A centralized and discrete automotive safety control system

Abstract

The present invention belongs to the field of automotive safety technologies and provides a centralized and discrete automotive safety control system. The system includes a vehicle control system and a module control system. The vehicle controller system includes a main control ECU, a first battery management module, and a first drive module. The module control system includes a Can communication module, an MCU module, a second battery management module, a sensor module, and a second drive module. The Can communication module communicates with the main control ECU. The first drive module is used to unlock and lock the safety devices of the vehicle. The second drive module is used to unlock the safety devices. The second battery management module is used to supply power to the module control system and the safety devices after receiving power from the first battery management module. The MCU module also runs a pre-trained recognition model to recognize the impact state of the vehicle based on the data fed back by the sensor module, so as to control the second drive module to directly unlock the safety devices. The present invention can solve the problem that the vehicle is prone to losing control and unable to unlock when encountering an accident.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle safety, and particularly to a centralized and discrete vehicle safety control system. Background Art

[0002] The existing vehicle control system consists of a single main control chip, circuit, battery, and each module. When the vehicle encounters an accident, once one or all of the main control, main line, and battery are damaged, the whole vehicle will be out of control, there are risks such as the car door cannot be opened, the tailgate cannot be opened, the window cannot be lowered, the seat belt cannot be unlocked, and the horn cannot sound an alarm, etc., thus increasing the difficulty of rescue and reducing the escape chance of the people in distress. Summary of the Invention

[0003] In view of the above technical problems, the present invention provides a centralized and discrete vehicle safety control system to solve the problem that the whole vehicle is easily out of control and cannot be unlocked when the vehicle encounters an accident.

[0004] Other features and advantages of the present disclosure will become apparent from the following detailed description, or be learned in part through the practice of the present disclosure.

[0005] The present invention discloses a centralized and discrete vehicle safety control system, including a vehicle control system and a module control system. The vehicle controller system includes a main control ECU, a first battery management module, and a first drive module. The module control system includes a Can communication module, an MCU module, a second battery management module, a sensor module, and a second drive module. The Can communication module communicates with the main control ECU. The first drive module is used to unlock and lock the safety devices of the vehicle. The second drive module is used to unlock the safety devices. The second battery management module is used to supply power to the module control system and the safety devices after receiving power from the first battery management module. The MCU module is connected to the safety devices to detect the states of the safety devices. The safety devices include at least one of a door lock module, a tailgate lock module, a sunroof lock module, a seat belt lock module, and a horn module. The sensor module includes at least one of a Hall effect sensor, a vibration sensor, an acceleration sensor, a gyroscope sensor, a light sensor, a chemical resistance sensor, a force sensor, and an impact sensor. The MCU module also runs a pre-trained recognition model to recognize the impact state of the vehicle according to the data fed back by the sensor module for:

[0006] When there is no impact, taking the command of the vehicle control system as the highest priority;

[0007] When an impact is detected, the lock state of the safety device is detected. When it is not unlocked, the Can communication module actively detects whether the communication with the main control ECU is normal and whether the first battery management module is powered normally. If all are normal, the main control ECU is requested to control the first drive module to unlock the safety device. If any one is abnormal, the second drive module is controlled to directly unlock the safety device.

[0008] Further, when the recognition model recognizes the impact state of the vehicle, it includes:

[0009] Based on the Hall effect sensor or GPS, measure the speed of the vehicle;

[0010] Based on the force sensor, the impact sensor, and the vibration sensor, determine the collision point and the severity of the collision of the vehicle;

[0011] Based on the Hall effect sensor, the acceleration sensor, and the gyroscope sensor, determine the rollover and fall states of the vehicle;

[0012] Based on the light sensor and the chemical resistance sensor, determine the size of the flame and the smoke level of the vehicle.

[0013] Further, when determining the collision point and the severity of the collision of the vehicle, it includes:

[0014] Place multiple force sensors at key parts of the vehicle. When the vehicle is impacted, determine the specific collision point of the vehicle according to the pressure changes of the force sensors at different positions;

[0015] Based on the impact sensor, determine the acceleration change of the vehicle during the collision, determine the collision impact force of the vehicle, so as to determine the severity of the collision;

[0016] Comprehensively determine whether the vehicle has a collision based on the instantaneous change values of the force sensor, the impact sensor, and the vibration sensor.

[0017] Further, when determining the rollover and fall states of the vehicle, it includes:

[0018] When installing the acceleration sensor, align the X-axis of the three axes of the acceleration sensor with the body axis of the vehicle, align the Z-axis with gravity, and the Y-axis is perpendicular to the body axis;

[0019] Calculate the roll angle and pitch angle of the vehicle around the X-axis and Y-axis to determine whether the vehicle rolls over, and based on the gyroscope sensor, calculate the speed of the vehicle during the rollover to determine the rollover speed of the vehicle;

[0020] When the acceleration in the Z-axis is greater than or equal to 9.7m / s 2When the accelerations along the Y-axis and X-axis are within the threshold range, it is determined that the vehicle is in a linear fall;

[0021] Sample the values of the three axes of the acceleration sensor at a frame interval of 10 ms. When the difference between the peaks of the accelerations in any two adjacent frames in any axis direction is greater than or equal to 15.5 m / s 2 and the vehicle speed measured by the Hall effect sensor undergoes a speed change exceeding the threshold and then becomes zero, it is determined that the vehicle is in a non-linear fall.

[0022] Furthermore, determining the ignition state of the vehicle includes:

[0023] Based on the radiation in different narrow or wide wavelength regions detected by the optical sensor to determine whether a flame is detected;

[0024] Based on the change in the material resistance of the chemiresistor sensor changed by the smoke to determine the smoke level.

[0025] Furthermore, the recognition model is one of a Gaussian mixture model, a NB tree, a decision tree, and a classification and regression tree. When training the recognition model, speed, flame size, collision point, smoke level, rollover state, fall type, and gravity magnitude are used as training features to classify different accident severities and accident types.

[0026] Furthermore, when the recognition model is a Gaussian mixture model, the Gaussian mixture model is composed of the weighted sum of multiple Gaussian components with normal distributions. Each Gaussian component has its own mean vector, covariance matrix, and corresponding mixing weight. The mean vector, the covariance matrix, and the mixing weight are learned from the corresponding training features. When training the Gaussian mixture model, feature vectors are obtained from the training features based on the expectation-maximization algorithm;

[0027] When the Gaussian mixture model classifies the accident severity, the data sampled by each sensor in the sensor module is input, and the accident severity of the input data is calculated. When calculating, it is based on the following formula:

[0028]

[0029] g(x|μ i , ∑ i ) is the probability density, μ i is the mean vector, ∑ i is the covariance matrix, w i is the mixing weight, θ is the given model parameter, x is the input feature vector. Among them, g(x|μ i , ∑ i ) can be expressed as:

[0030]

[0031] D is the dimension of the feature vector, and the mixing weight w i satisfies

[0032] Further, when the recognition model is an NB tree, when the NB tree classifies the accident severity, it is based on the following formula:

[0033]

[0034] x = (x 1 , x 2 , x 3 , …, x n ) is the input variable, including flame size, smoke level, rollover status, fall type, and gravity magnitude. y is the accident severity, and P(y) and P(x) are obtained from known training features and known accident severities:

[0035]

[0036] Then, according to the data sampled by each sensor in the input sensor module, the accident severity is obtained:

[0037]

[0038] Further, when the recognition model is a decision tree, when the decision tree classifies the accident type, it is based on the following formula:

[0039]

[0040] where D is the input variable with m different values, representing flame size, smoke level, rollover status, fall type, and gravity magnitude. Among them, the input variable is the dataset sampled by each sensor in the sensor module, and p i is the probability of selecting a specific change from i options. Then, the expected unpredictability of n results of D is calculated as:

[0041]

[0042] where c i is the number of observations, c is the total number of observations, and the information gain is the difference between the actual entropy and the expected entropy, expressed as: I = E(D) - EE.

[0043] Further, when the recognition model is a classification and regression tree, when the classification and regression tree classifies the accident type, it is based on the following formula:

[0044]

[0045] Among them, D is the data set sampled by each sensor in the sensor module, Q represents the number of classifications, Gini(D) represents the Gini index of D, and p i represents the proportion of records in the i-th category in the data D, and p i is calculated by the following formula:

[0046]

[0047] c j is the specified accident type.

[0048] The technical solution of the present disclosure has the following beneficial effects:

[0049] The present invention will add an independent battery supply and risk judgment mechanism to each safety control module. Even if each module in the safety device loses contact with the main control, it can make an independent judgment based on the sensor module and the MCU module, and thus make an unlocking action, greatly increasing the chance of escape. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 is a structural block diagram of a centralized and discrete automotive safety control system in an embodiment of this specification;

[0051] Figure 2 is a schematic structural diagram of a seat belt lock module in an embodiment of this specification. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0052] Now, the exemplary embodiments will be described more fully with reference to the accompanying drawings. However, the exemplary embodiments can be implemented in various forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that this disclosure will be more thorough and complete, and will fully convey the concept of the exemplary embodiments to those skilled in the art. The described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are provided to give a thorough understanding of the embodiments of the present disclosure. However, those skilled in the art will realize that the technical solutions of the present disclosure can be practiced without one or more of the specific details, or other methods, components, devices, steps, etc. can be adopted. In other cases, well-known technical solutions are not shown or described in detail to avoid obscuring the various aspects of the present disclosure.

[0053] In addition, the accompanying drawings are only schematic illustrations of the present disclosure. The same reference numerals in the drawings denote the same or similar parts, and thus repeated descriptions thereof will be omitted. Some of the block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities may be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.

[0054] As Figure 1 shown, an embodiment of the present specification provides a centralized and discrete automotive safety control system, including a vehicle control system 100 and a module control system 200. The vehicle controller system includes a main control ECU 101, a first battery management module 102, and a first drive module 103. The module control system 200 includes a Can communication module 201, an MCU module 202, a second battery management module 203, a sensor module 204, and a second drive module 205. The Can communication module 201 communicates with the main control ECU 101. The first drive module 103 is used to unlock and lock a safety device 300 of the vehicle. The second drive module 205 is used to unlock the safety device 300. The second battery management module 203 is used to supply power to the module control system 200 and the safety device 300 after receiving power from the first battery management module 102. The MCU module is connected to the safety device 300 to detect the state of the safety device 300. The safety device 300 includes at least one of a door lock module, a tailgate lock module, a sunroof lock module, a seat belt lock module, and a horn module. The sensor module 204 includes at least one of a Hall effect sensor, a vibration sensor, an acceleration sensor, a gyroscope sensor, an optical sensor, a chemical resistance sensor, a force sensor, and an impact sensor. The MCU module also runs a pre-trained recognition model to recognize the impact state of the vehicle based on the data fed back by the sensor module 204, for: when there is no impact, taking the command of the vehicle control system 100 as the highest priority; when an impact is recognized, detecting the lock state of the safety device 300. When it is not unlocked, actively detecting whether the communication with the main control ECU 101 is normal through the Can communication module 201, and detecting whether the first battery management module 102 is supplying power normally. If all are normal, requesting the main control ECU 101 to control the first drive module 103 to unlock the safety device 300. If any one is abnormal, controlling the second drive module 205 to directly unlock the safety device 300.

[0055] Among them, the prior art has an automatic unlocking function when the vehicle is out of control. However, if the out-of-control is caused by damage to the power system, bus or main controller, these functions will be directly cut off. Therefore, on the basis of the original centralized control, the module control system 200 can have the ability of self-judgment and control. After losing contact with the upper main controller, it can still work to ensure the safety of the occupants. That is, in the state where the vehicle cannot be automatically unlocked after being out of control, it can automatically open the door lock, tailgate lock, lower the window, eject the seat belt, sound the horn for alarm, etc.

[0056] In one embodiment, when the recognition model recognizes the impact state of the vehicle, it includes the following:

[0057] Based on the Hall effect sensor or GPS, measure the speed of the vehicle; based on the force sensor, the impact sensor, and the vibration sensor, determine the collision point and the severity of the collision of the vehicle; based on the Hall effect sensor, the acceleration sensor, and the gyroscope sensor, determine the rollover and falling state of the vehicle; based on the optical sensor and the chemical resistance sensor, determine the size of the flame and the level of smoke of the vehicle.

[0058] As a supplement, when determining the speed of the vehicle, if the Hall effect sensor is used to sense the wheel speed, the vehicle speed can be directly obtained.

[0059] As a supplement, the force sensor can detect physical pressure, extrusion and weight, so as to infer the collision point. Therefore, when determining the collision point and the severity of the collision of the vehicle, it includes: placing a plurality of the force sensors at key parts of the vehicle, and when the vehicle is impacted, determine the specific collision point of the vehicle according to the pressure changes of the force sensors at different positions; based on the impact sensor, determine the acceleration change of the vehicle during the collision, determine the collision impact force of the vehicle, so as to determine the severity of the collision; comprehensively determine whether the vehicle has a collision according to the instantaneous change values of the force sensor, the impact sensor and the vibration sensor.

[0060] As a supplement, when determining the rollover and falling state of the vehicle, it includes: when installing the acceleration sensor, align the X-axis of the three axes of the acceleration sensor with the vehicle body axis, align the Z-axis with gravity, and the Y-axis is perpendicular to the vehicle body axis; calculate the roll angle and pitch angle of the vehicle around the X-axis and Y-axis to determine whether the vehicle rolls over, and based on the gyroscope sensor, calculate the speed of the vehicle during the rollover to determine the rollover speed of the vehicle; when the acceleration in the Z-axis is greater than or equal to 9.7m / s 2 , and the accelerations in the Y-axis and X-axis are within the threshold range, determine that the vehicle is linearly falling; sample the values of the three axes of the acceleration sensor at a frame interval of 10ms, and the difference between the peaks of the accelerations of any two adjacent frames in any axis direction is greater than or equal to 15.5m / s2 When the vehicle speed measured by the Hall effect sensor undergoes a speed change exceeding a threshold and then changes to zero, it is determined that the vehicle has a non-linear fall.

[0061] Among them, when the roll angle and pitch angle of the vehicle around the X-axis and Y-axis exceed 90°, it means that the vehicle has overturned. A non-linear fall means that the vehicle falls while flipping, and it is after the vehicle is traveling at a certain speed and suddenly has an accident resulting in a stop, representing a relatively high accident level with a high severity. A linear fall means that the vehicle rises and falls vertically, and relatively speaking, the harm is smaller.

[0062] As a supplement, when determining the fire state of the vehicle, it includes: determining whether a flame is detected based on the radiation in different narrow or wide wavelength regions detected by the optical sensor; determining the smoke level based on the change in the material resistance of the chemiresistor sensor changed by the smoke.

[0063] In one embodiment, the recognition model is one or more of a Gaussian mixture model, an NB tree, a decision tree, and a classification and regression tree. When training the recognition model, speed, flame size, collision point, smoke level, rollover state, fall type, and gravity magnitude are used as training features to classify different accident severities and accident types.

[0064] Exemplarily, when the recognition model is a Gaussian mixture model, the Gaussian mixture model is composed of a weighted sum of multiple Gaussian components with normal distributions. Each Gaussian component has its own mean vector, covariance matrix, and corresponding mixture weight. The mean vector, the covariance matrix, and the mixture weight are learned from the corresponding training features. When training the Gaussian mixture model, feature vectors are obtained from the training features based on the expectation-maximization algorithm;

[0065] When the Gaussian mixture model performs accident severity classification, the data sampled by each sensor in the sensor module is input, and the accident severity of the input data is calculated. When calculating, it is based on the following formula:

[0066]

[0067] g(x|μ i ,∑ i ) is the probability density, μ i is the mean vector, ∑ i is the covariance matrix, w i is the mixture weight, λ is a given model parameter, x is the input feature vector, where g(x|μ i ,∑ i ) can be expressed as:

[0068]

[0069] D is the dimension of the feature vector, and the mixing weight w i satisfies

[0070] Exemplarily, when the recognition model is an NB tree, when the NB tree performs accident severity classification, it is based on the following formula:

[0071]

[0072] x = (x 1 , x 2 , x 3 , …, x n ) is the input variable, including flame size, smoke level, rollover status, fall type, gravity magnitude, y is the accident severity, and P(y) and P(x) are obtained from known training features and known accident severities:

[0073]

[0074] Then, according to the data sampled by each sensor in the input sensor module, the accident severity is obtained:

[0075]

[0076] Exemplarily, when the recognition model is a decision tree, when the decision tree performs accident type classification, it is based on the following formula:

[0077]

[0078] Among them, D is the input variable with m different values, representing flame size, smoke level, rollover status, fall type, gravity magnitude. Among them, the input variable is the dataset sampled by each sensor in the sensor module, and p i is the probability of selecting a specific change from i options. Then, the expected unpredictability of n results of D is calculated as:

[0079]

[0080] Among them, c i is the number of observations, c is the total number of observations, and the information gain is the difference between the actual entropy and the expected entropy, expressed as: I = E(D) - EE.

[0081] Exemplarily, when the recognition model is a classification and regression tree, when the classification and regression tree performs accident type classification, it is based on the following formula:

[0082]

[0083] Among them, D is the data set sampled by each sensor in the sensor module, Q represents the number of classifications, Gini(D) represents the Gini index of D, and p i represents the proportion of records of the i-th category in the data D, and p i is calculated by the following formula:

[0084]

[0085] c j is the specified accident type.

[0086] The recognition model can be one of a Gaussian mixture model, an NB tree, a decision tree, and a classification and regression tree, or a combination of two, such as a Gaussian mixture model combined with a classification and regression tree. One is used for accident severity classification, and the other is used for accident type recognition. In this way, the accident level can be determined comprehensively. After an accident occurs, according to the accident level, the MCU module sends requests for help to the corresponding fire department and medical emergency through its built-in 4G model, and drives the motors of the door lock module, tail door lock module, sunroof lock module, and seat belt lock module to rotate and unlock through the first drive module, and drives the horn module to sound an alarm. Among them, the seat belt lock module can be Figure 2 as shown, having a seat belt, a buckle 3011, a rotating motor 3012, a traction rope 3013, and a return spring 3014. When the rotating motor 3012 pulls the traction rope 3013, the buckle 3011 rotates away from the buckle 3015 of the seat belt, realizing the unlocking of the seat belt lock module.

[0087] The technical solution of the present disclosure has the following beneficial effects:

[0088] The present invention will add an independent battery supply and risk judgment mechanism to each safety control module. Even if each module in the safety device loses contact with the main control, it can still make judgments independently according to the sensor module and the MCU module, so as to make an unlocking action, greatly increasing the chance of escape.

[0089] Those skilled in the art will readily think of other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present disclosure, which follow the general principles of the present disclosure and include known common knowledge or conventional technical means in the technical field not disclosed in the present disclosure. The specification and embodiments are only regarded as exemplary, and the true scope and spirit of the present disclosure are pointed out by the claims.

Claims

1. A centralized and discrete automobile safety control system, characterized in that: The vehicle control system includes a vehicle control system and a module control system, wherein the vehicle control system includes a main control ECU, a first battery management module, and a first drive module, wherein the module control system includes a Can communication module, an MCU module, a second battery management module, a sensor module, and a second drive module, wherein the Can communication module communicates with the main control ECU, wherein the first drive module is used to unlock and lock the safety device of the vehicle, wherein the second drive module is used to unlock the safety device, wherein the second battery management module is used to power the module control system and the safety device after receiving power from the first battery management module, wherein the MCU module is connected to the safety device to detect the state of the safety device, wherein the safety device includes at least one of a door lock module, a tailgate lock module, a sunroof lock module, a seat belt lock module, and a horn module, wherein the sensor module includes at least one of a Hall effect sensor, a vibration sensor, an acceleration sensor, a gyroscope sensor, a light sensor, a chemical resistance sensor, a force sensor, and an impact sensor, wherein the MCU module also runs a pre-trained recognition model to recognize the collision state of the vehicle according to the data fed back by the sensor module, so as to: When there is no collision, the command of the vehicle control system has the highest priority; When a collision is identified, the lock state of the safety device is detected. When the device is not unlocked, the Can communication module is used to actively detect whether the communication with the main control ECU is normal, and whether the first battery management module is powered normally. If all are normal, the main control ECU is requested to control the first drive module to unlock the safety device. If any one of them is abnormal, the second drive module is controlled to directly unlock the safety device. When the recognition model recognizes the impact state of the car, it includes: measuring the speed of the car based on the Hall effect sensor or GPS; determining the collision point and the severity of the collision of the car based on the force sensor, the impact sensor, and the vibration sensor; determining the rollover and falling state of the car based on the Hall effect sensor, the acceleration sensor, and the gyroscope sensor; determining the flame size and smoke level of the car based on the light sensor and the chemical resistance sensor; When determining the rollover and falling state of a car, the method includes: when installing the acceleration sensor, aligning the X axis of the acceleration sensor with the body axis of the car, aligning the Z axis with gravity, and the Y axis perpendicular to the body axis; calculating the roll angle and pitch angle of the car around the X axis and the Y axis to determine whether the vehicle rolls over, and calculating the speed of the car when rolling based on the gyroscope sensor to determine the rollover speed of the car; when the acceleration of the Z axis is greater than or equal to 9.7 , and when the acceleration of the Y axis and the X axis is within the threshold range, it is determined that the car is falling linearly; the values ​​of the three axes of the acceleration sensor are sampled at a frame interval of 10ms, and the difference between the peak values ​​of the acceleration of any two adjacent frames in any axis direction is greater than or equal to 15.5 When the speed of the vehicle measured by the Hall effect sensor changes beyond the threshold and becomes zero, it is determined that the vehicle falls nonlinearly.

2. A centralized and discrete automobile safety control system according to claim 1, characterized in that: When determining the point of impact and severity of a car collision, include: Placing a plurality of the force sensors at key positions of the vehicle, and determining the specific collision point of the vehicle according to the pressure changes of the force sensors at different positions when the vehicle is hit; Based on the impact sensor, determine the acceleration change of the car when it collides, determine the collision impact force of the car, and determine the severity of the collision; The instantaneous change values ​​of the force sensor, the impact sensor and the vibration sensor are integrated to determine whether the vehicle has collided.

3. A centralized and discrete automobile safety control system according to claim 1, characterized in that: When determining the fire status of the car, include: Determining whether a flame is detected based on radiation in different narrow or wide wavelength regions detected by the optical sensor; The smoke level is determined based on the change in the resistance of the material of the chemiresistor sensor that is altered by the smoke.

4. A centralized and discrete automobile safety control system according to claim 1, characterized in that: The recognition model is one of a Gaussian mixture model, a NB tree, a decision tree and a classification regression tree. When training the recognition model, speed, flame size, collision point, smoke level, rollover state, fall type and gravity size are used as training features to classify different accident severities and accident types.

5. A centralized and discrete automobile safety control system according to claim 4, characterized in that: When the recognition model is a Gaussian mixture model, the Gaussian mixture model is composed of a weighted sum of multiple normally distributed Gaussian components, each of the Gaussian components has its own mean vector, covariance matrix and corresponding mixing weights, the mean vector, the covariance matrix and the mixing weights are learned from the corresponding training features, and when training the Gaussian mixture model, a feature vector is derived from the training features based on an expectation maximization algorithm; When the Gaussian mixture model is used to classify the severity of an accident, the data sampled by each sensor in the sensor module is input, and the severity of the accident of the input data is calculated based on the following formula: ； is the probability density, is the mean vector, is the covariance matrix, is the mixing weight, For given model parameters, is the input feature vector, where It can represent: ； D is the dimension of the feature vector, and the mixing weight satisfy .

6. A centralized and discrete automobile safety control system according to claim 4, characterized in that: When the identification model is a NB tree, the NB tree performs accident severity classification based on the following formula: ； The input variables include flame size, smoke level, rollover state, fall type, gravity, The severity of the accident, and From the known training characteristics and the known severity of the accident: ； The severity of the accident is obtained based on the data sampled by each sensor in the sensor module: 。 7. A centralized and discrete automobile safety control system according to claim 4, characterized in that: When the identification model is a decision tree, the decision tree performs accident type classification based on the following formula: ； Wherein, D is an input variable with m different values, representing the flame size, smoke level, rollover state, fall type, and gravity size, wherein the input variable is the data set sampled by each sensor in the sensor module, For The probability of choosing a specific variation among the options is then the expected unpredictability of the n outcomes of D is calculated as: ； in, is the number of observations, is the total number of observations, and the information gain is the difference between the actual entropy and the expected entropy, expressed as: .

8. A centralized and discrete automobile safety control system according to claim 4, characterized in that: When the identification model is a classification and regression tree, the classification and regression tree classifies the accident type based on the following formula: ； Wherein, D is the data set sampled by each sensor in the sensor module, Q represents the number of classifications, represents the Gini index of D, Indicates the first The proportion of records in each category, Calculated by the following formula: ； The specified incident type.

Images