until the acoustic impedance distribution converges.

[0014] Beneficial effects of this invention: This invention constructs a noise-structure correlation dataset by collecting geometric parameters of tire grooves and in-vehicle noise data under driving conditions through an on-board sensing system. It uses an acoustic feature extraction model to identify the mapping relationship between groove parameters and noise resonance, and combines acoustic topology optimization to generate a resonance risk map, accurately locating the 1000Hz sound energy concentration area. Using the resonance risk map as the optimization target, it uses a reinforcement learning algorithm to dynamically iterate the combination scheme of groove width and quantity, introduces a material acoustic database to pre-calculate the sound attenuation effect after the sound insulation cotton is attached, and verifies the feasibility of the parameters through finite element acoustic coupling simulation, outputting the groove optimization parameters. Based on the optimized parameters of the longitudinal grooves, the tire mold is digitally designed. Porous composite sound-absorbing cotton is attached to the inner wall of the longitudinal grooves in a gradient density manner according to the sound energy distribution to achieve targeted suppression of sound energy. A simulation system is built on the digital twin platform to simulate driving conditions and output 1000Hz sound pressure level attenuation data. The data is compared with the preset threshold and fed back to the data acquisition stage to start an adaptive optimization cycle, forming a continuous improvement mechanism. This dynamically adjusts the longitudinal groove design, effectively eliminating 1000Hz frequency resonance caused by fixed parameters, improving in-vehicle noise comfort, and overcoming the limitations of existing designs that are difficult to adapt to diverse vehicle models and road conditions. Attached Figure Description

[0015] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on the drawings without creative effort.

[0016] Figure 1 This is a flowchart of the tire design method of the present invention, which uses longitudinal grooves to attach sound-absorbing cotton to improve in-vehicle noise. Detailed Implementation

[0017] To make the technical solution of the present invention clearer, the present invention will be clearly and completely described below with reference to specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. The present invention provided by various embodiments will be described in detail below with reference to the accompanying drawings. To better understand the purpose of the present invention, the present invention will be described in further detail below.

[0018] This invention provides a tire design method for improving in-vehicle noise by attaching sound-absorbing cotton to the longitudinal grooves, comprising: Step 1: Collect the geometric parameters of the tire longitudinal grooves and the in-vehicle noise data under driving conditions through the vehicle-mounted sensing system, and construct a noise and structure correlation dataset with the 1000Hz frequency band as the core. Step 2: Input the noise and structure association dataset into the acoustic feature extraction model to identify the mapping relationship between longitudinal groove parameters and noise resonance, and use the longitudinal groove geometric parameters in the dataset to perform acoustic topology optimization to generate a resonance risk map that marks the area of ​​concentrated acoustic energy. Step 3: Using the resonance risk map as the optimization target, a reinforcement learning algorithm is used to dynamically iterate the combination scheme of longitudinal groove width and number. During the iteration process, the acoustic database of materials is introduced to pre-calculate the sound attenuation effect after the sound-absorbing cotton is attached, and the feasibility of the parameters is verified by finite element acoustic coupling simulation, and the longitudinal groove optimization parameters are output. Step 4: Based on the longitudinal groove optimization parameters, drive the digital design of the tire mold, and attach porous composite sound-absorbing cotton to the inner wall of the longitudinal groove in a gradient density manner according to the sound energy distribution indicated by the resonance risk spectrum. Step 5: Construct a simulation system in the digital twin platform, including a tire model, a vehicle model, and an environment model after the sound-absorbing foam is applied, to simulate driving conditions and output 1000Hz sound pressure level attenuation data; compare the 1000Hz sound pressure level attenuation data with a preset threshold, and feed the comparison result back to the data acquisition stage to guide supplementary data acquisition, and start the adaptive optimization loop.

[0019] This invention provides a tire design method to improve in-vehicle noise by attaching sound-absorbing foam to the longitudinal grooves. First, an onboard sensing system collects the geometric parameters of the tire's longitudinal grooves and in-vehicle noise data under driving conditions, constructing a noise-structure correlation dataset centered on the 1000Hz frequency band. Specifically, the onboard sensing system includes a laser scanner and an acoustic sensor array. The laser scanner acquires the dynamic sequence of the width, depth, and distribution angle of the tire's longitudinal grooves. The acoustic sensor array is arranged inside the vehicle compartment to collect noise signals and extract time-frequency data of the sound pressure level near 1000Hz. Simultaneously, real-time load and speed information is read from the vehicle bus and timestamped with road features output from a road surface texture recognition system. Wavelet denoising and principal component analysis are performed on the synchronized data to eliminate environmental interference and extract key features, thus forming a high-quality noise-structure correlation dataset. This step provides the raw data foundation for subsequent analysis, ensuring that subsequent processing is based on multi-source information under real driving conditions.

[0020] Subsequently, the noise-structure association dataset is input into the acoustic feature extraction model to identify the mapping relationship between longitudinal groove parameters and noise resonance. Acoustic topology optimization is then performed using the longitudinal groove geometric parameters in the dataset to generate a resonance risk map indicating areas of concentrated acoustic energy. The acoustic feature extraction model employs a pre-trained convolutional neural network, strengthening the weights of 1000Hz frequency band features through an attention mechanism, and outputting a resonance sensitivity feature vector. In parallel, the acoustic topology optimization uses a variable density method to solve the longitudinal groove layout problem aimed at minimizing acoustic energy concentration, generating the topology optimization result. The resonance sensitivity feature vector is compared with the topology optimization result, and regions with overlap exceeding a preset threshold are marked, ultimately generating a resonance risk map with weighted coefficients. This step, through collaborative analysis of deep learning and physical models, accurately locates acoustic resonance risk areas, providing a clear objective for parameter optimization.

[0021] Using the resonance risk spectrum as the optimization objective, a reinforcement learning algorithm is employed to dynamically iterate the combination of groove width and quantity. During the iteration process, a materials acoustics database is used to pre-calculate the sound attenuation effect after the sound-absorbing foam is applied. The feasibility of the parameters is verified through finite element acoustic coupling simulation, and the optimized groove parameters are output. The reinforcement learning algorithm defines a model with groove width and quantity as the state space and parameter adjustment direction as the action space, setting a reward function with a positive reward of reducing the sound pressure level by 1000Hz and a negative reward of structural interference. A deep deterministic policy gradient algorithm is used to explore the parameter space, and the complex impedance parameters of the sound-absorbing foam are obtained from the materials acoustics database. The feasibility of the parameters is verified collaboratively with finite element acoustic coupling simulation. This step achieves data-driven parameter optimization, taking into account both acoustic performance and structural constraints, ensuring the reliability of the optimized scheme in practical applications.

[0022] The tire mold is digitally designed based on the optimized parameters of the longitudinal grooves. Porous composite sound-absorbing foam is then applied to the inner wall of the grooves using a gradient density method, according to the acoustic energy distribution indicated by the resonance risk map. The digital design process converts the optimized parameters into a 3D CAD model, generating a digital mold file for robotic spraying. Based on the resonance risk map, the inner wall of the grooves is divided into high, medium, and low acoustic energy zones. A six-axis robot sprays high-density polyurethane foam sound-absorbing foam into the high-acoustic energy zones and low-density material into the low-acoustic energy zones. This step transforms the optimized parameters into a solid structure, achieving targeted acoustic energy suppression through a gradient application process, thereby improving the tire's acoustic performance.

[0023] A simulation system is built in the digital twin platform, including a tire model with sound-absorbing foam applied, a vehicle model, and an environmental model. This system simulates driving conditions and outputs 1000Hz sound pressure level attenuation data. The 1000Hz sound pressure level attenuation data is compared with a preset threshold, and the comparison result is fed back to the data acquisition stage to guide supplementary data collection, initiating an adaptive optimization loop. The simulation system integrates the tire model with sound-absorbing foam applied, the multibody dynamics equations of the vehicle suspension, and acoustic boundary element conditions to establish a multiphysics simulation model. Simulations are run under typical road spectra to obtain simulated attenuation curves for the 1000Hz sound pressure level. When the attenuation data falls below the preset threshold, suggestions for adjusting the longitudinal groove parameters are generated and sent to the data acquisition system to guide supplementary data collection. The new data is used to update the training set of the acoustic feature extraction model. This step forms a closed-loop optimization through virtual verification and feedback mechanisms, enabling the system to continuously improve and adapt to diverse driving conditions.

[0024] Step 1 involves capturing the dynamic changes in the width, depth, and distribution angle of tire longitudinal grooves in real time using a laser scanner. Simultaneously, an acoustic sensor array deployed within the vehicle compartment collects noise signals and extracts time-frequency data of sound pressure levels in the 1000Hz frequency band. This process is precisely timestamped and synchronized with real-time load and speed information read from the vehicle bus and road features output from the road texture recognition system, achieving spatiotemporal alignment of multi-source heterogeneous data. Subsequently, wavelet denoising is performed on the synchronized data to eliminate high-frequency interference, and principal component analysis is used to remove redundant information, ultimately generating a high-quality noise-structure correlation dataset. This series of operations ensures the integrity and accuracy of data acquisition, providing a reliable input foundation for subsequent analysis.

[0025] In step 2, when the pre-trained convolutional neural network processes the noise and structure association dataset, it prioritizes 1000Hz frequency band features through an attention mechanism, outputting a feature vector representing resonance sensitivity. Simultaneously, using the longitudinal groove geometric parameters from the same dataset, an acoustic topology optimization method is employed to solve for the longitudinal groove layout scheme aiming to minimize acoustic energy concentration, generating the topology optimization result. Subsequently, the feature vector output by the convolutional neural network is compared with the topology optimization result in terms of confidence. Regions with overlap exceeding a preset threshold are marked, and a resonance risk map with weighted coefficients is generated based on the marking results. This parallel processing mode integrates the advantages of both data-driven and physical models, improving the accuracy of resonance identification.

[0026] Step 3 first defines a reinforcement learning model with the width and number of longitudinal grooves as the state space and the parameter adjustment direction as the action space, and sets a reward function with a positive reward of reducing the sound pressure level by 1000Hz and a negative reward of structural interference. Then, a deep deterministic policy gradient algorithm is used to explore the parameter space. During this process, the complex impedance parameters of the sound-absorbing foam are obtained from a material acoustics database to simulate the sound attenuation effect after attachment. Finally, the exploration results are co-validated with finite element acoustic coupling simulation to ensure the feasibility of the optimized parameters under both theoretical calculations and engineering constraints. This dynamic iterative mechanism effectively balances acoustic performance and structural reliability.

[0027] Step 4: The optimized parameters of the longitudinal groove are converted into a 3D CAD model, generating a digital mold file suitable for robotic spraying. Based on the acoustic energy distribution indicated by the resonance risk spectrum, the inner wall of the longitudinal groove is divided into high, medium, and low acoustic energy zones. A six-axis robot is used to spray high-density polyurethane foam sound-absorbing material onto the high-acoustic energy zones, while low-density material is sprayed onto the low-acoustic energy zones. This gradient density application method achieves targeted suppression of acoustic energy distribution, ensuring spatial matching between the sound-absorbing material layout and the longitudinal groove structure.

[0028] Before the bonding process in step 4, the inner wall of the longitudinal groove is subjected to plasma activation treatment to enhance the adhesion strength of the sound-absorbing foam. During the bonding process, the thickness uniformity of the sound-absorbing foam is monitored in real time using an infrared thermal imager to avoid localized areas that are too thick or too thin. After bonding, a stepped temperature curing process is used to treat the interface, and internal defects are detected by X-ray flaw detection, thereby ensuring the structural integrity and long-term stability of the bonded layer. This three-stage process control significantly improves the reliability of the sound-absorbing foam bonding.

[0029] Step 5: The digital twin platform integrates the tire model after applying sound-absorbing foam, the vehicle suspension multibody dynamics equations, and acoustic boundary element conditions to establish a multiphysics simulation model. By running the simulation under typical road spectrum conditions, the simulated attenuation curve of the 1000Hz sound pressure level is obtained, thereby predicting the tire's noise performance in a virtual environment. This simulation process provides an alternative to physical trial and error, significantly shortening the development cycle.

[0030] Step 5 compares the simulated 1000Hz sound pressure level attenuation data with a preset threshold. When the attenuation data falls below the threshold, a suggestion for adjusting the longitudinal groove parameters is automatically generated and sent to the data acquisition system to guide supplementary data acquisition. Subsequently, the training set of the acoustic feature extraction model is updated using the supplementary acquired data, forming a closed-loop optimization chain from virtual verification to actual data acquisition. This adaptive mechanism enables the system to continuously optimize to adapt to changing operating conditions.

[0031] In step 2, the resonance sensitivity feature vector output by the convolutional neural network is compared with the acoustic impedance distribution map calculated by acoustic topology optimization. The feature vector weights of the overlapping regions are fed back into the iterative calculation of topology optimization to correct the acoustic impedance boundary conditions. The corrected topology optimization result is then used as prior knowledge for feature selection by the attention mechanism of the convolutional neural network, forming a cross-model cross-validation loop. This bidirectional feedback mechanism enhances the synergy between feature extraction and physical optimization.

[0032] The eigenvector weights corresponding to overlapping regions are extracted and mapped to adjustment coefficients for acoustic impedance boundary conditions. These adjustment coefficients are then substituted into the Helmholtz equation to correct the acoustic impedance boundary values. The corrected Helmholtz equation is solved using the finite difference method to obtain an updated acoustic impedance distribution map. Finally, the updated distribution map is used as input for the next topology optimization iteration, and the processes of weight extraction, boundary correction, and equation solving are repeated until the acoustic impedance distribution converges. This mathematical implementation ensures the accuracy and stability of the topology optimization.

[0033] This invention addresses the 1000Hz frequency resonance problem caused by fixed groove widths and quantities in existing tire longitudinal groove designs through a data-driven dynamic optimization method. First, a vehicle-mounted sensing system collects the geometric parameters of the tire longitudinal grooves and in-vehicle noise data during driving, constructing a noise-structure correlation dataset centered on the 1000Hz frequency band to obtain detailed characteristics of the resonance frequency. Next, the dataset is input into an acoustic feature extraction model to identify the mapping relationship between groove parameters and noise resonance. Acoustic topology optimization is then performed using the groove geometric parameters to generate a resonance risk map indicating areas of concentrated sound energy, accurately locating high-risk areas for 1000Hz resonance. Then, using the resonance risk map as the optimization target, a reinforcement learning algorithm dynamically iterates through combinations of groove widths and quantities. During the iteration process, a material acoustics database is used to pre-calculate the sound attenuation effect after applying sound-absorbing foam. The feasibility of the parameters is verified through finite element acoustic coupling simulation, outputting optimized groove parameters to achieve adaptive adjustment of groove parameters rather than a fixed design. The tire mold is digitally designed based on optimized longitudinal groove parameters. Porous composite sound-absorbing cotton is then applied to the inner wall of the longitudinal groove using a gradient density method according to the acoustic energy distribution indicated by the resonance risk spectrum. This synergistic effect of materials and structure directly suppresses 1000Hz acoustic energy concentration. Finally, a simulation system is built in a digital twin platform to simulate driving conditions and output 1000Hz sound pressure level attenuation data. This data is compared with preset thresholds, and the results are fed back to the data acquisition stage to guide supplementary data collection, forming an adaptive optimization loop. This continuously optimizes the longitudinal groove design to adapt to diverse operating conditions, thereby fundamentally eliminating resonance degradation caused by fixed parameters.

[0034] This invention addresses the problem of 1000Hz frequency resonance degradation caused by the fixed width and number of grooves in existing tire longitudinal groove designs, providing a data-driven dynamic optimization method. Specifically, an onboard sensing system first collects the geometric parameters of the tire longitudinal grooves and in-vehicle noise data under driving conditions, constructing a noise-structure correlation dataset centered on the 1000Hz frequency band. The onboard sensing system includes a laser scanner and an acoustic sensor array. The laser scanner acquires the dynamic sequence of the width, depth, and distribution angle of the tire longitudinal grooves in real time. The acoustic sensor array is deployed inside the vehicle to collect noise signals and extract time-frequency data of the sound pressure level near 1000Hz. Simultaneously, real-time load and speed information is read from the vehicle bus and timestamped with road features output from a road surface texture recognition system, forming a spatiotemporal alignment of multi-source data. Wavelet denoising is performed on the synchronized data to eliminate high-frequency interference, and principal component analysis is combined to extract key features, generating a high-quality noise-structure correlation dataset, providing a reliable input foundation for subsequent analysis.

[0035] A noise-structure correlation dataset is input into an acoustic feature extraction model to identify the mapping relationship between longitudinal groove parameters and noise resonance. Acoustic topology optimization is then performed using the longitudinal groove geometric parameters from the dataset to generate a resonance risk map indicating areas of concentrated acoustic energy. The acoustic feature extraction model employs a pre-trained convolutional neural network, which strengthens the weights of 1000Hz frequency band features through an attention mechanism, outputting a resonance sensitivity feature vector. In parallel, the acoustic topology optimization uses a variable density method to solve the longitudinal groove layout problem aimed at minimizing acoustic energy concentration, generating the topology optimization result. The confidence level of the resonance sensitivity feature vector is compared with the topology optimization result. Regions with overlap exceeding a preset threshold are marked, generating a weighted resonance risk map that accurately locates high-risk areas for 1000Hz resonance.

[0036] Using the resonance risk spectrum as the optimization objective, a reinforcement learning algorithm is employed to dynamically iterate the combination of longitudinal groove width and quantity. The reinforcement learning algorithm defines a model with the groove width and quantity as the state space and the parameter adjustment direction as the action space, setting a reward function that provides a positive reward for reducing the sound pressure level by 1000Hz and a negative reward for structural interference. A deep deterministic policy gradient algorithm is used to explore the parameter space, and the complex impedance parameters of the sound-absorbing foam are obtained from a materials acoustics database to pre-calculate the sound attenuation effect after attachment. The exploration results are coupled with finite element acoustic simulation to verify the feasibility of the parameters, outputting optimized groove parameters and achieving adaptive adjustment of the groove parameters.

[0037] The digital design of the tire mold is driven by optimized parameters of the longitudinal grooves. These parameters are converted into a 3D CAD model, generating digital mold files for robotic spraying. Based on the acoustic energy distribution indicated by the resonance risk spectrum, the inner wall of the longitudinal grooves is divided into high, medium, and low acoustic energy zones. A six-axis robot is used to spray high-density polyurethane foam sound-absorbing material onto the high-acoustic energy zones and low-density material onto the low-acoustic energy zones, achieving gradient density adhesion. Before adhesion, the inner wall of the longitudinal grooves undergoes plasma activation treatment to enhance adhesion. During the adhesion process, the uniformity of the sound-absorbing material thickness is monitored using an infrared thermal imager. After adhesion, a stepped temperature curing process is used to treat the interface, and X-ray flaw detection is used to detect defects, ensuring the structural integrity of the adhesion layer.

[0038] A simulation system was built in a digital twin platform, including a tire model, a vehicle model, and an environmental model after applying sound-absorbing foam. The system integrated the multibody dynamics equations of the vehicle suspension and acoustic boundary element conditions to establish a multiphysics simulation model. Simulations were run under typical road spectra to obtain simulated attenuation curves for the 1000Hz sound pressure level. The attenuation data of the 1000Hz sound pressure level output from the simulation was compared with a preset threshold. When the attenuation data fell below the threshold, suggestions for adjusting the longitudinal groove parameters were automatically generated and sent to the data acquisition system to guide supplementary data collection. The training set of the acoustic feature extraction model was updated using the supplementary data, forming an adaptive optimization loop from virtual verification to actual data acquisition. This continuously optimized the longitudinal groove design to adapt to diverse operating conditions, thereby eliminating resonance degradation caused by fixed parameters.

[0039] Embodiment 1 of this invention: In a high-speed passenger vehicle scenario, addressing the 1000Hz frequency resonance problem caused by fixed longitudinal groove parameters, this invention utilizes an onboard sensing system to collect geometric parameters of the tire longitudinal grooves and in-vehicle noise data, constructing a noise-structure correlation dataset. Specifically, a laser scanner acquires dynamic sequences of the longitudinal groove width, depth, and distribution angle; an acoustic sensor array collects time-frequency data of sound pressure levels near 1000Hz; and simultaneously integrates real-time load, speed information, and road surface texture features provided by the vehicle bus. After timestamp synchronization, wavelet denoising and principal component analysis are used to generate a high-quality dataset. This dataset is input into a pre-trained convolutional neural network, which strengthens the feature weights of the 1000Hz frequency band through an attention mechanism, outputting a resonance sensitivity feature vector. Simultaneously, acoustic topology optimization is performed using the longitudinal groove geometric parameters, employing a variable density method to solve for the longitudinal groove layout with minimum sound energy concentration, generating the topology optimization result. By comparing the feature vector with the topology optimization result, overlapping areas are marked to generate a resonance risk map with weighted coefficients. Using this pattern as the optimization target, a reinforcement learning algorithm is employed to dynamically iterate the combination of groove width and quantity. A deep deterministic gradient algorithm is used to explore the parameter space, and a materials acoustics database is used to pre-calculate the sound attenuation effect after the sound-absorbing foam is applied. Finite element acoustic coupling simulation is combined to verify the feasibility of the parameters, and optimized groove parameters are output. Based on these parameters, the tire mold is digitally designed, and the inner wall of the groove is divided into high, medium, and low-density regions according to sound energy distribution. A six-axis robot is used for gradient density application; high-density polyurethane foam sound-absorbing foam is sprayed onto high-sound-energy regions, while low-density materials are sprayed onto low-sound-energy regions. Plasma activation treatment is performed before application, and the thickness is monitored using an infrared thermal imager during application. After application, stepped temperature curing and X-ray flaw detection are used to ensure quality. A tire-vehicle-environment simulation model is built on a digital twin platform to simulate high-speed driving conditions. 1000Hz sound pressure level attenuation data is output and compared with a threshold. When the attenuation is insufficient, parameter adjustment suggestions are generated and fed back to the data acquisition stage to supplement data and update the model, forming an adaptive optimization loop that effectively suppresses resonance noise in high-speed scenarios.

[0040] Embodiment 2 of this invention: To meet the diverse vehicle and road condition adaptation requirements, this invention addresses the limitations of fixed designs through dynamic longitudinal groove parameter optimization. First, an onboard sensing system collects longitudinal groove geometric parameters and in-vehicle noise data for different vehicle models (e.g., SUVs and sedans) during driving, focusing on recording sound pressure level changes in the 1000Hz frequency band. This data, combined with real-time load and road texture information, constructs a multi-scenario noise and structure correlation dataset. The dataset is input into an acoustic feature extraction model. A convolutional neural network extracts vehicle-specific resonance features through an attention mechanism. Simultaneously, acoustic topology optimization analyzes the acoustic impedance matching defects of the longitudinal groove layout, generating resonance risk maps for different vehicle models. A reinforcement learning algorithm, targeting the map, iteratively optimizes the combination of longitudinal groove width and quantity. A material acoustics database is introduced to match the sound absorption characteristics of different sound-absorbing materials. Finite element simulation verifies the feasibility of the parameters under various road conditions (e.g., asphalt and gravel roads). Optimized parameters drive digital mold design, attaching porous composite sound-absorbing materials according to the acoustic energy distribution gradient, with increased attachment density in high-sound-energy areas for SUVs under high load characteristics. The digital twin platform integrates vehicle parameters and road spectrum, simulates and outputs 1000Hz sound pressure level attenuation data, compares it with threshold values, and provides adjustment suggestions to guide supplementary testing and model updates. Through closed-loop optimization, the system adaptively adjusts the longitudinal groove design, eliminating the risk of resonance under diverse conditions with fixed parameters, and improving ride comfort.

Claims

1. A tire design method for improving in-vehicle noise by attaching sound-absorbing cotton to the longitudinal grooves, characterized in that, include: Step 1: Collect the geometric parameters of the tire longitudinal grooves and the in-vehicle noise data under driving conditions through the vehicle-mounted sensing system, and construct a noise and structure correlation dataset with the 1000Hz frequency band as the core. Step 2: Input the noise and structure association dataset into the acoustic feature extraction model to identify the mapping relationship between longitudinal groove parameters and noise resonance, and use the longitudinal groove geometric parameters in the dataset to perform acoustic topology optimization to generate a resonance risk map that marks the area of ​​concentrated acoustic energy. Step 3: Using the resonance risk map as the optimization target, a reinforcement learning algorithm is used to dynamically iterate the combination scheme of longitudinal groove width and number. During the iteration process, the acoustic database of materials is introduced to pre-calculate the sound attenuation effect after the sound-absorbing cotton is attached, and the feasibility of the parameters is verified by finite element acoustic coupling simulation, and the longitudinal groove optimization parameters are output. Step 4: Based on the longitudinal groove optimization parameters, drive the digital design of the tire mold, and attach porous composite sound-absorbing cotton to the inner wall of the longitudinal groove in a gradient density manner according to the sound energy distribution indicated by the resonance risk spectrum. Step 5: Construct a simulation system in the digital twin platform, including a tire model, a vehicle model, and an environment model after the sound-absorbing foam is applied, to simulate driving conditions and output 1000Hz sound pressure level attenuation data; compare the 1000Hz sound pressure level attenuation data with a preset threshold, and feed the comparison result back to the data acquisition stage to guide supplementary data acquisition, and start the adaptive optimization loop.

2. The tire design method for improving vehicle interior noise by attaching sound-absorbing cotton to the longitudinal grooves according to claim 1, characterized in that, Step 1 includes: A dynamic sequence of the width, depth, and distribution angle of the tire longitudinal grooves was obtained using a laser scanner. At the same time, noise signals are collected by an acoustic sensor array placed inside the carriage, and time-frequency data of sound pressure level around 1000Hz are extracted; Real-time load and speed information is read from the vehicle bus and timestamped with the road features output by the road texture recognition system. Wavelet denoising and principal component analysis are performed on the synchronized data to generate a dataset that correlates noise and structure.

3. The tire design method for improving in-vehicle noise by attaching sound-absorbing cotton to the longitudinal grooves according to claim 1, characterized in that, Step 2 includes: A pre-trained convolutional neural network is used to process the dataset related to noise and structure. The weights of the 1000Hz frequency band features are strengthened through an attention mechanism, and the resonance sensitivity feature vector is output. Simultaneously, by utilizing the longitudinal groove geometric parameters in the noise-structure association dataset, the variable density method is used to solve the longitudinal groove layout topology optimization problem with the objective of minimizing acoustic energy concentration, and the topology optimization results are generated. The resonance sensitivity feature vector is compared with the topology optimization results, and regions with overlap exceeding a preset threshold are marked. A resonance risk map with weighted coefficients is generated based on the labeling results.

4. A tire design method for improving in-vehicle noise by attaching sound-absorbing cotton to the longitudinal grooves according to claim 3, characterized in that, Step 3 includes: Define a reinforcement learning model with the width and number of longitudinal grooves as the state space and the parameter adjustment direction as the action space; Set a reward function that gives a positive reward for reducing the sound pressure level by 1000 Hz and a negative reward for the occurrence of structural interference; Next, the parameter space is explored using a deep deterministic strategy gradient algorithm, and the complex impedance parameters of the sound-absorbing foam are obtained by calling the material acoustics database. The feasibility of the parameters will be verified by combining the exploration results with finite element acoustic coupling simulation.

5. A tire design method for improving in-vehicle noise by attaching sound-absorbing cotton to the longitudinal grooves according to claim 1, characterized in that, Step 4 includes: The longitudinal groove optimization parameters are converted into a 3D CAD model to generate a digital mold file for robotic spraying. Based on the resonance risk map, the inner wall of the longitudinal trench is divided into high, medium, and low acoustic energy zones; A six-axis robot is used to spray high-density polyurethane foam sound-absorbing material onto high-sound-energy areas and low-density material onto low-sound-energy areas.

6. A tire design method for improving in-vehicle noise by attaching sound-absorbing cotton to the longitudinal grooves according to claim 5, characterized in that, Step 4 also includes: Plasma activation treatment was performed on the inner wall of the longitudinal groove before attachment; The thickness of the sound-absorbing cotton is monitored using an infrared thermal imager during the application process; After the bonding is completed, a stepped temperature curing process is used, and defects are detected by X-ray flaw detection.

7. A tire design method for improving in-vehicle noise by attaching sound-absorbing cotton to the longitudinal grooves according to claim 1, characterized in that, Step 5 includes: A multiphysics simulation model was established by integrating the tire model after applying sound-absorbing foam, the multibody dynamics equations of the vehicle suspension, and the acoustic boundary element conditions. Simulations were run under typical road spectrum conditions to obtain simulated attenuation curves for the 1000Hz sound pressure level.

8. A tire design method for improving in-vehicle noise by attaching sound-absorbing cotton to the longitudinal grooves according to claim 7, characterized in that, Step 5 further includes: Compare the 1000Hz sound pressure level attenuation data with a preset threshold; When the attenuation data is lower than the preset threshold, a suggestion for adjusting the longitudinal groove parameters is generated; Suggestions for adjusting the longitudinal trench parameters were sent to the data acquisition system to guide supplementary data collection. The training set of the acoustic feature extraction model is updated using the supplementary collected data.

9. A tire design method for improving in-vehicle noise by attaching sound-absorbing cotton to the longitudinal grooves according to claim 1, characterized in that, Step 2 includes: The confidence level of the resonance sensitivity feature vector output by the convolutional neural network is compared with that of the acoustic impedance distribution map calculated by acoustic topology optimization. The eigenvector weights of the overlapping regions are fed back into the iterative calculation of topology optimization to correct the acoustic impedance boundary conditions. The corrected topology optimization results are used as prior knowledge for feature selection in the attention mechanism of convolutional neural networks.

10. A tire design method for improving in-vehicle noise by attaching sound-absorbing cotton to the longitudinal grooves according to claim 9, characterized in that, Step 2 also includes: Extract the feature vector weight values ​​corresponding to the overlapping regions, and map the weight values ​​to the adjustment coefficients of the acoustic impedance boundary conditions; Substitute the adjustment coefficient into the Helmholtz equation to correct the acoustic impedance boundary value in the equation. The modified Helmholtz equation was then solved using the finite difference method to obtain an updated acoustic impedance distribution map. The updated acoustic impedance distribution map is used as the input for the next topology optimization iteration. The weight extraction, boundary correction and equation solving processes are repeated until the acoustic impedance distribution converges.

Images