Motion state identification method and system for intelligent kneecap

Abstract

The invention provides a motion state recognition method and system for an intelligent kneecap, and the method comprises the following steps: S1, building a model: employing a random forest method to construct the model; s2, model deployment: firstly converting the established model into a C language deployed at an equipment end, then performing offline identification on the model, and calculating the probability of the motion state of each current item by the model by using an integrated decision tree method; and the optimal model performs weighting processing on the calculated original probability of each item motion state to obtain a final prediction probability, and the state with the maximum probability is identified as the current item motion state by the system. The model is deployed on the fixed intelligent kneecap, fine motion changes of the knee joint can be accurately captured, the recognition rate is high, the current motion states such as static, walking, running, stair climbing and special motion of a wearer can be accurately recognized, and therefore targeted protection measures are provided according to the current motion states, and health of the knee joint is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of knee pads, and in particular to a motion state recognition method for an intelligent knee pad. The present invention also relates to a motion state recognition system for an intelligent knee pad. Background Art

[0002] With the improvement of health awareness, more and more people are beginning to pay attention to knee joint protection in daily life. As an important load-bearing joint of the human body, the knee joint has random movement status in daily activities. For example, during walking, running, going up and down stairs, sitting and standing up, the load and impact on the knee joint are also different. However, the existing smart knee pads cannot provide effective protection for different movement states. Therefore, how to monitor the state of the knee joint in real time during daily activities and provide targeted protection measures according to different movement states is the key to ensuring the health of the knee joint.

[0003] In addition, existing motion state recognition technology relies on sensors at the wrist, waist and ankle, but it has limitations. Wrist and waist sensors are difficult to accurately capture subtle movements of the knee joint and lack detailed analysis of the knee joint state. Although ankle sensors are suitable for lower limb motion monitoring, the data mainly reflects foot movement and cannot accurately identify the specific state of the knee joint. Summary of the invention

[0004] The present invention aims to provide a method for identifying the motion state of an intelligent knee brace to overcome the deficiencies in the prior art.

[0005] In order to solve the above technical problems, the technical solution of the present invention is: a method for identifying the motion state of an intelligent kneepad, comprising the following steps:

[0006] S1. Model establishment: The random forest method is used to build the model, including the following steps:

[0007] S11, database acquisition, let the subject complete multiple motions continuously in the specified motion process, obtain the three-axis original motion information {gx, gy, gz} of the subject's knee, collect data with a specified depth of N as a column of data, collect multiple columns of data for each motion action, and then integrate these data into a complete data set with motion labels, and integrate the data sets of multiple subjects to form a database;

[0008] S12, data processing and training, extracting a certain column of data from the data set with action labels in the above database in equal proportion and combining them into corresponding data with project labels, using StandardScaler to standardize the data for each project, and then extracting the eigenvalues ​​of the three-axis acceleration from the standardized data; then importing the extracted eigenvalues ​​of the three-axis acceleration into the random forest, generating multiple decision trees through the Bootstrap sampling technology, and obtaining a model with a high accuracy through multiple trainings to achieve the establishment of the model;

[0009] S2, model deployment, including the following steps:

[0010] S21, model conversion, convert the established model into C language through tensorflow lite so that it can be deployed on the embedded device side;

[0011] S22, model offline identification, the deployment end of the smart knee brace worn on the knee is equipped with a three-axis accelerometer, which acquires the motion data of the knee in real time at a frequency of 100Hz, and the motion data is three-axis acceleration; the collected motion data will be pressed into the cache area of ​​the deployment end, and when the amount of data in the cache area reaches the specified depth N, a data set is obtained, triggering a model operation, extracting the characteristic value of the three-axis acceleration in the data set, and normalizing this characteristic value, and inputting it into the model in a standardized form; the model uses an integrated decision tree method to comprehensively analyze each characteristic value and calculate the original probability of the motion state of each item;

[0012] S23, the model output result is post-processed, and the calculated original probability of the motion state of each item is weighted to obtain the final predicted probability. The state with the highest probability is recognized by the system as the current motion state of the item.

[0013] Furthermore, the above-mentioned motion state recognition method of the intelligent knee brace acquires the original three-axis motion information {gx, gy, gz} of the subject's knee through IIC communication and three-axis accelerometer communication when the database is acquired, and the unit is m / s 2 The measurement range of the subject during exercise is ±3g, where g is the value of gravity acceleration.

[0014] Furthermore, in the above-mentioned motion state recognition method of the smart knee pad, the multiple motion movements that the subject needs to complete during the exercise process include: standing, sitting, turning, shaking legs, walking at a low speed, walking at a high speed, walking and stopping, walking sideways, jogging, sprinting, cycling, going upstairs, and going downstairs.

[0015] Furthermore, the motion state recognition method of the above-mentioned smart knee pad classifies and combines multiple motion movements completed by the subject during exercise into 5 items, namely: stillness, walking, running, climbing stairs, and special sports. The stillness items include standing, sitting still, turning around, and shaking legs. The walking items include low-speed walking, high-speed walking, stop-and-go, and sideways walking. The running items include jogging and sprinting. The climbing stairs items include going upstairs and going downstairs. Special sports items include cycling.

[0016] Furthermore, in the above-mentioned method for identifying the motion state of the intelligent knee brace, in step S12 and step S22, the characteristic values ​​of the three-axis acceleration include the maximum value, minimum value, average value, and standard deviation of the three-axis acceleration.

[0017] Furthermore, in the above-mentioned method for identifying the motion state of the intelligent knee brace, the method for extracting the characteristic value of the three-axis acceleration is:

[0018] For the X-axis direction, first extract the maximum and minimum values ​​in a column of data and record them as Xmax and Xmin respectively, then calculate the average value Xavg of a column of data and the standard deviation Xstd of the column of data. The calculation formula of the average value Xavg is as follows:

[0019]

[0020] The calculation formula of standard deviation Xstd is as follows:

[0021]

[0022] Where N is the depth of a column of data;

[0023] Then use the same method to extract Ymax, Ymin, Yavg, Ystd in the Y-axis direction and Zmax, Zmin, Zavg, Zstd in the Z-axis direction.

[0024] Furthermore, in the above-mentioned method for identifying the motion state of the intelligent knee brace, the weighted processing method in step S23 includes the following steps:

[0025] A weighted coefficient table is set according to the probability relationship between the past project motion state and the current project motion state. The weighted coefficient table is shown in Table 2. The table includes three weighted factors representing the probability relationship between the past project motion state and the current project motion state. The three weighted factors are W1, W2, and W3, which are arranged from large to small.

[0026] According to the past project movement status, the weighted coefficient table is checked to obtain the weighted factor of the current project movement status, and then the weighted probability is calculated. The weighted probability calculation formula is as follows:

[0027] P′ i =P i ×Wi

[0028] Among them, i is the project category, P i is the original probability of item category i, W i is the weighting factor of project category i, P i ′ is the weighted probability of item category i;

[0029] Then the weighted probabilities of all items are normalized, and the normalized calculation formula is as follows:

[0030]

[0031] Among them, P i ” is the normalized final probability of item category i, j is the category index number, P j ′ is the sum of the weighted probabilities of all project categories, and k is the total number of categories; after normalization, the final probability of all projects is obtained, among which the highest probability result is the current project motion state.

[0032] Furthermore, in the above-mentioned method for identifying the motion state of the intelligent knee pad, when the motion state of the past project was still, the weighting factors of the motion state of the current project being still, walking, running, climbing stairs, and special sports are W1, W1, W2, W2, and W3 respectively; when the motion state of the past project was walking, the weighting factors of the motion state of the current project being still, walking, running, climbing stairs, and special sports are W1, W1, W1, W1, and W1 respectively; when the motion state of the past project was running, the weighting factors of the motion state of the current project being still, walking, running, climbing stairs, and special sports are W1, W1, W1, W3, and W3 respectively; when the motion state of the past project was climbing stairs, the weighting factors of the motion state of the current project being still, walking, running, climbing stairs, and special sports are W1, W2, W3, W1, and W3 respectively; when the motion state of the past project was special sports, the weighting factors of the motion state of the current project being still, walking, running, climbing stairs, and special sports are W3, W3, W3, W3, and W1 respectively.

[0033] Furthermore, in the above-mentioned method for identifying the motion state of the intelligent knee pad, the three weighting factors W1, W2, and W3 are adjustable, the value range of W1 is 1.5-2.0, the value range of W2 is 1.0-1.5, and the value range of W3 is 0.5-1.0.

[0034] The present invention also provides a motion state recognition system for an intelligent knee brace, which is arranged on an intelligent knee brace worn at the patellar tendon position of the knee, and includes a data acquisition module, a data processing module, and a motion recognition module. The data acquisition module includes a three-axis accelerometer for collecting three-axis acceleration data {gx, gy, gz} of a human body motion process; the data processing module extracts characteristic values ​​of the collected data, imports the characteristic values ​​into a model in a database to calculate the probability of each motion state, and the motion recognition module weights the probability of each motion state to obtain the maximum probability and recognize the current motion state.

[0035] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention deploys a model on the smart kneepad, which is fixed on the knee, and can accurately capture subtle movement changes of the knee joint, achieve high-precision monitoring of the knee joint state, and have a high recognition rate. It can accurately identify the wearer's current motion states such as stillness, walking, running, climbing stairs, and special sports, thereby providing targeted protection measures according to the current motion state to ensure the health of the knee joint; and the model is a static model, which does not need to be updated frequently, and can be deployed on an MCU-level main controller to reduce the size of the smart kneepad. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0037] Figure 1 This is a schematic diagram of the wearing position of the smart kneepad of the present invention;

[0038] Figure 2 It is a relationship diagram of collected data of the motion state identification method of the intelligent knee brace of the present invention;

[0039] Figure 3 A schematic diagram of a model deployment process of a motion state recognition method of an intelligent knee brace of the present invention;

[0040] Figure 4 Table 1 is a table of classification of items of the present invention;

[0041] Figure 5 Table 2 is the weighted coefficient table of the present invention. DETAILED DESCRIPTION

[0042] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0043] Example 1

[0044] A method for identifying the motion state of an intelligent knee brace comprises the following steps:

[0045] S1. Model establishment: The random forest method is used to build the model, including the following steps:

[0046] S11. Database acquisition: obtain the three-axis original motion information {gx, gy, gz} of the subject's knee through IIC communication and three-axis accelerometer communication, the unit is m / s 2 , the measurement range of the subject during the exercise is ±3g, where g is the value of gravity acceleration; let the subject complete multiple movements continuously during the specified exercise process, obtain the three-axis original motion information {gx, gy, gz} of the subject's knee, collect data with a specified depth of N as a column of data, collect multiple columns of data for each movement, and then integrate these data into a complete data set with action labels. The integration of data sets of multiple subjects constitutes a database;

[0047] like Figure 2 As shown in the figure, the subjects need to complete multiple sports actions during exercise, including standing, sitting, turning, shaking legs, walking at low speed, walking at high speed, walking and stopping, walking sideways, jogging, running, cycling, going upstairs, and going downstairs. By collecting these actions, a complete data set of the subjects is finally constructed.

[0048] S12, data processing and training, extracting a certain column of data in proportion from the data set with action labels in the above database and combining them into corresponding data with project labels; classifying sports movements with similar exercise intensity, and classifying and combining multiple sports movements completed by the subjects during exercise into 5 items, namely: stillness, walking, running, climbing stairs, and special sports, among which the stillness items include standing, sitting still, turning around, and shaking legs; the walking items include slow walking, high-speed walking, stop-and-go, and sideways walking; the running items include jogging and sprinting; the climbing stairs items include going upstairs and going downstairs; and the special sports items include cycling.

[0049] The specific project classification is shown in Table 1;

[0050] Table 1:

[0051]

[0052]

[0053] For each project, StandardScaler is used to standardize the data, and then the eigenvalues ​​of the three-axis acceleration are extracted from the standardized data. The extracted eigenvalues ​​of the three-axis acceleration are then imported into the random forest, and multiple decision trees are generated through the Bootstrap sampling technology. A model with a high accuracy is obtained through multiple trainings to achieve the establishment of the model.

[0054] S2, model deployment, such as Figure 3 As shown, the following steps are included:

[0055] S21, model conversion, convert the established model into C language through tensorflow lite so that it can be deployed on the embedded device side;

[0056] S22, model offline recognition, such as Figure 1 As shown, the smart knee brace is worn on the knee. The deployment end of the smart knee brace is equipped with a three-axis accelerometer. The three-axis accelerometer obtains the motion data of the knee in real time at a frequency of 100 Hz. The motion data is three-axis acceleration. The collected motion data will be pressed into the cache area of ​​the deployment end. When the amount of data in the cache area reaches the specified depth N, a data set is obtained, triggering a model operation. The model can be operated using learning models such as PF, MLP, and SVM. Specifically, the eigenvalues ​​of the three-axis acceleration in the data set are extracted. The eigenvalues ​​include the maximum value, minimum value, average value, and standard deviation of the three directions of XYZ, a total of 12 eigenvalues, and the eigenvalues ​​are normalized and input into the model in a standardized form. Then the model uses an integrated decision tree method to comprehensively analyze each eigenvalue and calculate the original probability of the motion state of each item.

[0057] S23, the model output result is post-processed, and the calculated original probability of the motion state of each item is weighted to obtain the final predicted probability. The state with the highest probability is recognized by the system as the current motion state of the item. Through weighted processing, the misjudgment rate is reduced.

[0058] The characteristic values ​​of the three-axis acceleration include the maximum value, minimum value, average value, and standard deviation of the three-axis acceleration; and the characteristic value extraction method of the three-axis acceleration in step S12 and step S22 is the same, which is as follows:

[0059] For the X-axis direction, first extract the maximum and minimum values ​​in a column of data and record them as Xmax and Xmin respectively, then calculate the average value Xavg of a column of data and the standard deviation Xstd of the column of data. The calculation formula of the average value Xavg is as follows:

[0060]

[0061] The calculation formula of standard deviation Xstd is as follows:

[0062]

[0063] Where N is the depth of a column of data;

[0064] Then use the same method to extract Ymax, Ymin, Yavg, Ystd in the Y-axis direction and Zmax, Zmin, Zavg, Zstd in the Z-axis direction.

[0065] The present invention deploys a model on an intelligent kneepad, which is fixed on the knee and can accurately capture subtle movement changes of the knee joint, achieve high-precision monitoring of the knee joint state, have a high recognition rate, and can accurately identify the wearer's current motion states such as stillness, walking, running, climbing stairs, and special sports, thereby providing targeted protection measures according to the current motion state to ensure the health of the knee joint; and the model is a static model, which does not need to be updated frequently, and can be deployed on an MCU-level main controller to reduce the size of the intelligent kneepad.

[0066] Example 2

[0067] Based on the structure of Example 1, Figure 1-3 As shown, the weighted processing method in step S23 includes the following steps:

[0068] A weighting coefficient table is set according to the probabilistic relationship between the motion state of past items and the motion state of current items. The table contains three weighting factors representing the probabilistic relationship between the motion state of past items and the motion state of current items. The three weighting factors are W1, W2, and W3, which are arranged from large to small. The three weighting factors W1, W2, and W3 are adjustable according to actual effects. The value range of W1 is 1.5-2.0, the value range of W2 is 1.0-1.5, and the value range of W3 is 0.5-1.0.

[0069] When the past project movement state was still, the weighting factors of the current project movement state still, walking, running, climbing, and special sports are W1, W1, W2, W2, and W3 respectively; when the past project movement state was walking, the weighting factors of the current project movement state still, walking, running, climbing, and special sports are W1, W1, W1, W1, and W1 respectively; when the past project movement state was running, the weighting factors of the current project movement state still, walking, running, climbing, and special sports are W1, W1, W1, W3, and W3 respectively; when the past project movement state was climbing, the weighting factors of the current project movement state still, walking, running, climbing, and special sports are W1, W2, W3, W1, and W3 respectively; when the past project movement state was special sports, the weighting factors of the current project movement state still, walking, running, climbing, and special sports are W3, W3, W3, W3, and W1 respectively. See Table 2 of Weighted Coefficients for details.

[0070] Table 2:

[0071]

[0072]

[0073] According to the past project movement status, the weighted coefficient table is checked to obtain the weighted factor of the current project movement status, and then the weighted probability is calculated. The weighted probability calculation formula is as follows:

[0074] P′ i =P i ×W i

[0075] Among them, i is the project category, P i is the original probability of item category i, W i is the weighting factor of project category i, P i ′ is the weighted probability of item category i;

[0076] Then the weighted probabilities of all items are normalized, and the normalized calculation formula is as follows:

[0077]

[0078] Among them, P i ” is the normalized final probability of item category i, j is the category index number, P j ′ is the sum of the weighted probabilities of all project categories, and k is the total number of categories; after normalization, the final probability of all projects is obtained, among which the highest probability result is the current project motion state.

[0079] For example, if the current user changes from running to walking, the corresponding workflow of the entire recognition module is as follows:

[0080] First, the MCU communicates with the three-axis accelerometer to obtain the three-axis original motion information. After reaching the specified depth, feature extraction is performed to obtain 12 feature data, which are imported into the model for calculation. After the model calculation is completed, the original probability P of 5 specified motion states is output. i ; Assumption: P i ={0.1, 0.4, 0.3, 0.1, 0.1}, which are the original probabilities of stationary, walking, running, climbing stairs and special sports respectively. Let W1 = 2.0, W2 = 1.0, W3 = 0.5, and calculate the weighted probability P i ′={0.2, 0.8, 0.6, 0.05, 0.05}, after normalization, P i "={0.12, 0.47, 0.35, 0.03, 0.03}, among which, the final probability of the walking project motion state P"=0.8 / (0.2+0.8+0.6+0.05+0.05)=0.47, the probability of walking 0.47 is the largest, then the system outputs the walking project motion state as the current motion state.

[0081] Through this method, the smart knee pad can accurately identify the user's current movement state, so as to provide corresponding protection measures for the user's knee according to the current movement state.

[0082] The present invention also provides a motion state recognition system for an intelligent knee brace, which is arranged on an intelligent knee brace worn at the patellar tendon position of the knee, and includes a data acquisition module, a data processing module, and a motion recognition module. The data acquisition module includes a three-axis accelerometer for collecting three-axis acceleration data {gx, gy, gz} of a human body motion process; the data processing module extracts characteristic values ​​of the collected data, imports the characteristic values ​​into a model in a database to calculate the probability of each motion state, and the motion recognition module weights the probability of each motion state to obtain the maximum probability and recognize the current motion state.

[0083] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

[0084] In addition, it should be understood that although the present specification is described according to implementation modes, not every implementation mode contains only one independent technical solution. This description of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment may also be appropriately combined to form other implementation modes that can be understood by those skilled in the art.

Claims

1. A method for identifying the motion state of an intelligent kneepad, characterized in that: The following steps are involved: S1. Model establishment: The random forest method is used to build the model, including the following steps: S11, database acquisition, let the subject complete multiple motions continuously in the specified motion process, obtain the three-axis original motion information {gx, gy, gz} of the subject's knee, collect data with a specified depth of N as a column of data, collect multiple columns of data for each motion action, and then integrate these data into a complete data set with motion labels, and integrate the data sets of multiple subjects to form a database; S12, data processing and training, extracting a certain column of data from the data set with action labels in the above database in equal proportion and combining them into corresponding data with project labels, using StandardScaler to standardize the data for each project, and then extracting the eigenvalues ​​of the three-axis acceleration from the standardized data; then importing the extracted eigenvalues ​​of the three-axis acceleration into the random forest, generating multiple decision trees through the Bootstrap sampling technology, and obtaining a model with a high accuracy through multiple trainings to achieve the establishment of the model; S2, model deployment, including the following steps: S21, model conversion, convert the established model into C language through tensorflow lite so that it can be deployed on the embedded device side; S22, model offline identification, the deployment end of the smart knee brace worn on the knee is equipped with a three-axis accelerometer, which acquires the motion data of the knee in real time at a frequency of 100Hz, and the motion data is three-axis acceleration; the collected motion data will be pressed into the cache area of ​​the deployment end, and when the amount of data in the cache area reaches the specified depth N, a data set is obtained, triggering a model operation, extracting the characteristic value of the three-axis acceleration in the data set, and normalizing this characteristic value, and inputting it into the model in a standardized form; the model uses an integrated decision tree method to calculate the original probability of the motion state of each item; S23, the model output result is post-processed, and the calculated original probability of the motion state of each item is weighted to obtain the final predicted probability. The state with the highest probability is recognized by the system as the current motion state of the item.

2. The method for identifying the motion state of the intelligent knee brace according to claim 1, characterized in that: When acquiring the database, the three-axis original motion information {gx, gy, gz} of the subject's knee is obtained through IIC communication and communication with the three-axis accelerometer. The measurement range of the subject during movement is ±3g, where g is the gravity acceleration value.

3. The method for identifying the motion state of the intelligent knee brace according to claim 1, characterized in that: The subjects need to complete multiple sports movements during the exercise, including: standing, sitting still, turning around, shaking legs, walking at a low speed, walking at a high speed, walking and stopping, walking sideways, jogging, running, cycling, going upstairs, and going downstairs.

4. The method for identifying the motion state of the intelligent knee brace according to claim 3, characterized in that: The multiple sports movements completed by the subjects during exercise are classified into five items, namely: stillness, walking, running, stair climbing, and special sports. The stillness items include standing, sitting still, turning around, and shaking legs. The walking items include slow walking, high-speed walking, stop-and-go, and sideways walking. The running items include jogging and sprinting. The stair climbing items include going upstairs and going downstairs. Special sports items include cycling.

5. The method for identifying the motion state of the intelligent knee brace according to claim 1, characterized in that: The characteristic values ​​of the three-axis acceleration include the maximum value, minimum value, average value, and standard deviation of the three-axis acceleration.

6. The method for identifying the motion state of the intelligent knee brace according to claim 5, characterized in that: The method for extracting the eigenvalues ​​of the three-axis acceleration is: For the X-axis direction, first extract the maximum and minimum values ​​in a column of data, record them as Xmax and Xmin respectively, then calculate the average value Xavg of a column of data and the standard deviation Xstd of the column of data. The calculation formula of the average value Xavg is as follows: The calculation formula of standard deviation Xstd is as follows: Where N is the depth of a column of data; Then use the same method to extract Ymax, Ymin, Yavg, Ystd in the Y-axis direction and Zmax, Zmin, Zavg, Zstd in the Z-axis direction.

7. The method for identifying the motion state of the intelligent knee brace according to claim 1, characterized in that: The weighted processing method comprises the following steps: A weighting coefficient table is set according to the probability relationship between the past project motion state and the current project motion state. The table contains three weighting factors representing the probability relationship between the past project motion state and the current project motion state. The three weighting factors are W1, W2, and W3, which are arranged from large to small. According to the past project movement status, the weighted coefficient table is checked to obtain the weighted factor of the current project movement status, and then the weighted probability is calculated. The weighted probability calculation formula is as follows: P′ i =P i ×W i Among them, i is the project category, P i is the original probability of item category i, W i is the weighting factor of project category i, P i ′ is the weighted probability of item category i; Then the weighted probabilities of all items are normalized, and the normalized calculation formula is as follows: Among them, P i ” is the normalized final probability of item category i, j is the category index number, P j ′ is the sum of the weighted probabilities of all project categories, and k is the total number of categories; after normalization, the final probability of all projects is obtained, among which the highest probability result is the current project motion state.

8. The method for identifying the motion state of the intelligent knee brace according to claim 7, characterized in that: When the past project movement state was still, the weighting factors of the current project movement state still, walking, running, climbing stairs, and special sports are W1, W1, W2, W2, and W3 respectively; when the past project movement state was walking, the weighting factors of the current project movement state still, walking, running, climbing stairs, and special sports are W1, W1, W1, W1, and W1 respectively; When the past project motion status was running, the weighting factors of the current project motion status of stillness, walking, running, stair climbing and special sports were W1, W1, W1, W3, W3 respectively; when the past project motion status was stair climbing, the weighting factors of the current project motion status of stillness, walking, running, stair climbing and special sports were W1, W2, W3, W1, W3 respectively; when the past project motion status was special sports, the weighting factors of the current project motion status of stillness, walking, running, stair climbing and special sports were W3, W3, W3, W3, W1 respectively.

9. The method for identifying the motion state of the intelligent knee brace according to claim 7, characterized in that: The three weighting factors W1, W2 and W3 are adjustable, with W1 ranging from 1.5 to 2.0, W2 ranging from 1.0 to 1.5, and W3 ranging from 0.5 to 1.

0.

10. A motion state recognition system for an intelligent kneepad, characterized in that: The motion state recognition system is arranged on an intelligent knee brace worn at the patellar tendon position of the knee, and includes a data acquisition module, a data processing module, and a motion recognition module. The data acquisition module includes a three-axis accelerometer for collecting three-axis acceleration data {gx, gy, gz} of the human body's motion process; the data processing module extracts the characteristic values ​​of the collected data, imports the characteristic values ​​into a model in a database to calculate the probability of each motion state; the motion recognition module weights the probability of each motion state to obtain the maximum probability and recognizes the current motion state.

Images