Pavement roughness identification method, system and device based on suspension movement and vehicle

Abstract

The invention provides a road surface roughness recognition method, system and device based on suspension movement and a vehicle, and relates to the technical field of intelligent driving, and the method comprises the steps: obtaining a vehicle speed and a suspension height, carrying out the statistical processing of a preset time window, and obtaining a multi-dimensional characteristic quantity; comprise an average vehicle speed, a suspension speed root-mean-square value, a suspension acceleration root-mean-square value, a suspension speed absolute value arithmetic average value and a suspension acceleration absolute value arithmetic average value, and the dynamic relation between suspension movement and road excitation is comprehensively reflected; the road surface roughness grade is recognized in real time by combining the rapid reasoning capacity of the classification prediction model, the damping force of the shock absorber is dynamically adjusted according to the road surface roughness grade, the real-time control requirement of the active suspension is met, real-time adaptation of the suspension system and the road surface working condition is achieved, and the smoothness and safety of vehicle driving are effectively optimized.

Description

Technical Field

[0001] The present application relates to the field of intelligent driving technology, and in particular to a method, system, device and vehicle for identifying road roughness based on suspension movement. Background Art

[0002] Road roughness, a core indicator for evaluating road quality, directly impacts vehicle ride smoothness, safety, and suspension system life. With the development of intelligent connected vehicles and active suspension technology, accurate real-time identification of road roughness is crucial for scenarios such as active suspension control, autonomous driving decision-making, and road maintenance assessment. However, existing technologies suffer from the following major drawbacks: Visual sensing-based methods require the installation of additional cameras and image processors, resulting in high hardware costs and susceptibility to environmental factors such as lighting and weather. Their high computational complexity makes them difficult to meet real-time requirements. LiDAR-based methods rely on expensive equipment, resulting in high point cloud data redundancy and susceptibility to interference from road reflections and obstructions. Unsprung motion calibration-based methods require extensive real-vehicle calibration, which is time-consuming and labor-intensive. Furthermore, due to the strong correlation between vibration intensity and vehicle speed, separate calibrations are required for different vehicle speeds, resulting in poor generalization. Summary of the Invention

[0003] The present application provides a method, system, device and vehicle for identifying road roughness based on suspension movement to solve one or more technical problems existing in the prior art and at least provide a beneficial option or create conditions.

[0004] In one aspect, the present application provides a method for identifying road roughness based on suspension motion, comprising the following steps: Acquiring the vehicle speed and suspension height in real time; the suspension height represents the vertical displacement of the suspension; Performing statistical processing on the vehicle speed and the suspension height within a preset time window to calculate vehicle speed characteristic statistics and suspension motion characteristic statistics; the preset time window is set based on a data acquisition frequency; The vehicle speed characteristic statistics include the average vehicle speed; the suspension motion characteristic statistics include the root mean square value of the suspension speed, the root mean square value of the suspension acceleration, the arithmetic mean of the absolute value of the suspension speed, and the arithmetic mean of the absolute value of the suspension acceleration; Inputting the vehicle speed characteristic statistics and the suspension motion characteristic statistics into a trained classification prediction model, and outputting a road surface roughness grade; The damping force of the vehicle's shock absorber is dynamically adjusted according to the road surface roughness level to adapt to road conditions.

[0005] Furthermore, the classification prediction model includes a multi-layer perceptron model.

[0006] Furthermore, the average vehicle speed satisfies the following calculation formula: ; in, represents the average vehicle speed, represents the vehicle speed, represents the starting time of the preset time window, Indicates the end time of the preset time window, and the data acquisition frequency is 1000 Hz; The root mean square value of the suspension speed satisfies the following calculation formula: ; in, represents the RMS value of the suspension velocity; Represents the suspension movement speed, which satisfies the following calculation formula: ; in, represents the suspension height; The root mean square value of the suspension acceleration satisfies the following calculation formula: ; in, represents the root mean square value of the suspension acceleration; Represents the acceleration of the suspension movement, which satisfies the following calculation formula: ; The arithmetic mean of the absolute value of the suspension speed satisfies the following calculation formula: ; in, represents the arithmetic mean of the absolute values ​​of the suspension velocities; The arithmetic mean of the absolute values ​​of the suspension acceleration satisfies the following calculation formula: ; in, represents the arithmetic mean of the absolute values ​​of the suspension acceleration.

[0007] Furthermore, the classification prediction model is trained by the following steps: The vehicle continuously drives at different speeds for a predetermined time on a standard sample road surface, and obtains the vehicle speed and suspension height as vehicle data; the standard sample road surface includes a smooth road surface, a rough road surface, and a bumpy road surface; Dividing the vehicle data into multiple parts according to a preset time window, and calculating the vehicle speed feature statistics and suspension motion feature statistics corresponding to each part of the vehicle data as an input feature set; Dividing the input feature set into a training set and a validation set according to a predetermined ratio, wherein the training set accounts for a larger proportion than the validation set; The classification prediction model is trained with the training set, a preset number of iterations and a target error threshold are set, and the parameters of the classification prediction model are optimized with the validation set.

[0008] Furthermore, the classification prediction model is simulated and verified by the following steps: Constructing a road surface roughness excitation model and a 1 / 4 vehicle model; the road surface roughness excitation model is used to simulate road surface excitations of different roughness levels; the 1 / 4 vehicle model is used to simulate the dynamic response of the vehicle suspension system; Calculating the vehicle speed and suspension height of the 1 / 4 vehicle model under the excitation of the road surface roughness excitation model by simulation software; Performing statistical processing on the vehicle speed and the suspension height within a preset time window to calculate vehicle speed characteristic statistics and suspension motion characteristic statistics corresponding to the 1 / 4 vehicle model; Inputting the vehicle speed characteristic statistics and the suspension motion characteristic statistics into the classification prediction model to output a simulated road surface roughness grade; Verify the matching degree between the simulated road surface roughness level and the preset road surface roughness level, and confirm that the recognition accuracy of the classification prediction model meets the preset requirements.

[0009] On the other hand, the present application provides a road roughness identification system based on suspension motion, comprising a data acquisition module, a feature calculation module, a model identification module, and a vibration reduction control module; The data acquisition module is used to obtain the vehicle speed and suspension height in real time; the suspension height represents the vertical displacement of the suspension; The feature calculation module is used to perform statistical processing on the vehicle speed and the suspension height within a preset time window to calculate vehicle speed feature statistics and suspension motion feature statistics; the preset time window is set based on the data acquisition frequency; wherein the vehicle speed feature statistics include average vehicle speed; the suspension motion feature statistics include a root mean square value of suspension velocity, a root mean square value of suspension acceleration, an arithmetic mean of an absolute value of suspension velocity, and an arithmetic mean of an absolute value of suspension acceleration; The model recognition module is used to input the vehicle speed characteristic statistics and the suspension motion characteristic statistics into a trained classification prediction model, and output a road surface roughness grade; The vibration reduction control module is used to dynamically adjust the damping force of the vehicle's shock absorber according to the road surface roughness level to adapt to road conditions.

[0010] On the other hand, the present application provides a road surface roughness identification device based on suspension movement, comprising: a speed measuring instrument, an electronically controlled shock absorber module, a processor, and a memory; The speed measuring instrument is used to obtain the vehicle speed in real time; the electronically controlled shock absorber module is used to obtain the vehicle suspension height in real time; the suspension height represents the vertical displacement of the suspension; The memory is used to store a program; when the program is executed by the processor, the processor performs statistical processing on the vehicle speed and the suspension height within a preset time window, calculates vehicle speed feature statistics and suspension motion feature statistics, inputs the calculated statistics into a trained classification prediction model, and outputs a road surface roughness grade; The electronically controlled shock absorber module is further configured to dynamically adjust the damping force of the shock absorber to adapt to road conditions according to the road roughness level output by the classification prediction model.

[0011] Furthermore, the electronically controlled shock absorber module includes a front suspension height sensor, an electronically controlled adjustment unit, a shock absorber and an actuator; The front suspension height sensor is used to obtain the suspension height of the vehicle in real time; the front suspension height sensor includes a linear displacement sensor; The electronic control adjustment unit is in communication with the processor and is configured to receive the road surface roughness level output by the processor through the classification prediction model; The electronically controlled adjustment unit is also in communication with the actuator, and dynamically controls the actuator according to the road surface roughness level output by the classification prediction model, thereby adjusting the damping force of the shock absorber to adapt to the road surface conditions.

[0012] Furthermore, after receiving the road surface roughness level, the electronically controlled adjustment unit calls a preset damping force mapping table, matches the road surface roughness level with the damping force mapping table, obtains a target damping force value, and controls the actuator to adjust the actual damping force of the shock absorber to the target damping force value.

[0013] On the other hand, the present application provides a vehicle integrated with the road surface roughness identification system based on suspension movement as described in claim 6.

[0014] The beneficial effects of the present application are as follows: the present application provides a method for identifying road roughness based on suspension motion. By obtaining vehicle speed and suspension height, and statistically processing them within a preset time window, multi-dimensional feature quantities are obtained, including average vehicle speed, root mean square value of suspension velocity, root mean square value of suspension acceleration, arithmetic mean value of absolute value of suspension velocity, and arithmetic mean value of absolute value of suspension acceleration, which comprehensively reflect the dynamic relationship between suspension motion and road excitation. Combined with the rapid reasoning capability of the classification prediction model, the road roughness level is identified in real time, and the shock absorber damping force is dynamically adjusted accordingly to meet the real-time control requirements of the active suspension, realize real-time adaptation of the suspension system to the road conditions, and effectively optimize the smoothness and safety of vehicle driving. The present application also provides corresponding systems, devices, and vehicles. The beneficial effects of the systems, devices, and vehicles are similar to those of the method and will not be elaborated here.

[0015] Other features and advantages of the present application will be described in the following description, and in part will become apparent from the description, or will be understood by practicing the present application. The purposes and other advantages of the present application can be achieved and obtained through the structures particularly pointed out in the description, claims and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The accompanying drawings are used to provide a further understanding of the technical solution of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the technical solution of the present invention and do not constitute a limitation to the technical solution of the present invention.

[0017] Figure 1 is a flow chart of a method for identifying road roughness based on suspension motion provided by this application; Figure 2 It is a training performance curve diagram of the classification prediction model provided by this application; Figure 3 This is a schematic diagram of the simulation verification results of the classification prediction model provided by this application; Figure 4 This is a structural diagram of the road surface roughness identification system based on suspension motion provided by this application; Figure 5 This is a structural diagram of the road surface roughness identification device based on suspension movement provided in this application. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0019] The present application is further described below in conjunction with the accompanying drawings and specific embodiments. The described embodiments should not be considered as limiting the present application. All other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0020] In the following description, reference is made to “some embodiments”, which describes a subset of all possible embodiments, but it will be understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application pertains. The terms used herein are for the purpose of describing the embodiments of this application only and are not intended to limit this application.

[0022] Road roughness, or the degree of unevenness in the road, is a core indicator for evaluating road quality, directly impacting vehicle ride smoothness, safety, and the lifespan of the suspension system. With the rapid development of intelligent connected vehicles and active suspension technology, the need for real-time, accurate identification of road roughness levels is becoming increasingly urgent for dynamic control of active suspensions, such as adaptive damping force to enhance comfort, autonomous driving path planning and braking decisions, and road maintenance assessments. However, existing methods suffer from high hardware costs, strong environmental dependence, and long development cycles, making them difficult to meet practical application requirements. Therefore, an efficient and low-cost identification technology is urgently needed.

[0023] Recognition methods based on visual sensing use monocular or binocular cameras to capture road images, then use image feature extraction such as texture and edge detection or deep learning models such as CNN to compare with pre-existing models of different road roughness to output a roughness level. This method is mainly used for laboratory research or road surface assessment in specific scenarios such as sunny days and dry roads. There are three problems with this method: First, the hardware cost is high, requiring the installation of additional high-precision cameras and image processors such as GPUs, which increases the hardware cost of the entire vehicle. Second, it is highly dependent on the environment, and image quality is easily affected by lighting such as backlighting and low light at night, weather such as rain, snow, fog and haze, and road pollution such as water accumulation and oil stains, resulting in feature extraction failures or increased model inference errors. Third, it lacks real-time performance. The large amount of image data requires a lot of computing power for feature extraction and model inference, making it difficult to meet the millisecond-level response requirements of scenarios such as autonomous driving.

[0024] LiDAR-based recognition methods use a LiDAR to emit laser light and receive reflected signals, generating road surface point cloud data. The roughness level is determined based on statistical characteristics of point cloud height differences, such as variance and kurtosis. These methods are primarily used for high-precision road mapping or in scenarios requiring extremely high recognition accuracy, such as autonomous driving test sites. Their drawbacks are significant: First, hardware costs are high. The unit price of a LiDAR is generally over 10,000 yuan, and mechanical LiDAR, for example, can reach tens of thousands of yuan, limiting large-scale mass production applications. Second, data redundancy is significant. The point cloud data volume is enormous, reaching hundreds of thousands to millions of points per second, requiring additional storage and processing resources and increasing system load. Third, anti-interference capabilities are weak. The laser is easily affected by road surface reflections, such as mirror reflections from wet roads, or by obstructions such as fallen leaves and gravel, resulting in missing or distorted point cloud data, affecting recognition accuracy.

[0025] The identification method based on unsprung motion calibration uses acceleration sensors to collect vibration signals from unsprung masses such as wheels and axles. Based on the correlation between vibration intensity and road excitation, a mapping relationship between vibration intensity and roughness level is established through a large number of real-vehicle calibrations at different speeds and on different road surfaces. This method is mainly used for road surface evaluation of traditional passive suspension vehicles or scenarios that are sensitive to development costs. There are some problems with this method: First, it is highly dependent on vehicle speed, and vibration intensity is strongly correlated with vehicle speed. For example, for the same roughness road surface, the vibration intensity at high speed is significantly higher than that at low speed. Separate calibration is required for each vehicle speed, making it difficult to cover all operating conditions. Second, the generalization ability is poor. The calibration results are only applicable to specific vehicle models or suspension systems. For example, the unsprung mass and suspension stiffness of different vehicle models vary greatly. Recalibration is required when reusing across different models, which reduces development efficiency.

[0026] To address these challenges, this application provides a suspension-based road roughness identification method, system, device, and vehicle. This method acquires vehicle speed and suspension height in real time and calculates five characteristic parameters within a preset time window: average vehicle speed, RMS suspension velocity, RMS suspension acceleration, arithmetic mean of absolute suspension velocity, and arithmetic mean of absolute suspension acceleration. These parameters are then input into a classification prediction model trained with multi-speed data from a standard sample road surface. The model outputs a roughness grade and dynamically adjusts the shock absorber damping force based on the results. This method offers advantages such as hardware reuse, robustness to environmental interference, efficient development, and high accuracy. The supporting system comprises four modules: data acquisition, feature calculation, model identification, and vibration damping control, operating in tandem. Through a modular design, this method implements a closed-loop "data acquisition-feature extraction-model inference-damping adjustment" process, ensuring real-time performance. The identification device integrates a speed meter, an electronically controlled shock absorber module, a processor, and memory. Through hardware integration and software collaboration, these functions are integrated into the vehicle system, resulting in high hardware reuse and low deployment costs. Vehicles integrated with this system can optimize ride smoothness and improve safety by dynamically adjusting the suspension damping force. It can also assist traffic management departments in quickly locating maintenance sections and reducing maintenance costs by collecting road surface data in real time.

[0027] First, the road surface roughness identification method based on suspension movement provided by an embodiment of the present application will be described in detail with reference to the accompanying drawings.

[0028] Reference Figure 1 The implementation process of the road roughness identification method based on suspension movement provided in the embodiment of the present application includes but is not limited to the following steps.

[0029] Step S110 , obtaining the vehicle speed and suspension height in real time.

[0030] Among them, the suspension height represents the vertical displacement of the suspension.

[0031] In step S110, the vehicle's speed and suspension height are acquired in real time. This step forms the foundation of the entire recognition method. Sensor technology is used to monitor and record the vehicle's speed and the vertical displacement of the suspension system (i.e., suspension height) in real time. This data provides essential input for subsequent analysis, enabling the system to perceive the vehicle's dynamic response under different road conditions.

[0032] Step S120 , performing statistical processing on the vehicle speed and suspension height within a preset time window, and calculating vehicle speed characteristic statistics and suspension motion characteristic statistics.

[0033] The preset time window is set based on the data acquisition frequency. Vehicle speed characteristic statistics include average vehicle speed. Suspension motion characteristic statistics include the root mean square (RMS) value of suspension velocity, the root mean square value of suspension acceleration, the arithmetic mean of the absolute value of suspension velocity, and the arithmetic mean of the absolute value of suspension acceleration.

[0034] In step S120, vehicle speed and suspension height are statistically processed within a preset time window to calculate vehicle speed and suspension motion characteristic statistics. This step processes data within a specific time window to extract key characteristic values ​​that reflect road surface roughness. The time window here is set based on the data acquisition frequency, ensuring data stability and representativeness. The calculated features include average vehicle speed, RMS suspension velocity, RMS suspension acceleration, arithmetic mean of absolute suspension velocity, and arithmetic mean of absolute suspension acceleration. These statistics effectively capture the relationship between vehicle suspension motion and road conditions.

[0035] In step S130 , the vehicle speed characteristic statistics and the suspension motion characteristic statistics are input into the trained classification prediction model, and the road surface roughness grade is obtained as output.

[0036] In step S130, the vehicle speed and suspension motion feature statistics are input into the trained classification prediction model, which outputs the road surface roughness grade. In this step, the previously calculated features are used as input to a neural network model, which has been trained with a large amount of sample data and can accurately predict the current road surface roughness grade based on the input features. This process demonstrates the value of machine learning algorithms in identifying complex road conditions and enables automated assessment of road conditions.

[0037] Step S140 : dynamically adjusting the damping force of the vehicle's shock absorber according to the road surface roughness level to adapt to the road conditions.

[0038] In step S140, the vehicle's shock absorber damping force is dynamically adjusted based on the road surface roughness level to adapt to road conditions. This final step utilizes the road surface roughness information obtained in the previous steps to optimize the driving experience and safety by adjusting the vehicle's shock absorber damping force. Specifically, when a rough road surface is detected, the system automatically increases the shock absorber damping force to reduce vehicle vibration. Conversely, if the road surface is smooth, the damping force is appropriately reduced to improve ride comfort. This approach not only improves driving safety and comfort, but also extends the service life of the vehicle's suspension system.

[0039] In some embodiments of the present application, the classification prediction model includes a multi-layer perceptron model.

[0040] A multilayer perceptron is a feedforward artificial neural network model consisting of at least three levels of nodes: an input layer, hidden layers, and an output layer. Nodes in each layer are interconnected by weights, and, except for the input nodes, each node can transform its input using a nonlinear activation function. In the application scenario of road roughness grade identification, the input layer receives data features acquired from vehicle sensors, such as vehicle speed and suspension motion statistics. The hidden layers perform complex nonlinear transformations on these features to extract higher-level abstract features. Finally, the output layer provides a predicted road roughness grade.

[0041] The advantage of using a multilayer perceptron model lies in its powerful expressiveness and flexibility, adapting to a variety of complex data patterns and relationships. Furthermore, through extensive training data and appropriate adjustments (such as setting a maximum number of iterations and a target error threshold), the multilayer perceptron model can learn the mapping between input features and road roughness levels, thereby achieving highly accurate road condition recognition. This neural network-based approach not only improves recognition accuracy but also reduces the time and labor costs of extensive on-site calibration required by traditional methods. Therefore, it has high practical value and broad application prospects.

[0042] In some embodiments of the present application, the classification prediction model adopts a multi-layer perceptron structure, which is mainly composed of an input layer, a hidden layer, and an output layer. The input layer contains 5 nodes, corresponding to the five key feature quantities: average vehicle speed, root mean square value of suspension velocity, root mean square value of suspension acceleration, arithmetic mean of absolute value of suspension velocity, and arithmetic mean of absolute value of suspension acceleration; the hidden layer contains 3 nodes (the number of nodes can be adjusted according to actual needs), each of which is fully connected to all nodes in the input layer. After weighted summation and bias calculation, a nonlinear transformation is performed through an activation function to extract a high-order abstract representation of the input features; the output layer contains 1 node, which is fully connected to all nodes in the hidden layer. The abstract features output by the hidden layer are mapped to the final road surface roughness level (such as smooth, rough, bumpy) through linear transformation, thereby achieving classification prediction of road surface roughness.

[0043] In some embodiments of the present application, the average vehicle speed satisfies the following calculation formula (1): (1); In formula (1), Indicates the average speed, Indicates vehicle speed, Indicates the start time of the preset time window, Indicates the end time of the preset time window, and the data acquisition frequency is 1000 Hz.

[0044] Formula (1) is used to calculate the average vehicle speed within a preset time window. Specifically, the vehicle speed data within the time window is integrated and divided by the total number of sampling points within the time window (because the data acquisition frequency is 1000 Hz, the total number of points is 1000 multiplied by the time window length), and the average vehicle speed within the time period is finally obtained. This calculation method can effectively smooth the fluctuation of instantaneous vehicle speed, reflect the overall driving state of the vehicle within the time period, and avoid feature distortion caused by accidental changes such as temporary acceleration or deceleration. Since the intensity of suspension movement is closely related to vehicle speed (for example, the suspension vibration is more severe when driving at high speed on the same rough road surface), accurately extracting the average vehicle speed helps the classification prediction model learn the correlation between "vehicle speed-suspension movement-road roughness", thereby more accurately distinguishing the road roughness level at different speeds (such as smooth, rough, bumpy). In addition, this formula converts the original vehicle speed time series data into quantifiable feature quantities, which are input into the model together with other features such as the root mean square value of suspension speed and the root mean square value of acceleration, providing the model with complete information reflecting the vehicle driving state and serving as an important data basis for achieving efficient and accurate identification of road roughness.

[0045] In some embodiments of the present application, the root mean square value of the suspension velocity satisfies the following calculation formula (2): (2); In formula (2), Indicates the RMS value of the suspension velocity. represents the suspension movement speed, which satisfies the following calculation formula (3): (3); In formula (3), Indicates suspension height.

[0046] Formula (2) is used to calculate the root mean square (RMS) value of the suspension velocity. By averaging the sum of the squares of the suspension velocity within a time window and then taking the square root, it quantifies the intensity of the suspension's movement within that time period. Specifically, the numerator is the sum of the squares of all suspension velocities within the time window, and the denominator is the total number of sampling points within the time window. The RMS value obtained after the square root effectively captures the energy characteristics of the suspension velocity fluctuations: large velocity changes are amplified by the square, and the final result more realistically reflects the actual intensity of the suspension movement. This characteristic quantity is directly related to the roughness of the road surface. Bumpy roads cause frequent and violent suspension movement, resulting in a higher RMS value; smooth roads cause the suspension to move more gently, resulting in a lower RMS value. Therefore, the RMS value of the suspension velocity provides a key input basis for the classification prediction model, effectively distinguishing the differences in suspension response caused by different road roughnesses, and is an important statistical feature for accurate identification of road roughness.

[0047] In some embodiments of the present application, the root mean square value of the suspension acceleration satisfies the following calculation formula (4): (4); In formula (4), Indicates the RMS value of the suspension acceleration. represents the acceleration of the suspension movement, which satisfies the following calculation formula (5): (5).

[0048] Formula (4) is used to calculate the root mean square (RMS) value of the suspension acceleration. By averaging the sum of the squares of the suspension accelerations within a time window and then taking the square root, the intensity of the suspension acceleration fluctuation within that time window is quantified. Specifically, the numerator is the sum of the squares of all suspension accelerations within the time window, and the denominator is the total number of sampling points within the time window. The RMS value obtained after the square root can effectively capture the energy characteristics of the suspension acceleration changes: violent acceleration fluctuations will be amplified by the square, and the final result more realistically reflects the degree of impact on the suspension. This characteristic quantity is directly related to the roughness of the road surface. Bumpy roads will cause the suspension to accelerate or decelerate frequently and violently, resulting in a higher RMS value; smooth roads have a smoother suspension acceleration change and a lower RMS value. Therefore, the RMS value of the suspension acceleration provides a key input basis for the classification prediction model, which can effectively distinguish the differences in suspension response caused by different road roughness and is an important statistical feature for accurate identification of road roughness.

[0049] In some embodiments of the present application, the arithmetic mean of the absolute value of the suspension velocity satisfies the following calculation formula (6): (6); In formula (6), Indicates the arithmetic mean of the absolute values ​​of the suspension velocities.

[0050] Formula (6) is used to calculate the arithmetic mean of the absolute value of the suspension velocity. By accumulating the absolute value of the suspension velocity within the time window and taking the average, the average magnitude of the suspension velocity within that time period is quantified. This calculation method eliminates the influence of velocity direction and retains only the velocity amplitude information. It can more directly reflect the continuity and overall activity of the suspension movement: if the suspension moves frequently and violently (such as on bumpy roads), the average value of the absolute value of the velocity will be higher; if the suspension moves gently (such as on smooth roads), the average value of the absolute value of the velocity will be lower. Therefore, the arithmetic mean of the absolute value of the suspension velocity provides a key statistical feature reflecting the intensity of the suspension movement for the classification prediction model. Combined with other features such as the root mean square value of the suspension velocity, it can more comprehensively describe the interaction between the suspension and the road surface, effectively improving the model's recognition accuracy for roads of different roughness.

[0051] In some embodiments of the present application, the arithmetic mean of the absolute values ​​of the suspension acceleration satisfies the following calculation formula (7): (6); In formula (7), Indicates the arithmetic mean of the absolute values ​​of suspension acceleration.

[0052] Formula (7) is used to calculate the arithmetic mean of the absolute value of the suspension acceleration. By accumulating the absolute value of the suspension acceleration within the time window and taking the average, the average magnitude of the suspension acceleration within that time period is quantified. This calculation method eliminates the influence of the acceleration direction and retains only the acceleration amplitude information. It can more intuitively reflect the duration of the impact on the suspension: if the road surface is bumpy (such as uneven), the suspension will frequently experience large accelerations or decelerations, and the average value of the absolute value of the acceleration will be higher; if the road surface is smooth, the suspension acceleration changes smoothly, and the average value is lower. Therefore, the arithmetic mean of the absolute value of the suspension acceleration provides a key statistical feature for the classification prediction model to reflect the intensity of the suspension impact. Combined with other features such as the root mean square value of the suspension acceleration, it can more comprehensively describe the interaction between the suspension and the road surface, effectively improving the model's recognition accuracy for roads of different roughness.

[0053] In some embodiments of the present application, the classification prediction model is trained through the following steps.

[0054] In step 210, the vehicle continuously drives at different speeds for a predetermined time on a standard sample road surface, and obtains the vehicle speed and suspension height as vehicle data. The standard sample road surface includes a smooth road surface, a rough road surface, and a bumpy road surface.

[0055] In step 210, the vehicle is driven continuously on a standard sample road surface at various speeds for a predetermined time, and the vehicle speed and suspension height are acquired as vehicle data. Standard sample roads include smooth, rough, and bumpy surfaces. This process is a key step in preparing training data for the classification prediction model. By driving the vehicle on representative roads with varying roughness levels and recording the corresponding speed and suspension response data, the subsequent training dataset ensures that it covers a wide range of possible road conditions, thereby improving the model's generalization capabilities.

[0056] In step 220 , the vehicle data is divided into multiple parts according to a preset time window, and the vehicle speed feature statistics and suspension motion feature statistics corresponding to each part of the vehicle data are calculated as an input feature set.

[0057] In step 220, the vehicle data is segmented into multiple data sets based on a preset time window. The corresponding vehicle speed and suspension motion feature statistics for each data set are calculated as the input feature set. This step processes and extracts features from the raw data. By analyzing data within a specific time period, key feature values ​​that reflect road conditions are extracted. This processing approach not only reduces data complexity but also makes it easier for machine learning algorithms to understand and process the data, providing high-quality input for subsequent model training.

[0058] Step 230: Divide the input feature set into a training set and a validation set according to a predetermined ratio, wherein the training set accounts for a larger proportion than the validation set.

[0059] In step 230, the input feature set is divided into a training set and a validation set according to a predetermined ratio, with the training set accounting for a larger proportion than the validation set. This division ensures that there is sufficient data to train the model while also retaining a portion for validating model performance to avoid overfitting. By properly allocating the training set and validation set ratios, a balance can be found between ensuring model accuracy and generalization ability, thereby improving the overall performance of the model.

[0060] Step 240 , train the classification prediction model with the training set, set a preset number of iterations and a target error threshold, and optimize the parameters of the classification prediction model with the validation set.

[0061] In step 240, the classification prediction model is trained using the training set, with a preset number of iterations and target error threshold set. The parameters of the classification prediction model are then optimized using the validation set. In this step, the model is trained using the previously prepared training set, while certain stopping conditions (such as a maximum number of iterations and a target error threshold) are set to ensure that the training process completes while meeting accuracy requirements. The model parameters are continuously adjusted based on the validation set's feedback, ultimately resulting in a classification prediction model that performs well on the training set and generalizes well to unseen data. This process is the core of the entire training process and directly determines the model's ability to effectively identify different road roughness levels.

[0062] In some embodiments of the present application, the training process of the classification prediction model is as follows: first, on a standard sample road surface (including three types: smooth, rough, and bumpy), let the vehicle drive continuously for at least 500 seconds at different speeds, such as 20km / h, 40km / h, 60km / h, 80km / h, 100km / h, and 120km / h, to collect stable vehicle speed and suspension height data to obtain at least 150 minutes of valid data; then, the collected data of each road surface is divided into 600 parts according to a 5-second time window, and the five feature quantities of average vehicle speed, root mean square value of suspension speed, root mean square value of suspension acceleration, arithmetic mean of absolute value of suspension speed, and arithmetic mean of absolute value of suspension acceleration are calculated for each data to form an input feature set; then, the input feature set is divided into a training set (the first 480 parts) and a validation set (the last 120 parts) in proportion; finally, the model is trained with the training set to calculate the mean square error between the predicted value and the actual value; the maximum number of iterations (such as 100 times) and the target error threshold (such as 100 times) are set. ), the model parameters are adjusted through the validation set to optimize its ability to recognize roads of different roughness, and finally a classification prediction model that can accurately output the road roughness grade is obtained.

[0063] Reference Figure 2 , Figure 2 This graph shows the training performance curves of the classification prediction model. The horizontal axis represents the number of training epochs, ranging from 0 to 28; the vertical axis shows the mean squared error (MSE), using a logarithmic scale. As can be seen, the training, validation, and test curves all approach the target MSE by epoch 28, with the validation set MSE reaching the optimal value. Initially, the MSE decreases rapidly with the number of epochs, then levels off, indicating that the model is gradually converging to a stable state.

[0064] In some embodiments of the present application, the classification prediction model is simulated and verified through the following steps.

[0065] Step 310: construct a road surface roughness excitation model and a 1 / 4 vehicle model.

[0066] The road roughness excitation model is used to simulate road excitations of varying roughness levels, while the quarter-vehicle model is used to simulate the dynamic response of the vehicle suspension system. The quarter-vehicle model is a simplified vehicle dynamics model used to simulate the vertical motion response of the vehicle suspension system under road roughness excitation. By focusing on a single-wheel suspension system (including key components such as vehicle body mass, suspension springs, shock absorbers, and wheel mass), it simplifies the complex multi-degree-of-freedom calculations of the entire vehicle while preserving core dynamic characteristics such as suspension spring compression / extension and shock absorber damping force. This allows for efficient generation of simulation data such as suspension height, velocity, and acceleration, providing reliable virtual scene input for validating classification prediction models.

[0067] In step 310, a road surface roughness excitation model and a quarter-vehicle model are constructed. This step is fundamental to simulation verification. The road surface roughness excitation model is used to simulate road conditions of varying roughness levels, while the quarter-vehicle model approximates the dynamic behavior of the vehicle's suspension system. The combination of these two models allows for a realistic reproduction of the road surface excitation experienced by the vehicle during actual driving and the resulting suspension response, providing a reliable theoretical foundation for subsequent simulation analysis.

[0068] Step 320 , using simulation software to calculate the vehicle speed and suspension height of the 1 / 4 vehicle model under the excitation of the road surface roughness excitation model.

[0069] In step 320, simulation software calculates the vehicle speed and suspension height of the quarter-vehicle model under the road roughness excitation model. This step utilizes computer simulation technology to numerically calculate the vehicle's driving process on a simulated road surface, thereby obtaining key data such as vehicle speed changes and suspension displacement. This data reflects the vehicle's dynamic characteristics under different road conditions and is an important basis for evaluating the performance of the classification prediction model.

[0070] Step 330 : Perform statistical processing on the vehicle speed and suspension height within a preset time window to calculate the vehicle speed characteristic statistics and suspension motion characteristic statistics corresponding to the 1 / 4 vehicle model.

[0071] In step 330, the vehicle speed and suspension height are statistically processed within a preset time window to calculate the vehicle speed and suspension motion characteristic statistics corresponding to the 1 / 4 vehicle model. This step aims to extract key characteristic parameters that represent the road surface conditions from the raw simulation data, including average speed, root mean square value, and arithmetic mean of absolute values. These characteristics not only simplify the data structure but also enhance the model's ability to identify road surface roughness.

[0072] In step 340 , the vehicle speed characteristic statistics and the suspension motion characteristic statistics are input into the classification prediction model, and the simulated road surface roughness grade is output.

[0073] In step 340, the vehicle speed and suspension motion feature statistics are input into the classification prediction model, which outputs the simulated road surface roughness level. In this step, the trained classification prediction model automatically determines and outputs the current road surface roughness level based on the input feature data. This process validates the model's effectiveness in a virtual environment and demonstrates its potential for application in complex road condition recognition.

[0074] Step 350 : Verify the matching degree between the simulated road surface roughness level and the preset road surface roughness level, and confirm that the recognition accuracy of the classification prediction model meets the preset requirements.

[0075] In step 350, the simulated road roughness level is verified to match the preset road roughness level, confirming that the classification prediction model's recognition accuracy meets the preset requirements. The final verification step evaluates the model's recognition accuracy by comparing the model output with a known standard. Only when the model's predictions are highly consistent with the expected results can it be considered sufficiently practical and capable of reliably performing road roughness level recognition tasks in practical applications.

[0076] Reference Figure 3 , Figure 3 This figure shows the simulation validation results of the classification prediction model, where a validation result of 1 indicates a smooth road surface; a validation result of 2 indicates a rough road surface; and a validation result of 3 indicates a bumpy road surface. The horizontal axis represents data points, ranging from 0 to 1800; the vertical axis represents the identified road surface type. The recognition results for most data points in the figure are consistent with expectations, forming continuous horizontal bands corresponding to the three road surface types. Only a few data points show fluctuations, resulting in a high overall accuracy. This chart intuitively demonstrates the reliability and practicality of the model, demonstrating its suitability for road roughness recognition tasks in real vehicles.

[0077] Secondly, refer to Figure 4 The embodiment of the present application provides a road roughness identification system based on suspension movement, including a data acquisition module 410 , a feature calculation module 420 , a model identification module 430 and a vibration reduction control module 440 .

[0078] The data acquisition module 410 is used to obtain the vehicle speed and suspension height in real time. The suspension height represents the vertical displacement of the suspension.

[0079] Feature calculation module 420 is used to perform statistical processing on vehicle speed and suspension height within a preset time window to calculate vehicle speed feature statistics and suspension motion feature statistics. The preset time window is set based on the data acquisition frequency. Vehicle speed feature statistics include average vehicle speed. Suspension motion feature statistics include the RMS value of suspension velocity, the RMS value of suspension acceleration, the arithmetic mean of the absolute value of suspension velocity, and the arithmetic mean of the absolute value of suspension acceleration.

[0080] The model identification module 430 is used to input the vehicle speed feature statistics and the suspension motion feature statistics into the trained classification prediction model, and output the road surface roughness grade.

[0081] The vibration reduction control module 440 is used to dynamically adjust the damping force of the vehicle's shock absorber according to the road roughness level to adapt to the road conditions.

[0082] Furthermore, refer to Figure 5 , an embodiment of the present application provides a road roughness identification device based on suspension movement, including: a speed measuring instrument 510, an electronically controlled shock absorber module 520, a processor 530 and a memory 540.

[0083] Speedometer 510 is used to obtain real-time vehicle speed. It is one of the information input sources for the entire system, responsible for continuously collecting current vehicle speed data during driving. As a key parameter affecting vehicle dynamic response, speed provides essential context for subsequent feature extraction and model evaluation, helping to improve the accuracy of road roughness identification.

[0084] The electronically controlled shock absorber module 520 is used to obtain the vehicle's suspension height in real time; suspension height represents the vertical displacement of the suspension. Suspension height reflects the vertical vibration experienced by the vehicle under different road conditions and is a key physical quantity for identifying road roughness. High-precision, real-time suspension height measurement by the electronically controlled shock absorber module 520 can capture subtle changes caused by road surface unevenness, providing a high-quality data foundation for classification and prediction models.

[0085] The memory 540 is used to store programs; when the program is executed by the processor 530, the processor 530 performs statistical processing on the vehicle speed and suspension height within a preset time window, calculates the vehicle speed feature statistics and the suspension motion feature statistics, and inputs them into the trained classification prediction model to output the road roughness level.

[0086] Memory 540 stores programs and data. These programs contain a series of instructions for feature extraction, model invocation, and result output. When executing these programs, processor 530 statistically analyzes vehicle speed and suspension height data over a specific time period, extracting characteristic values ​​reflecting road surface conditions. The processor then uses a pre-trained classification prediction model for inference, ultimately outputting the current road surface roughness level.

[0087] In some embodiments of the present application, the electronically controlled shock absorber module 520 includes a front suspension height sensor 521 , an electronically controlled adjustment unit 522 , a shock absorber 523 and an actuator 524 .

[0088] The electronically controlled shock absorber module 520 is also used to dynamically adjust the damping force of the shock absorber 523 to adapt to road conditions based on the road roughness level output by the classification prediction model. This function demonstrates the system's closed-loop control capabilities. Once the model identifies the corresponding road roughness level, the system automatically adjusts the damping force of the shock absorber 523 based on this level, adapting the vehicle suspension to the current road conditions. This improves ride comfort and driving stability, achieving intelligent active suspension control.

[0089] The front suspension height sensor 521 is used to obtain the vehicle's suspension height in real time. It includes a linear displacement sensor. This sensor is specifically designed to monitor changes in the vertical displacement of the front suspension. The use of a linear displacement sensor ensures high measurement accuracy and fast response. This data is directly used to construct the input feature set and forms one of the essential data elements for the classification and prediction model to accurately identify road conditions.

[0090] The electronic control unit 522 is in communication with the processor 530 and is configured to receive the road surface roughness level output by the processor 530 using the classification prediction model. As a core component of the control system, the electronic control unit 522 is responsible for receiving road surface identification results from the processor 530. Through the communication interface with the processor 530, it obtains road surface roughness information in real time, providing a basis for subsequent control decisions regarding the shock absorber 523.

[0091] The electronically controlled adjustment unit 522 is also in communication with the actuator 524. Based on the road roughness level output by the classification prediction model, it dynamically controls the actuator 524 to adjust the damping force of the shock absorber 523 to adapt to the road conditions. This function completes a complete closed-loop process from identification to control. Based on the identification results, the electronically controlled adjustment unit 522 sends a control signal to the actuator 524, causing it to adjust the damping force parameters of the shock absorber 523, thereby optimizing the vehicle's driving performance on different road surfaces.

[0092] In some embodiments of the present application, after receiving the road roughness level, the electronic control adjustment unit 522 calls a preset damping force mapping table, matches the road roughness level with the damping force mapping table, obtains a target damping force value, and controls the actuator 524 to adjust the actual damping force of the shock absorber 523 to the target damping force value.

[0093] To simplify control logic and improve response efficiency, the system incorporates a damping force mapping table that predefines the optimal damping force values ​​for different roughness levels. The electronic control adjustment unit 522 quickly determines the target damping force through table lookup and drives the actuator 524 to complete the adjustment, ensuring stable and efficient adaptive control under complex and changing road conditions.

[0094] In summary, the device integrates multiple sensors, processors, memories, and actuators to build a complete closed-loop system for road surface recognition and active control, enabling intelligent perception of road roughness and dynamic adjustment of the vehicle suspension, thereby improving the driving comfort and safety of the entire vehicle.

[0095] In addition, an embodiment of the present application also provides a vehicle that integrates the aforementioned road roughness identification system based on suspension movement.

[0096] In summary, the suspension motion-based road roughness identification method, system, device, and vehicle provided in the embodiments of the present application have the following technical effects.

[0097] This method, based on the vehicle's existing electronically controlled shock absorber system's front suspension height sensor, achieves efficient recognition through a three-stage process. First, vehicle speed and suspension height are acquired in real time. Five feature quantities—average vehicle speed, RMS suspension velocity, RMS suspension acceleration, arithmetic mean of absolute suspension velocity, and arithmetic mean of absolute suspension acceleration—are calculated within a preset time window. This comprehensively reflects the dynamic relationship between suspension motion and road excitation. Second, these feature quantities are input into a trained classification prediction model, which is trained using multi-speed data from a standard sample road surface to output a roughness grade. Finally, the shock absorber damping force is dynamically adjusted based on the recognition results. This method reuses existing vehicle sensors, avoiding expensive hardware. The multi-dimensional features based on physical signals are resistant to environmental interference. Model training decouples road testing from development, resulting in high development efficiency. Simulation verification demonstrates high accuracy and robust real-world reliability.

[0098] The suspension-based road roughness identification system consists of four collaborative modules: data acquisition, feature calculation, model identification, and vibration damping control. The data acquisition module acquires vehicle speed and suspension height in real time. The feature calculation module performs statistical processing on these parameters within a preset time window, generating five feature quantities. The model identification module integrates a trained classification prediction model and outputs a roughness level based on the feature quantities. The vibration damping control module uses a preset damping force mapping table to match the target damping value based on the roughness level and controls the shock absorber actuator to adjust the actual damping force. This modular design achieves functional decoupling and efficient collaboration, ensuring real-time identification and control.

[0099] The suspension-based road roughness identification device consists of a speedometer, an electronically controlled shock absorber module, a processor, memory, and a shock absorber actuator. The speedometer acquires vehicle speed in real time. The electronically controlled shock absorber module includes a front suspension height sensor and an electronic control unit. The processor runs feature calculation and model inference algorithms to output a roughness level. The memory stores model parameters, a damping force mapping table, and program code. The shock absorber actuator adjusts the damping force according to instructions from the electronic control unit. Through hardware integration and software collaboration, the device integrates identification and control functions into the vehicle system, ensuring high hardware reusability and low deployment costs.

[0100] Vehicles equipped with this recognition system significantly improve ride smoothness and safety by identifying road roughness in real time and dynamically adjusting suspension damping. On smooth surfaces, damping is reduced to minimize vehicle jerkiness. On rough or bumpy surfaces, damping is increased to suppress significant suspension movement. This system also provides reliable road surface information to the autonomous driving system, optimizing braking and steering strategies and enhancing safety. The on-board system also collects real-time road roughness data, assisting traffic management authorities in quickly locating sections requiring maintenance, reducing road maintenance cycles and costs.

[0101] In some optional embodiments, the functions / operations mentioned in the block diagram may not occur in the order mentioned in the operation schematic diagram. For example, depending on the functions / operations involved, the two boxes shown in succession may actually be executed substantially simultaneously or the boxes can sometimes be executed in reverse order. In addition, the embodiments presented and described in the flow chart of the present application are provided in an exemplary manner for the purpose of providing a more comprehensive understanding of the technology. The disclosed method is not limited to the operations and logical flows presented herein. Optional embodiments are contemplated in which the order of the various operations is changed and the sub-operations described as a part of a larger operation are performed independently.

[0102] In addition, although the present application is described in the context of functional modules, it should be understood that, unless otherwise stated, one or more of the functions and / or features may be integrated into a single physical device and / or software module, or one or more functions and / or features may be implemented in separate physical devices or software modules. It is also understood that a detailed discussion of the actual implementation of each module is not necessary for understanding the present application. More specifically, given the properties, functions, and internal relationships of the various functional modules in the devices disclosed herein, the actual implementation of the module will be understood within the routine skills of an engineer. Therefore, a person skilled in the art will be able to implement the present application as set forth in the claims using ordinary techniques. It is also understood that the specific concepts disclosed are merely illustrative and are not intended to limit the scope of the present application, which is determined by the full scope of the appended claims and their equivalents.

[0103] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or the portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several programs that enable a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the various embodiments of the method of the present invention. The aforementioned storage medium includes various media that can store program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0104] The logic and / or steps represented in a flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable programs for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, a program execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can retrieve and execute a program from a program execution system, apparatus, or device). For purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, a program execution system, apparatus, or device.

[0105] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic devices), a portable computer disk cartridge (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting, or processing it in a suitable manner as necessary, and then storing it in a computer memory.

[0106] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable program execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used: a discrete logic circuit having logic gate circuits for implementing logic functions on data signals, an application-specific integrated circuit having suitable combinational logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0107] In the above description of this specification, reference to the terms "one embodiment / implementation," "another embodiment / implementation," or "certain embodiments / implementations" means that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in the embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0108] Although the embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the claims and their equivalents.

[0109] The above is a specific description of the preferred implementation of the present invention, but the present invention is not limited to the embodiments. Those skilled in the art can make various equivalent modifications or substitutions without violating the spirit of the present invention. These equivalent modifications or substitutions are all included in the scope defined by the claims of the present invention.

Claims

1. A road roughness identification method based on suspension motion, characterized in that: The following steps are involved: Acquiring the vehicle speed and suspension height in real time; the suspension height represents the vertical displacement of the suspension; Performing statistical processing on the vehicle speed and the suspension height within a preset time window to calculate vehicle speed characteristic statistics and suspension motion characteristic statistics; The preset time window is set based on the data collection frequency; The vehicle speed characteristic statistics include the average vehicle speed; the suspension motion characteristic statistics include the root mean square value of the suspension speed, the root mean square value of the suspension acceleration, the arithmetic mean of the absolute value of the suspension speed, and the arithmetic mean of the absolute value of the suspension acceleration; Inputting the vehicle speed characteristic statistics and the suspension motion characteristic statistics into a trained classification prediction model, and outputting a road surface roughness grade; The damping force of the vehicle's shock absorber is dynamically adjusted according to the road surface roughness level to adapt to road conditions.

2. The method for identifying road roughness based on suspension motion according to claim 1, characterized in that: The classification prediction model includes a multi-layer perceptron model.

3. The road surface roughness identification method based on suspension motion according to claim 1, characterized in that: The average vehicle speed satisfies the following calculation formula: ； in, represents the average vehicle speed, represents the vehicle speed, represents the starting time of the preset time window, Indicates the end time of the preset time window, and the data acquisition frequency is 1000 Hz; The root mean square value of the suspension speed satisfies the following calculation formula: ； in, represents the RMS value of the suspension velocity; Represents the suspension movement speed, which satisfies the following calculation formula: ； in, represents the suspension height; The root mean square value of the suspension acceleration satisfies the following calculation formula: ； in, represents the root mean square value of the suspension acceleration; Represents the acceleration of the suspension movement, which satisfies the following calculation formula: ； The arithmetic mean of the absolute value of the suspension speed satisfies the following calculation formula: ； in, represents the arithmetic mean of the absolute values ​​of the suspension velocities; The arithmetic mean of the absolute values ​​of the suspension acceleration satisfies the following calculation formula: ； in, represents the arithmetic mean of the absolute values ​​of the suspension acceleration.

4. The method for identifying road roughness based on suspension motion according to claim 1, characterized in that: The classification prediction model is trained by the following steps: The vehicle continuously drives at different speeds for a predetermined time on a standard sample road surface, and obtains the vehicle speed and suspension height as vehicle data; the standard sample road surface includes a smooth road surface, a rough road surface, and a bumpy road surface; Dividing the vehicle data into multiple parts according to a preset time window, and calculating the vehicle speed feature statistics and suspension motion feature statistics corresponding to each part of the vehicle data as an input feature set; Dividing the input feature set into a training set and a validation set according to a predetermined ratio, wherein the training set accounts for a larger proportion than the validation set; The classification prediction model is trained with the training set, a preset number of iterations and a target error threshold are set, and the parameters of the classification prediction model are optimized with the validation set.

5. The method for identifying road roughness based on suspension motion according to claim 1, characterized in that: The classification prediction model is simulated and verified by the following steps: Constructing a road surface roughness excitation model and a 1 / 4 vehicle model; the road surface roughness excitation model is used to simulate road surface excitations of different roughness levels; the 1 / 4 vehicle model is used to simulate the dynamic response of the vehicle suspension system; Calculating the vehicle speed and suspension height of the 1 / 4 vehicle model under the excitation of the road surface roughness excitation model by simulation software; Performing statistical processing on the vehicle speed and the suspension height within a preset time window to calculate vehicle speed characteristic statistics and suspension motion characteristic statistics corresponding to the 1 / 4 vehicle model; Inputting the vehicle speed characteristic statistics and the suspension motion characteristic statistics into the classification prediction model to output a simulated road surface roughness grade; Verify the matching degree between the simulated road surface roughness level and the preset road surface roughness level, and confirm that the recognition accuracy of the classification prediction model meets the preset requirements.

6. The road roughness recognition system based on suspension movement is characterized by: It includes data acquisition module, feature calculation module, model identification module and vibration reduction control module; The data acquisition module is used to obtain the vehicle speed and suspension height in real time; the suspension height represents the vertical displacement of the suspension; The feature calculation module is used to perform statistical processing on the vehicle speed and the suspension height within a preset time window to calculate vehicle speed feature statistics and suspension motion feature statistics; The preset time window is set based on the data acquisition frequency; wherein the vehicle speed characteristic statistics include the average vehicle speed; the suspension motion characteristic statistics include the root mean square value of the suspension velocity, the root mean square value of the suspension acceleration, the arithmetic mean of the absolute value of the suspension velocity, and the arithmetic mean of the absolute value of the suspension acceleration; The model recognition module is used to input the vehicle speed characteristic statistics and the suspension motion characteristic statistics into a trained classification prediction model, and output a road surface roughness grade; The vibration reduction control module is used to dynamically adjust the damping force of the vehicle's shock absorber according to the road surface roughness level to adapt to road conditions.

7. A road surface roughness identification device based on suspension movement, characterized in that: include: Speed ​​measuring instrument, electronically controlled shock absorber module, processor and memory; The speed measuring instrument is used to obtain the vehicle speed in real time; the electronically controlled shock absorber module is used to obtain the suspension height of the vehicle in real time; The suspension height represents the vertical displacement of the suspension; The memory is used to store a program; when the program is executed by the processor, the processor performs statistical processing on the vehicle speed and the suspension height within a preset time window, calculates vehicle speed feature statistics and suspension motion feature statistics, inputs the calculated statistics into a trained classification prediction model, and outputs a road surface roughness grade; The electronically controlled shock absorber module is further configured to dynamically adjust the damping force of the shock absorber to adapt to road conditions according to the road roughness level output by the classification prediction model.

8. The road surface roughness identification device based on suspension movement according to claim 7, characterized in that: The electronically controlled shock absorber module includes a front suspension height sensor, an electronically controlled adjustment unit, a shock absorber and an actuator; The front suspension height sensor is used to obtain the suspension height of the vehicle in real time; the front suspension height sensor includes a linear displacement sensor; The electronic control adjustment unit is in communication with the processor and is configured to receive the road surface roughness level output by the processor through the classification prediction model; The electronically controlled adjustment unit is also in communication with the actuator, and dynamically controls the actuator according to the road surface roughness level output by the classification prediction model, thereby adjusting the damping force of the shock absorber to adapt to the road surface conditions.

9. The road surface roughness identification device based on suspension movement according to claim 8, characterized in that: After receiving the road surface roughness level, the electronically controlled adjustment unit calls a preset damping force mapping table, matches the road surface roughness level with the damping force mapping table, obtains a target damping force value, and controls the actuator to adjust the actual damping force of the shock absorber to the target damping force value.

10. A vehicle, characterized in that: The road surface roughness identification system based on suspension movement as claimed in claim 6 is integrated.

Images