Child sleep detection method and sleep detection system in new energy vehicle intelligent cockpit

Abstract

This invention provides a method and system for detecting child sleep in a smart cockpit of a new energy vehicle. The sleep detection method includes: collecting respiratory signals from a target child sitting in a child seat, and verifying respiratory interruptions based on the respiratory signals to identify abnormal respiratory events; in response to the abnormal respiratory event to be verified, acquiring the target child's body surface temperature distribution information and the target child's body posture relative to the child seat; performing correlation analysis between the temporal characteristics of the abnormal respiratory event and the temporal temperature change sequence of the mouth and nose area in the body surface temperature distribution information to generate a suffocation risk verification result; and controlling the child seat to perform corresponding posture adjustments based on the risk level of the suffocation risk verification result. The technical problem solved by this invention is that existing child seats, as passive safety devices, have limited functionality and cannot handle the risk of abnormal breathing that occurs when a child is sleeping while traveling in a vehicle.

Description

Technical Field

[0001] This invention relates to the field of child car seat technology, and more specifically, to a method and system for detecting children's sleep in the intelligent cockpit of a new energy vehicle. Background Technology

[0002] With the increasing popularity of family car travel, it is becoming more and more common for children to ride in vehicles and fall asleep in their car seats during trips. Child car seats effectively ensure collision safety, but their static restraint design cannot address another hidden risk: suffocation during sleep.

[0003] Due to the vibrations and inertia of a moving vehicle, a child's body can easily slide unintentionally, causing their mouth and nose to be obstructed by the seat's shoulder straps, padding, or side supports, thus hindering their breathing. This risk occurs suddenly and is difficult for a guardian in the driver's seat to detect in real time.

[0004] However, the relevant technologies have at least one of the following problems: existing child seats, as passive safety devices, have a single function and cannot address the risk of abnormal breathing in children while they are sleeping in a car. Summary of the Invention

[0005] The technical problem solved by this invention is that existing child seats, as passive safety devices, have limited functions and cannot address the risk of abnormal breathing in children while they are sleeping in a car.

[0006] To address the aforementioned issues, this invention provides a method for detecting child sleep in a smart cockpit of a new energy vehicle. The method is applied to a child seat located within the smart cockpit. The method includes: collecting respiratory signals from a target child seated in the child seat and verifying respiratory interruption based on the respiratory signals to identify abnormal respiratory events; in response to the abnormal respiratory event to be verified, acquiring the target child's body surface temperature distribution information and the target child's body posture relative to the child seat; performing correlation analysis between the temporal characteristics of the abnormal respiratory event and the temporal temperature change sequence of the mouth and nose area in the body surface temperature distribution information to generate a suffocation risk verification result; and controlling the child seat to perform corresponding posture adjustments based on the risk level of the suffocation risk verification result. Specifically, when the risk level is high, the support angle of the child seat is adjusted to position the target child in a resuscitation position with their head higher than their chest.

[0007] Compared with existing technologies, the technical effects achieved by this solution are as follows: This invention collects and monitors the respiratory signals of a target child in a child car seat to determine whether the child is experiencing respiratory abnormalities. If so, it performs cross-validation in a vehicle environment by fusing and analyzing multi-source information such as respiratory signals, temperature of the mouth and nose area, and body posture, thus achieving accurate identification of suffocation risk. By directly linking the suffocation risk verification results with the car seat's mechanical mechanism, it can automatically generate and execute gradient adjustment commands based on different risk levels to perform physical interventions such as resuscitation positioning. This directly alleviates suffocation risk factors such as mouth and nose compression, achieving targeted intervention.

[0008] In one embodiment of the present invention, respiratory interruption verification is performed based on the respiratory signal to identify abnormal respiratory events, including: filtering and denoising the respiratory signal to extract a clean respiratory waveform; calculating the real-time respiratory frequency based on the respiratory signal; and determining abnormal respiratory events specifically includes: determining that the respiratory interruption determination condition is met when the amplitude of the clean respiratory waveform is continuously lower than a first amplitude threshold and reaches a first duration; and / or determining that the respiratory interruption determination condition is met when the real-time respiratory frequency is continuously lower than a first frequency threshold and reaches a first duration.

[0009] Compared with existing technologies, the technical effects achieved by adopting this technical solution are as follows: by using a pure breathing waveform amplitude below a first amplitude threshold or a real-time breathing frequency below a first frequency threshold, two interruption scenarios are covered, which improves the accuracy of breathing interruption determination compared with the single indicator of existing technologies; at the same time, by combining the amplitude and frequency of the breathing signal, it is more in line with the respiratory physiological characteristics of children and avoids ignoring potential suffocation risks due to a single indicator.

[0010] In one embodiment of the present invention, the respiratory interruption verification based on the respiratory signal to identify abnormal respiratory events further includes: calculating the signal-to-noise ratio (SNR) and signal quality index of the pure respiratory waveform; performing preliminary classification and credibility assessment of the abnormal respiratory events based on the SNR and signal quality index; if the SNR is lower than a first threshold or the signal quality index is lower than a first quality threshold, it is determined to be a low-quality signal segment, and the abnormal respiratory events in the signal segment are marked as unreliable events and removed; if the SNR is higher than the first threshold and the signal quality index is higher than the first quality threshold, it is determined that the abnormal respiratory event occurred in a high-quality signal segment and is marked as an event to be verified.

[0011] Compared with existing technologies, the technical effects achieved by this solution are as follows: by screening with both signal-to-noise ratio and signal quality index, interruptions of low-quality signal segments are marked as unreliable events and eliminated, reducing interference in subsequent fusion analysis from the source and improving the credibility of interruption events entering the verification stage; furthermore, only interruptions of high-quality signal segments are marked as events to be verified, reducing system energy consumption and improving overall detection efficiency.

[0012] In one embodiment of the present invention, the child sleep detection method further includes: collecting body movement signals of a target child in a child seat; the body movement signals include changes in the target child's body shape and posture in the child seat; analyzing respiratory abnormal events marked as events to be verified in combination with the body movement signals; if the intensity of the body movement signal exceeds the body movement intensity threshold, the event to be verified is determined to be a false interruption.

[0013] Compared with existing technologies, the technical effect achieved by adopting this technical solution is: by synchronously collecting body movement signals, it can be determined whether the event to be verified is caused by body movement, and false interruptions caused by turning over or displacement can be effectively eliminated.

[0014] In one embodiment of the present invention, the temporal characteristics of abnormal breathing events are correlated with the temporal changes in the temperature of the mouth and nose region in the body surface temperature distribution information to generate a suffocation risk verification result. This includes: identifying the mouth and nose region in the body surface temperature distribution map; calculating the temperature drop rate of the mouth and nose region over a continuous time period; if the temperature drop rate is greater than a preset rate threshold, the suffocation risk verification result is determined to be high risk; and / or identifying the cheek region in the body surface temperature distribution map; comparing the temperature of the mouth and nose region with the temperature of the cheek region; if the temperature of the mouth and nose region is lower than the temperature of the cheek region and reaches a preset temperature difference threshold, the suffocation risk verification result is determined to be high risk.

[0015] Compared with existing technologies, the technical effects achieved by this solution are as follows: by clarifying the criteria for determining high-risk asphyxiation in two directions—the rate of temperature drop being greater than a preset threshold and the temperature difference between the mouth and nose being greater than a preset threshold compared to the cheek temperature; by using the key physiological characteristic of temperature drop in the mouth and nose area, secondary biological verification of abnormal breathing events is performed, and only when breathing interruption and either of the above temperature characteristics are met simultaneously is it determined to be high-risk, thus excluding non-asphyxiation scenarios such as breathing interruption but normal temperature, or temperature drop but no breathing interruption.

[0016] In one embodiment of the present invention, the temporal characteristics of respiratory abnormalities are correlated with the temporal changes in the temperature of the mouth and nose region in the body surface temperature distribution information to generate a suffocation risk verification result. This further includes: if the temperature drop rate is less than a preset rate threshold but greater than a preset lower limit threshold for medium-risk rates, the suffocation risk verification result is determined to be medium-risk; and / or if the temperature of the mouth and nose region is lower than the temperature of the cheek region, but the temperature difference between the mouth and nose region and the cheek region is less than a preset temperature difference threshold and persists for a preset duration, the suffocation risk verification result is determined to be medium-risk; if the temperature drop rate is less than a preset lower limit threshold for medium-risk rates, and the temperature difference between the mouth and nose region and the cheek region is less than a preset temperature difference threshold, the suffocation risk verification result is determined to be low-risk.

[0017] Compared with existing technologies, the technical effects achieved by adopting this technical solution are: supplementing the medium-risk criteria for temperature drop rate or temperature difference between high risk and normal, and the low-risk criteria for no temperature abnormalities; forming a three-level risk system of low, medium and high.

[0018] In one embodiment of the present invention, the child seat is controlled to perform corresponding posture adjustments based on the risk level of the asphyxiation risk verification result, including: determining the sleep phase of an abnormal breathing event; if the abnormal breathing event occurs during the sleep-wake transition period, an additional verification process is initiated; the additional verification process includes: acquiring the micro-motion signal of the target child's periorbital area in the child seat; comparing the micro-motion signal with a pre-stored library of historical respiratory fluctuation patterns during eye-opening periods; if the frequency of the micro-motion pressure signal matches the historical wakefulness characteristics, the current state is determined to be a false positive, and the seat adjustment command is not triggered.

[0019] Compared with existing technologies, the technical effects achieved by this solution are as follows: by identifying sleep stages and comparing micro-movement signals around the eyes, it can distinguish between normal breathing fluctuations during the wakefulness transition and the risk of suffocation during sleep, thereby reducing false positive alarms.

[0020] In one embodiment of the present invention, the child seat is controlled to perform corresponding posture adjustments based on the risk level of the asphyxiation risk verification result, including: if the asphyxiation risk verification result is low risk, a first adjustment command is generated; the first adjustment command includes recording an event log and controlling the child seat to adjust the backrest to a first preset angle; if the asphyxiation risk verification result is medium risk, a second adjustment command is generated; the second adjustment command includes adjusting the backrest of the child seat from the first preset angle to a recovery position; if the asphyxiation risk verification result is high risk, a third adjustment command is generated; the third adjustment command includes adjusting the angle of the backrest and the height of the headrest in the child seat to form a recovery position in which the target child's head is higher than his chest.

[0021] Compared with existing technologies, the technical effects achieved by adopting this solution are as follows: It defines a three-level gradient adjustment strategy that precisely corresponds to the three-level risk system: Low risk: Slightly adjust the backrest angle to provide a more comfortable sleeping posture, reflecting the intelligence and humanization of the system; Medium risk: Adjust to the recovery position to prepare for possible worsening risks and achieve early physical intervention; High risk: Adjust the backrest and headrest in conjunction to form a standard recovery position with the head higher than the chest, directly creating conditions for an open airway for the child through the seat's own movement.

[0022] In one embodiment of the present invention, the sleep stage determination includes: analyzing the collected physiological signals to obtain heart rate variability, body movement frequency and intensity distribution characteristics; based on the heart rate variability, body movement frequency and intensity distribution characteristics, classifying and outputting the current sleep stage through a machine learning model, the sleep stage including deep sleep, light sleep, REM sleep and sleep-wake transition period.

[0023] Compared with existing technologies, the technical effect achieved by this solution is: by combining the physiological characteristics of heart rate variability and body movement frequency or intensity with machine learning models, accurate classification of sleep stages can be achieved.

[0024] In one embodiment of the present invention, a smart cockpit child sleep detection system for new energy vehicles is provided. The child sleep detection system can implement the child sleep detection method described in any of the above embodiments. The child sleep detection system includes: a child seat, comprising a backrest and a seating area, wherein the backrest is rotatable relative to the seating area; the backrest includes a backrest and a headrest; a first sensing unit, built into the backrest or headrest of the child seat, for collecting physiological state information of a target child sitting in the child seat; a second sensing unit, built into the headrest or the front guardrail structure of the child seat, for collecting a body surface temperature distribution map of the target child and the target child's body posture relative to the child seat; a decision module, communicatively connected to the first and second sensing units, for determining whether the target child in the child seat experiences an abnormal breathing event, and for performing a fusion analysis based on the abnormal breathing event, the temperature drop characteristics of the mouth and nose area in the body surface temperature distribution map, and the body posture to generate a suffocation risk verification result; and an execution module, communicatively connected to the decision module, for receiving and executing a seat adjustment command generated by the decision module based on the suffocation risk verification result to adjust the posture of the child seat.

[0025] Compared with existing technologies, the technical effects achieved by adopting this technical solution are: it can achieve the technical effects in any of the above examples, which will not be elaborated further here.

[0026] By adopting the technical solution of the present invention, the following technical effects can be achieved: (1) This invention collects the respiratory signals of the target child in the child seat and monitors the respiratory signals to determine whether there is an abnormal respiratory event in the child in the child seat; if so, it performs cross-validation in the vehicle environment by fusing and analyzing the respiratory signals, the temperature of the mouth and nose area and the body posture of the child, and realizes the accurate identification of suffocation risk; by directly linking the suffocation risk verification results with the seat mechanical mechanism, it can automatically generate and execute gradient adjustment commands according to different risk levels to carry out physical intervention for resuscitation position adjustment; directly relieve suffocation risk factors such as mouth and nose compression, and realize targeted intervention; (2) By screening with both signal-to-noise ratio and signal quality index, the interruption of low-quality signal segment is marked as an unreliable event and removed, thereby reducing the interference of subsequent fusion analysis from the source and improving the credibility of the interruption event entering the verification stage; and only the interruption of high-quality signal segment is marked as an event to be verified, thereby reducing system energy consumption and improving overall detection efficiency. (3) By synchronously collecting body motion signals, it can be determined whether the event to be verified is caused by body motion, which can effectively eliminate false interruptions caused by turning over or displacement, and greatly reduce the invalid activation rate of the infrared module. Attached Figure Description

[0027] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings to be used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Figure 1 A flowchart illustrating a method for detecting children's sleep in a smart cockpit of a new energy vehicle, provided as an embodiment of the present invention. Detailed Implementation

[0028] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0029] In the specific application scenarios of child safety seats in vehicles, children are in a restrained and relatively fixed space. While this provides a stable basis for continuous monitoring, it also introduces unique safety risks: during vehicle operation, vibrations, acceleration and deceleration inertia, and children's unconscious movements can easily cause their heads to tilt forward, slide sideways, or their bodies to slip, resulting in their mouths and noses being partially or completely covered by the seat's shoulder straps, side wing protectors, or soft padding, leading to breathing obstruction or even suffocation. These risks are highly insidious and sudden due to the noise and vibration interference of the driving environment and the difficulty for guardians in the driver's seat to observe closely in real time.

[0030] See Figure 1 , Figure 1 A flowchart illustrating a child sleep detection method in a smart cockpit of a new energy vehicle, provided by an embodiment of the present invention; specifically, the child sleep detection method is applied to a child seat, which is located in the smart cockpit; the child sleep detection method includes: S1: Collect respiratory signals from the target child sitting in the child seat and verify respiratory interruption based on the respiratory signals to identify abnormal respiratory events; S2: In response to the respiratory abnormality event to be verified, acquire the target child's body surface temperature distribution information and the target child's body shape and posture relative to the child seat; S3: Correlation analysis is performed between the temporal characteristics of abnormal respiratory events and the temporal changes in the temperature distribution of the mouth and nose area in the body surface temperature distribution information to generate asphyxiation risk verification results; S4: Based on the risk level of the suffocation risk verification results, control the child seat to perform the corresponding posture adjustment.

[0031] When the risk level is high, the support angle of the child seat is adjusted to position the child in a resuscitation position with their head higher than their chest.

[0032] In a specific example, the system collects and receives echo phase changes generated by the micro-movements of a child's chest and abdomen in a child seat; converts these phase change signals into real-time respiratory signals; the waveform amplitude is proportional to the respiratory depth, and the frequency is consistent with the respiratory rate. Based on this, the system sets the number of effective peak values ​​for the respiratory waveform within a real-time calculation window. When a series of consecutive respiratory peak values ​​are detected that are not valid and the waveform amplitude is consistently lower than the normal average amplitude, the event is immediately marked as a respiratory abnormality event. Only after receiving the respiratory abnormality event marker is the system captured of the temperature change process in the pre-defined target child's facial area; the system then performs region segmentation and extracts the temperature changes at the mouth and nose; by calculating the temperature drop characteristics of this area in consecutive frames, the respiratory abnormality event and the temperature drop characteristics are temporally aligned and analyzed: only when the time difference between the occurrence of the respiratory abnormality and the start of the temperature drop is within ±2 seconds, and their durations overlap, is a valid correlation determined, and a suffocation risk verification result is generated.

[0033] For example, when an abnormal breathing event is detected, the event is immediately marked with a timestamp T1; simultaneously, the starting timestamp T2 of the continuous drop in temperature in the mouth and nose area is recorded; by calculating the time difference between T1 and T2, if the difference is within ±2 seconds, and the duration of the abnormal breathing event overlaps with the duration of the temperature drop, then the two are determined to be physiologically related, generating a high-risk verification result; for example, if T1 is the 3rd second and T2 is the 4th second, with a difference of 1 second, the duration of the breathing interruption is from the 3rd to the 9th second, which overlaps with the duration of the temperature drop from the 4th to the 8th second, i.e., from the 4th to the 8th second, then the two are determined to be physiologically related, generating a high-risk asphyxiation verification result.

[0034] It should be noted that one of the direct physiological manifestations of suffocation is the drop in temperature in the mouth and nose area due to airflow obstruction, while other causes of respiratory signal interruption, such as clothing covering the chest, usually do not accompany this characteristic. Therefore, by acquiring the spatial distribution of body surface temperature, the mouth and nose area can be accurately located and its temperature change characteristics extracted.

[0035] Furthermore, respiratory interruption verification is performed based on the respiratory signals to identify abnormal respiratory events, including: The respiratory signal is filtered and denoised to extract a clean respiratory waveform; Calculate the real-time respiratory rate based on respiratory signals; The determination of abnormal respiratory events specifically includes: When the amplitude of the pure breathing waveform is continuously lower than the first amplitude threshold and reaches the first duration, the breathing interruption judgment condition is determined to be met. And / or when the real-time respiratory rate is continuously lower than the first frequency threshold and reaches the first duration, the respiratory interruption determination condition is determined to be met.

[0036] Specifically, the raw respiratory signal is preprocessed to eliminate the interference of environmental noise on the waveform morphology and retain the pure respiratory waveform that reflects chest and abdominal movements; and a real-time respiratory rate is generated by detecting the waveform period; on this basis, the waveform amplitude and respiratory rate are monitored. When the waveform amplitude is continuously lower than the first amplitude threshold and the duration exceeds the first duration, it is determined to be a sleep apnea event; when the respiratory rate is continuously lower than the first frequency threshold and the duration exceeds the first duration, it is determined to be a respiratory rhythm abnormality event.

[0037] For example, the first duration is defined as 3 seconds, the first amplitude threshold is 25% of the average amplitude of a normal respiratory waveform, and the first frequency threshold is 10 breaths / minute. When the waveform amplitude is detected to be below 25% of the threshold for 5 consecutive seconds, a respiratory apnea event is immediately triggered. Simultaneously, if the respiratory rate is detected to be below 10 breaths / minute for 5 consecutive seconds, a respiratory rhythm abnormality is triggered. A respiratory abnormality event alarm is generated when either condition is met, thus achieving dual detection of both complete respiratory arrest and severe bradyventricular contraction. Furthermore, it effectively reduces the probability of misjudgment caused by instantaneous distortion or local interference of sensor signals. Through the dual verification mechanism of waveform amplitude and respiratory rate, it can accurately distinguish between real respiratory abnormalities and transient signal fluctuations.

[0038] Furthermore, the process of verifying respiratory interruption based on the respiratory signal to identify abnormal respiratory events also includes: Calculate the signal-to-noise ratio and signal quality index of a clean breathing waveform; Preliminary classification and credibility assessment of abnormal respiratory events based on signal-to-noise ratio and signal quality index; If the signal-to-noise ratio is lower than the first threshold, or the signal quality index is lower than the first quality threshold, it is determined to be a low-quality signal segment, and the abnormal breathing events in the signal segment are marked as unreliable events and removed. If the signal-to-noise ratio is higher than the first threshold and the signal quality index is higher than the first quality threshold, then the respiratory abnormality event is determined to have occurred in the high-quality signal segment and is marked as an event to be verified.

[0039] Among them, the signal-to-noise ratio (SNR) refers to the energy ratio of the effective component to the noise component in the respiratory signal, which is used to quantify the degree of influence of the signal on the environmental electromagnetic interference or the noise of the equipment itself; the signal quality index refers to the integrity and stability index of the respiratory waveform, which is used to assess whether there is signal loss due to obstruction or body position shift during the sensor acquisition process; the low-quality signal segment refers to the signal range where the SNR or signal quality index does not meet the preset standard, and the high-quality signal segment refers to the signal range that meets both the SNR and signal quality index requirements.

[0040] Specifically, during respiratory signal monitoring, the raw signal is first filtered and denoised to obtain a clean respiratory waveform. The signal-to-noise ratio (SNR) and signal quality index are calculated in parallel. When strong electromagnetic interference or poor equipment contact exists, the SNR will significantly decrease. When a child's position changes, causing radar beam shift, the signal quality index will decrease accordingly. A dynamic threshold determination mechanism is employed.

[0041] For example, the signal-to-noise ratio (SNR) threshold can be set to the range of 8dB to 12dB, and the signal quality index (SQI) threshold can be set to the range of 0.7 to 0.9. When the SNR or SQI is below either threshold, the time period is automatically marked as a low-quality signal segment, and all abnormal breathing events within that time period are removed to avoid misjudgment due to signal distortion. Only when both the SNR and SQI are above the threshold will the corresponding abnormal breathing event be marked as an event to be verified, providing reliable input for subsequent infrared thermal imaging verification.

[0042] Furthermore, methods for detecting children's sleep also include: Synchronously collect the body movement signals of the target child in the child seat; the body movement signals include changes in the target child's body shape and posture in the child seat; Analyze respiratory abnormalities marked as events to be verified by combining body movement signals; If the intensity of the body motion signal exceeds the body motion intensity threshold, the event to be verified is determined to be a false interruption and will not trigger subsequent temperature acquisition and attitude adjustment.

[0043] Among them, body movement signal refers to the motion characteristic signal generated by synchronously capturing the child's limb activities, which is used to reflect the child's turning over, limb twitching and other movements during sleep; body movement intensity threshold refers to the pre-set critical value used to distinguish between normal body movement and disturbing violent movement, which is used to determine whether body movement is sufficient to cause respiratory signal distortion; and false interruption refers to abnormal respiratory signal caused by body movement interference rather than a real suffocation event.

[0044] Specifically, body movement signals are acquired simultaneously with respiratory signals to ensure consistency in data analysis when respiratory abnormalities occur. When a respiratory abnormality is marked as pending verification, the intensity of the body movement signal is extracted within the corresponding time period and compared with the body movement intensity threshold. If the body movement intensity exceeds the threshold, it indicates that there is intense limb activity during that time period. In this case, the interruption of the respiratory signal may be caused by body movement interference rather than actual suffocation, and therefore the current pending verification event is determined to be a false interruption. Through the correlation analysis between movement state and respiratory abnormality, a dual verification logic is constructed, and the thermal imaging verification process is only initiated when there is no intense body movement.

[0045] For example, the system detects a 5-second respiratory interruption, which is marked as an event to be verified because of the high signal-to-noise ratio during that period. Before thermal imaging verification, the body movement signals within those 5 seconds are traced back and their average intensity is calculated. If the calculated body movement signal intensity exceeds the body movement intensity threshold, it is determined that the respiratory interruption is very likely caused by signal distortion due to the child's violent turning or posture adjustment, rather than actual physiological asphyxia.

[0046] Furthermore, the temporal characteristics of abnormal respiratory events are correlated with the temporal changes in temperature in the oral and nasal regions of the body surface temperature distribution information to generate asphyxiation risk verification results: Identify the mouth and nose area in a body surface temperature distribution map; Calculate the rate of temperature decrease in the oral and nasal region over a continuous time period; If the rate of temperature decrease is greater than the preset rate threshold, the asphyxiation risk verification result is determined to be high risk; And / or identify the cheek region in a body surface temperature distribution map; Compare the temperature of the mouth and nose area with the temperature of the cheek area; If the temperature of the mouth and nose area is lower than that of the cheek area and reaches a preset temperature difference threshold, the suffocation risk verification result is determined to be high risk.

[0047] The temperature drop rate refers to how fast the temperature changes in the mouth and nose area. It can be calculated by measuring the temperature difference between two consecutive frames of thermal imaging data. This is used to calculate the rate of heat loss caused by apnea. The cheek area refers to the facial area adjacent to the mouth and nose area and not directly affected by respiratory airflow. It can be used as a reference area for temperature comparison to eliminate environmental interference.

[0048] Upon receiving a respiratory abnormality event to be verified, a facial thermal image is acquired. First, the mouth and nose area is located, and the rate of temperature decrease in this area over three consecutive frames is calculated. Simultaneously, the cheek area is automatically identified as a reference, and the real-time temperature difference between the mouth and nose and the cheek is calculated. Based on this, a parallel judgment logic is used. If the rate of temperature decrease is greater than a preset rate threshold, the asphyxiation risk verification result is determined to be high risk. If the temperature of the mouth and nose area is lower than the temperature of the cheek area to a preset temperature difference threshold, the asphyxiation risk verification result is determined to be high risk.

[0049] For example, a preset rate threshold is defined as 0.5℃ / second; a preset temperature difference threshold is -2℃. When the system detects an abnormal breathing event, the thermal imaging module shows that the temperature of the nose and mouth drops from 36.5℃ to 35.3℃ within 2 seconds, with a rate of decrease of 0.6℃ / second, exceeding the preset rate threshold. At the same time, the algorithm detects that the temperature of the cheek area is stable at 36.0℃. At this time, the temperature difference between the nose and mouth is -0.7℃, which is greater than the preset temperature difference threshold. Although the temperature difference condition is not triggered, a high-risk verification result will still be generated because the sudden temperature drop condition is met. Conversely, if the indoor air conditioning causes the ambient temperature to drop uniformly, the temperature of the nose and mouth drops from 36.0℃ to 35.0℃, with a rate of decrease that does not exceed the preset rate threshold, but the cheek temperature remains at 35.8℃. At this time, the temperature difference between the nose and mouth is -0.8℃, which is greater than the preset temperature difference threshold. The system will not determine high risk, thus effectively avoiding false alarms caused by environmental factors.

[0050] Furthermore, the temporal characteristics of abnormal respiratory events are correlated with the temporal changes in temperature in the oral and nasal regions of the body surface temperature distribution information to generate asphyxiation risk verification results, which also include: If the rate of temperature decrease is less than the preset rate threshold, but greater than the preset lower limit threshold for medium risk, then the asphyxiation risk verification result is determined to be medium risk. And / or if the temperature of the mouth and nose area is lower than that of the cheek area, but the temperature difference between the mouth and nose area and the cheek area is less than the preset temperature difference threshold and continues for a preset duration, then the suffocation risk verification result is determined to be medium risk. If the rate of temperature decrease is lower than the preset lower limit threshold for medium risk, and the temperature difference between the mouth and nose area and the cheek area is less than the preset temperature difference threshold, then the suffocation risk verification result is determined to be low risk.

[0051] Specifically, when the detected rate of temperature decrease is within the medium-risk range, a lower limit threshold is set to avoid misjudging slow temperature changes as high-risk, while a medium-risk assessment is made based on the rate change trend. When the temperature difference between the mouth, nose, and cheeks does not reach the high-risk threshold but there are persistent low-temperature characteristics, a potential medium-risk state is identified to prevent misjudgments caused by short-term temperature fluctuations. For cases where the rate is below the medium-risk rate lower limit threshold and the temperature difference is insufficient, a dual-condition constraint mechanism is used to filter out interfering factors such as changes in ambient temperature, accurately distinguishing between physiological fluctuations and real risk events.

[0052] For example, if the medium-risk rate range is defined as 0.2℃ / second to 0.5℃ / second, then the lower limit threshold for the medium-risk rate is 0.2℃ / second, and the medium-risk temperature difference condition is 1.0℃ to ≤2.0℃, lasting for at least 3 seconds. In the low-risk condition, the temperature decrease rate is 0.2℃ / second, and the preset temperature difference threshold is 1.0℃. When the temperature in the mouth and nose area drops from 36.4℃ to 35.6℃ within 4 seconds, with a decrease rate of 0.2℃ / second, and the cheek temperature stabilizes at 36.0℃, although the temperature difference is 0.4℃ / second... If the temperature does not reach the risk threshold, but the rate of decrease is within the lower limit of the medium-risk range, the system will activate a medium-risk warning. Conversely, if the ambient temperature decreases uniformly, causing the temperature of the mouth and nose to slowly drop from 36.0℃ to 35.2℃ at a rate of 0.2℃ / second, while the temperature of the cheeks simultaneously drops to 35.1℃, with a temperature difference of 0.1℃, the system will accurately determine it as low-risk by applying dual constraints: the rate of temperature decrease must be less than or equal to 0.2℃ / second and the preset temperature difference threshold must be ≤1.0℃. This effectively avoids false alarms caused by environmental interference.

[0053] Furthermore, based on the risk level of the suffocation risk verification results, the child car seat is controlled to perform corresponding posture adjustments, including: Determine the sleep phase in response to abnormal respiratory events; If the respiratory abnormality event occurs during the sleep-wake transition, an additional verification process is initiated; Additional verification procedures include: Acquire micro-motion signals of the target child's peri-ocular region in a child seat; The micro-motion signals are compared with a pre-stored library of historical respiratory fluctuation patterns during the eye-opening period; If the micro-motion signal matches the historical awakening characteristics, the current state is determined to be a false positive, and the seat adjustment command is not triggered.

[0054] Among them, the additional verification process refers to the event review mechanism triggered in a specific sleep stage to capture physiological characteristics related to eyelid movement; the historical respiratory fluctuation pattern library during eye-opening refers to a pre-established database of respiratory and eye movement correlation features in a waking state. Furthermore, micro-motion signals refer to signals collected by micro-motion sensors integrated into the headrest of the child seat or at the relative position of the child's face, reflecting minute movements of the child's eyelids or periocular muscles. When a child is awake or asleep, rapid eye movements or weak characteristic vibrations of the eyelids are transmitted to the micro-motion sensors through the facial soft tissues. During deep sleep or light sleep, these micro-motion signals essentially disappear or exhibit extremely low-frequency changes. Therefore, by comparing the detected micro-motion signals with a pre-collected historical signal database marked as being from awake periods, if the similarity exceeds a preset threshold, the current breathing interruption event is determined to be a false positive caused by the child's normal awakening activity, thereby avoiding the accidental triggering of seat adjustment commands and reducing unnecessary intervention.

[0055] Alternatively, the micro-motion sensor may be a piezoelectric thin-film sensor, a capacitive proximity sensor, or an optical reflective sensor.

[0056] Furthermore, based on the risk level of the suffocation risk verification results, the child car seat is controlled to perform corresponding posture adjustments, including: If the suffocation risk verification result is low risk, a first adjustment command is generated; the first adjustment command includes recording the event log and controlling the child seat to adjust the backrest to a first preset angle; If the suffocation risk verification result is medium risk, a second adjustment command is generated; the second adjustment command includes adjusting the back of the child seat from the first preset angle to the restored body position; If the asphyxiation risk verification result is high, a third adjustment command is generated; the third adjustment command includes adjusting the angle of the chair back and the height of the headrest in the child seat to form a resuscitation position in which the target child's head is higher than his chest.

[0057] Furthermore, once the suffocation risk verification result is low and the child seat has completed the first adjustment command, the child in the child seat will continue to be monitored.

[0058] Furthermore, if the suffocation risk verification result is medium risk, and the child seat executes the second adjustment command, the low-light indicator light in the child seat will flash and a secondary sound reminder will be triggered, and the warning information will be sent to the bound monitoring terminal.

[0059] Furthermore, if the suffocation risk verification result is high risk, and the child seat executes the third adjustment command, the child seat's high-intensity audible and visual alarm will be triggered immediately, and an emergency alarm message containing risk details and a real-time environmental screenshot will be sent to the monitoring terminal at the same time.

[0060] Among them, the monitoring terminal is the guardian or caregiver of the target child.

[0061] Among them, the risk level corresponding to the suffocation risk verification result refers to the quantitative classification of the suffocation possibility after multi-dimensional data analysis, which can effectively distinguish potential risks of different degrees of severity.

[0062] Furthermore, in the normal sitting posture of a child seat, the backrest and the seat are usually at 100°; in low-risk events, the first preset angle is 125° between the backrest and the seat, that is, the backrest is rotated 25° relative to the seat; in medium-risk events, the recovery position is 155°, that is, the backrest is rotated another 30° relative to the seat; in high-risk events, the recovery position with the head higher than the chest is 165°, at which point the backrest is rotated another 10° relative to the seat.

[0063] Based on specific work conditions, upon detecting an abnormal breathing event, a suffocation risk verification result is generated by fusing and analyzing the temperature change characteristics of the mouth and nose area. For low-risk events, it indicates that the child is in a slightly uncomfortable sleeping position. At this time, the chair back tilt angle, i.e., the first preset angle, is adjusted according to the first adjustment command to help the target child expand their chest cavity and improve breathing patency. The event time, sensor data, and other parameters are written to a log file in the local storage to maintain the current monitoring status without interruption. For medium-risk events, it indicates a real threat of suffocation. At this time, the chair back angle is adjusted from the first preset angle to the restored position so that the target child can be in a position conducive to ventilation, creating an unobstructed breathing environment. Environment; control the flashing frequency of the indicator light in the child seat, while playing short, intermittent alert sounds and sending a warning containing a risk level code to a preset mobile terminal; for high-risk events, indicating that the child is in serious suffocation danger; at this time, the linkage adjustment mechanism raises the headrest in the child seat to its highest point to create a position where the head is significantly higher than the chest, using gravity to help the tongue fall back and move forward, maximizing airway opening; simultaneously, the high-brightness flashing light and continuous alarm sound in the child seat are activated, and the camera is simultaneously called to capture images of the current environment, packaging and sending multimodal data including breathing waveforms, temperature distribution maps, and on-site photos to the monitoring terminal for the child's guardian or caregiver to handle and respond to.

[0064] Furthermore, the determination of the sleep phase includes: The collected physiological signals are analyzed to obtain the characteristics of heart rate variability, body motion frequency and intensity distribution; Based on the distribution characteristics of heart rate variability, body movement frequency and intensity, the current sleep stage is classified and output through a machine learning model. The sleep stages include deep sleep, light sleep, REM sleep, and sleep-wake transition.

[0065] Among them, heart rate variability refers to the degree of change in the interval between heartbeat cycles, which is used to reflect the activity state of the autonomic nervous system; body movement frequency and intensity distribution characteristics refer to the number of body movements and their amplitude changes per unit time, which are used to characterize the degree of motor inhibition in different sleep stages.

[0066] Among them, the machine learning model can adopt the forest classification model.

[0067] Based on the specific working method, when an abnormal respiratory event is detected, the system first determines whether the event occurred during the sleep-wake transition period by using heart rate variability parameters and body movement intensity distribution characteristics. If it is in this stage, the micro-motion pressure sensing device in the periorbital area is activated to collect the frequency characteristics of the pressure signal generated by eyelid movement. This signal is input into the historical respiratory fluctuation pattern database during the eye-opening period and compared with the typical waveform characteristics recorded in the waking state. When a high-frequency micro-motion signal is detected that is highly similar to the respiratory fluctuation pattern during the eye-opening period, the system determines that the child is in a state of brief wakefulness, and the interruption of breathing is a normal physiological phenomenon. This is classified as a false positive event, and the alarm process is blocked.

[0068] Based on the above, in one instance, the system detected a 6-second respiratory interruption, but the sleep stage recognition module indicated that the current period was a transitional phase between wakefulness and wakefulness. The additional verification process was then initiated, and the micro-motion pressure sensor detected a 3Hz micro-motion signal lasting 2 seconds in the periorbital area. The signal characteristics matched the historical patterns of children turning over with their eyes open in the historical respiratory fluctuation pattern database with a 90% match. Based on this, it was determined that the respiratory interruption was caused by the child's physical activity in the pre-awakening stage, rather than a suffocation event. Therefore, the gradient response mechanism was not triggered, and only the event log was recorded. Conversely, if the sleep stage recognition module detects a respiratory interruption during deep sleep and the thermal imaging verification is high-risk, the additional verification is skipped, and a high-risk response is executed directly: the high-decibel alarm is activated, and a "high-risk suffocation alarm" and a real-time thermal imaging screenshot are sent to the parent's mobile phone.

[0069] Furthermore, methods for detecting children's sleep also include: Collect normal respiratory, physical movement, and temperature data from children when they are in a known safe situation. Establish a personalized baseline model of physiological parameters and a threshold for the range of fluctuations for the child.

[0070] Among them, the known safe state refers to an environment where the child is in an environment with no risk of suffocation and stable physiological activity. This can be achieved by the guardian confirming that the child is in a normal sleeping posture and breathing evenly. This state serves as the baseline scenario for data collection. The personalized physiological parameter baseline model is a reference model established based on the physiological data continuously collected from individual children in a safe state. This can be achieved by identifying and extracting features from respiratory rate, body movement amplitude, and body surface temperature. This model can reflect the unique physiological fluctuation patterns of children. The fluctuation range threshold refers to the allowable range of individualized parameter changes. The threshold is dynamically updated as the child grows and the environment changes.

[0071] Specifically, when a child is in a safe state as confirmed by a guardian, respiratory sensors, body movement sensors, and temperature sensors simultaneously collect physiological signals over a continuous period of time. These signals are filtered and feature extracted to form an initial dataset. By correlating the periodic changes in respiratory rate, the distribution characteristics of body movement intensity, and temperature fluctuations in the mouth and nose area, a baseline model containing time-series features is established. During subsequent real-time monitoring, the currently detected physiological parameters are matched with the corresponding time period data in the baseline model. If the relevant parameters deviate from the individualized fluctuation threshold when an abnormal respiratory event occurs, cross-validation is performed using multi-sensor data to reduce the probability of misjudgment.

[0072] This application further proposes a child sleep detection system in the intelligent cockpit of a new energy vehicle. The child sleep detection system can implement the child sleep detection method in any of the above examples. The child sleep detection system includes: a child seat, a first sensing unit, a second sensing unit, a decision module, and an execution module. The child seat includes a backrest and a seating area, and the backrest can rotate relative to the seating area. The backrest includes a seat back and a headrest. The first sensing unit is built into the backrest or headrest of the child seat and is used to collect physiological state information of the target child sitting in the child seat. The second sensing unit is built into the headrest or the front guardrail structure of the child seat and is used to collect the body surface temperature distribution map of the target child and the body posture of the target child relative to the child seat. The decision module is communicatively connected to the first sensing unit and the second sensing unit. It is used to determine whether the target child in the child seat has a respiratory abnormality event, and to perform a fusion analysis based on the respiratory abnormality event, the temperature drop characteristics of the mouth and nose area in the body surface temperature distribution map, and the body posture to generate a suffocation risk verification result. The execution module is communicatively connected to the decision module and is used to receive and execute the seat adjustment command generated by the decision module to adjust the posture of the child seat according to the suffocation risk verification result.

[0073] Optionally, the first sensor is a millimeter-wave radar sensor.

[0074] Optionally, the second sensor is an infrared temperature sensor.

[0075] The alarm unit includes at least a low-light indicator light and a high-intensity audible and visual alarm.

[0076] Specifically, the first sensor is integrated into the backrest or headrest of the child seat, utilizing the rigid support of the child seat structure to maintain a fixed distance between the sensor and the child's body, thereby eliminating signal acquisition position shifts caused by the child's movements. When the respiratory rate signal is abnormal, the decision module sends an activation command to the second sensor, triggering it to collect temperature distribution data in the mouth and nose area. The lens of the second sensor is kept aligned through a reserved hole in the guardrail or headrest, ensuring that the temperature data and the respiratory signal interruption event are spatially correlated. The decision module determines whether there is a real risk of suffocation by comparing the time period of missing respiratory signals with the spatiotemporal characteristics of the sudden drop in temperature in the mouth and nose area. All modules adopt an embedded layout, making the sensor and child seat structure an integrated design, avoiding external devices from hindering the child's movements.

[0077] It should be noted that this invention is mainly aimed at children's sleep detection. The rotation of the backrest relative to the seating part in the seat body to adjust the angle between the two is a conventional technical means in this field, and the specific structure between the two will not be described in detail here.

[0078] While the present invention has been disclosed above, it is not limited thereto. Any person skilled in the art can make various modifications and alterations without departing from the spirit and scope of the invention; therefore, the scope of protection of the present invention should be determined by the scope defined in the claims.

Claims

1. A method for detecting children's sleep in the intelligent cockpit of a new energy vehicle, characterized in that, The child sleep detection method is applied to a child car seat, which is located in the smart cockpit; The methods for detecting children's sleep include: The respiratory signals of the target child sitting in the child seat are collected, and respiratory interruption is verified based on the respiratory signals to identify abnormal respiratory events; In response to the respiratory abnormality event to be verified, the body surface temperature distribution information of the target child and the body shape and posture of the target child relative to the child seat are obtained; The temporal characteristics of the respiratory abnormality events are correlated with the temporal changes in the temperature of the mouth and nose area in the body surface temperature distribution information to generate asphyxiation risk verification results. Based on the risk level of the asphyxiation risk verification result, the child seat is controlled to perform corresponding posture adjustments; wherein, when the risk level is high risk, the support angle of the child seat is adjusted to make the target child assume a resuscitation position with the head higher than the chest.

2. The method for detecting children's sleep according to claim 1, characterized in that, The step of verifying respiratory interruption based on the respiratory signal to identify abnormal respiratory events includes: The respiratory signal is filtered and denoised to extract a clean respiratory waveform; Calculate the real-time respiratory rate based on the respiratory signal; The determination of abnormal respiratory events specifically includes: When the amplitude of the pure breathing waveform is continuously lower than the first amplitude threshold and reaches the first duration, the breathing interruption determination condition is determined to be met. And / or when the real-time respiratory rate is continuously lower than the first frequency threshold and reaches the first duration, the respiratory interruption determination condition is determined to be met.

3. The method for detecting children's sleep according to claim 2, characterized in that, The step of verifying respiratory interruption based on the respiratory signal to identify abnormal respiratory events also includes: Calculate the signal-to-noise ratio and signal quality index of the pure breathing waveform; Based on the signal-to-noise ratio and the signal quality index, the respiratory abnormality events are preliminarily classified and their credibility is assessed. If the signal-to-noise ratio is lower than the first threshold, or the signal quality index is lower than the first quality threshold, it is determined to be a low-quality signal segment, and the respiratory abnormality event in the signal segment is marked as an unreliable event and removed. If the signal-to-noise ratio is higher than the first threshold and the signal quality index is higher than the first quality threshold, then the respiratory abnormality event is determined to have occurred in a high-quality signal segment and is marked as an event to be verified.

4. The method for detecting children's sleep according to claim 3, characterized in that, The method for detecting children's sleep also includes: The body movement signals of the target child in the child seat are collected synchronously; the body movement signals include the changes in the body shape and posture of the target child in the child seat; The abnormal respiratory events marked as events to be verified are analyzed in conjunction with the body movement signals. If the intensity of the body movement signal exceeds the body movement intensity threshold, the event to be verified is determined to be a false interruption.

5. The method for detecting children's sleep according to claim 1, characterized in that, The step of correlating the temporal characteristics of the respiratory abnormality event with the temporal temperature change sequence of the oral and nasal regions in the body surface temperature distribution information to generate a suffocation risk verification result includes: Identify the mouth and nose region in the body surface temperature distribution map; Calculate the rate of temperature decrease in the oral and nasal region over a continuous time period; If the rate of temperature decrease is greater than a preset rate threshold, the asphyxiation risk verification result is determined to be high risk; And / or identify the cheek region in the body surface temperature distribution map; The temperature of the mouth and nose area is compared with the temperature of the cheek area; If the temperature of the mouth and nose area is lower than the temperature of the cheek area and reaches a preset temperature difference threshold, the suffocation risk verification result is determined to be high risk.

6. The method for detecting children's sleep according to claim 5, characterized in that, The step of correlating the temporal characteristics of the respiratory abnormality event with the temporal temperature change sequence of the oral and nasal regions in the body surface temperature distribution information to generate a suffocation risk verification result further includes: If the rate of temperature decrease is less than a preset rate threshold, but greater than a preset lower limit threshold for medium-risk rates, then the asphyxiation risk verification result is determined to be medium-risk. And / or if the temperature of the mouth and nose area is lower than the temperature of the cheek area, but the temperature difference between the mouth and nose area and the cheek area is less than a preset temperature difference threshold and continues for a preset duration, then the suffocation risk verification result is determined to be medium risk. If the rate of temperature decrease is lower than the preset lower limit threshold for medium risk, and the temperature difference between the mouth and nose area and the cheek area is less than the preset temperature difference threshold, then the asphyxiation risk verification result is determined to be low risk.

7. The method for detecting children's sleep according to claim 6, characterized in that, The step of controlling the child seat to perform corresponding posture adjustments based on the risk level of the suffocation risk verification result includes: The sleep process is determined for the aforementioned abnormal breathing events; If the respiratory abnormality event occurs during the sleep-wake transition period, an additional verification process is initiated; The additional verification process includes: Acquire the micro-motion signals of the target child's periorbital area in the child seat; The micro-motion signal is compared with a pre-stored library of historical respiratory fluctuation patterns during the eye-opening period; If the micro-motion signal matches the historical awakening characteristics, the current state is determined to be a false positive, and the seat adjustment command is not triggered.

8. The method for detecting children's sleep according to claim 6, characterized in that, The step of controlling the child seat to perform corresponding posture adjustments based on the risk level of the suffocation risk verification result includes: If the suffocation risk verification result is low risk, a first adjustment command is generated; the first adjustment command includes recording the event log and controlling the child seat to adjust the backrest to a first preset angle; If the suffocation risk verification result is medium risk, a second adjustment command is generated; the second adjustment command includes adjusting the backrest of the child seat from the first preset angle to a restored body position; If the asphyxiation risk verification result is high risk, a third adjustment command is generated; the third adjustment command includes adjusting the angle of the chair back and the height of the headrest in the child seat to form the resuscitation position in which the target child's head is higher than his chest.

9. The method for detecting children's sleep according to claim 7, characterized in that, The sleep process determination includes: The collected physiological signals are analyzed to obtain the characteristics of heart rate variability, body motion frequency and intensity distribution; Based on the heart rate variability, body movement frequency, and intensity distribution characteristics, the current sleep stage is classified and output through a machine learning model. The sleep stage includes deep sleep, light sleep, REM sleep, and the sleep-wake transition period.

10. A child sleep monitoring system in the intelligent cockpit of a new energy vehicle, characterized in that, The child sleep detection system is capable of implementing the child sleep detection method as described in any one of claims 1 to 9, and the child sleep detection system comprises: A child seat; comprising a backrest and a seating area, wherein the backrest is rotatable relative to the seating area; the backrest includes a backrest and a headrest; The first sensing unit, built into the backrest or the headrest, is used to collect physiological state information of the target child sitting in the child seat; The second sensing unit is built into the front guard structure of the headrest or the child seat, and is used to collect the body surface temperature distribution map of the target child and the body shape and posture of the target child relative to the child seat; The decision module is communicatively connected to the first sensing unit and the second sensing unit; it is used to determine whether the target child in the child seat has an abnormal breathing event, and to perform a fusion analysis based on the abnormal breathing event, the temperature drop characteristics of the mouth and nose area in the body surface temperature distribution map, and the body posture to generate a suffocation risk verification result. An execution module, which is communicatively connected to the decision module, is used to receive and execute the seat adjustment command of the child seat generated by the decision module based on the suffocation risk verification result, so as to adjust the posture of the child seat.

Images