Intelligent skeleton control device

Abstract

The invention discloses an intelligent skeleton control device. Comprising a sensor module, an actuator module, a control module, a processing module and a safety braking module, the sensor module is electrically connected with the processing module, the processing module is electrically connected with the control module, the control module is electrically connected with the actuator module, and the safety braking module is electrically connected with the actuator module. The control module is further electrically connected with the safety braking module. The motion trend of the human body is decomposed in combination with the inclination detected by the gyroscope, so that the motion intention of each pressure sensor can be decomposed into motion information in the X-axis direction, the Y-axis direction and the Z-axis direction, all the pressure sensors are fused, and the fusion processing of the motion information is more accurate. The motion state can be conveniently analyzed and processed through model prediction control, and a motion instruction is generated; and under the condition of abnormity, the movement of the exoskeleton is stopped immediately through the safety braking module.

Description

Technical Field

[0001] The present invention belongs to the technical field of exoskeleton control, and more specifically, particularly relates to a skeleton intelligent control device. Background Art

[0002] Exoskeleton robot technology is a comprehensive technology that integrates sensing, control, information, fusion, and mobile computing to provide a wearable mechanical mechanism for people as operators. This article briefly introduces the development status and trends of exoskeleton world robot technology in the military field. It refers to a robot that is worn outside the human body, also known as a "wearable robot."

[0003] Generally speaking, I think that power-assisted machines can be divided into two categories: engineering power-assisted machines and biological power-assisted machines. Engineering power-assisted machines have been applied in automobiles (such as electric power steering systems), factories (such as forklifts, electric hoists) and various engineering machinery (such as excavators, cranes); while biological power-assisted machines are mainly exoskeleton power-assisted robots used by humans, which can be used in three major areas, namely military, civilian and medical. It can be said that the essence of power-assisted machines is to amplify the strength and movement speed of human beings several times or even thousands of times. Exoskeleton robot technology is a direct reflection of this essence. Combining the current research progress and human survival needs, we can look forward to the future technological development. To achieve power-assisted equipment, just like the clothes we wear, it will not only not hinder the movement of humans, but also amplify the target movement (strength and speed) of humans to the required target value according to the human brain consciousness. For example, a person cannot lift a car by himself, but after wearing an exoskeleton suit, a single individual can lift the car smoothly, and can also lift it to walk or run. In terms of future application areas, the future exoskeleton power-assist devices can be used in many fields including military, mining, industry, medical care, etc. Since the human body's functions will need to be continuously improved in the future, even far beyond its own limits, exoskeleton power-assist devices will become particularly important and eventually become indispensable products.

[0004] The existing technology has the problem that when the exoskeleton robot is in use, it is impossible to achieve precise control of the exoskeleton robot, that is, it is impossible to control and adjust the exoskeleton according to the human body's operations, that is, it is impossible to decompose the human body's operations, and it is not convenient to accurately control the analysis and processing of the exoskeleton on the X-axis, Y-axis and Z-axis. Therefore, we propose a skeleton intelligent control device. Summary of the invention

[0005] In view of the problems existing in the prior art, the purpose of the present invention is to provide a skeletal intelligent control device, which can detect the operating status of the human body through a pressure sensor, and perform three-axial analysis on the detected data information during the detection. After the analysis, the operation is controlled through model predictive control, and safe braking can be achieved to protect the safety of the wearer.

[0006] To achieve the above-mentioned object, the present invention provides the following technical solutions: a skeletal intelligent control device, comprising a sensor module, an actuator module, a control module, a processing module and a safety brake module, wherein the sensor module is electrically connected to the processing module, the processing module is electrically connected to the control module, the control module is electrically connected to the actuator module, and the control module is also electrically connected to the safety brake module;

[0007] The sensor module is used to collect data information about the user's motion state. The sensor module includes a gyroscope, a pressure sensor and a position sensor, which are used to detect the user's limb motion state. The motion state includes the gyroscope detecting the tilt angle of the user's limb, the exoskeleton detecting the user's motion intention through the pressure sensor, and the position sensor detecting the position information of the user's limb.

[0008] The processing module is used to receive the data information collected by the sensor module, and perform preprocessing after receiving the data information. The processing module includes denoising, cleaning, format conversion and normalization processing to realize denoising of the collected data information, and detect outliers and missing values ​​in the data information through cleaning, and fill the missing outliers and missing values ​​through the median value, so that the data information maintains integrity. The format conversion is used to convert the collected data information in different formats so that the data information in different formats can be in a unified format, and the normalization processing is used to map the data information between [0,1];

[0009] The control module is responsible for data processing and command output of data information, fusing and real-time processing of data information from the processing module, and generating motion instructions through algorithms. The real-time processing of data information is to decompose the collected motion intention according to the information of the gyroscope, so that the motion intention of each pressure sensor can be decomposed into motion information in the directions of X-axis, Y-axis and Z-axis, and then all pressure sensors are fused for analysis and processing of motion state through model predictive control, and motion instructions are generated;

[0010] The actuator module converts the motion instructions transmitted by the control module into physical actions to achieve control and regulation of the human exoskeleton. The actuators in the actuator module include a hydraulic drive, a pneumatic drive and a servo motor.

[0011] The safety braking module is used to immediately stop the movement of the exoskeleton in the event of an abnormality, that is, to perform calculations and determinations based on set pressure parameters and set position information, that is, after the data information collected by the pressure sensor exceeds the set pressure parameters, the position information collected by the position sensor is determined. If the exoskeleton cannot move or cannot reach the predicted position, it is determined to be an abnormal situation, and the movement of the exoskeleton is braked, and a warning is issued.

[0012] Specifically, the denoising algorithm used is Kalman filtering, which is used to remove time series signals with random noise and continuously update the estimated value of the signal according to the state model and the measured value:

[0013] The calculation formula of the Kalman filter is as follows:

[0014] State Model:

[0015]

[0016] P k|k-1 =AP k-1|k-1 A T +Q,

[0017] in, is the state prediction value, A is the state transfer matrix, the system dynamics from time k-1 to time k, is the state estimate at time k-1, B is the control input matrix, which describes how the input affects the change of the system state, and u k is the control input, P k|k-1 is the prediction covariance, Q is the process noise, P k-1|k-1 A represents the error covariance matrix at time k-1, which reflects the uncertainty of the state estimation. The smaller the error covariance matrix, the higher the confidence in the state estimation. Conversely, the larger the error covariance matrix, the higher the uncertainty in the state estimation. T represents the transpose of the state transfer matrix;

[0018] Update steps:

[0019]

[0020] Among them, K k is the Kalman gain, z k is the observed value, R is the measurement noise, H is the observation matrix, P k|k is the updated error covariance matrix.

[0021] Specifically, the cleaning performs anomaly detection on the data information through the following calculation formula:

[0022]

[0023] Among them, Y y is the data information output after calculation, x i is the input data, N is the total number of data, is the mean of the data information, is the median of the data information, σ is the standard deviation, k is the threshold, and is set to 2. If Y y >kσ, the data is judged as abnormal data and removed. y <kσ, the data information is judged to be normal data information.

[0024] Specifically, the missing values ​​and erroneous information in the cleaning are filled with the median value, and the calculation process is as follows:

[0025] When N is an odd number of data information:

[0026] in, Indicates the position of the sorted data information is Elements of

[0027] When N is an even number of data information:

[0028] in, and These are the two numbers in the middle of the sorted data.

[0029] Specifically, the calculation formula for the normalization process is as follows:

[0030] The data information is normalized and mapped between [0,1]:

[0031] Among them, X' is the mapped data information, X is the original data information, and X max is the maximum value in the original data information, X min It is the minimum value in the original data information.

[0032] Specifically, the data information collected by the gyroscope and the pressure sensor is analyzed and converted on the X-axis, Y-axis and Z-axis;

[0033] The gyroscope detects the direction and posture information of the exoskeleton, and weights are allocated to the X-axis, Y-axis and Z-axis according to the direction and posture information of the gyroscope. If the direction and posture information of the exoskeleton is large, the weight allocation is large, and if the direction and posture information of the exoskeleton is small, the weight allocation is small.

[0034] The pressure of each pressure sensor is broken down on the X-axis, Y-axis, and Z-axis as follows:

[0035] F h =w x F x +w y F y +w z F z ;

[0036] Among them, F h is the data information collected by the pressure sensor, w x 、w y 、w z are the weight values ​​in the X-axis, Y-axis and Z-axis directions respectively, x 、F y 、F z The data information is decomposed into the X-axis, Y-axis and Z-axis directions respectively.

[0037] Specifically, the data information on the X-axis, Y-axis and Z-axis is subjected to feature extraction:

[0038] The feature extraction calculation on the X-axis is as follows:

[0039]

[0040] The feature extraction calculation on the Y-axis is as follows:

[0041]

[0042] The feature extraction calculation on the Z axis is as follows:

[0043]

[0044] The features extracted by all pressure sensors are combined into a vector V, which is formed into a high-dimensional vector by simply concatenating all feature points:

[0045] V X =[F X1 ,F X2 ,F X3 ,…,F Xn ],V Y =[F Y1 ,F Y2 ,F Y3 ,…,FYn ] and V Z =[F Z1 ,F Z2 ,F Z3 ,…,F Zn ],in

[0046] f i is the extracted feature;

[0047] Normalize the feature vectors for consistency across applications:

[0048] Among them, μ is the mean of the feature and ρ is the standard deviation of the feature.

[0049] Specifically, the vector V is converted into a combined vector graph:

[0050]

[0051] Among them, m×j represents the total number of features, ij represents the jth feature point in the ith feature; the combined vector graph is integrated:

[0052]

[0053] Then, all feature points are accumulated by the accumulation method:

[0054]

[0055] After the pressure data information of all pressure sensors is analyzed and processed, and after integration processing, all the data information is accumulated and processed to facilitate the acquisition of movement trends in the X-axis, Y-axis and Z-axis directions.

[0056] Specifically, the calculation process of the model predictive control is as follows:

[0057] Assume the dynamic equation of the system is:

[0058] x(t+1)=Ax(t)+Bu(t);

[0059] Among them, x(t) is the system state, u(t) is the control input, that is, the movement trend in the X-axis, Y-axis and Z-axis directions, and A and B are the dynamic matrices of the system;

[0060] Minimize a cost function, which is the weighted sum of errors over a period of time in the future:

[0061]

[0062] Among them: Q and R are weight matrices that determine the weights of the error and control input, x ref is the reference trajectory,

[0063] The exoskeleton is controlled and regulated by predicting future outputs and determining current control inputs by solving an optimization problem.

[0064] Specifically, the safety brake module consists of pressure parameters and set position information, and the overall reliability of the system R total Calculated by the reliability of each component;

[0065] R total =1-(1-R1)(1-R2);

[0066] Among them, R1 and R2 are the reliability of two independent components;

[0067] are pressure parameters and position information respectively, and the pressure parameter F h is set to the maximum value in the historical data, and the position information is the position information calculated by the model predictive control;

[0068] By calculating the pressure parameters and position information, the data information of the other two independent components R1 and R2 is obtained, and then the overall reliability R is calculated. total , to achieve safe braking control of the exoskeleton.

[0069] Technical effects and advantages of the present invention:

[0070] The present invention detects the motion trend of a human body through a sensor module, and after detecting the motion trend of the human body, the motion trend of the human body is decomposed in combination with the inclination detected by a gyroscope, so that the motion intention of each pressure sensor can be decomposed into motion information in the directions of the X-axis, the Y-axis and the Z-axis, and then all the pressure sensors are fused and processed. For the fusion processing of the motion information, the features of the motion information on the three axes are extracted, and the feature information is vectorized, and then the vector is converted into a combined vector graph, and then an integral processing is performed, so as to facilitate the analysis and processing of the motion state through model predictive control, and generate motion instructions;

[0071] In the event of an abnormality, the movement of the exoskeleton is immediately stopped by the safety braking module, that is, the calculation and judgment are performed based on the set pressure parameters and the set position information. After the data information collected by the pressure sensor exceeds the set pressure parameters, the position information collected by the position sensor is judged. If the exoskeleton cannot move or cannot reach the predicted position, it is judged as an abnormal situation, and the movement of the exoskeleton is braked, and an alarm is issued.

[0072] Further features and advantages of the present invention will become apparent from the following detailed description of exemplary embodiments of the present invention with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] Figure 1 It is a schematic diagram of the system structure provided by the present invention. DETAILED DESCRIPTION

[0074] In order to make the purpose, technical scheme and advantages of the present invention clearer, the present invention is further described in detail below in conjunction with specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0075] like Figure 1 As shown, a skeleton intelligent control device provided by an embodiment of the present invention includes a sensor module, an actuator module, a control module, a processing module and a safety brake module, wherein the sensor module is electrically connected to the processing module, the processing module is electrically connected to the control module, the control module is electrically connected to the actuator module, and the control module is also electrically connected to the safety brake module;

[0076] The sensor module is used to collect data information about the user's motion state. The sensor module includes a gyroscope, a pressure sensor and a position sensor, which are used to detect the user's limb motion state. The motion state includes the gyroscope detecting the tilt angle of the user's limb, the exoskeleton detecting the user's motion intention through the pressure sensor, and the position sensor detecting the position information of the user's limb.

[0077] The processing module is used to receive the data information collected by the sensor module, and perform preprocessing after receiving the data information. The processing module includes denoising, cleaning, format conversion and normalization processing to realize denoising of the collected data information, and detect outliers and missing values ​​in the data information through cleaning, and fill the missing outliers and missing values ​​through the median value, so that the data information maintains integrity. The format conversion is used to convert the collected data information in different formats so that the data information in different formats can be in a unified format, and the normalization processing is used to map the data information between [0,1];

[0078] The control module is responsible for data processing and command output of data information, fusing and real-time processing of data information from the processing module, and generating motion instructions through algorithms. The real-time processing of data information is to decompose the collected motion intention according to the information of the gyroscope, so that the motion intention of each pressure sensor can be decomposed into motion information in the directions of X-axis, Y-axis and Z-axis, and then all pressure sensors are fused for analysis and processing of motion state through model predictive control, and motion instructions are generated;

[0079] The actuator module converts the motion instructions transmitted by the control module into physical actions to achieve control and regulation of the human exoskeleton. The actuators in the actuator module include a hydraulic drive, a pneumatic drive and a servo motor.

[0080] The safety braking module is used to immediately stop the movement of the exoskeleton in the event of an abnormality, that is, to perform calculations and determinations based on set pressure parameters and set position information, that is, after the data information collected by the pressure sensor exceeds the set pressure parameters, the position information collected by the position sensor is determined. If the exoskeleton cannot move or cannot reach the predicted position, it is determined to be an abnormal situation, and the movement of the exoskeleton is braked, and a warning is issued.

[0081] In this embodiment, preferably, the denoising algorithm is Kalman filtering, which is used to remove time series signals with random noise and continuously update the estimated value of the signal according to the state model and the measured value:

[0082] The calculation formula of the Kalman filter is as follows:

[0083] State Model:

[0084]

[0085] P k|k-1 =AP k-1|k-1 A T +Q,

[0086] in, is the state prediction value, A is the state transfer matrix, the system dynamics from time k-1 to time k, is the state estimate at time k-1, B is the control input matrix, which describes how the input affects the change of the system state, and u k is the control input, P k|k-1 is the prediction covariance, Q is the process noise, P k-1|k-1A represents the error covariance matrix at time k-1, which reflects the uncertainty of the state estimation. The smaller the error covariance matrix, the higher the confidence in the state estimation. Conversely, the larger the error covariance matrix, the higher the uncertainty in the state estimation. T represents the transpose of the state transfer matrix;

[0087] Update steps:

[0088]

[0089] Among them, K k is the Kalman gain, z k is the observed value, R is the measurement noise, H is the observation matrix, P k|k is the updated error covariance matrix;

[0090] It should be noted that the collected data information is filtered through Kalman filtering to improve the accuracy of the data information, facilitate subsequent analysis and processing of the data information, and use the state model to predict the current state and error covariance; the state estimate is corrected through observation data, and the Kalman gain is used to determine the weight between prediction and observation.

[0091] In this embodiment, preferably, the cleaning performs abnormality detection on the data information by the following calculation formula:

[0092]

[0093] Among them, Y y is the data information output after calculation, x i is the input data, N is the total number of data, is the mean of the data information, is the median of the data information, σ is the standard deviation, k is the threshold, and is set to 2. If Y y >kσ, the data is judged as abnormal data and removed. y <kσ, the data information is judged to be normal data information;

[0094] It should be noted that by cleaning the data information, detecting abnormal values ​​in the data information, and removing the abnormal values, the accuracy of the data information is improved, which facilitates subsequent analysis and processing of the data information.

[0095] In this embodiment, preferably, the missing values ​​and the error information in the cleaning are filled with the median value, and the calculation process is as follows:

[0096] When N is an odd number of data information:

[0097] in, Indicates the position of the sorted data information is Elements of

[0098] When N is an even number of data information:

[0099] in, and These are the two numbers in the middle of the sorted data information;

[0100] It should be noted that the missing values ​​and outliers in the data information are filled by the median value, so that the data information can maintain its integrity and facilitate subsequent analysis and processing of the data information.

[0101] In this embodiment, preferably, the calculation formula for the normalization process is as follows:

[0102] The data information is normalized and mapped between [0,1]:

[0103] Among them, X' is the mapped data information, X is the original data information, and X max is the maximum value in the original data information, X min is the minimum value in the original data information;

[0104] It should be noted that the data information is processed through normalization so that the data information can be mapped between [0, 1], reducing the dimension of the data information, making the data information convenient for subsequent calculation and processing, and reducing the pressure and complexity of the calculation.

[0105] In this embodiment, preferably, the data information collected by the gyroscope and the pressure sensor is analyzed and converted on the X-axis, Y-axis and Z-axis;

[0106] The gyroscope detects the direction and posture information of the exoskeleton, and weights are allocated to the X-axis, Y-axis and Z-axis according to the direction and posture information of the gyroscope. If the direction and posture information of the exoskeleton is large, the weight allocation is large, and if the direction and posture information of the exoskeleton is small, the weight allocation is small.

[0107] The pressure of each pressure sensor is broken down on the X-axis, Y-axis, and Z-axis as follows:

[0108] F h =w x F x +w y F y +w z F z ;

[0109] Among them, F h is the data information collected by the pressure sensor, w x 、w y 、w z are the weight values ​​in the X-axis, Y-axis and Z-axis directions respectively, x 、F y 、F z Decompose data information in the X-axis, Y-axis and Z-axis directions respectively;

[0110] It should be noted that the collected data information is decomposed through the gyroscope, and the pressure information collected by the pressure sensor is decomposed in the X-axis, Y-axis and Z-axis directions to obtain three-dimensional information on the movement trend of the human body, and according to the offset strength of the gyroscope, different proportions of weight distribution are achieved for the X-axis, Y-axis and Z-axis to realize the decomposed data information in the X-axis, Y-axis and Z-axis directions.

[0111] In this embodiment, preferably, the data information on the X-axis, Y-axis and Z-axis is subjected to feature extraction:

[0112] The feature extraction calculation on the X-axis is as follows:

[0113]

[0114] The feature extraction calculation on the Y-axis is as follows:

[0115]

[0116] The feature extraction calculation on the Z axis is as follows:

[0117]

[0118] The features extracted by all pressure sensors are combined into a vector V, which is formed into a high-dimensional vector by simply concatenating all feature points:

[0119] V X =[F X1 ,F X2 ,F X3 ,…,F Xn ],V Y =[F Y1 ,F Y2 ,F Y3 ,…,F Yn ] and V Z =[F Z1 ,F Z2 ,F Z3 ,…,F Zn ],in

[0120] fi is the extracted feature;

[0121] Normalize the feature vectors for consistency across applications:

[0122] Among them, μ is the mean of the feature, ρ is the standard deviation of the feature;

[0123] It should be noted that feature extraction is performed on the information in the X-axis, Y-axis and Z-axis directions to facilitate combining the feature points on the X-axis, Y-axis and Z-axis into vectors, and the vectors are standardized to improve the uniformity of the vectors.

[0124] In this embodiment, preferably, the vector V is converted into a combined vector graph:

[0125]

[0126] Among them, m×j represents the total number of features, and ij represents the jth feature point in the i-th feature;

[0127] Combine vector graphs for integration:

[0128]

[0129] Then, all feature points are accumulated by the accumulation method:

[0130]

[0131] After the pressure data information of all pressure sensors is analyzed and processed, and after integration processing, all the data information is accumulated to facilitate the acquisition of movement trends in the X-axis, Y-axis and Z-axis directions;

[0132] It should be noted that the combined vector graph is integrated, and all the integrated vectors are accumulated to obtain the movement trends in the X-axis, Y-axis and Z-axis directions, so as to facilitate the control and regulation of the movement of the exoskeleton.

[0133] In this embodiment, preferably, the calculation process of the model predictive control is as follows:

[0134] Assume the dynamic equation of the system is:

[0135] x(t+1)=Ax(t)+Bu(t);

[0136] Among them, x(t) is the system state, u(t) is the control input, that is, the movement trend in the X-axis, Y-axis and Z-axis directions, and A and B are the dynamic matrices of the system;

[0137] Minimize a cost function, which is the weighted sum of errors over a period of time in the future:

[0138]

[0139] Among them: Q and R are weight matrices that determine the weights of the error and control input, x ref is the reference trajectory,

[0140] The exoskeleton is controlled by predicting future outputs and determining current control inputs by solving an optimization problem.

[0141] It should be noted that the movement trend of the human body is predicted through model predictive control, and the exoskeleton is controlled and adjusted according to the prediction results, so that the exoskeleton can fit the surface of the human body, so that the exoskeleton can assist the human body in movement and enhance the human body.

[0142] In this embodiment, preferably, the safety brake module is composed of pressure parameters and set position information, and the overall reliability of the system R total Calculated by the reliability of each component;

[0143] R total =1-(1-R1)(1-R2);

[0144] Among them, R1 and R2 are the reliability of two independent components;

[0145] are pressure parameters and position information respectively, and the pressure parameter F h is set to the maximum value in the historical data, and the position information is the position information calculated by the model predictive control;

[0146] By calculating the pressure parameters and position information, the data information of the other two independent components R1 and R2 is obtained, and then the overall reliability R is calculated. total , to achieve safe braking control of the exoskeleton;

[0147] It should be noted that the safety braking module is used to control the braking of the exoskeleton, so that when the exoskeleton is prevented from being obstructed, safe braking can be achieved, thereby improving the safety of the wearer.

[0148] The specific operation process of this application:

[0149] When in use, the wearer wears the exoskeleton. When the wearer's limbs move, the sensor module on the exoskeleton detects the movement trend of the limbs. The gyroscope detects the tilt angle of the user's limbs. The exoskeleton detects the user's movement intention through the pressure sensor. The position sensor obtains the position information of the user's limbs. The collected data information is analyzed and processed by the processing module to reduce the noise of the collected data information. The outliers and missing values ​​in the data information are detected through cleaning, and the missing outliers and missing values ​​are filled by the median value to keep the data information complete. The format conversion is used to convert the collected data information in different formats so that the data information in different formats can be unified in format. The normalization processing is used to map the data information between [0,1];

[0150] After receiving the pre-processed data information, the control module processes the data information in real time by decomposing the collected movement intention according to the information of the gyroscope, so that the movement intention of each pressure sensor can be decomposed into movement information in the X-axis, Y-axis and Z-axis directions, and then all the pressure sensors are fused for processing, so as to analyze and process the movement state through model predictive control, and generate movement instructions, and then control the exoskeleton to move through the actuator module, so that the exoskeleton can move with the wearer's limbs, realize auxiliary movement and enhanced movement for the wearer, and in the event of an abnormality, the safety braking module can immediately stop the movement of the exoskeleton, that is, calculate and determine through the set pressure parameters and the set position information, that is, after the data information collected by the pressure sensor exceeds the set pressure parameters, the position information collected by the position sensor is determined. If the exoskeleton cannot move or cannot reach the predicted position, it is determined as an abnormal situation, and the movement of the exoskeleton is braked and an alarm is issued.

[0151] Finally, it should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A skeleton intelligent control device, characterized in that: The invention comprises a sensor module, an actuator module, a control module, a processing module and a safety brake module, wherein the sensor module is electrically connected to the processing module, the processing module is electrically connected to the control module, the control module is electrically connected to the actuator module, and the control module is also electrically connected to the safety brake module; The sensor module is used to collect data information about the user's motion state. The sensor module includes a gyroscope, a pressure sensor and a position sensor, which are used to detect the user's limb motion state. The motion state includes the gyroscope detecting the tilt angle of the user's limb, the exoskeleton detecting the user's motion intention through the pressure sensor, and the position sensor detecting the position information of the user's limb. The processing module is used to receive the data information collected by the sensor module, and perform preprocessing after receiving the data information. The processing module includes denoising, cleaning, format conversion and normalization processing to realize denoising of the collected data information, and detect outliers and missing values ​​in the data information through cleaning, and fill the missing outliers and missing values ​​through the median value, so that the data information maintains integrity. The format conversion is used to convert the collected data information in different formats so that the data information in different formats can be in a unified format, and the normalization processing is used to map the data information between [0,1]; The control module is responsible for data processing and command output of data information, fusing and real-time processing of data information from the processing module, and generating motion instructions through algorithms. The real-time processing of data information is to decompose the collected motion intention according to the information of the gyroscope, so that the motion intention of each pressure sensor can be decomposed into motion information in the directions of X-axis, Y-axis and Z-axis, and then all pressure sensors are fused for analysis and processing of motion state through model predictive control, and motion instructions are generated; The actuator module converts the motion instructions transmitted by the control module into physical actions to achieve control and regulation of the human exoskeleton. The actuators in the actuator module include a hydraulic drive, a pneumatic drive and a servo motor. The safety braking module is used to immediately stop the movement of the exoskeleton in the event of an abnormality, that is, to perform calculations and determinations based on set pressure parameters and set position information, that is, after the data information collected by the pressure sensor exceeds the set pressure parameters, the position information collected by the position sensor is determined. If the exoskeleton cannot move or cannot reach the predicted position, it is determined to be an abnormal situation, and the movement of the exoskeleton is braked, and a warning is issued.

2. A skeleton intelligent control device according to claim 1, characterized in that: The algorithm used in the denoising is Kalman filtering, which is used to remove time series signals with random noise and continuously update the estimated value of the signal according to the state model and the measured value: The calculation formula of the Kalman filter is as follows: State Model: P k|k-1 =AP k-1|k-1 From T +Q, in, is the state prediction value, A is the state transfer matrix, the system dynamics from time k-1 to time k, is the state estimate at time k-1, B is the control input matrix, which describes how the input affects the change of the system state, and u k is the control input, P k|k-1 is the prediction covariance, Q is the process noise, P k-1|k-1 A represents the error covariance matrix at time k-1, which reflects the uncertainty of the state estimation. The smaller the error covariance matrix, the higher the confidence in the state estimation. Conversely, the larger the error covariance matrix, the higher the uncertainty in the state estimation. T represents the transpose of the state transfer matrix; Update steps: K k =P k|k-1 H T (HP k|k-1 H T +R) -1 P k|k =(I-K k H)P k|k-1 Among them, K k is the Kalman gain, z k is the observed value, R is the measurement noise, H is the observation matrix, P k|k is the updated error covariance matrix.

3. A skeleton intelligent control device according to claim 1, characterized in that: The cleaning performs anomaly detection on the data information through the following calculation formula: Among them, Y y is the data information output after calculation, x i is the input data, N is the total number of data, is the mean of the data information, is the median of the data information, σ is the standard deviation, k is the threshold, and is set to 2. If Y y >kσ, the data is judged as abnormal data and removed. y <kσ, the data information is judged to be normal data information.

4. A skeleton intelligent control device according to claim 3, characterized in that: The missing values ​​and erroneous information in the cleaning are filled with the median value, and the calculation process is as follows: When N is an odd number of data information: in, Indicates the position of the sorted data information is Elements of When N is an even number of data information: in, and These are the two numbers in the middle of the sorted data.

5. The skeleton intelligent control device according to claim 1, characterized in that: The calculation formula for the normalization process is as follows: The data information is normalized and mapped between [0,1]: Among them, X' is the mapped data information, X is the original data information, and X max is the maximum value in the original data information, X min It is the minimum value in the original data information.

6. A skeleton intelligent control device according to claim 1, characterized in that: The data information collected by the gyroscope and the pressure sensor is analyzed and converted in the X-axis, Y-axis and Z-axis; The gyroscope detects the direction and posture information of the exoskeleton, and weights are allocated to the X-axis, Y-axis and Z-axis according to the direction and posture information of the gyroscope. If the direction and posture information of the exoskeleton is large, the weight allocation is large, and if the direction and posture information of the exoskeleton is small, the weight allocation is small. The pressure of each pressure sensor is broken down on the X-axis, Y-axis, and Z-axis as follows: F h =w x F x +w y F y +w z F z ; Among them, F h is the data information collected by the pressure sensor, w x 、w y 、w z are the weight values ​​in the X-axis, Y-axis and Z-axis directions respectively, x 、F y 、F z The data information is decomposed into the X-axis, Y-axis and Z-axis directions respectively.

7. A skeleton intelligent control device according to claim 6, characterized in that: The data information on the X-axis, Y-axis and Z-axis is subjected to feature extraction: The feature extraction calculation on the X-axis is as follows: The feature extraction calculation on the Y-axis is as follows: The feature extraction calculation on the Z axis is as follows: The features extracted by all pressure sensors are combined into a vector V, which is formed into a high-dimensional vector by simply concatenating all feature points: V X = [F X1 , F X2 , F X3 , …, F Xn , V Y = [F Y1 , F Y2 , F Y3 , …, F Yn and V Z = [F Z1 , F Z2 , F Z3 , …, F Zn , where f i is the extracted feature; Normalize the feature vectors for consistency across applications: Among them, μ is the mean of the feature and ρ is the standard deviation of the feature.

8. A skeleton intelligent control device according to claim 7, characterized in that: The vector V is converted into a combined vector graph: Among them, m×j represents the total number of features, and ij represents the jth feature point in the i-th feature; Combine vector graphs for integration: Then, all feature points are accumulated by the accumulation method: After the pressure data information of all pressure sensors is analyzed and processed, and after integration processing, all the data information is accumulated and processed to facilitate the acquisition of movement trends in the X-axis, Y-axis and Z-axis directions.

9. The skeleton intelligent control device according to claim 1, characterized in that: The calculation process of the model predictive control is as follows: Assume the dynamic equation of the system is: x(t+1)=Ax(t)+Bu(t); Among them, x(t) is the system state, u(t) is the control input, that is, the movement trend in the X-axis, Y-axis and Z-axis directions, and A and B are the dynamic matrices of the system; Minimize a cost function, which is the weighted sum of errors over a period of time in the future: Among them: Q and R are weight matrices that determine the weights of the error and control input, x ref is the reference trajectory, The exoskeleton is controlled and regulated by predicting future outputs and determining current control inputs by solving an optimization problem.

10. The intelligent skeleton control device according to claim 1, characterized in that: The safety brake module consists of pressure parameters and set position information, and the overall reliability of the system is R total Calculated by the reliability of each component; R total =1-(1-R1)(1-R2); Among them, R1 and R2 are the reliability of two independent components; are pressure parameters and position information respectively, and the pressure parameter F h is set to the maximum value in the historical data, and the position information is the position information calculated by the model predictive control; By calculating the pressure parameters and position information, the data information of the other two independent components R1 and R2 is obtained, and then the overall reliability R is calculated. total , to achieve safe braking control of the exoskeleton.

Images