Personnel tumble detection method fusing impulse and angle

Abstract

The invention provides a person falling detection method fusing impulse and angle, and relates to the technical field of intelligent monitoring, and the method comprises the steps: collecting three-dimensional acceleration data and three-dimensional angle change data in the motion process of a user side based on a wearable motion detection device; constructing a motion impulse calculation model, calculating motion impulse data according to the three-dimensional acceleration data, and extracting corresponding motion impulse features; constructing an angle change analysis model and obtaining angle change characteristics based on the three-dimensional angle change data; and a tumble detection fusion model is constructed, and a tumble detection result is generated according to the motion impulse characteristics and the angle change characteristics and is sent to a management end, so that the accuracy and reliability of tumble detection can be significantly improved, the misjudgment rate can be reduced, and a powerful guarantee is provided for personnel safety through fusion of multi-dimensional data.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent monitoring, and particularly to a method for detecting human falls by fusing impulse and angle. Background Art

[0002] As a common accidental event, if a human fall is not detected and processed in time, it may lead to serious consequences.

[0003] Existing fall detection methods have many deficiencies. For example, the detection method based on video surveillance is easily affected by environmental factors such as light and occlusion, and there is a risk of privacy infringement; while the detection method based on a single acceleration sensor often has insufficient detection accuracy and is prone to misjudging daily activities such as strenuous exercise and sudden turning as falls.

[0004] Therefore, it is necessary to provide a method for detecting human falls by fusing impulse and angle to solve the above technical problems. Summary of the Invention

[0005] To solve the above technical problems, the present invention provides a method for detecting human falls by fusing impulse and angle, which is used to solve the problems that existing fall detection methods cannot adapt to various environments, have a high risk of privacy infringement, and have insufficient detection accuracy and a high misjudgment rate.

[0006] The method for detecting human falls by fusing impulse and angle provided by the present invention includes:

[0007] Collecting three-dimensional acceleration data and three-dimensional angle change data during the movement of the user end based on a wearable motion detection device;

[0008] Constructing a motion impulse calculation model and calculating motion impulse data according to the three-dimensional acceleration data, and extracting corresponding motion impulse features;

[0009] Constructing an angle change analysis model and obtaining angle change features based on the three-dimensional angle change data;

[0010] Constructing a fall detection fusion model, generating a fall detection result according to the motion impulse features and the angle change features, and sending it to the management end.

[0011] Preferably, the wearable motion detection device is worn at the waist position of the user end;

[0012] The wearable motion detection device is equipped with an acceleration sensor and a gyroscope sensor. The acceleration sensor is used to collect the three-dimensional acceleration data, and the gyroscope sensor is used to collect the three-dimensional angle change data.

[0013] Preferably, constructing a motion impulse calculation model and calculating motion impulse data based on the three-dimensional acceleration data specifically includes:

[0014] Set a sampling period W, and divide the sampling period W into n sub-periods, and the duration of each sub-period k is

[0015] Within the sub-period k, calculate the average acceleration values in the x, y, and z axes respectively:

[0016]

[0017]

[0018] In the formula, respectively represent the average acceleration values in the x, y, and z axes within the sub-period k; Δt represents the duration of the sub-period k; a x (t), a y (t), a z (t) respectively represent the accelerations in the x, y, and z axes at time t;

[0019] Calculate the impulse increments in the x, y, and z axes within the sub-period k:

[0020]

[0021] In the formula, ΔI x，k , ΔI y，k , ΔI z，k respectively represent the impulse increments in the x, y, and z axes within the sub-period k;

[0022] Calculate the cumulative impulses in the x, y, and z axes within the sub-period k:

[0023] I x，k = I x，k-1 + ΔI x，k

[0024] I y，k = I y，k-1 + ΔI y，k

[0025] I z，k = I z，k-1 + ΔI z，k

[0026] In the formula, I x，k , I y，k , I z，k respectively represent the cumulative impulses in the x, y, and z axes within the sub-period k; I x,k-1 , I y,k-1 , I z,k-1Respectively represent the cumulative impulses in the x, y, and z axes within the sub-cycle k-1;

[0027] Based on the cumulative impulses in the x, y, and z axes within the sub-cycle k, determine that the cumulative impulses in the x, y, and z axes within the sampling period W are respectively I x = I x，n , I y = I y，n , I z = I z，n , that is, the motion impulse data.

[0028] Preferably, extract the motion impulse features based on the motion impulse data, specifically including:

[0029] Calculate the differences in the cumulative impulses in the x, y, and z axes between two adjacent sampling periods W:

[0030]

[0031] In the formula, ΔI x , ΔI y , ΔI z respectively represent the differences in the cumulative impulses in the x, y, and z axes between two adjacent sampling periods W; I x , I y , I z respectively represent the cumulative impulses in the x, y, and z axes within the sampling period W; respectively represent the cumulative impulses in the x, y, and z axes within the next sampling period W;

[0032] Calculate the impulse change rates in the x, y, and z axes:

[0033]

[0034]

[0035] In the formula, respectively represent the impulse change rates in the x, y, and z axes, that is, the motion impulse features.

[0036] Preferably, for the x axis, obtain the starting time u-1 and the ending time u of the sampling period W, and predict the angle estimation value at the ending time based on the Kalman filtering algorithm:

[0037]

[0038] In the formula, represents the angle estimation value at the ending time u; F represents the state transition matrix; represents the angle estimation value at the starting time u-1;

[0039] Predict the angular covariance matrix at the termination moment:

[0040] B u|u-1 = FB u-1|u-1 F T + Q

[0041] In the formula, B u|u-1 represents the angular covariance matrix of u at the termination moment; B u1|u-1 represents the angular covariance matrix of u - 1 at the starting moment; Q represents the process noise covariance matrix;

[0042] Obtain the actual angular value at the termination moment based on the three-dimensional angular change data, and calculate the measurement error at the termination moment:

[0043]

[0044] In the formula, e u represents the measurement error at the termination moment u; g u represents the actual angular value at the termination moment u; H represents the measurement matrix;

[0045] Calculate the Kalman gain at the termination moment:

[0046] K u = B u|u-1 H T (HB ulu-1 H T + R) -1

[0047] In the formula, K u represents the Kalman gain at the termination moment u; H T represents the transpose of the measurement matrix; R represents the measurement noise covariance matrix;

[0048] Update the angular estimated value at the termination moment:

[0049]

[0050] In the formula, represents the updated angular estimated value at the termination moment u;

[0051] Update the angular covariance matrix at the termination moment:

[0052] B u|u = (1 - K u H)B u|u-1

[0053] In the formula, B u|u represents the updated angular covariance matrix at the termination moment u.

[0054] Preferably, based on the angle estimation value at the starting moment u-1 and the updated angle estimation value at the ending moment u calculate the angular change rate in the x-axis direction:

[0055]

[0056] wherein, represents the angular change rate in the x-axis direction; represents the angle estimation value at the starting moment u-1; represents the updated angle estimation value at the ending moment u; W represents the sampling period;

[0057] The calculation processes of the angular change rates in the y and z axis directions are the same as above;

[0058] Summarize the angular change rates in the x, y, and z axis directions to obtain the angular change characteristics.

[0059] Preferably, generate a fusion input feature based on the motion impulse feature and the angular change feature and input it into the fall detection fusion model to output the fall detection result;

[0060] If the fall detection result indicates a fall event, trigger the fall alarm mechanism and send the fall detection result to the management terminal.

[0061] Preferably, further include, based on the adaptive threshold adjustment mechanism, automatically learn the real-time motion pattern of the user terminal according to the motion impulse feature and the angular change feature, and dynamically adjust the parameters of the motion impulse calculation model and the angular change analysis model based on the real-time motion pattern.

[0062] A personnel fall detection system integrating impulse and angle, the detection system includes:

[0063] A data acquisition module, configured to collect three-dimensional acceleration data and three-dimensional angular change data during the movement of the user terminal based on a wearable motion detection device;

[0064] An impulse extraction module, configured to construct a motion impulse calculation model and calculate motion impulse data according to the three-dimensional acceleration data, and extract the corresponding motion impulse features;

[0065] An angle acquisition module, configured to construct an angular change analysis model and obtain angular change features based on the three-dimensional angular change data;

[0066] A fall detection module, configured to construct a fall detection fusion model, generate a fall detection result according to the motion impulse feature and the angular change feature, and send it to the management terminal.

[0067] Compared with related technologies, a method for detecting human falls by integrating impulse and angle provided by the present invention has the following

[0068] Advantages:

[0069] The present invention can collect three-dimensional acceleration data and three-dimensional angle change data during the movement of the user terminal based on a wearable motion detection device; construct a motion impulse calculation model and calculate motion impulse data according to the three-dimensional acceleration data, and extract corresponding motion impulse characteristics; construct an angle change analysis model and obtain angle change characteristics based on the three-dimensional angle change data; construct a fall detection fusion model, generate a fall detection result according to the motion impulse characteristics and the angle change characteristics, and send it to the management terminal. Thus, by integrating multi-dimensional data, the accuracy and reliability of fall detection can be significantly improved, the false positive rate can be reduced, and strong protection can be provided for the safety of personnel.

[0070] The present invention uses a wearable motion detection device for data collection, without video monitoring, which fully protects the privacy of users and can be widely applied to various scenarios. The present invention can extract motion impulse characteristics and angle change characteristics, integrate multi-dimensional data of impulse and angle, and construct a fall detection fusion model for fall detection, thus effectively avoiding the detection limitations of a single data dimension and greatly improving the accuracy of fall detection. The present invention adopts an adaptive threshold adjustment mechanism to accurately distinguish falls from daily activities, automatically learn the real-time motion pattern of users, and dynamically adjust the model parameters, thereby reducing the false positive rate and significantly improving the accuracy and reliability of fall detection. The present invention is provided with an alarm mechanism, which can notify the management personnel in time and initiate subsequent rescue when a user fall event occurs, ensuring the life safety of users. Description of the Drawings

[0071] Figure 1 is a flowchart of a method for detecting human falls by integrating impulse and angle according to the present invention;

[0072] Figure 2 is a schematic diagram of the trigger of the fall alarm mechanism according to the present invention;

[0073] Figure 3 is a schematic diagram of the operation of the adaptive threshold adjustment mechanism according to the present invention;

[0074] Figure 4 is a system block diagram of a system for detecting human falls by integrating impulse and angle according to the present invention. Detailed Embodiments

[0075] The present invention will be further described below in conjunction with the drawings and embodiments.

[0076] Embodiment 1

[0077] As Figure 1As shown in the figure, a method for detecting human falls by integrating impulse and angle, the detection method comprising:

[0078] S1, collecting three-dimensional acceleration data and three-dimensional angle change data during the movement of the user end based on a wearable motion detection device;

[0079] Among them, the wearable motion detection device is integrated with a high-precision acceleration sensor and a gyroscope sensor, so as to accurately collect three-dimensional acceleration data and three-dimensional angle change data during the user's movement.

[0080] Specifically, the three-dimensional acceleration data is used to reflect the speed change rate of the human body in the x, y, and z axes, which is an important parameter for measuring the change of the human body's motion state; while the three-dimensional angle change data is used to reflect the rotation of the human body relative to the x, y, and z axes, which can intuitively show the change of the human body's posture.

[0081] S2, constructing a motion impulse calculation model and calculating motion impulse data according to the three-dimensional acceleration data, and extracting corresponding motion impulse characteristics;

[0082] Furthermore, a motion impulse calculation model can be constructed to perform integral operation on the collected three-dimensional acceleration data to accurately calculate the motion impulse data. Then, the impulse change rate, that is, the motion impulse characteristic, can be further extracted. The impulse change rate reflects the change rate of the impulse over time, which is used to judge whether there is an abnormal situation in the human body's motion state.

[0083] S3, constructing an angle change analysis model and obtaining angle change characteristics based on the three-dimensional angle change data;

[0084] It can be understood that an angle change analysis model can be constructed to obtain the angle change rate, that is, the angle change characteristic, based on the collected three-dimensional angle change data, so as to accurately judge the severity and trend of the change of the human body's posture.

[0085] S4, constructing a fall detection fusion model, generating a fall detection result according to the motion impulse characteristic and the angle change characteristic, and sending it to the management end.

[0086] Finally, the extracted motion impulse characteristics and angle change characteristics can be used as the input of the fall detection fusion model to generate a fall detection result and send it to the management end. The management end can be a server of a monitoring center, a terminal device of a caregiver, etc., so as to take corresponding measures in time.

[0087] In the specific implementation process, the wearable motion detection device is worn at the waist position of the user end;

[0088] The wearable motion detection device is equipped with an acceleration sensor and a gyroscope sensor. The acceleration sensor is used to collect the three-dimensional acceleration data, and the gyroscope sensor is used to collect the three-dimensional angular change data.

[0089] It should be noted that the wearable motion detection device is usually worn at the waist position of the user. Specifically, since the waist is near the center of gravity of the human body, it can comprehensively and effectively reflect the overall motion state of the human body. Therefore, compared with other parts, the data collected at the waist is less interfered by local limb movements and can more accurately present the comprehensive motion characteristics of the human body in various activities.

[0090] The wearable motion detection device integrates key acceleration sensors and gyroscope sensors. As a core component, the acceleration sensor can use micro-electromechanical system (MEMS) technology to collect three-dimensional acceleration data and determine the speed change rate and direction when the human body moves. The gyroscope sensor, on the other hand, accurately collects three-dimensional angular change data based on the principle of conservation of angular momentum to judge the change of the human body posture.

[0091] Building a motion impulse calculation model and calculating motion impulse data based on the three-dimensional acceleration data specifically includes:

[0092] Set a sampling period W and divide the sampling period W into n sub-periods, and the duration of each sub-period k is

[0093] Within the sub-period k, calculate the average acceleration values in the x, y, and z axes respectively:

[0094]

[0095] In the formula, respectively represent the average acceleration values in the x, y, and z axes within the sub-period k; Δt represents the duration of the sub-period k; a x (t), a y (t), a z (t) respectively represent the accelerations in the x, y, and z axes at time t;

[0096] Calculate the impulse increments in the x, y, and z axes within the sub-period k:

[0097]

[0098] In the formula, ΔI x，k 、ΔI y，k 、ΔI z，k respectively represent the impulse increments in the x, y, and z axes within the sub-period k;

[0099] Calculate the cumulative impulses in the x, y, and z axes within the sub - period k:

[0100] I x，k = I x，k -1+ΔI x，k

[0101] I y，k = I y，k-1 +ΔI y，k

[0102] I z，k = I z，k-1 +ΔI z，k

[0103] In the formula, I x，k 、I y，k 、I z，k respectively represent the cumulative impulses in the x, y, and z axes within the sub - period k; I x，k-1 、I y，k-1 、I z，k-1 respectively represent the cumulative impulses in the x, y, and z axes within the sub - period k - 1;

[0104] According to the cumulative impulses in the x, y, and z axes within the sub - period k, determine that the cumulative impulses in the x, y, and z axes within the sampling period W are respectively I x = I x，n 、I y = I y，n 、I z = I z，n , that is, the motion impulse data.

[0105] First, the sampling period can be set and divided into multiple sub - periods. Furthermore, the acceleration data can be finely processed to accurately capture the changes in acceleration in a short time, avoid losing key information due to too long a time interval, and improve the accuracy of impulse calculation.

[0106] Then, the average acceleration values of each axis can be calculated respectively within each sub - period, and further, the instantaneous fluctuation interference of the data is eliminated, accurately reflecting the motion state of the human body in different directions.

[0107] Furthermore, the impulse increments and cumulative impulses of each axis can be calculated based on the average acceleration values, so as to clearly show the cumulative process of impulses within each sub - period, which helps to deeply analyze the dynamic impulse changes during the human body movement process and provides more detailed and accurate impulse data for subsequent fall detection.

[0108] Finally, the cumulative impulse within the sampling period can be determined based on the cumulative impulse in each sub-period, and then the complete motion impulse data can be obtained. These data can comprehensively and accurately reflect the motion impulse of the human body within a sampling period, providing a reliable data basis for the fall detection of personnel by fusing impulse and angle, and effectively improving the accuracy and reliability of fall detection.

[0109] Extract the motion impulse features based on the motion impulse data, specifically including:

[0110] Calculate the differences in the cumulative impulses in the x, y, and z axes between two adjacent sampling periods W:

[0111]

[0112] In the formula, ΔI x 、ΔI y 、ΔI z respectively represent the differences in the cumulative impulses in the x, y, and z axes between two adjacent sampling periods W; I x 、I y 、I z respectively represent the cumulative impulses in the x, y, and z axes within the sampling period W; respectively represent the cumulative impulses in the x, y, and z axes within the next sampling period W;

[0113] Calculate the impulse change rates in the x, y, and z axes:

[0114]

[0115]

[0116] In the formula, respectively represent the impulse change rates in the x, y, and z axes, that is, the motion impulse features.

[0117] In practical applications, by calculating the differences in the cumulative impulses in each axis between two adjacent sampling periods, the dynamic changes of the impulse in a short time can be accurately captured.

[0118] In addition, by calculating the impulse change rate, the speed of impulse change can be further quantified, obtaining the intensity and change trend of human motion, facilitating the subsequent accurate identification of abnormal motion conditions such as falls, and effectively improving the accuracy of fall detection.

[0119] For the x axis, obtain the starting moment u - 1 and the ending moment u of the sampling period W, and predict the angle estimated value at the ending moment based on the Kalman filtering algorithm:

[0120]

[0121] Wherein, represents the estimated value of the angle at the termination time u; F represents the state transition matrix; represents the estimated value of the angle at the starting time u-1;

[0122] Predict the angle covariance matrix at the termination time:

[0123] B u|u-1 =FB u-1|u -1 F T+Q

[0124] Wherein, B u|u-1 represents the angle covariance matrix at the termination time u; B u|u-1 represents the angle covariance matrix at the starting time u-1; Q represents the process noise covariance matrix;

[0125] Obtain the actual value of the angle at the termination time based on the three-dimensional angle change data, and calculate the measurement error at the termination time:

[0126]

[0127] Wherein, e u represents the measurement error at the termination time u; g u represents the actual value of the angle at the termination time u; H represents the measurement matrix;

[0128] Calculate the Kalman gain at the termination time:

[0129] K u =B u|u-1 H T (HB u|u-1 H T +R) -1

[0130] Wherein, K u represents the Kalman gain at the termination time u; H T represents the transpose of the measurement matrix; R represents the measurement noise covariance matrix;

[0131] Update the estimated value of the angle at the termination time:

[0132]

[0133] Wherein, represents the updated estimated value of the angle at the termination time u;

[0134] Update the angle covariance matrix at the termination time:

[0135] B u|u =(1-K u H)B u|u-1 Wherein, Bu|u Represents the angular covariance matrix of the updated termination time u.

[0136] Based on the angular estimate at the starting time u-1 and the angular estimate at the updated termination time u Calculate the angular change rate in the x-axis direction:

[0137]

[0138] In the formula, represents the angular change rate in the x-axis direction; represents the angular estimate at the starting time u-1; represents the angular estimate at the updated termination time u; W represents the sampling period;

[0139] The calculation process of the angular change rates in the y and z axis directions is the same as above;

[0140] Summarize the angular change rates in the x, y, and z axis directions to obtain the angular change characteristics.

[0141] Among them, by predicting the angular estimate and the angular covariance matrix, the dynamic characteristics of the system and noise interference can be fully considered. After obtaining the actual angular value, the measurement error can be calculated and the angular estimate and covariance matrix can be updated to further correct the prediction result, making the finally obtained angular estimate closer to the true value.

[0142] Furthermore, based on the updated angular estimate, the angular change rate of each axis can be calculated, and thus the rotation speed of the human body in the three axes can be accurately reflected. The angular change rate is a key indicator for judging human body posture changes. Different motion states, such as normal walking, sudden turning, falling, etc., correspond to different angular change rates.

[0143] Finally, the angular change rates of the three axes can be summarized to obtain the angular change characteristics, providing comprehensive and detailed information for subsequent fall detection, helping to accurately identify abnormal motion conditions such as falls, and improving the accuracy and reliability of fall detection.

[0144] Generate a fusion input feature based on the motion impulse feature and the angular change feature and input it into the fall detection fusion model, and output the fall detection result;

[0145] If the fall detection result is that a fall event exists, trigger the fall alarm mechanism and send the fall detection result to the management terminal.

[0146] In practical applications, the motion impulse feature reflects the strength and trend of human motion, and the angle change feature reflects the change of human posture. By fusing the motion impulse feature and the angle change feature to generate input features and obtaining the detection result through the fall detection fusion model, the human motion state can be comprehensively and accurately analyzed, greatly improving the accuracy of fall detection.

[0147] Furthermore, as Figure 2 shown, once a fall event is detected, the fall alarm mechanism is immediately triggered and the result is sent to the management terminal, so as to ensure that relevant personnel can learn about the situation in a timely manner and take rescue measures in a timely manner, minimizing the personal injuries caused by falls, ensuring personal safety, and enhancing the reliability and practicality of fall detection.

[0148] As Figure 3 shown, a method for detecting human falls by fusing impulse and angle further includes, based on an adaptive threshold adjustment mechanism, automatically learning the real-time motion pattern of the user terminal according to the motion impulse feature and the angle change feature, and dynamically adjusting the parameters of the motion impulse calculation model and the angle change analysis model based on the real-time motion pattern.

[0149] It should be noted that there are differences in the motion habits and characteristics of different users, and the motion patterns of the same user also change at different time periods.

[0150] Through the adaptive threshold adjustment mechanism, the motion impulse feature and the angle change feature can be analyzed, the real-time motion pattern of the user terminal can be automatically learned, and the parameters of the motion impulse calculation model and the angle change analysis model can be dynamically adjusted, so as to better adapt to different users and different motion scenarios, avoid misjudgment caused by user individual differences or scenario changes, improve the accuracy and reliability of fall detection, and provide personalized and accurate safety protection for users.

[0151] Embodiment 2

[0152] As Figure 4 shown, a system for detecting human falls by fusing impulse and angle, the detection system includes:

[0153] A data acquisition module, configured to collect three-dimensional acceleration data and three-dimensional angle change data during the motion of the user terminal based on a wearable motion detection device;

[0154] An impulse extraction module, configured to construct a motion impulse calculation model and calculate motion impulse data according to the three-dimensional acceleration data, and extract corresponding motion impulse features;

[0155] An angle acquisition module, configured to construct an angle change analysis model and obtain angle change features based on the three-dimensional angle change data;

[0156] A fall detection module, configured to build a fall detection fusion model, generate a fall detection result according to the motion impulse feature and the angle change feature, and send the result to the management terminal.

[0157] Through the introduction of the above embodiments, the method for detecting a person's fall by fusing impulse and angle in the present invention can collect three-dimensional acceleration data and three-dimensional angle change data during the user's movement process based on a wearable motion detection device; build a motion impulse calculation model and calculate motion impulse data according to the three-dimensional acceleration data, and extract the corresponding motion impulse features; build an angle change analysis model and obtain angle change features based on the three-dimensional angle change data; build a fall detection fusion model, generate a fall detection result according to the motion impulse feature and the angle change feature, and send the result to the management terminal. Therefore, by fusing multi-dimensional data, the accuracy and reliability of fall detection can be significantly improved, the false positive rate can be reduced, and strong protection can be provided for the safety of personnel.

[0158] The present invention uses a wearable motion detection device for data collection, without video monitoring, which fully protects user privacy and can be widely applied to various scenarios. The present invention can extract motion impulse features and angle change features, fuse multi-dimensional data of impulse and angle, and build a fall detection fusion model for fall detection, thus effectively avoiding the detection limitations of a single data dimension and greatly improving the accuracy of fall detection. The present invention adopts an adaptive threshold adjustment mechanism to accurately distinguish falls from daily activities, automatically learn the user's real-time motion pattern, and dynamically adjust the model parameters, thereby reducing the false positive rate and significantly improving the accuracy and reliability of fall detection. The present invention is provided with an alarm mechanism, which can notify the management personnel in time and initiate subsequent rescue when a user fall event occurs, ensuring the life safety of the user.

[0159] This application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for realizing the functions specified in one Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0160] Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing relevant hardware through a program, and the program can be stored in a computer-readable storage medium. The storage medium includes read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc memories, magnetic disc memories, tape memories, or any other medium that can be used to carry or store data and is computer-readable.

[0161] It should also be noted that the term "including", "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, commodity or device. Without further limitation, an element defined by the statement "including one..." does not exclude the existence of another identical element in the process, method, commodity or device including the element.

Claims

1. A method for detecting human falls by integrating impulse and angle, characterized in that The detection method includes: Collecting three-dimensional acceleration data and three-dimensional angle change data during the movement of the user terminal based on a wearable motion detection device; Constructing a motion impulse calculation model and calculating motion impulse data according to the three-dimensional acceleration data, and extracting corresponding motion impulse characteristics; Constructing an angle change analysis model and obtaining angle change characteristics based on the three-dimensional angle change data; Constructing a fall detection fusion model, generating a fall detection result according to the motion impulse characteristics and the angle change characteristics, and sending it to the management terminal.

2. The method for detecting human falls by integrating impulse and angle according to claim 1, characterized in that, The wearable motion detection device is worn at the waist position of the user terminal; The wearable motion detection device is equipped with an acceleration sensor and a gyroscope sensor. The acceleration sensor is used to collect the three-dimensional acceleration data, and the gyroscope sensor is used to collect the three-dimensional angle change data.

3. A method for detecting human falls by integrating impulse and angle according to claim 1, characterized in that, The constructing of the motion impulse calculation model and calculating the motion impulse data according to the three-dimensional acceleration data specifically includes: Setting a sampling period W, and dividing the sampling period W into n sub-periods, and the duration of each sub-period k is Within the sub-period k, calculate the average acceleration values in the x, y, and z axes respectively: In the formula, respectively represent the average acceleration values in the x, y, and z axial directions within the sub-cycle k; Δt represents the duration of sub - period k; a x (t), a y (t), a z (t) represent the accelerations in the x, y, and z axes at time t, respectively; Calculate the impulse increments in the x, y, and z axes within the sub-period k: where, ΔI x,k , ΔI y,k , ΔI z,k respectively represent the impulse increments in the x, y, and z axial directions within the sub-cycle k; Calculate the cumulative impulses in the x, y, and z axes within the sub-period k: I x,k = I x,k-1 + ΔI x,k I y,k = I y,k-1 + ΔI y,k I z,k = I z,k-1 + ΔI z,k Wherein, I x，k , I y,k , I z,k respectively represent the cumulative impulses in the x, y, and z axial directions within the sub-period k; I x,k-1 , I y,k-1 , I z,k-1 respectively represent the cumulative impulses in the x, y, and z axial directions within the sub-period k-1; Determine that the cumulative impulses in the x, y, and z axial directions within the sampling period W are respectively I x = I x,n , I y = I y,n , I z = I z,n , that is, the motion impulse data.

4. A method for detecting human falls by integrating impulse and angle according to claim 3, characterized in that Extracting the motion impulse characteristics based on the motion impulse data specifically includes: Calculating the differences in the cumulative impulses in the x, y, and z axes within two adjacent sampling periods W; where ΔI x , ΔI y , and ΔI z respectively represent the differences in the cumulative impulses in the x, y, and z axes within two adjacent sampling periods W; I x , I y , and I z respectively represent the cumulative impulses in the x, y, and z axes within the sampling period W; respectively represent the accumulated impulses in the x, y, and z axial directions within the next sampling period W; Calculating the impulse change rates in the x, y, and z axes; In the formula, respectively represent the impulse change rates in the x, y, and z axial directions, that is, the motion impulse characteristics.

5. A method for detecting human falls by integrating impulse and angle according to claim 4, characterized in that, For the x axis, obtain the starting moment u-1 and the ending moment u of the sampling period W, and predict the angle estimated value at the ending moment based on the Kalman filtering algorithm: In the formula, represents the estimated value of the angle at the termination time u; F represents the state transition matrix; represents the angle estimation value of u-1 at the starting moment; Predict the angle covariance matrix at the ending moment: B u|u-1 = FB u-1|u-1 F T + Q where B u|u-1 represents the angular covariance matrix of u at the termination time; B u-1|u-1 represents the angular covariance matrix of u - 1 at the starting time; Q represents the process noise covariance matrix; Obtain the actual angle value at the ending moment based on the three-dimensional angle change data, and calculate the measurement error at the ending moment: where, e u represents the measurement error of u at the termination moment; g u represents the actual value of the angle of u at the termination moment; H represents the measurement matrix; Calculate the Kalman gain at the ending moment: K u = B u|u-1 H T (HB u|u-1 H T + R) -1 where K u represents the Kalman gain of u at the termination time; H T represents the transpose of the measurement matrix; R represents the measurement noise covariance matrix; Update the angle estimated value at the ending moment: wherein, represents the estimated value of the angle of the updated termination time u; Update the angle covariance matrix at the ending moment: R u|u = (1 - K u H)B u|u-1 where R u|u represents the angular covariance matrix of the updated termination time u.

6. A method for detecting human falls by integrating impulse and angle according to claim 5, characterized in that, Based on the angle estimation value at the starting moment u-1 and the updated angle estimation value at the ending moment u Calculate the angular rate of change in the x-axis direction: In the formula, represents the angular rate of change in the x-axis direction; represents the estimated angle value at the starting moment u - 1; represents the updated estimated angle value at the ending moment u; W represents the sampling period; The calculation process of the angle change rates in the y and z axes is the same as above; Summarize the angle change rates in the x, y, and z axes to obtain the angle change characteristics.

7. A method for detecting human falls by integrating impulse and angle according to claim 6, characterized in that Generate fusion input characteristics based on the motion impulse characteristics and the angle change characteristics and input them into the fall detection fusion model, and output the fall detection result; If the fall detection result indicates that a fall event exists, trigger a fall alarm mechanism and send the fall detection result to the management terminal.

8. A method for detecting human falls by integrating impulse and angle according to claim 1, characterized in that, It also includes, based on an adaptive threshold adjustment mechanism, automatically learning the real-time motion pattern of the user terminal according to the motion impulse characteristics and the angle change characteristics, and dynamically adjusting the parameters of the motion impulse calculation model and the angle change analysis model based on the real-time motion pattern.

9. A personnel fall detection system integrating impulse and angle, characterized in that, Applied to a personnel fall detection method that fuses impulse and angle as described in any one of claims 1-8, the detection system includes: A data acquisition module for collecting three-dimensional acceleration data and three-dimensional angle change data during the movement of the user terminal based on a wearable motion detection device; An impulse extraction module, which is used to build a motion impulse calculation model, calculate motion impulse data according to the three-dimensional acceleration data, and extract corresponding motion impulse features; An angle acquisition module, which is used to build an angle change analysis model and obtain angle change features based on the three-dimensional angle change data; A fall detection module, which is used to build a fall detection fusion model, generate a fall detection result according to the motion impulse features and the angle change features, and send it to the management terminal.

Images