CN113288084B

CN113288084B - Flexible exoskeleton system and method capable of monitoring multivariate physiological energy consumption of wearer

Info

Publication number: CN113288084B
Application number: CN202110625411.XA
Authority: CN
Inventors: 王宏; 金博丕; 李坦; 吴志伟; 张亮; 郭翔宇; 王震; 余亚辉
Original assignee: Northeastern University China
Current assignee: Northeastern University China
Priority date: 2021-06-04
Filing date: 2021-06-04
Publication date: 2022-04-15
Anticipated expiration: 2041-06-04
Also published as: CN113288084A

Abstract

The invention provides a flexible exoskeleton system and a method capable of monitoring multivariate physiological energy consumption of a wearer, wherein the whole structure adopts a soft shell to be fixed with the wearer, a wire spool is driven to rotate through a high-power steering engine, a Bowden cable wound on the wire spool is driven to be wound and unwound, and assistance and unloading to hip joints and ankle joints are realized; the flexible exoskeleton control module acquires motion information and identifies motion intention through the IMU, generates PWM signals according to identification results, drives the stay wire steering engine to rotate, acquires the tension at the tail end of the Bowden wire through the tension and compression sensor, and feeds back tension results to generate PWM signals to drive the stay wire steering engine to rotate; the multi-element physiological energy consumption analysis module collects physiological signals, carries out filtering, feature extraction and weighted fusion on the signals, carries out energy consumption calculation in an upper computer, transmits energy consumption calculation results to the microprocessor through a TCP (transmission control protocol), and feeds back the energy consumption results to generate PWM (pulse width modulation) signals to drive the stay wire steering engine to rotate; and simultaneously, real-time display is carried out on the basis of a labview energy consumption monitoring display interface.

Description

Flexible exoskeleton system and method capable of monitoring multivariate physiological energy consumption of wearer

Technical Field

The invention relates to the crossing technical field of biomedical engineering and mechanical electronic engineering, in particular to a flexible exoskeleton system and a method for monitoring multivariate physiological energy consumption of a wearer.

Background

The human-computer cooperation is a development trend of modern industry, and with the development of robot technology and biomedical engineering technology, the interactive cooperation performance of the exoskeleton robot and the human body is more consistent, the biggest characteristic is that the development of the exoskeleton focuses on the human-computer interaction of a biological layer more and more, and the motion intention recognition result is utilized to control the exoskeleton motion, so that the comfort and the safety of human body wearing are ensured while high-efficiency assistance is realized.

At present, some lower limb exoskeleton equipment exists at home and abroad, most of the equipment adopts hard shells, and certain assistance is provided for wearers through a mechanized structure. However, the rigid exoskeleton can increase the motion inertia of the extremity of the limb, thereby causing larger energy consumption of a wearer, and in addition, the rigidly connected joints have larger friction and unnatural joint motion, so that the energy in the gait process cannot be utilized to the maximum extent, and the gait energy utilization efficiency is lower.

The exoskeleton system is usually used for assisting rehabilitation training, cost can be greatly saved, the exoskeleton system can be competent for long-time repetitive assistance work, a wearer is responsible for environment perception and behavior decision making, motion instructions are provided for the exoskeleton system, the exoskeleton system needs to deduce the motion intention of a person through a sensor so as to follow the motion of a human body, the human-computer interaction system can work better in coordination, and besides, the feedback of the pulling force of a Bowden cable is needed, so that the injury of the exoskeleton system to the human body due to the fact that the pulling force of the exoskeleton system is too large is prevented. However, the existing exoskeleton system can only perform simple assistance training generally, tension feedback is not performed, and safety problems are not taken into consideration.

The exoskeleton has the effect of assisting a wearer in loading, the assisting effect can be called as the assistance performance, the evaluation of the assistance performance mainly measures the change of energy consumption before and after the wearer wears the exoskeleton, an important index for measuring the walking ability of a human body is the energy consumption, and the energy consumption is also an important index for monitoring the assistance efficiency of the exoskeleton. Reasonable evaluation of energy consumption of a wearer can be used as an important feedback basis of the exoskeleton system, the system is facilitated to realize more optimal control on bones, and most of the existing exoskeleton systems lack a reasonable and effective assistance performance evaluation module. The physiological signals are important measurement indexes of human body states, particularly for the condition of human body energy consumption, wherein the amplitude of the electromyographic signals can be increased and the frequency can be reduced along with the deepening of the human body energy consumption, and meanwhile, the average pulse, the maximum expiration inspiratory pressure and the respiratory frequency all show the increasing trend.

In conclusion, the existing exoskeleton system enables the lower limb joints to move unnaturally, cannot effectively solve the problems of overlarge human body metabolism consumption and the like caused in the movement process of specific people, does not perform tension feedback, cannot ensure safety, cannot perform energy consumption assessment, and is poor in real-time human-computer interaction.

Disclosure of Invention

Aiming at the defects of the prior art, the invention provides a flexible exoskeleton system capable of monitoring the multi-element physiological energy consumption of a wearer, which comprises a flexible exoskeleton, a motion signal acquisition unit and a physiological signal acquisition unit, wherein the flexible exoskeleton comprises a wire pulling mechanism, braces, a Bowden cable mechanism, a wearing soft shell and a working boot, the wire pulling mechanism comprises a wire pulling steering engine, a frame, a wire spool and an outer carbon plate, the outer carbon plate is fixed on the opposite side surface of the frame, two wire pulling steering engines are fixed on the frame, the output shaft of each wire pulling steering engine is connected with the central shaft of the wire spool, the wire spool is fixed on the outer carbon plate through a horizontal bearing seat, one end of the Bowden cable mechanism is wound on the wire spool, the other end of the Bowden cable mechanism is connected with the corresponding part of the wearing soft shell, the motion signal acquisition unit is used for acquiring acceleration signals of different parts of a human body and identifying the motion intention of the human body according to the acceleration signals, then controlling the movement of a stay wire steering engine in the flexible exoskeleton according to the movement intention recognition result; the physiological signal acquisition unit is used for acquiring myoelectric signals, pulse signals and respiratory signals of a human body, and performing multi-mode information fusion on the acquired signals to output an energy consumption value within a period of time for controlling the movement of a stay wire steering engine in the flexible exoskeleton; the bowden cable mechanism is driven by the stay wire steering engine to respectively drive the wearable soft shell and the working boot to move.

The wearable soft shell comprises a waist soft shell, a thigh soft shell and a shank soft shell, the Bowden cable mechanism comprises a Bowden cable mechanism A and a Bowden cable mechanism B, and the motion signal acquisition unit comprises an inertial attitude sensor-I, an inertial attitude sensor-II, an inertial attitude sensor-III, a tension and compression sensor, an amplifier, an AD conversion module and a microprocessor; the inertial attitude sensor-I is arranged in front of the chest of a wearer, the inertial attitude sensor-II is arranged on the outer side of the thigh on the right side, the inertial attitude sensor-III is arranged on the outer side of the shank on the right side, the tension and compression sensor is arranged on the working boot, the inner wire at one end of the Bowden cable mechanism A is wound on the wire spool, the inner wire wrapped by the outer wire sequentially passes through the waist soft shell, the thigh soft shell and the shank soft shell and is arranged at one end of the tension and compression sensor, and the other end of the tension and compression sensor is fixed on the working boot; an inner wire at one end of the Bowden wire mechanism B is wound on the wire spool, the inner wire wrapped by an outer wire penetrates through the waist soft shell and is fixed on the thigh soft shell, the inertial attitude sensor-I, the inertial attitude sensor-II and the inertial attitude sensor-III are respectively connected with the microprocessor in an RS-485 serial port communication mode, the tension and compression sensor is electrically connected with the amplifier, the amplifier is electrically connected with the AD conversion module, and the AD conversion module is connected with the microprocessor through a USB serial port;

the inertial attitude sensor-I is used for collecting acceleration signals at the chest of a wearer and transmitting the acceleration signals to the microprocessor;

the inertial attitude sensor-II is used for acquiring an acceleration signal of the outer side of the thigh on the right side of the wearer and transmitting the acceleration signal to the microprocessor;

the inertial attitude sensor-III is used for collecting an acceleration signal of the outer side of the right shank of the wearer and transmitting the acceleration signal to the microprocessor;

the tension and compression sensor is used for acquiring tension signals at the tail end of a wire in the Bowden cable mechanism A, and the tension signals are amplified by the amplifier and then transmitted to the AD conversion module;

the AD conversion module is used for converting the acquired amplified tension signal into a digital signal from an analog quantity and transmitting the digital signal to the microprocessor.

The physiological signal acquisition unit comprises a myoelectric sensor, an electric patch, a respiration sensor, a plastic hose, a pulse sensor and a wireless signal transmission module; the myoelectric sensor and the respiration sensor are worn on a wearer, the electric patch is attached to calf muscles and is electrically connected with the myoelectric sensor through a lead wire, the acquisition end of the respiration sensor is inserted into one end of a plastic hose, the plastic hose surrounds the chest of the wearer for a circle, the other end of the plastic hose is fixed on the respiration sensor, the pulse sensor is worn on the wrist of the wearer, and the myoelectric sensor, the respiration sensor and the pulse sensor are all transmitted to an upper computer through a wireless signal transmission module;

the myoelectric sensor is used for collecting myoelectric signals at the position of the crus and transmitting the myoelectric signals to the upper computer through the wireless signal transmission module;

the breathing sensor is used for collecting breathing signals of a human body and transmitting the breathing signals to the upper computer through the wireless signal transmission module;

the pulse sensor is used for collecting pulse signals of a human body and transmitting the pulse signals to the upper computer through the wireless signal transmission module;

the wireless signal transmission module is used for transmitting wireless signals;

the upper computer is used for filtering, characteristic extraction and multi-mode information fusion of the received electromyographic signals, respiratory signals and pulse signals, calculating the energy consumption value of a wearer within a period of time, transmitting the calculated energy consumption value to the microprocessor through a TCP/IP network communication protocol, and displaying the energy consumption value through an energy consumption monitoring display interface;

the microprocessor is used for carrying out median filtering processing on the received acceleration signals, then building a classifier based on an artificial neural network model, and carrying out movement intention identification by using the classifier according to the collected acceleration signals of different parts to obtain a movement intention identification result; the digital quantity signal of the received tension signal is subjected to median filtering processing firstly, then an actually measured tension value is extracted, and a tension feedback result is obtained by comparing the actually measured tension value with a preset tension value; the energy consumption feedback device is also used for comparing the energy consumption value within a period of time when the flexible exoskeleton is worn with the energy consumption value within the same period of time when the flexible exoskeleton is not worn to obtain an energy consumption feedback result; and generating a PWM signal for controlling the rotation state of the output shaft of the stay wire steering engine according to the movement intention recognition result, the tension feedback result and the energy consumption feedback result.

The microprocessor adopts RaspberryPi; the upper computer is a computer; the energy consumption monitoring display interface is realized by programming on a computer by labview; the IMU-I, IMU-II and the IMU-III are connected in a cascade mode, and a Modbus protocol is matched with an RS-485 physical serial port to carry out data transmission.

A method of flexible exoskeleton control using a flexible exoskeleton system having a plurality of physiological energy consumptions monitoring a wearer, comprising:

step 1: collecting respiratory signals, pulse signals and myoelectric signals at crus of a wearer within a period of time T when the wearer wears the flexible exoskeleton, simultaneously collecting acceleration signals of the front of the chest, the outer side of the thigh on the right side and the outer side of the crus on the right side of the wearer, and collecting tension signals at the tail ends of wires in the Bowden wire mechanism A;

step 2: respectively carrying out noise reduction processing on the acquired signals, wherein the acceleration signals are subjected to noise reduction processing through a median filter; the electromyographic signals are subjected to noise reduction treatment through a four-order Butterworth band-pass filter; the respiratory signal is firstly subjected to a forty-order FIR low-pass filter and then subjected to smooth filtering processing to reduce noise; selecting Sym5 wavelet as a basis function for the pulse signal, and denoising by using a wavelet decomposition and reconstruction-based method; the tension signal is subjected to noise reduction treatment through a median filter;

and step 3: performing feature extraction and multi-mode information fusion on the electromyographic signals, the respiratory signals and the pulse signals after noise reduction processing to obtain an energy consumption value E of a wearer in a period of time;

and 4, step 4: collecting a respiration signal, a pulse signal and a myoelectric signal at a shank in advance when a wearer does not wear the flexible exoskeleton for a period of time, and carrying out multi-mode information fusion to obtain an energy consumption value E' in the same time when the wearer does not wear the flexible exoskeleton;

and 5: comparing the energy consumption value E within a period of time when the flexible exoskeleton is worn with the energy consumption value E' within the same period of time when the flexible exoskeleton is not worn, and outputting an energy consumption feedback result;

step 6: extracting an actually measured tension value of the noise-reduced tension signal, comparing the actually measured tension value with a preset tension value, and outputting a tension signal feedback result;

and 7: and constructing a classifier based on the artificial neural network, identifying the movement intention by using the classifier according to the noise-reduced acceleration signal, and outputting a movement intention identification result.

And 8: and outputting PWM signals for controlling the rotation angle value of the output shaft of the stay wire steering engine according to the zone bits of the output state of the stay wire steering engine corresponding to the motion intention recognition result, the tension feedback result and the energy consumption feedback result and the preset zone bit priority level.

The step 3 comprises the following steps:

step 3.1: signal feature extraction, namely, performing time domain analysis on the electromyographic signals to extract feature parameters of the electromyographic signals: root mean square value RMS and electromyographic integral value IEMG; time domain analysis is carried out on the respiratory signal to extract characteristic parameters of the respiratory signal: period T1 and amplitude A; extracting time domain characteristic parameters of the pulse signals by using a differential threshold method: branch time of rise t₁Descending branch time t₂Height of the rising leg H₁Descending branch height H₂And a period T2;

step 3.2: the characteristic parameters of the three signals are standardized by a min-max method, so that the result value is mapped between 0 and 1,

in the formula, E_tCharacteristic parameter values representing the electromyographic signals acquired at time t, E_min、E_maxRepresenting the minimum and maximum values of a characteristic parameter, R, of the myoelectric signal over a period of sampling time_tCharacteristic parameter value, R, representing the breathing signal acquired at time t_min、R_maxRepresenting the minimum, maximum, P, of a value of a characteristic parameter of the respiratory signal over a period of sampling time_tA characteristic parameter value representing the pulse signal acquired at time t,P_min、P_maxthe minimum value and the maximum value of the characteristic parameter values of the pulse signals in a period of sampling time are represented;

wherein E is_t＝a₁×RMS+a₂X IEMG, wherein a₁Is a weighting factor, a, on the influence of the root mean square value, RMS, term₂Is a weight coefficient of influence on the myoelectric integral value IEMG term, and a₁+a₂＝1；

Rt＝b₁×T1+b₂X A, in the formula b₁Is a weight coefficient of influence on the term of the period T1, b₂Is a weight coefficient of the effect on the amplitude A term, and b₁+b₂＝1；

Pt＝c₁×t₁+c₂×t₂+c₃×H₁+c₄×H₂+c₅X T2, wherein c₁Is for the rising branch time t₁Weight coefficient of influence of term, c₂Is for the descending branch time t₂Weight coefficient of influence of term, c₃For the rising branch height H₁Weight coefficient of influence of term, c₄For descending branch height H₂Weight coefficient of influence of term, c₅Is a weight coefficient of influence on the term of the period T2, and c₁+c₂+c₃+c₄+c₅＝1；

Step 3.3: carrying out weighted fusion on the characteristic parameters of the three normalized signals by using a characteristic fusion strategy to obtain a fusion value f (E) of the three signals_tn,R_tn,P_tn)，

f(E_tn,R_tn,P_tn)＝a×E_tn+b×R_tn+c×P_tn

Wherein a, b, and c represent weighted values of the myoelectric signal, the respiratory signal, and the pulse signal, respectively, and satisfy a + b + c being 1;

step 3.4: the resulting fusion value f (E)_tn,R_tn,P_tn) The integral is multiplied by the weight W to obtain the energy consumption value E in a period of time,

where n' represents the total divided division phase within the wearer movement time period T, T_iIndicating the duration of the motion in the ith division stage.

The step 5 is specifically expressed as follows:

step 5.1: if E < (1-gamma) E', the flexible exoskeleton plays a power assisting effect on a wearer without changing the output state of the stay wire steering engine, wherein gamma represents a primary adjusting parameter;

step 5.2: if (1-gamma) E '< E < (1+ gamma) E', indicating that the flexible exoskeleton does not have the assistance effect on the wearer, the flag bit of the output state of the pull wire steering engine needs to be set to l0, and the PWM signal for changing the rotation of the output shaft of the pull wire steering engine comprises the following steps:

if (1+ alpha) E '< E < (1+ gamma) E', increasing the PWM value for controlling the stay wire steering engine to be (1+3 epsilon) times of the original value, wherein alpha represents a secondary regulation parameter, and epsilon represents an energy consumption feedback threshold value;

if (1-alpha) E '< E < (1+ alpha) E', increasing the PWM value for controlling the stay wire steering engine to be (1+2 epsilon) times of the original value;

if (1-gamma) E '< E < (1-alpha) E', increasing the PWM value for controlling the stay wire steering engine to be (1+ epsilon) times;

step 5.3: if E > (1+ gamma) E', the flexible exoskeleton plays a resistance effect on a wearer, the stay wire steering engine is controlled to enter a power-off protection state, and the flag bit of the output state of the stay wire steering engine needs to be set to l 1.

The step 6 is specifically expressed as: extracting an actually measured tension value F according to the tension signal acquired in real time_lMaximum value F of preset tension value_l' max or minimum value F_l' min is compared if F_l<F_l'min', increasing the PWM value for controlling the stay wire steering engine to be (1+ phi) times of the original value, and setting the flag bit of the output state of the stay wire steering engine to be l2, wherein phi represents a tension feedback threshold value; if F_l≥F_lAnd max controls the stay wire steering engine to enter a power-off protection state, and the flag bit of the output state of the stay wire steering engine needs to be set to l 3.

The step 7 comprises the following steps:

step 7.1: respectively extracting acceleration signals ACC _ X in the X-axis direction from the acquired acceleration signals;

step 7.2: adopting a sliding window algorithm to carry out gait phase division on the acceleration signal ACC _ x;

step 7.3: constructing a classifier based on an artificial neural network, taking the acceleration signal marked by the gait phase as a training sample set of the classifier to train the classifier, and stopping training when the maximum training times is reached to obtain the trained classifier;

step 7.4: the gait phase of the acquired acceleration signals is predicted by using the trained classifier, the prediction result of the gait phase is output, the flag bit of the output state of the stay wire steering engine is set to be l4, and the priority level of the flag bit is set to be l3 > l1 > l4 > l2 > l 0.

Said step 7.2 comprises:

step 7.2.1: setting the size N of a sliding window and the overlapping rate eta of the sliding window, and initializing the current state of a wearer to be a standing state;

step 7.2.2: extracting a phase division point of the acceleration signal, and marking a time point when m continuous amplitude values of a signal ACC _ x corresponding to the thigh outer side acceleration signal start to change in a sliding window as a phase division point A1; the sliding window continuously moves backwards, and the amplitude value of a signal ACC _ x corresponding to the acceleration signal of the outer side of the crus detected for the first time is omega_xThe time point of time is marked as phase division point a 2; wherein ω is_x10% g, where g represents acceleration of gravity; the sliding windows continue to move backward, and the time point when the amplitude of the thigh outer acceleration signal corresponding signal ACC _ x is detected for the second time in one sliding window is marked as the phase division point a 3; the sliding window continues to move backwards, and a time point when m continuous amplitude values of a signal ACC _ x corresponding to the chest acceleration signal are detected to be unchanged in one sliding window is marked as a phase division point A4;

step 7.2.3: dividing gait phases according to phase division points, wherein the corresponding gait phases between an initial time point 0 and a time point A1 are in a standing state, the corresponding gait phases between a time point A1 and a time point A2 are in a heel-off state, the corresponding gait phases between a time point A2 and a time point A3 are in a swinging state, and the corresponding gait phases between a time point A3 and a time point A4 are in a sole landing state; the gait phase from the time point a4 to the first time point a1 of the next cycle is the heel strike state.

The invention has the beneficial effects that:

the invention provides a flexible exoskeleton system and a method for monitoring multivariate physiological energy consumption of a wearer. The arrangement mode and the transmission mode can place the weight of the driving device on the outer side carbon plate A, so that a wearer is prevented from directly bearing the load; in addition, the material of the outer side carbon plate A is carbon fiber, so that the weight of the structure is reduced, and the back of a wearer is well protected; the control on the flexible exoskeleton system effectively combines the motion signal, the Bowden cable tension signal and the energy consumption value obtained by multi-mode information fusion calculation, has the functions of motion intention identification and feedback, detects the motion intention of a wearer in real time, provides assistance based on intention identification and feedback results, greatly improves the accuracy and the real-time of motion intention identification, and provides sufficient guarantee for the assistance efficiency of the exoskeleton robot; meanwhile, the tension and compression sensor can monitor the tension of the tail end of the wire in the flexible lower limb exoskeleton bowden cable mechanism A in real time, the tension is controlled in a proper range, the damage to a human body caused by the exceeding range is avoided, and the safety performance of an exoskeleton system is effectively improved; the control method disclosed by the invention integrates the physiological information signals of myoelectricity, respiration and pulse of the wearer to evaluate the energy consumption condition of the wearer, emphasizes the comprehensiveness of decision information, and has a better effect than that of only considering single physiological information characteristic.

Drawings

FIG. 1 is a block diagram of a flexible exoskeleton system with monitoring of multiple physiological energy consumptions of a wearer in accordance with the present invention;

fig. 2 is a general block diagram of the flexible exoskeleton capable of monitoring multiple physiological energy consumptions of a wearer according to the present invention, wherein (a) is a drawing of a cable mechanism in the flexible exoskeleton, (b) is a general outline diagram of the flexible exoskeleton, and (c) is a schematic view of the flexible exoskeleton during wearing;

fig. 3 is a schematic diagram of a control method of the present invention using a flexible exoskeleton system with monitoring of multiple physiological energy consumptions of a wearer, wherein (a) is a schematic diagram of the control method and (b) is a schematic diagram of the control method;

FIG. 4 is a flow chart of a control method of the present invention employing a flexible exoskeleton system with monitoring of multiple physiological energy consumptions of a wearer;

FIG. 5 is a schematic diagram of the wiring of the flexible exoskeleton system with monitoring of multiple physiological energy consumptions of a wearer in accordance with the present invention;

FIG. 6 is a flow chart of energy consumption feedback control in the present invention;

FIG. 7 is a flow chart of tension feedback control in the present invention;

FIG. 8 is a schematic diagram of multimodal information fusion in the present invention;

FIG. 9 is a schematic diagram of an upper computer control system interface according to the present invention;

in the figure, 1, a harness, 2, a pull wire structure, 201, an outer carbon plate B, 202, an European standard aluminum frame, 203, an outer carbon plate C, 204, a countersunk head bolt pair, 205, a horizontal bearing seat, 206, a wire spool, 207, an outer hexagonal bolt pair, 208, a flange plate, 209, an outer carbon plate A, 210, a pull wire steering engine, 211, an inner hexagonal bolt pair, 212, an inner carbon plate, 213, an outer carbon plate D, 3, a Bowden wire mechanism A, 301, an inner wire, 302, an outer wire, 4, a Bowden wire mechanism B, 5, a waist soft shell, 502, a buckle, 503, a nylon buckle belt, 6, a thigh soft shell, 604, a stainless steel single clamp head, a stainless steel lifting hole, 7, a shank soft shell, 8, a working boot, 801, a working boot with a hole, 802, a pull pressure sensor, 9, a physiological signal acquisition unit, 901, a myoelectric sensor, 902, an electric patch, 903, a breathing sensor, 904, a plastic hose, a breathing sensor, 905, a, The pulse monitoring device comprises a pulse sensor 10, a motion signal acquisition unit 1001, IMU-I, 1002, IMU-II, 1003, IMU-III, 11 and a power battery.

Detailed Description

The invention is further described with reference to the accompanying drawings 1-9 and the specific embodiments. The invention provides a flexible exoskeleton system capable of monitoring multivariate physiological energy consumption of a wearer, which comprises a flexible exoskeleton mechanism, wherein the whole structure adopts a soft shell to be fixed with the wearer, a wire reel is driven to rotate through a high-power steering engine, and a Bowden cable wound on the wire reel is driven to be wound and unwound, so that power assistance and unloading on hip joints and ankle joints are realized; the flexible exoskeleton control module acquires motion information and identifies motion intention through the IMU, generates PWM signals according to identification results, drives the stay wire steering engine to rotate, acquires the tension at the tail end of the Bowden wire through the tension and compression sensor, and feeds back tension results to generate PWM signals to drive the stay wire steering engine to rotate; the multi-element physiological energy consumption analysis module collects physiological signals, carries out filtering, feature extraction and weighted fusion on the signals, carries out energy consumption calculation in an upper computer, transmits energy consumption calculation results to the microprocessor through a TCP (transmission control protocol), and feeds back the energy consumption results to generate PWM (pulse width modulation) signals to drive the stay wire steering engine to rotate; and simultaneously, real-time display is carried out on the basis of a labview energy consumption monitoring display interface.

As shown in fig. 1-2, a flexible exoskeleton system capable of monitoring multivariate physiological energy consumption of a wearer comprises a flexible exoskeleton, a motion signal acquisition unit 10 and a physiological signal acquisition unit 9, wherein the flexible exoskeleton comprises a wire pulling mechanism 2, a back strap 1, a bowden wire mechanism a3, a bowden wire mechanism B4, a waist soft shell 5, a thigh soft shell 6, a shank soft shell 7 and a work boot 8; the wire drawing mechanism 2 comprises a wire drawing steering engine 210, an European standard aluminum frame 202, a flange plate 208, a countersunk head bolt pair 204, a wire winding disc 206, an outer carbon plate B201, an outer carbon plate C203, a horizontal bearing seat 205, an outer hexagon bolt pair 207, an outer carbon plate A209, a wire drawing steering engine 210, an inner hexagon bolt pair 211, an inner carbon plate 212, an outer carbon plate D213, an inner wire 301, an outer wire 302, a buckle 502, a nylon buckle belt 503, a stainless steel single chuck 604, a stainless steel hanging hole 605, a working boot with hole 801 and the European standard aluminum frame 202 are frames with a cubic structure built up by 12 aluminum profiles, the outer carbon plate A209 is fixed on the rear side of the European standard aluminum frame 202, the outer carbon plate B201 is respectively fixed on the lower side of the European standard aluminum frame 202, the outer carbon plate C203 and the outer carbon plate D213 are fixed on the left and right opposite side surfaces of the European standard aluminum frame 202, the two wire drawing engines 210 are fixed on the European standard aluminum frame 202 along the length direction of the aluminum profiles through the respective inner carbon plates 212, in order to ensure the stress balance and the reasonability of spatial distribution of the whole stay wire mechanism 2 during actual installation, output shafts of two stay wire steering engines with the same model are respectively close to the left lower side and the right upper side, the output shaft of each stay wire steering engine is connected with a central shaft of a wire reel 206 through a flange 208, the two wire reels 206 with the same size are respectively fixed on an outer carbon plate C203 and an outer carbon plate D213 through a horizontal bearing seat 205, the height of the outer carbon plate A209 is higher than that of an European standard aluminum frame 202, four rectangular through holes are formed in the outer carbon plate A209, a suspender 1 penetrates through the four through holes in the outer carbon plate A209, a wearer wears the stay wire mechanism 2 on the back through the suspender 1, a waist soft shell 5 is worn on the waist of the wearer, a shank soft shell 7 is worn on the position of the wearer, and a shank soft shell 6 is worn on the position of the thigh of the wearer; the power cell 11, which powers the entire flexible exoskeleton system, is mounted on the outer carbon plate a209, above the euro standard aluminum frame 202.

The motion signal acquisition unit is used for acquiring acceleration signals of different parts of a human body, identifying the motion intention of the human body according to the acceleration signals and then controlling the motion of the pull wire steering engine 210 in the flexible exoskeleton according to the motion intention identification result; the physiological signal acquisition unit is used for acquiring myoelectric signals, pulse signals and respiratory signals of a human body, and performing multi-mode information fusion on the acquired signals to output an energy consumption value within a period of time for controlling the movement of the pull-wire steering engine 210 in the flexible exoskeleton; the bowden cable mechanism is driven by the pull line steering engine 210 to respectively drive the wearable soft shell and the working boot to move.

The setting of the back of the human body on the wire drawing mechanism 2 is realized by the back belt 1 through the hole on the outer carbon plate 209, that is, 4 countersunk head bolt pairs 204 on the European standard aluminum frame 202 are fixed on the outer carbon plate B201 and the outer carbon plate A209, wherein the outer carbon plate C203 and the outer carbon plate C213 on both sides are connected and fixed on both sides of the system of the wire drawing mechanism 2 through the countersunk head bolt pairs 204, the horizontal bearing seat 205 is fixed on the outer carbon plate C203 through the 2 countersunk head bolt pairs 204, and the power supply battery 11 for supplying power to the system is fixed on the outer carbon plate A209 through the 4 countersunk head bolt pairs 204 and is positioned above the European standard aluminum frame 202.

The central shaft structure at one end of the wire spool 206 is clamped into one end of the horizontal bearing seat 205 and matched with the same, the other end of the central shaft of the wire spool 206 is connected with an output shaft of a stay wire steering engine 210 through a flange plate 208, the flange plate 208 is fixed through an inner hexagon bolt pair 211, the two stay wire steering engines 210 with the same model are respectively connected and fixed on an inner side carbon plate 212 through 4 inner hexagon bolt pairs 211, the inner side carbon plate 212 is connected and fixed on the left side and the right side of the European standard aluminum frame 202 through 4 inner hexagon bolt pairs 211, in order to keep the stress balance of the whole structure and reasonably utilize the installation space, the two stay wire steering engines are respectively installed on the left side and the right side of the European standard aluminum frame 202, and the central shafts of the stay wire steering engines are installed in a staggered manner; the inner wire 301 in the Bowden wire mechanism A3 and the inner wire 301 in the Bowden wire mechanism B4 are respectively wound on the wire reels 207 on the left side and the right side for fixation, the Bowden wire mechanism A3 and the Bowden wire mechanism B4 are divided into two parts, the inner wire 301 is positioned at the center of the Bowden wire, and the outer wire 302 wrapped on the inner wire 301 realizes the fixation and protection effects on the inner wire 301.

The waist soft shell 5 is fixed on the waist of a human body through two nylon buckle belts 503, and the Bowden cable mechanism A3 and the Bowden cable mechanism B4 are fixed through a plurality of buckles 502 arranged on the waist soft shell 5 through bolts; the thigh soft shell 6 is fixed on the thigh of a human body through two nylon buckle belts 503, and the Bowden cable mechanism A3 and the Bowden cable mechanism B4 are fixed by a buckle 502 on the thigh soft shell 6, wherein the end of the outer wire 302 in the Bowden cable mechanism B4 on the front side of the thigh is fixed by the buckle 502, and the end of the extended inner wire 301 passes through a stainless steel single clamp 604 and is bent to be fixed on the thigh soft shell 501 through a stainless steel hanging hole 505; the shank soft shell 7 is fixed on the shank of a human body through two nylon buckle belts 503, and the Bowden cable mechanism A3 is fixed by a buckle 502 on the shank soft shell 7; the waist soft shell 5, the thigh soft shell 6 and the shank soft shell 7 are all plastic soft shells, power is transmitted to the thigh soft shell 6 and the working boot 8 through the two Bowden wire mechanisms by the high-power wire-pulling steering engine 210 to drive the hip joint and the ankle joint of a wearer to move, when the wire-pulling mechanism 2 on the right side drives the wire reel 206 on the wire-pulling mechanism to rotate, the hip joint of the human body is driven to rotate by the inner wire 301 of the Bowden wire mechanism B4, so that the hip joint of the wearer is driven to stretch and bend, and when the wire-pulling mechanism on the left side drives the wire reel 207 on the wire-pulling mechanism to rotate, the ankle joint of the human body is driven to rotate by the inner wire 301 of the Bowden wire mechanism A3, so that the ankle joint of the wearer is driven to stretch and bend.

The motion signal acquisition unit 10 comprises an IMU-I1001, an IMU-II1002, an IMU-III1003, a tension and compression sensor 802, an amplifier, an AD conversion module and a microprocessor; IMU-I1001 is arranged on the chest of a wearer, IMU-II1002 is arranged on the outer side of the right thigh, IMU-III1003 is arranged on the outer side of the right shank, a tension and compression sensor 802 is arranged on a working boot 8 through a working boot strap hole 801, an inner wire 301 at one end of a Bowden wire mechanism A3 is wound on a wire spool 206, the inner wire 301 wrapped by an outer wire 302 sequentially passes through a buckle 502 on a waist soft shell 5, a thigh soft shell 6 and a shank soft shell 7 and is arranged on the tension and compression sensor 802 through a stainless steel single chuck, the inner wire 301 at one end of a Bowden wire mechanism B4 is wound on the wire spool 206, the inner wire 301 wrapped by the outer wire 302 sequentially passes through the buckle 502 on the waist soft shell 5 and the bayonet 502 on the thigh soft shell 6 and is fixed on the thigh soft shell 6 through the stainless steel single chuck, the IMU-I1001, the IMU-II1002 and the IMU-III1003 are respectively connected with microprocessing through RS-485 serial ports, the tension and compression sensor 802 is electrically connected with an amplifier, the amplifier is electrically connected with an AD conversion module, and the AD conversion module is connected with the microprocessor through a USB serial port;

the IMU-I is used for collecting acceleration signals of the chest of a wearer and transmitting the acceleration signals to the microprocessor;

the IMU-II is used for collecting acceleration signals of the outer side of the thigh on the right side of the wearer and transmitting the acceleration signals to the microprocessor;

the IMU-III is used for collecting acceleration signals of the outer side of the right shank of the wearer and transmitting the acceleration signals to the microprocessor;

The physiological signal acquisition unit comprises an electromyographic sensor 901, an electric patch 902, a respiratory sensor 903, a plastic hose 904, a pulse sensor 905 and a wireless signal transmission module; the myoelectric sensor 901 and the respiratory sensor 903 are both worn on a wearer, the electric patch 902 is attached to the muscles of the lower leg, the electric patch 902 is electrically connected with the myoelectric sensor 901 through a lead wire, the acquisition end of the respiratory sensor 903 is inserted into one end of a plastic hose 904, the plastic hose 904 surrounds the chest of the wearer for a circle, the other end of the plastic hose 904 is fixed on the respiratory sensor 903, a pulse sensor 905 is sleeved on the wrist of the wearer through a nylon thread gluing on the pulse sensor, the myoelectric sensor 901, the respiratory sensor 903 and the pulse sensor 905 are all transmitted to an upper computer through a wireless signal transmission module, the filtering, feature extraction and multi-mode information fusion of the myoelectric signal, the respiratory signal and the pulse signal are realized through the upper computer, the energy consumption value of the wearer within a period of time is output, and the upper computer transmits the calculated energy consumption value to a microprocessor through a TCP/IP network communication protocol, displaying the energy consumption value through an energy consumption monitoring display interface;

the Bowden cable mechanism A3 led out from the shank of a human body is positioned at the rear side of the shank, the tail end of the outer line 302 in the Bowden cable mechanism A3 is fixed by a buckle 502, the tail end of the inner line 302 in the Bowden cable mechanism A3 is fixed by bending a stainless steel single chuck 504 to penetrate through one end of a tension and compression sensor 802, and the other end of the tension and compression sensor 802 is fixed on a working boot strap hole 801 on a working boot 8.

The electromyographic sensor 901 is adhered to the back side of the thigh by magic tape, wherein the led-out electric patch 902 is adhered to the skin surface of the human body, and the position of the electric patch corresponds to the gastrocnemius muscle of the calf; the respiration sensor 903 is stuck to the chest by a magic tape positioned on the back, wherein a plastic hose 904 connected to the respiration sensor 903 is wound around the chest of the body for one circle; the pulse sensor 905 is sleeved on the wrist of the wearer through a nylon thread gluing on the pulse sensor; the IMU-I1001 is placed in front of the chest of a wearer through the magic tape on the IMU-I1002, the IMU-II1002 is placed on the thigh soft shell 6 through the magic tape on the IMU-II, the IMU-III1003 is placed on the shank soft shell 7 through the magic tape on the IMU-II, all IMUs are guaranteed to be mounted in the mode that the plane of a Pitch angle (namely a Pitch angle) is parallel to the sagittal plane of a human body, and the IMU-I, IMU-II and the IMU-III are three inertial posture sensors.

the wireless signal transmission module is used for transmitting wireless Bluetooth signals;

the upper computer is used for filtering, feature extraction and multimodal information fusion of the received electromyographic signals, respiratory signals and pulse signals, calculating an energy consumption value of a wearer in a period of time when the wearer wears the flexible exoskeleton (on the right side), transmitting the energy consumption result to the microprocessor through a TCP/IP network communication protocol, and displaying the energy consumption value through an energy consumption monitoring display interface, as shown in FIG. 8; the concrete expression is as follows:

in the formula, E_tCharacteristic parameter values representing the electromyographic signals acquired at time t, E_min、E_maxCharacteristic parameter representing an electrical myosignal over a period of timeMinimum and maximum values of the values, R_tCharacteristic parameter value, R, representing the breathing signal acquired at time t_min、R_maxRepresenting the minimum, maximum, P, of a value of a characteristic parameter of the breathing signal over a period of time_tCharacteristic parameter value, P, representing the pulse signal acquired at time t_min、P_maxThe minimum value and the maximum value of the characteristic parameter values of the pulse signals in a period of time are represented;

wherein E is_t＝a₁×RMS+a₂X IEMG, where a1 is the weight coefficient for the effect on the RMS term, a₂Is a weight coefficient of influence on the myoelectric integral value IEMG term, and a₁+a₂＝1；

f(E_tn,R_tn,P_tn)＝a×E_tn+b×R_tn+c×P_tn

step 3.4: the resulting fusion value f (E)_tn,R_tn,P_tn) The integral is multiplied by the weight W to obtain the energy consumption value E within a period of time T,

The microprocessor is used for carrying out median filtering processing on the received acceleration signals, then building a classifier based on an artificial neural network model, and carrying out movement intention recognition on corresponding parts by using the classifier according to the collected acceleration signals of different parts to obtain a movement intention recognition result, wherein the specific expression is as follows:

step 7.1: respectively extracting acceleration signals ACC _ X in the X-axis direction from the filtered acceleration signals; wherein the X-axis direction is parallel to the trunk of the wearer, the Y-axis direction points to the right side of the body, and the Z-axis direction points to the front side of the body;

step 7.4: predicting gait phase of the acquired acceleration signals by using the trained classifier, outputting a gait phase prediction result, and outputting a state flag position l4 by using a stay wire steering engine;

the microprocessor is also used for carrying out median filtering processing on the digital quantity signal of the received tension signal, extracting an actually measured tension value and comparing the actually measured tension value with a preset tension value to obtain a tension feedback result; the concrete expression is as follows:

extracting according to the real-time collected tension signalMeasuring tensile force value F_lMaximum value F of preset tension value_l' max or minimum value F_l' min is compared if F_l<F_lMin, increasing the PWM value for controlling the stay wire steering engine to be (1+ phi) times of the original value, and marking the output state of the stay wire steering engine at a mark position l2, wherein phi represents a tension feedback threshold value; if F_l≥F_lMax controls the stay wire steering engine to enter a power-off protection state, and the stay wire steering engine needs to output a mark position l3 of the state;

the microprocessor is also used for comparing the energy consumption value within a period of time when the flexible exoskeleton is worn with the energy consumption value within the same period of time when the flexible exoskeleton is not worn to obtain an energy consumption feedback result; the concrete expression is as follows:

and 4, step 4: collecting a respiration signal, a pulse signal and a myoelectric signal at a shank in advance when a wearer does not wear the flexible exoskeleton for a period of time, and carrying out multi-mode information fusion to obtain an energy consumption value E' in the period of time when the wearer does not wear the flexible exoskeleton;

and 5: comparing the energy consumption value E within a period of time when the flexible exoskeleton is worn with the energy consumption value E' within the same period of time when the flexible exoskeleton is not worn to obtain an energy consumption feedback result; the concrete expression is as follows:

step 5.2: if (1-gamma) E '< E < (1+ gamma) E', indicating that the flexible exoskeleton does not have the assistance effect on the wearer, the position l0 of the output state of the pull wire steering engine needs to be marked, and the PWM signal for changing the rotation of the output shaft of the pull wire steering engine comprises the following steps:

step 5.3: if E > (1+ gamma) E', the flexible exoskeleton plays a resistance effect on a wearer, the stay wire steering engine is controlled to enter a power-off protection state, and the stay wire steering engine needs to be output to a mark position l1 in the state;

and finally, generating a PWM (pulse-width modulation) signal by the microprocessor according to the movement intention recognition result, the pulling force feedback result and the energy consumption feedback result for controlling the rotation state of the output shaft of the stay wire steering engine, wherein the specific principle is that three control signals of the movement intention recognition result, the pulling force feedback result and the energy consumption feedback result are used for determining who preferentially outputs the control signals by setting the output priority level, and the control signals with high priority level are preferentially output as the output signals for controlling the stay wire steering engine.

The microprocessor is a RaspberryPi; the upper computer is a computer; the energy consumption monitoring display interface is realized by programming labview on a computer, and the acquisition value of a signal, the connection state of each sensor, the information of a wearer and the energy consumption value are displayed in real time through the energy consumption monitoring display interface; the IMU-I, IMU-II and the IMU-III are connected in a cascade mode, a Modbus protocol is adopted to match with an RS-485 physical serial port for data transmission, and interface terminals VCC are used for VCC, A is used for A, B is used for B, and GND is used for GND, wherein A, B is an IMU acquisition signal interface respectively; the default setting of the IMU inertial sensor is 0x50, and the address is increased by 1 when an IMU is added, so that the three communication addresses of the IMU inertial sensor are set as follows: 0x50, 0x51, 0x 52; meanwhile, the communication baud rate of the IMU needs to be set to be 460800 bps.

A method for controlling a flexible exoskeleton using the flexible exoskeleton system with monitoring multivariate physiological energy consumption of a wearer, as shown in figures 3-4, comprises:

step 1: collecting respiratory signals, pulse signals and myoelectric signals at crus of a wearer within a period of time T when the wearer wears the flexible exoskeleton, simultaneously collecting acceleration signals of the front of the chest, the outer side of the thigh on the right side and the outer side of the crus on the right side of the wearer, and collecting tension signals of inner wires in the Bowden wire mechanism A;

wherein the median filtering is expressed as:

Y(i)＝Med{x_i-v,…,x_i-1,x_i,x_i+1,…,x_i+v},i∈N',v＝(m-1)/2

wherein m is the number of points around the target value, and Y (i) is { { x in the sequence_i-v,…,x_i-1,x_i,x_i+1,…,x_i+vThe median value of }; m needs to be an odd number, and when m is 3, the noise reduction effect of the motion signal (the acceleration signal collected here) is most obvious, so that the signal is smoothed, and the original data of the acceleration is not changed too much.

And step 3: performing feature extraction and multi-mode information fusion on the electromyographic signals, the respiratory signals and the pulse signals after noise reduction processing to obtain an energy consumption value E of a wearer in a period of time; as shown in fig. 6, includes:

step 3.1: signal feature extraction, namely, performing time domain analysis on the electromyographic signals to extract feature parameters of the electromyographic signals: root mean square value RMS and electromyographic integral value IEMG; time domain analysis is carried out on the respiratory signal to extract characteristic parameters of the respiratory signal: period T1 and amplitude A; extracting time domain characteristic parameters of the pulse signals by using a differential threshold method: branch time of rise t₁For downturn time t₂Height of the rising leg H₁Descending branch height H₂And a period T2;

step 3.2: the multivariate physiological signals have different dimensions and units, and in order to eliminate the dimension influence among the multivariate signals, data standardization processing is needed to solve the comparability among data indexes. After data standardization processing is carried out on multivariate signal data, each index is in the same order of magnitude and is suitable for comprehensive comparison and evaluation, so that the characteristic parameters of three signals are standardized by a min-max method, the result value is mapped to 0-1,

in the formula, E_tCharacteristic parameter values representing the electromyographic signals acquired at time t, E_min、E_maxMinimum and maximum values of a parameter representing the characteristic of the myoelectric signal over a period of time, R_tCharacteristic parameter value, R, representing the breathing signal acquired at time t_min、R_maxRepresenting the minimum, maximum, P, of a value of a characteristic parameter of the breathing signal over a period of time_tCharacteristic parameter value, P, representing the pulse signal acquired at time t_min、P_maxThe minimum value and the maximum value of the characteristic parameter values of the pulse signals in a period of time are represented;

R_t＝b₁×T1+b₂X A, in the formula b₁Is a weight coefficient of influence on the term of the period T1, b₂Is a weight coefficient of the effect on the amplitude A term, and b₁+b₂＝1；

P_t＝c₁×t₁+c₂×t₂+c₃×H₁+c₄×H₂+c₅X T2, wherein c₁Is for the rising branch time t₁Weight coefficient of influence of term, c₂Is for the descending branch time t₂Weight coefficient of influence of term, c₃For the rising branch height H₁Weight coefficient of influence of term, c₄For descending branch height H₂Weight coefficient of influence of term, c₅Is a weight coefficient of influence on the term of the period T2, and c₁+c₂+c₃+c₄+c₅＝1；

Step 3.3: carrying out weighted fusion on the characteristic parameters of the three normalized signals by using a characteristic fusion strategy to obtain a fusion value f (E) of the three signals_tn,R_tn,P_tn) I.e. the unit energy consumption kcal/min/kg;

f(E_tn,R_tn,P_tn)＝a×E_tn+b×R_tn+c×P_tn

in the formula, a, b and c respectively represent weighted values of electromyographic signals, respiratory signals and pulse signals, the value range of a is 0.55-0.65, the value range of b is 0.15-0.25, the value range of c is 0.15-0.25, and a + b + c is 1;

where n' represents the total divided division phase within the wearer movement time period T, T_iRepresenting the duration of motion in the ith division stage;

and 4, step 4: collecting a respiration signal, a pulse signal and a myoelectric signal at a shank within a period of time T when a wearer does not wear the flexible exoskeleton in advance, and carrying out multi-mode information fusion to obtain an energy consumption value E' within the same time when the wearer does not wear the flexible exoskeleton;

and 5: comparing the energy consumption value E within a period of time T when the flexible exoskeleton is worn with the energy consumption value E' within the same period of time when the flexible exoskeleton is not worn, outputting an energy consumption feedback result and a corresponding zone bit of the output state of the stay wire steering engine, wherein the specific expression is as follows:

step 6: extracting an actually measured tension value of the tension signal after noise reduction, comparing the actually measured tension value with a preset tension value, and outputting a tension signal feedback result and a corresponding zone bit of the output state of the stay wire steering engine, as shown in fig. 7; the concrete expression is as follows: extracting an actually measured tension value F according to the tension signal acquired in real time_lMaximum value F of preset tension value_l' max or minimum value F_l' min is compared if F_l<F_lMin, increasing the PWM value for controlling the stay wire steering engine to be (1+ phi) times of the original value, and marking the output state of the stay wire steering engine at a mark position l2, wherein phi represents a tension feedback threshold value; if F_l≥F_lMax controls the stay wire steering engine to enter a power-off protection state, and the stay wire steering engine needs to output a mark position l3 of the state;

and 7: the classifier is constructed based on the artificial neural network, the classifier is used for recognizing the movement intention according to the acceleration signal after noise reduction, the recognition result of the movement intention is output, and the corresponding zone bit of the output state of the stay wire steering engine is specifically expressed as follows:

step 7.2: adopting a sliding window algorithm to carry out gait phase division on the signal ACC _ x; the method comprises the following steps:

step 7.2.3: dividing gait phases according to phase division points, wherein the corresponding gait phases between an initial time point 0 and a time point A1 are in a standing state, the corresponding gait phases between a time point A1 and a time point A2 are in a heel-off state, the corresponding gait phases between a time point A2 and a time point A3 are in a swinging state, and the corresponding gait phases between a time point A3 and a time point A4 are in a sole landing state; the corresponding gait phase from the time point A4 to the first time point A1 of the next cycle is the heel-strike state;

and 8: and outputting a PWM signal for controlling the rotation angle value of the output shaft of the stay wire steering engine according to the flag bit of the output state of the stay wire steering engine corresponding to the motion intention recognition result, the tension feedback result and the energy consumption feedback result and a preset flag bit priority level, wherein the flag bit priority level is set to l3 & gtl 1 & gtl 4 & gtl 2 & gtl 0.

The types of the components adopted in the embodiment are as follows: the inertial sensor IMU is WT901C485, the tension and compression sensor is HYLY-019, the myoelectric sensor is ZTEMG-1300, the respiration sensor is HKH-11C +, the pulse sensor is MAX30102, the microprocessor is raspberry RaspberryPi, the amplifier is Bluetooth, and the wireless transmission module is Bluetooth; the AD conversion module is AD7606, the display screen is MAKEBIT7 inch IPS high definition screen, a specific wiring schematic diagram is shown in FIG. 5, and an energy consumption monitoring display interface is compiled on an upper computer by labview as shown in FIG. 9;

the IMU inertial attitude sensor collects acceleration signals, a USB interface is connected with a raspberry type RaspberryPi in an RS-485 serial port communication mode, so that the collection and transmission of the acceleration signals are achieved, the signals are transmitted to the RaspberryPi and then subjected to filtering processing and then recognized as movement intentions, the recognition results are displayed and fed back through a portable display screen connected with the RaspberryPi through an HDMI interface, meanwhile, the microprocessing is also used for generating PWM signals according to the recognition results to serve as control signals of the stay wire steering engine, namely, the change of duty ratio is used for changing the rotation angle of an output shaft of the stay wire steering engine, so that the forward and reverse rotation of the high-power stay wire steering engine are achieved; the pull-press sensor acquires the pull force at the tail end of the Bowden cable A mechanism, the pull force is electrically connected with the AD conversion module through the amplifier, the data transmission between the AD conversion module and the RaspberryPi is realized through an SPI protocol, the data is filtered, a feedback result is output by judging whether the pull force value is within a preset range, and a PWM signal is generated according to the feedback result and is used for controlling a high-power pull-wire steering engine; it should be noted that in order to avoid the stay wire steering engine receiving multiple control signals at the same time, the priority level of each control signal needs to be set in advance, and the mode of setting the priority level of the output state flag bit of the stay wire steering engine is adopted to determine which control signal is preferentially output; a high-power stay wire steering engine drives a wire spool to control the folding and unfolding of the Bowden cable mechanism, so that the assisting and unloading of hip joints and ankle joints are realized; the IMU inertial sensors are connected in a cascade mode, and a Modbus protocol is matched with an RS-485 physical serial port to carry out data transmission;

the tension signal input by the port of the tension and compression sensor is pre-amplified by the amplifier, the tension signal data is amplified to the sampling range of the AD conversion module and input to the AD conversion module, and the tension signal data output by the AD conversion module is input to the RaspberryPi. The measuring range of the tension and compression sensor is 50Kg, the diameter size is 6mm, the characteristic curve of the tension and compression sensor is linear, the tension value is converted into a voltage value to represent, the sensitivity is 1.300mV/V, the tension and compression sensor is fixed at the heel of the working boot, and the tension borne by the tail end of the wire in the Bowden wire mechanism A is measured; the amplifier is a differential amplifier, and voltage signals v1 and v2 are collected by two channels when signals are collected. The two voltage signals are amplified by the differential amplifier, and the processed signals are output to an AD acquisition signal channel for AD conversion of the signals.

The external 12V power supply of pressure sensor drive draws, and the 12V power is connected with the pressure sensor through constant voltage constant current module, and wherein input constant voltage constant current power supply module adopts LM 2596.

The transmission process of the tension signal is as follows: after the RaspberryPi writes data into the sending data buffer, a clock signal is generated under the control of the RaspberryPi, the RaspberryPi controls the GPIO pin to pull down the level of the chip selection pin, after the CS pin of the AD7606 detects that the level is pulled down, the AD7606 transmission object is selected, the data is sent one bit at a time from the 8-bit shift register according to the sequence from a high bit (MSB) to a low bit (LSB), the RaspberryPi receives one bit at each clock cycle, 8 clock signal cycles pass, after the transmission is completed, the RaspberryPi reads the data from the receiving buffer and converts the data into the data through an electric signal to complete the whole transmission process.

When RaspberryPi and AD7606 are connected, the connecting pin is divided into two parts: the device comprises a control pin and a data transmission pin, wherein the control pin is mainly operated by RaspberryPi and used for configuring AD7606 parameters and controlling analog-to-digital conversion, and the data transmission pin is mainly connected with an analog-to-digital converter by adopting an SPI protocol.

The tension signal filtering is median filtering.

The tension feedback is designed as follows: controlling the pulling force to be in a proper range according to a preset pulling force range, and gradually increasing the position control amount if the measured pulling force is smaller than the expected pulling force minimum value to ensure the wearing effect of the flexible exoskeleton; if the measured pulling force is smaller than the expected pulling force maximum value, the flexible exoskeleton can enter a power-off protection state to prevent the wearer from being hurt due to the accident fault.

The electromyographic sensor with the model of ZTEMG-1300 is externally connected with two dipoles, a single-stage lead wire, an electric patch is pasted at the position of calf muscle and is connected with the port of the electromyographic sensor through the lead wire, the electromyographic sensor can measure the tension of the muscle, 7 voltages are used for amplifying and adjusting the frequency, and the electric signals of different muscles are accurately measured; wherein the unit of the electromyographic signal is mu V, the range of the electromyographic parameter is 0-60 mu V, and the precision is 0.01 mu V.

When a respiration signal is measured by using a respiration sensor, the plastic hose surrounds the chest of a wearer for a circle and is fixed through a port of the respiration sensor, and in an experiment, attention needs to be paid to the fact that a conduit cannot be bent or twisted sharply, because the data is inaccurate, the resistance of a human body changes due to thoracic motion in the respiration process, the change graph of the resistance value describes the dynamic waveform of respiration, and the respiration signal is obtained after the change graph of the resistance value is converted by a conversion circuit; wherein the unit of the measured respiratory signal is cm, the respiratory parameter range is +/-11 cm, the precision is 1mm, and the resolution is 0.2 mm.

The pulse sensor with the model of MAX30102 is fixed on a nylon belt through a nylon thread gluing on the pulse sensor to form a bracelet form, and the bracelet form is directly sleeved on the wrist of a wearer, wherein the unit of a measured pulse signal is bests/min, the pulse parameter range is 30-200bpm (bests/min), and the display resolution is 1 bpm.

The wireless transmission module transmits the collected myoelectric signals, respiratory signals and pulse signals to the upper computer by using Bluetooth.

The collected information needs to be subjected to noise reduction treatment, the electromyographic signals are filtered by a four-order Butterworth band-pass filter, and the pass band range is set to be 25Hz-200 Hz; the respiratory signal is firstly subjected to 40-order FIR low-pass filtering and then is subjected to smooth filtering; the pulse wave signal selects Sym5 wavelet as a basis function, and high-frequency noise and baseline drift are eliminated by using a method based on wavelet decomposition and reconstruction.

In addition, because the physiological information data come from different physiological information acquisition components, in order to enable the myoelectric, respiratory and pulse physiological information data to have the same time reference, and the acquired physiological information can be subjected to corresponding acquisition operation through information synchronization processing, the invention realizes data sharing by matching a shared memory based on semaphore, takes a timer as a resampling standard in the main process of data acquisition, simultaneously performs data segmentation on the original data in a circulating sliding time window mode, packages and stores the synchronized data, and periodically transmits the acquired data packets to a multi-mode information fusion module for processing and analysis by a physiological information acquisition main control.

When the device is used, the flexible exoskeleton is worn on the right lower limb of a wearer, the wearer can carry out normal daily activities, the exercise intention recognition result is observed in real time through the portable display screen, and whether the data is effective in real time is verified; and displaying the collected electromyographic signals, respiratory signals, pulse signals and the calculated energy consumption value on an energy consumption monitoring display interface in real time.

The exercise intention recognition is designed as follows: constructing a classifier based on an artificial neural network, wherein the classifier comprises 32 input layer nodes, 20 hidden layer nodes and n output nodes, and n represents the number of action types of behaviors to be classified; the data segmentation stage takes the form of a rectangular window of 250/30; finally, the recognition accuracy rate obtained by multiple classifier tests is 95.08% +/-0.44, and the action recognition time is less than 300 ms.

And selecting the acquired ACC _ x signal as a basis for analyzing each phase time point in one gait cycle. The invention adopts a sliding window algorithm to determine each gait cycle time point, thereby realizing the division of human gait phases. The size of the sliding window is selected to be N-30, the overlapping rate of the sliding window is set to be 50%, taking walking as an example, 5 gait phases are respectively: standing state, heel off the ground, swinging state, sole landing, heel landing. When the ACC _ x signal collected by the IMU at the thigh begins to change for 20 consecutive values within the sliding window, the stance phase is confirmed to be over,entering heel off phase; thereafter, the sliding window continues to move backward, when the value of ACC _ x signal at the position of the calf is detected for the first time as 1 (here ω is_xWhen the approximate value is 1), recording that the moment is in a swing state; the sliding windows continue to move backwards, and when the ACC _ x signal value at the thigh is detected to be the maximum value for the second time in one sliding window, the sole landing is recorded at the moment; and the sliding window continuously moves backwards, when detecting that 20 continuous ACC _ x signals at the chest are approximately unchanged in one sliding window, recording that the ACC _ x signals are heel landed at the moment, determining all key time points in one gait cycle, and dividing the gait phase in the gait cycle according to the key time position points determined by the sliding window.

The energy consumption feedback is designed as follows: measuring the energy consumption E 'of a wearer who does not wear the exoskeleton in advance for a certain time when the wearer moves, and taking a 20% value of the energy consumption E' as a difference value to define an energy consumption threshold, wherein the energy consumption upper threshold is 120% E ', and the energy consumption lower threshold is 80% E'; when an energy consumption monitoring experiment is carried out, measuring the energy consumption E of a wearer wearing the exoskeleton at the same time during movement, comparing the energy consumption E with a set threshold value, and if the energy consumption E is smaller than the set minimum threshold value, indicating that the flexible exoskeleton plays a power assisting effect on the wearer, wherein the feedback result is 'power assisting'; if the E is between the set maximum threshold and the set minimum threshold, the flexible exoskeleton does not have the assistance effect on a wearer, the feedback result is 'no assistance', and the rotation angle of the output shaft is increased; if E is larger than the set maximum threshold value, the flexible exoskeleton plays a resistance effect on a wearer, the feedback result is 'resistance', and the high-power steering engine is stopped for inspection.

Wherein the minimum time for measurement is 30 minutes, 45 minutes, 60 minutes and the like can be taken, the time does not exceed 105 minutes, and the measurement of energy consumption is influenced by too short or too long time.

The hardware connection state displayed on the energy consumption monitoring display interface displays the connection condition of the myoelectric sensor, the respiration sensor and the pulse sensor with the upper computer, and when the upper computer is successfully connected with the hardware of each sensor, the corresponding indicator light is on; and when the upper computer is not successfully connected with the hardware of each sensor, the corresponding indicator lamp is turned off.

The tester information displayed on the energy consumption monitoring display interface comprises a tester name and the date of birth, and each data is recorded by taking the tester name and the date of birth as labels.

The energy consumption analysis result displayed on the energy consumption monitoring display interface is displayed through an instrument panel, the instrument panel is divided into 0-150 scales, when a pointer on the instrument panel points to a certain scale, the energy consumption value of a wearer at the moment is represented, the unit is kcal, and the instrument panel is provided with a digital display and has 6 effective digits; the lower right corner has running time in min, which represents the total time from the start of the multi-element signal recording to the current state; the left side is the multi-element signal ratio which is represented by a horizontal pointer sliding rod numerical value, and the ratio of the multi-element signal can be changed by dragging the sliding rod.

The energy consumption monitoring display interface is provided with an electromyographic signal waveform diagram, a respiratory signal waveform diagram, a pulse signal waveform diagram and respective waveform display buttons, when the waveform display buttons are pressed, the corresponding waveform diagrams are displayed, the abscissa of the waveform diagram is time, the unit is t, the ordinate of the waveform diagram is amplitude, the unit of the electromyographic signal is mu V, the unit of the respiratory signal is cm, and the unit of the pulse signal is bests/min.

Claims

1. A flexible exoskeleton control method capable of monitoring multivariate physiological energy consumption of a wearer is realized by adopting a flexible exoskeleton system capable of monitoring multivariate physiological energy consumption of the wearer, the system comprises a flexible exoskeleton, a motion signal acquisition unit and a physiological signal acquisition unit, the flexible exoskeleton comprises a wire pulling mechanism, braces, a Bowden wire mechanism, a wearing soft shell and a working boot, the wire pulling mechanism comprises a wire pulling steering engine, a frame, a wire reel and an outer carbon plate, the outer carbon plate is fixed on the opposite side surface of the frame, two wire pulling steering engines are fixed on the frame, the output shaft of each wire pulling steering engine is connected with the central shaft of the wire reel, the wire reel is fixed on the outer carbon plate through a horizontal bearing seat, one end of the Bowden wire mechanism is wound on the wire reel, the other end of the Bowden wire mechanism is connected with the corresponding part of the wearing soft shell, the motion signal acquisition unit is used for acquiring acceleration signals of different parts of a human body and identifying the motion intention of the human body according to the acceleration signals, then controlling the movement of a stay wire steering engine in the flexible exoskeleton according to the movement intention recognition result; the physiological signal acquisition unit is used for acquiring myoelectric signals, pulse signals and respiratory signals of a human body, and performing multi-mode information fusion on the acquired signals to output an energy consumption value within a period of time for controlling the movement of a stay wire steering engine in the flexible exoskeleton; drive bowden cable mechanism through the steering wheel of acting as go-between and drive respectively and dress soft shell, work boots motion, dress the soft shell and include waist soft shell, thigh soft shell, shank soft shell, bowden cable mechanism includes two respectively for bowden cable mechanism A and bowden cable mechanism B, its characterized in that, the method includes:

step 1: collecting respiratory signals, pulse signals and myoelectric signals at crus of a wearer within a period of time T when the wearer wears the flexible exoskeleton, simultaneously collecting acceleration signals of the front of the chest, the outer side of the thigh on the right side and the outer side of the crus on the right side of the wearer, and collecting tension signals of wires in the Bowden wire mechanism A;

and 7: constructing a classifier based on an artificial neural network, identifying the movement intention by using the classifier according to the noise-reduced acceleration signal, and outputting a movement intention identification result;

2. The flexible exoskeleton control method of claim 1 wherein step 3 comprises:

in the formula, E_tCharacteristic parameter values representing the electromyographic signals acquired at time t, E_min、E_maxMinimum and maximum values of a parameter representing the characteristic of the myoelectric signal over a period of time, R_tCharacteristic parameter value, R, representing the breathing signal acquired at time t_min、R_maxRepresenting the minimum, maximum, P, of a value of a characteristic parameter of the breathing signal over a period of time_tIndicating acquisition at time tCharacteristic parameter value of pulse signal, P_min、P_maxThe minimum value and the maximum value of the characteristic parameter values of the pulse signals in a period of time are represented;

f(E_tn,R_tn,P_tn)＝a×E_tn+b×R_tn+c×P_tn

3. The flexible exoskeleton control method of claim 1 wherein step 5 is specifically expressed as:

4. The flexible exoskeleton control method of claim 1 wherein step 6 is specifically expressed as: extracting an actually measured tension value F according to the tension signal acquired in real time_lMaximum value F of preset tension value_l' max or minimum value F_l' min is compared if F_l<F_l'min', then PW controlling stay wire steering engineIncreasing the M value to be (1+ phi) times of the original value, and setting the flag bit of the output state of the stay wire steering engine to be l2, wherein phi represents a tension feedback threshold value; if F_l≥F_lAnd max controls the stay wire steering engine to enter a power-off protection state, and the flag bit of the output state of the stay wire steering engine needs to be set to l 3.

5. The flexible exoskeleton control method of claim 1 wherein said step 7 comprises:

6. The flexible exoskeleton control method of claim 5 wherein said step 7.2 comprises:

step 7.2.2: extracting a phase division point of the acceleration signal, and marking a time point when m continuous amplitude values of a signal ACC _ x corresponding to the thigh outer side acceleration signal start to change in a sliding window as a phase division point A1; the sliding window continuously moves backwards, and the amplitude value of a signal ACC _ x corresponding to the acceleration signal of the outer side of the crus detected for the first time is omega_xThe time point of time is marked as phase division point a 2; wherein ω is_x10% g, where g represents acceleration of gravity;the sliding windows continue to move backward, and the time point when the amplitude of the thigh outer acceleration signal corresponding signal ACC _ x is detected for the second time in one sliding window is marked as the phase division point a 3; the sliding window continues to move backwards, and a time point when m continuous amplitude values of a signal ACC _ x corresponding to the chest acceleration signal are detected to be unchanged in one sliding window is marked as a phase division point A4;