A method and device for detecting a living body in a vehicle

Abstract

The application relates to the technical field of in-vehicle detection, and provides a detection method and device for in-vehicle living beings. The method comprises the following steps: monitoring whether a current target vehicle enters a power-off locking state; if yes, continuously acquiring real-time response data of a sensor on an internal environment of a vehicle cabin of the target vehicle under the current power-off locking state; processing all real-time response data under the current power-off locking state to obtain current real-time data features; inputting the current real-time data features into a first support vector machine; acquiring a first output result of the first support vector machine; and if the first output result is that the internal environment of the vehicle cabin contains living beings, a safety strategy is adopted. The application does not increase the cost of high-cost sensors or processing algorithms, and accurately detects the in-vehicle living beings at a low cost, thereby improving the safety of vehicle use.

Description

Technical Field

[0001] This application relates to the field of in-vehicle detection technology, and in particular to a method and device for detecting living beings inside a vehicle. Background Technology

[0002] For in-vehicle occupant detection, the industry offers various solutions. Indirect detection methods, such as pressure / capacitive sensing of rear door opening and closing, are gradually being phased out due to accuracy issues. Currently, most mainstream automakers utilize the Doppler effect of radar or detection solutions directly through in-vehicle cameras. However, using separate radar or cameras for detection leads to high overall vehicle costs, and the use of in-vehicle cameras also makes it difficult to guarantee user privacy.

[0003] In addition to the methods mentioned above, using gas sensors for in-vehicle live animal detection is another detection method that has been explored. This method involves setting a detection threshold for carbon dioxide concentration. If the carbon dioxide sensor detects that the carbon dioxide concentration in the vehicle exceeds its specified threshold, it will issue a warning to remind the user.

[0004] However, this method also has its shortcomings. The increase in carbon dioxide concentration inside the vehicle requires an accumulation process, and the detection results are delayed. Furthermore, different biological individuals produce carbon dioxide at different rates and in different quantities. Using a fixed carbon dioxide concentration threshold for judgment will lead to inaccurate detection results.

[0005] Therefore, methods for determining carbon dioxide concentration based on a fixed detection threshold can introduce significant errors in detecting living organisms inside vehicles. Consequently, providing a solution to these technical problems is a current challenge for those skilled in the art. Summary of the Invention

[0006] In view of this, the present application provides a method and apparatus for detecting living beings inside a vehicle, in order to solve the problems of high cost and poor accuracy in the detection of living beings inside a vehicle in the prior art.

[0007] A first aspect of this application provides a method for detecting living beings inside a vehicle, comprising:

[0008] Monitor whether the target vehicle has entered the power-off locking state;

[0009] If so, continuously acquire real-time response data of the sensors to the interior environment of the target vehicle under the current power-down and lockout state; the sensors include gas sensors, and the gas sensors include at least one of carbon dioxide sensors and oxygen sensors;

[0010] The real-time data characteristics are obtained by processing all real-time response data under the current power-down lockout state. The real-time data characteristics include the current real-time response data, the rate of change of the response data corresponding to the current real-time response data, the curve integral value corresponding to all real-time response data, the relative steady-state average value of all real-time response data in a stable state, and the maximum response difference between the maximum and minimum values ​​of all real-time response data.

[0011] The current real-time data features are input into the first support vector machine; the first support vector machine is trained based on the first type of feature samples with the corresponding label of no living organisms in the interior environment of the carriage and the second type of feature samples with the corresponding label of living organisms in the interior environment of the carriage.

[0012] Obtain the first output result of the first support vector machine. If the first output result indicates that there is a living organism in the interior environment of the carriage, then adopt a safety strategy. The safety strategy includes one or more of the following: introducing external air strategy and issuing a warning message strategy.

[0013] A second aspect of this application provides a device for detecting living beings inside a vehicle, comprising:

[0014] The vehicle status module is used to monitor whether the current target vehicle has entered the power-off locking state; if so, it triggers the environmental sampling module.

[0015] An environmental sampling module is used to continuously acquire real-time response data of the sensors to the interior environment of the target vehicle under the current power-off and locked state; the sensors include gas sensors, which include at least one of carbon dioxide sensors and oxygen sensors;

[0016] The feature processing module is used to process all real-time response data under the current power-down lockout state to obtain the current real-time data features. The real-time data features include the current real-time response data, the rate of change of the response data corresponding to the current real-time response data, the curve integral value corresponding to all real-time response data, the relative steady-state average value of all real-time response data in a stable state, and the maximum response difference between the maximum and minimum values ​​of all real-time response data.

[0017] The vector machine module is used to input the current real-time data features into the first support vector machine; the first support vector machine is trained based on the first type of feature samples with the corresponding label of no living organisms in the interior environment of the carriage and the second type of feature samples with the corresponding label of living organisms in the interior environment of the carriage.

[0018] The action module is used to obtain the first output result of the first support vector machine. If the first output result indicates that there is a living organism in the interior environment of the carriage, a safety strategy is adopted. The safety strategy includes one or more of the following: introducing external air strategy and issuing a prompt message strategy.

[0019] 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 computer program to implement the steps of the above-described method.

[0020] Compared with the prior art, the beneficial effects of this application embodiment include at least the following: This application embodiment acquires real-time response data from sensors when the target vehicle is in a power-off locked state, and processes the real-time response data returned by each sensor to obtain real-time data features. The real-time data features are then input into a first support vector machine to obtain the output results. Since the first support vector machine is trained using two types of feature samples that are consistent with the form of real-time data features and contain multiple data features and labels corresponding to whether there are living beings in the vehicle's interior environment, the first support vector machine can determine the labels from multiple dimensions of data features. Therefore, the output results have high accuracy. Based on the output results, it is determined whether there are living beings in the interior environment of the current target vehicle, and corresponding safety strategies are taken when living beings are present. This application does not add expensive sensors or processing algorithms, and achieves accurate detection of living beings in the vehicle at a low cost, improving the safety of vehicle use. In later maintenance, the program algorithm can be directly modified and upgraded. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a schematic diagram illustrating one application scenario of this application.

[0023] Figure 2 This is a schematic flowchart of a method for detecting living beings inside a vehicle, provided in an embodiment of this application.

[0024] Figure 3 This is a schematic diagram illustrating the relationship between related components provided in an embodiment of this application;

[0025] Figure 4 This is a schematic diagram of the structure of a vehicle-mounted life detection device provided in an embodiment of this application;

[0026] Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0027] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0028] The following describes in detail, with reference to the accompanying drawings, a method and apparatus for detecting living beings inside a vehicle according to an embodiment of this application.

[0029] Figure 1 This is a schematic diagram illustrating an application scenario according to an embodiment of this application. The application scenario may include a first terminal device 101, a second terminal device 102, a third terminal device 103, a server 104, and a network 105.

[0030] The first terminal device 101, the second terminal device 102, and the third terminal device 103 can be hardware or software. When the first terminal device 101, the second terminal device 102, and the third terminal device 103 are hardware, they can be various electronic devices with displays and supporting communication with the server 104, including but not limited to vehicle systems, smartphones, tablets, laptops, and desktop computers. When the first terminal device 101, the second terminal device 102, and the third terminal device 103 are software, they can be installed in the aforementioned electronic devices. The first terminal device 101, the second terminal device 102, and the third terminal device 103 can be implemented as multiple software programs or software modules, or as a single software program or software module; this application embodiment does not impose any limitations on this. Furthermore, various applications can be installed on the first terminal device 101, the second terminal device 102, and the third terminal device 103, such as data processing applications, instant messaging tools, social platform software, search applications, shopping applications, etc.

[0031] Server 104 can be a server that provides various services, such as a backend server that receives requests sent by terminal devices with which it has established communication connections. This backend server can receive and analyze the requests sent by the terminal devices and generate processing results. Server 104 can be a single server, a server cluster consisting of several servers, or a cloud computing service center. This application embodiment does not limit this.

[0032] It should be noted that the server 104 can be either hardware or software. When the server 104 is hardware, it can be various electronic devices that provide various services to the first terminal device 101, the second terminal device 102, and the third terminal device 103. When the server 104 is software, it can be multiple software programs or software modules that provide various services to the first terminal device 101, the second terminal device 102, and the third terminal device 103, or it can be a single software program or software module that provides various services to the first terminal device 101, the second terminal device 102, and the third terminal device 103. This application embodiment does not impose any limitations on this.

[0033] Network 105 can be a wired network using coaxial cable, twisted pair, and fiber optic connection, or it can be a wireless network that enables interconnection of various communication devices without wiring, such as Bluetooth, Near Field Communication (NFC), and Infrared. This application embodiment does not limit this.

[0034] It should be noted that the specific types, quantities and combinations of the first terminal device 101, the second terminal device 102, the third terminal device 103, the server 104 and the network 105 can be adjusted according to the actual needs of the application scenario, and this application embodiment does not impose any restrictions on this.

[0035] Figure 2 This is a flowchart illustrating a method for detecting living beings inside a vehicle, as provided in an embodiment of this application. Figure 2 The detection method can be made by Figure 1 The first terminal device, second terminal device, third terminal device, or server executes the command. For example... Figure 2 As shown, the detection method includes:

[0036] S201: Monitor whether the current target vehicle has entered the power-off locking state;

[0037] S202: If so, continuously acquire real-time response data of the sensor to the interior environment of the target vehicle under the current power-down and lockout state; the sensor includes a gas sensor, which includes at least one of a carbon dioxide sensor and an oxygen sensor;

[0038] S203: Process all real-time response data under the current power-down lockout state to obtain the current real-time data characteristics; the real-time data characteristics include the current real-time response data, the rate of change of the response data corresponding to the current real-time response data, the curve integral value corresponding to all real-time response data, the relative steady-state average value of all real-time response data in a stable state, and the maximum response difference between the maximum and minimum values ​​of all real-time response data.

[0039] S204: Input the current real-time data features into the first support vector machine; the first support vector machine is trained based on the first type of feature samples with the corresponding label of no living organisms in the interior environment of the carriage and the second type of feature samples with the corresponding label of living organisms in the interior environment of the carriage.

[0040] S205: Obtain the first output result of the first support vector machine. If the first output result indicates that there is a living organism in the interior environment of the carriage, then adopt a safety strategy. The safety strategy includes one or more of the following: introducing external air strategy and issuing a warning message strategy.

[0041] Understandably, the system monitors the real-time vehicle status of the target vehicle. Once the target vehicle enters the power-off locking state, steps S202-S205 are continuously executed to continuously acquire real-time response data from the sensors to the interior environment of the vehicle. The real-time response data is processed to obtain corresponding real-time data features, which are then input into the first support vector machine. The first output of the first support vector machine is used to characterize the analysis result of the first support vector machine based on the current real-time data features to determine whether there are living beings inside the vehicle. If the first output result indicates that there are living beings inside the vehicle, corresponding safety strategies need to be adopted to ensure the safety of living beings inside the vehicle.

[0042] Compared to the detection algorithms of radar and cameras, this embodiment uses at least one of the commonly used gas sensors originally installed inside the vehicle compartment, which is significantly less expensive than radar and cameras. Compared to the original method of detection using carbon dioxide sensors and fixed detection thresholds, this embodiment uses multiple data features and support vector machines as the basis for detection, enabling effective detection of various biological individuals with different respiratory capacities, achieving high accuracy.

[0043] It is understandable that, such as Figure 3As shown, when the method of this embodiment is executed by the vehicle system, the state detection in step S201 is determined by the real-time state parameters of various parts of the target vehicle. The acquisition of the real-time state parameters is based on the communication network inside the target vehicle, and the communication network realizes the information transmission between the vehicle system and various parts. The sensors in step S202 are located inside the passenger compartment of the target vehicle and are used to detect the internal environment of the passenger compartment. The sensors include gas sensors, which include at least one of carbon dioxide sensors and oxygen sensors. The sensors may also include temperature sensors. The sensor data can also be transmitted to the vehicle system through the communication network. When taking safety strategies in step S205, the cabin area is involved. The specific measures to introduce external air involve... When a vehicle reaches cabin-level devices such as windows and air conditioning, the vehicle system sends corresponding action commands to these devices via a communication network to implement a strategy for introducing external air. Similarly, when issuing a prompt message strategy, if the prompt message strategy targets the current interior of the vehicle, it involves cabin-level devices such as the audio system. If the prompt message strategy targets devices outside the vehicle, such as the driver's or user's mobile terminal, it involves connected devices. The connected devices are used to establish an external network between the target vehicle and the cloud server. The cloud server connects to the mobile terminal in the external network. Based on the relay of the cloud server, the mobile terminal can ultimately receive the prompt message corresponding to the prompt message strategy. The mobile terminal can also remotely control the target vehicle.

[0044] Specifically, in step S201, when monitoring whether the target vehicle has entered the power-off locking state, the power-off locking state can be set according to the corresponding power-off locking scenario through multiple dimensions of the target vehicle's state parameters. When the real-time state parameters of the target vehicle meet the power-off locking scenario, it is determined that the target vehicle has entered the power-off locking state. For example, if the power-off locking scenario is the scenario where the owner locks the door and leaves, the corresponding state parameters should reflect that the target vehicle has been powered off and the owner has left the vehicle. This can be described by state parameters such as the vehicle power-on / off indicator, the driver monitoring system, and the status changes of the door corresponding to the driver's seat. As another example, if the power-off locking scenario is the scenario where the vehicle is powered off and there is no air circulation between the interior and exterior of the vehicle, the corresponding state parameters should reflect that the target vehicle has been powered off and all devices related to external air circulation are in a non-circulating state. This can be described by state parameters such as the vehicle power-on / off indicator and the device operating status indicator.

[0045] Therefore, the following electrically interlocking scenario, taking a situation where there is no air circulation between the inside and outside of the vehicle compartment as an example, includes the process of monitoring whether the current target vehicle has entered the electrically de-interlocking state:

[0046] Monitor the current vehicle status of the target vehicle;

[0047] If all mechanical ventilation devices of the target vehicle are turned off and the air conditioning of the target vehicle is not in external circulation mode, then the target vehicle is determined to be in the power-off locking state.

[0048] Mechanical ventilation systems include doors, windows, and sunroofs.

[0049] Understandably, if all mechanical ventilation devices in the target vehicle are closed—that is, the doors, windows, and sunroof are closed—outdoor air cannot freely exchange with the air inside the vehicle. Similarly, if the air conditioning is not in external recirculation mode (i.e., it is not working or is in internal recirculation mode), it is not actively exchanging air between the vehicle and the outside. Based on this, when the target vehicle enters the power-off locked state, there is no possibility of air exchange between the interior and exterior environments. If there are living organisms inside the vehicle, they are highly likely to suffer from heatstroke, oxygen deprivation, or suffocation.

[0050] Furthermore, considering that the air conditioner may be actively set to recirculation mode by the user under certain circumstances, the determination of the power-off locking state can be modified from the original determination that the target vehicle's air conditioner is not in external circulation mode to the determination that the target vehicle's air conditioner is not working, or a determination for a specific duration can be added when the air conditioner is in recirculation mode, that is:

[0051] If all mechanical ventilation devices of the target vehicle are turned off and the air conditioning of the target vehicle is not working in the current vehicle status, then the target vehicle is determined to be in the power-off locking state.

[0052] If, in the current vehicle status, all mechanical ventilation devices of the target vehicle are in the off state and the air conditioning of the target vehicle is in recirculation mode and remains in this mode for a period of time exceeding the preset duration, then the target vehicle is determined to have entered the power-off locking state.

[0053] Another conclusion can be drawn at this point: when the above conditions are not met, the target vehicle does not enter the power-off locking state. Specifically:

[0054] If, in the current vehicle status, any mechanical ventilation device of the target vehicle is turned on, or the target vehicle's air conditioning is in external circulation mode, then it is determined that the target vehicle has not entered the power-off locking state.

[0055] It is understandable that the above are some different methods for determining the power-off interlocking state. The specific settings can be made according to user needs and actual conditions, and there are no restrictions here.

[0056] It is understandable that after determining that the target vehicle has entered the power-off locking state, the real-time response data of the sensors is continuously acquired and processed to obtain the current real-time data characteristics. According to category, the sensors in this embodiment include at least a gas sensor, which includes at least one of a carbon dioxide sensor and an oxygen sensor.

[0057] Furthermore, the sensor may also include a temperature sensor; the process of continuously acquiring real-time response data of the sensor to the interior environment of the target vehicle under the current power-down locked state includes:

[0058] Continuously acquire real-time temperature response data of the temperature sensor to the interior environment of the target vehicle under the current power-down and lockout state;

[0059] Continuously acquire real-time gas response data of the gas sensor to the internal environment of the carriage under the current power-off and lockout state.

[0060] It is understandable that the detection accuracy of this embodiment will further increase as the number and types of sensors increase.

[0061] Furthermore, the number and specific location of various sensors are not limited here and can be set according to the actual vehicle configuration.

[0062] Furthermore, since real-time response data is continuously acquired, the current real-time response data is updated over time, while previous real-time response data is retained as historical real-time response data. As time changes, each sensor generates a set of real-time response data. Multiple sets of real-time response data from multiple similar sensors can be processed and integrated into a single set of real-time response data, or they can be processed as independent sets of real-time response data for subsequent processing, depending on the actual algorithm.

[0063] For each set of real-time response data, step S203 is executed to obtain the current real-time data characteristics of that set, namely, the current real-time response data, the rate of change of response data, the integral value of the curve, the relative steady-state average value, and the maximum response difference of that set.

[0064] Specifically, the rate of change of response data is the ratio of the difference between the real-time response data of the current sampling period and the real-time response data of the previous sampling period to the duration of a unit sampling period.

[0065] Similarly, the curve integral value is specifically the sum of all real-time response data under the current power-down lockout state.

[0066] Furthermore, when the real-time data characteristics include a relative steady-state average, the process of processing all real-time response data under the current power-down lockout state to obtain the relative steady-state average includes:

[0067] Calculate the window variance for the real-time response data in the sliding window; the window length of the sliding window includes a preset number of sampling periods, and the endpoint of the sliding window is the real-time response data of the current sampling period;

[0068] When the window variance is less than the preset variance value, it is determined that all real-time response data in the sliding window are in a stable state.

[0069] The average value of all real-time response data in a steady state is obtained as the relative steady-state average.

[0070] If the window variance is not less than the preset variance value, it is determined that all real-time response data in the sliding window are not in a stable state, and the current relative steady-state average value is set to an invalid state.

[0071] Taking a set of real-time response data as an example, after entering the power-down lockout state, the sensor's real-time response data is acquired once every unit sampling period. The real-time response data of the first sampling period is x1, the current sampling period is the i-th sampling period, and the real-time response data of the current sampling period is x. i ,at this time:

[0072] Let the rate of change of the response data be denoted as X. 1i The calculation process is as follows: Where T is the unit sampling period duration, that is, the duration of each sampling period.

[0073] Similarly, the value of the line integral is denoted as X. 2i The calculation process is as follows: Where j takes a value between 1 and i, meaning the summation of all currently collected real-time response data.

[0074] Accordingly, the maximum response difference is denoted as X. 3i The calculation process is X 3i =max(x1,...,x i )-min(x1,...,x i ), which means the difference between the maximum and minimum values ​​among all the real-time response data collected so far.

[0075] Among these, the relative steady-state average is more complex and requires first using a sliding window analysis to determine if a steady state has occurred, i.e., calculating the window variance s of the real-time response data within the sliding window. 2 The calculation process is as follows:

[0076]

[0077] Where N is the number of sampling periods corresponding to the window length, i.e., the preset number, and the value of j ranges from i-N+1 to i, meaning the range of real-time response data in the sliding window is x. i-N+1To x i Based on the above formula, the variance of the real-time response data within the sliding window is calculated, and the result is used as the window variance. This represents the average value of the real-time response data within the sliding window.

[0078] If the window variance is not less than the preset variance value, it is determined that the real-time response data in the current sliding window is not in a stable state, and the current relative steady-state average value is set to an invalid state, that is, let X... 4i =invalid;

[0079] When the window variance is less than the preset variance value, it can be determined that the real-time response data within the sliding window is in a stable state. The average value of all real-time response data currently in a stable state is calculated and denoted as the relative steady-state average value X. 4i It's important to note that the range for calculating the average is not limited to the sliding window. As time changes, the sliding window changes accordingly, and real-time response data that was originally within the sliding window will leave it. If all real-time response data within the sliding window is already in a stable state, even if the sliding window no longer includes earlier real-time response data as time changes, the earlier real-time response data that is considered to be in a stable state must still be considered when calculating the relative steady-state average. Therefore, we have:

[0080]

[0081] Where m ≥ N, meaning the number of real-time response data points used to calculate the relative steady-state average is not less than the sliding window length, and the value of j ranges from i-m+1 to i, x i-m+1 This marks the starting point of a stable state.

[0082] Furthermore, prior to step S203, the initial real-time response data can be preprocessed to improve data quality. Preprocessing operations include data cleaning, data integration, and data normalization. Data cleaning primarily improves data consistency by filling in missing values, smoothing noisy data, and smoothing or removing outliers. Data integration involves organizing data from multiple sources, such as multiple sets of real-time response data from similar sensors mentioned above. Considering the correlation between data, data is integrated into a single set of real-time response data, reducing data redundancy. Data normalization addresses the issue of different data sources leading to varying numerical representations in the real-time response data. This can be achieved by using the scale function to normalize each set of real-time response data. Data normalization includes:

[0083]

[0084] Where x k This represents the real-time response data after data normalization for the k-th sampling period, where k ranges from 1 to i, and c...k For the real-time response data of the k-th sampling period before data normalization processing, c max and c min These are the maximum and minimum data of the sensor, respectively. Specifically, c max and c min It can be the maximum and minimum values ​​of a fixed calibration data range when the sensor is applied to the target vehicle, or it can be the maximum and minimum values ​​of all current real-time response data. The specific value can be set according to the actual situation or user needs.

[0085] Understandably, the real-time response data after data normalization is all located between [-1,1], which effectively reduces the data differences between samples, and will significantly improve the convergence ability of the subsequent first support vector machine.

[0086] Therefore, before processing all real-time response data under the current power-down lockout state to obtain the current real-time data characteristics, the following steps are also included:

[0087] All real-time response data under the current power-down lockout state are preprocessed, including at least one of data cleaning and data normalization.

[0088] Understandably, the current real-time data features corresponding to the current collection period are input as a set of input feature vectors to the first support vector machine. The first support vector machine can output a first output result. If the first output result indicates that there are no living beings inside the carriage, no measures are taken. If the first output result indicates that there are living beings inside the carriage, safety measures need to be taken. The safety measures in this embodiment include a strategy of introducing external air by controlling the operation of corresponding devices in the target vehicle to exchange air between the inside and outside of the carriage, and a strategy of issuing a prompt message to the owner or current user of the target vehicle.

[0089] Furthermore, the external air introduction strategy specifically includes at least one of the following: setting at least one mechanical ventilation device to the open position and setting the air conditioning to external circulation mode. This external air introduction strategy ensures the safety of living beings currently inside the carriage in the most direct way, eliminating the risks of heatstroke, oxygen deficiency, and suffocation that would otherwise exist for living beings inside the carriage.

[0090] Furthermore, considering that introducing external air involves actively adjusting the target vehicle's systems, which may reduce the vehicle's safety (e.g., opening windows), a warning message strategy is generally preferred. The external air introduction strategy is only implemented when real-time data analysis indicates insufficient oxygen levels inside the vehicle, necessitating measures to replenish oxygen. Similarly, considering the vehicle's safety, the external air introduction strategy can prioritize setting the air conditioning to external recirculation mode. If the air conditioning malfunctions, at least one mechanical ventilation device should be opened. The sunroof is the preferred open mechanical ventilation device, followed by windows.

[0091] Understandably, when analyzing real-time data features to determine whether oxygen needs replenishment in the current carriage interior, a support vector machine (SVM) can also be used as the analysis subject. The training process and usage of the SVM are similar. Therefore, if the first output indicates the presence of living organisms inside the carriage, the process of implementing safety strategies includes:

[0092] If the first output result indicates that there are living organisms in the interior environment of the carriage, the current real-time data features are input into the second support vector machine; the second support vector machine is trained based on the third feature sample with the corresponding label of sufficient oxygen in the interior environment of the carriage and the fourth feature sample with the corresponding label of insufficient oxygen in the interior environment of the carriage.

[0093] Obtain the second output result of the second support vector machine. If the second output result indicates that the oxygen content in the interior environment of the carriage is insufficient, then adopt the strategy of introducing external air.

[0094] Understandably, considering the public's current concern about the health status of living beings inside enclosed vehicles, in order to alleviate public anxiety about the health status of living beings inside vehicles and to prevent bystanders from blindly attempting rescue, this embodiment, after adopting appropriate external air strategies, can output status prompts such as "This vehicle has activated external ventilation mode" through the central control display screen of the target vehicle, informing the public outside the target vehicle who are concerned about the living beings inside the vehicle that the living beings inside the vehicle are in a safe and stable state.

[0095] It is understood that step S201 of this embodiment is repeatedly executed according to a preset cycle. During the repeated execution, when step S201 monitors whether the current target vehicle has entered the power-off locking state, if the determination result is yes, step S202 is triggered; if the determination result is no, step S201 is returned to execution. After step S202 is triggered, steps S202-S205 will continue to be executed. If the first output result is that there are no living beings in the interior of the vehicle, it can be considered that there are no living beings in the interior of the current target vehicle, and the detection in the current parking cycle can be directly ended. However, in order to ensure that there is no false judgment that there are living beings in the interior of the vehicle, step S202 can be returned to execution, and the detection in the current parking cycle can be ended when multiple first output results are obtained that there are no living beings in the interior of the vehicle.

[0096] Furthermore, the following is the training process of the first support vector machine.

[0097] I. Sample collection.

[0098] Sample collection was divided into two categories: one category consisted of first-category samples where no living organisms were present inside the carriage, and the other category consisted of second-category samples where living organisms were present inside the carriage. Test vehicles with the same structure as the target vehicle were selected for sample collection. By setting the environmental temperature and the presence of living organisms in the test vehicle, real-time response test data of the test sensors inside the carriage was obtained.

[0099] In one specific embodiment, the first type of feature samples includes multiple first sample data corresponding to multiple environmental temperature curves, and the second type of feature samples includes multiple second sample data corresponding to multiple environmental temperature curves and multiple preset living organisms, respectively; the process of acquiring each second sample data includes:

[0100] The simulation device is set up according to the corresponding preset life form. The simulation device is set up in the test environment inside the test vehicle to simulate the ecological behavior of the preset life form.

[0101] Control the test vehicle to enter the power-off lockout state;

[0102] The ambient temperature of the test vehicle is controlled to change according to the ambient temperature curve, and the test sensors continuously acquire the second real-time response test data of the test environment inside the vehicle compartment.

[0103] The second real-time response test data is processed to obtain the corresponding second data features, which are used as the second sample data of the corresponding current ambient temperature curve and the preset living organism.

[0104] The process of obtaining each first sample data includes:

[0105] The test environment inside the train carriage was set to be devoid of living organisms;

[0106] Control the test vehicle to enter the power-off lockout state;

[0107] The ambient temperature of the test vehicle is controlled to change according to the ambient temperature curve, and the first real-time response test data of the test sensors to the test environment inside the vehicle is continuously acquired.

[0108] The first real-time response test data is processed to obtain the corresponding first data feature, which is then used as the first sample data.

[0109] The test vehicle and the target vehicle have the same structure, and the test sensors have the same structure and location as the sensors in the target vehicle.

[0110] Understandably, the simulation device is located inside the test vehicle's cabin environment to simulate the ecological behavior of a pre-defined living organism. This primarily includes respiratory behavior under different conditions, namely, the consumption of oxygen and the exhalation of carbon dioxide. The specific parameters of respiratory behavior vary depending on the type and state of the pre-defined living organism. Furthermore, the ecological behavior of the pre-defined living organism also includes changes in body temperature, which are correlated with oxygen content. The ambient temperature of the test vehicle includes at least one of the external ambient temperature and the internal temperature of the corresponding cabin test environment. Multiple ambient temperature curves may include temperature change curves corresponding to different temperature starting points and different rates of temperature change.

[0111] In one specific embodiment, the collection of the second type of feature samples aims to include as many different living organisms as possible. A simulation device is used to model infants (0-1 years), toddlers (1-3 years), children (4-12 years), adolescents (12-18 years), adults (18-60 years), the elderly (over 60 years), and pets, controlling different in-vehicle temperatures and collecting real-time response data from various test sensors. As the in-vehicle temperature increases, the heart rate of the pre-set living organism should rise, the carbon dioxide concentration should show a non-linear increasing trend, and the oxygen concentration should show a non-linear decreasing trend. The in-vehicle temperature can be set to 16℃, 26℃, 36℃, 38℃, 40℃, and above 40℃, respectively. That is, the ambient temperature curve is set as a temperature change curve with the in-vehicle temperature as the preset starting point and a temperature change rate of 0. If the in-vehicle temperature is less than 40℃, the sampling interval (unit sampling period) is set to 1 second; when the in-vehicle temperature is greater than 40℃, the sampling interval is set to 0.5 seconds.

[0112] The first type of feature sample collection was conducted under the scenario that the target vehicle was turned off, the windows were closed, and there were no living organisms in the interior environment. Similarly, real-time response data of the sensor was collected at different interior temperatures: 16℃, 26℃, 36℃, 38℃, 40℃, and above 40℃.

[0113] II. Data normalization processing.

[0114] To eliminate the influence of different numerical values, the scale function is used to normalize the first and second type feature samples measured in the sensor array. In the new dataset after normalization, the processed data values ​​are all between [-1, 1], which effectively reduces the data differences between samples. This will significantly improve the convergence performance of the model subsequently built.

[0115] III. Feature Extraction.

[0116] For the normalized first-class and second-class feature samples, feature extraction is performed respectively, that is, the corresponding response data change rate, curve integral value, relative steady-state average value, and maximum response difference are determined, thereby obtaining the corresponding first-class and second-class feature samples.

[0117] IV. Support Vector Machine Algorithm Recognition.

[0118] Support Vector Machines (SVMs) address classification problems involving two or more classes of samples by constructing a decision hyperplane based on the principle of minimizing structural risk. In this embodiment, the detection of living beings inside a vehicle is transformed into a binary classification problem: the absence of living beings inside the vehicle and the presence of living beings inside the vehicle, referred to as "no living beings" and "living beings." The first SVM is then trained using the first and second class feature samples.

[0119] It is understandable that both the first type of feature sample and the second type of feature sample include feature values ​​and labels. The label for the first type of feature sample is that there are no living organisms in the interior of the carriage, while the label for the second type of feature sample is that there are living organisms in the interior of the carriage.

[0120] Given a feature set X i , and the tag set y i ={+1, -1}, forming the feature sample set T = {(X i ,y i )}, where X i Let X represent an eigenvector. Suppose an eigenvector contains two data features, X. i =[x i1 ,x i2 ], where x i1 and x i2 Let y be two data features. Assuming that life detection is considered a linearly separable problem in a two-dimensional plane, if y i = +1 indicates the presence of living creatures; if y i = -1 indicates the absence of living organisms. This linearly separable problem can then be represented in vector form:

[0121] If yi =+1, then ω T X i +b>0;

[0122] If y i =-1, then ω T X i +b<0.

[0123] in, w and b are the hyperplane parameters, respectively.

[0124] hyperplane w T X i +b represents two classes of samples: "with living organisms" and "without living organisms." However, in reality, the distribution of feature samples in the feature space is not ideal, and sample points may have positional errors. Therefore, it is necessary to find the support vectors to the hyperplane w. T X i The hyperplane that maximizes the distance d from +b can be represented by the distance from the support vectors to the hyperplane as follows:

[0125]

[0126] Because the hyperplane represented by (w,b) and the hyperplane represented by (aw,ab) are the same plane, and a∈R, a is a parameter value that implements scaling (w,b) such that |w T X0 + b| = 1, X0 is a support vector, so the distance from the support vector X0 to the hyperplane is transformed into:

[0127]

[0128] Finding the maximum distance d is equivalent to minimizing ||w||. The constraint |w|| is derived by combining support and non-support vectors. T x i +b|≥1. Therefore, in the case of linear separability, the optimization problem of finding the optimal hyperplane can be expressed as:

[0129]

[0130] (w, b) are the parameters to be solved. For linearly separable cases, they can be solved using a convex quadratic programming problem. However, due to the complexity of the samples collected by the sensor array and the inherent limitations of the sensors themselves, linearly inseparable cases may exist. In this case, the linear model is insufficient, and the above optimization problem has no solution; that is, w and b do not satisfy all the above constraints. Therefore, for linearly inseparable models, the constraints need to be relaxed by introducing a non-negative slack variable δ. The modified support vector machine optimization is as follows:

[0131]

[0132] C is a penalty factor used to control the degree of punishment for misclassified samples, and it is selected based on the recognition rate of the test algorithm. If C = 10000, the solved δ approaches 0, and the resulting hyperplane remains consistent with the linearly separable case.

[0133] If we consider liveness detection as a linearly inseparable model, the core idea of ​​support vector machines is to use a mapping function... Mapping training samples from a low-dimensional feature space to a high-dimensional feature space transforms a low-dimensional linearly inseparable problem into a high-dimensional linearly separable problem. (Introduction...) The optimization issues are as follows:

[0134]

[0135] for The solution is obtained by introducing a kernel function. So that we don't know Even with explicit expression, classification problems involving nonlinear models can be solved. Because the objective function is convex and the constraint is linear, satisfying the strong duality theorem, the solution to the primal problem is equal to the solution to the dual problem. The primal problem is described as follows:

[0136]

[0137] The dual problem maxθ can be written in the following form:

[0138]

[0139] st:α i ≥0, β i ≥0;

[0140] α i and β i These are Lagrange multipliers under two different constraints;

[0141] For w, b, δ respectively i Taking the derivative and setting it to 0, the objective function for optimization becomes as follows:

[0142]

[0143] By solving the above equation, we obtain the classification hyperplane decision function, which is expressed as follows:

[0144]

[0145] The penalty factor C of the support vector machine is a trade-off between the classification model's complexity and training error, ensuring good generalization ability. The kernel function is selected as follows:

[0146]

[0147] The kernel function parameter σ affects the complexity of the data distribution in the high-dimensional feature space. The training and test sets are divided in a 7:3 ratio. The penalty factor C and kernel function parameter σ are determined using a grid search algorithm, selecting the optimal parameters C and σ that achieve the highest cross-validation accuracy. By solving the sign function sgn(), we can obtain f(x) = 1, indicating the presence of living organisms; and f(x) = -1, indicating the absence of living organisms.

[0148] It is understood that in this embodiment, the data feature types of the input data of the first support vector machine and the data feature types of the two types of feature samples during training are kept consistent to ensure that the first support vector machine that can be trained can be successfully applied to the detection process of this embodiment.

[0149] It is understandable that the training process of the second support vector machine is similar to that of the first support vector machine. However, the label settings for the third and fourth feature samples differ from those for the first and second feature samples, which will not be elaborated here.

[0150] The detection method in this embodiment has the following effects:

[0151] Using gas sensors for life detection significantly reduces costs compared to using radar or vision sensors.

[0152] Compared to detecting living beings inside a vehicle using fixed sensor thresholds, this embodiment uses a support vector machine-based intelligent algorithm to make judgments, which improves the accuracy of identification.

[0153] In later maintenance, the specific parameters of the support vector machine can be modified and upgraded.

[0154] The method of this application acquires real-time response data from sensors when the target vehicle is in a power-off locked state, and processes the real-time response data returned by each sensor to obtain real-time data features. The real-time data features are then input into a first support vector machine to obtain the output results. Since the first support vector machine is trained using two types of feature samples that are consistent with the form of the real-time data features and contain multiple data features and labels corresponding to whether there are living beings in the vehicle's interior environment, the first support vector machine can determine the labels from multiple dimensions of data features. Therefore, the output results have high accuracy. Based on the output results, it is determined whether there are living beings in the interior environment of the current target vehicle, and corresponding safety strategies are taken when living beings are present. This application does not add expensive sensors or processing algorithms, and achieves accurate detection of living beings in the vehicle at a low cost, thereby improving the safety of vehicle use.

[0155] All the above-mentioned optional technical solutions can be combined in any way to form the optional embodiments of this application, and will not be described in detail here. It should be understood that the sequence number of each step in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0156] The following are embodiments of the apparatus described in this application, which can be used to execute the embodiments of the method described in this application. For details not disclosed in the apparatus embodiments of this application, please refer to the embodiments of the method described in this application.

[0157] Figure 4 This is a schematic diagram of a vehicle-mounted life detection device provided in an embodiment of this application. Figure 4 As shown, the detection device includes:

[0158] The vehicle status module 401 is used to monitor whether the current target vehicle has entered the power-off locking state; if so, it triggers the environmental sampling module 402.

[0159] The environmental sampling module 402 is used to continuously acquire real-time response data of the sensors to the interior environment of the target vehicle under the current power-off and locked state; the sensors include gas sensors, or the sensors include the gas sensors and temperature sensors, and the gas sensors include at least one of carbon dioxide sensors and oxygen sensors;

[0160] The feature processing module 403 is used to process all real-time response data under the current power-down lockout state to obtain the current real-time data features. The real-time data features include the current real-time response data, the rate of change of the response data corresponding to the current real-time response data, the curve integral value corresponding to all real-time response data, the relative steady-state average value corresponding to all real-time response data in a stable state, and the maximum response difference between the maximum and minimum values ​​of all real-time response data.

[0161] The vector machine module 404 is used to input the current real-time data features into the first support vector machine; the first support vector machine is trained based on the first type of feature samples with the corresponding label of no living organisms in the interior environment of the carriage and the second type of feature samples with the corresponding label of living organisms in the interior environment of the carriage.

[0162] Action module 405 is used to obtain the first output result of the first support vector machine. If the first output result indicates that there is a living organism in the interior environment of the carriage, a safety strategy is adopted. The safety strategy includes one or more of the following: introducing external air strategy and issuing a prompt message strategy.

[0163] The device in this application acquires real-time response data from sensors when the target vehicle is in a power-off locked state. Based on the real-time response data returned by each sensor, it processes the data to obtain real-time data features. These features are then input into a first support vector machine (SVM) to obtain the output. Since the first SVM is trained using two types of feature samples—one containing multiple data features and the other labeled with labels corresponding to the presence or absence of living beings in the vehicle's interior—it can determine labels from multiple data feature dimensions. Therefore, the output has high accuracy. Based on the output, it determines whether living beings exist inside the target vehicle's interior and takes appropriate safety measures when they are present. This application does not add expensive sensors or processing algorithms, achieving accurate detection of living beings inside the vehicle at a lower cost, thus improving vehicle safety.

[0164] In an exemplary embodiment, the response data change rate is specifically the ratio of the difference between the real-time response data of the current sampling period and the real-time response data of the previous sampling period to the duration of a unit sampling period.

[0165] The curve integral value is specifically the sum of all real-time response data under the current power-down interlocking state.

[0166] In an exemplary embodiment, when the real-time data characteristics include a relative steady-state average, the process of processing all real-time response data under the current power-down lockout state to obtain the relative steady-state average includes:

[0167] Calculate the window variance for the real-time response data in the sliding window; the window length of the sliding window includes a preset number of sampling periods, and the endpoint of the sliding window is the real-time response data of the current sampling period;

[0168] When the window variance is less than the preset variance value, it is determined that all real-time response data in the sliding window are in a stable state.

[0169] The average value of all real-time response data in a steady state is obtained as the relative steady-state average.

[0170] If the window variance is not less than the preset variance value, it is determined that all real-time response data in the sliding window are not in a stable state, and the current relative steady-state average value is set to an invalid state.

[0171] In an exemplary embodiment, before the investment strategy gate processes all real-time response data under the current power-down lockout state to obtain the current real-time data characteristics, it further includes:

[0172] All real-time response data under the current power-down lockout state are preprocessed, including at least one of data cleaning and data normalization.

[0173] In one exemplary embodiment, the sensor further includes a temperature sensor; the environmental sampling module is specifically used for:

[0174] Continuously acquire real-time temperature response data of the temperature sensor to the interior environment of the target vehicle under the current power-down and lockout state;

[0175] Continuously acquire real-time gas response data of the gas sensor to the internal environment of the carriage under the current power-off and lockout state.

[0176] In an exemplary embodiment, the first type of feature samples includes multiple first sample data corresponding to multiple environmental temperature curves, and the second type of feature samples includes multiple second sample data corresponding to multiple environmental temperature curves and multiple preset living organisms, respectively; the process of obtaining each second sample data includes:

[0177] The simulation device is set up according to the corresponding preset life form. The simulation device is set up in the test environment inside the test vehicle to simulate the ecological behavior of the preset life form.

[0178] Control the test vehicle to enter the power-off lockout state;

[0179] The ambient temperature of the test vehicle is controlled to change according to the ambient temperature curve, and the test sensors continuously acquire the second real-time response test data of the test environment inside the vehicle compartment.

[0180] The second real-time response test data is processed to obtain the corresponding second data features, which are used as the second sample data of the corresponding current ambient temperature curve and the preset living organism.

[0181] The process of obtaining each first sample data includes:

[0182] The test environment inside the train carriage was set to be devoid of living organisms;

[0183] Control the test vehicle to enter the power-off lockout state;

[0184] The ambient temperature of the test vehicle is controlled to change according to the ambient temperature curve, and the first real-time response test data of the test sensors to the test environment inside the vehicle is continuously acquired.

[0185] The first real-time response test data is processed to obtain the corresponding first data feature, which is then used as the first sample data.

[0186] The test vehicle and the target vehicle have the same structure, and the test sensors have the same structure and location as the sensors in the target vehicle.

[0187] In one exemplary embodiment, the process of monitoring whether the current target vehicle has entered the power-off locking state includes:

[0188] Monitor the current vehicle status of the target vehicle;

[0189] If all mechanical ventilation devices of the target vehicle are turned off and the air conditioning of the target vehicle is not in external circulation mode, then the target vehicle is determined to be in the power-off locking state.

[0190] Mechanical ventilation systems include doors, windows, and sunroofs.

[0191] In one exemplary embodiment, the strategy of introducing external air specifically includes at least one of setting at least one mechanical ventilation device to an on state and setting the air conditioner to an external circulation mode.

[0192] In one exemplary embodiment, the action module is specifically used for:

[0193] If the first output result indicates that there are living organisms in the interior environment of the carriage, the current real-time data features are input into the second support vector machine; the second support vector machine is trained based on the third feature sample with the corresponding label of sufficient oxygen in the interior environment of the carriage and the fourth feature sample with the corresponding label of insufficient oxygen in the interior environment of the carriage.

[0194] Obtain the second output result of the second support vector machine. If the second output result indicates that the oxygen content in the interior environment of the carriage is insufficient, then adopt the strategy of introducing external air.

[0195] Figure 5 This is a schematic diagram of the electronic device 5 provided in an embodiment of this application. Figure 5 As shown, the electronic device 5 of this embodiment includes: a processor 501, a memory 502, and a computer program 503 stored in the memory 502 and executable on the processor 501. When the processor 501 executes the computer program 503, it implements the steps in the various method embodiments described above. Alternatively, when the processor 501 executes the computer program 503, it implements the functions of each module / unit in the various device embodiments described above.

[0196] Electronic device 5 can be a desktop computer, laptop, handheld computer, cloud server, or other electronic device. Electronic device 5 may include, but is not limited to, processor 501 and memory 502. Those skilled in the art will understand that... Figure 5 This is merely an example of electronic device 5 and does not constitute a limitation on electronic device 5. It may include more or fewer components than shown, or different components.

[0197] The processor 501 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.

[0198] The memory 502 can be an internal storage unit of the electronic device 5, such as a hard disk or RAM of the electronic device 5. The memory 502 can also be an external storage device of the electronic device 5, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, Flash Card, etc., equipped on the electronic device 5. The memory 502 can also include both internal and external storage units of the electronic device 5. The memory 502 is used to store computer programs and other programs and data required by the electronic device.

[0199] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0200] If an integrated module / unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program may include computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable storage medium may include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable storage medium may be appropriately added to or subtracted according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable storage medium does not include electrical carrier signals and telecommunication signals.

[0201] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for detecting living beings inside a vehicle, characterized in that, include: Monitor whether the target vehicle has entered the power-off locking state; If so, continuously acquire real-time response data of the sensors to the interior environment of the target vehicle under the current power-off locked state; the sensors include gas sensors and temperature sensors, and the gas sensors include at least one of carbon dioxide sensors and oxygen sensors; the real-time response data includes real-time gas response data and real-time temperature response data of the interior environment of the target vehicle. The current real-time data characteristics are obtained by processing all the real-time response data in the current power-down lockout state. The real-time data features include multiple features such as the current real-time response data, the rate of change of the response data corresponding to the current real-time response data, the curve integral value corresponding to all real-time response data, the relative steady-state average value corresponding to all real-time response data in a stable state, and the maximum response difference between the maximum and minimum values ​​among all real-time response data. The real-time data features are input into a first support vector machine; the first support vector machine is trained based on a first type of feature sample with the corresponding label that there are no living organisms in the interior environment of the carriage and a second type of feature sample with the corresponding label that there are living organisms in the interior environment of the carriage; the first type of feature sample includes multiple first sample data corresponding to multiple environmental temperature curves, and the second type of feature sample includes multiple second sample data corresponding to multiple environmental temperature curves and multiple preset living organisms respectively; Obtain the first output result of the first support vector machine. If the first output result indicates that there is a living organism in the interior environment of the carriage, then adopt a safety strategy. The safety strategy includes one or more of the following: introducing external air strategy and issuing a warning message strategy. The acquisition process for each of the second sample data includes: setting up a simulation device according to the corresponding preset life form, the simulation device being installed in the test environment inside the test vehicle to simulate the ecological behavior of the preset life form; controlling the test vehicle to enter the power-off lockout state; controlling the ambient temperature of the test vehicle to change according to the ambient temperature curve, and continuously acquiring the second real-time response test data of the test sensors to the test environment inside the test vehicle; processing the second real-time response test data to obtain the corresponding second data features as the second sample data corresponding to the current ambient temperature curve and the preset life form; The process of acquiring each of the first sample data includes: setting the test environment inside the carriage to be devoid of living organisms; controlling the test vehicle to enter the power-off lockout state; controlling the ambient temperature of the test vehicle to change according to the ambient temperature curve, and continuously acquiring the first real-time response test data of the test sensor to the test environment inside the carriage; processing the first real-time response test data to obtain the corresponding first data feature as the first sample data; The test vehicle has the same structure as the target vehicle, and the test sensor has the same structure and position as the sensor in the target vehicle.

2. The detection method according to claim 1, characterized in that, The rate of change of the response data is specifically the ratio of the difference between the real-time response data of the current sampling period and the real-time response data of the previous sampling period to the duration of a unit sampling period. The curve integral value is specifically the sum of all the real-time response data under the current power-down lockout state.

3. The detection method according to claim 1, characterized in that, When the real-time data features include the relative steady-state average value, the process of processing all the real-time response data in the current power-down lockout state to obtain the relative steady-state average value includes: Calculate the window variance for the real-time response data in the sliding window; the window length of the sliding window includes a preset number of sampling periods, and the endpoint of the sliding window is the real-time response data of the current sampling period; When the window variance is less than a preset variance value, it is determined that all the real-time response data in the sliding window are in a stable state. The average value of all the real-time response data in a stable state is taken as the relative steady-state average value. If the window variance is not less than the preset variance value, it is determined that all the real-time response data in the sliding window are not in the stable state, and the current relative steady-state average value is set to an invalid state.

4. The detection method according to claim 1, characterized in that, The process of monitoring whether the current target vehicle has entered the power-off locking state includes: Monitor the current vehicle status of the target vehicle; If, in the current vehicle state, all mechanical ventilation devices of the target vehicle are in the off state and the air conditioning of the target vehicle is not in the external circulation mode, then it is determined that the target vehicle has entered the power-off locking state. The mechanical ventilation system includes doors, windows, and sunroof.

5. The detection method according to claim 4, characterized in that, The strategy for introducing external air specifically includes at least one of the following: setting at least one of the mechanical ventilation devices to an open state and setting the air conditioner to an external circulation mode.

6. The detection method according to any one of claims 1 to 5, characterized in that, If the first output result indicates the presence of living organisms inside the carriage, the process of implementing a safety strategy includes: If the first output result indicates that there is a living organism in the interior environment of the carriage, the current real-time data features are input into the second support vector machine; the second support vector machine is trained based on the third feature sample with the corresponding label of sufficient oxygen in the interior environment of the carriage and the fourth feature sample with the corresponding label of insufficient oxygen in the interior environment of the carriage. Obtain the second output result of the second support vector machine. If the second output result indicates that the oxygen content in the interior environment of the carriage is insufficient, then adopt the strategy of introducing external air.

7. A device for detecting living beings inside a vehicle, used to perform the detection method according to any one of claims 1-6, characterized in that, include: The vehicle status module is used to monitor whether the current target vehicle has entered the power-off locking state; if so, it triggers the environmental sampling module. The environmental sampling module is used to continuously acquire real-time response data of the sensors to the interior environment of the target vehicle under the current power-off locking state; the sensors include gas sensors, and the gas sensors include at least one of carbon dioxide sensors and oxygen sensors; The feature processing module is used to process all the real-time response data in the current power-down lockout state to obtain the current real-time data features; the real-time data features include multiple of the following: the current real-time response data, the rate of change of the response data corresponding to the current real-time response data, the curve integral value corresponding to all the real-time response data, the relative steady-state average value corresponding to all the real-time response data in a stable state, and the maximum response difference between the maximum and minimum values ​​in all the real-time response data. The vector machine module is used to input the current real-time data features into the first support vector machine; the first support vector machine is trained based on a first type of feature sample with the corresponding label that there are no living organisms in the interior environment of the carriage and a second type of feature sample with the corresponding label that there are living organisms in the interior environment of the carriage. The action module is used to obtain the first output result of the first support vector machine. If the first output result indicates that there is a living organism in the interior environment of the carriage, a safety strategy is adopted. The safety strategy includes one or more of the following: introducing external air strategy and issuing a prompt message strategy.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 6.

Images