A gait analysis and anomaly detection system and method for an intelligent wearable device

Abstract

The present invention relates to the technical field of abnormal gait detection, and particularly to a gait analysis and abnormal detection system and method for intelligent wearable devices. The data acquisition module is used to collect motion, physiological data and road condition information; the moving speed calculation module is used to calculate the moving speed; the gait prediction module is used to obtain the predicted gait and predicted movement amount; the gait feature extraction module is used to extract joint vectors and joint angle features; the abnormal gait judgment module is used to judge whether there is an abnormal gait. The present invention constructs a gait space equation and uses a neural network model for gait prediction to capture the time dependence and non-linear relationship of the gait sequence, obtaining a more accurate prediction result; by comparing the similarity between the predicted gait features and the actual gait features, and between the predicted movement amount and the actual movement amount, and comprehensively using the methods of joint vectors, joint angles and movement amount features, it more comprehensively identifies various abnormal gait patterns, improving the accuracy and reliability of abnormal detection.

Description

Technical Field

[0001] The present invention relates to the technical field of abnormal gait detection, and particularly to a gait analysis and abnormal detection system for intelligent wearable devices and a gait analysis and abnormal detection method for intelligent wearable devices. Background Art

[0002] With the development of technology, intelligent wearable devices have been widely used in the fields of health monitoring, behavior analysis, etc. Among them, gait analysis is a very valuable technical means. Human gait contains a large amount of information about an individual's physical condition, balance ability, nervous system operation, etc. Through detailed analysis of gait, some health hazards can be detected, providing valuable basis for disease prevention, rehabilitation treatment, etc.

[0003] In gait analysis technology, the key is to extract and quantify gait characteristic parameters. Common gait characteristics include step frequency, step length, symmetry of double - leg swing, joint angle change, etc. These characteristics reflect the operation status of various joints, muscles, and the nervous system of the human body. By measuring and analyzing these characteristics, some abnormal gait patterns can be found, thereby inferring possible health problems.

[0004] Existing gait abnormal detection systems mainly collect motion data such as plantar pressure, step frequency, step length, etc. It is difficult to capture the time - dependence and non - linear relationship of gait sequences, resulting in inaccurate detection results. Summary of the Invention

[0005] The present invention provides a gait analysis and abnormal detection system and method for intelligent wearable devices to solve the defect of inaccurate detection results in the prior art.

[0006] On the one hand, the present invention provides a gait analysis and abnormal detection system for intelligent wearable devices, including:

[0007] A data acquisition module for collecting the motion data, physiological data, and road condition information of the subject.

[0008] A moving speed calculation module for establishing a spatial coordinate system based on the road surface where the subject is located as the reference plane, and calculating the moving speed of the subject according to the gait changes of the subject at two adjacent moments in the spatial coordinate system.

[0009] The gait prediction module is used to construct a gait space equation based on the position coordinates of the subject, obtain gait variables, take the derivative of the gait variables, and then convert the gait space equation into a gait transition equation. A prediction model based on a neural network is constructed, and the gait variable road condition information is input to output the next moment movement function. Any two movement points are selected in the movement function and expanded to obtain a gait prediction equation, which includes the predicted gait at the next moment and the predicted movement amount in the next time period.

[0010] The gait feature extraction module is used to extract the predicted gait features of the predicted gait at the next moment and the actual gait features of the actual gait of the subject at the next moment. The gait features include joint vectors and joint angle features.

[0011] The abnormal gait judgment module is used to obtain an abnormal detection score according to the similarity between the predicted gait features and the actual gait features and the similarity between the predicted movement amount and the actual movement amount, and compare the abnormal detection score with a preset abnormal detection score threshold. If the abnormal detection score exceeds the abnormal detection score threshold range, it is judged as an abnormal gait.

[0012] According to a gait analysis and abnormal detection system of an intelligent wearable device provided by the present invention, the moving speed includes a linear speed and an angular speed. The linear speed is the average value of the moving speed of the subject's left foot and the moving speed of the right foot. The angular speed is obtained according to the ratio of the angle turned by the subject at two adjacent moments to the time.

[0013] According to a gait analysis and abnormal detection system of an intelligent wearable device provided by the present invention, the gait space equation includes a position equation and a movement equation. The position equation is used to represent the position coordinates of the subject and the current deflection angle. The movement equation is used to represent the moving speed and the turning angle of the subject. The gait space equation is expressed as:

[0014] ;

[0015] In the formula, K represents the gait variable, x represents the abscissa, y represents the ordinate, z represents the vertical coordinate, δ represents the deflection angle, δ = (α, β, γ), S represents the movement variable, v represents the linear speed, represents the steering angle.

[0016] According to a gait analysis and abnormal detection system of an intelligent wearable device provided by the present invention, the gait transition equation is expressed as:

[0017] ;

[0018] In the formula, represents the lateral speed, represents the longitudinal speed, represents the vertical speed, represents the angular velocity, and d represents the distance between the movement trajectories of the left and right feet.

[0019] According to a gait analysis and anomaly detection system for an intelligent wearable device provided by the present invention, the gait prediction equation includes a predicted gait equation and a predicted movement equation. The predicted gait equation is used to characterize the variables of the current gait and movement amount to obtain the predicted gait of the subject at the next moment. The predicted movement equation is used to characterize the predicted movement amount in the next time period when the subject moves on an ideal movement trajectory. The gait prediction equation is expressed as:

[0020] ;

[0021] ;

[0022] In the formula, represents the predicted gait at the next moment, is the current gait, represents the variable of the movement amount, A represents the gait matrix, B represents the movement matrix, is the current movement amount, and C is the output matrix.

[0023] According to a gait analysis and anomaly detection system for an intelligent wearable device provided by the present invention, the predicted gait at the next moment includes the predicted gait amount and the predicted movement amount of the subject at the next moment. The predicted gait at the next moment is expressed as:

[0024] ;

[0025] In the formula, represents the predicted gait amount at the next moment, represents the predicted movement amount in the next time period.

[0026] According to a gait analysis and anomaly detection system for an intelligent wearable device provided by the present invention, the extraction process of the joint vector includes:

[0027] Obtain the three-dimensional coordinates of the center point of the subject's joint in the space coordinate system.

[0028] Set the three-dimensional coordinates of any two joints as P i = (x i , y i , z i ), P j = (x j , y j , z j ), then the joint vector of joints P i and P j is expressed as u ij = (x j - x i, y j -y i , z j -z i ).

[0029] Calculate the joint vectors of every two joints of the subject respectively to form a joint vector set U = {u ij | i, j = 1, 2, 3 ∧ i < j}.

[0030] The process of extracting joint angle features includes:

[0031] Set any two joint vectors u A and u B , then the angle feature between u A and u B is expressed as:

[0032] ;

[0033] Calculate the angle features between the joint vectors of the subject respectively to form an angle feature set R = {r uA，uB | u A , u B ∈U ∧ u A ≠ u B}.

[0034] According to a gait analysis and abnormality detection system of an intelligent wearable device provided by the present invention, the process of calculating the similarity between the predicted gait features and the actual gait features includes:

[0035] Use the Pearson correlation coefficient to calculate the similarity of the joint vectors in the predicted gait features and the actual gait features.

[0036] Use the cosine similarity algorithm to calculate the similarity of the joint angle features in the predicted gait features and the actual gait features.

[0037] Use the relative difference percentage to calculate the similarity of the predicted movement amount and the actual movement amount.

[0038] According to a gait analysis and abnormality detection system of an intelligent wearable device provided by the present invention, the process of determining whether the subject has abnormal gait includes:

[0039] Perform weighted summation on the joint vector similarity, joint angle similarity and movement amount similarity to obtain an abnormality detection score.

[0040] Compare the abnormality detection score with a preset abnormality detection score threshold. If the abnormality detection score exceeds the abnormality detection score threshold range, it is determined as abnormal gait.

[0041] On the other hand, the present invention also provides a gait analysis and abnormality detection method for an intelligent wearable device, including:

[0042] S1. Collect the motion data, physiological data, and road condition information of the subject.

[0043] S2. Calculate the moving speed of the subject, including:

[0044] S21. Establish a spatial coordinate system with the road surface where the subject is located as the reference plane.

[0045] S22. Calculate the moving speed of the subject according to the gait changes of the subject at two adjacent moments in the spatial coordinate system.

[0046] S3. Predict the gait of the subject at the next moment, including:

[0047] S31. Construct a gait space equation describing the gait change law according to the position coordinates of the subject to obtain gait variables.

[0048] S32. Take the derivative of the gait variables, and then convert the gait space equation into a gait transition equation.

[0049] S33. Construct a prediction model based on a neural network, input the gait variables and road condition information, and output the moving function at the next moment.

[0050] S34. Select any two moving points in the moving function and expand the two moving points to obtain a gait prediction equation, which includes the predicted gait at the next moment and the predicted moving amount in the next time period.

[0051] S4. Extract the predicted gait features of the predicted gait at the next moment and the actual gait features of the actual gait of the subject at the next moment. The gait features include joint vectors and joint angle features.

[0052] S5. Obtain an anomaly detection score according to the similarity between the predicted gait features and the actual gait features, and the similarity between the predicted moving amount and the actual moving amount.

[0053] S6. Compare the anomaly detection score with a preset anomaly detection score threshold. If the anomaly detection score exceeds the anomaly detection score threshold range, it is determined as an abnormal gait.

[0054] A gait analysis and anomaly detection system and method for an intelligent wearable device provided by the present invention collect the motion data of the subject, combine the physiological data and road condition information, provide rich input information for subsequent gait analysis and anomaly detection, and help to more accurately characterize the gait features of the subject. By establishing a moving speed calculation module based on the spatial coordinate system where the subject is located. The method based on the spatial coordinate system can better reflect the motion state of the subject in the actual environment and reduce the measurement error caused by the change of the measurement environment.

[0055] The present invention constructs a gait space equation, converts it into a gait transition equation, and then uses a neural network model for gait prediction. Compared with traditional statistical prediction models, it can better capture the time dependence and non-linear relationship of gait sequences, thereby obtaining more accurate prediction results.

[0056] The present invention compares the similarity between predicted gait features and actual gait features, and introduces the similarity between predicted movement amount and actual movement amount as the detection basis. By comprehensively using the methods of joint vector features, joint angle features and movement amount features, it can more comprehensively identify various abnormal gait patterns and improve the accuracy and reliability of abnormal detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0058] Figure 1 is a schematic structural diagram of a gait analysis and abnormal detection system of an intelligent wearable device provided by an embodiment of the present invention;

[0059] Figure 2 is a schematic flowchart of a gait analysis and abnormal detection method of an intelligent wearable device provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0060] To make the objectives, technical solutions and advantages of the present invention clearer, the following will clearly and completely describe the technical solutions in the present invention with reference to the drawings in the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art without creative efforts based on the embodiments of the present invention belong to the scope of protection of the present invention.

[0061] The following will be combined with Figure 1 - Figure 2 to describe a gait analysis and abnormal detection system and method of an intelligent wearable device of the present invention.

[0062] Figure 1 is a schematic structural diagram of a gait analysis and abnormal detection system of an intelligent wearable device provided by an embodiment of the present invention.

[0063] As Figure 1As shown in the figure, a gait analysis and anomaly detection system and method for a smart wearable device provided by an embodiment of the present invention, the execution subject can be a gait analysis and anomaly detection system for a smart wearable device, including: a data acquisition module, a moving speed calculation module, a gait prediction module, a gait feature extraction module, and an abnormal gait judgment module.

[0064] The data acquisition module is used to collect the movement data, physiological data, and road condition information of the subject. The movement data includes the position information, running direction, etc. of the subject. Among them, the position information can be obtained by installing a GPS locator, and the running direction can be obtained by installing a direction recognition detector. The physiological data includes the distance between the moving trajectories of the subject's left and right feet, etc. The movement data and road condition information can be obtained by installing corresponding sensors. For example, the road condition information can be obtained by installing multiple cameras, radar detectors, etc. on the subject.

[0065] The moving speed calculation module is used to establish a space coordinate system based on the road surface where the subject is located as the reference plane, map the corresponding gait coordinates in the space coordinate system according to the position coordinates and the traveling direction of the subject at each moment, and calculate the moving speed of the subject according to the gait changes of the subject at two adjacent moments.

[0066] Arbitrarily select a point on the moving road surface as the origin, and arbitrarily select three mutually perpendicular directions on the road surface as the X-axis, Y-axis, and Z-axis respectively to form a space coordinate system based on the road surface. Then the position of the subject relative to the origin is also the position coordinate, denoted as (x i ,y i ,z i ).

[0067] The moving speed includes linear speed and angular speed. The linear speed is the average value of the moving speed of the subject's left foot and the moving speed of the subject's right foot. The angular speed is obtained according to the ratio of the angle turned by the subject at two adjacent moments to the time.

[0068] Assume that the sampling time interval between two adjacent moments is very small. Through geometric relationships, it can be obtained that:

[0069] ;

[0070] Expanding the above formula gives:

[0071] ;

[0072] According to the speed formula, it can be obtained that:

[0073] ;

[0074] Then the formula expression of the moving speed is:

[0075] ;

[0076] ;

[0077] Wherein, v represents the linear velocity, ω represents the angular velocity, represents the right - foot velocity, represents the left - foot velocity, d represents the distance between the moving trajectories of the left and right feet, θ represents the angle turned between two adjacent moments, and represents the difference in the distances moved by the subject's left and right feet within two adjacent moments.

[0078] The gait prediction module is used to construct a gait space equation based on the position coordinates, obtain gait variables, take the derivative of the gait variables, and then convert the gait space equation into a gait transition equation. And construct a prediction model based on a neural network, input the gait variable road condition information, and output the moving function at the next moment. Arbitrarily select two moving points in the moving function and expand the two moving points to obtain a gait prediction equation, and then obtain the predicted gait of the subject at the next moment and the predicted movement amount in the next time period.

[0079] The gait space equation includes a position equation and a movement equation. The position equation is used to characterize the position coordinates and the current deflection angle of the subject. The movement equation is used to characterize the moving speed and the turning angle of the subject. The gait space equation is expressed as:

[0080] ;

[0081] Wherein, K represents the gait variable, x represents the abscissa, y represents the ordinate, z represents the vertical coordinate, δ represents the deflection angle, δ = (α, β, γ), S represents the movement variable, represents the steering angle.

[0082] The gait transition equation is expressed as:

[0083] ;

[0084] Wherein, represents the lateral velocity, represents the longitudinal velocity, represents the vertical velocity, represents the angular velocity of rotation.

[0085] Construct a prediction model based on a neural network, input the gait variable road condition information, and output the moving function at the next moment. Among them, the neural network includes multiple activation functions for predicting the gait amount at the next moment. The gait amount at the next moment is expressed through the moving function as:

[0086] ;

[0087] Wherein, is the movement function at time t + 1, and are the gait quantity and movement quantity at time t + 1 respectively.

[0088] Arbitrarily select two points in this movement function: and , where and are the gait quantities at arbitrarily selected times n and m respectively, and are the movement quantities at arbitrarily selected times n and m respectively.

[0089] Expand the two selected points on , that is:

[0090] ;

[0091] In the formula, is the gait quantity at time m - 1.

[0092] After sorting, we can get:

[0093] ;

[0094] Let the predicted value of the gait quantity , the predicted value of the movement quantity , where is the movement quantity at time m, is the movement quantity at time n. Introduce the sampling time T, and, , where is the predicted value of the movement quantity in the previous time period.

[0095] We can get:

[0096] ;

[0097] In the formula, represents the current gait quantity, represents the movement quantity, A 1 is the first gait matrix, B 1 is the first movement matrix.

[0098] In the process of gait prediction, it is necessary to know all the predicted quantities and gait quantities in the adjacent time periods, and then find the optimal value of the cost matrix. Therefore, it is necessary to construct a new gait quantity. The final gait prediction equation is expressed as:

[0099] ;

[0100] ;

[0101] In the formula, Indicates the predicted gait at the next moment, is the current gait, The variable representing the amount of movement, A represents the gait matrix, B represents the movement matrix, represents the current amount of movement, and C represents the output matrix.

[0102] The optimal solution of the gait prediction equation is obtained to get the predicted gait at the next moment.

[0103] Define Γ, Λ, Υ as intermediate transition matrices, and:

[0104] ;

[0105] ;

[0106] ;

[0107] Define ;

[0108] Then ;

[0109] Set the optimization objective function as:

[0110] ;

[0111] In the formula, represents the gait amount at the previous moment, represents the weight coefficient, represents the relaxation factor, E represents the identity matrix, M represents the gait weighting matrix, and G represents the movement weighting matrix.

[0112] Let the error term ,

[0113] ;

[0114] Model this problem as a quadratic programming problem with constraints, then:

[0115] ;

[0116] ;

[0117] In the formula, J represents the optimization function, Q represents the quadratic term matrix, represents the linear term matrix, , , , represent the constraint matrices.

[0118] Let ;

[0119] Then ;

[0120] Optimize and solve the predicted gait quantity and predicted movement quantity of the subject at the next moment, so as to obtain the predicted gait at the next moment. .

[0121] The gait feature extraction module is used to extract the predicted gait features of the predicted gait at the next moment and the actual gait features of the actual gait of the subject at the next moment. The gait features include joint vectors and joint angle features.

[0122] The extraction process of the joint vector includes:

[0123] Obtain the three-dimensional coordinates of the center point of the subject's joint in the space coordinate system.

[0124] In this embodiment, the joints may include hip joints, knee joints, ankle joints, etc.

[0125] Set the three-dimensional coordinates of any two joints as P i = (x i , y i , z i ), P j = (x j , y j , z j ), then the joint vector of joints P i and P j is represented as u ij = (x j - x i , y j - y i , z j - z i ).

[0126] Calculate the joint vectors of every two joints of the subject respectively to form a joint vector set U = {u ij |i, j = 1, 2, 3 ∧ i < j}.

[0127] The extraction process of the joint angle feature includes:

[0128] Set any two joint vectors u A and u B , then the angle feature between u A and u B is represented as:

[0129] ;

[0130] Calculate the angle features between the joint vectors of each joint of the subject respectively to form an angle feature set R = {r uA，uB |u A , uB ∈U ∧ u A ≠ u B}。

[0131] The abnormal gait judgment module is used to obtain an abnormal detection score based on the similarity between the predicted gait features and the actual gait features, and the similarity between the predicted movement amount and the actual movement amount, and compare the abnormal detection score with a preset abnormal detection score threshold. If the abnormal detection score exceeds the abnormal detection score threshold range, it is judged as an abnormal gait.

[0132] The process of calculating the similarity between the predicted gait features and the actual gait features includes:

[0133] Use the Pearson correlation coefficient to calculate the similarity of the joint vectors in the predicted gait features and the actual gait features. The formula is expressed as:

[0134] ;

[0135] In the formula, represents the actual joint vector, represents the predicted joint vector, represents the total number of joint vectors.

[0136] Use the cosine similarity algorithm to calculate the similarity of the joint angle features in the predicted gait features and the actual gait features. The formula is expressed as:

[0137] ;

[0138] In the formula, represents the angle feature between the joint vectors and in the actual gait, represents the angle feature between the joint vectors and in the predicted gait, represents the total number of angle features.

[0139] Use the relative difference percentage to calculate the similarity between the predicted movement amount and the actual movement amount. The formula is expressed as:

[0140] ;

[0141] In the formula, represents the predicted movement amount in the next time period, represents the actual movement amount in the next time period.

[0142] The process of judging whether the subject has abnormal gait includes:

[0143] The joint vector similarity, joint angle similarity, and movement deviation score are weighted and summed to obtain an anomaly detection score, which is expressed by the formula:

[0144] ;

[0145] In the formula, , and represent weight coefficients.

[0146] The anomaly detection score is compared with a preset anomaly detection score threshold. If the anomaly detection score exceeds the anomaly detection score threshold range, it is determined as an abnormal gait.

[0147] In summary, this embodiment provides a gait analysis and anomaly detection system for an intelligent wearable device. By collecting the movement data of the subject, combining physiological data and road condition information, it provides rich input information for subsequent gait analysis and anomaly detection, which helps to more accurately characterize the gait characteristics of the subject. By establishing a moving speed calculation module based on the space coordinate system where the subject is located. The method based on the space coordinate system can better reflect the movement state of the subject in the actual environment and reduce the measurement error caused by the change of the measurement environment. By constructing a gait space equation and converting it into a gait transition equation, and then using a neural network model for gait prediction. Compared with the traditional statistical prediction model, it can better capture the time dependence and non-linear relationship of the gait sequence, so as to obtain more accurate prediction results. It not only compares the similarity between the predicted gait characteristics and the actual gait characteristics, but also introduces the similarity between the predicted movement amount and the actual movement amount as the detection basis. The method of comprehensively using joint vector features, joint angle features, and movement amount features can more comprehensively identify various abnormal gait patterns and improve the accuracy and reliability of anomaly detection.

[0148] Based on the same general inventive concept, the present invention also protects a gait analysis and anomaly detection method for an intelligent wearable device. The following describes a gait analysis and anomaly detection method for an intelligent wearable device provided by the present invention. The gait analysis and anomaly detection method for an intelligent wearable device described below can be mutually referred to the gait analysis and anomaly detection system described above.

[0149] Figure 2 is a schematic flowchart of a gait analysis and anomaly detection method for an intelligent wearable device provided by an embodiment of the present invention.

[0150] As Figure 2 shown, the gait analysis and anomaly detection method for an intelligent wearable device includes:

[0151] S1. Collect the movement data, physiological data, and road condition information of the subject.

[0152] S2. Calculate the moving speed of the subject, including:

[0153] S21. Establish a spatial coordinate system based on the road surface where the subject is located as the reference plane.

[0154] S22. Calculate the moving speed of the subject according to the gait changes of the subject at two adjacent moments in the spatial coordinate system.

[0155] S3. Predict the gait of the subject at the next moment, including:

[0156] S31. Construct a gait space equation describing the law of gait change according to the position coordinates of the subject to obtain gait variables.

[0157] S32. Take the derivative of the gait variables, and then convert the gait space equation into a gait transition equation.

[0158] S33. Construct a prediction model based on a neural network, input the road condition information of the gait variables, and output the moving function at the next moment.

[0159] S34. Select any two moving points in the moving function and expand the two moving points to obtain a gait prediction equation, which includes the predicted gait at the next moment and the predicted moving amount in the next time period.

[0160] S4. Extract the predicted gait features of the predicted gait at the next moment and the actual gait features of the actual gait of the subject at the next moment. The gait features include joint vectors and joint angle features.

[0161] S5. Obtain an anomaly detection score according to the similarity between the predicted gait features and the actual gait features, and the similarity between the predicted moving amount and the actual moving amount.

[0162] S6. Compare the anomaly detection score with a preset anomaly detection score threshold. If the anomaly detection score exceeds the anomaly detection score threshold range, it is determined as an abnormal gait.

[0163] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the above technical solution, in essence, or the part that makes a contribution to the prior art can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0164] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A gait analysis and anomaly detection system for smart wearable devices, characterized in that: include: A data collection module is used to collect the subjects' exercise data, physiological data and road condition information; A moving speed calculation module, used to establish a spatial coordinate system based on the road surface where the subject is located as a reference plane, and calculate the moving speed of the subject according to the gait changes of the subject at two adjacent moments in the spatial coordinate system; A gait prediction module is used to construct a gait space equation according to the position coordinates of the subject, obtain gait variables, and derive the gait variables to convert the gait space equation into a gait transfer equation; and to construct a prediction model based on a neural network, input the gait variables and the road condition information, and output a movement function at the next moment; select any two movement points in the movement function, and expand the two movement points to obtain a gait prediction equation, wherein the gait prediction equation includes the predicted gait at the next moment and the predicted movement amount in the next time period; A gait feature extraction module, used to extract the predicted gait features of the predicted gait at the next moment and the actual gait features of the subject's actual gait at the next moment, wherein the gait features include joint vectors and joint angle features; The abnormal gait judgment module is used to obtain an abnormal detection score based on the similarity between the predicted gait feature and the actual gait feature, and the similarity between the predicted movement amount and the actual movement amount, and compare the abnormal detection score with a preset abnormal detection score threshold. If the abnormal detection score exceeds the abnormal detection score threshold range, it is judged as an abnormal gait.

2. The gait analysis and abnormality detection system for a smart wearable device according to claim 1, characterized in that: The moving speed includes linear speed and angular speed. The linear speed is the average of the moving speed of the subject's left foot and the moving speed of the right foot. The angular speed is obtained according to the ratio of the angle turned by the subject at two adjacent moments to the time.

3. The gait analysis and abnormality detection system for a smart wearable device according to claim 1, characterized in that: The gait space equation includes a position equation and a movement equation; the position equation is used to characterize the position coordinates and current deflection angle of the subject; the movement equation is used to characterize the movement speed and turning angle of the subject; the gait space equation is expressed as: ； Where K represents the gait variable, x represents the horizontal coordinate, y represents the vertical coordinate, z represents the vertical coordinate, δ represents the deflection angle, δ=(α,β,γ), S represents the movement variable, v represents the linear velocity, Indicates the steering angle.

4. The gait analysis and abnormality detection system for a smart wearable device according to claim 1, characterized in that: The gait transfer equation is expressed as: ； In the formula, represents the lateral velocity, represents the longitudinal velocity, represents the vertical velocity, represents the angular velocity, represents the steering angle, α, β, γ represent the components of the deflection angle δ on the X, Y, and Z axes, and d represents the distance between the moving trajectories of the left and right feet.

5. The gait analysis and abnormality detection system for a smart wearable device according to claim 1, characterized in that: The gait prediction equation includes a predicted gait equation and a predicted movement equation; the predicted gait equation is used to characterize the variables according to the current gait and movement amount, and obtain the predicted gait of the subject at the next moment; the predicted movement equation is used to characterize the predicted movement amount in the next period when the subject moves on the ideal movement trajectory; the gait prediction equation is expressed as: ； ； In the formula, represents the predicted gait at the next moment, is the current gait, represents the variable of movement, A represents the gait matrix, B represents the movement matrix, is the current movement amount, and C is the output matrix.

6. The gait analysis and abnormality detection system for a smart wearable device according to claim 1, characterized in that: The predicted gait at the next moment includes the predicted gait amount and predicted movement amount of the subject at the next moment; the predicted gait at the next moment Expressed as: ； In the formula, represents the predicted gait amount at the next moment, Indicates the predicted movement amount for the next period.

7. The gait analysis and abnormality detection system for a smart wearable device according to claim 1, characterized in that: The process of extracting the joint vector includes: Obtaining the three-dimensional coordinates of the center point of the subject's joint in the spatial coordinate system; Set the three-dimensional coordinates of any two joints to P i =(x i ,y i , z i ), P j =(x j ,y j , z j ), then the joint P i and P j The joint vector is represented as u ij =(x j -x i ,y j -y i , z j -z i ); Calculate the joint vectors of every two joints of the subject respectively to form a joint vector set U={u ij |i,j=1,2,3∧i <j}; The extraction process of the joint angle feature includes: Set any two joint vectors u A and u B , then u A and u B The angle feature between is expressed as: ； Calculate the angle features between the joint vectors of the subjects respectively to form the angle feature set R={r uA，uB |u A ,u B ∈U∧u A ≠u B }.

8. The gait analysis and abnormality detection system for a smart wearable device according to claim 7, characterized in that: The process of calculating the similarity between the predicted gait feature and the actual gait feature includes: Calculate the similarity between the predicted gait feature and the joint vectors in the actual gait feature using the Pearson correlation coefficient; Calculating the similarity between the predicted gait feature and the joint angle feature in the actual gait feature using a cosine similarity algorithm; The similarity between the predicted movement amount and the actual movement amount is calculated using a relative difference percentage.

9. The gait analysis and abnormality detection system for a smart wearable device according to claim 1, characterized in that: The process of determining whether a subject has gait abnormality includes: The joint vector similarity, joint angle similarity and movement amount similarity are weightedly summed to obtain the anomaly detection score; The abnormality detection score is compared with a preset abnormality detection score threshold, and if the abnormality detection score exceeds the abnormality detection score threshold range, it is determined to be an abnormal gait.

10. A gait analysis and anomaly detection method for a smart wearable device, using a gait analysis and anomaly detection system for a smart wearable device as claimed in any one of claims 1 to 9, characterized in that: Methods include: S1, collect the subjects' exercise data, physiological data and road condition information; S2. Calculate the moving speed of the subject, including: S21, establishing a spatial coordinate system based on the road surface where the subject is located as a reference plane; S22, calculating the moving speed of the subject according to the gait changes of the subject at two adjacent moments in the spatial coordinate system; S3. Predict the subject's gait at the next moment, including: S31. According to the position coordinates of the subjects, a gait space equation describing the gait change law is constructed to obtain gait variables; S32, deriving the gait variables, and then converting the gait space equation into a gait transfer equation; S33, constructing a prediction model based on a neural network, inputting gait variable road condition information, and outputting a movement function at the next moment; S34, selecting any two moving points in the moving function, and expanding the two moving points to obtain a gait prediction equation, wherein the gait prediction equation includes a predicted gait at the next moment and a predicted movement amount in the next period; S4, extracting the predicted gait features of the predicted gait at the next moment and the actual gait features of the subject's actual gait at the next moment, the gait features including joint vectors and joint angle features; S5. Obtain anomaly detection scores according to the similarity between the predicted gait features and the actual gait features, and the similarity between the predicted movement amount and the actual movement amount; S6. Compare the anomaly detection score with a preset anomaly detection score threshold. If the anomaly detection score exceeds the anomaly detection score threshold range, it is determined to be an abnormal gait.

Images