A method and system for predicting the probability of human rib fracture in a car crash

Abstract

The application provides a method and system for predicting the probability of human rib fracture in a car crash, by collecting state data in the car crash process; wherein the state data includes vehicle operation data, safety constraint data and occupant attribute data; based on the state data, the vehicle dynamics characteristics and the occupant static attribute characteristics are calculated; based on the vehicle dynamics characteristics and the occupant static attribute characteristics, the occupant rib fracture probability is calculated; based on the fracture probability, the crash handling decision is determined; the vehicle dynamics characteristics and the occupant static attribute characteristics are calculated according to the dynamics data of the car in the crash process and the occupant attribute data, the occupant rib fracture probability is calculated in combination with the vehicle dynamics characteristics and the occupant static attribute characteristics, so as to improve the prediction accuracy, and the crash handling decision is determined according to the fracture probability, so that the crash can be quickly responded, and the safety of the occupant is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of automobile collision processing, in particular to a method and system for predicting rib fracture probability of human body in automobile collision. BACKGROUND

[0002] In vehicle collision accidents, chest injury is one of the most fatal injuries, and rib fracture is the most common form of injury. Accurate and rapid prediction of rib fracture risk is of decisive significance for the precise triggering of intelligent restraint systems (such as adaptive airbags and reversible pre-tensioners) and the priority dispatch of emergency medical rescue after the accident.

[0003] Existing occupant injury assessment techniques mainly include physical-based simulation models, experience-based statistical models or simplified injury criteria (such as chest compression volume and viscous criteria). Among them, the physical-based simulation model uses finite element method (FEM) or multi-body dynamics (MBD) to establish a high-precision human body model for simulation. This method takes a very long time to calculate (several hours to several days), and completely cannot meet the real-time requirements of millisecond-level response in the collision moment, and is only suitable for post-analysis in the vehicle design stage; the experience-based statistical model uses traditional statistical methods such as logistic regression to establish the correlation between macroscopic parameters such as collision speed and vehicle deformation and injury risk. This model is difficult to capture the complex nonlinear dynamics in the collision process, and the prediction accuracy of rib fracture, which is caused by local load, is very low, and the generalization ability is poor; the simplified injury criterion usually requires special and expensive sensors (such as chest displacement sensors), which are difficult to popularize in mass-produced vehicles, and only provides overall evaluation, and cannot realize probabilistic prediction. Therefore, there is an urgent need for a method that can accurately and quickly predict the probability of occupant rib fracture in automobile collision. SUMMARY

[0004] In order to solve the above technical problems, the present application is proposed. The embodiments of the present application provide a method and system for predicting rib fracture probability of human body in automobile collision.

[0005] According to one aspect of the present application, a method for predicting rib fracture probability of human body in automobile collision is provided, comprising: collecting state data in the automobile collision process; wherein the state data includes vehicle operation data, safety restraint data and occupant attribute data, the vehicle operation data includes collision speed change, longitudinal acceleration, lateral acceleration and original acceleration signal, the safety restraint data includes seat belt anchor point force and webbing pull-out amount ，The occupant attribute data includes occupant height, weight, body mass index, age, gender, and precise sitting posture in the vehicle. Based on the state data, vehicle dynamics characteristics and occupant static attribute characteristics are calculated. Based on the vehicle dynamics characteristics and occupant static attribute characteristics, the probability of rib fracture is calculated. Based on the fracture probability, a collision handling decision is determined. The calculation of vehicle dynamics characteristics and occupant static attribute characteristics based on the state data includes: segmenting the vehicle operation data and safety constraint data using a sliding time window to obtain multiple segmented data; extracting time-domain features and frequency-domain features for each segmented data; wherein, the time-domain features include mean, variance, skewness, kurtosis, root mean square, absolute value integral, and zero-crossing rate; and the frequency-domain features include performing a fast Fourier transform on the signal within the sliding time window and extracting 0–200... Hz The amplitude and total energy of the first few main frequencies obtained in the frequency band; the vehicle dynamics characteristics are obtained by combining the time domain characteristics and the frequency domain characteristics; the occupant static attribute characteristics are calculated based on the occupant attribute data; wherein, the occupant static attribute characteristics represent the rib impact resistance of the corresponding occupant.

[0006] In one embodiment, calculating the fracture probability of the occupant's ribs based on the vehicle dynamics features and the occupant static attribute features includes: inputting the vehicle dynamics features and the occupant static attribute features into a trained prediction model to obtain the fracture probability of the occupant's ribs.

[0007] In one embodiment, the prediction model includes a temporal feature extractor, a perceptron, and a classifier, wherein the temporal feature extractor and the perceptron are connected in parallel; wherein, the step of inputting the vehicle dynamics features and the occupant static attribute features into the trained prediction model to obtain the fracture probability of the occupant's ribs includes: inputting the vehicle dynamics features and the occupant static attribute features into the temporal feature extractor and the perceptron, respectively, to obtain spatiotemporal features and static features; and then fusing the spatiotemporal features and the static features and inputting them into the classifier to obtain the fracture probability of the occupant's ribs.

[0008] In one embodiment, the temporal feature extractor includes a two-stage one-dimensional convolutional neural network and a global average pooling layer. The first-stage one-dimensional convolutional neural network includes 64 convolutional kernels of size 5, and the second-stage one-dimensional convolutional neural network includes 32 convolutional kernels of size 10. The first-stage one-dimensional convolutional neural network, the second-stage one-dimensional convolutional neural network, and the global average pooling layer are connected in series.

[0009] In one embodiment, the perceptron includes two fully connected layers, wherein the output dimension of the first fully connected layer is 16 and the output dimension of the second fully connected layer is 8.

[0010] In one embodiment, the classifier includes a random deactivation layer and a classification layer, wherein the random deactivation layer has a dropout rate of 0.5, and the classification layer includes an activation function.

[0011] In one embodiment, determining the collision handling decision based on the fracture probability includes: if the fracture probability is greater than a preset probability threshold, then triggering an emergency response.

[0012] In one embodiment, triggering an emergency response if the probability of fracture is greater than a preset probability threshold includes: if the probability of fracture is greater than a preset first probability threshold, pre-tightening the seat belt; if the probability of fracture is greater than a preset second probability threshold, switching the airbag deployment mode and sending a warning message; wherein the second probability threshold is greater than the first probability threshold.

[0013] According to another aspect of this application, a prediction system for the probability of rib fracture in a human body during a car collision is provided, comprising: a state data acquisition module for acquiring state data during the car collision process; wherein, the state data includes vehicle operation data, safety constraint data, and occupant attribute data, the vehicle operation data including collision speed change, longitudinal acceleration, lateral acceleration, and original acceleration signal, and the safety constraint data including seat belt anchor force and webbing pull-out amount. ， The occupant attribute data includes occupant height, weight, body mass index, age, gender, and precise sitting posture in the vehicle; the collision feature calculation module is used to calculate vehicle dynamics features and occupant static attribute features based on the state data; the fracture probability calculation module is used to calculate the fracture probability of the occupant's ribs based on the vehicle dynamics features and the occupant static attribute features; the processing strategy determination module is used to determine the collision processing decision based on the fracture probability; the collision feature calculation module is further configured to: segment the vehicle operation data and the safety constraint data using a sliding time window to obtain multiple segmented data; for each segmented data, extract time-domain features and frequency-domain features; wherein, the time-domain features include mean, variance, skewness, kurtosis, root mean square, absolute value integral, and zero-crossing rate, and the frequency-domain features include performing a fast Fourier transform on the signal within the sliding time window and extracting 0–200 Hz The amplitude and total energy of the first few main frequencies obtained in the frequency band; the vehicle dynamics characteristics are obtained by combining the time domain characteristics and the frequency domain characteristics; the occupant static attribute characteristics are calculated based on the occupant attribute data; wherein, the occupant static attribute characteristics represent the rib impact resistance of the corresponding occupant.

[0014] This application provides a method and system for predicting the probability of rib fracture in a human body during a car collision. The method involves collecting state data during the car collision process, including vehicle operation data, safety constraint data, and occupant attribute data. Based on the state data, vehicle dynamics characteristics and occupant static attribute characteristics are calculated. Based on these characteristics, the probability of rib fracture is calculated. Based on the fracture probability, a collision response decision is determined. The method improves prediction accuracy by combining the vehicle dynamics characteristics and occupant static attribute data with the rib fracture probability, thereby enabling rapid collision response and improving occupant safety. Attached Figure Description

[0015] The above and other objects, features, and advantages of this application will become more apparent from the more detailed description of the embodiments of this application in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of this application and form part of the specification. They are used together with the embodiments of this application to explain this application and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same components or steps.

[0016] Figure 1 This is a flowchart illustrating a method for predicting the probability of rib fractures in a human body during a car collision, provided in an exemplary embodiment of this application.

[0017] Figure 2 This is a schematic diagram of data comparison before and after collision pulse filtering provided in an exemplary embodiment of this application.

[0018] Figure 3 This is a schematic diagram of the structure of a prediction model provided in an exemplary embodiment of this application.

[0019] Figure 4 This is a flowchart illustrating a method for predicting the probability of rib fractures in a human body during a car collision, provided in another exemplary embodiment of this application.

[0020] Figure 5 This is a schematic diagram of the structure of a system for predicting the probability of rib fractures in a human body during a car collision, provided in an exemplary embodiment of this application. Detailed Implementation

[0021] Hereinafter, exemplary embodiments according to this application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments of this application. It should be understood that this application is not limited to the exemplary embodiments described herein.

[0022] Figure 1This is a flowchart illustrating a method for predicting the probability of rib fractures in a human body during a car collision, provided in an exemplary embodiment of this application. Figure 1 As shown, the method for predicting the probability of rib fractures in a car collision includes the following steps:

[0023] Step 110: Collect state data during the car collision process.

[0024] The status data includes vehicle operation data, safety constraint data, and occupant attribute data. Specifically, the vehicle operation data includes the collision speed change Δ obtained from the CAN bus. V Longitudinal acceleration a long ( t ), lateral acceleration a lat ( t ) and from deployed in vehicles B High-frequency (greater than 10) components along key force transmission paths such as columns, door sill beams, and seat rails. kHz Raw acceleration signals acquired by a triaxial accelerometer a x ( t ), a y ( t ), a z ( t Safety constraint data includes seatbelt anchor force obtained from seat-integrated sensors. F anchor ( t ) and webbing pull-out L webbing ( t Occupant attribute data includes occupant height from in-vehicle vision systems (such as cameras) or manually entered. H ,weight W , body mass index BMI ,age Age ,gender Gender Precise seating position in the vehicle (e.g., distance from the steering wheel / dashboard) D chest )wait.

[0025] After collecting the status data, this application preprocesses the status data. The specific processing methods include:

[0026] Synchronization and time-series alignment: Based on hardware trigger signals or software timestamps, all heterogeneous state data are synchronized and interpolated to form unified time-series data.

[0027] Filtering and noise reduction: Bandpass filtering (such as the Butterworth filter with a passband frequency of 5) is used for high-frequency acceleration data. Hz -1000 Hz ), retaining the dominant frequency components related to collision and chest response, and removing high-frequency noise and low-frequency drift; such as Figure 2 As shown, the original collision pulse signal collected from the accelerometer contains high-frequency noise and glitches. After being processed by a 5Hz-1000Hz bandpass filter (Butterworth filter), the signal has a smooth waveform, retains the main frequency component of the collision energy, and removes high-frequency noise and low-frequency drift.

[0028] After synchronizing and aligning the state data, and performing filtering and noise reduction, a rib response-related feature set is further constructed, which includes the following features:

[0029] Composite acceleration: ;

[0030] velocity change curve: ;

[0031] Collision Intensity Index (CSI): The collision intensity index is directly related to energy transfer;

[0032] Chest load direction ratio: This is used to determine the main direction of the load;

[0033] Energy decay rate: Calculates the energy decay characteristics of the rising and falling segments of the collision pulse;

[0034] Occupant-specific gain factor: based on occupant weight W and sitting distance D chest Calculate the scaling factor to personalize the equivalent intensity of the input pulse.

[0035] Step 120: Based on the state data, calculate the vehicle dynamics characteristics and occupant static attribute characteristics.

[0036] This application calculates vehicle dynamics characteristics and occupant static attribute characteristics based on state data and a calculated feature set, respectively, to comprehensively determine the probability of rib fractures for different occupants and the dynamic characteristics during the current collision process. Specifically, a sliding time window is used to segment the vehicle operation data and safety constraint data, resulting in multiple segmented data. For each segmented data, time-domain and frequency-domain features are extracted. The time-domain features include mean, variance, skewness, kurtosis, root mean square, absolute value integral, and zero-crossing rate. The frequency-domain features include performing a fast Fourier transform on the signal within the sliding time window and extracting 0–200... HzThe amplitude and total energy of the first few main frequencies obtained from the frequency bands are used to obtain the vehicle dynamics characteristics based on time-domain and frequency-domain characteristics. The static attribute characteristics of the occupants are calculated based on the occupant attribute data. Among them, the static attribute characteristics of the occupants represent the rib impact resistance of the corresponding occupants.

[0037] Step 130: Based on the vehicle dynamics characteristics and the occupant static attribute characteristics, calculate the probability of rib fracture in the occupant.

[0038] After calculating the vehicle dynamics characteristics and occupant static attributes, this application determines the ease or difficulty parameters of rib fracture based on the occupant's static attributes, and calculates the occupant's rib fracture probability by combining the vehicle dynamics characteristics during the current collision process.

[0039] Step 140: Determine the collision handling decision based on the fracture probability.

[0040] After calculating the probability of occupant rib fractures, corresponding collision handling decisions are made based on this probability to reduce the probability of occupant fractures or the severity of injuries, or to respond to emergency rescue measures in advance, so as to minimize personnel injury.

[0041] This application provides a method for predicting the probability of rib fracture in a human body during a car collision. The method involves collecting state data during the car collision process, including vehicle operation data, safety constraint data, and occupant attribute data. Based on the state data, vehicle dynamics characteristics and occupant static attribute characteristics are calculated. Based on these characteristics, the probability of rib fracture is calculated. Based on the fracture probability, a collision response decision is determined. The method improves prediction accuracy by combining the vehicle dynamics characteristics and occupant static attribute data with the rib fracture probability, thereby enabling rapid collision response and improving occupant safety.

[0042] In one embodiment, the specific implementation of step 120 above may be as follows: segmenting the vehicle operation data and safety constraint data using a sliding time window to obtain multiple segmented data; extracting time-domain features and frequency-domain features for each segmented data; comprehensively obtaining vehicle dynamic features based on the time-domain features and frequency-domain features; calculating occupant static attribute features based on occupant attribute data; wherein, the occupant static attribute features represent the rib impact resistance of the corresponding occupant.

[0043] Specifically, this application uses a fixed-length sliding time window (e.g., covering the first 20ms after the collision begins) to segment the preprocessed vehicle operation data and safety constraint data. For each window, three types of features are extracted. These time-domain and frequency-domain features are then concatenated to obtain vehicle dynamics features. Finally, the vehicle dynamics features and occupant static attribute features are concatenated to form a high-dimensional feature vector. X i :

[0044] Time-domain characteristics: mean, variance, skewness, kurtosis, root mean square (RMS), absolute value integral, zero-crossing rate, etc.

[0045] Frequency domain characteristics: Perform a Fast Fourier Transform (FFT) on the signal within the window to extract 0–200. Hz The amplitude and total energy of the first few dominant frequencies in the frequency band (related to the human thoracic cavity response frequency);

[0046] Passenger static attributes: including height H ,weight W ,age Age ,gender Gender , BMI chest distance D chest After standardization or one-hot encoding, it is incorporated into the feature vector.

[0047] This feature vector X i It comprehensively characterizes the intensity, dynamic characteristics, and individual risk factors of the current collision event, providing rich and structured input for subsequent models.

[0048] In one embodiment, step 130 can be implemented by inputting vehicle dynamics features and occupant static attribute features into a trained prediction model to obtain the probability of occupant rib fracture.

[0049] This application trains a hybrid deep learning model (prediction model) that deeply integrates spatiotemporal dynamic features and occupant static attributes to predict the probability of rib fractures in real time. This prediction model features a targeted architecture designed for the specific biomechanical mechanisms of rib fracture prediction, effectively solving core challenges such as multi-source heterogeneous data fusion and balancing real-time performance with accuracy. Furthermore, during actual collisions, vehicle dynamic features and occupant static attribute features are input into this prediction model to predict the probability of rib fractures.

[0050] In one embodiment, the prediction model includes a temporal feature extractor, a perceptron, and a classifier, with the temporal feature extractor and the perceptron connected in parallel. Specifically, step 130 can be implemented by inputting vehicle dynamics features and occupant static attribute features into the temporal feature extractor and the perceptron, respectively, to obtain spatiotemporal features and static features; then fusing the spatiotemporal features and static features and inputting them into the classifier to obtain the occupant's rib fracture probability.

[0051] like Figure 3 As shown, the prediction model includes a temporal feature extractor, a perceptron, and a classifier. The temporal feature extractor and the perceptron are connected in parallel. The input to the temporal feature extractor is a high-dimensional temporal signal within a sliding window (e.g., ...). a resultant ( t ), F anchor ( t These signals (e.g., collision loads) directly reflect the time-varying characteristics of the collision load.

[0052] In one embodiment, the temporal feature extractor includes a two-stage one-dimensional convolutional neural network and a global average pooling layer. The first-stage one-dimensional convolutional neural network includes 64 convolutional kernels of size 5, and the second-stage one-dimensional convolutional neural network includes 32 convolutional kernels of size 10. The first-stage one-dimensional convolutional neural network, the second-stage one-dimensional convolutional neural network, and the global average pooling layer are connected in series.

[0053] The temporal feature extractor used in this application specifically consists of two-stage one-dimensional convolutional neural network (1D-CNN) modules and a global average pooling layer connected in series. The first-stage 1D-CNN includes 64 convolutional kernels of size 5, which scan the input signal at high resolution, focusing on extracting collision pulse fluctuation features (such as high-frequency impact components). These features may be closely related to the initial response or stress concentration of the ribs. The second-stage 1D-CNN includes 32 convolutional kernels of size 10. The second-stage 1D-CNN receives the output of the first-stage 1D-CNN, enabling it to capture more global temporal patterns (such as the rise slope, duration, and overall shape of the pulse). These patterns reflect the overall energy level and load type of the collision. Each convolutional layer is followed by an activation function, specifically: ,in, Indicates the first The weights of each activation function, Indicates the first The output of each activation function Indicates the first The bias of each activation function, where * indicates a one-dimensional convolution operation, and ReLU is the activation function. Global Average Pooling transforms the feature map generated by convolution into a fixed-length, highly abstract feature vector. h cnn The temporal feature extractor provided in this application not only significantly reduces the number of parameters, which is beneficial for model acceleration, but also enhances the robustness of the model to the length of the input sequence.

[0054] In one embodiment, the perceptron includes two fully connected layers, wherein the output dimension of the first fully connected layer is 16 and the output dimension of the second fully connected layer is 8.

[0055] The perceptron provided in this application consists of a multilayer perceptron (MLP), specifically comprising two fully connected layers: the input to the perceptron is a occupant static attribute feature vector (such as height H, weight W, age, gender, etc.) processed by normalization or one-hot encoding; the first fully connected layer maps the input to a 16-dimensional space, and the second fully connected layer further maps it to an 8-dimensional space; through nonlinear transformation, the perceptron maps static attributes such as category and numerical value into a low-dimensional, dense semantic vector. h dense This semantic vector effectively encodes the impact of individual occupant differences on skeletal impact resistance.

[0056] In one embodiment, the classifier includes a random deactivation layer and a classification layer, wherein the random deactivation layer has a dropout rate of 0.5 and the classification layer includes an activation function.

[0057] The classifier consists of a random deactivation layer and a classification layer. The output vectors (spatiotemporal features and static attribute features) of the two branches (temporal feature extractor and perceptron) are concatenated. This concatenation operation achieves deep fusion of the external collision load and the occupant's inherent vulnerability at the feature level, providing the model with unprecedented comprehensive information. The fused features are then input into a fully connected classifier (16 neurons) with a Dropout layer (dropout rate set to 0.5) to prevent overfitting. Finally, the data is input into the classification layer, which outputs a scalar value between 0 and 1 through a Sigmoid activation function, representing the probability of rib fracture. P rib fracture .

[0058] The final model output expression is: P rib fracture = σ ( W f •[ h cnn ; h dense]+ b f );

[0059] in, σ It is the Sigmoid function. W f and b f These are the weights and biases of the classification layer, and [;] indicates vector concatenation.

[0060] To address the extreme class imbalance caused by the scarcity of positive fracture samples in real collision data during the training process of the aforementioned prediction model, this application employs focus loss, the formula of which is as follows:

[0061] ;

[0062] in, It is the damage function value. It is the probability that the model predicts a sample belongs to the positive class. It is a category balance factor (e.g.) The value range is [0.7, 0.9]. γ It is a regulator (e.g.) γ With a value range of [1.5, 3.0], this loss function can automatically reduce the weight of easily classified samples (a large number of non-fracture samples), so that the training process focuses on difficult-to-classify, rare fracture samples, thereby significantly improving the model's recall rate for high-risk cases.

[0063] In one embodiment, step 140 can be implemented as follows: if the probability of fracture is greater than a preset probability threshold, an emergency response is triggered.

[0064] In actual collision processes, the feature vectors generated in real time will be... X i Input into the deployed prediction model, calculate and output the current time value in real time. P rib fracture It integrates the SHAP (SHapley Additive exPlanations) analysis module, which calculates the SHAP value of each input feature for important prediction samples, greatly enhancing the reliability and actionability of the results.

[0065] In one embodiment, step 140 can be implemented as follows: if the probability of fracture is greater than a preset first probability threshold, the seat belt is pre-tightened; if the probability of fracture is greater than a preset second probability threshold, the airbag deployment mode is switched and a warning message is sent; wherein the second probability threshold is greater than the first probability threshold.

[0066] This application will predict probabilities. P rib fractureThe probability is compared with multiple preset thresholds (e.g., the first probability threshold T1 = 0.3, the second probability threshold T2 = 0.7). If... P rib fracture >T1, triggers Level 1 warning, pretensions seatbelt to eliminate webbing slack; if P rib fracture >T2 triggers a level 2 response, instructing the airbags to deploy earlier and with greater force, and sending a "high-risk chest injury" warning to the rescue center via the eCall system.

[0067] Figure 4 This is a flowchart illustrating a method for predicting the probability of rib fractures in a human body during a car collision, provided in another exemplary embodiment of this application. Figure 4 As shown, the method for predicting the probability of rib fractures in a car collision includes the following steps:

[0068] Step 510: Collect raw data.

[0069] This step is the same as step 110 above, and will not be repeated here.

[0070] Step 520: Data preprocessing.

[0071] This step is the same as the data preprocessing described above, and will not be repeated here.

[0072] Step 530: Feature extraction.

[0073] This step is the same as step 120 above, and will not be repeated here.

[0074] Step 540: Construct fusion features.

[0075] This step is the same as the method for obtaining fusion features described above, and will not be repeated here.

[0076] Step 550: Input the model to calculate the probability of fracture.

[0077] This step is the same as step 130 above, and will not be repeated here.

[0078] Step 560: Determine if it is greater than the threshold. If yes, proceed to step 570; otherwise, proceed to step 510.

[0079] This step is the same as the comparison of fracture probability and probability threshold mentioned above, and will not be repeated here.

[0080] Step 570: Trigger collision handling decision.

[0081] This step is the same as step 140 above, and will not be repeated here.

[0082] Figure 5This is a schematic diagram of the structure of a system for predicting the probability of rib fractures in a human body during a car collision, provided in an exemplary embodiment of this application. Figure 5 As shown, the prediction system 60 for the probability of rib fracture in a car collision includes: a state data acquisition module 61, used to acquire state data during the car collision process; wherein, the state data includes vehicle operation data, safety constraint data, and occupant attribute data; a collision feature calculation module 62, used to calculate vehicle dynamics features and occupant static attribute features based on the state data; a fracture probability calculation module 63, used to calculate the probability of rib fracture in the occupant based on the vehicle dynamics features and occupant static attribute features; and a processing strategy determination module 64, used to determine the collision processing decision based on the fracture probability.

[0083] This application provides a system for predicting the probability of rib fractures in a car collision. The system collects state data during the collision process via a state data acquisition module 61. This state data includes vehicle operation data, safety constraint data, and occupant attribute data. A collision feature calculation module 62 calculates vehicle dynamics characteristics and occupant static attribute characteristics based on the state data. A fracture probability calculation module 63 calculates the probability of rib fractures based on the vehicle dynamics characteristics and occupant static attribute characteristics. A processing strategy determination module 64 determines the collision handling decision based on the fracture probability. By calculating the vehicle dynamics characteristics and occupant static attribute characteristics based on the vehicle's dynamics data and occupant attribute data during the collision process, and combining these characteristics to calculate the rib fracture probability, the system improves prediction accuracy. Furthermore, by determining the collision handling decision based on the fracture probability, the system can respond quickly to collisions and improve occupant safety.

[0084] In one embodiment, the collision feature calculation module 62 can be further configured to: segment the vehicle operation data and safety constraint data using a sliding time window to obtain multiple segmented data; for each segmented data, extract time-domain features and frequency-domain features; wherein, the time-domain features include mean, variance, skewness, kurtosis, root mean square, absolute value integral, and zero-crossing rate, and the frequency-domain features include performing a fast Fourier transform on the signal within the sliding time window and extracting 0–200 Hz The amplitude and total energy of the first few main frequencies obtained from the frequency bands are used to obtain the vehicle dynamics characteristics based on time-domain and frequency-domain characteristics. The static attribute characteristics of the occupants are calculated based on the occupant attribute data. Among them, the static attribute characteristics of the occupants represent the rib impact resistance of the corresponding occupants.

[0085] In one embodiment, the fracture probability calculation module 63 can be further configured to input the vehicle dynamics features and the occupant static attribute features into the trained prediction model to obtain the fracture probability of the occupant's ribs.

[0086] In one embodiment, the prediction model includes a temporal feature extractor, a perceptron, and a classifier, with the temporal feature extractor and the perceptron connected in parallel; wherein, the fracture probability calculation module 63 can be further configured to: input vehicle dynamic features and occupant static attribute features into the temporal feature extractor and the perceptron respectively to obtain spatiotemporal features and static features; and input the fused spatiotemporal features and static features into the classifier to obtain the fracture probability of the occupant's ribs.

[0087] In one embodiment, the temporal feature extractor includes a two-stage one-dimensional convolutional neural network and a global average pooling layer. The first-stage one-dimensional convolutional neural network includes 64 convolutional kernels of size 5, and the second-stage one-dimensional convolutional neural network includes 32 convolutional kernels of size 10. The first-stage one-dimensional convolutional neural network, the second-stage one-dimensional convolutional neural network, and the global average pooling layer are connected in series.

[0088] In one embodiment, the perceptron includes two fully connected layers, wherein the output dimension of the first fully connected layer is 16 and the output dimension of the second fully connected layer is 8.

[0089] In one embodiment, the classifier includes a random deactivation layer and a classification layer, wherein the random deactivation layer has a dropout rate of 0.5 and the classification layer includes an activation function.

[0090] In one embodiment, the above-mentioned processing strategy determination module 64 can be further configured to trigger an emergency response if the probability of fracture is greater than a preset probability threshold.

[0091] In one embodiment, the above-mentioned processing strategy determination module 64 can be further configured to: if the probability of fracture is greater than a preset first probability threshold, then pre-tighten the seat belt; if the probability of fracture is greater than a preset second probability threshold, then switch the airbag deployment mode and send a warning message; wherein, the second probability threshold is greater than the first probability threshold.

[0092] In addition to the methods and apparatus described above, embodiments of this application may also be computer program products, which include computer program instructions that, when executed by a processor, cause the processor to perform the steps in the methods according to various embodiments of this application described in the "Exemplary Methods" section above.

[0093] The computer program product can be written in any combination of one or more programming languages ​​to perform the operations of the embodiments of this application. The programming languages ​​include object-oriented programming languages ​​such as Java and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computing device, partially on the user's computing device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0094] Furthermore, embodiments of this application may also be computer-readable storage media storing computer program instructions thereon, which, when executed by a processor, cause the processor to perform the steps in the methods according to various embodiments of this application described in the "Exemplary Methods" section above.

[0095] The computer-readable storage medium may be any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof.

[0096] The basic principles of this application have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in this application are merely examples and not limitations, and should not be considered as essential features of each embodiment of this application. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not limit the application to the necessity of employing the aforementioned specific details for implementation.

[0097] The block diagrams of devices, apparatuses, devices, and systems involved in this application are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, devices, and systems can be connected, arranged, and configured in any manner. Words such as “comprising,” “including,” “having,” etc., are open-ended terms meaning “including but not limited to,” and are used interchangeably with them. The terms “or” and “and” as used herein refer to the terms “and / or,” and are used interchangeably with them unless the context clearly indicates otherwise. The term “such as” as used herein refers to the phrase “such as but not limited to,” and is used interchangeably with it.

[0098] It should also be noted that in the apparatus, equipment, and methods of this application, the components or steps can be disassembled and / or recombined. These disassemblies and / or recombinations should be considered as equivalent solutions of this application.

[0099] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use this application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein can be applied to other aspects without departing from the scope of this application. Therefore, this application is not intended to be limited to the aspects shown herein, but rather to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0100] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of this application to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations thereof.

Claims

1. A method for predicting the probability of rib fractures in a human body during a car collision, characterized in that, include: Collect state data during a vehicle collision; wherein, the state data includes vehicle operation data, safety constraint data, and occupant attribute data, the vehicle operation data includes collision speed change, longitudinal acceleration, lateral acceleration, and original acceleration signal, and the safety constraint data includes seat belt anchor force and webbing pull-out amount. ， The occupant attribute data includes occupant height, weight, body mass index, age, gender, and precise sitting posture in the vehicle; Based on the state data, the vehicle dynamics characteristics and occupant static attribute characteristics are calculated; Based on the vehicle dynamics characteristics and the occupant static attributes, the probability of rib fracture in the occupant is calculated. Based on the fracture probability, a collision handling decision is determined; The vehicle dynamics characteristics and occupant static attribute characteristics calculated based on the state data include: The vehicle operation data and the safety constraint data are segmented using a sliding time window to obtain multiple segmented data. For each segment of data, time-domain features and frequency-domain features are extracted; wherein, the time-domain features include mean, variance, skewness, kurtosis, root mean square, absolute value integral, and zero-crossing rate; and the frequency-domain features include performing a Fast Fourier Transform on the signal within the sliding time window and extracting 0–200... Hz The amplitude and total energy of the first few main frequencies obtained in the frequency band; The vehicle dynamics characteristics are obtained by combining the time-domain features and the frequency-domain features. Based on the occupant attribute data, the occupant static attribute characteristics are calculated; wherein, the occupant static attribute characteristics represent the rib impact resistance of the corresponding occupant.

2. The method for predicting the probability of rib fractures in a human body during a car collision according to claim 1, characterized in that, The calculation of the rib fracture probability of the occupant based on the vehicle dynamics characteristics and the occupant static attribute characteristics includes: The vehicle dynamics features and the occupant static attributes are input into the trained prediction model to obtain the fracture probability of the occupant's ribs.

3. The method for predicting the probability of rib fractures in a human body during a car collision according to claim 2, characterized in that, The prediction model includes a temporal feature extractor, a perceptron, and a classifier, wherein the temporal feature extractor and the perceptron are connected in parallel; wherein, the step of inputting the vehicle dynamics features and the occupant static attribute features into the trained prediction model to obtain the fracture probability of the occupant's ribs includes: The vehicle dynamics features and the occupant static attribute features are respectively input into the temporal feature extractor and the perceptron to obtain spatiotemporal features and static features; The spatiotemporal features and the static features are fused and then input into the classifier to obtain the fracture probability of the occupant's ribs.

4. The method for predicting the probability of rib fractures in a human body during a car collision according to claim 3, characterized in that, The temporal feature extractor includes a two-stage one-dimensional convolutional neural network and a global average pooling layer. The first-stage one-dimensional convolutional neural network includes 64 convolutional kernels of size 5, and the second-stage one-dimensional convolutional neural network includes 32 convolutional kernels of size 10. The first-stage one-dimensional convolutional neural network, the second-stage one-dimensional convolutional neural network and the global average pooling layer are connected in series.

5. The method for predicting the probability of rib fractures in a human body during a car collision according to claim 3, characterized in that, The perceptron includes two fully connected layers, wherein the output dimension of the first fully connected layer is 16 and the output dimension of the second fully connected layer is 8.

6. The method for predicting the probability of rib fractures in a human body during a car collision according to claim 3, characterized in that, The classifier includes a random deactivation layer and a classification layer, wherein the dropout rate of the random deactivation layer is 0.5, and the classification layer includes an activation function.

7. The method for predicting the probability of rib fractures in a human body during a car collision according to claim 1, characterized in that, The process of determining the collision handling decision based on the fracture probability includes: If the probability of fracture is greater than a preset probability threshold, an emergency response is triggered.

8. The method for predicting the probability of rib fractures in a human body during a car collision according to claim 7, characterized in that, The step of triggering an emergency response if the probability of fracture is greater than a preset probability threshold includes: If the probability of fracture is greater than a preset first probability threshold, then the seat belt is pre-tightened; If the probability of fracture is greater than a preset second probability threshold, the airbag deployment mode is switched and a warning message is sent; wherein the second probability threshold is greater than the first probability threshold.

9. A prediction system for the probability of rib fracture in a human body during a car collision, characterized in that, include: The status data acquisition module is used to collect status data during a vehicle collision. This status data includes vehicle operation data, safety constraint data, and occupant attribute data. The vehicle operation data includes collision speed change, longitudinal acceleration, lateral acceleration, and the initial acceleration signal. The safety constraint data includes seatbelt anchor force and webbing pull-out amount. ， The occupant attribute data includes occupant height, weight, body mass index, age, gender, and precise sitting posture in the vehicle; The collision feature calculation module is used to calculate the vehicle dynamics features and occupant static attribute features based on the state data. The fracture probability calculation module is used to calculate the fracture probability of the occupant's ribs based on the vehicle dynamics characteristics and the occupant static attribute characteristics. The processing strategy determination module is used to determine a collision processing decision based on the fracture probability. The collision feature calculation module is further configured as follows: The vehicle operation data and the safety constraint data are segmented using a sliding time window to obtain multiple segmented data. For each segment of data, time-domain features and frequency-domain features are extracted; wherein, the time-domain features include mean, variance, skewness, kurtosis, root mean square, absolute value integral, and zero-crossing rate; and the frequency-domain features include performing a Fast Fourier Transform on the signal within the sliding time window and extracting 0–200... Hz The amplitude and total energy of the first few main frequencies obtained in the frequency band; The vehicle dynamics characteristics are obtained by combining the time-domain features and the frequency-domain features. Based on the occupant attribute data, the occupant static attribute characteristics are calculated; wherein, the occupant static attribute characteristics represent the rib impact resistance of the corresponding occupant.

Images