Method and system for determining triggering state of safety air bag and vehicle terminal

Abstract

The invention relates to the technical field of vehicle safety, and provides a method and system for determining the triggering state of an air bag restraint system and a vehicle terminal. Whether the vehicle has the collision event or not is detected through the collision detection factors, the corresponding collision detection results are obtained, different from direct fusion independent detection results, relevance between the collision detection results is further established through the relative detection results, feature fusion is conducted on the relative detection results, and the collision detection accuracy is improved. Therefore, the triggering state of the safety airbag is determined according to the comprehensive detection result obtained through feature fusion, by introducing the relative detection result and conducting feature fusion, the mode that all collision detection factors are regarded as isolated signals to be processed in the past is changed, internal logic association of different sensor signals in a collision event is effectively quantified, and the collision detection accuracy is improved. And the obtained comprehensive detection result can reflect the real collision state of the vehicle more accurately and comprehensively, so that the reliability of an airbag triggering decision is improved.

Description

Technical Field

[0001] This application relates to the field of vehicle safety technology, specifically to a method, system, and vehicle terminal for determining the triggering state of an airbag. Background Technology

[0002] Currently, as automotive electronic and electrical architectures gradually evolve towards a combination of central computing and regional control, the functions of traditionally distributed ECUs (Electronic Control Units) have been greatly simplified and integrated. This has significantly increased the computing power requirements, functional integration, and software complexity of regional controllers, thereby placing higher demands on vehicle safety functions, especially the response speed and reliability of safety measures such as door lock unlocking and airbag deployment. In particular, the airbag system typically uses sensors deployed throughout the vehicle to acquire collision detection factors, such as acceleration and rotation angle. When the value of the collision detection factor exceeds a preset trigger threshold, the system determines that a collision has occurred and triggers the airbag.

[0003] However, collision events involve complex physical processes. Changes in acceleration in different directions, structural deformation of different parts of the vehicle body, and the generation of impact pressure all exhibit inherent logical connections in terms of time series and intensity changes. For example, a forward collision event is inevitably accompanied by a sharp negative change in longitudinal acceleration and a surge in the signal from the front pressure sensor. However, in practical applications, various collision detection factors are often treated as independent signals, with only independent threshold judgments or simple linear weighting of the factor readings. This lacks analysis of the inherent correlations between them, resulting in an inaccurate assessment of whether a collision actually occurred. Ultimately, this leads to low reliability of airbag deployment decisions, failing to meet the stringent safety standards of the automotive industry. Summary of the Invention

[0004] To provide a basic understanding of some aspects of the disclosed embodiments, a brief summary is given below. This summary is not intended as a general commentary, nor is it intended to identify key / important components or describe the scope of protection of these embodiments, but rather as a prelude to the detailed description that follows.

[0005] In view of the shortcomings of the prior art described above, this application provides a method, system and vehicle terminal for determining the triggering state of an airbag, so as to improve the reliability of airbag triggering decisions.

[0006] This application provides a method for determining the triggering state of an airbag, comprising: acquiring multiple collision detection factors of a vehicle terminal; detecting whether a collision event has occurred in the vehicle terminal according to each of the collision detection factors, and obtaining a collision detection result corresponding to each of the collision detection factors; performing feature fusion based on the relative detection results between the collision detection results to obtain a comprehensive detection result; and determining the triggering state of the airbag based on the comprehensive detection result.

[0007] In one embodiment of this application, the collision detection of the vehicle terminal based on each collision detection factor to obtain a collision detection result corresponding to each collision detection factor includes at least one of the following: comparing the detection-related signals corresponding to the collision detection factor according to a preset detection-related signal threshold, and evaluating whether the vehicle terminal has experienced a collision event based on the comparison result to obtain a collision detection result corresponding to the collision detection factor, wherein the detection-related signals are obtained by data acquisition from the vehicle terminal; and using a preset collision detection model to evaluate the probability of the vehicle terminal experiencing a collision event based on the detection-related signals corresponding to the collision detection factor to obtain a collision detection result corresponding to the collision detection factor, wherein the collision detection model is obtained by training a preset neural network model.

[0008] In one embodiment of this application, feature fusion is performed based on the relative detection results between the collision detection results to obtain a comprehensive detection result, including: using each collision detection factor as the row index and column index of a first judgment matrix, and using the ratio between the collision detection results as the relative detection result, and filling it into the first judgment matrix; performing feature fusion on the first judgment matrix based on a preset feature fusion step to obtain a first fusion matrix, wherein the feature fusion step includes downsampling and convolution operations; calculating the first fusion matrix using the geometric mean method to obtain a first matrix quantization value, and mapping the first matrix quantization value to a preset first quantization value range to obtain a comprehensive detection result.

[0009] In one embodiment of this application, feature fusion is performed on the first judgment matrix based on a preset feature fusion step to obtain a first fusion matrix, including: using the first judgment matrix as a first current matrix; responding to the first current matrix, downsampling the first current matrix through a preset pooling layer to obtain a first intermediate matrix; performing convolution operation on the first current matrix and the first intermediate matrix to obtain a first target matrix; if the matrix dimension of the first target matrix meets a preset matrix dimension condition, then using the first target matrix as the first fusion matrix; if the matrix dimension of the first target matrix does not meet the matrix dimension condition, then using the first target matrix as a new first current matrix.

[0010] In one embodiment of this application, determining the triggering state of the airbag based on the comprehensive detection result includes: if the comprehensive detection result is greater than or equal to a preset detection result threshold, then setting the triggering state of the airbag to an activated state; if the comprehensive detection result is less than the detection result threshold, then setting the triggering state of the airbag to a deactivated state.

[0011] In one embodiment of this application, before determining the triggering state of the airbag based on the comprehensive detection result, the method further includes: acquiring multiple collision prediction factors of the vehicle terminal; evaluating the probability of a collision event occurring in the vehicle terminal based on each of the collision prediction factors to obtain a collision prediction result corresponding to each of the collision prediction factors, and performing feature fusion based on the relative prediction results between the collision prediction results to obtain a first prediction probability; evaluating the importance of each of the collision prediction factors, and determining the relative prediction weight of each of the collision prediction factors based on the relative importance between the importance levels, and performing weighted calculation on the collision prediction results based on the relative prediction weights to obtain a second prediction probability; combining the first prediction probability and the second prediction probability to calculate a comprehensive prediction result; if the comprehensive prediction result is greater than or equal to a preset prediction result threshold, then setting the pre-preparation state of the airbag to the execution state.

[0012] In one embodiment of this application, feature fusion is performed based on the relative prediction results between the collision prediction results to obtain a first prediction probability, including: using each of the collision prediction factors as the row index and column index of a second judgment matrix, and using the ratio between the collision prediction results as the relative prediction result, and filling it into the second judgment matrix; performing feature fusion on the second judgment matrix based on a preset feature fusion step to obtain a second fusion matrix; calculating the second fusion matrix using the geometric mean method to obtain a second matrix quantization value, and mapping the second matrix quantization value to a preset second quantization value interval to obtain the first prediction probability.

[0013] In one embodiment of this application, determining the relative prediction weights corresponding to each of the collision prediction factors based on the relative importance between the importance levels includes: using each of the collision prediction factors as the row index and column index of a third judgment matrix, and using the ratio between the importance levels as the relative importance, and filling it into the third judgment matrix; calculating the largest eigenvalue corresponding to the third judgment matrix, and calculating the eigenvector corresponding to the largest eigenvalue; normalizing the eigenvector to obtain a relative weight vector, wherein the relative weight vector includes the relative prediction weights corresponding to each of the collision prediction factors.

[0014] This application also provides a system for determining the triggering state of an airbag, comprising: an acquisition module for acquiring multiple collision detection factors of a vehicle terminal; a detection module for detecting whether a collision event has occurred in the vehicle terminal according to each of the collision detection factors, and obtaining a collision detection result corresponding to each of the collision detection factors; a fusion module for performing feature fusion based on the relative detection results between the collision detection results, and obtaining a comprehensive detection result; and a determination module for determining the triggering state of the airbag based on the comprehensive detection result.

[0015] This application also provides a vehicle terminal, including: a processor and a memory; the memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory to cause the vehicle terminal to perform the method described above.

[0016] The beneficial effects of this application are: By detecting whether a collision event has occurred for each collision factor separately, and obtaining their respective collision detection results, unlike directly fusing independent detection results, this approach further establishes the correlation between the collision detection results through relative detection results. Then, feature fusion is performed on the relative detection results, and the airbag triggering state is determined based on the comprehensive detection result obtained from feature fusion. In this way, by introducing relative detection results and performing feature fusion, the previous approach of treating each collision detection factor as an isolated signal is changed. The inherent logical correlation between different sensor signals in a collision event is effectively quantified, and the resulting comprehensive detection result can more accurately and comprehensively reflect the vehicle's true collision state, thereby improving the reliability of airbag triggering decisions. Attached Figure Description

[0017] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application. It is obvious that the drawings described below are merely some embodiments of this application, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.

[0018] In the attached diagram: Figure 1 This is a flowchart illustrating a method for determining the triggering state of an airbag in an embodiment of this application. Figure 2 This is a schematic diagram of the structure of a hierarchical model based on collision detection factors in an embodiment of this application; Figure 3 This is a schematic diagram of the structure of a control system based on an airbag controller in an embodiment of this application; Figure 4This is a schematic diagram of the structure of a hierarchical model based on collision prediction factors in an embodiment of this application; Figure 5 This is a schematic diagram of the structure of an airbag controller installed in a vehicle terminal according to an embodiment of this application; Figure 6 This is a schematic diagram of another control system based on an airbag controller in an embodiment of this application; Figure 7 This is a flowchart illustrating another method for determining the triggering state of an airbag in an embodiment of this application; Figure 8 This is a schematic diagram of the structure of an airbag triggering state determination system in an embodiment of this application; Figure 9 This is a structural schematic diagram of a vehicle terminal in an embodiment of this application. Detailed Implementation

[0019] The following specific examples illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. This application can also be implemented or applied through other different specific embodiments. Various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this application. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0020] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of this application. The drawings only show the components related to this application and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0021] In the following description, numerous details are explored to provide a more thorough explanation of embodiments of the present application. However, it will be apparent to those skilled in the art that embodiments of the present application may be practiced without these specific details. In other embodiments, well-known structures and devices are shown in block diagram form rather than in detail to avoid obscuring embodiments of the present application.

[0022] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this application described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion.

[0023] Unless otherwise stated, the term "multiple" means two or more.

[0024] In this application, the character " / " indicates that the objects before and after it are in an "or" relationship. For example, A / B means: A or B.

[0025] The term "and / or" describes an association between objects, indicating that three relationships can exist. For example, A and / or B means: A or B, or A and B.

[0026] Combination Figure 1 As shown, this application provides a method for determining the triggering state of an airbag, including: Step S101: Obtain multiple collision detection factors from the vehicle terminal; Step S102: Detect whether a collision event has occurred in the vehicle terminal according to each collision detection factor, and obtain the collision detection result corresponding to each collision detection factor. Step S103: Perform feature fusion based on the relative detection results between each collision detection result to obtain a comprehensive detection result; Step S104: Determine the triggering state of the airbag based on the comprehensive test results.

[0027] The airbag triggering state determination method provided in this application detects whether a collision event has occurred by detecting each collision detection factor separately, obtaining their respective collision detection results. Unlike directly fusing independent detection results, this method further establishes the correlation between the collision detection results through relative detection results, and then performs feature fusion on the relative detection results. Based on the comprehensive detection result obtained from feature fusion, the airbag triggering state is determined. By introducing relative detection results and performing feature fusion, this method changes the previous approach of treating each collision detection factor as an isolated signal. It effectively quantifies the inherent logical correlation between different sensor signals in a collision event, resulting in a more accurate and comprehensive comprehensive detection result that reflects the vehicle's true collision state, thereby improving the reliability of airbag triggering decisions.

[0028] Optionally, the collision detection results for each collision detection factor are obtained by detecting whether a collision event has occurred in the vehicle terminal according to each collision detection factor. This includes comparing the detection-related signals corresponding to the collision detection factors according to a preset detection-related signal threshold, and evaluating whether a collision event has occurred in the vehicle terminal based on the comparison results. The detection-related signals are obtained by collecting data from the vehicle terminal.

[0029] In some embodiments, a safe threshold range and a dangerous threshold range are determined based on a preset detection-related signal threshold. If the detection-related signal is within the safe threshold range, the collision detection result corresponding to the detection-related signal is 0. If the detection-related signal is within the dangerous threshold range, the signal deviation value of the detection-related signal relative to the detection-related signal threshold is calculated, and the collision detection result of the collision detection factor corresponding to the detection-related signal is obtained by matching based on the signal deviation value.

[0030] Optionally, the collision detection results corresponding to each collision detection factor are obtained by detecting whether a collision event occurs at the vehicle terminal according to each collision detection factor. This includes: using a preset collision detection model, evaluating the probability of a collision event occurring at the vehicle terminal based on the detection-related signals corresponding to the collision detection factors, and obtaining the collision detection results corresponding to the collision detection factors. The collision detection model is obtained by training a preset neural network model.

[0031] In some embodiments, a training sample set is constructed based on a vehicle event set; a preset neural network model is trained using the training sample set to obtain a collision detection model; and the collision detection model is used to perform risk assessment on the detection-related signals corresponding to each collision detection factor to obtain the collision detection results corresponding to each collision detection factor.

[0032] Optionally, feature fusion is performed based on the relative detection results between each collision detection result to obtain a comprehensive detection result, including: using each collision detection factor as the row index and column index of the first judgment matrix, and using the ratio between collision detection results as the relative detection result, and filling it into the first judgment matrix; performing feature fusion on the first judgment matrix based on a preset feature fusion step to obtain a first fusion matrix, wherein the feature fusion step includes downsampling and convolution operations; calculating the first fusion matrix using the geometric mean method to obtain a first matrix quantization value, and mapping the first matrix quantization value to a preset first quantization value range to obtain a comprehensive detection result.

[0033] Combination Figure 2 As shown, this application provides a hierarchical model based on collision detection factors, including multiple collision detection dimensions and collision detection factors corresponding to each collision detection dimension. The collision detection dimensions include displacement sensor dimensions, angle sensor dimensions, and vehicle environment dimensions. The collision detection factors corresponding to the displacement sensor dimension include vehicle relative displacement (e.g., X-axis relative displacement, Y-axis relative displacement, and Z-axis relative displacement), the collision detection factors corresponding to the angle sensor dimension include vehicle rotation angles (e.g., X-axis rotation angle, Y-axis rotation angle, and Z-axis rotation angle), and the collision detection factors corresponding to the vehicle environment dimension include vehicle smoke volume and vehicle temperature.

[0034] In some embodiments, the first The collision detection factor is relative to the first collision detection factor. The relative detection results of each collision detection factor are expressed as follows: ,in, For the first Collision detection results for each collision detection factor. For the first The collision detection results of each collision detection factor.

[0035] In some embodiments, at least a portion of the first judgment matrix corresponding to the relative detection results is shown in Table 1.

[0036] Table 1

[0037] Optionally, feature fusion is performed on the first judgment matrix based on a preset feature fusion step to obtain a first fusion matrix, including: using the first judgment matrix as a first current matrix; responding to the first current matrix, downsampling the first current matrix through a preset pooling layer to obtain a first intermediate matrix; performing convolution operation on the first current matrix and the first intermediate matrix to obtain a first target matrix; if the matrix dimension of the first target matrix meets a preset matrix dimension condition, then the first target matrix is ​​used as the first fusion matrix; if the matrix dimension of the first target matrix does not meet the matrix dimension condition, then the first target matrix is ​​used as a new first current matrix.

[0038] In some embodiments, the matrix dimension condition is 2×2.

[0039] In some embodiments, a first judgment matrix with a matrix dimension of 8×8 is used as the first current matrix T1. The first current matrix T1 is downsampled through a pooling layer to obtain a first intermediate matrix T2 with a matrix dimension of 4×4. A convolution is then performed on the first current matrix T1 and the first intermediate matrix T2 to obtain a first target matrix T3 with a matrix dimension of 4×4. The first target matrix T3 is used as the new first current matrix. The first current matrix T3 is downsampled through a pooling layer to obtain a first intermediate matrix T4 with a matrix dimension of 2×2. A convolution is then performed on the first current matrix T3 and the first intermediate matrix T4 to obtain a first target matrix T5 with a matrix dimension of 2×2. The first target matrix T5 is used as the first fusion matrix, and the geometric mean method is used to calculate the matrix quantization value corresponding to the first fusion matrix T5. Quantize the matrix values Mapped to and The comprehensive detection result is obtained by analyzing the quantitative value range between these values.

[0040] In some embodiments, average pooling is used to downsample the first current matrix to achieve matrix dimensionality reduction and feature selection, retaining the correlation features between relative probabilities and avoiding the loss of key features due to max pooling.

[0041] In some embodiments, a valid convolution operation is performed on the first current matrix and the first intermediate matrix to capture the deep correlation between matrix elements and realize feature correlation extraction.

[0042] In some embodiments, if the first fusion matrix T5 = [[ , ],[ , Then, the geometric mean method is used to calculate the first fusion matrix T5, ​​and the matrix quantization value corresponding to the first fusion matrix T5 is obtained. , where matrix quantization value The intensity value is relative to the detection result; through vehicle collision experiments, the minimum value of the matrix quantization value in a no-collision-risk scenario is taken as the intensity value. And taking the maximum value of the matrix quantization value in the collision risk scenario as... A linear normalization algorithm is used to quantize the matrix values. Mapped to and The comprehensive detection result is obtained by calculating the quantitative value range between these values, where the comprehensive detection result = .

[0043] Optionally, the triggering state of the airbag is determined based on the comprehensive test results, including: if the comprehensive test results are greater than or equal to a preset test result threshold, the triggering state of the airbag is set to the activated state; if the comprehensive test results are less than the test result threshold, the triggering state of the airbag is set to the deactivated state.

[0044] Combination Figure 3 As shown, this application provides a control system based on an airbag controller, including an algorithm module, a data acquisition module, an airbag controller, and an airbag. The algorithm module is used to execute the above-mentioned method for determining the trigger state of the airbag. The data acquisition module is used to acquire sensor signals corresponding to the collision detection factor. The airbag controller is used to trigger the airbag to deploy if the trigger state of the airbag is set to the open state. The airbag controller includes an ECU.

[0045] Optionally, before determining the airbag triggering state based on the comprehensive detection results, the method further includes: acquiring multiple collision prediction factors of the vehicle terminal; evaluating the probability of a collision event occurring at the vehicle terminal based on each collision prediction factor to obtain a collision prediction result corresponding to each collision prediction factor, and performing feature fusion based on the relative prediction results between the collision prediction results to obtain a first prediction probability; evaluating the importance of each collision prediction factor, and determining the relative prediction weight of each collision prediction factor based on the relative importance between the importance levels, and performing weighted calculation on the collision prediction results based on the relative prediction weights to obtain a second prediction probability; combining the first prediction probability and the second prediction probability to obtain a comprehensive prediction result; if the comprehensive prediction result is greater than or equal to a preset prediction result threshold, then setting the airbag's pre-preparation state to the execution state.

[0046] Combination Figure 4 As shown in the embodiment of this application, a hierarchical model based on collision prediction factors includes multiple collision prediction dimensions and collision prediction factors corresponding to each collision prediction dimension. The collision prediction dimensions include one or more of the following: user state dimension, vehicle state dimension, and environmental perception dimension. The collision prediction factors corresponding to the user state dimension include one or more of the following: in-vehicle environment perception, driving behavior perception, and user fatigue detection. The collision prediction factors corresponding to the vehicle state dimension include one or more of the following: driving speed, accelerator pedal travel, brake pedal travel, time-to-collision (TTC), and predicted collision angle. The collision prediction factors corresponding to the environmental perception dimension include one or more of the following: road surface perception, weather perception, user blind spot perception, and driving direction perception.

[0047] In some embodiments, the probability of a risk event occurring at the vehicle terminal is assessed based on each collision prediction factor to obtain a collision prediction result corresponding to each collision prediction factor, including: acquiring data from the vehicle terminal to obtain prediction-related signals corresponding to each collision prediction factor; assessing the probability of a collision event occurring at the vehicle terminal based on each prediction-related signal to obtain a prediction result vector, wherein the prediction result vector includes the collision prediction result corresponding to each collision prediction factor.

[0048] In some embodiments, a safe threshold range and a dangerous threshold range are determined based on a preset prediction correlation signal threshold. If the prediction correlation signal is located in the safe threshold range, the collision prediction result corresponding to the prediction correlation signal is 0. If the prediction correlation signal is located in the dangerous threshold range, the signal deviation value of the prediction correlation signal relative to the vehicle signal threshold is calculated, and the signal is matched according to the signal deviation value to obtain the collision prediction result of the collision prediction factor corresponding to the prediction correlation signal.

[0049] In some embodiments, a training sample set is constructed based on a vehicle event set; a preset neural network model is trained using the training sample set to obtain a collision prediction model; and the collision prediction model is used to assess the risk of the prediction-related signals corresponding to each collision prediction factor to obtain the collision prediction results corresponding to each collision prediction factor.

[0050] Optionally, feature fusion is performed based on the relative prediction results between collision prediction results to obtain a first prediction probability, including: using each collision prediction factor as the row index and column index of the second judgment matrix, and using the ratio between collision prediction results as the relative prediction result, and filling it into the second judgment matrix; performing feature fusion on the second judgment matrix based on a preset feature fusion step to obtain a second fusion matrix; calculating the second fusion matrix using the geometric mean method to obtain a second matrix quantization value, and mapping the second matrix quantization value to a preset second quantization value range to obtain the first prediction probability.

[0051] In some embodiments, the first The collision prediction factor relative to the first collision prediction factor The relative prediction results of each collision predictor are expressed as follows: ,in, For the first Collision prediction results for each collision predictor factor. For the first Collision prediction results for each collision predictor.

[0052] In some embodiments, at least a portion of the second judgment matrix corresponding to the relative prediction results is shown in Table 2.

[0053] Table 2

[0054] In some embodiments, the second judgment matrix is ​​used as the second current matrix; in response to the second current matrix, the second current matrix is ​​downsampled through a preset pooling layer to obtain a second intermediate matrix; the second current matrix and the second intermediate matrix are convolved through a preset convolutional layer to obtain a second target matrix; if the matrix dimension of the second target matrix meets the preset matrix dimension condition, the second target matrix is ​​used as the second fusion matrix; if the matrix dimension of the second target matrix does not meet the matrix dimension condition, the second target matrix is ​​used as the new second current matrix.

[0055] In some embodiments, the matrix dimension condition is 2×2.

[0056] In some embodiments, a second judgment matrix with a matrix dimension of 8×8 is used as the second current matrix T6. The second current matrix T6 is downsampled through a pooling layer to obtain a second intermediate matrix T7 with a matrix dimension of 4×4. A convolution is then performed on the second current matrix T6 and the second intermediate matrix T7 to obtain a second target matrix T8 with a matrix dimension of 4×4. The second target matrix T8 is used as the new second current matrix. A pooling layer is then used to downsample the second current matrix T8 to obtain a second intermediate matrix T9 with a matrix dimension of 2×2. A convolution is then performed on the second current matrix T8 and the second intermediate matrix T9 to obtain a second target matrix T10 with a matrix dimension of 2×2. The second target matrix T10 is used as the second fusion matrix, and the geometric mean method is used to calculate the second matrix quantization value corresponding to the second fusion matrix T10. Quantize the second matrix value Mapped to and The first predicted probability is obtained by quantizing the range between the given values.

[0057] In some embodiments, average pooling is used to downsample the second current matrix to achieve matrix dimensionality reduction and feature selection, retaining the correlation features between relative probabilities and avoiding the loss of key features due to max pooling.

[0058] In some embodiments, a valid convolution operation is performed on the second current matrix and the second intermediate matrix to capture the deep correlation between matrix elements and achieve feature correlation extraction.

[0059] In some embodiments, if the second fusion matrix T10 = [[ , ],[ , Then, the geometric mean method is used to calculate the second fusion matrix T10, and the quantized value of the second matrix corresponding to the second fusion matrix T10 is obtained. Among them, the quantization value of the second matrix This represents the intensity value relative to the predicted result; through vehicle collision experiments, the minimum value of the second matrix quantization value in a no-collision-risk scenario is set as... And the maximum value of the second matrix quantization value in the collision risk scenario is taken as The second matrix is ​​quantized using a linear normalization algorithm. Mapped to and The first predicted probability is obtained by quantizing the range between the given values, where the first predicted probability = .

[0060] In some embodiments, evaluating the importance of each collision prediction factor includes: acquiring a vehicle event set, wherein the vehicle event set includes multiple vehicle historical events, and the vehicle historical events include historical vehicle signals of each collision prediction factor; extracting multiple time-series data segments from the vehicle historical events and labeling each vehicle historical event with a collision tag, wherein the collision tag is used to characterize whether the vehicle historical event belongs to a preset collision event type; inputting each time-series data segment into a time-series model with an attention mechanism, and determining the attention score corresponding to each collision prediction factor based on the attention weight distribution output by the time-series model, wherein the time-series model is obtained by training a preset artificial intelligence model; and determining the factor importance corresponding to each collision prediction factor based on each attention score, wherein the factor importance is used to characterize the importance of each collision prediction factor.

[0061] In some embodiments, vehicle history events record historical vehicle signals for multiple collision prediction factors, including collision time, vehicle speed, fatigue monitoring, road friction, etc. Each vehicle history event is labeled with a collision tag. If the vehicle history event ultimately leads to a collision, or triggers the vehicle's emergency braking mechanism, the collision tag for that vehicle history event is set to 1, indicating that the vehicle history event belongs to the collision event type; otherwise, the collision tag is set to 0. An AI model based on the Transformer architecture is used for training. The time-series data fragments obtained in the previous step are used as model input, and their corresponding collision level tags are used as the model's prediction targets, resulting in a time-series model with an attention mechanism. This time-series model can output corresponding collision tags based on new time-series data fragments. Simultaneously, the attention mechanism of the time-series model can reveal which collision prediction factors the model "pays more attention to" when making classification decisions. The attention score for each collision prediction factor is determined, and each attention score is normalized to obtain the factor importance corresponding to each collision prediction factor.

[0062] Optionally, based on the relative importance among the different levels of importance, the relative prediction weights corresponding to each collision prediction factor are determined, including: using each collision prediction factor as the row index and column index of the third judgment matrix, and using the ratio between the different levels of importance as the relative importance, and filling it into the third judgment matrix; calculating the largest eigenvalue corresponding to the third judgment matrix, and calculating the eigenvector corresponding to the largest eigenvalue; normalizing the eigenvector to obtain the relative weight vector, wherein the relative weight vector includes the relative prediction weights corresponding to each collision prediction factor.

[0063] In some embodiments, the first The collision prediction factor relative to the first collision prediction factor The relative importance of each collision predictor is expressed as: ,in, For the first Factor importance of each collision predictor. For the first The factor importance of each collision prediction factor is used to generate a portion of the third judgment matrix based on the relative importance of each risk factor, as shown in Table 3.

[0064] Table 3

[0065] In some embodiments, the first verification rule is expressed as .

[0066] In some embodiments, the second verification rule is expressed as

[0067] In some embodiments, the third verification rule is expressed as [ , , , , , , , ]= [ , , , , , , , ].

[0068] In some embodiments, the relative weight vector corresponding to the third judgment matrix is ​​calculated using the AHP (Analytic Hierarchy Process) algorithm, wherein the AHP algorithm includes the eigenvalue method and the approximation algorithm.

[0069] In some embodiments, the eigenvalue method includes calculating a third judgment matrix. The largest eigenvalue According to the third judgment matrix and the largest eigenvalue Solve the third judgment matrix The eigenvectors are then normalized so that the sum of all elements is 1. The resulting column vector is the relative weight vector. .

[0070] In some embodiments, the relative weight vector is normalized. The sum of the vector elements in the vector is set to 100%, where the relative weight vector... It reflects the relative importance relationship between the various collision prediction factors and is a weighted micro-representation based on the AHP algorithm.

[0071] In some embodiments, the approximation algorithm includes: normalizing each column of the third judgment matrix, summing the normalized matrix by row, and normalizing the vector obtained after summing again to obtain an approximate value of the relative weight vector.

[0072] In some embodiments, a comprehensive prediction result is obtained by combining the first prediction probability and the second prediction probability, including: the comprehensive prediction result. ,in, The probability weight for the first predicted probability. The first predicted probability, The probability weights for the second predicted probability. This represents the second predicted probability.

[0073] In some embodiments, if the pre-preparation state of the airbag is set to the execution state, at least one of the following is performed: controlling the airbag controller to switch from a dormant state to an activated state; controlling the power supply voltage of the airbag power supply circuit according to a preset airbag detonation voltage via the airbag controller, wherein the airbag power supply circuit is used to supply power to the airbag at the vehicle terminal; charging the capacitor of the airbag igniter via the airbag controller, wherein the airbag igniter is used to trigger the gas generator to inflate the airbag; and reallocating the computing resources of the airbag controller.

[0074] In some embodiments, the airbag controller is configured at the vehicle terminal as follows: Figure 5 As shown, the airbag controller, as the core of the airbag control system, is in a low-power standby mode under normal conditions to reduce the energy consumption of the vehicle. When a collision signal is received, the airbag power supply circuit is activated, and the capacitor of the airbag igniter is charged through high voltage triggering, so that the airbag igniter triggers the gas generator to inflate the airbag. Based on this, this application adds a risk assessment strategy, so that the airbag controller is prepared in advance when a collision risk may occur, thereby reducing the time required to trigger the airbag.

[0075] Combination Figure 6 As shown, this application provides a control system for an airbag controller, including an airbag controller, an airbag, a vehicle power supply, a risk assessment module, a collision sensor, and an electronic control unit (ECU).

[0076] The airbag controller includes a microcontroller unit (MCU) and a backup power supply.

[0077] The microcontroller unit is used to receive the pre-preparation signal sent by the risk assessment module, perform pre-preparation actions related to the airbag, and receive the vehicle collision signal sent by the collision sensor to issue the deployment command to the airbag.

[0078] Backup power supply, used to supply power to the airbag controller when the vehicle's power supply fails.

[0079] Airbags, including front airbags and side airbags, are used to rapidly inflate via a gas generator to protect passengers inside the vehicle.

[0080] Vehicle power supply, used to power the airbag controller.

[0081] The risk assessment module is used to wake up the airbag controller and send a pre-preparation command to the airbag controller if the overall risk probability is greater than the preset risk probability threshold.

[0082] Collision sensors are used to send vehicle collision signals to the airbag controller.

[0083] The electronic control unit (ECU) is connected to the airbag controller via a CAN bus. The ECU is used to send vehicle status information (such as vehicle speed, steering wheel angle, braking status, etc.) as an auxiliary basis for the airbag controller to make collision judgments.

[0084] Combination Figure 7 As shown, this application provides a method for determining the triggering state of an airbag, including: Step S701: Obtain multiple collision prediction factors from the vehicle terminal; Step S702: Evaluate the probability of a collision event occurring at the vehicle terminal based on the collision prediction factor, obtain the collision prediction result, and perform feature fusion based on the relative prediction results between the collision prediction results to obtain the first prediction probability. Step S703: Determine the relative prediction weights based on the relative importance of each collision prediction factor, and then perform a weighted calculation on the collision prediction results based on the relative prediction weights to obtain the second prediction probability. Step S704: Combine the first prediction probability and the second prediction probability to calculate the comprehensive prediction result; Step S705: Determine whether the comprehensive prediction result is greater than or equal to the prediction result threshold. If yes, proceed to step S706; otherwise, proceed to step S702. Step S706: Implement the airbag pre-preparation strategy; Step S707: Obtain multiple collision detection factors from the vehicle terminal; Step S708: Detect whether a collision event has occurred at the vehicle terminal according to each collision detection factor, and obtain the collision detection results corresponding to each collision detection factor. Step S709: Perform feature fusion based on the relative detection results between each collision detection result to obtain a comprehensive detection result; Step S710: Determine the airbag's triggering status based on the comprehensive test results; Step S711: Determine whether the airbag is in the activated state. If yes, proceed to step S712; otherwise, proceed to step S702. Step S712: Deploy the airbag.

[0085] The airbag triggering state determination method provided in this application detects whether a collision event has occurred by detecting each collision detection factor separately, obtaining their respective collision detection results. Unlike directly fusing independent detection results, this method further establishes the correlation between the collision detection results through relative detection results, and then performs feature fusion on the relative detection results. Based on the comprehensive detection result obtained from feature fusion, the airbag triggering state is determined. By introducing relative detection results and performing feature fusion, this method changes the previous approach of treating each collision detection factor as an isolated signal. It effectively quantifies the inherent logical correlation between different sensor signals in a collision event, resulting in a more accurate and comprehensive comprehensive detection result that reflects the vehicle's true collision state, thereby improving the reliability of airbag triggering decisions.

[0086] Combination Figure 8 As shown, this application provides a system for determining the triggering state of an airbag, including an acquisition module 801, a detection module 802, a fusion module 803, and a determination module 804.

[0087] The acquisition module 801 is used to acquire multiple collision detection factors of the vehicle terminal.

[0088] The detection module 802 is used to detect whether a collision event has occurred at the vehicle terminal according to each collision detection factor, and to obtain the collision detection result corresponding to each collision detection factor.

[0089] The fusion module 803 is used to perform feature fusion based on the relative detection results between each collision detection result to obtain a comprehensive detection result.

[0090] The determination module 804 is used to determine the triggering state of the airbag based on the comprehensive test results.

[0091] The airbag triggering state determination system provided in this application detects whether a collision event has occurred by detecting each collision detection factor separately, obtaining their respective collision detection results. Unlike directly fusing independent detection results, this system further establishes the correlation between the collision detection results through relative detection results, and then performs feature fusion on the relative detection results. Based on the comprehensive detection result obtained from feature fusion, the airbag triggering state is determined. By introducing relative detection results and performing feature fusion, this approach changes the previous method of treating each collision detection factor as an isolated signal. It effectively quantifies the inherent logical correlation between different sensor signals in a collision event, resulting in a more accurate and comprehensive comprehensive detection result that reflects the vehicle's true collision state, thereby improving the reliability of airbag triggering decisions.

[0092] This application also provides a vehicle terminal, including: a processor and a memory; the memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory, so that the vehicle terminal performs the above-described method.

[0093] Figure 9 A schematic diagram of a computer system suitable for implementing the vehicle terminal embodiments of this application is shown. It should be noted that... Figure 9 The computer system 900 of the vehicle terminal shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.

[0094] like Figure 9 As shown, the computer system 900 includes a Central Processing Unit (CPU) 901, which can perform various appropriate actions and processes, such as executing the methods described in the above embodiments, based on programs stored in Read-Only Memory (ROM) 902 or programs loaded from storage portion 908 into Random Access Memory (RAM) 903. The RAM 903 also stores various programs and data required for system operation. The CPU 901, ROM 902, and RAM 903 are interconnected via a bus 904. An Input / Output (I / O) interface 905 is also connected to the bus 904.

[0095] The following components are connected to I / O interface 905: an input section 906 including a keyboard, mouse, etc.; an output section 907 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 908 including a hard disk, etc.; and a communication section 909 including a network interface card such as a LAN (Local Area Network) card, modem, etc. The communication section 909 performs communication processing via a network such as the Internet. A drive 910 is also connected to I / O interface 905 as needed. Removable media 911, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., are installed on drive 910 as needed so that computer programs read from them can be installed into storage section 908 as needed.

[0096] The vehicle terminal disclosed in this embodiment includes a processor, a memory, a transceiver, and a communication interface. The memory and communication interface are connected to the processor and transceiver and complete communication between them. The memory is used to store computer programs, the communication interface is used for communication, and the processor and transceiver are used to run the computer programs, enabling the vehicle terminal to perform the various steps of the above method. The above description and drawings fully illustrate the embodiments of this disclosure to enable those skilled in the art to practice them. Other embodiments may include structural, logical, electrical, procedural, and other changes. The embodiments represent only possible variations. Unless explicitly required, individual components and functions are optional, and the order of operation may vary. Parts and subsamples of some embodiments may be included in or replace parts and subsamples of other embodiments. Moreover, the terminology used in this application is only for describing embodiments and is not intended to limit the claims. As used in the description of embodiments and claims, the singular forms “a,” “an,” and “the” are intended to equally include the plural forms unless the context clearly indicates otherwise. Similarly, the term “and / or” as used in this application means including one or more of the associated listed items and all possible combinations thereof. Additionally, when used in this application, the term "comprise" and its variations "comprises" and / or "comprising" refer to the presence of stated subsamples, wholes, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other subsamples, wholes, steps, operations, elements, components, and / or groups thereof. Without further limitations, an element defined by the phrase "comprising a..." does not exclude the presence of other identical elements in the process, method, or apparatus that includes the element. In this document, each embodiment may focus on the differences from other embodiments, and similar or identical parts between embodiments can be referred to mutually. For methods, products, etc., disclosed in the embodiments, if they correspond to the method section disclosed in the embodiments, the relevant parts can be referred to the description of the method section.

[0097] 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 in 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. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0098] The methods and products (including but not limited to devices and equipment) disclosed in the embodiments herein can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For instance, the division of units may be merely a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some sub-samples may be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces, and the indirect coupling or communication connection of devices or units may be electrical, mechanical, or other forms. Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to implement this embodiment according to actual needs. Furthermore, the functional units in this application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0099] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of the systems, methods, and computer program products according to this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing the specified logical function. In some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. In the descriptions corresponding to the flowcharts and block diagrams in the accompanying drawings, the operations or steps corresponding to different blocks may also occur in a different order than those disclosed in the description; sometimes there is no specific order between different operations or steps. For example, two consecutive operations or steps may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. Each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

Claims

1. A method for determining the triggering state of an airbag, characterized in that, include: Acquire multiple collision detection factors from the vehicle terminal; Based on each of the collision detection factors, the vehicle terminal is detected to determine whether a collision event has occurred, and the collision detection result corresponding to each of the collision detection factors is obtained. Feature fusion is performed based on the relative detection results among the various collision detection results to obtain a comprehensive detection result; The triggering state of the airbag is determined based on the comprehensive test results.

2. The method according to claim 1, characterized in that, Based on each of the collision detection factors, the vehicle terminal is detected to determine whether a collision event has occurred, and the collision detection results corresponding to each of the collision detection factors are obtained, including at least one of the following: The detection-related signals corresponding to the collision detection factor are compared according to a preset detection-related signal threshold, so as to evaluate whether the vehicle terminal has a collision event based on the comparison result, and obtain the collision detection result corresponding to the collision detection factor. The detection-related signals are obtained by data acquisition from the vehicle terminal. Using a preset collision detection model, the probability of a collision event occurring at the vehicle terminal is evaluated based on the detection-related signals corresponding to the collision detection factors, and the collision detection result corresponding to the collision detection factors is obtained. The collision detection model is obtained by training a preset neural network model.

3. The method according to claim 1, characterized in that, Feature fusion is performed based on the relative detection results among the various collision detection results to obtain a comprehensive detection result, including: Each of the collision detection factors is used as the row index and column index of the first judgment matrix, and the ratio between the collision detection results is used as the relative detection result and filled into the first judgment matrix; The first judgment matrix is ​​fused based on a preset feature fusion step to obtain a first fusion matrix, wherein the feature fusion step includes downsampling and convolution operations; The first fusion matrix is ​​calculated using the geometric mean method to obtain the first matrix quantization value, and the first matrix quantization value is mapped to a preset first quantization value range to obtain the comprehensive detection result.

4. The method according to claim 3, characterized in that, Based on a preset feature fusion step, the first judgment matrix is ​​fused to obtain a first fusion matrix, including: Use the first judgment matrix as the first current matrix; In response to the first current matrix, the first current matrix is ​​downsampled through a preset pooling layer to obtain the first intermediate matrix; The first target matrix is ​​obtained by performing a convolution operation on the first current matrix and the first intermediate matrix; If the matrix dimension of the first target matrix meets the preset matrix dimension condition, then the first target matrix is ​​used as the first fusion matrix; If the matrix dimension of the first target matrix does not meet the matrix dimension condition, then the first target matrix is ​​used as the new first current matrix.

5. The method according to claim 1, characterized in that, The airbag triggering state is determined based on the comprehensive test results, including: If the overall detection result is greater than or equal to the preset detection result threshold, the trigger state of the airbag is set to the open state. If the overall detection result is less than the detection result threshold, the triggering state of the airbag is set to the off state.

6. The method according to any one of claims 1 to 5, characterized in that, Before determining the airbag triggering state based on the comprehensive test results, the method further includes: Obtain multiple collision prediction factors for the vehicle terminal; The probability of a collision event occurring at the vehicle terminal is evaluated based on each of the collision prediction factors to obtain the collision prediction result corresponding to each of the collision prediction factors. Feature fusion is then performed based on the relative prediction results between the collision prediction results to obtain the first prediction probability. The importance of each of the collision prediction factors is evaluated, and the relative prediction weight of each of the collision prediction factors is determined according to the relative importance between the importance factors. The collision prediction results are then weighted according to the relative prediction weights to obtain a second prediction probability. The comprehensive prediction result is obtained by combining the first prediction probability and the second prediction probability. If the comprehensive prediction result is greater than or equal to the preset prediction result threshold, then the pre-preparation state of the airbag is set to the execution state.

7. The method according to claim 6, characterized in that, Based on the relative prediction results among the collision prediction results, feature fusion is performed to obtain a first prediction probability, including: Each of the collision prediction factors is used as the row index and column index of the second judgment matrix, and the ratio between the collision prediction results is used as the relative prediction result and filled into the second judgment matrix; The second judgment matrix is ​​fused based on the preset feature fusion steps to obtain the second fusion matrix; The second fusion matrix is ​​calculated using the geometric mean method to obtain the second matrix quantization value, and the second matrix quantization value is mapped to a preset second quantization value interval to obtain the first prediction probability.

8. The method according to claim 6, characterized in that, Based on the relative importance among the aforementioned importance levels, determine the relative prediction weights corresponding to each of the collision prediction factors, including: Each of the collision prediction factors is used as the row index and column index of the third judgment matrix, and the ratio between the importance levels is used as the relative importance level and filled into the third judgment matrix; Calculate the largest eigenvalue corresponding to the third judgment matrix, and calculate the eigenvector corresponding to the largest eigenvalue; The feature vector is normalized to obtain a relative weight vector, wherein the relative weight vector includes the relative prediction weights corresponding to each of the collision prediction factors.

9. A system for determining the triggering state of an airbag, characterized in that, include: The acquisition module is used to acquire multiple collision detection factors from the vehicle terminal. The detection module is used to detect whether a collision event has occurred in the vehicle terminal according to each of the collision detection factors, and to obtain the collision detection results corresponding to each of the collision detection factors. The fusion module is used to perform feature fusion based on the relative detection results between the various collision detection results to obtain a comprehensive detection result; The determination module is used to determine the triggering state of the airbag based on the comprehensive detection results.

10. A vehicle terminal, characterized in that, include: Processor and memory; The memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory to cause the vehicle terminal to perform the method as described in any one of claims 1-8.

Images