A human fall detection method based on millimeter wave radar

Through the combination of Fourier transform and three-dimensional digital twin model, the movement characteristics of the knee and ankle joints are used to solve the misjudgment problem in the detection of human falls in millimeter wave radar, and the detection accuracy is improved. It is suitable for the elderly or disabled people. fall monitoring.

CN118806269BActive Publication Date: 2024-11-22HEBEI RUIJING ENERGY TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411259382.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-10
Publication Date
2024-11-22
Estimated Expiration
2044-09-10

AI Technical Summary

Technical Problem

The existing millimeter-wave radar human fall detection technology is prone to misjudgment when human body moves, resulting in low detection accuracy, especially in fall detection by the elderly or disabled.

Method used

The human body's movement characteristics were extracted through Fourier transform, and a three-dimensional digital twin model was constructed, combining the difference duration and height drop speed of the movement synchronization between the knee and ankle joints, and initially determining the risk of falling, and adjusting the number of training layers and radar signals of the movement characteristics The pulse repetition frequency improves detection accuracy.

Benefits of technology

It realizes rapid detection and accurate identification of human falls, reduces the rate of misjudgment, and improves the accuracy of millimeter-wave radar in fall detection of elderly people or disabled people.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118806269B_ABST
    Figure CN118806269B_ABST
Patent Text Reader

Abstract

The present invention relates to the field of millimeter wave radar technology, and in particular to a human fall detection method based on millimeter wave radar. The human fall detection method based on millimeter wave radar of the present invention can monitor the movement characteristics of a human body in real time through the millimeter wave radar, and can timely discover the emergency situation of falling through human fall detection. In an emergency, the millimeter wave radar can quickly detect the abnormal situation of human falling, and can determine the movement synchronization difference duration between the knee joint and the ankle joint through the movement characteristics of the human body. By determining whether the fall risk meets the risk rate standard, the occurrence of missed judgment is reduced, and the proportion of the number of errors in human fall recognition can reduce the occurrence of misjudgment, thereby improving the accuracy of human fall detection.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of millimeter wave radars, and in particular to a method for detecting a human fall based on millimeter wave radars. Background Art

[0002] For the elderly or disabled, falls can lead to a significant decline in their self-care ability and can also cause disability. As many as 22.6% of falls among the elderly occur in the bathroom and toilet at home. Since bathing and going to the toilet involve personal privacy, digital cameras cannot be installed. Radar fall detection devices solve this problem, but radar detection is prone to misjudgment when the human body is in motion. How to improve the accuracy of radar detection is an urgent problem to be solved.

[0003] The prior art publication number is CN116106855B, which provides a fall detection method and a fall detection device, by acquiring an echo signal collected by a millimeter-wave radar, wherein the echo signal is a continuous millimeter-wave signal modulated by the millimeter-wave radar to a target object in a target area, and then received by the millimeter-wave radar after being reflected by the target object. Based on a preset processing rule, the echo signal is converted and processed to obtain multiple frames of point cloud data in a time series, wherein the point cloud data includes multiple point cloud information, based on a preset tracking rule, the point cloud information in the multiple point cloud data is clustered and tracked to obtain multiple associated point clouds in a time series of at least one target object, based on a preset judgment rule, according to multiple associated points of the target object, a fall judgment result of the target object is determined, and when the fall judgment result indicates that the target object is in a fall posture, the converted data is input into a fall recognition network model, and a fall recognition result of the target object is output. However, the invention does not involve the problem of how to improve the accuracy of the fall detection device. Summary of the invention

[0004] To this end, the present invention provides a human fall detection method based on millimeter wave radar, so as to overcome the problem of low accuracy of human fall detection caused by false alarms due to misjudgment of target human body movement characteristics, including spatial position information of each joint of the human body at several moments and movement speed of each joint at several moments in the prior art.

[0005] To achieve the above object, the present invention provides a method for detecting a human fall based on a millimeter wave radar, comprising:

[0006] Step S1, starting the millimeter wave radar to transmit millimeter wave signals to the target detection area and receive echo signals returned from the target detection area;

[0007] Step S2, determining the position characteristics of the human body in the target detection area according to the echo signal;

[0008] Step S3, performing Fourier transform on the echo signal to extract the motion characteristics of the human body;

[0009] Step S4, determining the movement synchronization difference duration between the knee joint and the ankle joint according to the movement characteristics of the human body, so as to preliminarily determine the risk of falling;

[0010] Step S5, if it is preliminarily determined that the fall risk exists, a fall risk alarm is issued, and the number of recognition types of human body motion characteristics is adjusted;

[0011] Step S6, constructing a three-dimensional digital twin model of the human body according to the human body position characteristics and the human body motion characteristics, and performing data training on the human body motion characteristics to optimize the three-dimensional digital twin model;

[0012] Step S7, determining a human fall recognition result according to the height descent speed of the three-dimensional digital twin model;

[0013] Step S8, determining the accuracy of fall detection by the millimeter wave radar according to the error ratio of the fall recognition of the human body;

[0014] Step S9: If it is determined that the accuracy of the fall detection does not meet the requirements, the number of training layers of the human body motion characteristics is adjusted.

[0015] Or, the pulse repetition frequency of the millimeter wave radar is adjusted according to the attenuation amplitude of the radar signal.

[0016] Further, determining the position feature of a human body in the target detection area according to the echo signal includes:

[0017] Obtaining the time taken for the echo signal to be transmitted and returned;

[0018] Calculating the distance between each joint of the human body and the millimeter wave radar according to the time taken for the transmission to return;

[0019] The human body position feature includes the distance between each joint of the human body and the millimeter wave radar at each moment.

[0020] Furthermore, the movement characteristics of the human body include spatial position information of each joint at several moments and movement speed of each joint at several moments; wherein the spatial position information includes spatial coordinates of each joint at several moments.

[0021] Furthermore, the motion synchronization difference duration between the knee joint and the ankle joint is the absolute value of the difference in duration for the knee joint and the ankle joint to adjust from a previous motion state to a next motion state, respectively;

[0022] The previous motion state and the next motion state are adjacent motion states, and the motion state includes the moving direction, the spatial position and the instantaneous moving speed at the spatial position of the knee joint or the ankle joint.

[0023] Furthermore, the process of preliminarily determining the risk of falling according to the duration of the movement synchronization difference between the knee joint and the ankle joint includes:

[0024] Collect the duration of movement synchronization difference between the knee joint and the ankle joint;

[0025] Compare the motion synchronization difference duration between the knee joint and the ankle joint with the preset synchronization difference duration;

[0026] If the movement synchronization difference duration between the knee joint and the ankle joint is greater than the preset synchronization difference duration, it is preliminarily determined that the fall risk does not meet the requirements, and the number of recognition types of human body movement characteristics is increased.

[0027] Furthermore, the increase in the number of recognition types of human motion features is determined by the difference between the motion synchronization difference duration between the knee joint and the ankle joint and the preset synchronization difference duration.

[0028] Furthermore, the human body fall recognition result is determined based on the height descent speed of the three-dimensional digital twin model, including:

[0029] Obtaining a height descent speed of the three-dimensional digital twin model;

[0030] Comparing the height descent speed of the three-dimensional digital twin model with a preset height descent speed;

[0031] If the height descent speed of the three-dimensional digital twin model is greater than the preset height descent speed, it is determined that the human body has fallen;

[0032] Among them, the height descent speed is the ratio of the total height that the three-dimensional digital twin model descends in a single sampling cycle to the duration of the single sampling cycle.

[0033] Further, the process of determining the accuracy of fall detection by the millimeter wave radar according to the error ratio of the fall recognition includes:

[0034] Collecting the total number of fall recognitions and the number of errors in the fall recognitions;

[0035] Compare the percentage of the number of errors in fall recognition with the preset percentage of the number of errors;

[0036] If the error ratio of the fall recognition is greater than the preset error ratio, it is determined that the accuracy of the fall detection of the millimeter wave radar does not meet the requirements;

[0037] The error ratio of the fall recognition is the ratio of the error numbers of several fall recognition processes to the total number of fall recognitions.

[0038] Furthermore, adjusting the number of training layers of the human motion feature includes:

[0039] Comparing the error frequency ratio of the fall recognition with the preset first error frequency ratio and the preset second error frequency ratio respectively;

[0040] If the number of errors in the fall recognition is greater than the preset second error number ratio, increasing the number of training layers of the motion characteristics of the human body;

[0041] The number of training layers of the motion characteristics of the human body is the total number of trainings of the motion characteristics of the human body that meet the preset training conditions; the adjustment range of the number of training layers of the motion characteristics of the human body is determined by the difference between the error number ratio of fall recognition and the preset second error number ratio;

[0042] Among them, the preset training condition is that the movement characteristics of the human body are mapped at least once through data training; for a single mapping, the subsequent data training process of generating mapping selects the data in the last generated mapping to continue training and form a new mapping based on the last generated mapping.

[0043] Further, the process of adjusting the pulse repetition frequency of the millimeter wave radar according to the attenuation amplitude of the radar signal includes:

[0044] If the error number ratio of the fall recognition is greater than the preset first error number ratio and less than or equal to the preset second error number ratio, it is preliminarily determined that the degree of environmental interference to the millimeter-wave radar does not meet the requirements, and the signal strength of the millimeter-wave signal and the signal strength of the echo signal are collected;

[0045] Calculate the radar signal attenuation amplitude according to the signal strength of the millimeter wave signal and the signal strength of the echo signal;

[0046] Compare the radar signal attenuation amplitude with the preset attenuation amplitude;

[0047] If the attenuation amplitude of the radar signal is greater than the preset attenuation amplitude, it is determined that the degree of environmental interference to the millimeter-wave radar does not meet the requirements, and the pulse repetition frequency of the millimeter-wave radar is increased;

[0048] The adjustment amplitude of the pulse repetition frequency of the millimeter wave radar is determined by the difference between the radar signal attenuation amplitude and the preset attenuation amplitude.

[0049] Compared with the prior art, the beneficial effects of the present invention lie in that the present invention can monitor the motion characteristics of the human body in real time through the millimeter-wave radar, the millimeter-wave radar can quickly detect abnormal human falls, and the duration of the motion synchronization difference between the knee joint and the ankle joint and the preliminarily judge the risk of falling can be determined through the motion characteristics of the human body, and the accuracy of fall detection of the millimeter-wave radar is determined by the proportion of the number of errors in human fall recognition, and by adjusting the number of training layers of the human body's motion characteristics, the influence of the decrease in the accuracy of fall detection caused by the insufficient accuracy of the training of the human body's motion characteristics is reduced; by adjusting the pulse repetition frequency of the millimeter-wave radar according to the attenuation amplitude of the radar signal, the influence of the excessive attenuation of the intensity of the millimeter-wave radar signal caused by the excessive humidity in the surrounding detection environment of the millimeter-wave radar is reduced, thereby improving the accuracy of human fall detection.

[0050] Furthermore, the human motion feature detection by millimeter-wave radar can monitor and analyze the daily activities of the elderly in real time, detect fall emergencies in time, and improve the accuracy of fall detection.

[0051] Furthermore, by collecting the difference in duration of movement synchronization between the knee joint and the ankle joint, the risk of falling can be accurately analyzed and determined, thereby improving the accuracy of millimeter-wave radar human fall detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 This is a flowchart of the steps of a method for detecting a human fall based on millimeter wave radar according to an embodiment of the present invention;

[0053] Figure 2 A point cloud image of a target detection area of ​​a human fall detection method based on millimeter wave radar according to an embodiment of the present invention;

[0054] Figure 3 A point cloud connection diagram of a target detection area of ​​a human fall detection method based on millimeter wave radar according to an embodiment of the present invention;

[0055] Figure 4 This is a flow chart of a preliminary determination of fall risk based on the duration of the movement synchronization difference between the knee joint and the ankle joint according to an embodiment of the present invention. DETAILED DESCRIPTION

[0056] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0057] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the protection scope of the present invention.

[0058] It should be noted that, in the description of the present invention, terms such as "up", "down", "left", "right", "inside", "outside" and the like indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings. This is merely for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it should not be understood as a limitation on the present invention.

[0059] In addition, it should be noted that in the description of the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0060] See also Figures 1 to 4 As shown, they are respectively a flow chart of steps of a human fall detection method based on millimeter wave radar according to an embodiment of the present invention, a point cloud diagram of a target detection area, a point cloud connection diagram of a target detection area, and a determination flow chart of a preliminary determination of a fall risk according to a motion synchronization difference duration between a knee joint and an ankle joint according to an embodiment of the present invention; the present invention provides a human fall detection method based on millimeter wave radar, comprising:

[0061] Step S1, starting the millimeter wave radar to transmit millimeter wave signals to the target detection area and receive echo signals returned from the target detection area;

[0062] Step S2, determining the position characteristics of the human body in the target detection area according to the echo signal;

[0063] Step S3, performing Fourier transform on the echo signal to extract the motion characteristics of the human body;

[0064] Step S4, determining the movement synchronization difference duration between the knee joint and the ankle joint according to the movement characteristics of the human body, so as to preliminarily determine the risk of falling;

[0065] Step S5, if it is preliminarily determined that the fall risk exists, a fall risk alarm is issued, and the number of recognition types of human body motion characteristics is adjusted;

[0066] Step S6, constructing a three-dimensional digital twin model of the human body according to the human body position characteristics and the human body motion characteristics, and performing data training on the human body motion characteristics to optimize the three-dimensional digital twin model;

[0067] Step S7, determining a human fall recognition result according to the height descent speed of the three-dimensional digital twin model;

[0068] Step S8, determining the accuracy of fall detection by the millimeter wave radar according to the error ratio of the fall recognition of the human body;

[0069] Step S9: If it is determined that the accuracy of the fall detection does not meet the requirements, the number of training layers of the human body motion characteristics is adjusted.

[0070] Or, the pulse repetition frequency of the millimeter wave radar is adjusted according to the attenuation amplitude of the radar signal.

[0071] In implementation, the application scenario of the embodiment of the present invention is the scenario of preventing the elderly from falling in the bathroom or toilet. The present invention can reduce the probability of misjudgment of fall detection due to different degrees of bending of different parts of the human body posture by dividing the object to be detected into several areas. By real-time monitoring the posture of the object to be detected during human movement, the difference between the posture of the human body during movement and the fall action can be accurately identified, thereby improving the accuracy of fall detection. By performing data training on the movement characteristics of the human body, the three-dimensional digital twin model of the human body can be updated or optimized, thereby improving the accuracy of fall detection.

[0072] Specifically, the process of constructing a three-dimensional digital twin model of the human body based on the position characteristics and motion characteristics of the human body is as follows: filtering, denoising and other processing are performed on the collected raw data to improve data quality and ensure the accuracy of subsequent processing; key feature points of the human body, such as head, shoulders, elbows, hands, waist, knees and feet, are extracted from the preprocessed data; a three-dimensional skeleton model of the human body is constructed using the extracted feature points in combination with the principles of human kinematics. This step requires the use of three-dimensional reconstruction technology in computer vision and graphics; based on the three-dimensional skeleton, a human body surface model is generated through an interpolation algorithm or a deep learning-based method to form a more complete three-dimensional human body model; the action, position and movement status of the three-dimensional digital twin model are updated in real time according to the motion characteristics of the human body.

[0073] Specifically, three-dimensional reconstruction technology, interpolation algorithm and computer vision are conventional technical means well known to those skilled in the art, and the above process will not be described in detail.

[0074] Specifically, in step S6, training the motion characteristics of the human body to construct a three-dimensional digital twin model of the human body includes continuously collecting human motion data, including data under different motion postures, actions and environmental conditions, and collecting data through millimeter wave radar. Then, the collected data is denoised, cleaned and aligned for preprocessing, and then, a convolutional neural network deep learning model is used for feature extraction; finally, a three-dimensional digital twin model of the human body is constructed based on the extracted human motion feature data.

[0075] Specifically, the pulse repetition frequency of the millimeter-wave radar is the number of times the millimeter-wave radar transmits pulses per unit time.

[0076] Preferably, the best application scenario of the human fall detection method based on millimeter wave radar is to detect the movement status of people in the bathroom or toilet; millimeter wave radar can penetrate non-metallic materials such as clothes, and combine the Fourier transform and Doppler effect of the echo signal to obtain the movement speed, acceleration, direction and trajectory of the elderly.

[0077] Specifically, obtaining the movement speed of each joint point of the human body based on Fourier transform and Doppler effect is a conventional technical means well known to those skilled in the art, and the specific process of the conversion process and Doppler effect and Fourier transform will not be repeated here.

[0078] Specifically, the joints include shoulder, elbow, wrist, hip, knee and ankle joints.

[0079] In implementation, the present invention can monitor the movement characteristics of the human body in real time through the millimeter-wave radar. The millimeter-wave radar can quickly detect abnormal situations where the human body falls. The movement characteristics of the human body can determine the duration of the movement synchronization difference between the knee joint and the ankle joint and preliminarily determine the risk of falling. The accuracy of the millimeter-wave radar's fall detection is determined by the proportion of the number of errors in the human body's fall recognition. By adjusting the number of training layers of the human body's motion characteristics, the influence of the decrease in the accuracy of fall detection due to the insufficient accuracy of the training of the human body's motion characteristics is reduced. By adjusting the pulse repetition frequency of the millimeter-wave radar according to the attenuation amplitude of the radar signal, the influence of the excessive attenuation of the millimeter-wave radar signal intensity due to the excessive humidity in the surrounding detection environment of the millimeter-wave radar is reduced, thereby improving the accuracy of human fall detection.

[0080] Specifically, determining the human body position feature within the target detection area according to the echo signal includes:

[0081] Obtaining the time taken for the echo signal to be transmitted and returned;

[0082] Calculating the distance between each joint of the human body and the millimeter wave radar according to the time taken for the transmission to return;

[0083] The human body position feature includes the distance between each joint of the human body and the millimeter wave radar at each moment.

[0084] Specifically, the movement characteristics of the human body include the spatial position information of each joint at several moments and the moving speed of each joint at several moments; wherein the spatial position information includes the spatial coordinates of each joint at several moments.

[0085] Specifically, the motion synchronization difference duration between the knee joint and the ankle joint is the absolute value of the difference in duration for the knee joint and the ankle joint to adjust from a previous motion state to a next motion state, respectively;

[0086] The previous motion state and the next motion state are adjacent motion states, and the motion state includes the moving direction, the spatial position and the instantaneous moving speed at the spatial position of the knee joint or the ankle joint.

[0087] Specifically, the process of preliminarily determining the risk of falling based on the duration of the movement synchronization difference between the knee joint and the ankle joint includes:

[0088] Collect the duration of movement synchronization difference between the knee joint and the ankle joint;

[0089] Compare the motion synchronization difference duration between the knee joint and the ankle joint with the preset synchronization difference duration;

[0090] If the movement synchronization difference duration between the knee joint and the ankle joint is less than or equal to the preset synchronization difference duration, it is preliminarily determined that the fall risk meets the requirements;

[0091] If the movement synchronization difference duration between the knee joint and the ankle joint is greater than the preset synchronization difference duration, it is preliminarily determined that the fall risk does not meet the requirements, and the number of recognition types of human body movement characteristics is increased.

[0092] Specifically, the types of recognition of human body movement features include leaning, running, lying down, bending over, raising the knees, and stretching.

[0093] In the implementation, the preset synchronization difference duration is 0.020 seconds. In the implementation, the spatial coordinates of each joint are point cloud data. The millimeter wave radar detects objects in the surrounding environment by sending and receiving radar waves, and then converts these data into point cloud data, including the position coordinate information of the target in three-dimensional space.

[0094] Specifically, the increase in the number of recognition types of human motion features is determined by the difference between the motion synchronization difference duration between the knee joint and the ankle joint and the preset synchronization difference duration.

[0095] If the difference between the motion synchronization difference duration between the knee joint and the ankle joint and the preset synchronization difference duration is less than or equal to the preset synchronization difference duration difference, the number of recognition types of the motion characteristics of the human body is adjusted using the first quantity adjustment coefficient;

[0096] If the difference between the motion synchronization difference duration between the knee joint and the ankle joint and the preset synchronization difference duration is greater than the preset synchronization difference duration difference, the second quantity adjustment coefficient is used to adjust the number of recognition types of human motion characteristics.

[0097] In implementation, the number of recognition types of human motion features is adjusted based on the initial number standard 3 of the recognition types of human motion features. The adjustment formula for the number of recognition types of human motion features is F=f+β n ; F is the number of recognition types of human motion features after adjustment, f is the number of recognition types of human motion features before adjustment, β n is the nth quantity adjustment coefficient, β1=1; β2=2; the preset synchronization difference duration difference is 0.001 seconds; when the motion synchronization difference duration between the knee joint and the ankle joint is 0.022 seconds, the preset synchronization difference duration is 0.020 seconds, and the difference between the motion synchronization difference duration between the knee joint and the ankle joint and the preset synchronization difference duration is 0.02 seconds, the number of recognition types of human motion features F=f+β n =3+2=5.

[0098] Specifically, the human body fall recognition results determined based on the height descent speed of the three-dimensional digital twin model include:

[0099] Obtaining a height descent speed of the three-dimensional digital twin model;

[0100] Comparing the height descent speed of the three-dimensional digital twin model with a preset height descent speed;

[0101] If the height descent speed of the three-dimensional digital twin model is greater than the preset height descent speed, it is determined that the human body has fallen;

[0102] Among them, the height descent speed is the ratio of the total height that the three-dimensional digital twin model descends in a single sampling cycle to the duration of the single sampling cycle.

[0103] Specifically, the process of determining the accuracy of fall detection by the millimeter wave radar according to the error ratio of the fall recognition includes:

[0104] Collecting the total number of fall recognitions and the number of errors in the fall recognitions;

[0105] Comparing the error ratio of fall recognition with the preset first error ratio;

[0106] If the error number ratio of the fall recognition is greater than the preset first error number ratio, it is determined that the accuracy of the fall detection of the millimeter wave radar does not meet the requirements;

[0107] If the error number ratio of the fall recognition is greater than the preset first error number ratio, it is determined that the accuracy of the fall detection of the millimeter wave radar meets the requirements;

[0108] The error ratio of the fall recognition is the ratio of the error numbers of several fall recognition processes to the total number of fall recognitions.

[0109] Specifically, the errors in fall recognition include incorrect position of the recognized human body, incorrect determination of whether the human body has fallen, and incorrect identification of the person who has fallen.

[0110] Specifically, adjusting the number of training layers of the human motion feature includes:

[0111] Comparing the error frequency ratio of the fall recognition with the preset first error frequency ratio and the preset second error frequency ratio respectively;

[0112] If the number of errors in the fall recognition is greater than the preset second error number ratio, increasing the number of training layers of the motion characteristics of the human body;

[0113] The number of training layers of the motion characteristics of the human body is the total number of trainings of the motion characteristics of the human body that meet the preset training conditions; the adjustment range of the number of training layers of the motion characteristics of the human body is determined by the difference between the error number ratio of fall recognition and the preset second error number ratio;

[0114] Among them, the preset training condition is that the movement characteristics of the human body are mapped at least once through data training; for a single mapping, the subsequent data training process of generating mapping selects the data in the last generated mapping to continue training and form a new mapping based on the last generated mapping.

[0115] If the difference between the error frequency ratio of the fall recognition and the preset second error frequency ratio is less than or equal to the preset difference, the training layers of the human body motion characteristics are adjusted using the first layer adjustment coefficient;

[0116] If the difference between the error frequency ratio of the fall recognition and the preset second error frequency ratio is greater than the preset difference, the training layers of the human body motion characteristics are adjusted using the second layer adjustment coefficient;

[0117] In the implementation, the preset first error times account for 2%, the preset second error times account for 5%; the preset difference is 0.4%, the first layer adjustment coefficient is 1.5, and the second layer adjustment coefficient is 2; the training layer adjustment formula of the human body's motion characteristics is M=m×α n ; Where M is the number of training layers of the adjusted human motion features, m is the number of training layers of the current human motion features, and the current number of training layers of the human motion features is 8, α n is the adjustment coefficient of the nth layer. When the error rate of fall recognition accounts for 5.6%, the training layer number of human motion features is M=m×α n =8×2=16 layers, that is, the number of training layers for human motion features increases from 8 layers to 16 layers.

[0118] Specifically, the process of adjusting the pulse repetition frequency of the millimeter wave radar according to the attenuation amplitude of the radar signal includes:

[0119] If the error number ratio of the fall recognition is greater than the preset first error number ratio and less than or equal to the preset second error number ratio, it is preliminarily determined that the degree of environmental interference to the millimeter-wave radar does not meet the requirements, and the signal strength of the millimeter-wave signal and the signal strength of the echo signal are collected;

[0120] Calculate the radar signal attenuation amplitude according to the signal strength of the millimeter wave signal and the signal strength of the echo signal;

[0121] Compare the radar signal attenuation amplitude with the preset attenuation amplitude;

[0122] If the attenuation amplitude of the radar signal is greater than the preset attenuation amplitude, it is determined that the degree of environmental interference to the millimeter-wave radar does not meet the requirements, and the pulse repetition frequency of the millimeter-wave radar is increased;

[0123] The adjustment amplitude of the pulse repetition frequency of the millimeter wave radar is determined by the difference between the radar signal attenuation amplitude and the preset attenuation amplitude.

[0124] Specifically, the pulse repetition frequency of millimeter-wave radar mainly serves to increase the average power of the radar, improve the output signal-to-noise ratio of the radar, and optimize the performance of the radar system. By increasing the transmission pulse repetition frequency, more echoes can be obtained during the dwell time of the radar illuminating the target under the condition of a certain antenna rotation speed; by accumulating these echoes, the output signal-to-noise ratio of the radar can be improved, thereby improving the detection performance of the radar.

[0125] Specifically, the process of increasing the pulse repetition frequency of the millimeter wave radar is:

[0126] If the difference between the radar signal attenuation amplitude and the preset attenuation amplitude is less than or equal to the preset radar signal attenuation amplitude difference, adjusting the pulse repetition frequency of the millimeter wave radar using the first frequency adjustment coefficient;

[0127] If the difference between the radar signal attenuation amplitude and the preset attenuation amplitude is greater than the preset radar signal attenuation amplitude difference, the second frequency adjustment coefficient is used to adjust the pulse repetition frequency of the millimeter wave radar.

[0128] In practice, the pulse repetition frequency of the current millimeter-wave radar is 1000 Hz, the preset radar signal attenuation amplitude difference is 10 decibels, the first frequency adjustment coefficient is 1.16; the second frequency adjustment coefficient is 1.22; the pulse repetition frequency adjustment formula of the millimeter-wave radar is G=g×γ n , where G is the increased pulse repetition frequency of the millimeter-wave radar; g is the current pulse repetition frequency of the millimeter-wave radar. The preset radar signal attenuation amplitude is 80 decibels. When the radar signal attenuation amplitude is 100 decibels, the increased pulse repetition frequency of the millimeter-wave radar is G=g×γ n =1000×1.22=1220 Hz.

[0129] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

Claims

1. A method for detecting human falls based on millimeter wave radar, characterized in that: include: Step S1, starting the millimeter wave radar to transmit millimeter wave signals to the target detection area and receive echo signals returned from the target detection area; Step S2, determining the position characteristics of the human body in the target detection area according to the echo signal; Step S3, performing Fourier transform on the echo signal to extract the motion characteristics of the human body; Step S4, determining the movement synchronization difference duration between the knee joint and the ankle joint according to the movement characteristics of the human body, so as to preliminarily determine the risk of falling; Step S5, if it is preliminarily determined that the fall risk exists, a fall risk alarm is issued, and the number of recognition types of human body motion characteristics is adjusted; Step S6, constructing a three-dimensional digital twin model of the human body according to the human body position characteristics and the human body motion characteristics, and performing data training on the human body motion characteristics to optimize the three-dimensional digital twin model; Step S7, determining a human fall recognition result according to the height descent speed of the three-dimensional digital twin model; Step S8, determining the accuracy of fall detection by the millimeter wave radar according to the error ratio of the fall recognition of the human body; Step S9: If it is determined that the accuracy of the fall detection does not meet the requirements, the number of training layers of the human body motion characteristics is adjusted. Or, the pulse repetition frequency of the millimeter wave radar is adjusted according to the attenuation amplitude of the radar signal.

2. The method for detecting human fall based on millimeter wave radar according to claim 1, characterized in that: Determining the position characteristics of a human body within the target detection area according to the echo signal includes: Obtaining the time taken for the echo signal to be transmitted and returned; Calculating the distance between each joint of the human body and the millimeter wave radar according to the time taken for the transmission to return; The human body position feature includes the distance between each joint of the human body and the millimeter wave radar at each moment.

3. The method for detecting human fall based on millimeter wave radar according to claim 2, characterized in that: The movement characteristics of the human body include the spatial position information of each joint at a number of moments and the moving speed of each joint at a number of moments; wherein the spatial position information includes the spatial coordinates of each joint at a number of moments.

4. The method for detecting human fall based on millimeter wave radar according to claim 3, characterized in that: The motion synchronization difference duration between the knee joint and the ankle joint is the absolute value of the difference in duration used for the knee joint and the ankle joint to adjust from a previous motion state to a next motion state respectively; The previous motion state and the next motion state are adjacent motion states, and the motion state includes the moving direction, the spatial position and the instantaneous moving speed at the spatial position of the knee joint or the ankle joint.

5. The method for detecting human fall based on millimeter wave radar according to claim 4, characterized in that: The process of initially determining the risk of falling based on the duration of the movement synchronization difference between the knee and ankle joints includes: Collect the duration of movement synchronization difference between the knee joint and the ankle joint; Compare the motion synchronization difference duration between the knee joint and the ankle joint with the preset synchronization difference duration; If the movement synchronization difference duration between the knee joint and the ankle joint is greater than the preset synchronization difference duration, it is preliminarily determined that the fall risk does not meet the requirements, and the number of recognition types of human body movement characteristics is increased.

6. The method for detecting human fall based on millimeter wave radar according to claim 5, characterized in that: The increase in the number of recognition types of human motion features is determined by the difference between the motion synchronization difference duration between the knee joint and the ankle joint and the preset synchronization difference duration.

7. The method for detecting human fall based on millimeter wave radar according to claim 6, characterized in that: The human body fall recognition results determined based on the height descent speed of the three-dimensional digital twin model include: Obtaining a height descent speed of the three-dimensional digital twin model; Comparing the height descent speed of the three-dimensional digital twin model with a preset height descent speed; If the height descent speed of the three-dimensional digital twin model is greater than the preset height descent speed, it is determined that the human body has fallen; Among them, the height descent speed is the ratio of the total height that the three-dimensional digital twin model descends in a single sampling cycle to the duration of the single sampling cycle.

8. The method for detecting human fall based on millimeter wave radar according to claim 7, characterized in that: The process of determining the accuracy of fall detection by the millimeter wave radar according to the error ratio of the fall recognition includes: Collecting the total number of fall recognitions and the number of errors in the fall recognitions; Comparing the error ratio of fall recognition with the preset first error ratio; If the error number ratio of the fall recognition is greater than the preset first error number ratio, it is determined that the accuracy of the fall detection of the millimeter wave radar does not meet the requirements; If the error number ratio of the fall recognition is greater than the preset first error number ratio, it is determined that the accuracy of the fall detection of the millimeter wave radar meets the requirements; The error ratio of the fall recognition is the ratio of the error numbers of several fall recognition processes to the total number of fall recognitions.

9. The method for detecting human fall based on millimeter wave radar according to claim 8, characterized in that: Adjusting the number of training layers of the human motion feature includes: Comparing the error frequency ratio of the fall recognition with the preset first error frequency ratio and the preset second error frequency ratio respectively; If the number of errors in the fall recognition is greater than the preset second error number ratio, increasing the number of training layers of the motion characteristics of the human body; The number of training layers of the motion characteristics of the human body is the total number of trainings of the motion characteristics of the human body that meet the preset training conditions; the adjustment range of the number of training layers of the motion characteristics of the human body is determined by the difference between the error number ratio of fall recognition and the preset second error number ratio; Among them, the preset training condition is that the movement characteristics of the human body are mapped at least once through data training; for a single mapping, the subsequent data training process of generating mapping selects the data in the last generated mapping to continue training and form a new mapping based on the last generated mapping.

10. The method for detecting human fall based on millimeter wave radar according to claim 9, characterized in that: The process of adjusting the pulse repetition frequency of the millimeter wave radar according to the attenuation amplitude of the radar signal includes: If the error number ratio of the fall recognition is greater than the preset first error number ratio and less than or equal to the preset second error number ratio, it is preliminarily determined that the degree of environmental interference to the millimeter-wave radar does not meet the requirements, and the signal strength of the millimeter-wave signal and the signal strength of the echo signal are collected; Calculate the radar signal attenuation amplitude according to the signal strength of the millimeter wave signal and the signal strength of the echo signal; Compare the radar signal attenuation amplitude with the preset attenuation amplitude; If the attenuation amplitude of the radar signal is greater than the preset attenuation amplitude, it is determined that the degree of environmental interference to the millimeter-wave radar does not meet the requirements, and the pulse repetition frequency of the millimeter-wave radar is increased; The adjustment amplitude of the pulse repetition frequency of the millimeter wave radar is determined by the difference between the radar signal attenuation amplitude and the preset attenuation amplitude.