Safety reminding method, system and equipment based on environmental perception and medium

Abstract

The invention provides a safety reminding method, system and equipment based on environmental perception and a medium. The method comprises the following steps: acquiring various environmental parameters of different dimensions of an environment in which a monitored object is located; based on a preset weight coefficient, performing fusion calculation according to the various environmental parameters to obtain a comprehensive risk index representing the overall risk of the environment; the weight coefficient is used for reflecting the influence degree of the environmental parameters on the overall environmental risk; determining the risk level of the current environment according to the comprehensive risk index and a comparison result between the numerical value of the at least one environmental parameter and a preset risk threshold value; and executing a corresponding reminding operation according to the risk level. According to the invention, active and intelligent safety monitoring and intervention on the environment where the monitored object is located are realized, and the accuracy and safety of risk identification are improved.

Description

Technical Field

[0001] This invention relates to the field of smart wearable device technology, and specifically to a safety alert method, system, device, and medium based on environmental perception. Background Technology

[0002] With the rapid development of the Internet of Things (IoT) and smart hardware technologies, children's smartwatches have become an important tool for ensuring children's safety. Currently, similar products on the market mainly focus on basic functions such as real-time location tracking, two-way calling, electronic fences, and pedometer tracking. These functions effectively help parents track their children's location and basic status, alleviating parental anxiety to some extent. However, the safety threats children face in daily life and learning stem not only from the uncertainty of their location but also from potential risks in their environment. For example, prolonged exposure to high noise levels, air pollution, or unsuitable lighting can cause irreversible damage to their hearing, respiratory system, and vision.

[0003] Existing technologies include devices that attempt to integrate environmental monitoring functions. For example, some smart devices are integrating single PM2.5 or noise sensors to detect specific environmental parameters. Other solutions propose accessing urban air quality monitoring networks via mobile apps to obtain macro-environmental data for the area where children reside. These technological solutions indicate that the industry has begun to pay attention to the impact of environmental factors on safety and is attempting to address these issues through technological means.

[0004] However, these existing technological solutions still have significant problems and limitations. First, they are limited in their monitoring dimensions, often targeting only one or two environmental parameters, failing to provide a comprehensive risk assessment of the complex environments children encounter. Second, they lack effective data fusion and intelligent analysis capabilities, typically simply comparing sensor readings to fixed thresholds, which is prone to false alarms due to instantaneous fluctuations and cannot identify trends of escalating risk. Third, the alert mechanisms are rigid, failing to provide differentiated alert strategies based on risk levels and the target audience (children or parents), resulting in a poor user experience. Finally, existing systems generally lack self-learning and adaptive capabilities, unable to dynamically optimize based on children's individual tolerance and parents' actual feedback, leading to a decline in accuracy and practicality over long-term use. Therefore, there is an urgent need for a comprehensive solution capable of multi-dimensional environmental perception, intelligent risk assessment, and personalized adaptive alerts. Summary of the Invention

[0005] In view of this, it is necessary to provide a safety alert method, system, device and medium based on environmental perception to solve the technical problem that the existing technology lacks the ability to perceive and respond to environmental risks in a real time, comprehensively and intelligently.

[0006] To address the aforementioned technical problems, in a first aspect, the present invention provides a safety alert method based on environmental perception, wherein the industrial equipment includes a human-machine interface, and includes the following steps: Obtain environmental parameters from multiple dimensions of the environment in which the monitored individual is located; Based on preset weighting coefficients, a comprehensive risk index characterizing the overall environmental risk is obtained by integrating and calculating various environmental parameters; the weighting coefficients are used to reflect the degree of influence of the environmental parameters on the overall environmental risk. The risk level of the current environment is determined based on the comparison results of the comprehensive risk index and at least one of the environmental parameters with the preset risk threshold. Execute the corresponding reminder action based on the risk level.

[0007] In one possible implementation, the environmental parameters include noise, air quality, light intensity, temperature, and humidity; the comprehensive risk index characterizing the overall environmental risk is obtained by fusing and calculating various environmental parameters based on preset weighting coefficients, including: The comprehensive risk index is obtained by combining each of the environmental parameters with its corresponding weight coefficients; wherein, the weight coefficients corresponding to each of the environmental parameters are dynamically configured according to at least one of the following influencing factors; the influencing factors include the scenario type in which the monitored object is located, user feedback information received for historical reminder operations, and historical activity pattern data of the monitored object.

[0008] In one possible implementation, obtaining multiple environmental parameters of the environment in which the monitored object is located includes: Environmental state data is obtained by sampling, and the sampling frequency of the environmental state data is related to the fluctuation state of the environmental state data. The environmental parameters are obtained by preprocessing the environmental state data.

[0009] In one possible implementation, obtaining multiple environmental parameters of the environment in which the monitored object is located further includes: Based on the sliding time window algorithm, the rate of change of the environmental parameters after filtering and outlier processing within a preset time interval is calculated. When the environmental parameter is outside the preset safety range and the rate of change exceeds the preset rate threshold, the current risk level determined based on the environmental parameter will be updated to the next risk level; wherein the risk represented by the next risk level is higher than the current risk level.

[0010] In one possible implementation, determining the risk level of the current environment based on a comparison of the values ​​of the comprehensive risk index and at least one of the environmental parameters with a preset risk threshold includes: The comprehensive risk index is compared with a preset risk threshold range; the preset risk threshold range includes an upper limit value and a lower limit value of the risk index. The environmental parameters are compared with their respective risk ranges; the risk ranges include a safe range, a second risk range, and a first risk range; the first risk range and the second risk range do not overlap. When the environmental parameter is within its corresponding first risk range, or the comprehensive risk index is greater than or equal to the upper limit of the risk index, or the environmental parameter is within its corresponding second risk range and the duration exceeds a preset duration, the risk level of the current environment is determined to be the first level. When the environmental parameter is within its corresponding second risk range and the duration does not exceed the preset duration, or when the comprehensive risk index is greater than or equal to the lower limit of the risk index and less than or equal to the upper limit of the risk index, the risk level of the current environment is determined to be the second level; the risk represented by the second level is lower than that of the first level. When all the environmental parameters are within the safe range and the comprehensive risk index is less than the lower limit of the risk index, the risk level of the current environment is determined to be a safe level.

[0011] In one possible implementation, the step of performing the corresponding reminder operation based on the risk level includes: When the risk level is Level 1, control the smart wearable device of the monitored object to perform the first risk warning operation, and control the smart wearable device of the guardian to perform the emergency reminder operation. When the risk level is Level 2, the smart wearable device of the monitored object is controlled to perform a second risk warning operation, and the smart wearable device of the guardian is controlled to perform a silent reminder operation; the reminder intensity of the second risk warning operation is lower than the reminder intensity of the first risk warning operation.

[0012] One possible implementation also includes: Obtain feedback information from the guardian regarding the reminder operation, including confirmation of risk and ignoring false alarms; Based on the feedback information, the number of consecutive false alarms for the same type of risk alert is counted; When the number of consecutive false alarms reaches a preset number, the risk threshold corresponding to the environmental parameter is increased, and the adjusted risk threshold is not lower than the preset safety standard limit.

[0013] Secondly, the present invention also provides a safety alert system based on environmental perception, comprising: The data acquisition and processing module is used to acquire environmental parameters of various dimensions of the environment in which the monitored object is located; The fusion calculation module is used to perform fusion calculations based on preset weighting coefficients and various environmental parameters to obtain a comprehensive risk index that characterizes the overall environmental risk; the weighting coefficients are used to reflect the degree of influence of the environmental parameters on the overall environmental risk. The risk assessment module is used to determine the risk level of the current environment based on the comparison results of the comprehensive risk index and at least one of the environmental parameters with a preset risk threshold. The control module is used to perform corresponding reminder operations based on the risk level.

[0014] Thirdly, the present invention also provides an electronic device, including a memory and a processor, wherein, The memory is used to store programs; The processor, coupled to the memory, is used to execute the program stored in the memory to implement the steps in the environment-aware security alert method described in any of the above implementations.

[0015] Fourthly, the present invention also provides a computer-readable storage medium for storing a computer-readable program or instruction, which, when executed by a processor, can implement the steps of the environment-aware security alert method described in any of the above implementations.

[0016] The beneficial effects of this invention are as follows: The environmental perception-based safety alert method, system, device, and medium provided by this invention firstly achieve comprehensive and three-dimensional perception of environmental safety factors through the collaborative work of multi-source heterogeneous sensors, overcoming the limitations of single-parameter assessment. Data normalization provides a foundation for subsequent fusion calculations of different data, transforming multiple environmental parameters into a single, quantitative comprehensive risk index, greatly simplifying the decision-making complexity of risk assessment. Based on various types of environmental parameters collected by multi-source heterogeneous sensors and the comprehensive risk index obtained through fusion calculation, the risk level of the current environment is assessed and determined, which is more in line with the actual environmental conditions and improves the accuracy of current environmental safety judgments. For example, during outdoor activities on smoggy days, the system can automatically focus on air pollution risks. Furthermore, the dual judgment mechanism (environmental parameters + comprehensive risk index) takes into account both sudden severity risks (such as sudden loud noise) and cumulative composite risks (such as multiple parameters slightly exceeding the standard), ensuring the timeliness of the alarm while avoiding the one-sidedness of the assessment, significantly reducing the false alarm rate of the system, and making the alarm more convincing. Furthermore, differentiated alert procedures are implemented for different risk levels, ensuring that necessary warnings are delivered while reducing unnecessary interference with those under guardianship (such as children) and guardians (such as parents). This provides safety assurance while improving the user experience, allowing guardians to intervene in a timely manner and take measures such as remote guidance or on-site assistance. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying 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.

[0018] Figure 1 A flowchart illustrating an embodiment of the environmental awareness-based safety alert method provided by the present invention; Figure 2 A flowchart illustrating another embodiment of the environmental awareness-based safety alert method provided by the present invention; Figure 3 A flowchart illustrating another embodiment of the environmental awareness-based safety alert method provided by the present invention; Figure 4 A flowchart illustrating another embodiment of the environmental awareness-based safety alert method provided by the present invention; Figure 5 A flowchart illustrating another embodiment of the environmental awareness-based safety alert method provided by the present invention; Figure 6A schematic diagram of an embodiment of the environmental perception-based safety alert system provided by the present invention; Figure 7 A schematic diagram of an embodiment of the electronic device provided by the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0020] In the description of the embodiments of the present invention, unless otherwise stated, "multiple" means two or more. "And / or" describes the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can represent three situations: A exists alone, A and B exist simultaneously, and B exists alone.

[0021] The terms "first," "second," etc., used in the embodiments of this invention are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a technical feature defined with "first" or "second" may explicitly or implicitly include at least one of that feature.

[0022] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0023] This invention provides a safety alert method, system, device, and medium based on environmental perception, which are described below.

[0024] Figure 1 A flowchart illustrating an embodiment of the environment-aware safety alert method provided by the present invention is shown below. Figure 1 As shown, the environmental awareness-based safety alert method includes the following steps: S101. Obtain environmental parameters of various dimensions of the environment in which the monitored object is located.

[0025] Specifically, the monitored individuals include children and the elderly, while the guardians include parents and medical personnel. By deploying multiple sensor nodes, various environmental parameters of the monitored individuals' environment are acquired in real time. The sensors can be connected to the data acquisition terminal via wired (RS485, Modbus) or wireless (LoRa, NB-IoT, Wi-Fi, ZigBee) methods. The data acquisition terminal, electronic devices, or servers can preprocess the collected raw environmental data to obtain environmental parameters. These environmental parameters include noise, air quality, light intensity, temperature, and humidity, as described in the embodiments below, and may also include air pressure, altitude, and electromagnetic field strength.

[0026] S102. Based on preset weighting coefficients, a comprehensive risk index representing the overall environmental risk is obtained by integrating various environmental parameters; the weighting coefficients are used to reflect the degree of influence of environmental parameters on the overall environmental risk.

[0027] Specifically, the weighting coefficients reflect the degree of influence of each environmental parameter on the overall environmental risk. Initial weighting coefficients can be set by environmental safety experts based on the hazard of the parameters. Alternatively, based on historical pollution events or anomaly records, principal component analysis (PCA) or entropy weighting can be used to calculate the contribution of each parameter to the risk event, with the contribution used as the initial weighting coefficient. Machine learning models (such as random forests and XGBoost) can also be used to train parameter importance and periodically update the weighting coefficients. A comprehensive risk index is obtained by weighted summation of various types of environmental data based on the preset weighting coefficients.

[0028] S103. Determine the current environmental risk level based on the comparison results of the comprehensive risk index and at least one environmental parameter with the preset risk threshold.

[0029] Specifically, the comprehensive risk index is mapped to discrete risk levels. Furthermore, different types of environmental parameters have pre-defined risk ranges. Risk levels can be three (such as the safety level, first level, and second level in the example below) or four (such as low risk level I, moderate risk level II, higher risk level III, and high risk level IV). Different risk levels have corresponding risk ranges; that is, there is a correspondence between risk ranges and risk levels. By matching the comprehensive risk index and environmental parameters with their respective risk ranges, the risk level of the current environment can be determined based on the matched risk ranges. Alternatively, a multi-dimensional risk threshold table can be directly created, storing the correspondence between each risk level and risk range.

[0030] S104. Execute the corresponding reminder operation according to the risk level.

[0031] Specifically, the multi-dimensional risk threshold table can also store the correspondence between each risk level and each alert action. Users can look up the corresponding alert action based on the risk level and then issue a risk alert based on the retrieved alert action.

[0032] In summary, the environmental perception-based safety alert method provided in this invention first achieves comprehensive and three-dimensional perception of environmental safety factors through the collaborative work of multi-source heterogeneous sensors, overcoming the limitations of single-parameter assessment. Data normalization provides a foundation for subsequent fusion calculations of different data, transforming multiple environmental parameters into a single, quantifiable comprehensive risk index, greatly simplifying the decision-making complexity of risk assessment. Based on various types of environmental parameters collected by multi-source heterogeneous sensors and the comprehensive risk index obtained through weighted calculation, the risk level of the current environment is assessed and determined, which is more in line with the actual environmental conditions and improves the accuracy of current environmental safety judgments. For example, during outdoor activities on smoggy days, the system can automatically focus on air pollution risks. Furthermore, the dual judgment mechanism (environmental parameters + comprehensive risk index) takes into account both sudden severity risks (such as sudden loud noise) and cumulative composite risks (such as multiple parameters slightly exceeding the standard), ensuring the timeliness of the alert while avoiding the one-sidedness of the assessment, significantly reducing the false alarm rate of the system, and making the alert more convincing. Furthermore, differentiated alert procedures are implemented for different risk levels, ensuring that necessary warnings are delivered while reducing unnecessary interference with those under guardianship (such as children) and guardians (such as parents). This provides safety assurance while improving the user experience, allowing guardians to intervene in a timely manner and take measures such as remote guidance or on-site assistance.

[0033] Because fixed weighting coefficients cannot accurately reflect the true risk level, risk assessments are inaccurate, alerts are untimely, or false alarm rates are high. To address these issues, in some embodiments of the present invention, environmental parameters include noise, air quality, light intensity, temperature, and humidity; step S102 includes: The comprehensive risk index is obtained by combining each environmental parameter with its corresponding weight coefficient. The weight coefficient of each environmental parameter is dynamically configured according to at least one of the following influencing factors: the scenario type of the monitored object, the user feedback information received for historical reminder operations, and the historical activity pattern data of the monitored object.

[0034] Specifically, environmental parameters in different dimensions include noise, air quality, light intensity, temperature, and humidity. Environmental parameters can also include specific sound signatures (such as glass breaking, screams, and cries for help), ambient light color temperature / blue light intensity, ultraviolet radiation intensity, air pressure, altitude, electromagnetic field strength, and meteorological information such as rain and snow. Air quality can include PM2.5, formaldehyde concentration, carbon monoxide concentration, pollen / allergen concentration, and specific VOCs (such as benzene and toluene) concentration.

[0035] After obtaining the above environmental parameters, the comprehensive risk index R can be calculated using the following weighted formula (1): R = W1 × N' + W2 × A' + W3 × L' + W4 × T' + W5 × H' (1); where N', A', L', T', and H' are the normalized environmental parameters (noise, air quality, light intensity, temperature, and humidity), and W1 to W5 are the weighting coefficients. The data substituted into formula (1) for calculation is the normalized value of the five preprocessed core environmental parameters.

[0036] After obtaining the above environmental parameters, the high-risk parameters can also be given greater influence through nonlinear functions (such as square, root, exponent), that is, the comprehensive risk index R is calculated by the following weighted square root formula (2): R = √(W1 × N') 2 +W2×A' 2 +W3×L� 2 +W4×T' 2 +W5×H' 2 (2); where N', A', L', T', and H' are the normalized environmental parameters (noise, air quality, light, temperature, and humidity), and W1 to W5 are weighting coefficients. The data substituted into formula (2) for calculation are the normalized values ​​of the five preprocessed core environmental parameters. Formula (2) can more significantly amplify the impact of a single severely exceeding parameter on the overall index. Even if other parameters are normal, as long as one parameter (such as noise) is abnormally high, the R value will increase significantly, which is more in line with the intuition of the weakest link effect in safety protection.

[0037] Considering the synergistic or offsetting effects between different environmental parameters, after obtaining the above environmental parameters, the comprehensive risk index R can also be calculated by the following formula (3): R = W1×N' + W2×A' + W3×L' + W4×T' + W5×H' + W12×N'×A' (3); where N', A', L', T', and H' are the normalized environmental parameters (noise, air quality, light, temperature, and humidity), W1 to W5 and W12 are weighting coefficients, and W12×N×A is the interaction term. For example, it can be set that when high noise and high air pollution occur simultaneously (imagine a busy traffic intersection), the comprehensive health hazard is greater than the simple sum of the individual hazards of the two. The comprehensive risk index R calculated by formula (3) can more scientifically simulate the combined effect of multiple risk factors in a complex environment.

[0038] After obtaining the above environmental parameters, calculate the deviation of each environmental parameter Zi=(Xi-μi) / σi. The comprehensive risk index R can be calculated using the following formula (4): R=W1×ZN+W2×ZA+W3×ZL+W4×ZT+W5×ZH (4); where Zi represents the deviation of the current environmental parameter i from the historical normal level, Xi is the current original value of environmental parameter i, and the value of Xi is N, A, L, T, H, where N, A, L, T, H are environmental parameters (noise, air quality, light, temperature, and humidity), μi is the historical mean of environmental parameter i over a period of time, σi is the historical standard deviation of environmental parameter i, and W1 to W5 are weighting coefficients. Formula (4) can sensitively detect anomalies relative to an individual's historical benchmark.

[0039] In any of the above calculation methods, the weight coefficients corresponding to the environmental parameters are dynamically configured based on at least one of the following influencing factors: the scene type of the monitored object, user feedback information received regarding historical reminder operations, and historical activity pattern data of the monitored object. That is, the weight coefficients are related to at least one of the scene type, user feedback information, and historical activity pattern data. In other words, the weight vector [W1, W2, W3, W4, W5] can be dynamically adjusted in real time based on one or more of the following influencing factors. One influencing factor is the scene type, which can be determined and identified based on various methods, such as GPS positioning (indoor / outdoor), activity status identified by accelerometers (stationary / moving), or a combination of these data. Then, the weight coefficients are adjusted according to the scene type and a preset weight template is invoked. For example, since indoor lighting and air quality have the greatest impact on comfort and health, if the current scene type is indoor, the weight coefficients of W3 (lighting) and W2 (air quality) are increased. For example: (W1, W2, W3, W4, W5) = (0.15, 0.30, 0.30, 0.15, 0.10). Since air pollution (such as smog) and ultraviolet radiation (strong light) are the main risks outdoors, if the current scene type is outdoor, increase the weight of W2 (air quality) and maintain a relatively high weight for W3 (light). For example: (0.15, 0.35, 0.25, 0.15, 0.10). Since children are more sensitive to noise (such as on a playground) and temperature (too hot or too cold) when exercising, if the current scene type is sports scene, increase the weights of W1 (noise) and W4 (temperature). For example: (0.30, 0.20, 0.10, 0.30, 0.10).

[0040] Another influencing factor is user feedback. The system records the guardian's confirmation / false alarm feedback for each alert. If an alert triggered by a certain environmental parameter (such as formaldehyde in a home area) is confirmed multiple times in a short period, the system slightly increases its weight coefficient for that area to make its assessment more sensitive. Conversely, if it is repeatedly marked as a false alarm, its weight coefficient is cautiously decreased, provided that national safety standards are met, to achieve personalized de-interference. Another influencing factor is historical activity pattern data. The system analyzes the long-term location-time logs of the monitored subjects (such as children) and identifies frequently visited activity sites such as schools, parks, and homes through clustering algorithms. A personalized weight coefficient file is established for each frequently visited activity site. For example, if it is identified that a child is active in a park every afternoon, the system automatically uses the outdoor scene weight coefficient template for that time and location, and may additionally increase the weight coefficient of pollen concentration (a sub-item of air quality A) based on historical data. In summary, the comprehensive risk index R is calculated in real time based on the final weight coefficient determined in real time and the above calculation formula.

[0041] In this embodiment, by fusing environmental parameters with different physical meanings to calculate a comprehensive risk index R, the subsequent risk level determination logic is greatly simplified. This allows the system to make decisions from a global, quantitative perspective, avoiding potential conflicts and complexities when handling multiple parameter rules. Furthermore, the weighting coefficients can be dynamically adjusted based on scenario type, user feedback, and historical activity patterns. Through scenario recognition, the assessment model can automatically focus on the most significant risk source (e.g., indoors focusing on air and light, outdoors focusing on air and ultraviolet radiation), making the assessment results more consistent with objective risk conditions. Setting the weighting coefficients to be related to scenario type ensures that the system can set different values ​​for the weighting coefficients under different scenario types, significantly improving the accuracy and scenario relevance of risk assessment. For example, during outdoor activities on smoggy days, the system can automatically focus on air pollution risk without being interfered with by the more important light factor in indoor scenarios. This scenario adaptability makes the risk assessment results more reasonable, accurate, and more consistent with the risk distribution in the real world.

[0042] By further incorporating user feedback to adjust weighting coefficients, limited computing power and attention are directed towards the most likely and critical risk dimensions. This not only improves computational efficiency but also directly optimizes the user experience: it reduces frequent and ineffective reminders caused by minor fluctuations in non-critical risk parameters, ensuring that each reminder is more targeted and valuable, thereby enhancing users' trust and reliance on the product.

[0043] Furthermore, by incorporating historical patterns to adjust weighting coefficients, the system learns from feedback and data mining to adapt to these historical patterns and evolves itself. It automatically adjusts weighting coefficients to reflect changes in the monitored individuals (e.g., a child transitioning from kindergarten to primary school) and seasonal changes (e.g., the spring pollen season), eliminating the need for manual resets or frequent calibrations. This enhances the adaptive capabilities and long-term applicability of risk assessment, enabling the system to learn and adapt to the monitored individuals' living environment and sensitivities, achieving accurate assessments and fundamentally reducing misjudgments.

[0044] In summary, by adjusting the weighting coefficients, the system's sensitivity to different risks can be easily customized and optimized, enhancing its flexibility and interpretability. For example, to create a watch that specifically focuses on hearing protection, simply increase the noise weight W1 accordingly. The calculation process of the comprehensive risk index R is transparent and traceable, making the risk assessment results easy to understand and trust, facilitating the identification of major risk sources and guiding targeted improvement measures. If an alarm is issued, the system can analyze which environmental parameter(s) contributed the main risk due to their high weight and large value, thus providing users with a clearer explanation and enhancing the system's credibility.

[0045] Furthermore, the comprehensive risk index obtained through combined calculations exhibits excellent scalability. As shown in the air quality indicator (A), it can accommodate multi-level data fusion. First, data from multiple sensors of the same type (PM2.5, formaldehyde, VOCs) are fused into a secondary index (A). Then, this index is fused as a whole with other environmental parameters (such as noise N) at a higher level. This allows the system to integrate as much environmental information as possible, considering five core environmental factors: noise, air quality, light, temperature, and humidity. This avoids the limitations of single-parameter assessments and covers multiple health risks, including auditory health, respiratory health, visual comfort, and body temperature regulation. It provides comprehensive environmental health protection for monitored individuals, helping caregivers fully understand the overall quality of their environment and supporting targeted protective measures for different health dimensions. Furthermore, the combined calculation framework supports the addition of new environmental parameters and weighting dimensions. It operates stably with the limited computing resources of smart wearable devices, facilitating updates to assessment standards based on new health research findings. For example, it reserves interfaces for future additions of new monitoring indicators such as ultraviolet radiation and other environmental parameters. Timely warnings when environmental quality begins to decline allow for preventative measures, representing a technological leap from simple environmental monitoring to comprehensive health risk assessment, and significantly enhancing safety.

[0046] In continuous environmental monitoring, it is difficult to balance power consumption and data integrity due to fixed-frequency sampling, and the accuracy of subsequent risk assessment is affected by noise and interference in the raw sensor data. To address these issues, in some embodiments of the present invention, such as... Figure 2As shown, obtaining environmental parameters of various dimensions of the monitored object's environment includes the following steps: S201. Environmental state data is obtained by sampling. The sampling frequency of the environmental state data is related to the fluctuation state of the environmental state data.

[0047] Specifically, raw environmental data is collected synchronously or periodically through a multi-source sensor array integrated on the smartwatch. Environmental parameters include noise, air quality, light intensity, temperature, and humidity, as described in the embodiments below. The multi-source sensor array includes a MEMS microphone array, a multi-pollutant air quality sensor, a photodiode array, thermocouples, and capacitive sensors. The MEMS microphone array collects ambient sound pressure levels to obtain noise. The multi-pollutant air quality sensor can be used to measure air quality and may include an integrated laser scattering module to measure PM2.5 concentration, an electrochemical module to measure formaldehyde concentration, and a metal oxide semiconductor module to measure VOCs concentration. The photodiode array is used to collect ambient light intensity. Thermocouples and capacitive sensors collect ambient temperature and relative humidity, respectively.

[0048] The system does not collect environmental parameters at a fixed frequency, but intelligently adjusts the sampling rate based on environmental stability. During system initialization or in a stable environment, a base sampling frequency (e.g., sampling once every 10 seconds) is used. Collecting environmental parameters based on this base sampling frequency ensures basic data updates while minimizing power consumption. Furthermore, the system calculates in real-time the fluctuations of one or more key environmental parameters (such as noise levels in decibels or PM2.5 concentration) within a recent time window (e.g., the past 30 seconds). If the fluctuations indicate a highly volatile environment, the system immediately and automatically increases the sampling frequency from the base frequency (e.g., 10 seconds / sample) to a higher frequency (e.g., 2 seconds / sample). Of course, if the environmental parameters stabilize again and remain so for a period of time, the system automatically reverts to the base sampling frequency to conserve power.

[0049] S202. Preprocess the environmental status data to obtain environmental parameters.

[0050] Specifically, preprocessing includes at least one of filtering, outlier handling, and normalization. The purpose of filtering is to remove transient, meaningless random noise and instantaneous spikes from the data. The moving average filtering algorithm is primarily used. That is, the current sampled value is averaged with several previous historical values, and this average is used as the valid value at that moment. For example, for a sudden, instantaneous impact on a noise sensor, this moving average filtering algorithm can effectively smooth out the spike, making it closer to the true environmental noise level. Moving average filtering effectively removes instantaneous interference, such as errors from instantaneous impacts on noise sensors or brief shadows on light sensors. Alternatively, Kalman filtering can be used, for example, to dynamically correct random errors in air quality sensors, improving the accuracy of long-term monitoring data. This significantly improves data reliability and suppresses noise data.

[0051] The purpose of outlier handling is to identify and remove abnormal data points that deviate significantly from the normal range due to factors such as occasional sensor errors or strong external interference. It primarily employs the statistically based 3σ (three Sigma) criterion for outlier handling. That is, the system maintains a short-term historical data window and calculates its mean (μ) and standard deviation (σ). For a new sampled value x, if |x - μ| > 3σ, the value is considered an outlier and will be removed. Removed or replaced outlier data points will not be included in subsequent calculations but will be marked and stored in the log for system diagnosis and adaptive learning. This outlier detection and appropriate replacement mechanism ensures the integrity, continuity, and availability of the data sequence.

[0052] The purpose of normalization is to map data collected from different sensors, which have different physical meanings and dimensions, to the same numerical range, laying the foundation for subsequent data fusion and comprehensive comparison. Linear normalization is mainly used. That is, a "minimum-maximum" measurement range with actual physical meaning is preset for each environmental parameter, and mapping is performed using the following formula: Normalized value = (Original measured value - Lower limit of range) / (Upper limit of range - Lower limit of range). All environmental parameters are mapped to the standardized interval [0, 1], eliminating dimensional differences and providing a fair basis for weighted fusion. Risk sensitivity: The normalization method based on the safety range provides higher resolution at the risk boundary, enhancing the sensitivity of risk assessment. The normalized data facilitates horizontal comparison and trend analysis, supporting more accurate risk judgment.

[0053] For example, the noise range is 30dB to 130dB. When the noise level is measured at 80dB, the normalized value is (80-30) / (130-30) = 0.5. The PM2.5 range is 0μg / m³ to 500μg / m³. When the PM2.5 level is measured at 250μg / m³, the normalized value is (250-0) / (500-0) = 0.5. The illumination range is 300lux (weak light reference) to 50000lux (strong light reference). When the illumination level is measured at 1000lux, the normalized value is (1000-300) / (50000-300) ≈ 0.014.

[0054] In this embodiment, low-frequency sampling is used when the environment is stable, significantly reducing the operating load on the sensor and processor, and extending the battery life of the children's smartwatch. Automatic switching to high-frequency sampling during sudden environmental changes enables more accurate capture of instantaneous risk changes, avoiding the omission of critical danger signals (such as sudden loud noises or spikes in pollutant concentrations) due to insufficient sampling, thus improving the system's real-time performance and reliability. Furthermore, filtering and outlier handling (such as removal or replacement) effectively remove noise and glitches from the data, providing stable and reliable input for risk assessment, significantly improving data quality, and fundamentally reducing the probability of false alarms and missed alarms. Furthermore, normalization transforms physical quantities with different units such as dB, μg / m³, and lux into dimensionless values ​​between 0 and 1. This makes different types of environmental parameters comparable, enabling multi-source information fusion. For example, 0.5 for noise and 0.5 for PM2.5 are comparable, thus making it possible to perform weighted calculation of the comprehensive risk index (R = W1N + W2A + ...) in the embodiments below. This is a key step in achieving multi-dimensional collaborative risk assessment. Furthermore, multiple anomaly detection mechanisms (including filtering, outlier handling, and normalization) prevent a single erroneous reading from affecting the overall risk assessment adaptive learning. This optimizes system performance while ensuring data quality, laying a solid data foundation for accurate environmental risk assessment.

[0055] Because static threshold assessments cannot identify environmental degradation trends, risk warnings are delayed. To address this issue, in some embodiments of the present invention, Figure 3 A flowchart illustrating another embodiment of the environment-aware safety alert method provided by the present invention is shown below. Figure 3 As shown, after obtaining various environmental parameters of the environment in which the monitored object is located, the process further includes: S301. Based on the sliding time window algorithm, calculate the rate of change of environmental parameters after filtering and outlier processing within a preset time interval.

[0056] Specifically, the sliding time window algorithm processes high-quality environmental parameter sequences (such as PM2.5 concentration values) that have already been filtered for outliers and filtered out of the background. The system sets a fixed, relatively short sliding time window (i.e., a preset time interval, such as 30 seconds). This sliding time window always includes all valid data points (high-quality data points that have been filtered for outliers and filtered out of the background) within the most recent 30 seconds. On this dataset within the sliding time window, a linear regression analysis is performed with time as the independent variable and the environmental parameter value as the dependent variable. The slope of the resulting line is used as the rate of change. The linear regression fits a line y=mx+c that best represents the trend of the data points, where y is the environmental parameter (e.g., PM2.5 concentration), x is time, and m is the slope of the line. This slope m is the rate of change, representing how many units per second the environmental parameter has increased or decreased on average within the sliding time window. To facilitate comparison between different environmental parameters, this rate of change is often converted to a percentage rate of change (i.e., the current rate of change relative to the current parameter value). The rate of change is a precise and calculable indicator that provides an objective and reliable basis for subsequent rule-based judgments, avoiding subjective assumptions. Alternatively, the variance of any environmental parameter can be calculated as the rate of change within this sliding time window dataset.

[0057] S302. When environmental parameters are outside the preset safety range and the rate of change exceeds the preset rate threshold, the current risk level determined by the environmental parameters will be updated to the next risk level; wherein the risk represented by the next risk level is higher than the current risk level.

[0058] Specifically, the dual triggering conditions for updating the risk level from the current risk level to the next risk level include condition one (state condition) and condition two (trend condition). Condition one (state condition) means that the value of the environmental parameter is outside the preset safe range, which means that the environment is already in a "mild risk" or "high risk" state. Condition two (trend condition) means that the calculated rate of change exceeds a preset rate threshold. For example, the variance of any environmental parameter continuously exceeds a preset variance threshold, or the absolute value of its slope is greater than a certain slope limit (such as 5% / second), indicating that the environmental parameter is increasing in a more dangerous direction. When both conditions are met simultaneously, the system determines that there is an escalating risk trend. At this time, the system will not maintain the original risk level, but will initiate an upgrade procedure to update the current risk level determined based on the environmental parameters to the next risk level. If the currently determined safe level is the current risk level, it will be updated to the mild risk level as the next risk level. If the currently determined mild risk level is the current risk level, it will be updated to the high risk level as the next risk level. The logical judgment process can be summarized as: IF (parameter value ∉ safety range) AND (rate of change > preset threshold) THEN Risk level = current risk level + n levels, where n is a positive integer greater than or equal to 1.

[0059] In this embodiment, when environmental parameters are outside the preset safety range, but the system detects a rate of change exceeding a preset threshold, indicating an escalating risk, it triggers a high-risk alarm in advance. This provides valuable reaction time for the monitored individuals and their guardians, enabling early and escalation warnings. For example, when PM2.5 concentration starts to rise rapidly from 80 μg / m³ (mild risk) at a rate of 7% per second, even though the concentration has not yet reached the high-risk threshold of 150 μg / m³, the rate of change exceeds the preset threshold, updating the mild risk level to a high-risk level. This ensures that an escalation alarm is triggered only when the environment is not only poor but also rapidly deteriorating, effectively preventing frequent false alarms caused by normal fluctuations in environmental parameters near the risk threshold, and enhancing the seriousness and credibility of the alarm. Furthermore, the risk escalation mechanism based on sliding time windows and rate of change analysis injects forward-looking and intelligent core capabilities into the entire safety alert system. By monitoring the speed of environmental deterioration, it achieves a strategic shift from "post-event response" to "in-event intervention" and even "early warning," greatly improving the safety level of children facing dynamically changing environments.

[0060] For example, the sampling frequency can be dynamically adjusted according to different scenarios. For instance, in a normal environment, sampling occurs once every 10 seconds; in complex environments (such as when drastic fluctuations are detected), sampling automatically increases to once every 2 seconds. When the environment is stable, low-frequency sampling is used to obtain a representative average value while saving power. When the environment changes abruptly, it automatically switches to high-frequency sampling to ensure that key fluctuation information is captured, providing a data foundation for calculating accurate average values ​​and conducting trend analysis. For example, the current PM2.5 concentration is 100 μg / m³ (according to the threshold table, this falls within the low-risk range). A low-risk alert should be triggered. The system calculates using a sliding window and finds that the PM2.5 concentration has risen sharply from 30 μg / m³ to 100 μg / m³ in the past minute, with a change rate far exceeding 5% / s, marking it as an escalating risk trend. The system upgrades the current risk level from low-risk to high-risk. Levels of alerts. The children's watch will immediately trigger strong alerts such as flashing red, continuous vibration, and voice announcements, and parents will also receive emergency notifications. This allows the system to issue the highest level of warning before pollution concentrations become extremely dangerous, enabling proactive protection. Even if the current risk level remains unchanged, the indicator of an escalating risk trend (indicating that the rate of change of environmental parameters exceeds a preset threshold) will change the form of the alert, making it more urgent. When the risk level is low but there is an escalating risk trend, the system may shorten the alert interval, changing the original single short vibration to periodic repetitive vibrations. An "upward arrow" animation will be added to the yellow icon, and the text prompt will be changed to "The environment is rapidly deteriorating, please be careful!" Without causing excessive panic (or escalating to high risk), richer contextual information is conveyed to children and parents, prompting them to be more vigilant. This significantly reduces the system's response delay, enabling it to react promptly to rapidly deteriorating environments to improve safety.

[0061] To address the problem that single-dimensional or static rule-based risk assessments are prone to misjudgment and cannot achieve multi-condition collaborative judgment, in some embodiments of the present invention, such as... Figure 4 As shown, the risk level of the current environment is determined by comparing the values ​​of the comprehensive risk index and at least one of the environmental parameters with a preset risk threshold, including: S401. Compare the comprehensive risk index R with the preset risk threshold; the preset risk threshold range includes the upper limit value R_high and the lower limit value R_low of the risk index.

[0062] Specifically, the system presets two levels of judgment criteria as the basis for decision-making. The first level of judgment criteria compares the comprehensive risk index R with preset risk thresholds to determine whether the comprehensive risk index R falls within the range of the preset risk thresholds. The preset risk thresholds include a lower limit R_low and an upper limit R_high. The lower limit R_low is the threshold for triggering a mild risk in the comprehensive risk index R, for example, 0.3. The upper limit R_high is the threshold for triggering a high risk in the comprehensive risk index R, for example, 0.6.

[0063] S402. Compare the environmental parameters with their respective risk ranges; the risk ranges include the safe range, the second risk range, and the first risk range; the first risk range and the second risk range do not overlap.

[0064] Specifically, the second level of judgment criteria compares a certain environmental parameter (noise, PM2.5, formaldehyde, etc.) with its corresponding safe range, first-risk range, and second-risk range to determine whether the environmental parameter falls within its corresponding risk range. Each environmental parameter has three independently defined non-overlapping numerical ranges (including the safe range, first-risk range, and second-risk range). An environmental parameter is considered safe when it falls within the safe range. An environmental parameter is considered to have a severe risk when it falls within the first-risk range (high risk). An environmental parameter is considered to have a potential risk when it falls within the second-risk range (low risk). The second-risk range does not overlap with either the safe range or the first-risk range, and its boundaries are clearly defined.

[0065] S403. When an environmental parameter is within its corresponding first risk range, or the comprehensive risk index R is greater than or equal to the upper limit value of the risk index R_high, or the environmental parameter is within its corresponding second risk range and the duration exceeds the preset duration, the risk level of the current environment is determined to be the first level.

[0066] Specifically, the system checks whether environmental parameters meet the severe risk triggering conditions. Once any one of the severe risk triggering conditions is met, the current environmental risk level is immediately determined to be Level 1. Severe risk triggering conditions include: Condition 1: A single environmental parameter severely exceeds the standard, meaning any environmental parameter is within its corresponding first risk range (high risk); Condition 2: Extremely high overall risk, meaning the overall risk index R ≥ the upper limit of the risk index R_high; Condition 3: Mild risk remains unresolved for a long time, meaning any environmental parameter is within its corresponding second risk range (mild risk) and the duration exceeds a preset duration (e.g., 5 minutes). Meeting any one of the severe risk triggering conditions (i.e., meeting any one of conditions 1 to 3) determines the current environmental risk level to be Level 1, for example, PM2.5 concentration at 100 μg / m³ for 6 minutes.

[0067] S404. When environmental parameters are within their corresponding second risk range and the duration does not exceed the preset duration, or when the comprehensive risk index R is greater than or equal to the lower limit of the risk index R_low and less than or equal to the upper limit of the risk index R_high, the risk level of the current environment is determined to be the second level; the risk represented by the second level is lower than that of the first level.

[0068] Specifically, the environmental parameters are checked to see if they meet the mild risk triggering conditions. If any one of the mild risk triggering conditions is met, the current environment's risk level is immediately determined to be Level 2. Mild risk triggering conditions include: Condition 4: A single parameter slightly exceeds the limit but briefly, meaning any environmental parameter is within its corresponding second risk range (mild risk), but the duration does not exceed the preset duration (5 minutes); Condition 5: The overall risk is at the alert level, meaning the lower limit of the risk index R_low ≤ the overall risk index R < the upper limit of the risk index R_high. Meeting any one of the mild risk triggering conditions (i.e., meeting either condition 4 or 5) determines the current environment's risk level to be Level 1.

[0069] S405. When all environmental parameters are within the safe range and the comprehensive risk index R is less than the lower limit of the risk index R_low, the risk level of the current environment is determined to be a safe level.

[0070] Specifically, the environmental parameters are checked to see if they meet the safety trigger conditions. Only when the safety trigger conditions are met can the current environment be determined to be at a safe risk level. The safety trigger conditions include condition 6: all environmental parameters are within their respective safe ranges; and condition 7: the comprehensive risk index R < the lower limit of the risk index R_low. When all safety trigger conditions are met (i.e., conditions 6 and 7 are met simultaneously), the current environment can be determined to be at a safe risk level.

[0071] For example, a multi-dimensional risk assessment mapping table based on environmental parameters is shown in Table 1:

[0072] As shown in Table 1, the risk assessment logic is as follows: if all environmental parameters are within the safe range and the comprehensive risk index R < 0.3, the current environment's risk level is determined to be safe. If any environmental parameter is within the slightly risky range, or the comprehensive risk index 0.3 ≤ R < 0.6, the current environment's risk level is determined to be slightly risky. If any environmental parameter is within the highly risky range, or the comprehensive risk index R ≥ 0.6, or a single slightly risky condition persists for more than 5 minutes, the current environment's risk level is determined to be highly risky.

[0073] In this embodiment, a comprehensive and thorough risk assessment mechanism is constructed through precise combinations of conditions. This ensures that the system can effectively identify risks regardless of their form, greatly reducing the possibility of missed alarms, significantly lowering the false alarm rate, and improving alarm credibility. Furthermore, using duration as a necessary condition for escalating a minor risk to a high risk is an efficient false alarm prevention mechanism. It effectively filters out transient, occasional environmental disturbances (such as brief noise or dust caused by a passing truck), preventing the system from overreacting. Only when the risk truly persists will a higher-level alarm be triggered, making each high-risk alarm more serious and instructive. Furthermore, single-parameter veto (such as directly determining high risk) ensures the system has the fastest response speed to any single aspect of severity risk, conforming to the core principles of security protection. Comprehensive risk indicator assessment can identify composite, synergistic risks caused by multiple parameters slightly deviating from ideal values, without any single parameter severely exceeding the limit. This achieves a complementary advantage between single-parameter veto and comprehensive risk indicator assessment, making risk assessment both sensitive and comprehensive. Furthermore, the risk level determination mechanism, by establishing decision rules with multiple thresholds, multiple paths, and time-sensitive judgments, successfully transforms a complex environmental risk assessment problem into a stable, reliable, and highly intelligent automated process. This ensures that every alarm is issued after rigorous logical deduction, thereby providing timely and reliable safety protection for the monitored objects to the greatest extent possible.

[0074] To address the issues of limited risk alert methods and the inability to differentiate between guardians and wards when providing risk warnings, some embodiments of this invention perform corresponding alert operations based on risk levels, including: When the risk level is Level 1, control the smart wearable device of the person under supervision to perform the first risk warning operation, and control the smart wearable device of the guardian to perform the emergency reminder operation.

[0075] Specifically, when the risk level is high risk (Level 1), the first risk warning and emergency alert actions are executed. The design goal of the first alert action is to attract the attention of the monitored individual and notify the guardian to intervene. The monitored individual's smart wearable device is controlled to execute a more escalated warning than the second risk warning action. The first risk warning action includes any one or more of visual, tactile, and auditory cues. Visual cues can display a red, flashing icon and full-screen text warning (e.g., "Danger! PM2.5 concentration is extremely high!") on the smart wearable device screen to attract the monitored individual's attention with the strongest visual impact. Tactile cues can control the smart wearable device to trigger a continuous vibration pattern (e.g., vibrate for 2 seconds, pause for 1 second, repeat 3 times) to create a strong tactile alarm. Auditory cues can simultaneously trigger a clear voice broadcast (e.g., "The current PM2.5 concentration is too high; it is recommended to wear a mask immediately and stay away"). To prevent the monitored individual from ignoring the alert due to habit, the frequency of voice broadcasts can adaptively increase according to the duration of the risk (for example, initially broadcasting once every 30 seconds, then shortening to once every 15 seconds if the risk persists for 2 minutes). Additionally, the system can control the guardian's smart wearable device to perform emergency alerts, sending a strong alert notification to the monitoring app linked to the guardian's smart wearable device (such as a mobile phone). This strong alert notification is accompanied by a sound and vibration and pops up at the top of the screen, ensuring the guardian is immediately aware of it even when using other applications. In addition to risk details, the strong alert notification simultaneously displays real-time changes in environmental parameters, a high-risk level indicator, and the monitored individual's real-time location map, providing more comprehensive decision-making information. The system can also control the guardian's or the monitored individual's smart wearable device to automatically dial preset emergency contacts. To improve call connection rates, a looping call logic can be set (e.g., if the first contact fails to answer after 3 attempts, the call is automatically transferred to the next contact), ensuring that the emergency reaches at least one emergency contact.

[0076] When the risk level is Level 2, the smart wearable device of the monitored person is controlled to perform a second risk warning operation, and the smart wearable device of the guardian is controlled to perform a silent reminder operation; the reminder intensity of the second risk warning operation is lower than that of the first risk warning operation.

[0077] Specifically, when the risk level is low (Level 2), a second risk warning and a silent reminder are executed. The design goal of these actions is to attract the attention of the monitored individual while notifying the guardian without causing significant disruption. The monitored individual's smart wearable device is controlled to execute the second risk warning, which includes visual and / or tactile cues. Visual cues may display a yellow icon (such as an exclamation mark or an icon corresponding to the environment type) above the screen of the smart wearable device or in a designated area, accompanied by concise text (e.g., "The environment is slightly noisy; please protect your hearing" or "Air quality is slightly polluted"). Tactile cues may trigger a short vibration (e.g., lasting 0.5 seconds) on the smart wearable device. This level of intensity is sufficient for the monitored individual to perceive the warning without causing panic or severely interfering with their ongoing activities (such as studying). Additionally, the guardian's smart device is controlled to execute a silent reminder, which is achieved by pushing a silent notification to the monitoring app linked to the guardian's smart wearable device (such as a mobile phone). This silent notification appears in the notification bar / message center of the smart wearable device, but it will not trigger a ringtone, strong vibration, or full-screen pop-up. It is only visible when the user is viewing the smart wearable device. The silent notification includes the risk type, specific environmental parameter measurements (such as "Noise: 82dB"), and the real-time location information of the monitored person. This allows the caregiver to be aware of the situation without having to immediately interrupt their work.

[0078] In this embodiment, through hierarchical reminders, the system effectively differentiates between scenarios that require attention and those that demand immediate action, avoiding alarm fatigue caused by excessive harassment of the monitored object and the guardian in the event of minor risks. This ensures that when a high-risk alarm arrives, users can pay sufficient attention and respond immediately, implementing a precise and user-friendly reminder strategy and greatly enhancing the user experience. Further, a dual protection closed-loop of proactive avoidance on the monitored object side and remote monitoring on the guardian side is constructed. The reminder on the monitored object side aims to enable the monitored object to have a risk awareness and guide them to take proactive avoidance actions (such as leaving a noisy environment and wearing a mask), cultivating their self-protection ability. The reminder on the guardian side provides the ultimate safety guarantee. In particular, the strong notification in the event of high risks ensures that the guardian can be informed immediately regardless of their location, enabling them to remotely guide the monitored object or go to the scene in person to form a three-dimensional and internally and externally linked safety protection network. Further, the distinction between silent reminders and strong reminders for the guardian reflects the system's respect for the guardian's work and life scenarios and intelligent judgment. Only information synchronization is performed in the event of minor risks, while forced intervention occurs in the event of high risks, perfectly balancing the contradictory requirements of maintaining contact and reducing unnecessary interference. Through different reminder methods, the balance between information transmission and interference control is achieved. Further, the reminder intensity of the second risk reminder operation is lower than that of the first risk reminder operation, ensuring the effective delivery of high-risk reminder warnings. For example, on the monitored object side, multiple sensory channels such as vision (red flashing) + touch (strong vibration) + hearing (voice) are stimulated simultaneously, which can maximally break through the attention barrier of the monitored object and ensure that the alarm is effectively received even in a playing or noisy environment, enhancing the protection efficiency in critical situations. Further, the hierarchical reminder operation plan is a human-computer interaction system that fully considers the psychological and scenario needs of different users. Through precise hierarchical strategies and multi-terminal linked execution methods, it ensures that safety information can be delivered to the people who need it most with the most appropriate intensity and at the most appropriate time, thus transforming the front-end perception and risk assessment results into practical and effective safety protection actions.

[0079] To solve the problems of frequent false alarms and poor user experience caused by the inability to learn user feedback. In some embodiments of the present invention, as Figure 5 shown, it further includes: S501. Obtain the feedback information of the guardian on the reminder operation, and the feedback information includes confirmed risks and false alarm ignores.

[0080] Specifically, after each risk alert is sent to the guardian's app, the system provides a simple interactive interface for the guardian to provide feedback. This typically includes two clear feedback messages: one confirming the risk, meaning the guardian acknowledges the alert and believes there is indeed a risk in the environment; and the other ignoring a false alarm, meaning the guardian believes the alert is a false alarm and there is no corresponding risk in the current environment (for example, the monitored individual briefly passed through a renovation area, triggering the alert, but did not stay there for an extended period).

[0081] S502. Based on the feedback information, count the number of consecutive false alarms for the same type of risk alert.

[0082] Specifically, the system maintains an independent false alarm counter for different types of risks (such as noise risk, air quality risk, and light risk). When a guardian ignores a false alarm for a certain type of risk (such as noise risk), the false alarm counter for that type of risk increments by 1. If the same type of risk warning appears again and is marked as a false alarm ignored, the counter will continue to increment. Once the guardian confirms the risk for that type of risk warning, or the system determines it as a clear and valid risk (such as the risk continuing until a higher-level alarm is triggered), the false alarm counter for that type is immediately reset to 0. This design ensures that only consecutive false alarms will trigger adjustments.

[0083] S503. When the number of consecutive false alarms reaches the preset number, the risk threshold of the corresponding environmental parameter is increased, and the adjusted risk threshold is not lower than the preset safety standard limit.

[0084] Specifically, when the number of consecutive false alarms for a certain type of risk (such as noise risk) reaches a preset number (e.g., 3 times), the system determines that the current environmental parameter's risk threshold is too sensitive to the monitored object and its usual environment. The system will automatically and moderately increase the risk threshold corresponding to that environmental parameter. The adjusted risk threshold = old threshold × (1 + adjustment coefficient). The adjustment coefficient is a preset, fixed percentage value, such as 0.05 (5%) or 0.1 (10%). The adjustment coefficient is used to quantitatively and slightly increase the threshold when the system deems the current risk threshold too sensitive; essentially, it is a hyperparameter controlling the learning rate. The adjustment coefficient determines the step size for each threshold adjustment. A larger adjustment coefficient results in a larger adjustment magnitude per instance; a smaller adjustment coefficient results in a more refined adjustment process. The adjustment coefficient is preset and configured by the system developers.

[0085] For example, taking the high-risk noise threshold (initially 85dB) as an example, the monitored individual passes through a brief construction area on their way home from school, where the noise briefly rises to 88dB. The system issues high-risk alerts at 85dB (based on trend judgment) and 86dB, both of which are marked as false alarms and ignored by the guardian. When the third alert at 84dB is ignored again, the false alarm counter reaches 3. The adjusted risk threshold = 85dB × (1 + 0.06) ≈ 90.1dB. After this, the noise must exceed 90.1dB for the system to classify it as high-risk. Before the adjustment, the system frequently generated emergency alarms that disturbed residents in the construction area, which was extremely bothersome to the guardian. The guardian believed that such brief exposure did not warrant triggering a high-risk alarm. The adjusted high-risk noise threshold is 85 × 1.06 = 90.1dB. In this way, the adjusted system has learned to tolerate brief, controllable high noise levels along the monitored individual's daily route, while still remaining vigilant against truly severe and persistent noise (such as 95dB).

[0086] Of course, the system has preset national or industry safety standard limits as the bottom line for adjustment. These preset safety standard limits are the national or industry safety standard limits themselves. Any automatic adjustment must follow one principle: the adjusted risk threshold must not be lower than the preset safety standard limit. That is, before applying the adjusted risk threshold, the system will check whether the adjusted risk threshold is less than or equal to the preset safety standard limit. If so, the adjusted risk threshold is used. If not, the preset safety standard limit is used as the adjusted risk threshold to ensure that the safety bottom line is not breached. Example: Suppose the national standard stipulates that the preset safety standard limit for noise exposure of a monitored object for a long period is 90dB. Then, no matter how many false alarms there are, the system will never lower the high-risk threshold below 90dB, ensuring that the system's adaptive optimization always prioritizes the absolute health and safety of the monitored object. Since 90.1dB > 90dB, meeting the requirement of "not lower than" the safety standard, the adjusted risk threshold is equal to 90.1dB.

[0087] In this embodiment, because each monitored individual has different living environments and risk tolerance (e.g., urban monitored individuals may have a higher tolerance for noise than rural monitored individuals), this mechanism enables the system to learn and adapt to the specific living environment of each monitored individual. This reduces the disturbance caused by frequent false alarms in their daily environment, making the product smarter and more considerate, thereby enhancing user stickiness. Furthermore, when the number of consecutive false alarms reaches a preset number, the risk threshold of the corresponding environmental parameter is increased, and the adjusted risk threshold is not lower than the preset safety standard limit. This continuous self-optimization and updating of the risk threshold maintains high accuracy in risk assessment over the long term, achieving personalized system adaptation and significantly improving user experience and acceptance. By dynamically adjusting the threshold, potential deviations in the initial model can be compensated for, and environmental changes during the monitored individual's growth (such as moving or changing schools) can be adapted, ensuring high accuracy in risk assessment throughout the entire usage lifecycle. While providing flexibility, the system adheres to safety bottom lines, avoiding overcorrection. It cleverly balances the convenience requirement of reducing false alarms with the core principle of absolute safety, preventing the risk of missed alarms due to over-optimization from the design source, and ensuring the reliability of the protection function. Furthermore, by incorporating guardian feedback, the system constructs a human-machine collaborative optimization loop. Every confirmation or omission by the guardian helps train the system, integrating the guardian into the decision-making loop and enhancing the system's credibility and interactivity. This not only improves the system's accuracy but also allows guardians to feel they have control and influence over the device, thus fostering stronger trust in the system. This feedback-based adaptive learning approach endows the monitored individual's safety protection system with continuous evolution capabilities. By collecting user feedback, analyzing false alarms, and adjusting thresholds under strict safety constraints, the system ensures that it becomes increasingly accurate and reliable over long-term use, ultimately achieving truly personalized safety protection. Moreover, recognizing that each monitored individual's activity environment and family risk tolerance differ, the algorithm enables the system to tailor safety standards, significantly improving the user experience and ensuring that critical alarms are taken seriously.

[0088] In some real-time methods, a lightweight random forest model is deployed on the smart wearable device terminal of the monitored individual to learn the individual's risk tolerance characteristics. Due to the limited computing resources of the smart wearable device terminal, the lightweight random forest model needs to be lightweighted. It is trained using historical data within a preset time period (e.g., 3 months), collecting environmental parameters (noise, air quality, light intensity, temperature, humidity, etc.) and corresponding guardian feedback (confirmed risk or false alarm ignored) for the monitored individual within this preset time period. Each data sample includes environmental parameters and a label (whether it was confirmed as a risk by the guardian). Obviously erroneous data is removed, and missing values ​​are filled (e.g., using the average of the preceding and following data). Guardian feedback is used to label the data. If the guardian confirms the risk, the label is 1 (risk exists); if the guardian ignores it (false alarm), the label is 0 (acceptable risk). Note that only data with guardian feedback is used for training. Key features are extracted from the environmental parameters, including basic features, time features, location features, and personalized features. Basic features include instantaneous values, average values ​​(sliding window, e.g., 30 seconds), fluctuation range (maximum value - minimum value), trend slope, etc. Temporal features may include time periods of day (e.g., morning, noon, evening) and days of the week. Location features include extracting frequently visited locations (e.g., school, home, park) as category features. However, considering privacy and computational resources, clustered location labels may be used. Personalized features consider the historical responses of the monitored individual, such as the average risk value at the same location and time period over the past week. Lightweighting of the random forest involves reducing the number of trees (e.g., 10-20 trees) and limiting the tree depth (e.g., a maximum depth of 5-8). This allows for splitting using fewer features (e.g., the square root of the number of features). Of course, lightweight versions of decision trees (single trees) or gradient boosting trees (LightGBM or XGBoost) could also be considered, but random forests are easier to train in parallel and less prone to overfitting, making them more suitable for end-user devices. Based on a preset training frequency (e.g., weekly), the lightweight random forest model is trained once on a smart wearable device using data samples within a preset duration (i.e., a rolling window, always maintaining data from the most recent 3 months). Due to limited resources, online learning or incremental learning methods can be used, but random forests themselves do not support incremental learning, therefore periodic retraining is required. When retraining, you can use the previously trained lightweight random forest model as a base, or retrain entirely. Since the environment is safe most of the time, false positives (negative samples) may far outnumber confirmed risks (positive samples). You can downsample negative samples or oversample positive samples. Adjust the class weights to make the lightweight random forest model focus more on positive samples. After training, compare the new model with the old model to ensure that the new model's performance on the validation set (data from the most recent week) is no worse than the old model, and then update the model parameters.The trained model parameters (tree structure and split points) are stored on a smart wearable device for real-time risk assessment.

[0089] In some real-time modes, the system learns the unique risk tolerance patterns of each monitored individual. Inference is performed on the smart wearable device, protecting privacy and responding quickly. Weekly updates continuously adapt to changes in user behavior. A lightweight design allows operation with limited computing resources. Combined with spatiotemporal context, it provides accurate risk assessments. This approach transforms smart wearable devices from simple security monitoring into personalized health companions that truly understand the unique needs of each monitored individual. Assuming the system has already learned the monitored individual's activity patterns through sensor data such as GPS and accelerometers, combined with time information, such as identifying frequently visited places like home, school, and parks, and regular routes to and from school, the system now needs to dynamically adjust the scenario weight coefficients (W1-W5) in the risk assessment based on these activity patterns. The scenario weight coefficients are optimized based on the monitored individual's current activity patterns (location, time, activity type) to make the risk assessment more relevant to the current scenario. First, the system needs to identify the monitored individual's current scenario (e.g., at home, at school, on their way to school, playing in the park, etc.). A set of preset weight coefficients (W1-W5) is then set for each predefined scenario. These preset weights are set based on the importance of typical risk factors in the scenario. After identifying the current scenario, the preset weights corresponding to that scenario are used. Simultaneously, the weights can be fine-tuned based on historical data regarding false alarms reported by guardians in that scenario (e.g., through the previously mentioned adaptive learning module). The current location of the monitored individual is determined using GPS positioning data, Wi-Fi signals, and base station positioning, combined with time information (such as school hours) and movement status (using accelerometers to determine walking, running, or stationary states) to identify the scenario. For example, on weekdays from 7:00 to 8:00 AM, the route from home to school is identified as "on the way to school." If within the school's latitude and longitude range and during school hours, it is identified as "at school." If within the park's latitude and longitude range and the individual is actively moving, it is identified as "playing in the park." If at home in the evening, it is identified as "at home." Weight coefficients are preset based on the risk importance of various environmental factors in different scenarios. For example: At home, the focus might be more on indoor air quality (formaldehyde, VOCs) and noise (whether it affects rest), with a higher weight for W2 (air quality), a medium weight for W1 (noise), and a lower weight for W3 (lighting), as indoor lighting is usually controllable. At school, the focus might be more on noise (affecting learning) and air quality (classroom ventilation), with higher weights for W1 (noise) and W2 (air quality). On the way to / from school, the focus might be more on noise (traffic noise) and air quality (road exhaust), with higher weights for W1 (noise) and W2 (air quality). Since it's outdoors, lighting (W3) might also need to be considered (e.g., strong light glare). Playing in a park, the focus might be more on lighting (whether it's too strong) and temperature (whether it's suitable), with higher weights for W3 (lighting) and W4 (temperature). In addition to using preset weights, the weights can be fine-tuned based on feedback from the guardian.For example, in a given scenario, if the system repeatedly issues alerts for a certain risk but is flagged as false alarms by the guardian, the weight of that risk can be reduced in that scenario. Specifically, for each scenario, the guardian's feedback on the alerts is recorded. If an alert for a certain environmental parameter (such as noise) is repeatedly flagged as a false alarm, the weight of noise in that scenario is appropriately reduced (e.g., reduced by 0.05, while other weights are adjusted proportionally to maintain a total of 1). This is based on real-world activity patterns, not simple location labels. The weight coefficient for each monitored individual is based on their unique behavioral patterns. The weight coefficient can be adjusted in real-time according to context such as season, weather, and time. The weight coefficient can be adjusted based on explicit scenario classification and feature analysis. By recalibrating the weight coefficients, it is ensured that the weight coefficients remain optimal at all times.

[0090] In some implementations, sensors may age over time, causing readings to deviate from their true values. To address this, a baseline can be established based on historical normal data. Then, it's monitored whether sensor readings consistently deviate from the baseline; if so, a calibration coefficient is generated for compensation. This involves collecting historical normal data and calculating a baseline (e.g., mean or median) for each sensor parameter. Sensor readings are periodically (e.g., daily) checked for consistent deviations from the baseline. If a consistent deviation is found (e.g., PM2.5 sensor readings are consistently 10% higher), a calibration coefficient is generated. This coefficient is then used to compensate for the sensor readings. Assuming the sensors function normally during the initial phase (e.g., the first three months), data from this period is collected as historical normal data. For each sensor, a baseline value can be calculated. Considering that environmental conditions may vary over different time periods, baselines can be established by grouping by hour or day to account for diurnal and seasonal variations. For example, for a PM2.5 sensor, data from the same time period each day for three months (e.g., 8 AM to 9 AM) can be collected, and the baseline value for that period (which could be the mean or median) can be calculated. This results in 24 baseline values ​​(one per hour). Sensor data is continuously collected during system operation. To detect deviations, the current sensor reading is compared to the baseline value for the corresponding period. A threshold (e.g., 10%) is set; if the deviation between the current reading and the baseline value consistently exceeds this threshold (e.g., a deviation exceeding 10% for 30 consecutive days), the sensor is considered to have drifted. Assuming the baseline value is B and the current reading is X, the calibration factor can be calculated as: Calibration Factor = B / X. Multiplying the current reading by the calibration factor corrects it to the baseline level. It is assumed that the historical baseline is accurate and that environmental conditions have not fundamentally changed. Therefore, if a permanent change in the environment occurs (e.g., a factory relocation leading to a fundamental improvement in air quality), the baseline should be updated. However, this application assumes sensor aging, so the environment itself has not changed. Continuous deviations are used to identify and avoid miscalibration caused by temporary environmental changes. In the data processing module, the sensor reading is multiplied by the calibration factor before being used for subsequent risk assessment. In summary, a baseline is established based on historical normal data, and a calibration factor is automatically generated to compensate when the sensor output value continuously deviates from the baseline, reducing the impact of hardware aging.

[0091] To better implement the environment-aware safety alert method in this invention embodiment, based on the environment-aware safety alert method, correspondingly, as follows: Figure 6 As shown, the present invention also provides a safety alert system 600 based on environmental perception, comprising: The data acquisition and processing module is used to acquire environmental parameters of various dimensions of the environment in which the monitored object is located; The fusion calculation module is used to perform fusion calculations based on preset weighting coefficients and various environmental parameters to obtain a comprehensive risk index that characterizes the overall environmental risk; the weighting coefficients are used to reflect the degree of influence of environmental parameters on the overall environmental risk; The risk assessment module is used to determine the current environmental risk level based on the comparison results of the comprehensive risk index and at least one environmental parameter with the preset risk threshold. The control module is used to execute corresponding alert actions based on the risk level.

[0092] The environment-aware safety alert system 600 provided in the above embodiments can realize the technical solutions described in the above environment-aware safety alert method embodiments. The specific implementation principles of each module or unit can be found in the corresponding content of the above environment-aware safety alert method embodiments, which will not be repeated here.

[0093] like Figure 7 As shown, the present invention also provides an electronic device. This electronic device 700 includes a processor 701, a memory 702, and a display 703. The above only illustrates some components of the electronic device 700; however, it should be understood that it is not required to implement all the illustrated components, and more or fewer components may be implemented instead.

[0094] In some embodiments, processor 701 may be a central processing unit (CPU), microprocessor, or other data preprocessing chip, used to run program code stored in memory 702 or process data, such as the environment-aware safety alert method of the present invention.

[0095] In some embodiments, processor 701 may be a single server or a group of servers. The server group may be centralized or distributed. In some embodiments, processor 701 may be local or remote. In some embodiments, processor 701 may be implemented on a cloud platform. In one embodiment, the cloud platform may include a private cloud, public cloud, hybrid cloud, community cloud, distributed cloud, intranet, multi-cloud, etc., or any combination thereof.

[0096] In some embodiments, memory 702 may be an internal storage unit of an electronic device, such as a hard disk or memory of the electronic device. In other embodiments, memory 702 may also be an external storage device of electronic device 700, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., provided on electronic device 700.

[0097] Furthermore, the memory 702 may include both internal storage units of the electronic device 700 and external storage devices. The memory 702 is used to store application software and various types of data installed on the electronic device 700.

[0098] In some embodiments, display 703 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen. Display 703 is used to display information from electronic device 700 and to display a visual user interface. Components of electronic device 700 communicate with each other via a system bus.

[0099] In one embodiment, when the processor 701 executes the data parsing rule generation program in the memory 702, the following steps can be implemented: Obtain environmental parameters from multiple dimensions of the environment in which the monitored individual is located; Based on preset weighting coefficients, a comprehensive risk index representing the overall environmental risk is obtained by integrating various environmental parameters; the weighting coefficients are used to reflect the degree of influence of environmental parameters on the overall environmental risk. The risk level of the current environment is determined by comparing the values ​​of comprehensive risk indicators and at least one environmental parameter with the preset risk threshold. Execute the corresponding alert action based on the risk level.

[0100] It should be understood that when the processor 701 executes the data parsing rule generation program in the memory 702, in addition to the functions mentioned above, it can also perform other functions, as detailed in the description of the corresponding method embodiments above.

[0101] Furthermore, the embodiments of the present invention do not specifically limit the type of the electronic device 700 mentioned. The electronic device 700 can be a mobile phone, tablet computer, personal digital assistant (PDA), wearable device, laptop computer, or other portable electronic device. Exemplary embodiments of portable electronic devices include, but are not limited to, portable electronic devices running iOS, Android, Microsoft, or other operating systems. The aforementioned portable electronic device can also be other portable electronic devices, such as a laptop computer with a touch-sensitive surface (e.g., a touch panel). It should also be understood that in some other embodiments of the present invention, the electronic device may not be a portable electronic device, but a desktop computer with a touch-sensitive surface (e.g., a touch panel).

[0102] Accordingly, this application also provides a computer-readable storage medium for storing computer-readable programs or instructions. When the programs or instructions are executed by the processor 701, they can implement the steps or functions of the environment-aware security alert method provided in the above-described method embodiments.

[0103] Those skilled in the art will understand that all or part of the processes of the methods described in the above embodiments can be implemented by a computer program instructing related hardware (such as processor 701, controller, etc.), and the computer program can be stored in a computer-readable storage medium. The computer-readable storage medium may be a disk, optical disk, read-only memory, or random access memory, etc.

[0104] The present invention provides a detailed description of a safety alert method, system, device, and medium based on environmental perception. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, those skilled in the art will recognize that there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A safety reminding method based on environment perception, characterized in that, The method comprises the steps of: obtaining a plurality of different dimension environment parameters of an environment in which a guardian object is located; based on a preset weight coefficient, performing fusion calculation on each type of the environment parameters to obtain a comprehensive risk index representing the overall risk of the environment; the weight coefficient is used to reflect the influence degree of the environment parameters on the overall environmental risk; determining the risk level of the current environment according to the comparison result of the comprehensive risk index and the numerical value of at least one of the environment parameters with a preset risk threshold; performing a corresponding reminding operation according to the risk level. 2.The safety reminding method based on environment perception according to claim 1, characterized in that, The environment parameters include noise, air quality, illumination, temperature and humidity; the fusion calculation of each type of the environment parameters based on the preset weight coefficient to obtain the comprehensive risk index representing the overall risk of the environment comprises: combining each of the environment parameters with its corresponding weight coefficient to obtain the comprehensive risk index; wherein the weight coefficient corresponding to each of the environment parameters is dynamically configured according to at least one of the following influence factors; the influence factors include the scene type in which the guardian object is located, the user feedback information received for historical reminding operations, and the historical activity regularity data of the guardian object. 3.The safety alert method based on environmental perception according to claim 1, characterized in that, The obtaining of the plurality of environment parameters of the environment in which the guardian object is located comprises: sampling to obtain environment state data, the sampling frequency of the environment state data being related to the fluctuation state of the environment state data; preprocessing the environment state data to obtain the environment parameters. 4.The safety alert method based on environmental perception according to claim 1, characterized in that, After obtaining the plurality of environment parameters of the environment in which the guardian object is located, the method further comprises: based on a sliding time window algorithm, calculating the change rate of the environment parameters after filtering and abnormal value processing within a preset time interval; when the environment parameter is outside the preset safe range and the change rate exceeds a preset rate threshold, updating the current risk level determined according to the environment parameter to a next risk level; wherein the risk level represented by the next risk level is higher than that of the current risk level. 5.The safety alert method based on environmental perception according to claim 1, wherein, The determination of the risk level of the current environment according to the comparison result of the comprehensive risk index and the numerical value of at least one of the environment parameters with a preset risk threshold comprises: comparing the comprehensive risk index with a preset risk threshold range; the preset risk threshold range includes an upper limit value of the risk index and a lower limit value of the risk index; comparing the environment parameters with their respective corresponding risk ranges; the risk range includes a safe range, a second risk range and a first risk range; the first risk range and the second risk range have no overlap; when the environment parameter is within its corresponding first risk range, or the comprehensive risk index is greater than or equal to the upper limit value of the risk index, or the environment parameter is within its corresponding second risk range and the duration exceeds a preset duration, determining that the risk level of the current environment is a first level; determining that the risk level of the current environment is a second level when the environmental parameter is within its corresponding second risk range and the duration does not exceed a preset duration, or the comprehensive risk indicator is greater than or equal to the lower limit of the risk indicator and less than or equal to the upper limit of the risk indicator; the risk represented by the second level is lower than the first level; determining that the risk level of the current environment is a safe level when all the environmental parameters are within the safe range and the comprehensive risk indicator is less than the lower limit of the risk indicator. 6.The safety alert method based on environmental perception according to claim 1, wherein, the corresponding reminding operation is performed according to the risk level, including: when the risk level is the first level, controlling the intelligent wearable device of the guardian object to perform a first risk prompting operation, and controlling the intelligent wearable device of the guardian to perform an emergency reminding operation; when the risk level is the second level, controlling the intelligent wearable device of the guardian object to perform a second risk prompting operation, and controlling the intelligent wearable device of the guardian to perform a silent reminding operation; the reminding strength of the second risk prompting operation is lower than that of the first risk prompting operation. 7.The safety alert method based on environmental perception according to claim 1, wherein, Further comprising: obtaining feedback information of the guardian on the reminding operation, the feedback information including confirmation of risk and false alarm ignoring; according to the feedback information, counting the number of consecutive false alarms for the same type of risk reminding; when the number of consecutive false alarms reaches a preset number, increasing the risk threshold of the environmental parameter, and the adjusted risk threshold is not lower than a preset safe standard limit.

8. An environment-aware based safety alert system, characterized by, including: a collection processing module for obtaining a plurality of different dimension environmental parameters of the environment in which the guardian object is located; a fusion calculation module for obtaining a comprehensive risk indicator representing the overall risk of the environment by fusion calculation based on a preset weight coefficient according to each type of environmental parameter; the weight coefficient is used to reflect the influence degree of the environmental parameter on the overall environmental risk; a risk assessment module for determining the risk level of the current environment according to the comparison result of the comprehensive risk indicator and the numerical value of at least one environmental parameter with the preset risk threshold; a control module for performing a corresponding reminding operation according to the risk level.

9. An electronic device, comprising: including a memory and a processor, wherein the memory is used to store programs; the processor is coupled with the memory and is used to execute the programs stored in the memory to realize the steps in the safety reminding method based on environment perception in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, a computer readable program or instruction is stored, and the program or instruction is executed by the processor to realize the steps in the safety reminding method based on environment perception in any one of claims 1 to 7. a computer readable program or instruction is stored, and the program or instruction is executed by the processor to realize the steps in the safety reminding method based on environment perception in any one of claims 1 to 7.

Images