A method and system for identifying energy absorption working conditions of a speed bump

Abstract

The application provides a speed bump energy absorption working condition recognition method and system, and relates to the technical field of intelligent transportation systems.The method collects vehicle weight, speed and vertical acceleration data in real time through a laser radar and a load cell, constructs a full-quantity feature data set, first calculates kinetic energy change and vertical impact energy, and after the cumulative energy reaches a threshold, stores the speed and acceleration data before and after the speed bump according to vehicle weight, performs Fourier transform on the acceleration data, extracts the low-frequency energy proportion, and constructs a quadratic regression model combined with the speed before and after the speed bump to predict the speed after passing through.The energy proportion abnormal threshold is set by statistical stratification of the mean and standard deviation of the energy proportion database, and when the low-frequency energy proportion of a new vehicle is lower than the energy proportion abnormal threshold and the speed prediction error is greater than or equal to 15%, it is determined that the energy absorption structure fails, and the dynamic monitoring and accurate diagnosis of the health state of the speed bump are realized.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation systems, specifically to a method and system for identifying the energy absorption conditions of speed bumps. Background Technology

[0002] Speed ​​bumps are road safety facilities that force vehicles to slow down through physical structures. They are widely used in urban roads, around schools, residential areas, and highway entrances and exits. Their core function is to reduce vehicle speed by absorbing energy and attenuating vibrations. When a vehicle drives onto a speed bump, the wheels come into contact with it, and the vehicle's kinetic energy is gradually consumed in the process, reducing the risk of traffic accidents and protecting road and bridge structures.

[0003] Traditional health monitoring of speed bump energy-absorbing structures relies mainly on manual inspections or periodic maintenance, which suffers from slow response times and high costs. It also struggles to promptly identify fatigue failure of energy-absorbing materials or structural damage, leading to decreased speed bump cushioning performance, increased vehicle bump risk, and road safety hazards. Existing technologies lack real-time correlation analysis between vehicle dynamic parameters and the mechanical response of speed bumps, making it impossible to infer the state of the energy-absorbing structure from the motion characteristics of vehicles passing over it. This is particularly problematic in mixed-vehicle traffic scenarios, where it's difficult to distinguish the different impacts of various loads on the structure. Furthermore, existing methods lack sufficient research on the frequency domain characteristics and energy distribution of vibration signals, failing to effectively combine time-frequency domain data to establish failure criteria, thus limiting the accuracy of anomaly detection. Therefore, there is an urgent need for an intelligent identification method based on vehicle dynamic data fusion analysis, which, through multi-dimensional feature modeling and hierarchical threshold determination, can achieve real-time and accurate monitoring and failure early warning of speed bump energy-absorbing conditions.

[0004] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for identifying the energy absorption condition of speed bumps, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] A method for identifying energy absorption conditions at speed bumps, comprising the following steps:

[0008] Step 1: Install a weighing sensor in front of the speed bump to obtain the vehicle weight, use LiDAR to obtain the vehicle's speed and acceleration data before and after the speed bump, and construct a full feature dataset by associating and matching the vehicle's weight, speed and acceleration data before and after the speed bump.

[0009] Step 2: Extract the vehicle weight, speed before and after the speed bump, and acceleration data from the full feature dataset to calculate the cumulative impact energy. Once the cumulative impact energy reaches the preset impact energy threshold, classify different vehicle models according to their weight and construct a hierarchical feature database for the speed before and after the speed bump data of different vehicle models.

[0010] Step 3: The hierarchical feature database is updated weekly. Fourier transform is performed on the acceleration data to calculate the proportion of low-frequency energy. The speed before the speed bump and the corresponding speed after the speed bump within the preset range are extracted to build a regression model and output the predicted speed after the speed bump.

[0011] Step 4: Calculate the mean and standard deviation of the low-frequency energy ratio for each layer of the feature database. Calculate the energy ratio anomaly threshold based on the mean and standard deviation. If the low-frequency energy ratio of a new vehicle is lower than the energy ratio anomaly threshold, and the error between the speed after the speed bump and the regression model is higher than the error threshold, the speed bump energy absorption structure is deemed to have failed.

[0012] Furthermore, in the speed and acceleration data of the vehicle before and after the speed bump obtained using lidar, the acceleration data is obtained at a sampling frequency f. s Vertical acceleration a with timestamps obtained at 200Hz z t0 is the initial time before the vehicle touches the speed bump, and t is the final time after the vehicle passes the speed bump.

[0013] Furthermore, the method for calculating the cumulative impact energy by extracting the vehicle's weight, speed before and after the speed bump, and acceleration data from the full feature dataset is as follows:

[0014] First, extract the vehicle's weight and speed before and after the speed bump from the full feature dataset, and calculate the change in kinetic energy:

[0015]

[0016] In the formula, m represents the weight of the vehicle in the full feature dataset, v1 represents the speed before the speed bump, v2 represents the speed after the speed bump, and ΔKE represents the change in kinetic energy.

[0017] Then extract the acceleration data from the full feature dataset and calculate the vertical velocity:

[0018]

[0019] In the formula, v z (t) represents the vertical velocity of the vehicle as it passes over the speed bump, v z0 This represents the initial vertical velocity, with a value of 0, a z (τ) represents the vertical acceleration at time τ;

[0020] Calculate the impact energy in the vertical direction:

[0021] E z =∫(m·α) z ·v z )dt

[0022] In the formula, E z Represented as impact energy in the vertical direction;

[0023] Calculate the impact energy based on the vertical impact energy and kinetic energy change:

[0024] E b =E z +ΔKE

[0025] In the formula, E b Represented as impact energy;

[0026] The cumulative impact energy E is obtained by summing the impact energies of all vehicles in the full feature dataset. total .

[0027] Furthermore, after the cumulative impact energy reaches a preset impact energy threshold, the method for classifying different vehicle models according to vehicle weight is as follows:

[0028] Preset impact energy threshold Th E =100000, when the cumulative impact energy E total ≥Th E When the weight is m≤1500kg, vehicles are classified as light vehicles, vehicles with 1500kg<m≤3500kg are classified as medium vehicles, and vehicles with m>3500kg are classified as heavy vehicles.

[0029] Furthermore, the method for calculating the proportion of low-frequency energy by performing a Fourier transform on the acceleration data is as follows:

[0030] Extract the vertical acceleration data sequence from the initial time t0 before the vehicle contacts the speed bump to the final time t after the vehicle passes the speed bump {a z Calculate the length of the vertical acceleration data sequence (t)}:

[0031] N = (t - t0) × f s

[0032] In the formula, N represents the length of the vertical acceleration data sequence;

[0033] For the extracted vertical acceleration data sequence {a z Apply Hanning window:

[0034] {a z window (t)}={a z(t)×ω(t)}

[0035] In the formula, {a z window Let ω(t) represent the windowed vertical acceleration data sequence, and let ω(t) represent the Hanning window function. Where t = 0, 1, ..., N-1;

[0036] The windowed vertical acceleration data sequence is padded with zeros to the nearest power of 2 to obtain a′. z window (t), processed using Fast Fourier Transform:

[0037]

[0038] In the formula, A′ z (k) represents the frequency domain complex sequence of vertical acceleration data, padded with zeros to the nearest power of 2, where j represents the imaginary unit. It is represented as a complex exponential basis function, where k = 0, 1, ..., N′-1, and N′ represents the length of the vertical acceleration data sequence padded with zeros to the nearest power of 2;

[0039] Then, from A′ z Extract the first N points from (k):

[0040] A z (k)=A′ z (k) (k=1,2,...,N-1)

[0041] In the formula, A z (k) represents a frequency domain complex sequence of vertical acceleration data;

[0042] Define the low-frequency range as: 0≤f≤10Hz, and calculate the energy of each frequency component:

[0043] E(k)=|A z (k)| 2

[0044] In the formula, E(k) represents the energy of each frequency component;

[0045] Calculate the low-frequency range index:

[0046]

[0047] In the formula, k low Represented as a low-frequency range index;

[0048] Calculate the proportion of low-frequency energy:

[0049]

[0050] In the formula, Plow Expressed as the proportion of low-frequency energy.

[0051] Furthermore, the method of extracting the speed before the speed bump within a preset range interval and the corresponding speed after the speed bump to construct a regression model and output the predicted speed after the speed bump is as follows:

[0052] The preset range interval of the speed before the speed bump is: [c, d], where 0 < c < d. The following processing is performed on the hierarchical feature database: Extract the speed before the speed bump within the range interval [c, d] and the corresponding speed after the speed bump, and construct a quadratic regression model:

[0053]

[0054] In the formula, represents the predicted speed after the speed bump at the h-th layer, β0 (h) , β1 (h) , β2 (h) represents the regression coefficient at the h-th layer, where h = 1, 2, 3, corresponding to the light vehicle feature database, medium vehicle feature database, and heavy vehicle feature database respectively;

[0055] The least squares method is used to estimate the regression coefficients, and the loss function is:

[0056]

[0057] In the formula, L( h) (β0 (h) , β1 (h) , β2 (h) ) represents the loss function at the h-th layer, and find β0 that minimizes the loss function (h) , β1 (h) , β2 (h) , where v 1i (h) represents the speed before the speed bump of the i-th vehicle in the h-th layer feature database, v 2i (h) represents the speed after the speed bump of the i-th vehicle in the h-th layer feature database, where i represents the sample index, i = 1, 2,..., n h , n h represents the total number of vehicles in the h-th layer feature database.

[0058] Furthermore, for the hierarchical feature database, calculate the mean and standard deviation of the low-frequency energy proportion respectively. The method of calculating the energy proportion anomaly threshold based on the mean and standard deviation is:

[0059] For the h-th layer feature database, extract the low-frequency energy proportion of all vehicles and calculate the mean of the low-frequency energy proportion:

[0060]

[0061] In the formula, Let μ represent the low-frequency energy percentage of the i-th vehicle in the h-th layer feature database. h Let be the mean of the feature database of the h-th layer;

[0062] Calculate the standard deviation of the low-frequency energy proportion:

[0063]

[0064] In the formula, σ h It is represented by the standard deviation of the proportion of low-frequency energy in the h-th layer feature database;

[0065] The energy proportion anomaly threshold is calculated based on the mean and standard deviation of the low-frequency energy proportion in the h-th layer feature database:

[0066]

[0067] In the formula, This is represented as the abnormal threshold for energy proportion in the h-th layer feature database.

[0068] Furthermore, when the low-frequency energy ratio of the new vehicle is lower than the abnormal energy ratio threshold, and the error between the speed after the speed bump and the regression model is higher than the error threshold, the method for determining the failure of the speed bump energy-absorbing structure is as follows:

[0069] Set error threshold Th error =15%, for new vehicles, the calculated low-frequency energy percentage is 15%. Determine the layer h to which it belongs based on its weight, and calculate the velocity prediction error:

[0070]

[0071] In the formula, Error new This is expressed as the speed prediction error. This represents the vehicle's speed after hitting the new speed bump;

[0072] when Error new ≥Th error When this happens, the speed bump's energy-absorbing structure is deemed to have failed.

[0073] Additionally, a speed bump energy absorption condition identification system is provided. This system is used to execute the aforementioned speed bump energy absorption condition identification method, including:

[0074] The data acquisition module is used to install a weighing sensor in front of the speed bump to obtain the vehicle weight, and use LiDAR to obtain the vehicle's speed and acceleration data before and after the speed bump. The vehicle's weight, speed and acceleration data before and after the speed bump are correlated and matched to construct a full feature dataset.

[0075] The vehicle model stratification module is used to extract the weight, speed before and after the speed bump, and acceleration data of the vehicles in the full feature dataset to calculate the cumulative impact energy. After the cumulative impact energy reaches the preset impact energy threshold, different vehicle models are divided according to the vehicle weight, and a stratified feature database is constructed for the speed before and after the speed bump of different vehicle models.

[0076] The feature analysis module is used to update the hierarchical feature database weekly, perform Fourier transform on acceleration data, calculate the proportion of low-frequency energy, extract the speed before the speed bump and the corresponding speed after the speed bump within a preset range, construct a regression model, and output the predicted speed after the speed bump.

[0077] The anomaly detection module is used to calculate the mean and standard deviation of the low-frequency energy ratio for the hierarchical feature database, and calculate the energy ratio anomaly threshold based on the mean and standard deviation. When the low-frequency energy ratio of a new vehicle is lower than the energy ratio anomaly threshold, and the error between the speed after the speed bump and the regression model is higher than the error threshold, the speed bump energy absorption structure is determined to be faulty.

[0078] Compared with the prior art, the beneficial effects of the present invention are:

[0079] This invention combines the cumulative impact energy threshold with the vehicle weight to dynamically divide the vehicle model into hierarchical feature libraries, independently analyzes energy absorption characteristics, avoids the limitations of single parameters and static thresholds, and improves the accuracy of energy absorption condition identification. This invention also updates the hierarchical database weekly, performs Fourier transform on the acceleration data of new vehicles to extract the proportion of low-frequency energy, and calculates the speed prediction error by combining the output results of the regression model. The dual criteria enable early warning of energy absorption structure failure. Attached Figure Description

[0080] Figure 1 This is a schematic diagram of the overall method flow of the present invention;

[0081] Figure 2 This is a summary diagram of the cumulative impact energy of the present invention;

[0082] Figure 3 This is a diagram for determining the energy absorption conditions of a speed bump according to the present invention.

[0083] Figure 4 This is a schematic diagram of the overall system modules of the present invention. Detailed Implementation

[0084] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0085] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0086] Example:

[0087] Please see Figures 1 to 3 The present invention provides a technical solution:

[0088] A method for identifying energy absorption conditions at speed bumps, comprising the following steps:

[0089] Step 1: Install a weighing sensor in front of the speed bump to obtain the vehicle weight, use LiDAR to obtain the vehicle's speed and acceleration data before and after the speed bump, and construct a full feature dataset by associating and matching the vehicle's weight, speed and acceleration data before and after the speed bump.

[0090] A quartz pressure sensor, HBM C16A, is installed 10 meters before the speed bump. When a vehicle passes through the weighing area, the pressure sensor outputs a voltage signal, which is converted into the vehicle's weight data. A lidar is mounted on a gantry, with its detection range aligned with the center point of the speed bump, and a sampling frequency f is set. s =200Hz, assign a unique ID to each vehicle, obtain the speed and acceleration data of each vehicle before and after the speed bump, and provide a synchronization clock for the weighing sensor and lidar. The acceleration data is the vertical acceleration a with a timestamp obtained below. z The initial time t0 before the vehicle touches the speed bump and the final time t after the vehicle passes the speed bump are used to construct a full feature dataset by associating and matching the vehicle's unique ID, weight, speed before and after the speed bump, and acceleration data.

[0091] Step 2: Extract the vehicle weight, speed before and after the speed bump, and acceleration data from the full feature dataset to calculate the cumulative impact energy. Once the cumulative impact energy reaches the preset impact energy threshold, classify different vehicle models according to their weight and construct a hierarchical feature database for the speed before and after the speed bump data of different vehicle models.

[0092] First, extract the vehicle's weight and speed before and after the speed bump from the full feature dataset, and calculate the change in kinetic energy:

[0093]

[0094] In the formula, m represents the weight of the vehicle in the full feature dataset, in kg; v1 represents the speed before the speed bump; v2 represents the speed after the speed bump, in m / s; ΔKE represents the change in kinetic energy, reflecting the energy loss of the vehicle when passing over the speed bump. The absolute value ensures that the absolute value is calculated regardless of whether the speed increases or decreases. The greater the vehicle mass, the greater the change in kinetic energy; the greater the speed difference, the greater the change in kinetic energy. ΔKE is related to m, Positively correlated with Negative correlation;

[0095] Then extract the acceleration data from the full feature dataset and calculate the vertical velocity:

[0096]

[0097] In the formula, v z (t) represents the vertical velocity of the vehicle when it passes over the speed bump, reflecting the dynamic response of the suspension system to vertical vibration. z0 This represents the initial vertical velocity, with a value of 0, a z (τ) represents the vertical acceleration at time τ, which is directly obtained from the lidar data. τ is the integral variable, representing any time between t and t0. The specific calculation process is as follows:

[0098] Calculate the time step:

[0099]

[0100] In the formula, Δt represents the time step, with the unit being ms. Where j = 1, 2, ..., n1-1, represents the index of the discrete-time series, corresponding to the sequence number of the vertical acceleration data point acquired by the lidar, i.e. The vertical velocity was calculated using the trapezoidal integral method.

[0101]

[0102] In the formula, Let the vertical velocity of the j-th acceleration data point be denoted as . Let the vertical acceleration be represented by the j-th acceleration data point. Let the vertical acceleration be represented by the (j-1)th acceleration data point. The final vertical velocity can be obtained using this formula. That is, v z (t);

[0103] Calculate the impact energy in the vertical direction:

[0104] E z =∫(m·a z ·v z )dt

[0105] In the formula, E z Expressed as vertical impact energy, it reflects the energy absorbed by the suspension system and can quantify the damping ability of speed bumps on vertical vibrations. The greater the vehicle weight, the greater the impact energy; the greater the acceleration or velocity, the greater the impact energy. These three factors collectively affect the energy integral result, E. z With m·a z ·v z All showed a positive correlation;

[0106] The total energy is a linear superposition of the kinetic energy change and the vertical energy absorption. Therefore, the impact energy is calculated based on the vertical impact energy and the kinetic energy change.

[0107] E b =E z +ΔKE

[0108] In the formula, E b Expressed as impact energy, it comprehensively reflects the total energy loss of a vehicle passing over a speed bump and assesses the impact load borne by the speed bump, E b With E z Both ΔKE and ΔKE are positively correlated;

[0109] The cumulative impact energy E is obtained by summing the impact energies of all vehicles in the full feature dataset. total The energy absorption capacity of speed bumps gradually decreases with the number of uses and the accumulation of impact energy. The accumulated impact energy directly quantifies its fatigue loss. An impact energy threshold Th is preset. E The numerical value needs to take into account the frequency of vehicle traffic in actual applications. The corresponding monitoring period is approximately one month, ensuring that the number of vehicle passes is between 500 and 1000. For example, in mixed urban and suburban road sections, an impact energy threshold Th can be set. E =150,000,000 J. If it is at a highway entrance with a large number of heavy vehicles, the impact energy threshold Th can be increased further based on experience. EThis ensures that the number of times a vehicle passes through the speed bump is maintained at 500-1000 times, which is more in line with actual traffic conditions. It also avoids the impact of data sparsity on subsequent model training. Sufficient data limits the impact range of new data and avoids misjudgment caused by new data when the energy absorption effect of the speed bump decreases. Table 1 shows the calculation results of kinetic energy change, vertical impact energy and impact energy using the above formula after collecting the weight, speed and acceleration data of 40 sets of samples of vehicles. The data is then summarized into a table. L, M and H in vehicle type correspond to light vehicle, medium vehicle and heavy vehicle, respectively.

[0110]

[0111] Table 1 Cumulative Impact Energy Data

[0112] like Figure 2 As shown, the impact energy E of the 40 sample vehicles tota; Reaching 4905485J satisfies the design requirement of 500-1000 accumulations. Only when the accumulated energy is large enough can the data volume be sufficient to achieve the purpose of vehicle-specific statistics, avoiding model bias due to insufficient samples. When E total ≥Th E At that time, vehicles with m≤1500kg were classified as light vehicles, vehicles with 1500kg<m≤3500kg were classified as medium vehicles, and vehicles with m>3500kg were classified as heavy vehicles. The classification range is consistent with the internationally accepted commercial vehicle classification standard. The greater the mass, the more quadratically the impact energy increases at the same speed. A hierarchical feature database was constructed based on the speed and acceleration data before and after the speed bump for different vehicle models. By classifying and statistically analyzing the impact energy distribution of different vehicle models, the prediction of the deceleration effect of the speed bump is more objective, and the degree of influence of different vehicle models on the speed bump can be distinguished.

[0113] Step 3: The hierarchical feature database is updated weekly. Fourier transform is performed on the acceleration data to calculate the proportion of low-frequency energy. The speed before the speed bump and the corresponding speed after the speed bump within the preset range are extracted to build a regression model and output the predicted speed after the speed bump.

[0114] The system collects relevant vehicle data from various data sources, including weighing sensors and lidar, for the new week. This data includes vehicle weight, speed before and after speed bumps, and acceleration data. The data is initially classified according to the vehicle type, namely light vehicles, medium vehicles, and heavy vehicles, and then integrated into a hierarchical feature database. Weekly data updates allow the model to learn the latest vehicle behavior patterns and speed bump energy absorption characteristics.

[0115] From the hierarchical feature database, vertical acceleration data sequences from the initial time t0 before the vehicle contacts the speed bump to the final time t after the vehicle passes the speed bump were extracted from light vehicles, medium vehicles, and heavy vehicles. z(t)}, layered processing ensures independent analysis of the dynamic characteristics of different vehicle models, improving the accuracy of frequency domain analysis in subsequent Fourier transform processes. For example, low-frequency vibrations are more significant in light vehicles, so the length of the vertical acceleration data sequence is calculated:

[0116] N = (t - t0) × f s

[0117] In the formula, N represents the length of the vertical acceleration data sequence;

[0118] Apply a Hanning window to the extracted vertical acceleration data sequence:

[0119] {a z window (t)}={a z (t)×ω(t)}

[0120] In the formula, {a z window Let ω(t) represent the windowed vertical acceleration data sequence, and let ω(t) represent the Hanning window function. Where t = 0, 1, ..., N-1, thereby suppressing spectral leakage and improving frequency domain resolution;

[0121] For the windowed vertical acceleration data sequence {a z window (t)} is padded with zeros to the nearest power of 2, allowing for frequency domain analysis of the padded sequence using an efficient Fast Fourier Transform algorithm. For example, when N = 200, to determine the required padded length, we need to find the nearest power of 2 greater than 200. This means adding 56 zeros to the end of the windowed vertical acceleration data sequence to obtain a′. z window (t), processed using Fast Fourier Transform:

[0122]

[0123] In the formula, A′ z (k) represents the frequency domain complex sequence of vertical acceleration data, padded with zeros to the nearest power of 2, where j represents the imaginary unit. It is represented as a complex exponential basis function, where k = 0, 1, ..., N′-1, N′, N′ represents the length of the vertical acceleration data sequence padded with zeros to the nearest 2, which is the power of the result.

[0124] Then, from A′ z Extract the first N points from (k):

[0125] A z (k)=A′ z (k)(k=1,2,...,N-1)

[0126] In the formula, A z(k) represents a frequency domain complex sequence of vertical acceleration data;

[0127] The low-frequency range is defined as 0 ≤ f ≤ 10 Hz. The dynamic response of the speed bump energy-absorbing structure is mainly reflected in the low-frequency range. The attenuation of the energy absorption effect, such as material aging or structural damage, will directly affect the energy distribution of low-frequency vibrations. When the energy-absorbing structure fails, the energy of low-frequency vibrations will be significantly reduced, while high-frequency vibrations are not sensitive to this. The energy of each frequency component is calculated as follows:

[0128] E(k)=|A z (k)| 2

[0129] In the formula, E(k) represents the energy of each frequency component;

[0130] Calculate the low-frequency range index:

[0131]

[0132] In the formula, k low Represented as a low-frequency range index;

[0133] Calculate the proportion of low-frequency energy:

[0134]

[0135] In the formula, P low Expressed as the proportion of low-frequency energy, speed bumps force the vehicle's suspension system to compress and rebound. Therefore, the proportion of low-frequency energy can reflect the overall vibration characteristics of the vehicle when it passes over the speed bump. The larger the proportion of low-frequency energy, the more energy is absorbed and dissipated by the suspension system.

[0136] Speed ​​bumps are designed to safely decelerate low-to-medium speed vehicles. Vehicles exceeding this speed fall outside their designed energy absorption range, exhibiting significant differences compared to normal speeds. Including this data would cause the model to learn atypical operating conditions, reducing the accuracy of predictions for the designed target operating conditions. The speed range before the speed bump is pre-defined as [c, d], determined by the desired deceleration effect, where 0 < c < d. For example, the speed range before a speed bump on urban roads could be set to [5, 20]. The following processing is performed on the hierarchical feature database: extracting the speeds before the speed bump within the [c, d] range and their corresponding speeds after the speed bump, avoiding overemphasis on a few outlier samples. The relationship between the speed after the speed bump and the speed before it is non-linear, and the regression coefficients vary greatly among different vehicle models, requiring independent training to construct a quadratic regression model.

[0137]

[0138] In the formula, Let β0 represent the predicted speed after the speed bump at layer h. (h) β1 (h) β2 (h) Let h be the regression coefficient of the h-th layer, where h = 1, 2, 3, which correspond to the light vehicle feature database, medium vehicle feature database, and heavy vehicle feature database, respectively. For example, in actual experiments, the deceleration effect of heavy vehicles passing at high speed is more significant, and the prediction will be more accurate when h = 3.

[0139] The least squares method is used to estimate the regression coefficients, where the loss function is:

[0140]

[0141] In the formula, L (h) (β0 (h) β1 (h) β2 (h) Let β be the loss function of the h-th layer. Find the β0 that minimizes the loss function. (h) β1 (h) β2 (h) , where v 1i (h) Let v represent the speed of the i-th vehicle before the speed bump in the h-th layer feature database. 2i (h) Let represent the speed of the i-th vehicle after the speed bump in the h-th layer feature database, where i represents the sample index, i = 1, 2, ..., n. h n h Let represent the total number of vehicles in the h-th layer feature database. The loss function is defined as the sum of the squared differences between the observed and predicted values, directly measuring the model's prediction bias. By minimizing this loss, the model can fit the data to the greatest extent, minimizing the difference between the predicted and actual values.

[0142] Step 4: Calculate the mean and standard deviation of the low-frequency energy ratio for the hierarchical feature database. Calculate the abnormal energy ratio threshold based on the mean and standard deviation. If the low-frequency energy ratio of a new vehicle is lower than the abnormal energy ratio threshold, and the error between the speed after the speed bump and the regression model is higher than the error threshold, the speed bump energy absorption structure is deemed to have failed.

[0143] The layered processing approach takes into account the differences between vehicle models, making the method adaptable to various types of vehicles and improving the system's versatility and practicality. For the h-th layer feature database, the low-frequency energy ratio of all vehicles is extracted, and the mean of the low-frequency energy ratio is calculated.

[0144]

[0145] In the formula, i represents the proportion of low-frequency energy of the i-th vehicle in the h-th layer feature database, μ h Let be the mean of the feature database of the h-th layer;

[0146] Calculate the standard deviation of the low-frequency energy proportion:

[0147]

[0148] In the formula, σ h It is represented by the standard deviation of the proportion of low-frequency energy in the h-th layer feature database;

[0149] The energy proportion anomaly threshold is calculated based on the mean and standard deviation of the low-frequency energy proportion in the h-th layer feature database:

[0150]

[0151] In the formula, This represents the abnormal energy percentage threshold in the h-th layer feature database, and the data typically approximates a normal distribution. In a normal distribution, approximately 95% of the data falls within the range of the mean plus or minus two standard deviations. Therefore, using the mean and standard deviation to describe the distribution characteristics of low-frequency energy percentage is reasonable and can identify data points where the low-frequency energy percentage is significantly lower than the normal level.

[0152] Set the error threshold Tg error =15%. A 15% error is widely considered a reasonable boundary for distinguishing between normal and abnormal operating conditions, and it is consistent with long-term observations of the energy absorption performance of speed bumps. Relying solely on the low-frequency energy percentage to determine whether an energy-absorbing structure has failed may lead to misjudgment. By introducing speed prediction error as another criterion, the performance of the energy-absorbing structure can be evaluated more comprehensively. If the speed prediction error also exceeds the threshold, it indicates that the energy-absorbing structure has indeed encountered a problem. For new vehicles, the calculated low-frequency energy percentage... Determine the layer (g) to which it belongs based on its weight, and calculate the velocity prediction error:

[0153]

[0154] In the formula, Error new This is expressed as the speed prediction error. The speed is expressed as the new vehicle speed after the speed bump. The greater the difference between the actual speed and the predicted speed after the speed bump, the greater the speed prediction error, showing a positive correlation. This difference can be used to assess the degree of difference between the actual deceleration effect of the speed bump and the expected model, thus serving as a condition for determining the failure of the speed bump energy-absorbing structure. Error new ≥Th errorWhen determining the failure of the speed bump energy-absorbing structure, Table 2 shows the statistics of 40 sets of low-frequency energy proportion, speed prediction error, and failure determination for the light vehicle stratified database. At this time, the mean low-frequency energy proportion μ h =74%, standard deviation σ h =4%, and the abnormal threshold at this time is calculated using the above formula. The summary will generate a data table.

[0155] Serial Number Low-frequency energy percentage (%) Speed ​​prediction error (%) Failure determination 1 75 1.6 no 2 74 1.9 no 3 76 2.1 no 4 73 3.8 no 5 77 2.4 no 6 72 5 no 7 75 1.6 no 8 74 1.9 no 9 76 6.4 no 10 71 1.9 no 11 78 2.4 no 12 73 3.3 no 13 75 3.2 no 14 74 3.9 no 15 77 4.3 no 16 70 15.1 no 17 79 7.1 no 18 72 5 no 19 75 3.2 no 20 74 5.9 no 21 76 8.5 no 22 68 20.8 no 23 78 4.8 no 24 71 6.7 no 25 75 4.8 no 26 74 9.8 no 27 77 2.1 no 28 66 30.2 no 29 79 9.5 no 30 70 10 no 31 75 6.5 no 32 74 11.8 no 33 76 12.8 no 34 65 33.9 yes 35 78 11.9 no 36 71 11.7 no 37 75 8.1 no 38 74 13.7 no 39 77 14.9 no 40 64 39.6 yes

[0156] Table 2. Determination Table for Energy Absorption Structure of Speed ​​Bumps

[0157] like Figure 3 As shown, the changes in the proportion of low-frequency energy and the speed prediction error in 40 sets of data are illustrated. The two horizontal lines represent the abnormal threshold and error threshold of the energy proportion, respectively. The proportion of low-frequency energy reflects the absorption efficiency of the speed bump's energy-absorbing structure for low-frequency vibrations. A proportion below 66% indicates a significant decrease in energy absorption efficiency. The speed prediction error reflects the deviation between the actual deceleration effect of the speed bump and the expected result. A proportion above 15% indicates abnormal deceleration performance. Relying solely on the proportion of low-frequency energy may lead to misjudgment due to vehicle abnormalities, such as suspension failure. However, when the speed prediction error is higher than 15%, such interference can be ruled out. Therefore, when both conditions are met, it can be quickly determined that the speed bump's energy-absorbing structure has failed and can no longer effectively buffer vehicle impacts.

[0158] Please see Figure 4 The present invention also provides a speed bump energy absorption condition identification system, the system being used to execute the above-described speed bump energy absorption condition identification method, comprising:

[0159] The data acquisition module is used to install a weighing sensor in front of the speed bump to obtain the vehicle weight, and use LiDAR to obtain the vehicle's speed and acceleration data before and after the speed bump. The vehicle's weight, speed and acceleration data before and after the speed bump are correlated and matched to construct a full feature dataset.

[0160] The vehicle model stratification module is used to extract the weight, speed before and after the speed bump, and acceleration data of the vehicles in the full feature dataset to calculate the cumulative impact energy. After the cumulative impact energy reaches the preset impact energy threshold, different vehicle models are divided according to the vehicle weight, and a stratified feature database is constructed for the speed before and after the speed bump of different vehicle models.

[0161] The feature analysis module is used to update the hierarchical feature database weekly, perform Fourier transform on acceleration data, calculate the proportion of low-frequency energy, extract the speed before the speed bump and the corresponding speed after the speed bump within a preset range, construct a regression model, and output the predicted speed after the speed bump.

[0162] The anomaly detection module is used to calculate the mean and standard deviation of the low-frequency energy ratio for the hierarchical feature database, and calculate the energy ratio anomaly threshold based on the mean and standard deviation. When the low-frequency energy ratio of a new vehicle is lower than the energy ratio anomaly threshold, and the error between the speed after the speed bump and the regression model is higher than the error threshold, the speed bump energy absorption structure is determined to be faulty.

[0163] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0164] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0165] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0166] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for identifying the energy absorption condition of speed bumps, characterized in that, The specific steps include: Step 1: Install a weighing sensor in front of the speed bump to obtain the vehicle weight, use LiDAR to obtain the vehicle's speed and acceleration data before and after the speed bump, and construct a full feature dataset by associating and matching the vehicle's weight, speed and acceleration data before and after the speed bump. Step 2: Extract the vehicle weight, speed before and after the speed bump, and acceleration data from the full feature dataset to calculate the cumulative impact energy. Once the cumulative impact energy reaches the preset impact energy threshold, classify different vehicle models according to their weight and construct a hierarchical feature database for the speed before and after the speed bump data of different vehicle models. Step 3: The hierarchical feature database is updated weekly. Fourier transform is performed on the acceleration data to calculate the proportion of low-frequency energy. The speed before the speed bump and the corresponding speed after the speed bump within the preset range are extracted to build a regression model and output the predicted speed after the speed bump. Step 4: Calculate the mean and standard deviation of the low-frequency energy ratio for each layer of the feature database. Calculate the energy ratio anomaly threshold based on the mean and standard deviation. If the low-frequency energy ratio of a new vehicle is lower than the energy ratio anomaly threshold, and the error between the speed after the speed bump and the regression model is higher than the error threshold, the speed bump energy absorption structure is deemed to have failed. In using lidar to acquire speed and acceleration data of a vehicle before and after a speed bump, the acceleration data is obtained at a sampling frequency of... The vertical acceleration with timestamps obtained below The initial moment before the vehicle touches the speed bump The moment the vehicle stops after passing over the speed bump ; First, extract the vehicle's weight and speed before and after the speed bump from the full feature dataset, and calculate the change in kinetic energy: In the formula, This represents the weight of the vehicles in the full feature dataset. This represents the speed before the speed bump. This represents the speed after the speed bump. Expressed as the change in kinetic energy; Then extract the acceleration data from the full feature dataset and calculate the vertical velocity: In the formula, This represents the vertical speed of the vehicle as it passes over the speed bump. Represented as the initial vertical velocity, with a value of , Represented as time Vertical acceleration; Calculate the impact energy in the vertical direction: In the formula, Represented as impact energy in the vertical direction; Calculate the impact energy based on the vertical impact energy and kinetic energy change: In the formula, Represented as impact energy; The cumulative impact energy is obtained by summing the impact energies of all vehicles in the full feature dataset. .

2. The method for identifying the energy absorption condition of a speed bump according to claim 1, characterized in that: Once the cumulative impact energy reaches a preset impact energy threshold, the method for classifying different vehicle models according to vehicle weight is as follows: Preset impact energy threshold When the cumulative impact energy At that time, The vehicles are classified as light vehicles. Vehicles are classified as medium-sized vehicles. The vehicles are classified as heavy vehicles.

3. The method for identifying the energy absorption condition of a speed bump according to claim 1, characterized in that: The method for calculating the proportion of low-frequency energy by performing a Fourier transform on the acceleration data is as follows: Capture the initial moment before the vehicle touches the speed bump. The time after the vehicle passes over the speed bump Vertical acceleration data sequence Calculate the length of the vertical acceleration data sequence: In the formula, This is expressed as the length of the vertical acceleration data sequence; The extracted vertical acceleration data sequence Apply Hanning window: In the formula, This is represented as a windowed sequence of vertical acceleration data. Represented as the Hanning window function, and ,in, ; The windowed vertical acceleration data sequence is padded with zeros to the nearest power of 2 to obtain... The process is handled using Fast Fourier Transform: In the formula, This is represented as a frequency domain complex sequence of vertical acceleration data sequences padded with zeros to the nearest power of 2. Represented as the imaginary unit, It is represented as a complex exponential basis function, where, This represents the length of the vertical acceleration data sequence padded with zeros to the power of the nearest 2; After that, from Extracting the first Points: In the formula, Represented as a frequency domain complex sequence of vertical acceleration data; The low-frequency range is defined as: Calculate the energy of each frequency component: In the formula, This represents the energy of each frequency component. Calculate the low-frequency range index: In the formula, Represented as a low-frequency range index; Calculate the proportion of low-frequency energy: In the formula, This is expressed as the proportion of low-frequency energy.

4. The method for identifying the energy absorption condition of a speed bump according to claim 1, characterized in that: The method for extracting the speed before and after a speed bump within a preset range to construct a regression model and output the predicted speed after the speed bump is as follows: The preset speed range before the speed bump is: ,in, The following processing is performed on the hierarchical feature database: extracting features from... A quadratic regression model is constructed using the speeds before and after the speed bumps within a given range. In the formula, Represented as the first Predict the speed after the speed bump. Represented as the first The regression coefficients of the layer, where, These correspond to the light vehicle feature database, medium vehicle feature database, and heavy vehicle feature database, respectively. The least squares method is used to estimate the regression coefficients, where the loss function is: In the formula, Represented as the first Layer loss function, find the one that minimizes the loss function. ,in, Represented as the first Layer feature database The speed of each vehicle before the speed bump. Represented as the first Layer feature database The speed of each vehicle after the speed bump, among which... Represented as a sample index, Represented as the first The total number of vehicles in the layer feature database.

5. The method for identifying the energy absorption condition of a speed bump according to claim 4, characterized in that: The method for calculating the mean and standard deviation of the low-frequency energy proportion for each hierarchical feature database, and then calculating the energy proportion anomaly threshold based on the mean and standard deviation, is as follows: For the first From the layer feature database, extract the low-frequency energy ratio of all vehicles and calculate the mean of the low-frequency energy ratio: In the formula, Represented as the first Layer feature database The proportion of low-frequency energy in each vehicle Represented as the first The mean of the layer feature database; Calculate the standard deviation of the low-frequency energy proportion: In the formula, Represented as the first Standard deviation of the proportion of low-frequency energy in the layer feature database; Based on the The energy proportion anomaly threshold is calculated by using the mean and standard deviation of the low-frequency energy proportion in the layer feature database. In the formula, Represented as the first Abnormal threshold for energy percentage in the layer feature database.

6. The method for identifying the energy absorption condition of a speed bump according to claim 5, characterized in that: The method for determining the failure of the speed bump energy absorption structure when the low-frequency energy ratio of a new vehicle is lower than the abnormal energy ratio threshold, and the error between the speed after the speed bump and the regression model is higher than the error threshold, is as follows: Set error threshold For the new vehicle, the proportion of low-frequency energy was calculated. Determine the layer based on its weight. Calculation speed prediction error: In the formula, This is expressed as the speed prediction error. This represents the vehicle's speed after hitting the new speed bump; when When this happens, the speed bump's energy-absorbing structure is deemed to have failed.

7. A speed bump energy absorption condition identification system, characterized in that: The system is used to execute a speed bump energy absorption condition identification method according to any one of claims 1-6, including: The data acquisition module is used to install a weighing sensor in front of the speed bump to obtain the vehicle weight, and use LiDAR to obtain the vehicle's speed and acceleration data before and after the speed bump. The vehicle's weight, speed and acceleration data before and after the speed bump are correlated and matched to construct a full feature dataset. The vehicle model stratification module is used to extract the weight, speed before and after the speed bump, and acceleration data of the vehicles in the full feature dataset to calculate the cumulative impact energy. After the cumulative impact energy reaches the preset impact energy threshold, different vehicle models are divided according to the vehicle weight, and a stratified feature database is constructed for the speed before and after the speed bump of different vehicle models. The feature analysis module is used to update the hierarchical feature database weekly, perform Fourier transform on acceleration data, calculate the proportion of low-frequency energy, extract the speed before the speed bump and the corresponding speed after the speed bump within a preset range, construct a regression model, and output the predicted speed after the speed bump. The anomaly detection module is used to calculate the mean and standard deviation of the low-frequency energy ratio for the hierarchical feature database, and calculate the energy ratio anomaly threshold based on the mean and standard deviation. When the low-frequency energy ratio of a new vehicle is lower than the energy ratio anomaly threshold, and the error between the speed after the speed bump and the regression model is higher than the error threshold, the speed bump energy absorption structure is determined to be faulty.

Images