An arm bending degree detection method based on internet of things

Abstract

The application discloses an arm bending degree detection method based on an Internet of Things, relates to the motion capture technical field, and the detected person collects data by wearing an arm detection device, and generates a feedback signal to establish a wearing model according to the user data; the sensor data collected by the arm detection device is subjected to multi-point analysis processing, the picture image data recorded by a real-time display drawing board is subjected to frame-by-frame analysis processing, image cutting processing and an arm model is established; the arm model stored in a cloud database is matched, if the matching succeeds, the corresponding bending degree is obtained, and if the matching fails, the bending degree is calculated according to the generated arm model.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of motion capture, in particular to an arm bending degree detection method based on Internet of Things. BACKGROUND

[0002] Motion capture (Mocap) is a technology related to size measurement, positioning and orientation determination of objects in a physical space, which is data that can be directly understood and processed by a computer. A tracker is arranged at a key part of a moving object, the position of the tracker is captured by a Motion capture system, and three-dimensional space coordinate data is obtained after computer processing. When the data is recognized by the computer, it can be applied in the fields of animation production, gait analysis, biomechanics and ergonomics.

[0003] The motion capture technology is used to obtain the bending degree of the arm part. The collected bending degree data can be used for subsequent ergonomics analysis. The traditional arm bending degree detection is performed by arranging corresponding pressure sensors, speed sensors and angle sensors in the joint area, non-joint area and combined area of the arm. The arrangement mode is fixed and cannot be adjusted according to different arm shapes after the arrangement is completed. After the data in the sensor is obtained, the bending degree can only be obtained in a single way of calculating data. The bending degree is indirectly obtained by considering the access to the cloud and matching with the existing arm model with similar degree, so as to reduce the data processing amount and obtain the bending degree without calculation. The error of the bending degree calculated by using one-dimensional data is large. How to process multiple-dimensional data and reduce the error by obtaining the average value of multiple-dimensional data are problems faced by the current arm bending degree detection. SUMMARY

[0004] In order to solve the above problems, the purpose of the present application is to provide an arm bending degree detection method based on Internet of Things.

[0005] The purpose of the present application can be realized by the following technical scheme: an arm bending degree detection method based on Internet of Things, comprising the following steps:

[0006] Step S1: a data collection of a detected person wearing an arm detection device, and the detected person inputs self user data to establish a wearing model combined with a feedback signal, wherein the arm detection device comprises a sensor system and a real-time display board;

[0007] Step S2: multi-point analysis and processing of sensor data collected by the arm detection device, frame-by-frame analysis and processing, image cutting processing and arm model establishment of the drawing board image data recorded by the real-time display board;

[0008] Step S3: By networking, the user data and arm model uploaded by various other detected persons in the cloud database are acquired, the login unit is set to input the user data and the associated arm model thereof, the arm model generated from the user data is matched with the arm model in the cloud database, if the matching is successful, the bending degree corresponding to the arm model in the cloud database is acquired as the bending degree of the arm detection this time; if the matching fails, the bending degree of the arm detection this time is acquired according to the arm model generated by the detection, and the arm model generated by the detection this time and the user data are uploaded to the cloud database for storage.

[0009] Further, the process of the detected person wearing the arm detection device for data collection includes:

[0010] The arm detection device is composed of a sensor system and a real-time display drawing board;

[0011] The sensor system is composed of a plurality of pressure sensors, speed sensors and angle sensors, and signals of corresponding sensor types are collected;

[0012] The real-time display drawing board is used for real-time display of drawing board image data, establishment of different models and display of feedback signals.

[0013] Further, the process of the sensor system collecting signals of various types includes:

[0014] The pressure signal of the non-joint region of the arm is collected by the pressure sensor, the speed signal of the joint region and the non-joint region of the arm is collected by the speed sensor, and the angle signal of the joint region is collected by the angle sensor;

[0015] The pressure signal, the speed signal and the angle signal are processed by the micro-motion unit in the arm detection device to divide the pressure signal, the speed signal and the angle signal into a plurality of sub-signals;

[0016] The pressure signal, the speed signal and the angle signal are provided with a calibration threshold and a standard numerical range;

[0017] When the values collected by various sensors are greater than or equal to the calibration threshold, automatic calibration is performed, and when the values of the sub-signals collected are greater than or equal to or less than or equal to the standard numerical range, automatic rejection is performed.

[0018] Further, the process of generating the wearing model includes:

[0019] The detected person inputs the user data through the input unit of the arm detection device;

[0020] The self-user data includes arm length, arm girth, wrist girth, distance from joint rotation part to wrist and arm pressure signal in static state;

[0021] First, an initial wearing model is established according to self-user data, and then a feedback signal is generated according to the arm pressure signal in static state in the self-user data and the pressure signal;

[0022] According to the feedback signal, corresponding adjustment operations are performed on the initial wearing model, and after all the feedback signals corresponding to the adjustment operations are completed, the initial wearing model is changed into a wearing model.

[0023] Further, the process of multi-point analysis and processing of the sensor data includes:

[0024] The number of sub-signals collected by each type of sensor is obtained and an upward rounding operation is performed, and a corresponding number of data folders is generated according to the obtained integer;

[0025] Each data folder is used as a connection point, and the connection points are interconnected two by two, and when all the connection points are interconnected, a pressure signal network architecture, a speed signal network architecture and an angle signal network architecture are generated;

[0026] The connection points are analyzed, and if the data in the connection points is abnormal, it is marked as F, the total number of connection points is obtained, the total number of connection points is used as the denominator, the number of abnormal connection points is used as the numerator to obtain a ratio, and if the ratio is greater than or equal to 1 / 2, the network architecture of the corresponding type is destroyed;

[0027] Each type of network architecture with a ratio less than 1 / 2 is saved and used as basic data for generating a drawing board image data.

[0028] Further, the process of frame-by-frame analysis and image cutting of the drawing board image data includes:

[0029] Each type of network architecture is obtained and drawing board image data is generated and displayed on a real-time display drawing board;

[0030] The drawing board image data is generated once every set time interval T, and a plurality of drawing board image data generated by T is obtained, and each drawing board image data is used as a frame;

[0031] Each frame of drawing board image data is divided into a central region, a sub-central region and an edge region;

[0032] The area of the central region of each frame of drawing board image data is calculated, and two areas are obtained each time, and the smaller one is used as the denominator and the larger one is used as the numerator to obtain an overlap rate, and the drawing board image data is cut according to the overlap rate;

[0033] Obtain a period of drawing board image data, according to the overlap rate of adjacent time interval drawing board image data, directly remove the overlap part of the drawing board image data of the secondary center area, and retain the non-overlapping area of the edge area.

[0034] Further, the process of establishing the arm model comprises:

[0035] Establish an initial arm model, take the processed drawing board image data as input data, remove the data of the current time from the input data and mark it as historical data;

[0036] The initial arm model is continuously trained to generate a training data set according to the historical data, and the initial arm model is adjusted in real time according to the training data set;

[0037] After several adjustments, an arm model is generated, and the arm model is saved and uploaded to a cloud database.

[0038] Further, the process of connecting the cloud database comprises:

[0039] Get the connection permission of the cloud database, set a login unit, and the login unit is used to input the user data and the arm model;

[0040] If the connection permission of the cloud database is obtained, the login unit is logged into the cloud database, and if the connection permission cannot be obtained, an application operation is performed;

[0041] After obtaining the connection permission, the mobile phone of the detected person receives a permission approval email, and the permission approval email includes a login account and a login password;

[0042] The detected person logs into the cloud database through the login unit according to the login account and the login password, and can continue to perform a matching operation;

[0043] The application operation sends a to-be-confirmed application email to the mobile phone of the detected person, and the to-be-confirmed application email is provided with an application link, and the temporary access account and the one-time temporary private key are obtained by clicking the application link. The detected person realizes the connection with the cloud database and performs a matching operation through the temporary access account and the one-time temporary private key.

[0044] Further, the process of the matching operation comprises:

[0045] Take the user data and the associated arm model as to-be-matched information, and take the arm model and the user data already stored in the cloud database as target information;

[0046] Set persistent resource pools Y1 and Y2 to store the to-be-matched information and the target information respectively, and the arm model and the user data in the resource pool Y1 are stored in the form of a two-dimensional array.

[0047] The first bit of the two-dimensional array stores an arm model, and the second bit stores self-user data, and the reading of the first bit arm model is sequentially performed;

[0048] The read arm model is sequentially transmitted to the resource pool Y2, and all arm models in the resource pool Y2 are obtained and matched with the received arm model;

[0049] Select several reference points as feature points, traverse and record the number of coincident feature points of each arm model stored in the cloud database and the received arm model, and select the arm model with the highest number as the confirmation target;

[0050] The confirmation target is associated with a corresponding bending degree, and the bending degree is transmitted to the real-time display board for display as the bending degree of this arm detection;

[0051] If the arm model is not matched after the traversal ends, jump to the bending degree calculation interface of the real-time display board for calculation operation.

[0052] Further, the process of the bending degree calculation operation includes:

[0053] The calculation operation is performed according to the arm model on the real-time display board, the arm model includes vertical data points and horizontal data points, the vertical data points are taken as reference points, and the horizontal data points are taken as orthogonal points;

[0054] A reference line and an orthogonal line are respectively generated through the reference points and the orthogonal points, the reference line has only one and always remains horizontal, the orthogonal line has multiple and intersects with the reference line, and the included angle between the orthogonal line and the reference line is a bending angle;

[0055] Different orthogonal lines intersect with the reference line correspondingly to have corresponding bending angles and values, the value of each bending angle is obtained, and the values are summed after removing the highest value and the lowest value;

[0056] The value sum is averaged to generate an average value, which is the bending degree of this detection;

[0057] The bending degree is obtained and uploaded to the cloud database, the self-user data and the arm model in the cloud database are merged to form a mapping relationship of bending degree-user data-arm model;

[0058] The related information corresponding to the mapping relationship is persistently stored in the cloud database as a new bending degree for matching operation by other detected persons.

[0059] Compared with the prior art, the arm detection device generates a feedback signal through real-time display of the drawing board, generates a wearing model according to the feedback signal, and adjusts the arm detection device accordingly, so that the collected data is more accurate; through the setting of the two mails, the data collection of the device holder and the temporary detector can be carried out, and the mails also ensure that the data is not easy to be stolen; a special matching mechanism is set, when there is a matching arm model in the cloud database, the corresponding bending degree is directly obtained, the time for calculating the bending degree is reduced, if the corresponding arm model cannot be matched in the cloud database, a more accurate bending degree is obtained by forming a plurality of bending angles and values through the reference line and the orthogonal line, taking the average value after eliminating the maximum value and the minimum value, and the generated bending degree is uploaded to the cloud database, which can be used by other detected persons subsequently, so that the purpose of arm bending degree detection is achieved. BRIEF DESCRIPTION OF DRAWINGS

[0060] Figure 1 The flowchart of the present application. DETAILED DESCRIPTION

[0061] As shown in Figure 1 A method for detecting arm bending degree based on Internet of Things, comprising the following steps:

[0062] Step S1: the detected person wears an arm detection device to collect data, and inputs the user data of the detected person to establish a wearing model in combination with the feedback signal, wherein the arm detection device comprises a sensor system and a real-time display drawing board;

[0063] Step S2: the sensor data collected by the arm detection device is subjected to multi-point analysis processing, the drawing board image data recorded by the real-time display drawing board is subjected to frame-by-frame analysis processing, image cutting processing and arm model establishment;

[0064] Step S3: various user data and arm models uploaded in the cloud database by other detected persons are obtained through networking, a login unit is set to input the user data and the associated arm model of the detected person, the arm model generated from the user data and the arm model in the cloud database are subjected to matching operation, if the matching is successful, the bending degree corresponding to the arm model in the cloud database is obtained as the bending degree of the present arm detection; if the matching fails, the bending degree of the present arm detection is obtained according to the arm model generated by the detection, and the arm model generated by the present detection and the user data of the detected person are uploaded to the cloud database for storage.

[0065] The detected person wears an arm detection device, and the arm detection device is composed of a sensor system and a real-time display drawing board;

[0066] The sensor system comprises a plurality of pressure sensors, speed sensors and angle sensors, the pressure sensors are arranged in the non-joint region of the arm of the detected person to collect pressure signals, the speed sensors are arranged in the joint region and non-joint region of the arm to collect speed signals generated by the swinging of the arm, and the angle sensors are arranged in the elbow bending small region of the joint region to record the angle signals generated by the swinging arm;

[0067] The non-joint region of the arm comprises an A region and a B region, all the sensors in the A region form an A-level sensor sub-network, and the B region corresponds to a B-level sensor sub-network;

[0068] The pressure signals, speed signals and angle signals are processed by the micro-motion unit in the arm detection device to be micro-moved, the pressure signals, speed signals and angle signals are divided into a plurality of subdivided sub-signals and are numbered i, j and k respectively, wherein i∈1, 2, 3,..., n, j∈1, 2, 3,..., m, k∈1, 2, 3,..., p, wherein m, n and p are natural numbers;

[0069] The pressure signals, speed signals and angle signals are all provided with a calibration threshold and a standard numerical range, when the values collected by the various sensors in the sensor system are greater than or equal to the calibration threshold, the sensors are automatically calibrated, when the values of the sub-signals collected are greater than or equal to or less than or equal to the standard numerical range, the sub-signals are automatically excluded, and the sub-signals with abnormal values after the exclusion are summarized;

[0070] The detected person inputs the self-user data through the input unit of the arm detection device, and the self-user data comprises the arm length, arm girth, wrist girth, distance from the joint rotating part to the wrist and arm pressure signal in the static state;

[0071] The data transmission between the sensor system and the real-time display board is real-time and bidirectional, an initial wearing model is established according to the self-user data, and a feedback signal is generated according to the arm pressure signal in the static state in the self-user data and the pressure signal generated by the swinging of the arm;

[0072] The initial wearing model is adjusted according to the feedback signal, and the feedback signal comprises “CL”, “DL” and “EL”, and the corresponding adjustment operations are as follows:

[0073] When the feedback signal is “CL”, the real-time display board prompts information, and the information content is “shrink the wearing device”;

[0074] When the feedback signal is “DL”, the real-time display board prompts information, and the information content is “expand the wearing device”;

[0075] When the feedback signal is "EL", the real-time display panel prompts information, and the information content is "the optimal wearing state is reached, and the wearing model is generated!";

[0076] When the adjustment operation corresponding to all the feedback signals is completed, the initial wearing model is changed into the wearing model;

[0077] After the wearing adjustment of the arm detection device worn by the detected person is completed according to the wearing model, the sensor data is processed by multi-point analysis, the panel image data generated by the real-time display panel is processed by frame-by-frame analysis and image segmentation, and the arm model is established;

[0078] It should be noted that the feedback signal prompt information content is displayed on the real-time display panel, and the wearing adjustment of the arm detection device is performed, so that the data collection is more accurate.

[0079] Specifically, the process of processing the sensor data by multi-point analysis includes:

[0080] The sensor data includes pressure signals, speed signals and angle signals, and the sub-signals generated after the sensor data micro-motion processing are obtained;

[0081] The numbers n, m and p are obtained, and the numbers are rounded up to 3, the data folders corresponding to the obtained integers are generated, and the number of sub-signals in the last generated data folder is not full;

[0082] Each data folder is used as a connection point, the connection points are interconnected between each other, and each connection point can be connected to at most three connection points, and when all the connection points are interconnected, the pressure signal network architecture, the speed signal network architecture and the angle signal network architecture are generated;

[0083] The connection points are analyzed by multi-point, if the data in the connection points is abnormal, it is marked as F, the total number of connection points is obtained, and is recorded as M, if the ratio of F / M is greater than or equal to 1 / 2, the corresponding network architecture is destroyed;

[0084] The network architecture with the ratio of F / M less than 1 / 2 is saved and used as the basic data for generating the panel image data;

[0085] Specifically, the process of processing the panel image data by frame-by-frame analysis and image segmentation includes:

[0086] The panel image data generated by the arm swing bending is recorded on the real-time display panel, the panel image data is formed based on the network architecture, and the panel image data is generated once every set time interval T, therefore, a plurality of panel image data is generated in a period of time, and each panel image data is one frame;

[0087] The drawing board image data is divided into a center region, a sub-center region and an edge region by analyzing each frame;

[0088] The area of the center region of each frame of drawing board image data is calculated and recorded as S1, S2,..., Sn. Each time, two areas are obtained in sequence, and the larger one is used as the denominator and the smaller one is used as the numerator to obtain an overlap rate. The drawing board image data is cut according to the overlap rate;

[0089] The drawing board image data of a period of time is obtained, and the non-overlapping region of the edge region is retained by directly removing the overlapping part of the drawing board image data of the sub-center region according to the overlap rate of the drawing board image data of adjacent time intervals;

[0090] Specifically, the process of establishing the arm model includes:

[0091] An initial arm model is established, and the processed drawing board image data is used as input data, which does not include data of the current time, i.e., historical data;

[0092] The initial arm model is continuously trained to generate a training data set according to the historical data, and the initial arm model is adjusted in real time according to the training data set;

[0093] After multiple adjustments, the initial arm model generates a final required arm model, and the arm model is saved and uploaded to a cloud database;

[0094] The connection permission of the cloud database is obtained, and a login unit is set, which is used to input self-user data and the arm model;

[0095] First, it is judged whether the connection permission is obtained. If the connection permission of the cloud database is obtained, the login unit is logged into the cloud database. If the connection permission cannot be obtained, an application operation is performed;

[0096] The permission obtained is a permission email received by the mobile phone of the detected person. The received permission email includes a login account and a login password;

[0097] The detected person logs into the cloud database through the login unit according to the login account and the login password, and continues the matching operation;

[0098] The application operation is performed when the connection permission cannot be obtained. A to-be-confirmed application email is sent to the mobile phone of the detected person. The to-be-confirmed application email is provided with an application link. Clicking the application link can obtain a temporary access account and a one-time temporary private key. The detected person realizes the connection with the cloud database through the temporary access account and the one-time temporary private key;

[0099] It should be noted that the cloud database connection permission is judged first, and then the login unit is used to log in to obtain data information, which prevents others from stealing data information. The purpose of confirming the setting of the application email is to enable the arm detection device to be worn by others to continue, which is equivalent to temporary use.

[0100] When the access to the cloud database is successful, a matching operation is performed, which takes the user data and the associated arm model as the matching information, and the arm model and the user data already stored in the cloud database as the target information;

[0101] Persistent resource pools Y1 and Y2 are set, Y1 is used to store the matching information, and Y2 is used to store the target information. The arm model and the user data in the resource pool Y1 are stored by setting a two-dimensional array, the first bit of which stores the arm model, the second bit stores the user data, and the first bit of the arm model is read in sequence;

[0102] The read arm model is transmitted to the resource pool Y2 in order, all arm models in the resource pool Y2 are obtained and matched with the transmitted arm model, a number of reference points are selected as feature points, and the number of coincident feature points between each arm model already stored in the cloud database and the transmitted arm model is recorded. The arm model with the highest number is selected as the confirmation target;

[0103] The confirmation target is associated with a corresponding bending degree, which is transmitted to the real-time display board for display as the bending degree of this arm detection. This is a matching success case;

[0104] If the arm model is not matched after the iteration is completed, the bending degree calculation interface of the real-time display board is jumped to for calculation operation;

[0105] The calculation operation is performed according to the recorded arm model on the real-time display board, the arm model includes vertical data points and horizontal data points, the vertical data points are used as a number of reference points, and the horizontal data points are used as orthogonal points;

[0106] The reference points and orthogonal points are connected in a one-way sequence to generate reference lines and orthogonal lines, respectively. The reference line has only one and always remains horizontal, and the orthogonal lines have multiple and intersect with the reference line. The angle between the orthogonal line and the reference line is called the bending angle;

[0107] The parallel orthogonal lines are marked as the first orthogonal line, the second orthogonal line, the third orthogonal line,..., and the Zth orthogonal line;

[0108] Different orthogonal lines intersect the reference line correspondingly have corresponding bending angles and values, denoted as ∠1=X1, ∠2=X2, ∠3=X3, …, ∠Z=XZ, the value of each bending angle is obtained, and the highest value and the lowest value are removed from the value and summed, denoted as SUM;

[0109] The sum SUM is subjected to an average value operation, and the average value is denoted as Ave, Ave=SUM / (Z-2), and the generated average value is the bending degree of this detection, denoted as H;

[0110] The bending degree H recorded on the real-time display drawing board is obtained and uploaded to the cloud database, the self user data and the arm model in the cloud database are merged to form a mapping relationship of bending degree-user data-arm model, the related information corresponding to the mapping relationship is persistently stored in the cloud database, a new bending degree is generated, and can be matched for use by other detected persons next time;

[0111] It needs to be further explained that the provided matching operation mechanism can obtain the corresponding bending degree when there is a corresponding arm model in the cloud database, thereby reducing the time consumed for obtaining the bending degree, when there is no matching, a plurality of bending angles and corresponding values generated by the reference line and the orthogonal line are obtained, the influence of the maximum and minimum end point values is eliminated, the sum is summed and the average value is taken, the accuracy of the bending angle calculation is improved, and the data associated with the formed mapping relationship is uploaded, thereby facilitating the use of other detected persons subsequently.

[0112] The above embodiments are only used to illustrate the technical method of the present application and are not limited, although the present application is described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present application can be modified or replaced equivalently without departing from the spirit and scope of the technical method of the present application.

Claims

1. A method for detecting arm bending degree based on the Internet of Things, characterized in that, Includes the following steps: Step S1: The subject wears the arm detection device to collect data. The subject inputs their own user data and combines it with feedback signals to establish a wearing model. The arm detection device includes a sensor system and a real-time display board. Step S2: Perform multi-point analysis and processing on the sensor data collected by the arm detection device, and perform frame-by-frame analysis, image segmentation, and arm model establishment on the real-time display canvas image data recorded by the canvas. Step S3: Obtain user data and arm models uploaded to the cloud database by other subjects through the network, set up the login unit to input its own user data and its associated arm model, and perform a matching operation between the arm model generated by its own user data and the arm model in the cloud database. If the matching is successful, the curvature corresponding to the arm model in the cloud database is obtained as the curvature of the arm detection. If the match fails, the bending degree of the arm detected is obtained based on the arm model generated by the detection, and the arm model generated by the detection and the user data are uploaded to the cloud database for storage. The matching operation process includes: Use its own user data and its associated arm model as the matching information, and use the arm model and user data already stored in the cloud database as the target information. Persistent resource pools Y1 and Y2 are set up to store the information to be matched and the target information, respectively. The arm model and the user's own data in resource pool Y1 are stored by setting a two-dimensional array. The first position of the two-dimensional array stores the arm model, and the second position stores the user's own data. The arm model of the first position is read sequentially. The read arm models are transmitted sequentially to resource pool Y2, and all arm models in resource pool Y2 are obtained and their feature points are matched with the received arm models. Select several reference points as feature points, traverse and record the number of feature points that overlap between each arm model already stored in the cloud database and the received arm model, and take the arm model with the highest number of overlap points as the confirmation target. The confirmed target is associated with a corresponding degree of curvature, and this degree of curvature is transmitted to the real-time display panel for display as the degree of curvature detected in this arm test. If no arm model is matched after the traversal is completed, the process will jump to the real-time display of the bending degree calculation interface on the canvas for calculation. The process of calculating the curvature includes: The calculation operation is performed based on the arm model on the real-time display board. The arm model includes vertical data points and horizontal data points. The vertical data points are used as reference points, and the horizontal data points are used as orthogonal points. A baseline and an orthogonal line are generated using a reference point and an orthogonal point, respectively. There is only one baseline, which is always horizontal. There are multiple orthogonal lines running side by side, which intersect the baseline. The angle between the orthogonal line and the baseline is the curvature angle. Different orthogonal lines intersect the baseline and have corresponding bending angles and values. Obtain the value of each bending angle, remove the highest and lowest values, and then sum the values. The average value is calculated by taking the sum of the values, and the resulting average value is the curvature of the measured value. The curvature is obtained and uploaded to the cloud database. The user data and arm model in the cloud database are merged to form a mapping relationship between curvature, user data and arm model. The relevant information corresponding to this mapping relationship is persistently stored in the cloud database as a new curvature for other subjects to perform matching operations.

2. The method for detecting arm bending degree based on the Internet of Things according to claim 1, characterized in that, The process of data collection by the subject wearing the arm detection device includes: The arm detection device consists of a sensor system and a real-time display panel; The sensor system consists of a number of pressure sensors, speed sensors and angle sensors, which collect signals corresponding to the sensor types. The real-time display canvas is used for real-time display of canvas image data, creation of different models, and display of feedback signals.

3. The method for detecting arm bending degree based on the Internet of Things according to claim 2, characterized in that, The process of acquiring various types of signals by the sensor system includes: Pressure sensors collect pressure signals from non-joint areas of the arm, speed sensors collect speed signals from joint and non-joint areas of the arm, and angle sensors collect angle signals from joint areas. The pressure signal, speed signal, and angle signal are micro-motion processed by the micro-motion unit in the arm detection device, dividing the pressure signal, speed signal, and angle signal into several sub-signals; The pressure signal, velocity signal, and angle signal are all equipped with calibration thresholds and standard value ranges. Automatic calibration is performed when the values ​​collected by various sensors are greater than or equal to the calibration threshold, and automatic rejection is performed when the collected sub-signal values ​​are greater than or equal to or less than or equal to the standard value range.

4. The method for detecting arm bending degree based on the Internet of Things according to claim 2, characterized in that, The process of generating the wearing model includes: The person being tested inputs their own user data through the input unit set up on the arm detection device; The user data includes arm length, arm circumference, wrist circumference, distance from the joint rotation point to the wrist, and arm pressure signal in a static state. First, an initial wearing model is established based on the user's own data. Then, a feedback signal is generated by combining the arm pressure signal in the static state with the pressure signal in the user's own data. Based on the feedback signals, the initial wearing model is adjusted accordingly. After all the adjustment operations corresponding to the obtained feedback signals are completed, the initial wearing model is transformed into a wearing model.

5. The method for detecting arm bending degree based on the Internet of Things according to claim 4, characterized in that, The process of performing multi-point analysis on the sensor data includes: Obtain the number of sub-signals collected by each type of sensor and round up, then generate a corresponding number of data folders based on the obtained integers; Each data folder is treated as a connection point, and these connection points are interconnected. Once all connection points are interconnected, pressure signal network architecture, speed signal network architecture, and angle signal network architecture are generated. Perform multi-point analysis on the connection points. If the data in the connection point is abnormal, mark it as F. Obtain the total number of connection points. Use the total number of connection points as the denominator and the number of abnormal connection points as the numerator to obtain the ratio. If the ratio is greater than or equal to 1 / 2, destroy the network architecture of the corresponding type. Network architectures with a ratio less than 1 / 2 are saved and used as the basis for generating canvas image data.

6. The method for detecting arm bending degree based on the Internet of Things according to claim 5, characterized in that, The process of performing frame-by-frame analysis and image segmentation on the canvas image data includes: Acquire various network architectures and generate canvas image data to display on a real-time display canvas; The canvas image data is generated once every set time interval T. Several canvas image data generated by several T are obtained, and each canvas image data is taken as a frame. Each frame of canvas image data is divided into a central region, a secondary central region, and an edge region; The area of ​​the central region of each frame of canvas image data is calculated. Two areas are obtained in sequence each time. The larger value is used as the denominator and the smaller value is used as the numerator to obtain the overlap rate. The canvas image data is then segmented according to the overlap rate. Acquire canvas image data over a period of time. Based on the overlap rate of canvas image data in adjacent time intervals, directly remove the overlapping parts of the canvas image data in the secondary center region, and retain the non-overlapping parts in the edge region.

7. The method for detecting arm bending degree based on the Internet of Things according to claim 6, characterized in that, The process of creating the arm model includes: An initial arm model is established, and the processed canvas image data is used as input data. The current time data is removed from the input data and marked as historical data. The initial arm model is continuously trained based on historical data to generate a training dataset, and the initial arm model is adjusted in real time based on the training dataset. After several adjustments, an arm model is generated, and the arm model is saved and uploaded to the cloud database.

8. The method for detecting arm bending degree based on the Internet of Things according to claim 7, characterized in that, The process of connecting to the cloud database includes: Obtain connection permissions to the cloud database and set up a login unit, which is used to input user data and arm model; If connection permissions to the cloud database are obtained, then log in to the cloud database using the login unit; otherwise, apply for permission. After obtaining connection permission, the tested user's mobile phone receives a permission grant email, which includes the login account and login password. The person being tested logs into the cloud database using their login account and password through the login unit and can then continue the matching process. The application process sends a confirmation email to the user's mobile phone. The confirmation email contains an application link. Clicking the application link obtains a temporary access account and a one-time temporary private key. The user then uses the temporary access account and the one-time temporary private key to connect to the cloud database and perform matching operations.

Images