Search

CN-122025006-A - Self-adaptive training parameter adjusting method for hand rehabilitation exoskeleton

CN122025006ACN 122025006 ACN122025006 ACN 122025006ACN-122025006-A

Abstract

The invention discloses a self-adaptive training parameter adjusting method of hand rehabilitation exoskeleton, which relates to the technical field of rehabilitation medical appliances, and comprises the specific steps of deploying a multi-mode sensing system to synchronously acquire full-dimension original signals of a patient and building a hand exercise rhythm feature library; the method comprises the steps of preprocessing, characteristic quantization and analysis to obtain a plurality of indexes, judging conflict and type through intention-conflict decoding, constructing a self-adaptive parameter adjustment model, executing a flexible solution strategy, setting a closed-loop adjustment period, circularly executing steps to finish scene parameter adjustment and implement advanced rehabilitation training, and generating adaptive training parameters by constructing a comprehensive and accurate patient state sensing system, accurately judging intention and capacity matching states, avoiding training risks, improving safety and suitability, constructing a closed-loop adjustment mechanism, combining scene training and advanced training, dynamically adjusting training, mobilizing patient enthusiasm, accelerating hand function recovery and improving rehabilitation effect and efficiency.

Inventors

  • WANG HAOBO
  • JIANG ZHITAO

Assignees

  • 中国计量大学

Dates

Publication Date
20260512
Application Date
20260130

Claims (10)

  1. 1. The self-adaptive training parameter adjusting method for the hand rehabilitation exoskeleton is characterized by comprising the following specific steps of: The method comprises the steps of S1, synchronously collecting multi-mode sensing signals, namely deploying a multi-mode sensing system, configuring a unified signal collecting protocol, synchronously collecting all-dimensional original signals of physiological fatigue, pain, pre-pain, exercise rhythm of a healthy hand, rehabilitation exercise intention and body capacity of a patient, extracting daily action rhythm characteristics of the healthy hand and building a healthy hand exercise rhythm characteristic library; S2, preprocessing and characteristic quantization are carried out on the full-dimensional original signal to obtain physiological fatigue, pain quantized value, pre-pain characteristic value, synchronous deviation value of exercise rhythm of healthy hand and rehabilitation exercise intention strength, and body capacity comprehensive parameters are obtained through calculation of a body capacity comprehensive parameter quantization formula; S3, intent-conflict decoding and type judging, namely inputting comprehensive parameters of the intent strength and the physical ability of rehabilitation exercise into a rehabilitation intent-conflict decoding logic layer, obtaining the matching degree through a rehabilitation intent-conflict matching degree calculation formula, and judging whether intent-ability instantaneous conflict and specific conflict types exist according to a preset threshold and a risk grading standard; s4, constructing a multi-dimensional coupling self-adaptive parameter adjustment model, calculating an adjustment coefficient through a multi-dimensional coupling training parameter adjustment formula, completing exoskeleton training parameter calculation, and executing a differential flexible solution strategy according to a conflict judgment result; And S5, setting 500ms as a closed-loop regulation period, circularly executing the steps, driving the exoskeleton driving module to synchronously respond to the regulation parameters, completing scene coupling parameter regulation of eating, dressing, writing and washing scenes, and implementing advanced rehabilitation training by combining the function recovery condition of the affected hand.
  2. 2. The method for adjusting self-adaptive training parameters of hand rehabilitation exoskeleton according to claim 1, wherein in step S1, the multi-mode sensing system is deployed by attaching a surface myoelectric sensor, a wrist wearing heart rate variability sensor, an exoskeleton key joint embedded deployment film pressure sensor, a patient face front fixed facial expression recognition module, a hand wearing flexible IMU bracelet and a patient hand exoskeleton integrated motion sensing unit to a finger core cardiac muscle group of a patient, configuring a uniform millisecond triggering type signal acquisition protocol, synchronously acquiring myoelectricity and heart rate variability signals representing physiological fatigue of the patient, joint impedance and facial microexpressive signals representing pain and pre-pain, action tracks and time sequence signals representing movement rhythms of the patient hand, myoelectricity strength and change trend signals representing rehabilitation exercise intention, joint power and impedance feedback signals representing body capacity according to the signal acquisition protocol.
  3. 3. The self-adaptive training parameter adjusting method for the hand rehabilitation exoskeleton is characterized in that in the step S2, the noise reduction processing adopts three modes of self-adaptive filtering, moving average filtering and wavelet denoising, the noise reduction processing is respectively adaptive to different original signals, physiological fatigue and rehabilitation exercise intention related signals adopt 5Hz-500Hz band-pass filtering, pain and pre-pain related signals adopt wavelet denoising, the number of decomposition layers is 3-5, the exercise rhythm related signals of the rehabilitation hands adopt moving average filtering, the sliding window size is 5-10 sampling points, the normalization processing adopts a linear normalization method, all denoised signals are uniformly mapped to a [0,1] value interval, when the characteristics are extracted and quantized and analyzed, physiological fatigue is obtained through weighted fusion of an myoelectric signal MPF mean value attenuation rate and heart rate variability time domain characteristics, a pain quantized value and a pre-pain characteristic value are obtained through human-computer interaction force peak value and micro-expression amplitude variation characteristic quantization, and rehabilitation intention exercise intensity is obtained through myoelectric signal intensity peak value and rising slope.
  4. 4. The method for adjusting adaptive training parameters of a hand rehabilitation exoskeleton of claim 1, wherein in step S2, the mathematical expression of the body ability synthesis parameter quantization formula is: wherein S is a body capacity comprehensive parameter, k Z is a joint impedance correction factor, W Z 、W F 、W Pp 、W Ff is each index weight coefficient, Z is a real-time joint impedance value, Z max 、Z min is a maximum and minimum value of the joint impedance S which is individually calibrated for the patient, F is the normalized physiological fatigue, P p is a normalized pre-pain characteristic value, F f is a real-time limb force feedback value, and F fmax is a maximum value of the limb force feedback which is individually calibrated for the patient.
  5. 5. The method for adjusting adaptive training parameters of a hand rehabilitation exoskeleton according to claim 1, wherein in step S3, the rehabilitation intention-conflict decoding logic layer is divided into a signal fusion layer and a conflict determination layer, the signal fusion layer performs data alignment and feature weighted fusion on comprehensive parameters of rehabilitation exercise intention strength and physical ability, fusion weights are 0.4-0.5, the preset threshold and risk classification standard is specifically that a preset intention-ability conflict matching degree threshold is 0.8, an intention-ability conflict matching degree is M, wherein When no intent-capability transient conflict is determined, The risk classification standard is combined with the rehabilitation exercise intention intensity I, the physical ability comprehensive parameter S and the risk correction factor k R to define And is sometimes a high intent-low ability-high risk conflict, When a medium intent-low capability-medium risk conflict, And is an extremely high risk conflict.
  6. 6. The method for adjusting adaptive training parameters of a hand rehabilitation exoskeleton of claim 5, wherein in step S3, the mathematical expression of the rehabilitation intent-collision matching calculation formula is: Wherein M is rehabilitation intention-conflict matching degree, k R is a risk correction factor, the risk correction factor is calibrated by a pre-pain characteristic value P p , T a is an action type adaptation coefficient and calibrated by a specific action type of rehabilitation exercise intention, I is normalized rehabilitation exercise intention strength, S is a body capacity comprehensive parameter, and alpha is an intention-capacity deviation correction coefficient.
  7. 7. The self-adaptive training parameter adjusting method for the hand rehabilitation exoskeleton of claim 1, wherein in the step S4, the multi-dimensional coupling self-adaptive parameter adjusting model is of a three-layer neural network structure and comprises an input layer, a hidden layer and an output layer, wherein the input layer receives six parameters including physiological fatigue degree, pain quantized value, exercise rhythm synchronous deviation value of a patient, rehabilitation exercise intention strength, body capacity comprehensive parameters and matching degree, 2-3 neuron layers are arranged on the hidden layer, each neuron layer comprises 10-15 neurons, a ReLU activation function is adopted, the output layer outputs exoskeleton training parameter adjusting value, in a differential flexibility solution strategy, in the initial stage of an action track initiated by the intention of a patient, a guiding auxiliary force with the amplitude of 5-15 DEG and the strength of 3-8N is applied in a fusion mode for 0.8S-1.2S, the joint movement degree is guided safely, then the joint movement degree is lifted according to a step length of 5% -10%/100%, the step length is lifted according to a step length of 3% -5%/100%, the power assisting movement speed is lifted according to a step length of the stroke intention, in the step length of the graph is lifted according to a step length of 3% -5%/100%, and the step length of the step length is lifted according to the step length of the normal power assisting time is reduced by 20% -20% when the step length is not required, and the speed is adjusted according to the normal step length of the step length is set, and the step is reduced by the step speed of the step-20%.
  8. 8. The method for adjusting the self-adaptive training parameters of the hand rehabilitation exoskeleton according to claim 1, wherein in the step S4, the adjusting coefficient with the value range of [0.1,1.0] is calculated through a multi-dimensional coupling training parameter adjusting coefficient formula, specific values of the exoskeleton boosting size, the movement speed, the training amplitude, the intermittent time, the boosting torque and the stress triggering time sequence are correspondingly determined according to the adjusting coefficient, a differential flexibility resolving strategy is executed according to the conflict judging result of the step S3, if the conflict judging result is high intention-low capability-high risk conflict, the exoskeleton is controlled to execute joint preparatory stretching motion with the amplitude of 5-15 degrees, the strength of 3N-8N and the duration of 0.8S-1.2S, then the exoskeleton parameters are gradually adjusted to complete the motion intention of a patient, if the conflict is medium intention-low capability-stroke risk conflict, the exoskeleton foundation and the movement speed are adjusted, then the motion intention of the patient is followed, and if the conflict is not generated, the auxiliary training motion is directly controlled according to the parameters corresponding to the adjusting coefficient.
  9. 9. The method for adjusting adaptive training parameters of a hand rehabilitation exoskeleton of claim 1, wherein in step S4, the mathematical expression of the multi-dimensional coupling training parameter adjustment coefficient formula is: The method comprises the steps of (1) obtaining a multi-dimensional coupling training parameter adjustment coefficient, wherein K is a multi-dimensional coupling training parameter adjustment coefficient, W M 、W S 、W F 、W P 、W D is a weight coefficient of each dimension, the sum of all weight coefficients is 1;M, S is a calculated rehabilitation intention-conflict matching degree, S is a calculated body capacity comprehensive parameter, F is a normalized physiological fatigue degree, P is a normalized real-time pain quantized value, the value range is [0,1], D is a healthy hand movement rhythm synchronization deviation value, and D max is a rhythm synchronization deviation maximum value of patient individual calibration.
  10. 10. The self-adaptive training parameter adjusting method for the hand rehabilitation exoskeleton according to claim 1, wherein in step S5, the specific process of the full-process real-time closed-loop adjustment is that the following operations are sequentially executed in each 500ms closed-loop period, wherein the multi-mode sensing system synchronously collects full-dimensional original signals of a patient, the noise reduction, normalization, feature extraction and quantization analysis are carried out on the collected signals for 100-200ms to generate standardized core feature parameters, the body capacity comprehensive parameters, the intention-conflict matching degree and the multi-dimensional coupling training parameter adjusting coefficients are respectively calculated for 200-350ms to complete conflict judgment, the exoskeleton driving module receives adjusting parameters and strategy instructions, synchronously executes parameter adjustment and flexible solution strategies, monitors the exoskeleton parameter adjusting state, the physiological state and the action state of the patient in real time for 450-500ms, records adjusting results and monitoring data, provides basis for the next closed-loop period, and steady state errors of closed-loop adjustment do not exceed 3%.

Description

Self-adaptive training parameter adjusting method for hand rehabilitation exoskeleton Technical Field The invention relates to the technical field of rehabilitation medical appliances, in particular to a self-adaptive training parameter adjusting method for hand rehabilitation exoskeleton. Background The hand dysfunction can seriously influence the daily life self-care ability and social engagement degree of a patient, heavy burden is brought to individuals, families and society, along with the deep fusion of rehabilitation medicine and engineering technology, hand rehabilitation exoskeleton is used as important auxiliary rehabilitation equipment, the hand rehabilitation exoskeleton is gradually a key means for improving the hand movement function, the equipment provides targeted movement assistance and training guidance for the patient through the combination of a mechanical structure and a sensing technology, the gradual recovery of the damaged neuromuscular function is helped, currently, the demands of rehabilitation medicine on the training refinement and individuation are increasingly improved, the clinical demands are difficult to meet by simply relying on a fixation program or an exoskeleton system which is manually adjusted, the technical direction of sensing the state of the patient in a multi-dimensional manner and adapting training parameters in real time is gradually focused in the industry, the dynamic optimization of rehabilitation training is realized by integrating physiological signals, movement intention, body capacity and other multi-aspect information, the training process is more attached to the individual condition of the patient, and the rehabilitation effect and safety are improved. The traditional hand rehabilitation exoskeleton related technology has various limitations, is difficult to adapt to complex requirements of rehabilitation training, at a signal sensing level, most traditional systems only acquire motion or physiological signals with single dimension, cannot comprehensively capture key information such as physiological fatigue, pain state, motion rhythm, rehabilitation intention and the like of a patient, so that deviation exists in judgment of the real state of the patient, in terms of parameter adjustment, the traditional technology adopts a preset fixed parameter or simple linear adjustment mode, dynamic matching relation between the rehabilitation intention of the patient and the body capacity is not fully considered, flexible adjustment of training intensity, speed and assistance mode according to the real-time state is difficult, part of systems lack effective conflict recognition and resolution mechanisms, discomfort and even safety risks are easily caused when the rehabilitation intention of the patient is not matched with the body bearing capacity, meanwhile, the traditional technology generally lacks a closed-loop adjustment mechanism, real-time feedback and optimization cannot be performed on the training effect and the change of the state of the patient, the training process is stiff, in addition, the training scene of most traditional systems is separated from the actual daily life, and the training result is difficult to be effectively converted into practical hand functions of the patient, and the rehabilitation efficiency and the patient compliance are required to be improved. Disclosure of Invention The invention aims to make up the defects of the prior art, and provides a self-adaptive training parameter adjusting method for hand rehabilitation exoskeleton, which is characterized in that full-dimensional signals of physiology, movement, intention and the like of a patient are synchronously acquired through a multi-mode sensing system, core parameters are analyzed through preprocessing and feature quantification, the matching state and conflict type of intention and capability are judged by means of rehabilitation intention-conflict decoding logic, an adjusting coefficient is calculated through a multi-dimensional coupling self-adaptive parameter adjusting model, the training parameters are optimized through matching with a differential flexible solution strategy, full-flow closed-loop adjustment is realized in a fixed period, the state change of the patient is dynamically adapted, daily scenery and advanced training are combined, the exoskeleton training is accurately attached to individual demands, rehabilitation safety and adaptability are improved, and the hand dysfunction patient is efficiently assisted to recover the movement function. The invention provides a self-adaptive training parameter adjusting method for hand rehabilitation exoskeleton, which aims to solve the technical problems and comprises the following specific steps: The method comprises the steps of S1, synchronously collecting multi-mode sensing signals, namely deploying a multi-mode sensing system, configuring a unified signal collecting protocol, synchronously collecting all-dimens