CN116849643A - Method for detecting falling of wearable equipment based on neural network - Google Patents
Method for detecting falling of wearable equipment based on neural network Download PDFInfo
- Publication number
- CN116849643A CN116849643A CN202310750679.5A CN202310750679A CN116849643A CN 116849643 A CN116849643 A CN 116849643A CN 202310750679 A CN202310750679 A CN 202310750679A CN 116849643 A CN116849643 A CN 116849643A
- Authority
- CN
- China
- Prior art keywords
- data
- neural network
- user
- module
- frequency domain
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 54
- 238000000034 method Methods 0.000 title claims abstract description 33
- 230000033001 locomotion Effects 0.000 claims abstract description 50
- 230000006399 behavior Effects 0.000 claims abstract description 40
- 238000001514 detection method Methods 0.000 claims abstract description 38
- 238000004458 analytical method Methods 0.000 claims abstract description 34
- 238000003062 neural network model Methods 0.000 claims abstract description 21
- 230000001133 acceleration Effects 0.000 claims abstract description 17
- 238000007781 pre-processing Methods 0.000 claims abstract description 6
- 230000003993 interaction Effects 0.000 claims description 16
- 230000036541 health Effects 0.000 claims description 15
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 12
- 230000006870 function Effects 0.000 claims description 12
- 238000007405 data analysis Methods 0.000 claims description 11
- 230000009467 reduction Effects 0.000 claims description 8
- 238000012549 training Methods 0.000 claims description 8
- 238000013527 convolutional neural network Methods 0.000 claims description 7
- 238000011156 evaluation Methods 0.000 claims description 6
- 230000003862 health status Effects 0.000 claims description 6
- 238000012952 Resampling Methods 0.000 claims description 5
- 230000008859 change Effects 0.000 claims description 5
- 230000002708 enhancing effect Effects 0.000 claims description 4
- 230000008569 process Effects 0.000 description 5
- 230000005540 biological transmission Effects 0.000 description 4
- 230000001965 increasing effect Effects 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 238000004140 cleaning Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 238000013500 data storage Methods 0.000 description 2
- 238000007726 management method Methods 0.000 description 2
- 230000008447 perception Effects 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 208000017667 Chronic Disease Diseases 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000003012 network analysis Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B21/00—Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
- G08B21/02—Alarms for ensuring the safety of persons
- G08B21/04—Alarms for ensuring the safety of persons responsive to non-activity, e.g. of elderly persons
- G08B21/0407—Alarms for ensuring the safety of persons responsive to non-activity, e.g. of elderly persons based on behaviour analysis
- G08B21/043—Alarms for ensuring the safety of persons responsive to non-activity, e.g. of elderly persons based on behaviour analysis detecting an emergency event, e.g. a fall
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/103—Detecting, measuring or recording devices for testing the shape, pattern, colour, size or movement of the body or parts thereof, for diagnostic purposes
- A61B5/11—Measuring movement of the entire body or parts thereof, e.g. head or hand tremor, mobility of a limb
- A61B5/1112—Global tracking of patients, e.g. by using GPS
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/103—Detecting, measuring or recording devices for testing the shape, pattern, colour, size or movement of the body or parts thereof, for diagnostic purposes
- A61B5/11—Measuring movement of the entire body or parts thereof, e.g. head or hand tremor, mobility of a limb
- A61B5/1116—Determining posture transitions
- A61B5/1117—Fall detection
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/68—Arrangements of detecting, measuring or recording means, e.g. sensors, in relation to patient
- A61B5/6801—Arrangements of detecting, measuring or recording means, e.g. sensors, in relation to patient specially adapted to be attached to or worn on the body surface
- A61B5/6802—Sensor mounted on worn items
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/74—Details of notification to user or communication with user or patient ; user input means
- A61B5/746—Alarms related to a physiological condition, e.g. details of setting alarm thresholds or avoiding false alarms
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B21/00—Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
- G08B21/02—Alarms for ensuring the safety of persons
- G08B21/04—Alarms for ensuring the safety of persons responsive to non-activity, e.g. of elderly persons
- G08B21/0438—Sensor means for detecting
- G08B21/0446—Sensor means for detecting worn on the body to detect changes of posture, e.g. a fall, inclination, acceleration, gait
Abstract
The invention discloses a method for detecting falling of wearable equipment based on a neural network, and relates to the technical field of gesture recognition; the method comprises the steps of deploying a fall detection system to a neural network-based wearable device, wherein the fall detection system comprises a sensor module, an analysis module, a storage module and an alarm module, collecting acceleration, angular velocity and geographic position information through the sensor module respectively, comprehensively acquiring user movement behavior information, preprocessing the user movement behavior information through the analysis module, inputting preprocessed data features into a neural network model to classify according to fall behaviors and different fall states, wherein the classified movement states comprise normal states, fall events and lying events, storing generated data and collected information through the storage module, judging the movement states of a user to be the normal states, the fall events or the lying events through the alarm module according to movement state classification and identification of the analysis module, and sending early warning information to the user and emergency contact staff if the fall events occur.
Description
Technical Field
The invention discloses a method, relates to the technical field of gesture recognition, and particularly relates to a method for detecting falling of wearable equipment based on a neural network.
Background
As the population ages and chronic diseases increase, wearable devices become an increasingly popular way of health monitoring. However, the existing wearable device can only provide simple steps, heart rate and other data, fall detection only depends on devices such as a pressure sensor or a camera to identify gestures and video images, and the problems of easy false alarm or missing report, inaccurate data, environment limitation, portability, high user cost and the like exist.
Disclosure of Invention
Aiming at the problems in the prior art, the invention provides a method for detecting falling of wearable equipment based on a neural network, which is used for analyzing and judging the data by collecting the motion data of a user and using a neural network model, so as to realize the timely identification and processing of falling events and improve the safety and health condition of the user. Meanwhile, the invention also provides various alarm modes and data analysis services so as to better ensure the safety and health of users.
The specific scheme provided by the invention is as follows:
the invention provides a method for detecting falling of wearable equipment based on a neural network, which comprises deploying a falling detection system to the wearable equipment based on the neural network, wherein the falling detection system comprises a sensor module, an analysis module, a storage module and an alarm module,
the acceleration, angular velocity and geographic position information are respectively collected through the sensor module, the movement behavior information of the user is comprehensively obtained,
preprocessing user movement behavior information through an analysis module, forming a data set by the preprocessed user movement behavior information, adding a data interaction item operation to the data set, then performing data frequency domain expansion, firstly performing digitize boxing operation to the data set in the data interaction item operation to expand the data quantity, stacking the original data set and the boxed data by using an hstack function, constructing frequency domain information by adopting fast Fourier transform, expanding data time domain signals in the data set to the frequency domain, expanding data characteristics, enhancing the relativity of data expression,
inputting the preprocessed data features into a neural network model for feature dimension reduction to obtain denoised data features, inputting the denoised data features into a classifier for classification, wherein the classifier is trained through a convolutional neural network, continuously optimizes the neural network structure of parameters by using an Adam algorithm, classifies according to falling behaviors and different falling states, classifies motion states including a normal state, a falling event and a lying event,
the storage module stores the generated data and the collected information for subsequent data analysis and statistics, constructs a user portrait level chart, forms a report for the user exercise health data for the feedback of the recent exercise and health status of the user,
and the alarm module is used for judging whether the behavior state of the user is a normal state, a falling event or a lying event according to the movement state classification and identification of the analysis module, sending an alarm if the falling event occurs, and sending early warning information to the user and emergency contact persons.
Preferably, in the method for fall detection by a wearable device based on a neural network, the collecting acceleration, angular velocity and geographic position information by the sensor module includes: and acquiring acceleration, angular velocity and geographic position information according to the accelerometer, the gyroscope and the GPS-like geographic position sensor respectively.
Preferably, in the method for performing fall detection by using a wearable device based on a neural network, the performing, by using an analysis module, data frequency domain expansion includes: constructing frequency domain information according to the fast Fourier change, two-dimensionally expanding the signals on the separated frequency domain scale in a windowing mode, obtaining a time sequence according to the separated time sequence signals by using a resampling method on the time sequence, and arranging the time sequence in the space dimension to form preprocessed signals so as to finish the expansion of the data time domain signals in the data set to the frequency domain.
Preferably, in the method for fall detection by a wearable device based on a neural network, the neural network structure of parameters is continuously optimized by an analysis module by using Adam algorithm, including: and iteratively training a neural network model, optimizing parameters, optimizing a neural network structure, and evaluating the neural network structure according to the accuracy, precision and recall rate of the classified evaluation indexes.
The invention also provides a system for performing fall detection by using the wearable device based on the neural network, which comprises a sensor module, an analysis module, a storage module and an alarm module,
the sensor module respectively collects the acceleration, the angular velocity and the geographic position information, comprehensively acquires the movement behavior information of the user,
the analysis module preprocesses the user movement behavior information, forms the preprocessed user movement behavior information into a data set, adds a data interaction item operation to the data set, then expands the data frequency domain, the data set is firstly subjected to a digitize boxing operation in the data interaction item operation, expands the data quantity, stacks the original data set and the boxed data by using an hstack function, adopts fast Fourier transform to construct frequency domain information, expands the data time domain signal in the data set to the frequency domain, expands the data characteristics, enhances the relativity of the data expression,
inputting the preprocessed data features into a neural network model for feature dimension reduction to obtain denoised data features, inputting the denoised data features into a classifier for classification, wherein the classifier is trained through a convolutional neural network, continuously optimizes the neural network structure of parameters by using an Adam algorithm, classifies according to falling behaviors and different falling states, classifies motion states including a normal state, a falling event and a lying event,
the storage module stores the generated data and the collected information for subsequent data analysis and statistics, and constructs a user portrayal level chart, forms a report for the user's exercise health data, for the user's recent exercise and health status feedback,
the alarm module is used for judging whether the behavior state of the user is a normal state, a falling event or a lying event according to the movement state classification identification of the analysis module, sending an alarm if the falling event occurs, and sending early warning information to the user and emergency contact persons.
Preferably, in the system for fall detection by a wearable device based on a neural network, the sensor module collects acceleration, angular velocity and geographic position information respectively, including: and acquiring acceleration, angular velocity and geographic position information according to the accelerometer, the gyroscope and the GPS-like geographic position sensor respectively.
Preferably, in the system for performing fall detection by using a wearable device based on a neural network, the data frequency domain expansion performed by the analysis module includes: constructing frequency domain information according to the fast Fourier change, two-dimensionally expanding the signals on the separated frequency domain scale in a windowing mode, obtaining a time sequence according to the separated time sequence signals by using a resampling method on the time sequence, and arranging the time sequence in the space dimension to form preprocessed signals so as to finish the expansion of the data time domain signals in the data set to the frequency domain.
Preferably, in the system for fall detection by wearable equipment based on neural network, the analysis module uses Adam algorithm to continuously optimize the neural network structure of parameters, including: and iteratively training a neural network model, optimizing parameters, optimizing a neural network structure, and evaluating the neural network structure according to the accuracy, precision and recall rate of the classified evaluation indexes.
The invention has the advantages that:
the invention provides a method for detecting falling of wearable equipment based on a neural network, which can realize more accurate and reliable falling detection and has the following advantages:
1. based on the neural network model, the accuracy and reliability of fall detection can be effectively improved.
2. The data acquisition and transmission are carried out on the wearable equipment, the data acquisition and transmission are not limited by time and space, the operation is simple and convenient, and the perception of the user is low.
3. Multiple different sensors and neural network models can be adopted, so that the requirements and conditions of different users can be met, and the system is more flexible.
4. The fall detection system has the functions of real-time response and automatic early warning, and can remind users and related personnel in time, so that the harm caused by falling is reduced.
5. The data storage and analysis function can monitor and analyze the motion state of the user for a long time, and the motion dimension image of the user is constructed by combining the data cleaning and the visual icon mode, so that more comprehensive and effective health management service can be provided.
Drawings
FIG. 1 is a schematic diagram of the application flow of the method of the present invention.
Detailed Description
The present invention will be further described with reference to the accompanying drawings and specific examples, which are not intended to be limiting, so that those skilled in the art will better understand the invention and practice it.
The invention provides a method for detecting falling of wearable equipment based on a neural network, which comprises deploying a falling detection system to the wearable equipment based on the neural network, wherein the falling detection system comprises a sensor module, an analysis module, a storage module and an alarm module,
the acceleration, angular velocity and geographic position information are respectively collected through the sensor module, the movement behavior information of the user is comprehensively obtained,
preprocessing user movement behavior information through an analysis module, forming a data set by the preprocessed user movement behavior information, adding a data interaction item operation to the data set, then performing data frequency domain expansion, firstly performing digitize boxing operation to the data set in the data interaction item operation to expand the data quantity, stacking the original data set and the boxed data by using an hstack function, constructing frequency domain information by adopting fast Fourier transform, expanding data time domain signals in the data set to the frequency domain, expanding data characteristics, enhancing the relativity of data expression,
inputting the preprocessed data features into a neural network model for feature dimension reduction to obtain denoised data features, inputting the denoised data features into a classifier for classification, wherein the classifier is trained through a convolutional neural network, continuously optimizes the neural network structure of parameters by using an Adam algorithm, classifies according to falling behaviors and different falling states, classifies motion states including a normal state, a falling event and a lying event,
the storage module stores the generated data and the collected information for subsequent data analysis and statistics, constructs a user portrait level chart, forms a report for the user exercise health data for the feedback of the recent exercise and health status of the user,
and the alarm module is used for judging whether the behavior state of the user is a normal state, a falling event or a lying event according to the movement state classification and identification of the analysis module, sending an alarm if the falling event occurs, and sending early warning information to the user and emergency contact persons.
The invention collects the motion data of the user, analyzes and judges the data by using the neural network model, realizes the timely identification and processing of the falling event, and improves the safety and health condition of the user. Meanwhile, the invention also provides an alarm and data analysis service, in particular to a method for realizing fall detection on a small wearable device so as to better ensure the safety and health of a user.
In a specific application, in some embodiments of the method of the invention, the fall detection is achieved by the following procedure:
step 1: a fall detection system is deployed to a neural network-based wearable device, the fall detection system comprising a sensor module, an analysis module, a storage module, and an alarm module.
Step 2: and respectively collecting acceleration, angular velocity and geographic position information through a sensor module, and comprehensively acquiring the movement behavior information of the user. The sensor module can comprise various sensors such as an accelerometer, a gyroscope and a magnetometer, and can acquire user movement behavior information from multiple dimensions such as acceleration, angular velocity and geographic position in real time, and the user movement behavior information is used as raw data of a fall detection system applied to the wearable equipment based on the neural network, so that input data is provided for a subsequent network analysis model.
Step 3: the user movement behavior information is preprocessed through the analysis module, the preprocessed user movement behavior information is formed into a data set, wherein the data preprocessing is mainly responsible for carrying out preliminary processing and cleaning on collected data, and the data preprocessing comprises the steps of adding data interaction items to the data set, and then carrying out data frequency domain expansion. In the data set increasing data interaction item operation, the data set is subjected to the digitize boxing operation first, the data quantity is increased, the size of the box body number can be adjusted to hundreds from tens, the original data set and the boxed data are stacked by using an hstack function, so that the two data are interacted, the generated result is the interaction characteristics of the two data, the generated result is that the number of new data points is unchanged, but the characteristic quantity is correspondingly increased along with the increase of the box body number, and the linear characteristics of falling data are enhanced to a certain extent.
Besides the linear features, the features implicit in the whole signal are highlighted. And constructing frequency domain information by adopting fast Fourier transform, expanding a data time domain signal in a data set to a frequency domain, expanding data characteristics and enhancing the correlation of data expression. Preferably, the data frequency domain expansion by the analysis module includes: constructing frequency domain information according to the fast Fourier change, two-dimensionally expanding the signals on the separated frequency domain scale in a windowing mode, obtaining a time sequence according to the separated time sequence signals by using a resampling method on the time sequence, and arranging the time sequence in the space dimension to form preprocessed signals so as to finish the expansion of the data time domain signals in the data set to the frequency domain. P (j, k) represents a single pixel point in the image, j=1, … …, M; k=1, … …, M; l (i), i=1, …, M, represents the length of the original signal, and the formula is as follows:
after pretreatment, the data expression dimension can be effectively increased, and the data quality can be improved, so that the motion state of a user can be reflected better. And then transmitting the processed data to a regular self-encoder in the neural network model for data dimension reduction and denoising, so that the feature extraction is facilitated.
Step 4: the preprocessed data features are input into a neural network model for feature dimension reduction, denoised data features are obtained, denoised data features are input into a classifier for classification, the classifier is trained through a convolutional neural network, the neural network structure of parameters is continuously optimized by using an Adam algorithm, classification is carried out according to falling behaviors and different falling states, and classified motion states comprise a normal state, a falling event and a lying event.
Preferably, the neural network structure for continuously optimizing parameters by the analysis module through Adam algorithm includes: and iteratively training a neural network model, optimizing parameters, optimizing a neural network structure, and evaluating the neural network structure according to the accuracy, precision and recall rate of the classified evaluation indexes. In the optimization process, in order to reduce the error of the neural network as much as possible, the difference between the actually output label y_ik and the predicted label d_ik is reduced as much as possible, the weight omega and the deviation b are continuously updated in the training process, the loss function adopts the mean square error, and if the difference between the model prediction result and the actual classification is small, the calculated mean square variance is small. According to the data category, the classification result is divided into three categories of normal movement state, falling event and lying event, and the classified evaluation indexes adopt conventional parameters including Accuracy (Accuracy), precision (Precision) and Recall rate (Recall). In the training process, parameters of the model are adaptively adjusted according to the network model prediction classification result, and the recognition accuracy of the model is improved. Training is carried out by using a continuously-expanded falling sample data set, and a neural network model is trained through multiple iterations, so that model parameters are optimized, and the accuracy and generalization capability of the model are improved.
Step 5: the generated data and the collected information are stored through the storage module for subsequent data analysis and statistics, a user portrait level chart is constructed, and a report is formed for the user exercise health data for the feedback of the recent exercise and health condition of the user.
Step 6: and the alarm module is used for judging whether the behavior state of the user is a normal state, a falling event or a lying event according to the movement state classification and identification of the analysis module, sending an alarm if the falling event occurs, and sending early warning information to the user and emergency contact persons.
The invention can also provide guarantee for the safety aspect of factory machines according to actual conditions such as physical state detection for factory large machines; in addition, communication services of the wearable device and the server can be combined with the latest internet of things technology to construct an internet of things mode based on the user scene completely.
The invention also provides a system for performing fall detection by using the wearable device based on the neural network, which comprises a sensor module, an analysis module, a storage module and an alarm module,
the sensor module respectively collects the acceleration, the angular velocity and the geographic position information, comprehensively acquires the movement behavior information of the user,
the analysis module preprocesses the user movement behavior information, forms the preprocessed user movement behavior information into a data set, adds a data interaction item operation to the data set, then expands the data frequency domain, the data set is firstly subjected to a digitize boxing operation in the data interaction item operation, expands the data quantity, stacks the original data set and the boxed data by using an hstack function, adopts fast Fourier transform to construct frequency domain information, expands the data time domain signal in the data set to the frequency domain, expands the data characteristics, enhances the relativity of the data expression,
inputting the preprocessed data features into a neural network model for feature dimension reduction to obtain denoised data features, inputting the denoised data features into a classifier for classification, wherein the classifier is trained through a convolutional neural network, continuously optimizes the neural network structure of parameters by using an Adam algorithm, classifies according to falling behaviors and different falling states, classifies motion states including a normal state, a falling event and a lying event,
the storage module stores the generated data and the collected information for subsequent data analysis and statistics, and constructs a user portrayal level chart, forms a report for the user's exercise health data, for the user's recent exercise and health status feedback,
the alarm module is used for judging whether the behavior state of the user is a normal state, a falling event or a lying event according to the movement state classification identification of the analysis module, sending an alarm if the falling event occurs, and sending early warning information to the user and emergency contact persons.
The content of information interaction and execution process between the modules in the system is based on the same concept as the method embodiment of the present invention, and specific content can be referred to the description in the method embodiment of the present invention, which is not repeated here.
Likewise, the system of the invention allows for a more accurate and reliable fall detection, with the following advantages:
1. based on the neural network model, the accuracy and reliability of fall detection can be effectively improved.
2. The data acquisition and transmission are carried out on the wearable equipment, the data acquisition and transmission are not limited by time and space, the operation is simple and convenient, and the perception of the user is low.
3. Multiple different sensors and neural network models can be adopted, so that the requirements and conditions of different users can be met, and the system is more flexible.
4. The fall detection system has the functions of real-time response and automatic early warning, and can remind users and related personnel in time, so that the harm caused by falling is reduced.
5. The data storage and analysis function can monitor and analyze the motion state of the user for a long time, and the motion dimension image of the user is constructed by combining the data cleaning and the visual icon mode, so that more comprehensive and effective health management service can be provided.
It should be noted that not all the steps and modules in the above processes and the system structures are necessary, and some steps or modules may be omitted according to actual needs. The execution sequence of the steps is not fixed and can be adjusted as required. The system structure described in the above embodiments may be a physical structure or a logical structure, that is, some modules may be implemented by the same physical entity, or some modules may be implemented by multiple physical entities, or may be implemented jointly by some components in multiple independent devices.
The above-described embodiments are merely preferred embodiments for fully explaining the present invention, and the scope of the present invention is not limited thereto. Equivalent substitutions and modifications will occur to those skilled in the art based on the present invention, and are intended to be within the scope of the present invention. The protection scope of the invention is subject to the claims.
Claims (8)
1. A method for detecting falling by wearable equipment based on a neural network is characterized in that a falling detection system is deployed to the wearable equipment based on the neural network, the falling detection system comprises a sensor module, an analysis module, a storage module and an alarm module,
the acceleration, angular velocity and geographic position information are respectively collected through the sensor module, the movement behavior information of the user is comprehensively obtained,
preprocessing user movement behavior information through an analysis module, forming a data set by the preprocessed user movement behavior information, adding a data interaction item operation to the data set, then performing data frequency domain expansion, firstly performing digitize boxing operation to the data set in the data interaction item operation to expand the data quantity, stacking the original data set and the boxed data by using an hstack function, constructing frequency domain information by adopting fast Fourier transform, expanding data time domain signals in the data set to the frequency domain, expanding data characteristics, enhancing the relativity of data expression,
inputting the preprocessed data features into a neural network model for feature dimension reduction to obtain denoised data features, inputting the denoised data features into a classifier for classification, wherein the classifier is trained through a convolutional neural network, continuously optimizes the neural network structure of parameters by using an Adam algorithm, classifies according to falling behaviors and different falling states, classifies motion states including a normal state, a falling event and a lying event,
the storage module stores the generated data and the collected information for subsequent data analysis and statistics, constructs a user portrait level chart, forms a report for the user exercise health data for the feedback of the recent exercise and health status of the user,
and the alarm module is used for judging whether the behavior state of the user is a normal state, a falling event or a lying event according to the movement state classification and identification of the analysis module, sending an alarm if the falling event occurs, and sending early warning information to the user and emergency contact persons.
2. A method of fall detection by a neural network-based wearable device according to claim 1, characterized in that the collecting acceleration, angular velocity and geographical position information, respectively, by means of a sensor module comprises: and acquiring acceleration, angular velocity and geographic position information according to the accelerometer, the gyroscope and the GPS-like geographic position sensor respectively.
3. A method for fall detection by a neural network-based wearable device according to claim 1, characterized in that the data frequency domain expansion by the analysis module comprises: constructing frequency domain information according to the fast Fourier change, two-dimensionally expanding the signals on the separated frequency domain scale in a windowing mode, obtaining a time sequence according to the separated time sequence signals by using a resampling method on the time sequence, and arranging the time sequence in the space dimension to form preprocessed signals so as to finish the expansion of the data time domain signals in the data set to the frequency domain.
4. A method for fall detection by a neural network-based wearable device according to claim 1, characterized in that the neural network structure with parameters continuously optimized by an analysis module using Adam algorithm comprises: and iteratively training a neural network model, optimizing parameters, optimizing a neural network structure, and evaluating the neural network structure according to the accuracy, precision and recall rate of the classified evaluation indexes.
5. A system for carrying out fall detection by wearable equipment based on a neural network is characterized in that a fall detection system is deployed to the wearable equipment based on the neural network, the fall detection system comprises a sensor module, an analysis module, a storage module and an alarm module,
the sensor module respectively collects the acceleration, the angular velocity and the geographic position information, comprehensively acquires the movement behavior information of the user,
the analysis module preprocesses the user movement behavior information, forms the preprocessed user movement behavior information into a data set, adds a data interaction item operation to the data set, then expands the data frequency domain, the data set is firstly subjected to a digitize boxing operation in the data interaction item operation, expands the data quantity, stacks the original data set and the boxed data by using an hstack function, adopts fast Fourier transform to construct frequency domain information, expands the data time domain signal in the data set to the frequency domain, expands the data characteristics, enhances the relativity of the data expression,
inputting the preprocessed data features into a neural network model for feature dimension reduction to obtain denoised data features, inputting the denoised data features into a classifier for classification, wherein the classifier is trained through a convolutional neural network, continuously optimizes the neural network structure of parameters by using an Adam algorithm, classifies according to falling behaviors and different falling states, classifies motion states including a normal state, a falling event and a lying event,
the storage module stores the generated data and the collected information for subsequent data analysis and statistics, and constructs a user portrayal level chart, forms a report for the user's exercise health data, for the user's recent exercise and health status feedback,
the alarm module is used for judging whether the behavior state of the user is a normal state, a falling event or a lying event according to the movement state classification identification of the analysis module, sending an alarm if the falling event occurs, and sending early warning information to the user and emergency contact persons.
6. The system for fall detection by a neural network-based wearable device of claim 5, wherein the sensor module gathers acceleration, angular velocity, and geographic location information, respectively, comprising: and acquiring acceleration, angular velocity and geographic position information according to the accelerometer, the gyroscope and the GPS-like geographic position sensor respectively.
7. The neural network-based fall detection system of wearable devices of claim 5, wherein the analysis module performs data frequency domain expansion comprising: constructing frequency domain information according to the fast Fourier change, two-dimensionally expanding the signals on the separated frequency domain scale in a windowing mode, obtaining a time sequence according to the separated time sequence signals by using a resampling method on the time sequence, and arranging the time sequence in the space dimension to form preprocessed signals so as to finish the expansion of the data time domain signals in the data set to the frequency domain.
8. The system for fall detection of a neural network-based wearable device of claim 5, wherein the analysis module uses Adam's algorithm to continuously optimize the neural network structure of parameters, comprising: and iteratively training a neural network model, optimizing parameters, optimizing a neural network structure, and evaluating the neural network structure according to the accuracy, precision and recall rate of the classified evaluation indexes.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310750679.5A CN116849643A (en) | 2023-06-25 | 2023-06-25 | Method for detecting falling of wearable equipment based on neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310750679.5A CN116849643A (en) | 2023-06-25 | 2023-06-25 | Method for detecting falling of wearable equipment based on neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116849643A true CN116849643A (en) | 2023-10-10 |
Family
ID=88220812
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310750679.5A Pending CN116849643A (en) | 2023-06-25 | 2023-06-25 | Method for detecting falling of wearable equipment based on neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116849643A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117423210A (en) * | 2023-12-19 | 2024-01-19 | 西南医科大学附属医院 | Nursing is with disease anti-drop intelligent response alarm system |
-
2023
- 2023-06-25 CN CN202310750679.5A patent/CN116849643A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117423210A (en) * | 2023-12-19 | 2024-01-19 | 西南医科大学附属医院 | Nursing is with disease anti-drop intelligent response alarm system |
| CN117423210B (en) * | 2023-12-19 | 2024-02-13 | 西南医科大学附属医院 | Nursing is with disease anti-drop intelligent response alarm system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20190294869A1 (en) | Object behavior anomaly detection using neural networks | |
| KR102111894B1 (en) | A behavior pattern abnormality discrimination system and method for providing the same | |
| JP7470203B2 (en) | Analysis and selection in aquaculture | |
| CN112418117A (en) | Small target detection method based on unmanned aerial vehicle image | |
| JP2012518846A (en) | System and method for predicting abnormal behavior | |
| WO2009137118A1 (en) | System architecture and process for assessing multi-perspective multi-context abnormal behavior | |
| WO2021068781A1 (en) | Fatigue state identification method, apparatus and device | |
| Benito-Picazo et al. | Deep learning-based video surveillance system managed by low cost hardware and panoramic cameras | |
| EP3349142B1 (en) | Information processing device and method | |
| CN116849643A (en) | Method for detecting falling of wearable equipment based on neural network | |
| US20200394418A1 (en) | Image processing method, an image processing apparatus, and a surveillance system | |
| Raj et al. | IoT-based real-time poultry monitoring and health status identification | |
| EP3759789B1 (en) | System and method for audio and vibration based power distribution equipment condition monitoring | |
| WO2018171875A1 (en) | Control device, system and method for determining the perceptual load of a visual and dynamic driving scene | |
| CN109117771B (en) | System and method for detecting violence events in image based on anchor nodes | |
| CN114218992A (en) | Abnormal object detection method and related device | |
| CN113887318A (en) | Embedded power violation detection method and system based on edge calculation | |
| US20220130109A1 (en) | Centralized tracking system with distributed fixed sensors | |
| WO2020039559A1 (en) | Information processing device, information processing method, and work evaluation system | |
| WO2021036669A1 (en) | Intelligent vibration detection method and related products | |
| KR102289212B1 (en) | Fault diagnosis apparatus and method based on artificial intelligence technology | |
| CN112529836A (en) | High-voltage line defect detection method and device, storage medium and electronic equipment | |
| CN109409173A (en) | Driver's state monitoring method, system, medium and equipment based on deep learning | |
| CN116823884A (en) | Multi-target tracking method, system, computer equipment and storage medium | |
| CN113743184B (en) | Abnormal Behavior Crowd Detection Method and Device Based on Element Mining and Video Analysis |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |