Multi-level road noise optimization analysis method and device

Abstract

The invention relates to the technical field of road noise analysis, in particular to a multi-level road noise optimization analysis method and device, and the method comprises the steps: carrying out the preset road noise target and influence factor analysis of a vehicle, so as to obtain a level decomposition architecture, and combining a global optimization algorithm, a local optimization algorithm and a Transformer network, so as to obtain a road noise target; establishing and training a road noise machine learning prediction model; a real vehicle road test is carried out on the vehicle to collect various vibration data and driver right ear noise data, and corresponding vibration acceleration load signals are collected; and inputting the collected data into the road noise machine learning prediction model so as to output a multi-level road noise analysis result of the vehicle. Therefore, the problems that an existing road noise analysis method mainly focuses on a test or simulation technology, historical data containing key design features are ignored, the risk of a local optimal solution for directly solving the road noise problem through global optimization is high, and the global search capability is limited are solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of road noise analysis, and particularly relates to a multi-level road noise optimization analysis method and device. BACKGROUND

[0002] Traditional mechanism type working means of NVH mainly includes two aspects of test technology and CAE technology. As a basic method for promoting the progress of automobile technology, the test method mainly studies the verification of problems and solutions through test specifications, test equipment and test sites.

[0003] Through the test method to study the road noise problem, there are problems of difficult to obtain the entity parameters of the structure, long time-consuming of multiple single repeated tests, easy to produce errors, low work efficiency, high test cost and the like. While through the simulation problem to study the road noise problem, the complexity of the vehicle body system and the suspension system and the application of a large amount of composite materials, the physical mechanism is complex, the modeling precision is too low, the modeling parameters are difficult to obtain, the modeling efficiency is low and the like are still more prominent, which urgently need to be solved. SUMMARY

[0004] The present application provides a multi-level road noise optimization analysis method and device to solve the problems that the existing road noise analysis method is usually based on test or simulation technology, ignores the historical data containing key design features, and has high risk of local optimal solution of road noise problem solved directly by global optimization, limited global search ability and the like.

[0005] The first aspect embodiment of the present application provides a multi-level road noise optimization analysis device, including the following steps: performing preset road noise target and influence factor analysis on a target vehicle to obtain corresponding analysis data, and constructing a hierarchical decomposition architecture corresponding to the target vehicle according to the analysis data, and based on the hierarchical decomposition architecture, a preset global optimization algorithm, a local optimization algorithm and a Transformer network, establishing a road noise machine learning prediction model, and training the road noise machine learning prediction model; performing real vehicle road test on the target vehicle to collect corresponding multiple vibration data and driver right ear noise data, and collecting corresponding vibration acceleration load signals through a preset steering knuckle position sensor, wherein the multiple vibration data includes steering knuckle vibration data, suspension vibration data and vehicle body vibration data; inputting the multiple vibration data, the driver right ear noise data and the vibration acceleration load signals into the trained road noise machine learning prediction model to output a multi-level road noise analysis result corresponding to the target vehicle.

[0006] According to the technical means, the embodiment of the present application can effectively reduce the test and simulation cost, improve the efficiency of road noise analysis, realize the rapid and accurate analysis of road noise problems, and help designers to evaluate the noise effect of different parameter combinations in a short time, thereby greatly reducing human errors.

[0007] Optionally, in an embodiment of the present application, the preset road noise target and influence factor analysis of the target vehicle is performed to obtain corresponding analysis data, and a hierarchical decomposition architecture corresponding to the target vehicle is constructed according to the analysis data, including: based on a preset transfer path analysis method and a knowledge graph theory, the preset road noise target and influence factor analysis of the target vehicle is performed to obtain corresponding analysis data, so as to divide a chassis system level, a vehicle body system level and an in-vehicle noise response level according to the analysis data; based on a front deflector, a front floor, a front wall system sheet metal, a rear floor, a front longitudinal beam, a rear longitudinal beam and a rear roof of the target vehicle, a road noise response analysis of the target vehicle is performed to obtain corresponding target influence factors; based on the target influence factors and a preset TPA strategy, a plurality of target parameters corresponding to the chassis system level, the vehicle body system level and the in-vehicle noise response level are determined, and a hierarchical decomposition architecture of the target vehicle is constructed according to the plurality of target parameters, the target influence factors, the chassis system level, the vehicle body system level and the in-vehicle noise response level.

[0008] According to the technical means, the embodiment of the present application can accurately analyze road noise influence factors relying on scientific methods, construct a clear hierarchical architecture, focus on key vehicle body parts, and combine TPA strategies to clearly present the road noise transmission path and provide accurate and clear data basis for subsequent road noise optimization.

[0009] Optionally, in an embodiment of the present application, the hierarchical decomposition architecture, a preset global optimization algorithm, a local optimization algorithm and a Transformer network are used to establish a road noise machine learning prediction model, and the road noise machine learning prediction model is trained, including: a plurality of subsystems in the chassis system level and the vehicle body system level of the hierarchical decomposition architecture are modeled to obtain corresponding sub-models, and the vehicle body system level and the in-vehicle noise response level in the hierarchical decomposition architecture are modeled to obtain a corresponding road noise whole vehicle model, so as to construct the road noise machine learning prediction model based on the Transformer network, the sub-models and the road noise whole vehicle model; the parameters of the sub-models are optimized by the global optimization algorithm, and the whole vehicle model optimal parameter searching operation of the road noise whole vehicle model is performed by using the local optimization algorithm, so as to obtain the trained road noise machine learning prediction model.

[0010] Based on the above technical means, the embodiments of this application combine local optimization and global optimization, so that local optimization is applied to multiple sub-models of the chassis system-body system, and global optimization is applied to the road noise vehicle model of the body system-in-vehicle noise response. This can effectively improve the learning efficiency of the Transformer model for the hidden nonlinear laws of road noise, help designers evaluate the noise effect of different parameter combinations in a short time, and enable designers to use optimization algorithms to automatically recommend the best solution, greatly reducing human error, improving the efficiency and accuracy of design solutions, and realizing rapid and accurate analysis of road noise problems.

[0011] Optionally, in one embodiment of this application, the step of conducting a real-vehicle road test on the target vehicle to collect various vibration data and driver's right ear noise data, and collecting corresponding vibration acceleration load signals through a preset steering knuckle position sensor, wherein the various vibration data include steering knuckle vibration data, suspension vibration data, and vehicle body vibration data, includes: selecting the test environment, test equipment, and test conditions for the real-vehicle road test, and conducting a real-vehicle road test on the target vehicle based on the test environment, the test equipment, and the test conditions to collect various vibration data and driver's right ear noise data; determining the arrangement principle of the steering knuckle sensor, and arranging the steering knuckle sensor at the steering knuckle position of the front and rear suspensions of the target vehicle according to the arrangement principle, so as to collect the vibration acceleration load signal of the target vehicle through the steering knuckle sensor.

[0012] Based on the above technical means, the embodiments of this application ensure the accuracy and completeness of vibration, noise and other data acquisition through standardized test procedures and sensor arrangement requirements, and ensure the reliability of data through full on-site control, providing high-quality load excitation data support for subsequent road noise CAE simulation and optimization.

[0013] A second aspect of this application provides a multi-level road noise optimization analysis device, comprising: a modeling module, used to perform preset road noise target and influencing factor analysis on a target vehicle to obtain corresponding analysis data, and to construct a hierarchical decomposition architecture corresponding to the target vehicle based on the analysis data, and to establish a road noise machine learning prediction model based on the hierarchical decomposition architecture, a preset global optimization algorithm, a local optimization algorithm and a Transformer network, and to train the road noise machine learning prediction model; a testing module, used to conduct real-vehicle road tests on the target vehicle to collect corresponding various vibration data and driver's right ear noise data, and to collect corresponding vibration acceleration load signals through a preset steering knuckle position sensor, wherein the various vibration data includes steering knuckle vibration data, suspension vibration data and body vibration data; and an analysis module, used to input the various vibration data, the driver's right ear noise data and the vibration acceleration load signals into the trained road noise machine learning prediction model to output the multi-level road noise analysis results corresponding to the target vehicle.

[0014] Optionally, in one embodiment of this application, the modeling module includes: an acquisition unit, used to perform preset road noise target and influencing factor analysis on the target vehicle based on a preset transmission path analysis method and knowledge graph theory to obtain corresponding analysis data, so as to divide the chassis system level, body system level and in-vehicle noise response level according to the analysis data; a road noise response analysis unit, used to perform road noise response analysis on the target vehicle based on the front spoiler, front floor, front bulkhead sheet metal, rear floor, front longitudinal beam, rear longitudinal beam and rear roof, so as to obtain corresponding target influencing factors; and a determination unit, used to determine multiple target parameters corresponding to the chassis system level, body system level and in-vehicle noise response level based on the target influencing factors and a preset TPA strategy, and to construct the hierarchical decomposition architecture of the target vehicle according to the multiple target parameters, the target influencing factors, the chassis system level, body system level and in-vehicle noise response level.

[0015] Optionally, in one embodiment of this application, the modeling module further includes: a construction unit, configured to model multiple subsystems in the chassis system level and the body system level of the hierarchical decomposition architecture to obtain corresponding sub-models, and to model the body system level and the in-vehicle noise response level in the hierarchical decomposition architecture to obtain corresponding road noise vehicle models, so as to construct the road noise machine learning prediction model based on the Transformer network, the sub-models and the road noise vehicle models; and an optimization unit, configured to optimize the parameters of the sub-models using the global optimization algorithm, and to perform a vehicle model optimal parameter search operation on the road noise vehicle models using the local optimization algorithm, so as to obtain the trained road noise machine learning prediction model.

[0016] Optionally, in one embodiment of this application, the test module includes: a selection unit, used to select the test environment, test equipment, and test conditions for the actual vehicle road test, and to conduct an actual vehicle road test on the target vehicle based on the test environment, the test equipment, and the test conditions to collect various vibration data and driver's right ear noise data; and a collection unit, used to determine the arrangement principle of the steering knuckle sensor, and to arrange the steering knuckle sensor at the front and rear suspension steering knuckle positions of the target vehicle according to the arrangement principle, so as to collect the vibration acceleration load signal of the target vehicle through the steering knuckle sensor.

[0017] A third aspect of this application provides an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the multi-level road noise optimization analysis method as described in the above embodiments.

[0018] A fourth aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the multi-level road noise optimization analysis method described above.

[0019] A fifth aspect of this application provides a computer program product, including a computer program that is executed to implement the above-described multi-level road noise optimization analysis method.

[0020] Therefore, the embodiments of this application have the following beneficial effects: The embodiments of this application analyze the target vehicle for preset road noise targets and influencing factors to obtain corresponding analysis data. Based on the analysis data, a hierarchical decomposition architecture corresponding to the target vehicle is constructed. A road noise machine learning prediction model is established based on the hierarchical decomposition architecture, preset global optimization algorithms, local optimization algorithms, and a Transformer network, and the road noise machine learning prediction model is trained. Real-vehicle road tests are conducted on the target vehicle to collect various vibration data and driver's right ear noise data. Vibration acceleration load signals are collected through a preset steering knuckle position sensor. The various vibration data include steering knuckle vibration data, suspension vibration data, and body vibration data. These vibration data, driver's right ear noise data, and vibration acceleration load signals are input into the trained road noise machine learning prediction model to output the multi-level road noise analysis results corresponding to the target vehicle. This application combines historical road noise data to study and optimize existing or future road noise problems, thereby effectively reducing testing and simulation costs, improving the efficiency of road noise analysis, and enabling rapid and accurate analysis of road noise problems. This helps designers evaluate the noise effects of different parameter combinations in a short time and greatly reduces human error. This solves the problems of existing road noise analysis methods, which usually rely on experimental or simulation techniques, neglect historical data containing key design features, and have high risks and limited global search capabilities when directly solving road noise problems through global optimization.

[0021] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0022] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is a flowchart of a multi-level road noise optimization analysis method provided according to an embodiment of this application; Figure 2 A schematic diagram of a multi-level road noise decomposition architecture provided in this application embodiment; Figure 3 This is a schematic diagram of a suspension component provided in an embodiment of this application; Figure 4 A schematic diagram of a vehicle body provided in an embodiment of this application; Figure 5 A schematic diagram of the basic structure of a Transformer provided in an embodiment of this application; Figure 6 A schematic diagram of the execution logic of an ACO-Transformer is provided for embodiments of this application; Figure 7 A schematic diagram of the execution logic of a PSO-Transformer provided in an embodiment of this application; Figure 8 This is an example diagram of a multi-level road noise optimization analysis device according to an embodiment of this application; Figure 9 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0023] Among them, 10-multi-level road noise optimization analysis device; 100-modeling module, 200-experiment module, 300-analysis module; 901-memory, 902-processor, 903-communication interface. Detailed Implementation

[0024] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, 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.

[0025] The following description, with reference to the accompanying drawings, describes a multi-level road noise optimization analysis method and apparatus according to embodiments of this application. Addressing the problems mentioned in the background section, this application provides a multi-level road noise optimization analysis method. In this method, a preset road noise target and influencing factor analysis is performed on a target vehicle to obtain corresponding analysis data. Based on the analysis data, a hierarchical decomposition architecture corresponding to the target vehicle is constructed. A road noise machine learning prediction model is established based on the hierarchical decomposition architecture, a preset global optimization algorithm, a local optimization algorithm, and a Transformer network. The road noise machine learning prediction model is then trained. A real-vehicle road test is conducted on the target vehicle to collect various vibration data and driver's right ear noise data. A preset steering knuckle position sensor is used to collect corresponding vibration acceleration load signals. The various vibration data include steering knuckle vibration data, suspension vibration data, and vehicle body vibration data. The various vibration data, driver's right ear noise data, and vibration acceleration load signals are input into the trained road noise machine learning prediction model to output the multi-level road noise analysis results corresponding to the target vehicle. This application combines historical road noise data to study and optimize existing or future road noise problems, thereby effectively reducing testing and simulation costs, improving the efficiency of road noise analysis, and enabling rapid and accurate analysis of road noise issues. This helps designers quickly evaluate the noise effects of different parameter combinations and greatly reduces human error. Therefore, it solves the problems of existing road noise analysis methods, which typically rely on testing or simulation techniques, neglecting historical data containing key design features, and the high risk of finding local optima and limited global search capabilities when directly addressing road noise problems through global optimization.

[0026] Specifically, Figure 1 This is a flowchart of a multi-level road noise optimization analysis method provided in an embodiment of this application.

[0027] like Figure 1 As shown, this multi-level road noise optimization analysis method includes the following steps: In step S101, a preset road noise target and influencing factor analysis is performed on the target vehicle to obtain corresponding analysis data. Based on the analysis data, a hierarchical decomposition architecture corresponding to the target vehicle is constructed. Based on the hierarchical decomposition architecture, preset global optimization algorithm, local optimization algorithm and Transformer network, a road noise machine learning prediction model is established and trained.

[0028] The embodiments of this application first conduct a systematic analysis of the target value of road noise and the influencing factors for the target vehicle, output complete analysis results, and build a hierarchical decomposition architecture of vehicle road noise based on the results. Then, this architecture is integrated with preset global optimization algorithms, local optimization algorithms and Transformer networks to construct a road noise machine learning prediction model, and the training and optimization of the model are completed.

[0029] Therefore, the embodiments of this application can accurately locate the core influencing factors of road noise and integrate multi-algorithm and network models to improve the accuracy and generalization ability of road noise prediction, providing a reliable basis for vehicle road noise optimization.

[0030] Optionally, in one embodiment of this application, a preset road noise target and influencing factor analysis is performed on the target vehicle to obtain corresponding analysis data. Based on the analysis data, a hierarchical decomposition architecture corresponding to the target vehicle is constructed, including: performing a preset road noise target and influencing factor analysis on the target vehicle based on a preset transmission path analysis method and knowledge graph theory to obtain corresponding analysis data, and dividing the chassis system level, body system level, and in-vehicle noise response level according to the analysis data; performing road noise response analysis on the target vehicle based on the front spoiler, front floor, front bulkhead sheet metal, rear floor, front longitudinal beam, rear longitudinal beam, and rear roof to obtain corresponding target influencing factors; determining multiple target parameters corresponding to the chassis system level, body system level, and in-vehicle noise response level based on the target influencing factors and a preset TPA strategy, and constructing a hierarchical decomposition architecture for the target vehicle based on the multiple target parameters, target influencing factors, chassis system level, body system level, and in-vehicle noise response level.

[0031] In practical implementation, embodiments of this application can analyze road noise targets and influencing factors based on transmission path analysis methods and knowledge graph theory, ultimately dividing the road noise into a hierarchical decomposition architecture including chassis system (third level), body system (second level), and noise response (first level). The road noise hierarchical architecture is as follows: Figure 2 As shown in the diagram. This architecture includes the transmission paths of key influencing factors and incorporates seven components—front spoiler, front floor, front bulkhead sheet metal, rear floor, front longitudinal beams (including front shock absorbers), rear longitudinal beams (including rear shock absorbers), and rear roof—to analyze road noise response. The variations in the subframe and body are shown below. Figure 3 and Figure 4 As shown.

[0032] Secondly, the embodiments of this application can combine a three-level road noise decomposition architecture to obtain the main influencing factors (i.e., target influencing factors) in road noise problem analysis, and combine the TPA method to determine the main parameters (i.e., target parameters) included in the three levels. The first level is noise response; the second level (body system) is acceleration response, IPI (original dynamic stiffness), and NTF (noise transfer function) at each connection point. The connection points (left side) include: left front shock absorber, left upper control arm front mounting point, left upper control arm rear mounting point, etc. The front subframe front mounting point, front subframe rear mounting point, rear subframe front mounting point, rear subframe rear mounting point, and rear shock absorber; the third level (chassis system) consists of the acceleration response at each connection point, the IPI at the sheet metal parts, and the VTF (Vibration Transfer Function) from the passive end to the active end. The connection points (left side) include: left front shock absorber, left upper control arm front mounting point, left upper control arm rear mounting point, front subframe front mounting point, front subframe rear mounting point, rear subframe front mounting point, rear subframe rear mounting point, and rear shock absorber.

[0033] Therefore, the embodiments of this application can rely on scientific methods to accurately analyze the factors affecting road noise, construct a clear hierarchical architecture, focus on key vehicle body components, and combine the TPA strategy to clarify the core parameters of each level, thereby clearly presenting the road noise transmission path and providing accurate and clear data basis for subsequent road noise optimization.

[0034] Optionally, in one embodiment of this application, a road noise machine learning prediction model is established based on a hierarchical decomposition architecture, a preset global optimization algorithm, a local optimization algorithm, and a Transformer network, and the road noise machine learning prediction model is trained. This includes: modeling multiple subsystems in the chassis system level and the body system level of the hierarchical decomposition architecture to obtain corresponding sub-models; and modeling the body system level and the in-vehicle noise response level in the hierarchical decomposition architecture to obtain corresponding road noise vehicle models. Based on the Transformer network, the sub-models, and the road noise vehicle models, a road noise machine learning prediction model is constructed. The parameters of the sub-models are optimized using a global optimization algorithm, and the optimal parameters of the vehicle model are searched using a local optimization algorithm to obtain the trained road noise machine learning prediction model.

[0035] It should be noted that the embodiments of this application establish a road noise machine learning prediction model based on a hierarchical architecture, and model 16 sub-models of the chassis system-body system based on Transformer, while using PSO for model parameter optimization, thereby achieving high-efficiency and high-precision modeling of complex path sub-models. For a whole vehicle model of the body system-noise response, modeling is performed based on the Transformer structure, and ACO is used to find the optimal parameters of the whole vehicle model, thereby better addressing the complex optimization problems of road noise chassis and body noise.

[0036] The following embodiments of this application provide a detailed description of the Transformer basic algorithm, the PSO optimization algorithm, and the ACO ant colony algorithm.

[0037] Among them, Transformer is a deep learning model based on an attention mechanism. Multi-head attention uses multiple parallel attention sub-modules to represent and weight input information from different perspectives, thereby improving the model's ability to perceive contextual features and its expressive diversity. Single-head attention performs three linear transformations on the vector at each position in the input sequence to generate a query matrix Q, a key matrix K, and a value matrix V. A schematic diagram of the multi-head attention mechanism is shown below. Figure 5 As shown, the calculation formula is as follows:

[0038] Where Q, K, and V represent the Query, Key, and Value matrices, respectively; d k This represents the dimension of the key vector, used to scale the dot product to stabilize the softmax function.

[0039] In a multi-headed attention system, this mechanism operates in parallel. h Next, forming hThere are several attention heads, each using an independent linear transformation matrix to map the input to a different subspace. Self-attention computation is then performed independently in each subspace. The specific details are shown in the following equation:

[0040] in, , , Indicates the first i The linear transformation matrix of the head.

[0041] After completing the self-attention operations for all heads, their outputs are concatenated and fused using a unified linear transformation to integrate information from multiple subspaces, ultimately outputting the multi-head attention result, as shown in the following equation:

[0042] in, This represents the weight matrix of the output layer; This indicates a splicing operation.

[0043] Secondly, Ant Colony Optimization (ACO) is a swarm intelligence optimization method based on the foraging behavior of ants in nature. It simulates the cooperative search using pheromones during foraging, forming a stochastic optimization mechanism relying on positive feedback and parallel search. In the initial stage of the algorithm, several "artificial ants" randomly select paths in the solution space, gradually constructing feasible solutions; the quality of the paths is measured by the objective function and indirectly reflected by pheromone concentration. ACO has certain advantages in both local feature exploration and global feature learning. Introducing ACO to optimize key hyperparameter combinations of the Transformer can better address the complex optimization problems of the whole vehicle model. The ACO-Transformer structure diagram is shown below. Figure 6 As shown in the specific formula of the ant colony algorithm, during the optimization process, the steps of the second and third formulas below are repeated until the maximum number of iterations or the quality of the solution no longer improves. Finally, the optimal path is retained as the solution to the problem, as shown in the following formula:

[0044] in, side( i , j pheromone concentration; As a heuristic factor; To control the weights of pheromones and heuristic factors.

[0045]

[0046]

[0047] in, (0, 1), Indicates the first k Only ants on the path ( i , j The pheromones released on the surface are usually defined as Q / ( Q It is a constant. (where is the path length of the k-th ant).

[0048] Furthermore, Particle Swarm Optimization (PSO) is a type of evolutionary algorithm that simulates the swarm intelligence of birds foraging or fish swimming. It initially consists of a swarm of N particles, each represented by three metrics: position, velocity, and fitness. Each particle represents a potential solution and adjusts its position based on its own experience and that of the best particle in the swarm. In other words, each particle is guided by its current position and historical best positions, and is also influenced by the best position in the swarm. During the search process, it continuously adjusts its velocity and position, gradually approaching the optimal solution. The particle velocity and position updates are shown in the following equation:

[0049] in, This indicates the position of the particle in the search space; This represents the rate of change of a particle's position, when the particle... i When the iteration reaches the optimal position, it is denoted as When the population reaches its optimal position, it is denoted as... k is the current iteration step; Inertial weights; These are two non-negative constants, called acceleration factors; It is a random number in (0, 1).

[0050] Understandably, the rapid development of machine learning technology across various industries has provided new approaches to solving road noise problems. Manufacturers typically retain a large amount of historical test and simulation data during the R&D process. Introducing the experience gained from this historical data into the development of new vehicle models can improve the utilization rate of this data and enable rapid analysis of road noise issues. The embodiments of this application can highly utilize the manufacturer's historically accumulated data, thereby improving the efficiency of road noise problem analysis and reducing research costs. Based on this historical data, the influencing factors of the road noise problem are decomposed according to the transmission path, forming a hierarchical decomposition architecture. Based on the Transformer basic prediction model, PSO (Particle Swarm Optimization) and ACO (Ant Colony Optimization) are introduced at each level for detailed and global optimization, ultimately achieving refined optimization of the chassis-to-suspension path, accurately adjusting local problems in noise response prediction, and effectively optimizing the vehicle body noise response at the global level from the body system to the noise response path. Especially when facing complex multi-dimensional noise sources, ACO's global exploration can effectively find better solutions.

[0051] Compared to traditional experimental and simulation methods for analyzing road noise, machine learning technology, by learning from a large amount of historical data, can effectively capture complex nonlinear relationships and high-dimensional features, automatically identify the complex correlation between features and road noise performance, and provide more accurate and reliable predictions. Furthermore, compared to most existing methods that use a single optimization algorithm for global optimization, this application's embodiments combine local and global optimization strategies, thereby more efficiently addressing the complex optimization problem of road noise across chassis-suspension-in-vehicle noise. The established road noise prediction model has significant advantages in terms of efficiency, robustness, and low cost.

[0052] Therefore, the embodiments of this application combine local optimization and global optimization, so that local optimization is applied to the 16 sub-models of the chassis system-body system, and global optimization is applied to the road noise vehicle model of the body system-in-vehicle noise response. This can effectively improve the learning efficiency of the Transformer model for the hidden nonlinear laws of road noise, help designers evaluate the noise effect of different parameter combinations in a short time, and enable designers to use optimization algorithms to automatically recommend the best solution, greatly reducing human error, improving the efficiency and accuracy of design solutions, and realizing rapid and accurate analysis of road noise problems.

[0053] In step S102, a real-vehicle road test is conducted on the target vehicle to collect various vibration data and driver's right ear noise data, and to collect corresponding vibration acceleration load signals through a preset steering knuckle position sensor. The various vibration data include steering knuckle vibration data, suspension vibration data and vehicle body vibration data.

[0054] In step S103, various vibration data, driver's right ear noise data, and vibration acceleration load signals are input into the trained road noise machine learning prediction model to output the multi-level road noise analysis results corresponding to the target vehicle.

[0055] Furthermore, embodiments of this application can collect various vibration data such as those from the steering knuckle, suspension, and body, as well as noise data from the driver's right ear, through real-vehicle road tests of the target vehicle. Vibration acceleration load signals can be obtained using a steering knuckle position sensor, and these data can be input into a trained road noise machine learning prediction model to output multi-level road noise analysis results for the vehicle.

[0056] Therefore, the embodiments of this application can ensure the authenticity and reliability of the information input to the model based on real vehicle data, and combine dedicated sensors to accurately capture key load signals, so that the multi-level results output by the model can provide comprehensive and accurate data support for the localization and optimization of road noise problems.

[0057] Optionally, in one embodiment of this application, a real-vehicle road test is conducted on the target vehicle to collect various vibration data and driver's right ear noise data, and a corresponding vibration acceleration load signal is collected through a preset steering knuckle position sensor. The various vibration data include steering knuckle vibration data, suspension vibration data, and vehicle body vibration data. This includes: selecting the test environment, test equipment, and test conditions for the real-vehicle road test, and conducting a real-vehicle road test on the target vehicle based on the test environment, test equipment, and test conditions to collect various vibration data and driver's right ear noise data; determining the arrangement principle of the steering knuckle sensor, and arranging the steering knuckle sensor at the steering knuckle position of the front and rear suspensions of the target vehicle according to the arrangement principle, so as to collect the vibration acceleration load signal of the target vehicle through the steering knuckle sensor.

[0058] In the specific implementation process, in order to collect vibration data of the steering knuckle, suspension and body, as well as noise data of the driver's right ear inside the vehicle, the embodiments of this application need to conduct real vehicle road tests. The test environment is a rough road surface (cobblestone road surface with a diameter of about 20-30mm). The test equipment includes LMS data acquisition front end, microphone, and acceleration sensor. The test condition is a constant speed of 60km / h.

[0059] The sensors used in the test were mainly placed in locations such as the steering knuckle, seat, steering wheel, and ear. After placement, detailed position coordinates needed to be recorded for labeling in the finite element model. Sensors were placed at the steering knuckles of the front and rear suspensions of the vehicle to collect vibration acceleration load signals. The CAE (Computer-Aided Engineering) model of the vehicle in the simulation environment maintained the same point locations as the test.

[0060] It should be noted that the following should be considered when arranging the steering knuckle position sensor: 1) Ideally, five sensors should be placed at each steering knuckle position; 2) The orientation of the sensors should be aligned with the overall vehicle orientation as much as possible; 3) The sensors should be placed in locations with high rigidity whenever possible; 4) It should be ensured that at least one sensor is not on the same plane as the other three; 5) The locations of the sensors should be measured, photographed, and recorded in detail to facilitate their identification in the model; 6) The test sensors must correspond one-to-one with the sensors in the finite element model, and the order must be consistent; 7) The sensor coordinate system must be set accurately; 8) Ensure that the positions of the sensors on the left and right wheels are consistent.

[0061] Furthermore, in the embodiments of this application, the following information should be recorded during the vehicle and test site preparation: vehicle model information, presence or absence of a sunroof, powertrain model, whether the air conditioning is turned off, weather conditions, vehicle speed, number of test system channels, tire model, size, tire pressure, etc. During the test, all channels should be tested in one go if possible. If there are not enough channels, the front and rear wheels should be measured twice. In addition, the consistency and validity of the test data should be confirmed on-site.

[0062] In addition, CAE engineers should track the entire process of road noise load excitation acquisition test on site, and confirm the location of sensor points and the validity and consistency of test data on site to avoid the situation where data problems are found after leaving the test site after the test.

[0063] Therefore, the embodiments of this application ensure the accuracy and completeness of vibration, noise and other data acquisition through standardized test procedures and sensor arrangement requirements, and ensure the reliability of data through full on-site control, providing high-quality load excitation data support for subsequent road noise CAE simulation and optimization.

[0064] According to the multi-level road noise optimization analysis method proposed in this application, the target vehicle is subjected to preset road noise targets and influencing factors analysis to obtain corresponding analysis data. Based on the analysis data, a hierarchical decomposition architecture corresponding to the target vehicle is constructed. Based on the hierarchical decomposition architecture, preset global optimization algorithms, local optimization algorithms, and Transformer networks, a road noise machine learning prediction model is established and trained. The target vehicle is subjected to real-vehicle road tests to collect various vibration data and driver's right ear noise data. The corresponding vibration acceleration load signal is collected through a preset steering knuckle position sensor. The various vibration data include steering knuckle vibration data, suspension vibration data, and body vibration data. The various vibration data, driver's right ear noise data, and vibration acceleration load signal are input into the trained road noise machine learning prediction model to output the multi-level road noise analysis results corresponding to the target vehicle. This application combines historical road noise data to study and optimize existing or future road noise problems, thereby effectively reducing testing and simulation costs, improving the efficiency of road noise analysis, and enabling rapid and accurate analysis of road noise problems. This helps designers evaluate the noise effects of different parameter combinations in a short time and greatly reduces human error.

[0065] Secondly, the multi-level road noise optimization analysis device proposed according to the embodiments of this application is described with reference to the accompanying drawings.

[0066] Figure 8 This is a block diagram of a multi-level road noise optimization analysis device according to an embodiment of this application.

[0067] like Figure 8 As shown, the multi-level road noise optimization analysis device 10 includes: a modeling module 100, an experimental module 200, and an analysis module 300.

[0068] The modeling module 100 is used to perform preset road noise target and influencing factor analysis on the target vehicle to obtain corresponding analysis data. Based on the analysis data, it constructs a hierarchical decomposition architecture corresponding to the target vehicle. Based on the hierarchical decomposition architecture, preset global optimization algorithm, local optimization algorithm and Transformer network, it establishes a road noise machine learning prediction model and trains the road noise machine learning prediction model.

[0069] The test module 200 is used to conduct real-vehicle road tests on the target vehicle to collect various vibration data and driver's right ear noise data, and to collect corresponding vibration acceleration load signals through a preset steering knuckle position sensor. The various vibration data include steering knuckle vibration data, suspension vibration data and body vibration data.

[0070] The analysis module 300 is used to input various vibration data, driver's right ear noise data and vibration acceleration load signals into the trained road noise machine learning prediction model, so as to output the multi-level road noise analysis results corresponding to the target vehicle.

[0071] Optionally, in one embodiment of this application, the modeling module 100 includes: an acquisition unit, a road noise response analysis unit, and a determination unit.

[0072] The acquisition unit is used to perform preset road noise target and influencing factor analysis on the target vehicle based on the preset transmission path analysis method and knowledge graph theory, so as to obtain the corresponding analysis data and divide the chassis system level, body system level and in-vehicle noise response level according to the analysis data.

[0073] The road noise response analysis unit is used to perform road noise response analysis on the target vehicle based on the front spoiler, front floor, front bulkhead sheet metal, rear floor, front longitudinal beam, rear longitudinal beam and rear roof to obtain the corresponding target influencing factors.

[0074] The determination unit is used to determine multiple target parameters corresponding to the chassis system level, body system level, and in-vehicle noise response level based on the target influencing factors and the preset TPA strategy, and to construct the hierarchical decomposition architecture of the target vehicle based on the multiple target parameters, target influencing factors, chassis system level, body system level, and in-vehicle noise response level.

[0075] Optionally, in one embodiment of this application, the modeling module 100 further includes a construction unit and an optimization unit.

[0076] The building unit is used to model multiple subsystems in the chassis system level and body system level of the hierarchical decomposition architecture to obtain the corresponding sub-models, and to model the body system level and in-vehicle noise response level in the hierarchical decomposition architecture to obtain the corresponding road noise vehicle model. Based on the Transformer network, sub-models and road noise vehicle model, a road noise machine learning prediction model is constructed.

[0077] The optimization unit is used to optimize the parameters of the sub-model using a global optimization algorithm and to search for the optimal parameters of the whole vehicle model using a local optimization algorithm, so as to obtain the trained road noise machine learning prediction model.

[0078] Optionally, in one embodiment of this application, the test module 200 includes a selection unit and a data acquisition unit.

[0079] The selection unit is used to select the test environment, test equipment and test conditions for the actual vehicle road test, and to conduct the actual vehicle road test on the target vehicle based on the test environment, test equipment and test conditions to collect various vibration data and driver's right ear noise data.

[0080] The acquisition unit is used to determine the arrangement principle of the steering knuckle sensor, and arrange the steering knuckle sensor at the steering knuckle position of the front and rear suspension of the target vehicle according to the arrangement principle, so as to acquire the vibration acceleration load signal of the target vehicle through the steering knuckle sensor.

[0081] It should be noted that the foregoing explanation of the multi-level road noise optimization analysis method embodiment also applies to the multi-level road noise optimization analysis device of this embodiment, and will not be repeated here.

[0082] The multi-level road noise optimization analysis device proposed in this application includes a modeling module 100, which is used to perform preset road noise target and influencing factor analysis on the target vehicle to obtain corresponding analysis data, and construct a hierarchical decomposition architecture corresponding to the target vehicle based on the analysis data. Based on the hierarchical decomposition architecture, preset global optimization algorithm, local optimization algorithm and Transformer network, a road noise machine learning prediction model is established and trained. The testing module 200 is used to conduct real-vehicle road tests on the target vehicle to collect various vibration data and driver's right ear noise data, and to collect corresponding vibration acceleration load signals through a preset steering knuckle position sensor. The various vibration data include steering knuckle vibration data, suspension vibration data and body vibration data. The analysis module 300 is used to input the various vibration data, driver's right ear noise data and vibration acceleration load signals into the trained road noise machine learning prediction model to output the multi-level road noise analysis results corresponding to the target vehicle. This application combines historical road noise data to study and optimize existing or future road noise problems, thereby effectively reducing testing and simulation costs, improving the efficiency of road noise analysis, and enabling rapid and accurate analysis of road noise problems. This helps designers evaluate the noise effects of different parameter combinations in a short time and greatly reduces human error.

[0083] Figure 9 A schematic diagram of the structure of an electronic device provided in an embodiment of this application. The electronic device may include: The memory 901, the processor 902, and the computer program stored on the memory 901 and capable of running on the processor 902.

[0084] When the processor 902 executes the program, it implements the multi-level road noise optimization analysis method provided in the above embodiments.

[0085] Furthermore, electronic devices also include: Communication interface 903 is used for communication between memory 901 and processor 902.

[0086] The memory 901 is used to store computer programs that can run on the processor 902.

[0087] The memory 901 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device.

[0088] If the memory 901, processor 902, and communication interface 903 are implemented independently, then the communication interface 903, memory 901, and processor 902 can be interconnected via a bus to complete communication between them. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of representation, Figure 9 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0089] Optionally, in a specific implementation, if the memory 901, processor 902, and communication interface 903 are integrated on a single chip, then the memory 901, processor 902, and communication interface 903 can communicate with each other through an internal interface.

[0090] The processor 902 may be a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application.

[0091] This application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the above-described multi-level road noise optimization analysis method.

[0092] This application also provides a computer program product, including a computer program, which, when executed, is used to implement the above-described multi-level road noise optimization analysis method.

[0093] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0094] 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 at least one of that feature. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0095] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or N executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0096] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0097] It should be understood that the various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, the N steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. If implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0098] Those skilled in the art will understand that all or part of the steps of the methods described in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it includes one or a combination of the steps of the method embodiments.

[0099] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0100] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application.

Claims

1. A multi-level road noise optimization analysis method, characterized by, The method comprises the following steps: performing preset road noise target and influencing factor analysis on a target vehicle to obtain corresponding analysis data, and constructing a hierarchical decomposition architecture corresponding to the target vehicle according to the analysis data, and establishing a road noise machine learning prediction model based on the hierarchical decomposition architecture, a preset global optimization algorithm, a local optimization algorithm and a Transformer network, and training the road noise machine learning prediction model; performing real vehicle road test on the target vehicle to collect corresponding multiple vibration data and driver right ear noise data, and collecting corresponding vibration acceleration load signals through a preset steering knuckle position sensor, wherein the multiple vibration data include steering knuckle vibration data, suspension vibration data and vehicle body vibration data; inputting the multiple vibration data, the driver right ear noise data and the vibration acceleration load signals into the trained road noise machine learning prediction model to output a multi-level road noise analysis result corresponding to the target vehicle.

2. The multilevel road noise optimization analysis method of claim 1, wherein, The preset road noise target and influencing factor analysis on the target vehicle to obtain corresponding analysis data, and the construction of the hierarchical decomposition architecture corresponding to the target vehicle according to the analysis data, comprises: performing preset road noise target and influencing factor analysis on the target vehicle based on a preset transfer path analysis method and knowledge graph theory to obtain corresponding analysis data, and dividing chassis system levels, vehicle body system levels and in-vehicle noise response levels according to the analysis data; performing road noise response analysis on the target vehicle based on a front deflector, a front floor, a front wall system sheet metal, a rear floor, a front longitudinal beam, a rear longitudinal beam and a rear roof of the target vehicle to obtain corresponding target influencing factors; determining multiple target parameters corresponding to the chassis system levels, the vehicle body system levels and the in-vehicle noise response levels based on the target influencing factors and a preset TPA strategy, and constructing a hierarchical decomposition architecture of the target vehicle according to the multiple target parameters, the target influencing factors, the chassis system levels, the vehicle body system levels and the in-vehicle noise response levels.

3. The multi-level road noise optimization analysis method of claim 2, wherein, The establishment of the road noise machine learning prediction model based on the hierarchical decomposition architecture, the preset global optimization algorithm, the local optimization algorithm and the Transformer network, and the training of the road noise machine learning prediction model, comprises: modeling multiple subsystems in the chassis system levels and the vehicle body system levels of the hierarchical decomposition architecture to obtain corresponding submodels, and modeling the vehicle body system levels and the in-vehicle noise response levels in the hierarchical decomposition architecture to obtain a corresponding road noise whole vehicle model, and constructing the road noise machine learning prediction model based on the Transformer network, the submodels and the road noise whole vehicle model; performing parameter optimization on the submodels through the global optimization algorithm, and performing whole vehicle model optimal parameter searching operation on the road noise whole vehicle model by using the local optimization algorithm to obtain a trained road noise machine learning prediction model.

4. The multi-level road noise optimization analysis method of claim 3, wherein, The target vehicle is subjected to real vehicle road test to collect corresponding multiple vibration data and driver right ear noise data, and a preset steering knuckle position sensor is used to collect corresponding vibration acceleration load signals, wherein the multiple vibration data include steering knuckle vibration data, suspension vibration data and vehicle body vibration data, comprising: The test environment, test equipment and test conditions of the real vehicle road test are selected, and based on the test environment, test equipment and test conditions, the target vehicle is subjected to real vehicle road test to collect corresponding multiple vibration data and driver right ear noise data; The arrangement principle of the steering knuckle sensor is determined, and the steering knuckle sensor is arranged at the front and rear suspension steering knuckle positions of the target vehicle according to the arrangement principle to collect the vibration acceleration load signals of the target vehicle through the steering knuckle sensor.

5. A multi-level road noise optimization analysis apparatus, characterized by, Comprising: The modeling module is used for preset road noise target and influencing factor analysis of the target vehicle to obtain corresponding analysis data, and based on the analysis data, the hierarchical decomposition architecture of the target vehicle is constructed, and based on the hierarchical decomposition architecture, a preset global optimization algorithm, a local optimization algorithm and a Transformer network, a road noise machine learning prediction model is established, and the road noise machine learning prediction model is trained; The test module is used for real vehicle road test of the target vehicle to collect corresponding multiple vibration data and driver right ear noise data, and a preset steering knuckle position sensor is used to collect corresponding vibration acceleration load signals, wherein the multiple vibration data include steering knuckle vibration data, suspension vibration data and vehicle body vibration data; The analysis module is used for inputting the multiple vibration data, the driver right ear noise data and the vibration acceleration load signals into the trained road noise machine learning prediction model to output the multi-level road noise analysis result corresponding to the target vehicle.

6. The multi-level road noise optimization analysis apparatus according to claim 5, wherein The modeling module comprises: The acquisition unit is used for preset road noise target and influencing factor analysis of the target vehicle based on a preset transfer path analysis method and knowledge graph theory to obtain corresponding analysis data, and to divide chassis system level, vehicle body system level and in-vehicle noise response level according to the analysis data; The road noise response analysis unit is used for road noise response analysis of the target vehicle based on the front deflector, front floor, front wall system sheet metal, rear floor, front longitudinal beam, rear longitudinal beam and rear roof of the target vehicle to obtain corresponding target influencing factors; The determination unit is used for determining multiple target parameters corresponding to the chassis system level, vehicle body system level and in-vehicle noise response level based on the target influencing factors and a preset TPA strategy, and constructing the hierarchical decomposition architecture of the target vehicle according to the multiple target parameters, the target influencing factors, the chassis system level, the vehicle body system level and the in-vehicle noise response level.

7. The multi-level road noise optimization analysis apparatus according to claim 6, wherein The modeling module comprises: A construction unit is configured to model a plurality of subsystems in the chassis system level and the body system level of the hierarchical decomposition architecture to obtain corresponding sub-models, and model the body system level and the in-vehicle noise response level in the hierarchical decomposition architecture to obtain a corresponding road noise whole vehicle model, and construct the road noise machine learning prediction model based on the Transformer network, the sub-models and the road noise whole vehicle model; An optimization unit is configured to perform parameter optimization on the sub-models by using the global optimization algorithm, and perform whole vehicle model optimal parameter searching operation on the road noise whole vehicle model by using the local optimization algorithm, to obtain a trained road noise machine learning prediction model.

8. An electronic device, comprising: Comprise: A memory, a processor and a computer program stored on the memory and executable on the processor, the processor executing the program to implement the multi-level road noise optimization analysis method according to any one of claims 1-4.

9. A computer readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor to implement the multi-level road noise optimization analysis method according to any one of claims 1-4.

10. A computer program product comprising a computer program, characterized in that, The computer program is executed to implement the multi-level road noise optimization analysis method according to any one of claims 1-4. The computer program is executed to implement the multi-level road noise optimization analysis method according to any one of claims 1-4.

Images