CN114190921A - Gait phase recognition method capable of adapting to variable step frequency walking - Google Patents

Gait phase recognition method capable of adapting to variable step frequency walking Download PDF

Info

Publication number
CN114190921A
CN114190921A CN202010908495.3A CN202010908495A CN114190921A CN 114190921 A CN114190921 A CN 114190921A CN 202010908495 A CN202010908495 A CN 202010908495A CN 114190921 A CN114190921 A CN 114190921A
Authority
CN
China
Prior art keywords
phase
foot
gait
data
zero
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202010908495.3A
Other languages
Chinese (zh)
Other versions
CN114190921B (en
Inventor
赵新刚
谈晓伟
张弼
赵明
姚杰
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shenyang Institute of Automation of CAS
Original Assignee
Shenyang Institute of Automation of CAS
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Shenyang Institute of Automation of CAS filed Critical Shenyang Institute of Automation of CAS
Priority to CN202010908495.3A priority Critical patent/CN114190921B/en
Publication of CN114190921A publication Critical patent/CN114190921A/en
Application granted granted Critical
Publication of CN114190921B publication Critical patent/CN114190921B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/103—Detecting, measuring or recording devices for testing the shape, pattern, colour, size or movement of the body or parts thereof, for diagnostic purposes
    • A61B5/11—Measuring movement of the entire body or parts thereof, e.g. head or hand tremor, mobility of a limb
    • A61B5/112—Gait analysis
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/103—Detecting, measuring or recording devices for testing the shape, pattern, colour, size or movement of the body or parts thereof, for diagnostic purposes
    • A61B5/1036—Measuring load distribution, e.g. podologic studies
    • A61B5/1038—Measuring plantar pressure during gait
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/68—Arrangements of detecting, measuring or recording means, e.g. sensors, in relation to patient
    • A61B5/6801—Arrangements of detecting, measuring or recording means, e.g. sensors, in relation to patient specially adapted to be attached to or worn on the body surface
    • A61B5/6802—Sensor mounted on worn items
    • A61B5/6804—Garments; Clothes
    • A61B5/6807—Footwear
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7203—Signal processing specially adapted for physiological signals or for diagnostic purposes for noise prevention, reduction or removal
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7235—Details of waveform analysis
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7235—Details of waveform analysis
    • A61B5/725—Details of waveform analysis using specific filters therefor, e.g. Kalman or adaptive filters

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Surgery (AREA)
  • Animal Behavior & Ethology (AREA)
  • Veterinary Medicine (AREA)
  • Public Health (AREA)
  • Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Biophysics (AREA)
  • Pathology (AREA)
  • Biomedical Technology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Medical Informatics (AREA)
  • Molecular Biology (AREA)
  • Physiology (AREA)
  • Signal Processing (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Psychiatry (AREA)
  • Dentistry (AREA)
  • Oral & Maxillofacial Surgery (AREA)
  • Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)

Abstract

The invention relates to a gait phase recognition method suitable for variable step frequency walking, which is characterized in that an inertia measurement unit is placed on a foot to collect human motion data; designing a foot extreme value detection algorithm capable of carrying out self-adjustment of structural parameters according to the step frequency and amplitude characteristics to realize stable extraction of the extreme value point characteristics of the foot motion data; a foot zero-speed detection algorithm fused with the self-adaptive oscillator is designed, so that the foot zero-speed state characteristics are accurately extracted; and classifying and analyzing the extracted foot characteristic values by using a finite state machine, thereby realizing the accurate identification of four phases of the foot touchdown phase, the standing end phase and the pre-swing phase in the walking mode with variable step frequency. The invention ensures that the identification process of the foot gait phase is not limited to the walking mode under the laboratory environment and the fixed step frequency, effectively improves the gait phase identification accuracy under the complex environment and the motion mode, and is beneficial to the application and development of gait analysis in the field of medical rehabilitation and rehabilitation robots.

Description

Gait phase recognition method capable of adapting to variable step frequency walking
Technical Field
The invention relates to the technical field of biological signal identification, in particular to a gait phase identification method suitable for variable step frequency walking.
Background
Along with the development of wearable technology, more and more intelligent wearable equipment is developed and is used for recording and monitoring various physiological information of a human body all weather and all around. The human gait characteristic is taken as a frequently-occurring dissimilarity of human physiological information, so that the identification and detection of the human gait phase characteristic have important significance.
The gait phase is an important component of human gait characteristics and is often used for diagnosing the disease degree of dyskinesia people, and the identification of the gait phase is also often applied to the design of a control system of a rehabilitation robot, aiming at improving the active perception capability of the rehabilitation robot. The gait phase recognition work is mainly divided into two parts, namely the design of the wearable sensor and the research of a recognition algorithm. The wearable sensors generally adopt plantar pressure sensors and an inertia measurement unit at present, the plantar pressure sensors can weaken the research difficulty of an identification algorithm and simplify the identification process of gait phases, but the stability is poor, the service life is short, and the requirement of long-term use is difficult to meet; the inertia measurement unit is convenient to wear, the hardware maturity is high, the motion data of the feet of the human body can be measured in real time, but the stability of the data is easily influenced by the motion mode of the human body and the walking environment, and therefore a stable and effective gait phase recognition algorithm is difficult to design. In the aspect of identification algorithm, the existing human gait phase identification method generally lacks the adaptability to the application environment, for example, when the subject walks at a variable step frequency, the existing method is difficult to adapt to the step frequency-variable walking mode, so that the gait phase of the subject cannot be stably identified and segmented. Therefore, even though the inertia measurement unit can conveniently and quickly provide effective data, the application and development of human gait phase analysis in the field of medical rehabilitation and rehabilitation robots are limited by the defects of the existing identification method.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides the gait phase recognition method which is suitable for variable step frequency walking, the accuracy of the gait phase recognition in a complex environment and a motion mode is improved, the recognition method is not required to be adjusted off-line by human intervention, and the requirements of the medical treatment, the robot and other fields on real-time detection and monitoring of human gait phases can be effectively met.
The technical scheme adopted by the invention for realizing the purpose is as follows:
a gait phase recognition method suitable for variable step frequency walking comprises the following steps:
collecting foot gait data in the exercise process and carrying out data preprocessing;
extracting extreme point characteristics of the foot sagittal plane rotation angle and the foot sagittal plane rotation angular velocity by using a self-adaptive extreme detection algorithm;
estimating the zero-speed state characteristics of the foot by using a zero-speed detection algorithm of a fusion self-adaptive oscillator;
and identifying the gait phase of the foot by using the extracted characteristic value and combining a finite state machine.
The foot gait data in the movement process is collected through an inertia measuring unit, the inertia measuring unit is placed on the outer side of an ankle joint of a human body, and the inertia measuring unit does not generate relative displacement with the foot in the movement process of the foot.
The foot gait data comprises foot sagittal plane rotation angle, angular velocity and acceleration.
The data preprocessing comprises the following steps: filtering and estimating a direction angle of the acquired data; the filtering adopts low-pass filtering, and the direction angle estimation adopts a gradient descent method.
The method for extracting the foot motion extreme point features by using the self-adaptive extreme detection algorithm comprises the following steps:
1) for the preprocessed foot data, the head and tail data point information of the sliding window is utilized to preliminarily detect whether the current sliding window contains the extreme point pn；
2) If so, the adaptive acknowledgement gain is used to perform the preliminary detectionJudging the extreme point, if the extreme point is reached
Figure BDA0002662366910000021
Or
Figure BDA0002662366910000022
Confirming the current maximum value point or minimum value point, updating the extreme value window in an iterative manner, and returning to the step 1) to search again if not;
the extremum window at time n may be defined as:
Figure BDA0002662366910000023
wherein, TeThe coefficient is a constant coefficient,
Figure BDA0002662366910000024
an extremum window at time N-1, NeIs the data length of the extremum window;
Figure BDA0002662366910000025
for the value of the confirmed extreme point, { } represents the sequential splicing combination of the internal data.
The preliminary detection extreme point pnThe method comprises the following steps: when the head and tail data point values in two continuous sliding windows satisfy:
Figure BDA0002662366910000031
and is
Figure BDA0002662366910000032
Time, maximum value point pnHas already been preliminarily searched;
or,
Figure BDA0002662366910000033
and is
Figure BDA0002662366910000034
Time, minimum value point pnHas been already covered byPreliminary search is carried out;
the formula for the maximum or minimum point is:
Figure BDA0002662366910000035
wherein,
Figure BDA0002662366910000036
n is a value of the time of day,
Figure BDA0002662366910000037
for the ith data point of the sliding window at time N, NdIs the data length of the sliding window and,
Figure BDA0002662366910000038
can be expressed as a numerical integral of the sliding window at time n and the step size of the sliding window is 1 data point.
The estimation of the foot zero-speed state characteristics by using the zero-speed detection algorithm of the fusion adaptive oscillator comprises the following steps:
modeling and estimating a foot sagittal plane rotation angle signal through an adaptive oscillator, carrying out adaptive modification on an index value threshold of a zero-speed detection algorithm by using a step frequency parameter of an estimated model, and judging the zero-speed state when an index value S of the zero-speed detection algorithm meets the following formula:
S≤Tzupt(n)
Figure BDA0002662366910000039
wherein, Tzupt-NPIs a fixed threshold value omega obtained by adopting the traditional Neyman-Pearson methodnFoot step frequency characteristics estimated for adaptive oscillators, czAnd kzIs a constant coefficient related to the gait characteristics of the foot, TzuptAnd (n) is the adaptive threshold at time n.
The combined finite state machine divides the foot gait phase into:
four gait phases: a grounding phase, a standing end phase, a pre-swing phase and a different phase which are 5 states are used as a state chain of the finite state machine; the abnormal phase is used for connecting the pre-swing phase and the standing phase, and is used for avoiding the blockage caused by the transition of the finite state machine state due to the fact that the touchdown phase cannot be identified.
The correspondence between the extracted characteristic value and the gait phase includes:
t1: the leaving of the touchdown phase and the entering of the standing phase correspond to the foot in a zero-speed state;
t2: the leaving of the standing phase and the entering of the standing end phase correspond to the feet in a non-zero speed state;
t3: the leaving of the end-of-stance phase and the entering of the pre-swing phase correspond to the plantar flexion extreme point of the sagittal angular velocity of the foot;
t4: the leaving of the pre-swing phase and the entering of the touchdown phase correspond to the dorsiflexion extreme point of the sagittal angle of the foot;
t5: the leaving of the end phase of standing and the entering of the standing phase correspond to the foot in the zero-speed state.
A gait phase recognition system adaptable to variable stride frequency walking, comprising: the gait phase recognition method comprises a controller, a memory and an inertia measurement unit, wherein the inertia measurement unit collects gait data and sends the gait data to the controller, the memory stores a program, the controller loads the program to execute the steps of the gait phase recognition method which can adapt to step frequency changing walking according to any one of claims 1 to 9, and therefore gait features are extracted, and the finite state machine is combined to recognize the gait phase of the foot.
The invention has the following beneficial effects and advantages:
the invention provides a gait phase recognition method capable of adapting to variable step frequency walking by utilizing a wearable inertial measurement unit, the method can accurately and quickly recognize the gait phase characteristics of feet in a variable step frequency walking mode of a subject, manual off-line debugging or on-line regulation is not needed, the self-adaptive capacity is strong, the environmental constraint requirement is low, and the gait characteristics of a human body can be effectively described.
Drawings
FIG. 1 is a schematic view of the inertial measurement unit mounting and measurement locations during the practice of the present invention;
FIG. 2 is a schematic flow diagram of the method of the present invention;
FIG. 3 is a graph of an embodiment of an extremum detecting algorithm in the present invention;
FIG. 4 is a graph of an embodiment of the zero speed detection algorithm of the present invention;
FIG. 5 is a flow chart of the finite state machine state transition in the gait phase recognition method of the invention;
FIG. 6 is a diagram illustrating a correspondence between a finite state machine state transition event and a foot feature according to the present invention;
figure 7 is a graph of gait phase identification accuracy of the invention in a subject's step-frequency-varying locomotor mode.
Detailed Description
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in detail below. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention. This invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein, but rather should be construed as modified in the spirit and scope of the present invention as set forth in the appended claims.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. The terminology used in the description of the invention herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.
As shown in figure 1, the inertial measurement unit is placed on the outer side of the ankle joint of the human body, has no generality, and can be placed on any other part of the foot of the human body for measuring the motion data of the foot of the human body. The inertial measurement unit can be placed on the outer side of the shoe by using a fixing clip (as shown in fig. 1), so that the inertial measurement unit does not generate relative displacement with the foot of the human body in the normal movement process of the human body.
Fig. 2 shows a flow chart of gait phase recognition according to the invention.
The implementation process of the invention can be divided into two parts, namely feature extraction and phase identification. The feature extraction algorithm is composed of an extreme value detection algorithm and a zero-speed detection algorithm. Before feature extraction and phase identification, low-pass filtering is performed on data of the inertial measurement unit to eliminate noise interference. The filter here uses a 3 rd order Butterworth filter, the parameters being selected to be a passband cut-off frequency of 20Hz, a stopband cut-off frequency of 45Hz, a passband maximum attenuation of 1dB, and a stopband minimum attenuation of 15 dB. In addition, the method for estimating the direction angle of the inertial measurement unit adopts a gradient descent method to reduce the drift of the direction angle.
Extreme value detection algorithm
The extreme value detection algorithm is used for detecting a numerical extreme value point (maximum value or minimum value) of the data of interest, firstly, the data of interest (the rotation angle and the angular velocity of the foot sagittal plane are selected here, and other types of data are also available) is collected from the inertial measurement unit, and the data of interest is divided into a sliding window as follows:
Figure BDA0002662366910000061
wherein n is a value of the time of day,
Figure BDA0002662366910000062
the ith data point (foot sagittal rotation angle and angular velocity) of the sliding window at the time N, NdIs the data length of the sliding window and,
Figure BDA0002662366910000063
can be expressed as a numerical integral of the sliding window at time n and the step size of the sliding window is 1 data point. The maximum and minimum search is described as an example, and the first step of the extremum detection algorithm is to initially search the extremum point position (extremum window at time n) and the value p of the datanThe process can be described as follows:
when the head and tail data point values in two continuous sliding windows satisfy:
Figure BDA0002662366910000064
time, maximum value point pnHas already been preliminarily searched;
or,
Figure BDA0002662366910000065
and is
Figure BDA0002662366910000066
Time, minimum value point pnHas already been preliminarily searched;
that is, when the values of the head and tail data points in two consecutive sliding windows satisfy the above formula, it can be considered that the extreme point has been initially searched, and the maximum value point or the minimum value can be expressed as follows:
Figure BDA0002662366910000067
wherein p isnThe numerical value of the extreme point is represented. Since the human foot motion data is susceptible to the environment, the second step is the confirmation of the extreme point to reduce the influence of the walking environment and the unknown noise on the extreme search, and the confirmation process can be described as follows:
if the extreme point
Figure BDA0002662366910000068
The current maximum point is confirmed;
if the extreme point
Figure BDA0002662366910000071
The current minimum value point is confirmed;
wherein
Figure BDA0002662366910000072
An extremum window of time N-1, of length NeWith sliding window, extremum window for storing historyConfirmed extreme point data. T iseThe constant coefficient can be selected according to the requirement (0.6 is selected here). mean (—) represents the mean of all data values in the extremum window. If the extreme value searching algorithm meets the formula, the extreme value point is considered to be searched and confirmed as a correct extreme value point, and then the next numerical iteration process of the extreme value point can be carried out, so that the adaptability of the extreme value searching algorithm to different human bodies and environments is improved. If the formula is not satisfied, the searched extreme point is considered to be a pseudo extreme point and is caused by numerical noise and the like, and at this time, the initial searching process of the extreme point is returned to the first step to continue. The numerical iteration process in which the extreme points are identified can be described as follows:
Figure BDA0002662366910000073
wherein
Figure BDA0002662366910000074
Wherein the function
Figure BDA0002662366910000075
Meaning that if the argument x in the parenthesis is greater than the argument a, the function equals the argument b, otherwise equals the argument x. std (—) represents the numerical variance of all data points within the extremum window. At this point, the extremum window at time n may be iteratively updated as:
Figure BDA0002662366910000076
wherein { } represents a sequentially spliced combination of internal data.
The algorithm implementing pseudo-code for the maximum detection embodiment is shown in table 1, and the results for this embodiment are shown in fig. 3, where the dashed line represents the confirmation gain, i.e.,
Figure BDA0002662366910000077
the circles represent the correct extreme points identified, and the solid lines of periodic oscillations represent the rotation angles of the foot sagittal plane, i.e., the raw input data for the extreme point detection. In addition, as can be seen from the embedded graph, when the historical extreme point value gradually decreases, the amplitude of the confirmation gain also decreases, so that the accuracy of the extreme value search algorithm and the adaptability to different human bodies and environments can be ensured.
TABLE 1 maximum detection Algorithm
Figure BDA0002662366910000081
Zero-speed detection algorithm
The zero-speed detection algorithm is used for detecting whether the inertial measurement unit is in a static state or not, and the gait modes of different human bodies and the gait modes of the same human body in different environments are different, but the only gait event which can definitely exist is the static state of the foot. In any circumstance where the person is walking, the foot of one leg must remain stationary to swing the other leg to move the body forward. Therefore, a stable resting state (hereinafter referred to as a zero velocity state) for detecting the human foot (i.e., the fixed position of the inertial measurement unit) can provide useful information for identification of the gait phase. The zero-speed detection algorithm can be described as follows:
Figure BDA0002662366910000082
wherein
Figure BDA0002662366910000083
Wherein S is an index value of a zero-speed detection algorithm, Tzupt-NPThe threshold value is determined by a Neyman-Pearson method, and when the index value is lower than the threshold value, the inertial measurement unit can be considered to be in a zero-speed state at present.
Figure BDA0002662366910000091
The noise variance of the accelerometer and gyroscope respectively,
Figure BDA0002662366910000092
accelerometer and gyroscope values, N, respectively, measured by the inertial measurement unit at time izAnd n and g are respectively the length of a time window, a time value and a modulus value of a gravity vector. However, the walking step frequency of the human body is constantly changed, and experiments find that in the face of constantly changed step frequency characteristics, a fixed threshold value is difficult to meet the requirement of zero-speed state stable identification, so that a self-adaptive oscillator is introduced to extract the human body step frequency characteristics on line, and a model of the self-adaptive oscillator can be described as follows:
Figure BDA0002662366910000093
Figure BDA0002662366910000094
Figure BDA0002662366910000095
Figure BDA0002662366910000096
wherein j is ∈ [1, K ∈ >]K denotes the number of parallel oscillators, ωnAnd (3) the step frequency characteristic value estimated by the oscillator at the moment n is referred to, and v and eta are learning factors to determine the learning speed of the oscillator.
Figure BDA0002662366910000097
To learn the signal, wherein θnIs an input signal, which is selected here as the angle of rotation of the foot in the sagittal plane of the human body,
Figure BDA0002662366910000098
the estimated value of the input signal can be expressed as follows:
Figure BDA0002662366910000099
wherein
Figure BDA00026623669100000910
And
Figure BDA00026623669100000911
the phase and amplitude parameters of the ith oscillator at time n,
Figure BDA00026623669100000912
is a non-oscillating term. Through the process of online learning and estimation, the step frequency characteristic omeganParameters of the estimation model, which are input signals, can be gradually extracted and used here to perform online adjustment of the threshold of the null velocity detection algorithm as follows
Figure BDA00026623669100000913
Wherein T iszupt(n) is an adaptive threshold for time n, czAnd kzIs a constant coefficient and is related to the gait characteristics of human body. As shown in FIG. 4, when the subject walks at three different speeds of 3kmh-1, 6 km.h-1 and 9 km.h-1, the adaptive oscillator can estimate the step frequency characteristics of human walking on line (as shown in the middle graph of FIG. 4), which are 0.68Hz, 0.96Hz and 1.36Hz, respectively, and the proposed zero-speed detection algorithm can adjust the threshold value T of the index value S on line according to the estimated step frequency characteristicszupt. Furthermore, as can be seen from the transition time of 6 km. h-1 to 9 km. h-1, if the index value threshold value T is setzuptWithout adaptive adjustment, it is difficult to accurately measure the zero-velocity state of the foot inertia measurement unit (i.e., the index value is smaller than the threshold value).
And the second part is to construct a finite state machine to identify four gait phases of a human body grounding phase, a standing end phase and a pre-swing phase by utilizing the two feature extraction algorithms of extreme value search and zero-speed detection.
As shown in fig. 5, in addition to the four gait phases to be identified, a heteromorphic phase is additionally added because the human gait phase contact is the gait phase most susceptible to external factors. The sagittal angle of the foot corresponding to touchdown may be maximum when walking on flat ground, and minimum angle when walking down stairs. The reason for establishing the abnormal phase is to prevent the abnormal blocking phenomenon of the phase transition caused by the continuous failure of the transition event T4 from the pre-swing phase to the ground contact phase.
For the transition event of the finite state machine (the corresponding diagram of the transition event and the gait feature is shown in fig. 6), T1, T2, T5 and T7 correspond to the zero speed detection algorithm, T1, T5 and T7 are event satisfied when the foot is in the zero speed state, and T2 is event satisfied when the foot leaves the zero speed state. The event satisfaction conditions of T3 and T4 correspond to extremum detection algorithms, that is, the extremum points of the foot sagittal plane rotation angle and the foot sagittal plane angular velocity are respectively corresponded to, and if the extremum points are detected, the event satisfaction can be considered. While the satisfaction condition of the event T6 transitioning to the off-state phase is defined as follows:
Figure BDA0002662366910000101
wherein t iselapsedIndicating the relative time length, T, of the relative latest pre-wobble phasegThe coefficient is a constant coefficient,
Figure BDA0002662366910000102
representing a gait cycle window, a sliding window and an extreme value window, wherein the gait cycle window is used for storing historical gait cycle values, NgRepresenting the length of the gait cycle window.
Gait phase recognition algorithm verification is carried out by taking gait data of a subject walking for 60s at three different speeds of 3 km.h < -1 >, 6 km.h < -1 > and 9 km.h < -1 > as an embodiment, and accuracy description is carried out by taking the gait phase collected by a sole pressure insole as a reference and correctly recognizing the gait phase in a limited time (10% of the time length of a gait cycle is selected) as false recognition.
When the subject walks at variable step frequency according to the requirements, the extreme value detection algorithm and the zero speed detection algorithm can be adaptive to the walking step frequency of the subject, and the online adjustment of the algorithm framework (fig. 3 and 4) can be continuously performed according to the step frequency characteristics, so that the gait characteristics of the subject are accurately extracted. The recognition accuracy is shown in fig. 7, wherein the average recognition accuracy of the standing phase and the pre-swing phase is respectively 99.1% and 99.7%, the two phases can be applied to two main phases of standing and swinging of human gait in the field of segmented medical treatment, and the ratio of the two main phases of standing and swinging can represent the degrees of diseases and dyskinesia of some patients, so that the two main phases are important, but the current recognition accuracy can completely meet the requirements of the field of medical rehabilitation on gait analysis of patients. The average identification accuracy rates of the touchdown phase and the standing end phase are 93.8% and 96.6% respectively, as mentioned above, the touchdown phase is most susceptible to the influence of a walking mode, but due to the establishment of the abnormal phase, the finite state machine can still transit to the standing phase through the abnormal phase in 6.2% of the period of mistaken identification of the touchdown phase, and therefore the condition transition is prevented from being blocked, and the identification of the subsequent gait phase is prevented from being influenced.
While the foregoing is directed to the preferred embodiment of the present invention, it will be understood by those skilled in the art that various changes and modifications may be made without departing from the spirit and scope of the invention as defined in the appended claims.

Claims (10)

1. A gait phase recognition method suitable for variable step frequency walking is characterized by comprising the following steps:
collecting foot gait data in the exercise process and carrying out data preprocessing;
extracting extreme point characteristics of the foot sagittal plane rotation angle and the foot sagittal plane rotation angular velocity by using a self-adaptive extreme detection algorithm;
estimating the zero-speed state characteristics of the foot by using a zero-speed detection algorithm of a fusion self-adaptive oscillator;
and identifying the gait phase of the foot by using the extracted characteristic value and combining a finite state machine.
2. The gait phase recognition method according to claim 1, characterized in that the gait data of the foot during the movement is collected by an inertial measurement unit which is placed outside the ankle joint of the human body and which does not displace relative to the foot during the movement of the foot.
3. The gait phase identification method according to claim 1, characterized in that the foot gait data includes foot sagittal plane rotation angle, angular velocity and acceleration.
4. The gait phase identification method according to claim 1, characterized in that the data preprocessing comprises: filtering and estimating a direction angle of the acquired data; the filtering adopts low-pass filtering, and the direction angle estimation adopts a gradient descent method.
5. The gait phase recognition method according to claim 1, wherein the extracting the foot motion extreme point feature by using the adaptive extreme detection algorithm comprises:
1) for the preprocessed foot data, the head and tail data point information of the sliding window is utilized to preliminarily detect whether the current sliding window contains the extreme point pn；
2) If yes, the initial detected extreme point is judged by using the self-adaptive confirmation gain, and if yes, the initial detected extreme point is judged
Figure FDA0002662366900000011
Or
Figure FDA0002662366900000012
The current maximum value point or minimum value point is confirmedIteratively updating the extremum window, otherwise returning to the step 1) to search again;
the extremum window at time n may be defined as:
Figure FDA0002662366900000021
wherein, TeThe coefficient is a constant coefficient,
Figure FDA0002662366900000022
an extremum window at time N-1, NeIs the data length of the extremum window;
Figure FDA0002662366900000023
for the value of the confirmed extreme point, { } represents the sequential splicing combination of the internal data.
6. The gait phase identification method according to claim 5, characterized in that the preliminary detection extreme point pnThe method comprises the following steps: when the head and tail data point values in two continuous sliding windows satisfy:
Figure FDA0002662366900000024
and is
Figure FDA0002662366900000025
Time, maximum value point pnHas already been preliminarily searched;
or,
Figure FDA0002662366900000026
and is
Figure FDA0002662366900000027
Time, minimum value point pnHas already been preliminarily searched;
the formula for the maximum or minimum point is:
Figure FDA0002662366900000028
wherein,
Figure FDA0002662366900000029
n is a value of the time of day,
Figure FDA00026623669000000210
for the ith data point of the sliding window at time N, NdIs the data length of the sliding window and,
Figure FDA00026623669000000211
can be expressed as a numerical integral of the sliding window at time n and the step size of the sliding window is 1 data point.
7. The gait phase recognition method according to claim 1, wherein the estimating the foot zero velocity state feature by using the zero velocity detection algorithm of the fusion adaptive oscillator comprises:
modeling and estimating a foot sagittal plane rotation angle signal through an adaptive oscillator, carrying out adaptive modification on an index value threshold of a zero-speed detection algorithm by using a step frequency parameter of an estimated model, and judging the zero-speed state when an index value S of the zero-speed detection algorithm meets the following formula:
S≤Tzupt(n)
Figure FDA0002662366900000031
wherein, Tzupt-NPIs a fixed threshold value omega obtained by adopting the traditional Neyman-Pearson methodnFoot step frequency characteristics estimated for adaptive oscillators, czAnd kzIs a constant coefficient related to the gait characteristics of the foot, TzuptWhen (n) is nAn adaptive threshold at the moment.
8. The gait phase identification method according to claim 1, characterized in that the foot gait phase is divided into, in combination with a finite state machine:
four gait phases: a grounding phase, a standing end phase, a pre-swing phase and a different phase which are 5 states are used as a state chain of the finite state machine; the abnormal phase is used for connecting the pre-swing phase and the standing phase, and is used for avoiding the blockage caused by the transition of the finite state machine state due to the fact that the touchdown phase cannot be identified.
9. The gait phase identification method according to claim 1, wherein the correspondence between the extracted feature value and gait phase includes:
t1: the leaving of the touchdown phase and the entering of the standing phase correspond to the foot in a zero-speed state;
t2: the leaving of the standing phase and the entering of the standing end phase correspond to the feet in a non-zero speed state;
t3: the leaving of the end-of-stance phase and the entering of the pre-swing phase correspond to the plantar flexion extreme point of the sagittal angular velocity of the foot;
t4: the leaving of the pre-swing phase and the entering of the touchdown phase correspond to the dorsiflexion extreme point of the sagittal angle of the foot;
t5: the leaving of the end phase of standing and the entering of the standing phase correspond to the foot in the zero-speed state.
10. A gait phase recognition system adaptable to variable stride frequency walking, comprising: the gait phase recognition method comprises a controller, a memory and an inertia measurement unit, wherein the inertia measurement unit collects gait data and sends the gait data to the controller, the memory stores a program, the controller loads the program to execute the steps of the gait phase recognition method which can adapt to step frequency changing walking according to any one of claims 1 to 9, and therefore gait features are extracted, and the finite state machine is combined to recognize the gait phase of the foot.
CN202010908495.3A 2020-09-02 2020-09-02 Gait phase identification method capable of adapting to variable-step-frequency walking Active CN114190921B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010908495.3A CN114190921B (en) 2020-09-02 2020-09-02 Gait phase identification method capable of adapting to variable-step-frequency walking

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010908495.3A CN114190921B (en) 2020-09-02 2020-09-02 Gait phase identification method capable of adapting to variable-step-frequency walking

Publications (2)

Publication Number Publication Date
CN114190921A true CN114190921A (en) 2022-03-18
CN114190921B CN114190921B (en) 2024-05-17

Family

ID=80644279

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010908495.3A Active CN114190921B (en) 2020-09-02 2020-09-02 Gait phase identification method capable of adapting to variable-step-frequency walking

Country Status (1)

Country Link
CN (1) CN114190921B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115429257A (en) * 2022-10-08 2022-12-06 华南理工大学 Gait cycle phase recognition method based on IMU and adaptive oscillator algorithm

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104899842A (en) * 2015-06-29 2015-09-09 济南大学 Ranked-order adaptively extremum median filtering method for remote line structured light image
WO2018081986A1 (en) * 2016-11-03 2018-05-11 浙江大学 Wearable device and real-time step length measurement method for device
CN108186021A (en) * 2017-12-22 2018-06-22 北京中科汇成科技有限公司 A kind of gait zero-speed detection method and system based on multimodal information fusion
CN109635820A (en) * 2018-08-06 2019-04-16 浙江大学 The construction method of Parkinson's disease bradykinesia video detection model based on deep neural network
US20200149757A1 (en) * 2016-02-18 2020-05-14 Johnson Controls Technology Company Noise-adaptive extremum-seeking controller

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104899842A (en) * 2015-06-29 2015-09-09 济南大学 Ranked-order adaptively extremum median filtering method for remote line structured light image
US20200149757A1 (en) * 2016-02-18 2020-05-14 Johnson Controls Technology Company Noise-adaptive extremum-seeking controller
WO2018081986A1 (en) * 2016-11-03 2018-05-11 浙江大学 Wearable device and real-time step length measurement method for device
CN108186021A (en) * 2017-12-22 2018-06-22 北京中科汇成科技有限公司 A kind of gait zero-speed detection method and system based on multimodal information fusion
CN109635820A (en) * 2018-08-06 2019-04-16 浙江大学 The construction method of Parkinson's disease bradykinesia video detection model based on deep neural network

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
XIAOWEI TAN; BI ZHANG; GUANGJUN LIU; XINGANG ZHAO;: "Cadence-Insensitive Soft Exoskeleton Design With Adaptive Gait State Detection and Iterative Force Control", 《IEEE TRANSACTIONS ON AUTOMATION SCIENCE AND ENGINEERING》, 26 March 2021 (2021-03-26) *
戴洪德;李松林;周绍磊;全闻捷: "基于伪标准差和N-P准则的行人导航零速检测", 《中国惯性技术学报》, 31 December 2018 (2018-12-31) *
赵新刚;谈晓伟;张弼: "柔性下肢外骨骼机器人研究进展及关键技术分析", 《机器人》, 30 June 2020 (2020-06-30) *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115429257A (en) * 2022-10-08 2022-12-06 华南理工大学 Gait cycle phase recognition method based on IMU and adaptive oscillator algorithm
CN115429257B (en) * 2022-10-08 2024-09-06 华南理工大学 Gait cycle phase identification method based on IMU and adaptive oscillator algorithm

Also Published As

Publication number Publication date
CN114190921B (en) 2024-05-17

Similar Documents

Publication Publication Date Title
CN114469073B (en) Gait analysis and abnormality detection method based on wearable sensor
WO2015113915A1 (en) Improvements in the detection of walking in measurements of the movement of a user
WO2016180653A1 (en) Pedometer and method for analyzing motion data
EP3090685A1 (en) Pedometer and method for analyzing motion data
CN115655265B (en) Pedestrian navigation indoor positioning method based on inertial sensing
CN104613965B (en) A kind of step-by-step movement pedestrian navigation method based on bidirectional filtering smoothing technique
Wang et al. Estimate spatial-temporal parameters of human gait using inertial sensors
CN107194193A (en) A kind of ankle pump motion monitoring method and device
CN108592907A (en) A kind of quasi real time step-by-step movement pedestrian navigation method based on bidirectional filtering smoothing technique
CN108955719B (en) Step counting detection method and system based on wrist type device
CN110319840A (en) Conjugate gradient attitude algorithm method towards abnormal gait identification
CN114190921A (en) Gait phase recognition method capable of adapting to variable step frequency walking
CN112244820A (en) Method for measuring running gait by using three-axis accelerometer
Song et al. Adaptive neural fuzzy reasoning method for recognizing human movement gait phase
CN114440883B (en) Pedestrian positioning method based on foot and leg micro-inertia measurement unit
WO2023108315A1 (en) Gait analysis and anomaly detection method based on wearable step sensor
CN114224325A (en) Gait analysis system and method for calculating joint moment and angle by using inertial sensor
Brzostowski Novel approach to human walking speed enhancement based on drift estimation
KR102192451B1 (en) Smart shoes based on recognition of combined walking action and data processing method hereof
Ghobadi et al. A robust automatic gait monitoring approach using a single IMU for home-based applications
KR101371655B1 (en) System for detecting zero velocity interval and method for detecting zero velocity interval
CN113445992B (en) Method and device for processing movement displacement of oil pumping unit
CN113229806A (en) Wearable human body gait detection and navigation system and operation method thereof
CN116206358A (en) Lower limb exoskeleton movement mode prediction method and system based on VIO system
CN111735478B (en) LSTM-based pedestrian real-time navigation zero-speed detection method

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant