An emergency distress signaling method, device, and electronic equipment based on BeiDou short message service.

Abstract

This invention discloses an emergency distress signaling method, device, and electronic equipment based on BeiDou short message service. The method includes: real-time acquisition of raw sensor data from the user; data fusion processing of the raw sensor data to obtain attitude angle, movement speed, and activity status; inputting the attitude angle, movement speed, and activity status into a preset fall detection model and a stationary timeout detection model to identify whether the user is in an abnormal state; if the user is in an abnormal state, obtaining distress information and sending it to a rescue center for emergency assistance. This application can intelligently identify the user's unexpected state and automatically send distress signals, even when the user is unconscious, it can still provide timely assistance.

Description

Technical Field

[0001] This invention relates to the field of outdoor safety and emergency rescue, and in particular to an emergency distress signaling method, device, and electronic equipment based on BeiDou short message service. Background Technology

[0002] Currently, the safety of outdoor workers mainly relies on two methods: traditional manual distress signals and manual safety checks. However, both of these methods have significant drawbacks. Traditional manual distress signals require the user to be conscious and actively operate them. If a fall, impact, or other accident causes unconsciousness, effective distress signals cannot be sent. Furthermore, manual operation is susceptible to psychological factors in emergency situations, potentially leading to errors or delays.

[0003] Manual safety checks rely on scheduled contact or pre-set checkpoints, resulting in limited frequency and an inability to achieve real-time monitoring. For solo outdoor workers, in the event of an accident, the anomaly is often only discovered at the next scheduled contact time, missing the optimal rescue opportunity.

[0004] In addition, most existing emergency rescue devices lack intelligent recognition functions and cannot automatically determine the user's true status, which can easily lead to false alarms or missed alarms, affecting rescue efficiency and resource allocation. Summary of the Invention

[0005] This invention provides an emergency distress signaling method, device, and electronic equipment based on BeiDou short message service, which can intelligently identify the user's unexpected state and automatically send distress signals.

[0006] In one aspect of the present invention, an emergency distress signaling method based on BeiDou short message service is provided. The method includes: Use sensors to collect raw sensor data from users in real time; The raw sensor data is fused to obtain the attitude angle, motion speed, and activity status. The posture angle, the movement speed, and the activity state are input into a preset fall detection model and a stationary timeout detection model to identify whether the user is in an abnormal state. If the user is in an abnormal state, the distress message will be obtained and sent to the rescue center for emergency assistance.

[0007] In another aspect of the present invention, an emergency distress signaling device based on BeiDou short message service is provided. The device includes: The sensor acquisition module is configured to acquire raw sensor data from the user in real time using sensors; The data fusion module is configured to perform data fusion processing on the raw sensor data to obtain attitude angle, motion speed and activity state; An anomaly detection module is configured to input the posture angle, the movement speed, and the activity state into a preset fall detection model and a stationary timeout detection model to identify whether the user is in an abnormal state. The distress communication module is configured to receive distress information and send it to the rescue center for emergency assistance if the user is in an abnormal state.

[0008] In another aspect of the present invention, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the various steps of the above-described emergency distress method based on BeiDou short messages.

[0009] The beneficial effects of this invention are as follows: It utilizes sensors to collect raw sensor data from users in real time; it performs data fusion processing on the raw sensor data to obtain posture angle, movement speed, and activity status; it inputs the posture angle, movement speed, and activity status into preset fall detection models and static timeout detection models to identify whether the user is in an abnormal state; if the user is in an abnormal state, it obtains distress information and sends it to the rescue center for emergency assistance. This application can intelligently identify the user's accidental state and automatically send distress signals, even when the user is unconscious, providing timely rescue and all-weather safety protection for outdoor workers. Attached Figure Description

[0010] Figure 1 This is a flowchart of an emergency distress method based on BeiDou short message service, according to an embodiment of the present invention. Figure 2 This is another flowchart of an emergency distress method based on BeiDou short messages according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of an emergency distress device based on BeiDou short message service according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the structure of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0011] To explain in detail the technical content, objectives, and effects of the present invention, the following description is provided in conjunction with the embodiments and accompanying drawings.

[0012] In related technologies, outdoor operations such as mountain engineering construction, geological exploration, power line inspection, and forest fire prevention patrols are increasingly being carried out in various outdoor work scenarios. Workers are often in complex environments such as mountains, deserts, and forests, and face safety risks such as falls, impacts, and loss of consciousness due to sudden illness. This places higher demands on the safety protection measures for outdoor workers.

[0013] Currently, safety assurance for outdoor workers mainly relies on two methods: traditional manual distress signals and manual safety checks. However, both of these methods have significant technical shortcomings and cannot meet the all-weather, high-reliability safety requirements of outdoor operations. Traditional manual distress signals depend on the user being conscious and actively operating them. When outdoor workers experience falls, impacts, or other accidents and become unconscious, they cannot trigger a distress signal, rendering effective rescue impossible. Furthermore, in emergency situations, workers' mental state is easily affected, and even if conscious, manual operation can easily lead to errors and delays, missing crucial rescue time. Manual safety checks primarily rely on scheduled contact and pre-set checkpoints to monitor worker status. Limited by manpower and geographical constraints, the frequency of checks is significantly restricted, making real-time monitoring of workers' movement and vital signs impossible. Especially for solo outdoor workers, once an accident occurs, abnormalities are often only discovered at the next scheduled contact or checkpoint, easily missing the optimal rescue opportunity and increasing the risk of injury or death.

[0014] In addition, existing emergency rescue devices generally lack effective intelligent status recognition functions, making it impossible to accurately and automatically judge the actual movement and physical status of outdoor workers. In practical applications, there is a tendency for abnormal statuses to be missed and normal statuses to be falsely reported. This not only leads to accidents not being detected in time due to missed reports, but also causes ineffective consumption of rescue resources due to false reports, seriously affecting the overall efficiency of emergency rescue and the rational allocation of rescue resources.

[0015] In summary, existing outdoor safety measures suffer from technical problems such as reliance on manual operation, inability to monitor in real time, and lack of intelligent recognition capabilities. These limitations make it difficult to effectively address various safety risks in outdoor operations. There is an urgent need for a technical solution that can automatically identify unexpected situations and automatically request help without human intervention, in order to improve the intelligence and automation level of outdoor work safety and provide reliable emergency rescue protection for outdoor workers.

[0016] To address the aforementioned problems, this application provides an emergency distress signaling method, device, and electronic equipment based on BeiDou short message service. The following is a detailed description of the emergency distress signaling method based on BeiDou short message service.

[0017] The emergency distress method based on BeiDou short messages in this application can be integrated into hardware carriers such as wearable devices, vehicle terminals, and portable inspection equipment. It is applicable to all outdoor work and activity scenarios without terrestrial mobile communication network coverage and where there is a risk of falling or loss of contact due to stationary movement. It can effectively solve the technical pain points of traditional distress methods that rely on manual operation and cannot communicate without a network. It provides all-weather, intelligent emergency distress protection for personnel in various outdoor scenarios and has broad industrial application value and practical use value.

[0018] The following details the emergency distress method based on BeiDou short messages in this invention, with reference to the appendix. Figure 1 and attached Figure 2 This includes steps 110-140.

[0019] Step 110: Use sensors to collect raw sensor data from users in real time.

[0020] In some embodiments, the sensor employs a high-precision 9-axis inertial measurement unit (IMU) sensor, which integrates a three-axis accelerometer, a three-axis gyroscope, and a three-axis magnetometer. It has a sampling frequency of 100Hz and a data accuracy of 16 bits, and can collect real-time data on the user's three-axis acceleration (range ±16g), three-axis angular velocity (range ±2000dps), and three-axis magnetic field strength during the user's movement. It is the core sensing component for realizing the monitoring of the user's movement status.

[0021] In some embodiments, raw sensor data refers to physical quantity electrical signal conversion data directly acquired by the sensor without any processing. Raw sensor data includes triaxial acceleration data, triaxial angular velocity data, and triaxial magnetic field strength data.

[0022] In some embodiments, before acquiring raw sensor data from the user in real time using sensors, the process includes: performing sensor calibration, communication testing, battery power checking, and BeiDou signal acquisition checking. Sensor calibration includes: performing zero-point calibration, sensitivity calibration, and error compensation on the IMU inertial measurement unit sensor to ensure that the raw data acquired by the sensor is accurate and free of systematic errors. Communication testing includes: performing communication link testing on the BeiDou short message communication module and the BeiDou positioning module to verify whether the communication connection between the module and the BeiDou satellite network is normal, ensuring that subsequent positioning and distress message transmission functions are available. Battery power checking includes: detecting the remaining power of the device's built-in battery; if the power is lower than a preset operating threshold (e.g., 20%), a low power alarm is issued, prompting the user to charge; if the power is higher than the threshold, the power is considered sufficient. BeiDou signal acquisition checking includes: verifying whether the BeiDou positioning module can normally capture BeiDou satellite signals to ensure accurate acquisition of the user's location information; if the signal strength is lower than a preset threshold, the user is prompted to move to an area with good satellite signal.

[0023] In this way, through a full-dimensional self-check of sensor calibration, communication testing, battery power check, and BeiDou signal acquisition check, it ensures that all core functional modules are in normal working condition before the equipment enters monitoring mode. This effectively avoids risks such as inaccurate monitoring and failed distress calls caused by problems such as uncalibrated sensors, communication module failures, low battery, and missing satellite signals. It improves the stability and reliability of the entire distress call method and ensures that the equipment can always play an effective role in safety monitoring and distress call during outdoor work.

[0024] Step 120: Perform data fusion processing on the original sensor data to obtain attitude angle, motion speed and activity status.

[0025] Data fusion processing refers to the process of integrating, filtering, and processing multi-dimensional and multi-type raw sensor data using specific algorithms. Attitude angles refer to the spatial attitude parameters of a user wearing sensors, including pitch, roll, and yaw angles; changes in these values ​​can reflect whether the user experiences abrupt attitude changes or other abnormal situations. Motion speed refers to the user's real-time movement speed calculated by integrating acceleration data. Activity state refers to a qualitative description of the user's motion behavior based on sensor data processing, including states such as motion, stillness, and abrupt attitude changes.

[0026] In some embodiments, data fusion processing of the raw sensor data to obtain attitude angles, motion speed, and activity status includes: fusing triaxial acceleration data, triaxial angular velocity data, and triaxial magnetic field strength data using a Kalman filter algorithm to calculate the user's attitude angles, motion speed, and activity status in real time. The processor calls the Kalman filter algorithm program, taking the triaxial acceleration data, triaxial angular velocity data, and triaxial magnetic field strength data as algorithm inputs. Through the prediction, update, and correction steps of the Kalman filter, data errors caused by sensor noise and external environmental interference are eliminated. The three types of data are fused and calculated to obtain the user's attitude angles (pitch angle, roll angle, yaw angle), motion speed, and activity status in real time, with the calculation results updated in real time.

[0027] In this way, the Kalman filter algorithm is used to fuse multi-source sensor data, which effectively improves the accuracy and stability of attitude angle, motion speed and activity state calculation results, avoids the limitations of single sensor data and judgment errors caused by noise interference, lays a data foundation for the accuracy of subsequent abnormal state identification, and reduces the probability of model misjudgment and omission.

[0028] Step 130: Input the posture angle, the movement speed, and the activity state into the preset fall detection model and the stationary timeout detection model to identify whether the user is in an abnormal state.

[0029] In some embodiments, the fall detection model is an abnormal state recognition model trained based on machine learning algorithms (such as support vector machines, SVM). The training samples include various movement patterns such as normal walking, running, falling, and slipping. It can determine whether a user has experienced a fall based on posture angle and movement speed characteristic parameters, and the model's recognition accuracy is no less than 95%. The stationary timeout detection model is an abnormal state recognition model based on a preset threshold and the correlation of physiological parameters. It can determine whether a user has experienced a stationary abnormality exceeding the safe time limit based on the user's activity status and heart rate detection results.

[0030] In this embodiment, the drop detection model can be, but is not limited to, the following two types, including: Kernel-optimized Support Vector Machine (SVM) model: A binary SVM model using radial basis function (RBF) kernels is preferred. The kernel function maps low-dimensional motion features to a high-dimensional space, constructing the optimal classification hyperplane for normal motion and abnormal falls. The model has a simple structure, strong learning ability with small samples, achieves high accuracy without requiring a large number of training samples, and consumes very little computational power during inference. It can be directly deployed on low-power embedded terminals and is suitable for real-time monitoring scenarios in outdoor wearable devices.

[0031] Random Forest (RF) Ensemble Learning Model: This binary classification random forest model employs an ensemble of multiple CART decision trees. Multiple training subsets are generated through bootstrap sampling, and each subset is used to train a single decision tree. The final classification conclusion is output based on the voting results of all trees. The model exhibits strong resistance to overfitting, automatically identifies the contribution weights of different motion features to fall detection, and demonstrates excellent robustness to sensor noise data and anomalous data in complex outdoor scenarios, effectively reducing the false alarm probability of normal motion behavior.

[0032] Model input data: The time-series feature sequence obtained after data fusion processing in step 120. A single input sample is continuous feature data within a preset sliding time window (e.g., 2 seconds, sampling frequency 100Hz, corresponding to 200 sampling points), including: ① Attitude and angle characteristics: real-time values ​​and rate of change per unit time of pitch angle, roll angle, and yaw angle; ② Kinematic characteristics: real-time velocity, peak acceleration, mean acceleration, and variance of acceleration; ③ Auxiliary features: fused and filtered values ​​of triaxial acceleration, triaxial angular velocity, and triaxial magnetic field strength.

[0033] Model output data includes: ①Category tags: 0 represents normal movement state, 1 represents abnormal fall state; ② Confidence score: A probability value between 0 and 1, representing the reliability of the model's classification. When the confidence score is greater than or equal to a preset threshold (e.g., 0.9), the final classification result is output.

[0034] ③ Standardized distress message package: In step 140 below, a standardized distress message package will be continuously sent to the rescue center in a loop.

[0035] The core training process of the drop detection model constructed with the support vector machine (SVM) model includes: (1) Model initialization: Set the model kernel function to RBF kernel, initialize the optimization range of penalty coefficient C and kernel function parameter gamma, and use 5-fold cross-validation to optimize hyperparameters to avoid overfitting and underfitting; (2) Model iterative training: Input the preprocessed training set data into the initialization model, solve the optimal classification hyperplane through the sequence minimum optimization (SMO) algorithm, minimize the classification error, and complete the initial training of the model; (3) Model validation and hyperparameter optimization: Validate the model and optimize the hyperparameters. The validation set data is input into the initially trained model, and the core evaluation indicators of recognition accuracy, precision, recall, and F1 score are calculated. Based on the validation results, the penalty coefficient C and gamma parameters are optimized through the grid search algorithm until the model's recognition accuracy on the validation set is ≥98% and the recall is ≥99%. (4) Model testing and solidification: The optimized model is tested for generalization ability using independent test set data to ensure that the model's recognition accuracy on the test set is not less than 95% and the false alarm rate is ≤2%. After the test is passed, the model parameters are solidified and converted into a model file that can be deployed on an embedded terminal.

[0036] The core training process of the fall detection model constructed by the random forest (RF) ensemble learning model is as follows: (1) Model initialization: Set the number of decision trees to 100, the maximum depth of the trees to 10, the minimum number of split samples for the smallest node to 5, and the minimum number of leaf samples to 2. Use the Gini coefficient as the evaluation index for node splitting. (2) Model iterative training: Generate an independent training subset for each decision tree by using bootstrap sampling. Complete the training of a single decision tree in parallel and recursively complete node splitting until the preset stopping condition is reached. (3) Model verification and optimization: Verify the model performance based on out-of-bag (OOB) data and validation set data, calculate feature importance, remove redundant features, optimize the number and depth parameters of decision trees, reduce model complexity, and improve generalization ability. (4) Model testing and solidification: Use test set data to complete the final performance test to ensure that the model recognition accuracy is not less than 95%. After the test is passed, solidify the model parameters and complete the conversion of the model deployment file.

[0037] In this embodiment, the static timeout detection model can adopt, but is not limited to, the following two types, including: Multi-condition rule-integrated decision tree model: Based on the CART algorithm, this binary decision tree model uses preset thresholds as base nodes and integrates multiple dimensions of conditions, including activity state, rest duration, and physiological parameters, to construct classification decision paths. The model is highly interpretable, requires no complex iterative training, and can be directly implemented in embedded terminals via logic code. It consumes very little computing power, and its standby power consumption is negligible, making it perfectly suited for low-power standby monitoring scenarios.

[0038] Logistic Regression (LR) Binary Classification Model: This model employs a logistic regression model with L2 regularization. The sigmoid activation function maps multi-dimensional input features to a probability range of 0-1, enabling binary classification between normal stationary states and abnormal stationary states with timeouts. The model has a minimal structure, a very small number of parameters, and fast inference speed, achieving millisecond-level inference on low-power MCUs. Furthermore, its training and deployment processes are simple, making it suitable for the low-power requirements of long-term continuous monitoring in outdoor equipment.

[0039] The input data for the static timeout detection model includes: continuous feature data within a preset long sliding time window (e.g., 5 minutes, sampling frequency 1Hz, corresponding to 300 sampling points), specifically including: ①Activity status characteristics: duration of continuous stillness of the user, number of posture changes per unit time, and frequency of motion state switching; ② Vital signs: real-time heart rate, heart rate variability, heart rate data validity rate, and body temperature. ③ Auxiliary features: fluctuation variance of user posture angle, mean and variance of acceleration.

[0040] The output data of the static timeout detection model includes: ①Category tags: 0 represents normal static state, 1 represents abnormal static state with timeout; ② Confidence score: A probability value between 0 and 1, representing the reliability of the model's classification. When the confidence score is greater than or equal to a preset threshold (e.g., 0.85), the final classification result is output.

[0041] ③ Standardized distress message package: In step 140 below, a standardized distress message package will be continuously sent to the rescue center in a loop.

[0042] The core training process of the static timeout detection model constructed using the CART decision tree model includes: (1) Feature and label definition: The core features are continuous static duration, heart rate fluctuation rate, and posture fluctuation variance. The classification label is “normal static / abnormal static timeout”. The optimal splitting threshold range for each feature is determined. (2) Model initialization: A binary decision tree is constructed based on the CART algorithm. The maximum depth of the tree is set to 5, the minimum number of leaf node samples is set to 10, and the Gini coefficient is used as the evaluation index for node splitting. (3) Iterative training of the model: The training set data is input into the initialization model, the Gini coefficient of each feature is calculated recursively, and the Gini coefficient is selected. The feature with the smallest coefficient and the corresponding threshold are used as the basis for node splitting to complete the growth and construction of the decision tree; (4) Model pruning and optimization: The cost complexity pruning (CCP) algorithm is used to prune the trained decision tree. The pruning parameters are optimized by the validation set data to remove overfitting branches and improve the generalization ability of the model; (5) Model testing and solidification: The optimized model is tested by the test set data to ensure that the recognition accuracy of the model on the test set is not less than 96% and the false alarm rate is ≤1.5%; After the test is passed, the decision path of the decision tree is converted into condition judgment code that can be executed on the embedded terminal to complete the model solidification.

[0043] The core training process of the static timeout detection model constructed with logistic regression model includes: (1) Model initialization: Initialize the model weight coefficients and bias terms, set the optimization range of L2 regularization coefficients, and use 5-fold cross-validation to optimize hyperparameters to avoid model overfitting; (2) Model iterative training: Input the training set data into the initialization model, use binary cross-entropy as the loss function, and iteratively optimize the model weights and bias terms through gradient descent algorithm to minimize the classification loss until the loss function converges; (3) Model validation and optimization: Input the validation set data into the trained model, calculate the core evaluation index, optimize the regularization coefficient, and balance the model's fitting ability and generalization ability; (4) Model testing and solidification: Use the test set data to complete the final performance test to ensure that the model recognition accuracy is not less than 95%. After the test is passed, solidify the model weights and bias terms and convert them into code that can be deployed on embedded terminals.

[0044] In some embodiments, an abnormal state refers to an abnormal state that may endanger the life of a user during outdoor work. Abnormal states include fall anomalies and stationary timeout anomalies. The step of inputting the posture angle, movement speed, and activity state into a preset fall detection model and a stationary timeout detection model to identify whether the user is in an abnormal state includes: inputting the posture angle and movement speed into the fall detection model to determine whether a fall anomaly has occurred; inputting the activity state into the stationary timeout detection model to determine whether a stationary timeout anomaly has occurred; if either a fall anomaly or a stationary timeout anomaly is detected, it indicates that the user is in an abnormal state. For fall anomaly judgment: the processor inputs the posture angle and movement speed into the fall detection model in real time, and the fall detection model identifies and judges whether the user has experienced a fall anomaly based on preset judgment logic and thresholds. For stationary timeout anomaly judgment: the processor inputs the activity state into the stationary timeout detection model in real time, and the stationary timeout detection model identifies and judges whether the user has experienced a stationary timeout anomaly based on preset thresholds and physiological parameter association conditions. For comprehensive judgment of abnormal status: The processor receives the judgment results of the drop detection model and the stationary timeout detection model. If either model determines that there is a corresponding abnormality (drop abnormality or stationary timeout abnormality), the processor comprehensively determines that the user is in an abnormal state and immediately triggers the subsequent rescue process; if both models determine that there is no abnormality, the monitoring state continues.

[0045] This approach divides the overall anomaly identification into two specific judgments: fall anomalies and stationary timeout anomalies, enabling accurate identification of different types of outdoor accidents. Simultaneously, the use of an "OR logic" comprehensive judgment method ensures that any dangerous anomaly can be identified promptly, further improving the comprehensiveness and timeliness of anomaly identification.

[0046] In some embodiments, inputting the posture angle and the movement speed into the fall detection model to determine whether a fall anomaly has occurred includes: calculating acceleration in real time based on the movement speed; if the peak value of the acceleration exceeds a preset acceleration threshold within a preset first time period, and the user remains stationary within a preset second time period after the acceleration reaches its peak value, then a fall anomaly is determined to have occurred. The fall detection model receives the user's real-time movement speed transmitted by the processor, and calculates the user's movement acceleration in real time based on the differential relationship between acceleration and velocity. The acceleration calculation result is updated synchronously with the movement speed. The fall detection model retrieves preset first time period, acceleration threshold, and second time period parameters, and monitors the calculated acceleration in real time: if the peak value of the acceleration exceeds a preset acceleration threshold (e.g., 8g) within the preset first time period (e.g., 10 seconds), then a subsequent stationary state verification step is triggered; if the peak value of the acceleration does not exceed the threshold, then it is determined that no fall anomaly has occurred. Static state verification steps: Timing starts at the point when the acceleration reaches its peak. The fall detection model continuously monitors the user's motion state. If the user remains stationary within a preset second time period (e.g., 30 seconds), or remains stationary for an extended period, it is determined that a fall anomaly has occurred. If the user resumes motion within the second time period, it is determined that there is no fall anomaly, and the alarm is deactivated.

[0047] In this way, the abnormal fall is judged by the sudden acceleration and subsequent stillness. This is close to the actual movement characteristics of users falling in outdoor work. There will be obvious acceleration peaks during the fall, and after the fall, they will usually be in a still state due to injury. This judgment logic realizes the accurate identification of abnormal falls, effectively distinguishes the acceleration fluctuations in the fall from normal movement (such as rapid turning and jumping), reduces the probability of false alarms caused by acceleration changes in normal movement, and improves the recognition accuracy of the fall detection model.

[0048] In some embodiments, inputting the attitude angle and the movement speed into the fall detection model to determine whether a fall anomaly has occurred includes: calculating the abrupt change angle in real time based on the attitude angle at the previous moment and the attitude angle at the current moment; if the abrupt change angle exceeds a preset second threshold, and the user remains stationary for a preset third time period after the abrupt change angle reaches the preset second threshold, then a fall anomaly is determined to have occurred. The fall detection model receives the user's attitude angles (pitch angle, roll angle, yaw angle) at the previous moment and the current moment transmitted by the processor, calculates the abrupt change angle of each attitude angle according to the angle difference calculation rules, and then synthesizes them to obtain the overall attitude abrupt change angle of the user. The fall detection model retrieves the preset second threshold (e.g., 60 degrees) and the third time period parameter, and monitors the calculated abrupt change angle in real time: if the abrupt change angle exceeds the preset second threshold (e.g., 60 degrees), then the subsequent stationary state verification step is triggered; if the abrupt change angle does not exceed the second threshold, then it is determined that no fall anomaly has occurred. Static state verification: Timing starts at the point when the sudden angle reaches the second threshold. The fall detection model continuously monitors the user's movement state. If the user remains stationary within the preset third time period (e.g., 15 seconds), it is determined that a fall anomaly has occurred. If the user returns to normal posture and movement state within the third time period, it is determined that there is no fall anomaly and the alarm is deactivated.

[0049] In this way, the abnormality of a fall is judged by the sudden change in posture angle and subsequent stillness. This is consistent with another core motion feature of user falls in outdoor operations. During the fall, the body will undergo violent posture twisting, resulting in obvious angle changes, and after landing, it will remain still in an abnormal posture. This judgment logic complements the acceleration judgment logic, realizing comprehensive recognition of different fall scenarios (such as forward tilting fall and side roll fall). This further improves the scenario adaptability and recognition accuracy of the fall detection model, and effectively avoids the problem of missed detection in some fall scenarios where the acceleration peak is not obvious but the posture change is significant.

[0050] In some embodiments, inputting the activity state into the resting timeout detection model to determine whether a resting timeout anomaly has occurred includes: if the user's continuous resting time exceeds a preset third threshold, and no heart rate data is detected within the continuous resting time period, then a resting timeout anomaly is determined. The resting timeout detection model receives the user's real-time activity state transmitted by the processor, continuously monitors and times the user's resting state, and records the user's continuous resting time. The resting timeout detection model retrieves a preset third threshold (e.g., 30 minutes). If the user's continuous resting time exceeds the third threshold, a subsequent heart rate data verification step is triggered; if the continuous resting time does not exceed the threshold, it is determined that no resting timeout anomaly has occurred. After the user's continuous resting time exceeds the third threshold, the resting timeout detection model retrieves heart rate data collected by the vital signs monitoring module. If no valid heart rate data is detected within the continuous resting time period, a resting timeout anomaly is determined; if valid heart rate data is detected, it is determined that the user is actively resting, not in an abnormal state, and the resting timer is reset.

[0051] In this way, the abnormality of rest timeout is judged by the continuous rest duration and the absence of heart rate data. This is close to the characteristics of rest timeout caused by accidents (such as coma or loss of consciousness) during outdoor operations. It effectively distinguishes between users' active rest (with valid heart rate data) and accidental rest (without valid heart rate data), avoids false alarms caused by users' normal rest, and ensures timely identification of accidents such as loss of consciousness. This improves the practicality and accuracy of the rest timeout detection model.

[0052] Step 140: If the user is in the abnormal state, obtain the distress information and send it to the rescue center for emergency assistance.

[0053] In some embodiments, obtaining distress information and sending it to the rescue center for emergency assistance includes: obtaining the user's location information, vital sign data, and the timestamp of the abnormal state occurrence; integrating the location information, vital sign data, and timestamp to generate distress information; and sending the distress information to the rescue center for emergency assistance using BeiDou short message communication technology. When the processor determines that the user is in an abnormal state, it immediately triggers multi-module data acquisition: obtaining the user's current location information, including longitude (accurate to 6 decimal places), latitude (accurate to 6 decimal places), and altitude, through a BeiDou-3 dual-frequency positioning module (such as the HD8120 chip from Huada BeiDou Company), with a positioning accuracy better than 3 meters and an altitude measurement accuracy better than 5 meters; collecting the user's current vital sign data, including heart rate (using photoplethysmography technology, accuracy ±2 bpm) and body temperature (using infrared thermometry technology, accuracy ±0.2℃), through a vital sign monitoring module; and recording the precise timestamp of the abnormal state occurrence through a GPS time synchronization module (based on GPS time synchronization, accuracy down to the millisecond level). The processor integrates the acquired location information, vital signs data, and timestamps according to a preset standardized format, while adding auxiliary information such as the device's unique identifier (e.g., ID), device battery level, and anomaly type to generate a structured and standardized distress message packet, ensuring data integrity and uniform format. The BeiDou short message distress message format is shown in Table 1 below. The processor transmits the generated distress message packet to the BeiDou short message communication module (e.g., the HG-BDS3 module from Haige Communications). This module, through the BeiDou satellite network, can send 120 bytes of data at a time, with a communication success rate >95%, and automatically sends the distress message packet to a preset dedicated rescue center number, enabling emergency distress message transmission in environments without terrestrial network coverage. After the device sends the distress message for the first time, the processor immediately controls the device's local audible and visual alarm module to operate, driving the LED light to flash continuously at a frequency of 2Hz, and simultaneously controlling the buzzer to sound continuously at a fixed audio frequency of 3kHz, issuing a local alarm prompt. While activating the local audible and visual alarm, the processor controls the Beidou short message communication module to continuously send standardized distress information packets to the rescue center at fixed time intervals of 30 seconds until the processor receives a rescue response confirmation signal from the rescue center, or until the device's built-in battery is depleted and unable to continue operating.

[0054] Table 1

[0055] In this way, by integrating key data such as location, vital signs, and time, comprehensive and accurate rescue data is provided to the rescue center. This enables the center to quickly locate the user, assess their vital signs and the duration of any abnormalities, and develop targeted rescue plans. Simultaneously, the use of BeiDou short message communication technology solves the communication challenges of outdoor environments without terrestrial mobile communication network coverage, ensuring that distress messages can be transmitted over long distances with a high success rate. This achieves dual rescue protection through accurate positioning and effective communication, significantly improving the efficiency and targeting of emergency rescue efforts. Local audible and visual alarms can alert nearby workers or passersby to provide timely on-site assistance. The 30-second interval cyclical transmission effectively avoids information loss caused by satellite signal interference or transmission interruptions during single transmissions, ensuring the rescue center can continuously and stably receive distress messages, greatly improving the timeliness, effectiveness, and success rate of emergency rescue.

[0056] The emergency distress method based on BeiDou short messages described in this application relies on the short message communication characteristics of the BeiDou-3 satellite navigation system, the motion state perception capability of the IMU inertial measurement unit, and the anomaly recognition capability of artificial intelligence algorithms. It overcomes the coverage limitations of terrestrial mobile communication networks, achieving automatic identification of unexpected situations and proactive distress calls. It can be widely applied to outdoor work and activity scenarios where there is no terrestrial communication network coverage, personnel are working alone, or the working environment is complex and there is a risk of falls or loss of contact. According to another aspect of the invention… Figure 3 This is a schematic diagram illustrating the structure of an emergency distress device based on BeiDou short message service according to an embodiment of the present invention. (Refer to...) Figure 3 The device 300 includes a sensor acquisition module 310, a data fusion module 320, an anomaly identification module 330, and a distress communication module 340.

[0057] The sensor acquisition module 310 is configured to acquire raw sensor data from the user in real time using the sensor.

[0058] The data fusion module 320 is configured to perform data fusion processing on the raw sensor data to obtain attitude angle, motion speed and activity status.

[0059] The anomaly identification module 330 is configured to input the posture angle, the movement speed and the activity state into a preset fall detection model and a stationary timeout detection model to identify whether the user is in an abnormal state.

[0060] The distress communication module 340 is configured to acquire distress information and send it to the rescue center for emergency assistance if the user is in the abnormal state.

[0061] In some embodiments, the device 300 further includes a power management module, a vital signs monitoring module, a BeiDou-3 positioning module, an audible and visual alarm module, a processor, and a memory. The power management module provides stable power to all power-consuming modules of the device, integrating an energy storage battery, power conversion, power detection, and low-power management unit. It monitors the power level in real time and triggers a low-power mode to ensure outdoor endurance and stable operation of each module. The vital signs monitoring module collects the wearer's heart rate, body temperature, and other physiological parameters in real time, employing photoplethysmography (PPG) and infrared thermometry technologies. This provides high data accuracy, offering core physiological evidence for abnormal condition assessment and distress signals, and transmits the data to the processor in real time. The BeiDou-3 positioning module achieves precise positioning based on the BeiDou-3 satellite system, overcoming the limitations of terrestrial networks. It acquires latitude, longitude, and altitude information with an accuracy better than 3 meters, generating millisecond-level timestamps to provide high-precision location and time data for distress signals. The audible and visual alarm module is a local alarm execution unit, including an LED light-emitting unit and a buzzer sound-emitting unit. Upon triggering a distress call, it flashes the LED at a 2Hz frequency and sounds the buzzer at a 3kHz audio frequency, providing local visual and auditory warnings until a rescue response is received or the battery is depleted. The processor, as the core control and computing hub of the device, adapts to multi-module communication, completes sensor data fusion, intelligent identification of abnormal states, and coordinates the entire process of command issuance, including distress information integration, BeiDou short message transmission, audible and visual alarm activation, and cyclical transmission of distress signals. The memory is used for long-term storage of algorithm programs, detection models, and preset thresholds, and temporarily caches various real-time acquired and processed data, providing data support for processor operations, while also storing device operating status data.

[0062] According to another aspect of the invention, Figure 4 This is a schematic diagram illustrating the structure of an electronic device 400 according to an embodiment of the present invention. (Refer to...) Figure 4 The electronic device 400 includes a memory 410, a processor 420, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements each step of the emergency distress method based on Beidou short message service.

[0063] The beneficial effects of the device and electronic equipment of the present invention are the same as those of the methods described above, and will not be repeated here.

[0064] In summary, this application utilizes on-device sensors to collect real-time information on the user's motion state and posture changes, constructing fall detection and static timeout detection models to identify abnormal user states. It then automatically sends a rescue signal using BeiDou short message communication, transmitting distress information including precise location and vital signs to the rescue center. This achieves a closed-loop process from user motion state monitoring to anomaly identification and automatic distress signaling, overcoming the limitations of traditional manual distress devices that rely on the user's conscious awareness and active operation. Even if the user falls unconscious due to an accident, the distress signaling process can be completed automatically, providing proactive, all-weather safety protection for outdoor workers, effectively shortening rescue response time, and improving the success rate of outdoor accident rescues. Upon device startup, a comprehensive self-check is performed, including sensor calibration, communication testing, battery level checks, and BeiDou signal acquisition checks. This ensures that all core functional modules are in normal working order before entering monitoring mode, effectively mitigating risks such as inaccurate monitoring and failed distress calls caused by uncalibrated sensors, communication module malfunctions, low battery, or missing satellite signals. This enhances the stability and reliability of the entire distress call method, guaranteeing that the device can consistently provide effective safety monitoring and distress calls for outdoor workers. When collecting sensor data, a Kalman filter algorithm is used to fuse multi-source sensor data, effectively improving the accuracy and stability of attitude angle, motion speed, and activity status calculations. This avoids the limitations of single-sensor data and judgment errors caused by noise interference, laying a data foundation for accurate subsequent anomaly identification and reducing the probability of model misjudgment and missed detection. During anomaly identification, the overall anomaly identification is divided into two specific judgments: fall anomalies and stationary timeout anomalies, enabling accurate identification of different types of outdoor accidents. Simultaneously employing an OR logic-based comprehensive judgment method ensures that any dangerous or abnormal state can be identified in a timely manner, further enhancing the comprehensiveness and timeliness of abnormal state identification. During the transmission of distress messages, by integrating key data such as location, vital signs, and time, comprehensive and accurate rescue data is provided to the rescue center. This enables the rescue center to quickly locate the user's position, assess the user's vital signs and the duration of the abnormality, and formulate a targeted rescue plan. Furthermore, the use of BeiDou short message communication technology solves the communication problem in outdoor areas without terrestrial mobile communication network coverage, ensuring that distress messages can be transmitted over long distances with a high success rate. This achieves dual rescue protection of accurate positioning and effective communication, significantly improving the efficiency and targeting of emergency rescue. Local audible and visual alarms can alert workers or passersby in the surrounding area to provide timely on-site assistance; the 30-second interval cyclical transmission effectively avoids information loss caused by satellite signal interference or transmission interruptions during a single transmission, ensuring that the rescue center can continuously and stably receive distress messages, significantly improving the timeliness, effectiveness, and success rate of emergency rescue.

[0065] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent modifications made based on the content of the present invention's specification and drawings, or direct or indirect applications in related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. An emergency distress signaling method based on BeiDou short message service, characterized in that, include: Use sensors to collect raw sensor data from users in real time; The raw sensor data is fused to obtain the attitude angle, motion speed, and activity status. The posture angle, the movement speed, and the activity status are input into a preset fall detection model and a stationary timeout detection model to identify whether the user is in an abnormal state. If the user is in the aforementioned abnormal state, a distress message will be obtained and sent to the rescue center for emergency assistance.

2. The method according to claim 1, characterized in that, The raw sensor data includes triaxial acceleration data, triaxial angular velocity data, and triaxial magnetic field strength data; The process of fusing the raw sensor data to obtain the attitude angle, motion speed, and activity state includes: By fusing triaxial acceleration data, triaxial angular velocity data, and triaxial magnetic field strength data using the Kalman filter algorithm, the user's attitude angle, motion speed, and activity status can be calculated in real time.

3. The method according to claim 1, characterized in that, The step of inputting the posture angle, the movement speed, and the activity state into a preset fall detection model and a stationary timeout detection model to identify whether the user is in an abnormal state includes: The posture angle and the movement speed are input into the fall detection model to determine whether a fall anomaly has occurred. The activity status is input into the static timeout detection model to determine whether a static timeout anomaly has occurred. If the fall anomaly or the stationary timeout anomaly is detected, it indicates that the user is in an abnormal state.

4. The method according to claim 3, characterized in that, The step of inputting the posture angle and the movement speed into the fall detection model to determine whether a fall anomaly has occurred includes: Acceleration is calculated in real time based on the stated velocity of motion; If the peak value of the acceleration exceeds a preset acceleration threshold within a preset first time period, and the user remains stationary within a preset second time period after the acceleration reaches its peak value, then a fall anomaly is determined to have occurred.

5. The method according to claim 3, characterized in that, The step of inputting the posture angle and the movement speed into the fall detection model to determine whether a fall anomaly has occurred includes: The sudden change angle is calculated in real time based on the attitude angle at the previous moment and the attitude angle at the current moment. If the angle of change exceeds a preset second threshold, and the user remains stationary for a preset third time period after the angle of change reaches the preset second threshold, then it is determined that a fall has occurred.

6. The method according to claim 3, characterized in that, The step of inputting the activity status into the static timeout detection model to determine whether a static timeout anomaly has occurred includes: If the user's continuous static time exceeds a preset third threshold, and no heart rate data of the user is detected during the continuous static time period, it is judged as a static timeout abnormality.

7. The method according to claim 1, characterized in that, The process of obtaining distress information and sending it to the rescue center for emergency assistance includes: Obtain the user's location information, vital signs data, and the timestamp of the occurrence of the abnormal state; The location information, vital signs data, and timestamp are integrated to generate a distress message; The distress message was sent to the rescue center using BeiDou short message communication technology for emergency assistance.

8. The method according to claim 1, characterized in that, Before the process of using sensors to collect raw sensor data from users in real time includes: Perform sensor calibration, communication testing, battery power check, and BeiDou signal acquisition check.

9. An emergency distress device based on BeiDou short message service, characterized in that, include: The sensor acquisition module is configured to acquire raw sensor data from the user in real time using sensors; The data fusion module is configured to perform data fusion processing on the raw sensor data to obtain attitude angle, motion speed and activity state; An anomaly detection module is configured to input the posture angle, the movement speed, and the activity state into a preset fall detection model and a stationary timeout detection model to identify whether the user is in an abnormal state. The distress communication module is configured to acquire distress information and send it to the rescue center for emergency assistance if the user is in the abnormal state.

10. An electronic device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements each step of the emergency distress method based on BeiDou short message as described in any one of claims 1 to 8.

Images