Airbag control method, device, electronic device and storage medium

Abstract

This application relates to the field of vehicle technology, and in particular to a method, device, electronic device, and storage medium for controlling airbags. The method includes: identifying the current collision condition of the vehicle; acquiring current collision data corresponding to the current collision condition within a preset sampling interval; matching a corresponding target airbag deployment algorithm based on the current collision data; and controlling the vehicle to deploy the airbag when the current collision data meets preset ignition conditions. This solves the problems of collision control algorithms in related technologies struggling to distinguish the characteristics of different collision conditions and meet provided RTTFs, and the increased complexity of prediction models due to numerous auxiliary algorithms, thus reducing the robustness of the deployment algorithm. By constructing a support vector machine-based airbag deployment algorithm through model training and feature selection, the method effectively distinguishes the characteristics of different collision conditions and improves the robustness of the airbag deployment algorithm.

Description

Technical Field

[0001] This application relates to the field of vehicle technology, and in particular to a method, device, electronic device and storage medium for controlling an airbag. Background Technology

[0002] With the widespread use of vehicles, vehicle safety performance has become one of the important factors for users when making a choice. Among them, the car airbag is a passive safety device in the passive safety system of a car, which provides a supplementary safety measure and is often used as a last resort in a collision accident. Since it is directly related to the life safety of users, it is particularly important whether the airbag can be detonated at the right time.

[0003] In related technologies, when a vehicle is involved in a frontal collision, such as Figure 1 As shown, most systems receive and analyze acceleration signals from the front collision sensor and built-in sensors through the airbag controller. Without user intervention, the controller sends ignition commands to restraint systems such as airbags and seatbelt warning devices, while simultaneously triggering fuel and power cut-off devices to disconnect the fuel system circuit and electrical system circuit, thereby achieving the purpose of protecting personnel and system safety.

[0004] However, the collision control algorithms used in related technologies have difficulty distinguishing the characteristics of different collision conditions and meeting the provided RTTF (Required Time To Fire). At the same time, a large number of auxiliary algorithms increase the complexity of the prediction model, thereby reducing the robustness of the ignition algorithm, which urgently needs to be addressed. Summary of the Invention

[0005] This application provides a method, device, electronic device, and storage medium for controlling airbags, in order to solve the problems that the collision control algorithms used in related technologies are difficult to distinguish the characteristics of different collision conditions and meet the provided RTTF, while a large number of auxiliary algorithms increase the complexity of the prediction model, thereby reducing the robustness of the detonation algorithm.

[0006] The first aspect of this application provides a method for controlling an airbag, including the following steps:

[0007] Identify the vehicle's current collision condition;

[0008] Obtain current collision data corresponding to the current collision condition within a preset sampling interval, match the corresponding target airbag deployment algorithm based on the current collision data, and determine whether the current collision data meets the preset ignition conditions through the target airbag deployment algorithm.

[0009] If the current collision data meets the preset ignition conditions, then the vehicle's airbags are deployed.

[0010] Optionally, before matching the corresponding target airbag deployment algorithm based on the current collision data, the method further includes:

[0011] Obtain the preset ignition conditions corresponding to each preset collision condition of the vehicle;

[0012] The airbag deployment training dataset is obtained according to the preset ignition conditions corresponding to each preset collision condition. The airbag deployment training dataset is input into the preset airbag prediction deployment model to obtain the current support vector. Based on the current support vector, the feature data corresponding to each preset collision condition is determined.

[0013] The airbag deployment algorithm for each preset collision condition is constructed based on the feature data corresponding to each preset collision condition.

[0014] Optionally, obtaining the airbag deployment training dataset based on the preset ignition conditions corresponding to each preset collision condition includes:

[0015] Obtain the collision data matrix corresponding to each preset collision condition, and simulate the mirror collision data matrix corresponding to each preset collision condition based on the collision data matrix corresponding to each preset collision condition;

[0016] The mirror collision data corresponding to each preset collision condition is obtained based on the mirror collision data matrix corresponding to each preset collision condition, and the scaling ratio of the collision data and the mirror collision data corresponding to each preset collision condition is determined based on the preset ignition conditions corresponding to each preset collision condition.

[0017] Based on the scaling ratio of the collision data and mirror collision data corresponding to each preset collision condition, the collision data and mirror collision data of the corresponding preset collision condition are scaled, and the collision data matrix set corresponding to each preset collision condition is output. Then, the collision data matrix set corresponding to each preset collision condition is traversed, and the first filtered data and the second filtered data corresponding to each preset collision condition are output.

[0018] The target vector of the first filtered data corresponding to each preset collision condition is obtained at the target time. Based on the target vector, the open / closed state of the airbag deployment algorithm corresponding to each preset collision condition is determined. When the airbag deployment algorithm is in the open state, the target filtered data of the second filtered data of the preset collision condition corresponding to the open state in the airbag deployment algorithm for each preset collision condition is obtained at the target time. The collision feature data at the target time is calculated based on the target filtered data and the collision data corresponding to each preset collision condition.

[0019] The feature vector of the collision feature data is determined, and when the collision feature data sample is generated at the target time, the collision feature data training sample at the target time is obtained based on the collision feature data sample and the feature vector of the collision feature data. The collision feature data training sample and all running times of the airbag deployment algorithm corresponding to each preset collision condition are traversed to obtain the airbag deployment training dataset corresponding to each preset collision condition.

[0020] Optionally, before obtaining the collision feature data training samples for the target time based on the collision feature data samples and the feature vectors of the collision feature data, the method further includes:

[0021] If the collision feature data sample is not generated at the target time, the airbag deployment algorithm corresponding to each preset collision condition is controlled to enter the next moment of the target time, and the step of obtaining the target vector of the first filtered data corresponding to each preset collision condition at the target time continues to be executed.

[0022] Optionally, determining the current feature data corresponding to each preset collision condition based on the current support vector includes:

[0023] Based on each preset collision condition, the vehicle's ignition condition and non-ignition condition are obtained, and a first feature data belonging to the ignition condition and a first label sample in the support vector is obtained, and a second feature data belonging to the non-ignition condition and a second label sample in the support vector is obtained.

[0024] The first feature data and the second feature data are examined to obtain the difference value between the first feature data and the second feature data. When the difference value is less than a preset threshold, the current feature data corresponding to each preset collision condition is determined based on the difference value.

[0025] The airbag control method according to embodiments of this application identifies the current collision condition of the vehicle, obtains the current collision data corresponding to the current collision condition within a preset sampling interval, matches the corresponding target airbag deployment algorithm based on the current collision data, and controls the vehicle to deploy the airbag when the current collision data meets preset ignition conditions. This solves the problems of collision control algorithms used in related technologies, such as difficulty in distinguishing the characteristics of different collision conditions and meeting the provided RTTF, and the increased complexity of prediction models due to numerous auxiliary algorithms, thereby reducing the robustness of the deployment algorithm. By constructing a support vector machine-based airbag deployment algorithm through model training and feature selection, the method effectively distinguishes the characteristics of different collision conditions and improves the robustness of the airbag deployment algorithm.

[0026] A second aspect of this application provides a control device for an airbag, comprising:

[0027] The identification module is used to identify the current collision condition of the vehicle;

[0028] The judgment module is used to acquire the current collision data corresponding to the current collision condition within a preset sampling interval, match the corresponding target airbag deployment algorithm according to the current collision data, and determine whether the current collision data meets the preset ignition conditions through the target airbag deployment algorithm.

[0029] The control module is used to control the vehicle to deploy the airbags if the current collision data meets the preset ignition conditions.

[0030] Optionally, before matching the corresponding target airbag deployment algorithm based on the current collision data, the determination module further includes:

[0031] The acquisition unit is used to acquire the preset ignition conditions corresponding to each preset collision condition of the vehicle;

[0032] The determining unit is used to obtain the airbag deployment training dataset according to the preset ignition conditions corresponding to each preset collision condition, input the airbag deployment training dataset into the preset airbag prediction deployment model to obtain the current support vector, and determine the feature data corresponding to each preset collision condition based on the current support vector.

[0033] The construction unit is used to construct the airbag deployment algorithm corresponding to each preset collision condition based on the feature data corresponding to each preset collision condition.

[0034] Optionally, the determining unit includes:

[0035] The first acquisition subunit is used to acquire the collision data matrix corresponding to each preset collision condition, and simulate the mirror collision data matrix corresponding to each preset collision condition based on the collision data matrix corresponding to each preset collision condition.

[0036] The second acquisition subunit is used to obtain the mirror collision data corresponding to each preset collision condition according to the mirror collision data matrix corresponding to each preset collision condition, and to determine the scaling ratio of the collision data and the mirror collision data corresponding to each preset collision condition based on the preset ignition conditions corresponding to each preset collision condition.

[0037] The first traversal subunit is used to scale the collision data and mirror collision data of the corresponding preset collision condition based on the scaling ratio of the collision data and mirror collision data of each preset collision condition, output the collision data matrix set of each preset collision condition, and traverse the collision data matrix set of each preset collision condition to output the first filtered data and the second filtered data of each preset collision condition.

[0038] The third acquisition subunit is used to acquire the target vector of the first filtered data corresponding to each preset collision condition at the target time, determine the opening / closing state of the airbag deployment algorithm corresponding to each preset collision condition based on the target vector, and when the airbag deployment algorithm is in the open state, acquire the target filtered data of the second filtered data of the preset collision condition corresponding to the open state of the airbag deployment algorithm in each preset collision condition at the target time, and calculate the collision feature data at the target time based on the target filtered data and the collision data corresponding to each preset collision condition.

[0039] The second traversal subunit is used to determine the feature vector of the collision feature data, and when generating the collision feature data sample at the target time, obtain the collision feature data training sample at the target time based on the collision feature data sample and the feature vector of the collision feature data, and traverse all running times of the collision feature data training sample and the airbag deployment algorithm corresponding to each preset collision condition to obtain the airbag deployment training dataset corresponding to each preset collision condition.

[0040] Optionally, before obtaining the collision feature data training samples at the target time based on the collision feature data samples and the feature vectors of the collision feature data, the second traversal subunit further includes:

[0041] The control sub-component is configured to, if the collision feature data sample is not generated at the target time, control the airbag deployment algorithm corresponding to each preset collision condition to enter the next moment of the target time and continue to execute the step of obtaining the target vector of the first filtered data corresponding to each preset collision condition at the target time.

[0042] Optionally, the determining unit includes:

[0043] The fourth acquisition subunit is used to acquire the ignition condition and non-ignition condition of the vehicle based on each preset collision condition, and to acquire the first feature data of the first label sample belonging to the ignition condition in the support vector, and to acquire the second feature data of the second label sample belonging to the non-ignition condition in the support vector.

[0044] A subunit is defined to examine the first feature data and the second feature data to obtain the difference value between the first feature data and the second feature data, and when the difference value is less than a preset threshold, to determine the current feature data corresponding to each preset collision condition based on the difference value.

[0045] The airbag control device according to the embodiments of this application identifies the current collision condition of the vehicle, obtains the current collision data corresponding to the current collision condition within a preset sampling interval, matches the corresponding target airbag deployment algorithm based on the current collision data, and controls the vehicle to deploy the airbag when the current collision data meets the preset ignition conditions. This solves the problems of collision control algorithms used in related technologies, such as difficulty in distinguishing the characteristics of different collision conditions and meeting the provided RTTF, and the increased complexity of prediction models due to numerous auxiliary algorithms, thereby reducing the robustness of the deployment algorithm. By constructing a support vector machine-based airbag deployment algorithm through model training and feature selection, the device effectively distinguishes the characteristics of different collision conditions and improves the robustness of the airbag deployment algorithm.

[0046] A third aspect of this application provides an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the airbag control method as described in the above embodiments.

[0047] A fourth aspect of this application provides a computer-readable storage medium storing computer instructions for causing the computer to perform the airbag control method as described in the above embodiments.

[0048] A fifth aspect of this application provides a computer program product, including a computer program that is executed to implement the airbag control method described in the above embodiments.

[0049] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through practice of the present application. Attached Figure Description

[0050] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0051] Figure 1 This is a schematic diagram showing the distribution of peripheral sensors and ECUs (Electronic Control Units) for related technologies.

[0052] Figure 2This is a schematic diagram of the FRB (Full-width Rigid Barrier, 100% overlap rigid barrier collision test) collision condition for related technologies.

[0053] Figure 3 This is a schematic diagram of a frontal collision with offset for related technologies.

[0054] Figure 4 This is a schematic diagram of the frontal collision condition of the drill bit in the relevant technology.

[0055] Figure 5 This is a flowchart of a method for controlling an airbag according to an embodiment of this application;

[0056] Figure 6 This is a flowchart illustrating the training process of a prediction model according to an embodiment of this application.

[0057] Figure 7 An internal flowchart for generating a model from training data according to one embodiment of this application;

[0058] Figure 8 This is a flowchart of a feature filtering process according to an embodiment of this application;

[0059] Figure 9 This is a block diagram of an airbag control device according to an embodiment of this application;

[0060] Figure 10 This is a schematic diagram of the structure of an electronic device according to an embodiment of this application. Detailed Implementation

[0061] The embodiments of this application are described in detail below. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0062] The following description, with reference to the accompanying drawings, outlines an airbag control method, apparatus, electronic device, and storage medium according to embodiments of this application. Addressing the issues mentioned in the background art, such as the difficulty in distinguishing the characteristics of different collision conditions and meeting the provided RTTF (Real-Time Tolerance) in collision control algorithms, and the increased complexity of prediction models due to numerous auxiliary algorithms, thus reducing the robustness of the airbag deployment algorithm, this application provides an airbag control method. In this method, the current collision condition of the vehicle is identified, current collision data corresponding to the current collision condition is obtained within a preset sampling interval, a corresponding target airbag deployment algorithm is matched based on the current collision data, and the vehicle is controlled to deploy the airbag when the current collision data meets preset ignition conditions. This solves the problems of the collision control algorithms in the related art being unable to distinguish the characteristics of different collision conditions and meet the provided RTTF, and the increased complexity of prediction models due to numerous auxiliary algorithms, thus reducing the robustness of the airbag deployment algorithm. By constructing a support vector machine-based airbag deployment algorithm through model training and feature selection, the characteristics of different conditions are effectively distinguished, improving the robustness of the airbag deployment algorithm.

[0063] Specifically, before introducing the embodiments of this application, we will first introduce the airbag deployment strategies in related technologies, such as... Figure 1 As shown, the strategy adopted by most OEMs is that when a frontal collision occurs, the airbag controller receives and analyzes the acceleration signals from the front collision sensor and the built-in sensor. Without user intervention, it sends ignition commands to restraint systems such as airbags and seat belt warning devices, and at the same time triggers the fuel cut-off and power cut-off devices to disconnect the fuel system circuit and the electrical system circuit, thereby achieving the purpose of protecting the safety of people and systems.

[0064] Furthermore, during the development of airbag controllers, it is necessary to match the parameters in the collision control algorithm with the vehicle model. That is, in order to meet the RTTF (Real-Time Tolerance Function) of a specific vehicle model, the collision control algorithm used by the controller and its parameters, usually the threshold line, need to be adjusted. Currently used collision control algorithms use relatively simple features, which often make it difficult to distinguish different collision conditions.

[0065] For example, such as Figure 2 The 16FRB and 50FRB shown represent collisions at 16 km / h and 50 km / h respectively, where the vehicle's frontal collision with a rigid barrier results in 100% overlap. 16FRB is generally considered a low-speed collision, requiring no triggering of any restraint systems. Figure 3 As shown, 64SORB (Small Overlap Rigid Barrier) represents a vehicle undergoing an offset frontal collision with a rigid barrier at 64 km / h; Figure 4As shown, 35URT (Under Riding) The scenario of a vehicle crashing into a truck at 35 km / h is described as a medium-to-high-speed collision (35URT). This requires issuing an ignition command to the restraint system within a suitable timeframe. This is achieved by calculating the integral value of the acceleration in the x-direction from the sensors in the ECU and combining this with the acceleration of the front impact sensor and the built-in sensors at the corresponding moment to determine if each ignition circuit needs to be triggered. However, within the 35URT ignition timeframe, the 16FRB signal is often larger than that of the 35URT. Technical personnel need to sift through numerous data features for calibration, making the airbag controller performance more susceptible to the experience and skill level of the technicians, resulting in a significant workload and a longer calibration process. Furthermore, the use of numerous auxiliary algorithms increases the overall complexity of the model. For example, the ignition algorithm in the ECU includes a frontal collision auxiliary algorithm module, which uses the acceleration on the y-axis of the ECU's accelerometer and the difference in acceleration between the left and right front impact sensors for auxiliary judgment. The large number of auxiliary algorithms increases the overall complexity of the model, leading to decreased robustness of the ignition algorithm and increased computational resource consumption.

[0066] Therefore, in actual calibration, simple logic often fails to fully meet the RTTF (Release Time To Fulfillment) for various operating conditions. In order for the controller to distinguish as many operating conditions as possible and meet the RTTF as much as possible, relevant technicians need to select suitable features from a large number of features for calibration. This not only brings a huge workload to the relevant technicians and results in a long calibration process (usually more than 7 weeks), but also makes the performance of the airbag controller more susceptible to the influence of the experience and skill level of the relevant technicians.

[0067] On the other hand, machine learning-based collision control algorithms can alleviate the aforementioned problems, but directly applying machine learning models to airbag controllers also presents robustness and interpretability issues. Complex machine learning models can achieve excellent performance on training samples, but they do not necessarily have good generalization properties. When the nature of the collision signal changes slightly, the model's prediction results may show huge differences. In addition, compared to building a detonation algorithm by combining simple logic, the more complex calculation mechanism inside the machine learning model is not conducive to after-sales service interpretation.

[0068] Therefore, based on the aforementioned issues, this application provides a feature selection technique for airbag control algorithms based on support vector machines. This technique integrates traditional calibration processes with machine learning algorithms, offering a high degree of automation and a short development cycle. It can reduce the impact of the skill level and experience of technical personnel on airbag performance. Furthermore, the trained model is not directly used for airbag deployment but is instead used for feature selection to obtain features that effectively distinguish different operating conditions. This reduces the time required for technical personnel to select features from a feature library. Additionally, the deployment algorithm constructed by combining these features exhibits good robustness and interpretability. The specific implementation process will be described in detail in the following embodiments.

[0069] Specifically, Figure 5 This is a schematic flowchart illustrating a method for controlling an airbag according to an embodiment of this application.

[0070] like Figure 5 As shown, the airbag control method includes the following steps:

[0071] In step S501, the current collision condition of the vehicle is identified.

[0072] Specifically, the airbag control method involved in this application embodiment generates corresponding control strategies based on different vehicle models and different operating conditions. Therefore, in order to facilitate control accuracy, this application embodiment first needs to obtain collision data (including acceleration in the x and y directions of the front collision sensor and the ECU) for multiple collision conditions for calibration.

[0073] In step S502, the current collision data corresponding to the current collision condition within the preset sampling interval is obtained, the corresponding target airbag deployment algorithm is matched according to the current collision data, and the target airbag deployment algorithm is used to determine whether the current collision data meets the preset ignition conditions.

[0074] Optionally, before matching the corresponding target airbag deployment algorithm based on the current collision data, the method further includes: obtaining preset ignition conditions for the vehicle under each preset collision condition; obtaining an airbag deployment training dataset based on the preset ignition conditions for each preset collision condition; inputting the airbag deployment training dataset into a preset airbag prediction deployment model to obtain the current support vector; and determining the feature data corresponding to each preset collision condition based on the current support vector; and constructing an airbag deployment algorithm corresponding to each preset collision condition based on the feature data corresponding to each preset collision condition.

[0075] Among them, the preset collision conditions, preset sampling range and preset ignition conditions can all be set by those skilled in the art according to the ignition requirements in the actual test process, or they can be obtained through computer simulation, and no specific limitation is made here.

[0076] Specifically, the preset sampling interval in this embodiment can be from 200ms before the collision to 500ms after the collision, and the sampling frequency of the collision data is reduced to 2000Hz to adapt to the calculation frequency of the airbag ECU processor. In multiple possible collision scenarios for the vehicle, the corresponding preset ignition conditions are obtained based on the identified current collision scenario, i.e., the RTTF set by relevant technicians, and the corresponding current collision data, such as the acceleration of the left front collision sensor, the acceleration of the right front collision sensor, the x-axis acceleration in the ECU, and the y-axis acceleration in the ECU, during the current collision scenario. Based on the aforementioned current collision data, data is collected within this preset sampling interval, and... The airbag deployment training dataset is obtained based on the preset ignition conditions corresponding to each preset collision condition. Then, the airbag deployment training dataset is input into the preset airbag prediction deployment model to obtain the current support vector. The current support vector includes two classes: positive and negative. That is, the airbag deployment training dataset corresponding to each preset collision condition needs to be constructed based on the RTTF time point, and the feature data corresponding to each preset collision condition is determined in the current support vector to deploy the airbag based on the RTTF time point (label 1), and the airbag does not need to be deployed based on the RTTF time point (label 0), thereby obtaining the airbag deployment algorithm corresponding to each preset collision condition.

[0077] Optionally, the airbag deployment training dataset is obtained based on the preset ignition conditions corresponding to each preset collision condition, including: obtaining the collision data matrix corresponding to each preset collision condition; simulating the mirror collision data matrix corresponding to each preset collision condition based on the collision data matrix corresponding to each preset collision condition; obtaining the mirror collision data corresponding to each preset collision condition based on the mirror collision data matrix corresponding to each preset collision condition; scaling the collision data and mirror collision data of the corresponding preset collision condition based on the scaling ratio of the collision data and mirror collision data of the corresponding preset collision condition; outputting the collision data matrix set corresponding to each preset collision condition; and iterating through the collision data matrix set corresponding to each preset collision condition to output the first airbag deployment training dataset corresponding to each preset collision condition. First, filter data and second, filter data; obtain the target vector of the first filter data corresponding to each preset collision condition at the target time; determine the open / closed state of the airbag deployment algorithm corresponding to each preset collision condition based on the target vector; when the airbag deployment algorithm is open, obtain the target filter data of the second filter data of the preset collision condition corresponding to the airbag deployment algorithm in the open state at the target time; calculate the collision feature data at the target time based on the target filter data and the collision data corresponding to each preset collision condition; determine the feature vector of the collision feature data; when generating collision feature data samples at the target time, obtain the collision feature data training samples at the target time based on the collision feature data samples and the feature vectors of the collision feature data; and iterate through all running times of the airbag deployment algorithm corresponding to each preset collision condition to obtain the airbag deployment training dataset corresponding to each preset collision condition.

[0078] Optionally, before obtaining the collision feature data training sample at the target time based on the collision feature data sample and the feature vector of the collision feature data, the method further includes: if no collision feature data sample is generated at the target time, then controlling the airbag deployment algorithm corresponding to each preset collision condition to enter the next moment of the target time, and continuing to execute the step of obtaining the target vector of the first filtered data corresponding to each preset collision condition at the target time.

[0079] Specifically, such as Figure 6 As shown, the embodiments of this application first perform model training. The model training process mainly includes a training data generation module and an SVM (Support Vector Machine) model training module, wherein, as... Figure 7As shown, the training data generation module includes a data mirroring module S1, a margin construction module S2, a low-pass filter module S3, an algorithm triggering and exit module S4, a feature extraction module S5, and a label generation module S6. The above six modules perform relevant feature extraction and calculation to finally obtain the airbag deployment training dataset corresponding to each preset collision condition.

[0080] Furthermore, the data mirroring module S1 in this embodiment does not include control conditions and parameters; the parameter of the margin construction module S2 is the scaling ratio r, which changes according to the requirements of the RTTF, i.e., the preset ignition conditions given by relevant technicians. When the RTTF requires ignition, the scaling ratio r can be 1; when the RTTF requires no ignition, the scaling ratio r should be greater than 1. For example, the scaling ratio r can be 1, 1.05, 1.1, 1.15, 1.2, 1.25 or 1.3. It should be noted that the above parameter scaling ratios can be adjusted according to the needs of relevant technical personnel. The purpose of scaling the parameters is to improve the robustness of the model. When the parameters are scaled within a certain range, the model can still maintain the properties in RTTF. The low-pass filter module S3 contains two second-order low-pass filters with cutoff frequencies of 60Hz and 14Hz, respectively. The control parameters of the algorithm triggering and exiting module S4 are the trigger threshold, the exit threshold, and the required exit duration. The algorithm will be activated when the acceleration in the x-direction of the front collision sensor or ECU is greater than the trigger threshold. If the acceleration in the x-direction of the front collision sensor and ECU is lower than the exit threshold for a period of time greater than the required duration, the algorithm will exit and no feature extraction calculation will be performed. The feature extraction module S5 does not contain control conditions or parameters. The control parameter of the label generation module S6 is the extension time t. + With t - If the burst interval required by RTTF is [t1, t2], then the tag generation module S6 will set [max(t1-t2] as the burst interval. - ,0),t2+t + The samples within the specified time period are used as the positive class, and the output is a training dataset of airbags that includes both positive and negative samples (labeled 0 or 1).

[0081] Specifically, this embodiment first acquires collision data (including acceleration in the x and y directions of the front collision sensor and the ECU) for multiple preset collision conditions used for calibration. The sampling interval of the collision data can be set from 200ms before the collision to 500ms after the collision, and the sampling frequency of the collision data is reduced to 2000Hz to match the calculation frequency of the airbag ECU processor. Simultaneously, preset ignition conditions given by relevant technicians are acquired. Since both the collision data and the preset ignition conditions are inputs required by relevant technicians for calibration work, this embodiment does not require additional data or documentation. Figure 6 and Figure 7 As shown, the collision data corresponding to each preset collision condition and the preset ignition conditions corresponding to each preset collision condition are input into the training data generation module. Based on the collision data corresponding to each preset collision condition and the preset ignition conditions corresponding to each preset collision condition, the collision data matrix S corresponding to each preset collision condition is obtained. raw,k And the collision data matrix S raw,k The data is input into the data mirroring module S1 to obtain the mirrored collision data matrix S corresponding to each preset collision condition. mir，k And the mirrored collision data matrix S corresponding to each preset collision condition. mir，k Let it be S mir,k =[a r,k ,a l，k , a x，k -a y,k Simultaneously, it outputs the mirror collision data corresponding to each preset collision condition.

[0082] in, This represents the collision data matrix for the k-th operating condition, with rows representing time points and columns representing different signals, a l，k For the acceleration of the left front collision sensor, a r，k For the acceleration of the right-side front collision sensor, a x，k a is the acceleration in the x-axis direction in the ECU. y，k -a represents the y-axis acceleration in the ECU. y，k for a y，k The opposite number.

[0083] Secondly, the mirrored collision data and the collision data corresponding to each preset collision condition obtained above are input into the margin construction module S2. The margin construction module S2 determines the scaling ratio r of the collision data corresponding to each preset collision condition based on RTTF. For example, when RTTF requires detonation, the scaling ratio r can be 1; when RTTF requires no detonation, the scaling ratio r should be greater than 1. For example, the scaling ratio r can be 1, 1.05, 1.1, 1.15, 1.2, 1.25, or 1.3. Then, based on the ratio corresponding to each preset collision condition... The collision data is scaled to output a set of collision data matrices corresponding to each preset collision condition. And the set of collision data matrices corresponding to each preset collision condition output by the margin construction module S2. The process involves iterating through the data and then filtering the collision data set S using a low-pass filter S3, outputting the first filtered data S corresponding to each preset collision condition. trig The second filtered data S corresponding to each preset collision conditionfilt , where the first filtered data S trig The cutoff frequency is 14Hz, and the second filter data S filt The cutoff frequency is 60Hz.

[0084] Furthermore, this application's embodiments are based on the first filtered data S corresponding to each preset collision condition. trig The second filtered data S corresponding to each preset collision condition filt The time points of the airbag deployment algorithm are iterated to obtain the first filtered data S corresponding to each preset collision condition. trig At the current time t, i.e., the target time t, the target vector S trig,t Where, the target vector S trig，t It is a row vector of length 4, and the target vector S is... trig，t The input is sent to the algorithm triggering and exiting module S4 to determine the opening and closing state of the airbag deployment algorithm based on the target vector corresponding to each preset collision condition. The target vector S... trig，t The system includes a first target threshold, a second target threshold, a third target threshold, and a fourth target threshold. If any one of the first, second, or third target thresholds is greater than or equal to the target activation threshold, the airbag deployment algorithm is activated. If the airbag deployment algorithm is activated at target time t, and the first, second, and third target thresholds are all less than the target exit threshold within a preset duration interval, the airbag deployment algorithm is deactivated. At this time, the algorithm triggering and exit module S4 outputs the on / off state of the airbag deployment algorithm at target time t. If the airbag deployment algorithm is deactivated at target time t, it is determined whether the airbag deployment algorithm has reached the termination time point. If the airbag deployment algorithm has reached the termination time point, the system continues to traverse the above-mentioned collision data matrix set and subsequent related steps. Otherwise, the airbag deployment algorithm is controlled to enter the next moment of target time t for calculation. If the airbag deployment algorithm is activated at target time t, the second filtered data S corresponding to each preset collision condition is obtained. filt The target filtered data S at target time t filt,t and the target filtered data S filt，t The input is fed into the feature extraction module S5, which uses S... fiilt，t In addition, collision feature data at target time t is calculated based on vehicle acceleration and other data stored during the airbag deployment algorithm runtime. At this point, relevant technicians can, based on experience, determine the features to be calculated after balancing computational resources and performance; that is, the collision feature data corresponding to each preset collision condition. The most extreme case is to calculate all features in the feature library and denote the feature vector of the calculated collision feature data as... Among them, f 1，t f is the first feature calculated at the target time t. 2，t Let f be the second feature calculated at the target time t, ... n，t For the nth feature calculated at the target time t, the feature extraction module S5 outputs the feature vector x of the collision feature data corresponding to each preset collision condition. t .

[0085] Furthermore, when the airbag deployment algorithm is in the deployed state at the target time t, the tag generation module S6 will determine whether collision feature data samples need to be generated at the target time t based on the RTTF corresponding to each preset collision condition. For example, if the RTTF requires no deployment, then the feature vector x of the collision feature data at all times corresponding to each preset collision condition will be generated. t Both require synthesizing collision feature data training samples and obtaining the labels yt of the collision feature data training samples, and then... t Recorded as 0; if RTTF requires the explosion to occur within the time interval [t1, t2] after the collision, then the interval [max(t1-t2] is 0. - ,0),t2+t + Collision feature data training sample labels y within ] t It is denoted as 1, and the time is less than max(t1-t). - Collision feature data sample label y (0) t If the collision feature data at other times is recorded as 0, and the collision feature data at other times is not used to synthesize collision feature data training samples, then the airbag deployment algorithm corresponding to each preset collision condition is controlled to enter the next time of the target time, and the step of obtaining the target vector of the first filtered data corresponding to each preset collision condition at the target time continues.

[0086] Finally, in this embodiment of the application, the feature vector x of the collision feature data generated by the feature extraction module S5 t The labels y of the training samples are the collision feature data generated by the label generation module S6. t By combining the data, we obtain the collision feature data training samples (x) at the target time t. t ,y t ), and use the collision feature data to train samples (x) t ,y t The data is added to the preset training set, and all running times of the airbag deployment algorithm corresponding to each preset collision condition are traversed to finally obtain a deployment training dataset of airbags containing a large number of positive and negative class (labeled 1 or 0) samples.

[0087] Furthermore, such as Figure 6As shown in the embodiment of this application, the airbag deployment training dataset obtained above is input into the SVM model training module. A binary classification algorithm is constructed using the SVM model with slack variables provided by the SVM model training module. The optimization problem can be expressed as:

[0088]

[0089] st0≤a n ≤C

[0090]

[0091] Where N is the total number of collision feature data samples, y n Let be the label of the nth collision feature data sample, and C be a parameter given by relevant technical personnel. The larger C is, the fewer misclassified samples are allowed. n With a m X is the parameter to be optimized. m Let x be the feature of the m-th sample. n Let k be the feature of the nth sample, and k(·,·) be the kernel function. Here, a Gaussian kernel function can be used, with the following expression:

[0092]

[0093] γ is the scale parameter defined by relevant technical personnel, and x and z are feature vectors. The sequential minimum optimization algorithm is used to solve the above quadratic programming problem. If the optimal solution includes a... n If the value is greater than 0, then the corresponding collision feature data sample (X) n ,y n The SVM model training module outputs all support vectors. Where, N S N represents the number of support vectors. f This represents the number of collision feature data.

[0094] In step S503, if the current collision data meets the preset ignition conditions, the vehicle is controlled to deploy the airbags.

[0095] Optionally, determining the current feature data corresponding to each preset collision condition based on the current support vector includes: acquiring the vehicle's ignition condition and non-ignition condition based on each preset collision condition, and acquiring first feature data belonging to the ignition condition with a first label sample in the support vector, and second feature data belonging to the non-ignition condition with a second label sample in the support vector; verifying the first feature data and the second feature data to obtain the difference value between the first feature data and the second feature data, and determining the current feature data corresponding to each preset collision condition based on the difference value when the difference value is less than a preset threshold.

[0096] The preset airbag deployment prediction model can be selected by those skilled in the art based on actual collision detection needs, and is not specifically limited here.

[0097] Specifically, in this embodiment of the application, after obtaining the airbag deployment training dataset corresponding to each preset collision condition, the airbag deployment training dataset corresponding to each preset collision condition needs to be input into a preset airbag prediction deployment model, such as an SVM model training module, to obtain the current support vector corresponding to each preset collision condition. Based on the current support vector, feature filtering is performed on the feature data to determine the feature data corresponding to each preset collision condition, so as to construct the airbag deployment algorithm corresponding to each preset collision condition based on the feature data corresponding to each preset collision condition.

[0098] Specifically, if Figure 8 As shown, the feature selection process mainly includes a support vector selection module and a feature filtering module. The support vector selection module does not contain control conditions or parameters. The parameter of the feature filtering module is a threshold value of p, which is used to determine whether the feature data corresponding to each preset collision condition is different based on the p value. When the calculated p value is less than the preset threshold (which can be a calibrated threshold or a threshold set by relevant technical personnel), it is considered that the corresponding feature has a significant difference between the two types of support vectors, and the feature data corresponding to each preset collision condition is output.

[0099] Specifically, the input to the support vector screening module in this application embodiment includes the support vectors SV output by the SVM model training module, and conflicting detonation and non-detonation scenarios provided by those skilled in the art. The first feature data in the support vectors SV that is labeled 1 and belongs to a detonation scenario is denoted as SV. p That is, the feature data (positive class support vector) of the airbag that needs to be deployed. The second feature data in SV that has a label of 0 and belongs to the non-deployment condition is denoted as SV. n This refers to the feature data (negative class support vectors) that do not require airbag deployment, and the SV is output by the support vector filtering module. p With SV n At this point, the feature filtering module performs SV filtering. p and SV n For each preset collision condition, a statistical t-test is performed on the current feature data, and the p-value of the current feature data for each preset collision condition is calculated. The p-value of the nth feature is denoted as p. n When p nIf the value is less than a preset threshold, it indicates that the positive and negative support vectors have a significant difference in this feature. The current feature data corresponding to each preset collision condition is output to construct an airbag deployment algorithm based on the current feature data corresponding to each preset collision condition. When the current collision data corresponding to each preset collision condition meets the preset ignition conditions corresponding to each preset collision condition, the airbag deployment algorithm is used to control the vehicle to deploy the airbag.

[0100] In summary, based on the detailed discussion of the embodiments of this application above, the following beneficial effects are achieved:

[0101] (1) The feature screening technology of the airbag control algorithm based on support vector machine proposed in this application is different from the traditional rule-based calibration algorithm. This technology can screen out features that can effectively distinguish different working conditions from the training samples generated based on RTTF more efficiently through machine learning algorithms.

[0102] (2) This application differs from the detonation algorithm based entirely on machine learning. After obtaining the prediction model, it does not directly use the obtained prediction model for airbag detonation. Instead, it uses the prediction model to perform feature screening to obtain features that can effectively distinguish different working conditions. The detonation algorithm is constructed by combining these features, which has good robustness and interpretability.

[0103] The airbag control method according to embodiments of this application identifies the current collision condition of the vehicle, obtains the current collision data corresponding to the current collision condition within a preset sampling interval, matches the corresponding target airbag deployment algorithm based on the current collision data, and controls the vehicle to deploy the airbag when the current collision data meets preset ignition conditions. This solves the problems of collision control algorithms used in related technologies, such as difficulty in distinguishing the characteristics of different collision conditions and meeting the provided RTTF, and the increased complexity of prediction models due to numerous auxiliary algorithms, thereby reducing the robustness of the deployment algorithm. By constructing a support vector machine-based airbag deployment algorithm through model training and feature selection, the method effectively distinguishes the characteristics of different collision conditions and improves the robustness of the airbag deployment algorithm.

[0104] Next, the airbag control device according to the embodiments of this application is described with reference to the accompanying drawings.

[0105] Figure 9 This is a block diagram of the airbag control device according to an embodiment of this application.

[0106] like Figure 9 As shown, the control device 10 of the airbag includes: an identification module 100, a judgment module 200 and a control module 300.

[0107] The identification module 100 is used to identify the current collision condition of the vehicle.

[0108] The judgment module 200 is used to acquire the current collision data corresponding to the current collision condition within the preset sampling interval, match the corresponding target airbag detonation algorithm based on the current collision data, and determine whether the current collision data meets the preset ignition conditions through the target airbag detonation algorithm.

[0109] The control module 300 is used to control the vehicle to deploy the airbags if the current collision data meets the preset ignition conditions.

[0110] Optionally, before matching the corresponding target airbag deployment algorithm based on the current collision data, the determination module 200 further includes:

[0111] The acquisition unit is used to acquire the preset ignition conditions corresponding to each preset collision condition of the vehicle.

[0112] The determination unit is used to obtain the airbag deployment training dataset according to the preset ignition conditions corresponding to each preset collision condition, input the airbag deployment training dataset into the preset airbag prediction deployment model, obtain the current support vector, and determine the feature data corresponding to each preset collision condition based on the current support vector.

[0113] The construction unit is used to construct the airbag deployment algorithm corresponding to each preset collision condition based on the feature data corresponding to each preset collision condition.

[0114] Optionally, the determined unit includes:

[0115] The first acquisition subunit is used to acquire the collision data matrix corresponding to each preset collision condition, and to simulate the mirror collision data matrix corresponding to each preset collision condition based on the collision data matrix corresponding to each preset collision condition.

[0116] The second acquisition subunit is used to obtain the mirror collision data corresponding to each preset collision condition according to the mirror collision data matrix corresponding to each preset collision condition, and to determine the scaling ratio of the collision data and the mirror collision data corresponding to each preset collision condition based on the preset ignition conditions corresponding to each preset collision condition.

[0117] The first traversal subunit is used to scale the collision data and mirror collision data of the corresponding preset collision condition based on the scaling ratio of the collision data and mirror collision data of each preset collision condition, output the collision data matrix set of each preset collision condition, and traverse the collision data matrix set of each preset collision condition to output the first filtered data and the second filtered data of each preset collision condition.

[0118] The third acquisition subunit is used to acquire the target vector of the first filtered data corresponding to each preset collision condition at the target time, determine the opening / closing state of the airbag deployment algorithm corresponding to each preset collision condition based on the target vector, and when the airbag deployment algorithm is in the open state, acquire the target filtered data of the second filtered data of the preset collision condition corresponding to the airbag deployment algorithm in the open state at the target time, and calculate the collision feature data at the target time based on the target filtered data and the collision data corresponding to each preset collision condition.

[0119] The second traversal subunit is used to determine the feature vector of the collision feature data. When generating the collision feature data sample at the target time, it obtains the collision feature data training sample at the target time based on the collision feature data sample and the feature vector of the collision feature data. It then traverses all running times of the airbag deployment algorithm corresponding to each preset collision condition to obtain the airbag deployment training dataset corresponding to each preset collision condition.

[0120] Optionally, before obtaining the collision feature data training samples at the target time based on the collision feature data samples and the feature vectors of the collision feature data, the second traversal subunit further includes:

[0121] The control sub-component is used to control the airbag deployment algorithm corresponding to each preset collision condition to enter the next moment of the target time if no collision feature data sample is generated at the target time, and continue to execute the step of obtaining the target vector of the first filtered data corresponding to each preset collision condition at the target time.

[0122] Optionally, the determined unit includes:

[0123] The fourth acquisition subunit is used to acquire the vehicle's ignition condition and non-ignition condition based on each preset collision condition, and to acquire the first feature data of the first label sample belonging to the ignition condition in the support vector, and the second feature data of the second label sample belonging to the non-ignition condition in the support vector.

[0124] A sub-unit is defined to examine the first feature data and the second feature data, obtain the difference value between the first feature data and the second feature data, and determine the current feature data corresponding to each preset collision condition based on the difference value when the difference value is less than a preset threshold.

[0125] The airbag control device according to the embodiments of this application identifies the current collision condition of the vehicle, obtains the current collision data corresponding to the current collision condition within a preset sampling interval, matches the corresponding target airbag deployment algorithm based on the current collision data, and controls the vehicle to deploy the airbag when the current collision data meets the preset ignition conditions. This solves the problems of collision control algorithms used in related technologies, such as difficulty in distinguishing the characteristics of different collision conditions and meeting the provided RTTF, and the increased complexity of prediction models due to numerous auxiliary algorithms, thereby reducing the robustness of the deployment algorithm. By constructing a support vector machine-based airbag deployment algorithm through model training and feature selection, the device effectively distinguishes the characteristics of different collision conditions and improves the robustness of the airbag deployment algorithm.

[0126] Figure 10 A schematic diagram of the structure of an electronic device provided in an embodiment of this application. The electronic device may include:

[0127] The memory 1001, the processor 1002, and the computer program stored on the memory 1001 and capable of running on the processor 1002.

[0128] When the processor 1002 executes the program, it implements the airbag control method provided in the above embodiments.

[0129] Furthermore, electronic devices also include:

[0130] Communication interface 1003 is used for communication between memory 1001 and processor 1002.

[0131] The memory 1001 is used to store computer programs that can run on the processor 1002.

[0132] The memory 1001 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device.

[0133] If the memory 1001, processor 1002, and communication interface 1003 are implemented independently, then the communication interface 1003, memory 1001, and processor 1002 can be interconnected via a bus to complete communication between them. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be divided into address buses, data buses, control buses, etc. For ease of representation, Figure 10 Only one thick line is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0134] Optionally, in a specific implementation, if the memory 1001, processor 1002, and communication interface 1003 are integrated on a single chip, then the memory 1001, processor 1002, and communication interface 1003 can communicate with each other through an internal interface.

[0135] The processor 1002 may be a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application.

[0136] This embodiment also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described airbag control method.

[0137] This embodiment also provides a computer program product, including a computer program that is executed to implement the airbag control method of the above embodiment.

[0138] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0139] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of technical features indicated. Thus, a feature specified as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of this application, "N" means at least two, for example, two, three, etc., unless otherwise specifically defined.

[0140] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more N executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0141] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which programs can be printed, because programs can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0142] It should be understood that the various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, the N steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0143] Those skilled in the art will understand that all or part of the steps of the methods described in the above embodiments can be implemented by a program instructing related hardware, and the program can be stored in a computer-readable storage medium. When executed, the program includes one or a combination of the steps of the method embodiments.

[0144] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0145] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application.

Claims

1. A method for controlling an airbag, characterized in that, Includes the following steps: Identify the vehicle's current collision condition; Obtain current collision data corresponding to the current collision condition within a preset sampling interval, match the corresponding target airbag deployment algorithm based on the current collision data, and determine whether the current collision data meets the preset ignition conditions through the target airbag deployment algorithm. If the current collision data meets the preset ignition conditions, then control the vehicle to deploy the airbags; The process includes, before matching the target airbag deployment algorithm based on the current collision data, obtaining the current support vector from the preset airbag prediction deployment model, and determining the feature data corresponding to each preset collision condition based on the current support vector; and constructing the airbag deployment algorithm corresponding to each preset collision condition based on the feature data corresponding to each preset collision condition. The step of determining the current feature data corresponding to each preset collision condition based on the current support vector includes: acquiring the vehicle's ignition condition and non-ignition condition based on each preset collision condition; acquiring first feature data belonging to the ignition condition with a first label sample in the support vector; and acquiring second feature data belonging to the non-ignition condition with a second label sample in the support vector; verifying the first feature data and the second feature data to obtain the difference value between the first feature data and the second feature data; and determining the current feature data corresponding to each preset collision condition based on the difference value when the difference value is less than a preset threshold.

2. The method according to claim 1, characterized in that, Before matching the corresponding target airbag deployment algorithm based on the current collision data, the algorithm further includes: Obtain the preset ignition conditions corresponding to each preset collision condition of the vehicle; The airbag deployment training dataset is obtained based on the preset ignition conditions corresponding to each preset collision condition. The airbag deployment training dataset is then input into the preset airbag prediction deployment model to obtain the current support vector.

3. The method according to claim 1 or 2, characterized in that, The step of obtaining the airbag deployment training dataset based on the preset ignition conditions corresponding to each preset collision condition includes: Obtain the collision data matrix corresponding to each preset collision condition, and simulate the mirror collision data matrix corresponding to each preset collision condition based on the collision data matrix corresponding to each preset collision condition; The mirror collision data corresponding to each preset collision condition is obtained based on the mirror collision data matrix corresponding to each preset collision condition, and the scaling ratio of the collision data and the mirror collision data corresponding to each preset collision condition is determined based on the preset ignition conditions corresponding to each preset collision condition. Based on the scaling ratio of the collision data and mirror collision data corresponding to each preset collision condition, the collision data and mirror collision data of the corresponding preset collision condition are scaled, and the collision data matrix set corresponding to each preset collision condition is output. Then, the collision data matrix set corresponding to each preset collision condition is traversed, and the first filtered data and the second filtered data corresponding to each preset collision condition are output. The target vector of the first filtered data corresponding to each preset collision condition is obtained at the target time. Based on the target vector, the open / closed state of the airbag deployment algorithm corresponding to each preset collision condition is determined. When the airbag deployment algorithm is in the open state, the target filtered data of the second filtered data of the preset collision condition corresponding to the open state in the airbag deployment algorithm for each preset collision condition is obtained at the target time. The collision feature data at the target time is calculated based on the target filtered data and the collision data corresponding to each preset collision condition. The feature vector of the collision feature data is determined, and when the collision feature data sample is generated at the target time, the collision feature data training sample at the target time is obtained based on the collision feature data sample and the feature vector of the collision feature data. The collision feature data training sample and all running times of the airbag deployment algorithm corresponding to each preset collision condition are traversed to obtain the airbag deployment training dataset corresponding to each preset collision condition.

4. The method according to claim 3, characterized in that, Before obtaining the collision feature data training samples for the target time based on the collision feature data samples and the feature vectors of the collision feature data, the method further includes: If the collision feature data sample is not generated at the target time, the airbag deployment algorithm corresponding to each preset collision condition is controlled to enter the next moment of the target time, and the step of obtaining the target vector of the first filtered data corresponding to each preset collision condition at the target time continues to be executed.

5. A control device for an airbag, characterized in that, include: The identification module is used to identify the current collision condition of the vehicle; The judgment module is used to acquire the current collision data corresponding to the current collision condition within a preset sampling interval, match the corresponding target airbag deployment algorithm according to the current collision data, and determine whether the current collision data meets the preset ignition conditions through the target airbag deployment algorithm. The control module is used to control the vehicle to deploy the airbags if the current collision data meets the preset ignition conditions. The judgment module further includes, before matching the target airbag deployment algorithm according to the current collision data, the judgment module includes: a determination unit, which obtains the current support vector obtained from the preset airbag prediction deployment model and determines the feature data corresponding to each preset collision condition based on the current support vector; and a construction unit, which is used to construct the airbag deployment algorithm corresponding to each preset collision condition according to the feature data corresponding to each preset collision condition. The determining unit includes: a fourth acquisition subunit, configured to acquire, based on each preset collision condition, the vehicle's ignition condition and non-ignition condition, and acquire, in the support vector, a first feature data of a first label sample belonging to the ignition condition, and a second feature data of a second label sample belonging to the non-ignition condition; and a determining subunit, configured to examine the first feature data and the second feature data to obtain a difference value between the first feature data and the second feature data, and, when the difference value is less than a preset threshold, determine the current feature data corresponding to each preset collision condition based on the difference value.

6. An electronic device, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, the processor executing the program to implement the airbag control method as described in any one of claims 1-4.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, The program is executed by the processor to implement the airbag control method as described in any one of claims 1-4.

8. A computer program product, comprising a computer program, characterized in that, The computer program is executed to implement the airbag control method as described in any one of claims 1-4.

Images