CN113705649A - Hand tremor detection method and system based on EMD-SVD feature extraction - Google Patents
Hand tremor detection method and system based on EMD-SVD feature extraction Download PDFInfo
- Publication number
- CN113705649A CN113705649A CN202110959839.8A CN202110959839A CN113705649A CN 113705649 A CN113705649 A CN 113705649A CN 202110959839 A CN202110959839 A CN 202110959839A CN 113705649 A CN113705649 A CN 113705649A
- Authority
- CN
- China
- Prior art keywords
- data
- fingertip
- acceleration
- tremor
- time sequence
- 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
Links
- 206010044565 Tremor Diseases 0.000 title claims abstract description 108
- 238000001514 detection method Methods 0.000 title claims abstract description 53
- 238000000605 extraction Methods 0.000 title claims abstract description 29
- 230000006870 function Effects 0.000 claims abstract description 75
- 238000000034 method Methods 0.000 claims abstract description 48
- 239000011159 matrix material Substances 0.000 claims abstract description 31
- 238000012545 processing Methods 0.000 claims abstract description 17
- 230000001133 acceleration Effects 0.000 claims description 117
- 238000013145 classification model Methods 0.000 claims description 52
- 238000012549 training Methods 0.000 claims description 35
- 238000000354 decomposition reaction Methods 0.000 claims description 20
- 238000012360 testing method Methods 0.000 claims description 20
- 230000000694 effects Effects 0.000 claims description 15
- 230000004927 fusion Effects 0.000 claims description 11
- 238000007781 pre-processing Methods 0.000 claims description 9
- 230000008569 process Effects 0.000 claims description 6
- 238000010801 machine learning Methods 0.000 abstract description 5
- 239000013598 vector Substances 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 238000004458 analytical method Methods 0.000 description 4
- 230000009467 reduction Effects 0.000 description 4
- 238000011156 evaluation Methods 0.000 description 3
- 238000012216 screening Methods 0.000 description 3
- KGSSUTVUTPLSQW-UHFFFAOYSA-N Robustone Chemical compound C1=C2OCOC2=CC(C2=COC=3C=C4OC(C=CC4=C(O)C=3C2=O)(C)C)=C1 KGSSUTVUTPLSQW-UHFFFAOYSA-N 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 201000010099 disease Diseases 0.000 description 2
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 2
- 201000006517 essential tremor Diseases 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 210000000707 wrist Anatomy 0.000 description 2
- 206010073210 Dystonic tremor Diseases 0.000 description 1
- 208000002972 Hepatolenticular Degeneration Diseases 0.000 description 1
- 208000012902 Nervous system disease Diseases 0.000 description 1
- 208000018737 Parkinson disease Diseases 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000002490 cerebral effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000010224 classification analysis Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000002790 cross-validation Methods 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 238000002567 electromyography Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 230000002981 neuropathic effect Effects 0.000 description 1
- 230000010355 oscillation Effects 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000000611 regression analysis Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000000638 solvent extraction Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/2431—Multiple classes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06F18/2132—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on discrimination criteria, e.g. discriminant analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/251—Fusion techniques of input or preprocessed data
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
A hand tremor detection method and system based on EMD-SVD feature extraction relates to the technical field of machine learning and is used for solving the problem that an existing hand tremor detection system cannot rapidly and accurately classify collected hand data effectively. The technical points of the invention comprise: the method comprises the steps of collecting a time sequence of hand data by using a hand tremor detection device, carrying out signal processing based on an EMD method to obtain an intrinsic mode function in the time sequence, extracting singular values in an intrinsic mode function matrix by using an SVD method to obtain effective characteristics of the hand tremor data, and finally constructing a multi-classification strategy through an SVM classifier to classify the characteristics so as to achieve the purpose of hand tremor detection. The invention can be applied to clinical medical treatment to judge whether the hand of a patient has tremor.
Description
Technical Field
The invention relates to the technical field of machine learning, in particular to a hand tremor detection method and system based on EMD-SVD feature extraction.
Background
Hand tremor can be seen in various nervous system diseases, such as essential tremor, Parkinson's disease, hepatolenticular degeneration, dystonic tremor, peripheral neuropathic tremor, cerebellar tremor, and others. The frequency and amplitude of tremor varies in most cases from disease to disease, e.g., essential tremor is usually 8-10Hz, and Parkinson's tremor is usually 4-6 Hz. Electromyography can be applied to objectively evaluate the tremor frequency and amplitude in recent years, but the examination has certain invasiveness, and the development of a non-invasive tremor detection means is urgently needed.
The existing tremor detection methods are mainly completed based on special detection devices, for example, an acceleration sensor is adopted to obtain tremor information of a tested person in the movement process, or surface electromechanical signals are collected for tremor analysis, and the detection methods realize quantitative measurement of tremor signals, but have the limitations: different devices or modules need to be designed for detecting the tremor of fingers, wrists and arms by using a detection device based on equipment such as an acceleration sensor, and a patient needs to wear the detection device to carry out daily activities, so that the use obstacle is large; the special instrument for detecting the physiological signal of the tremor patient is expensive generally, and the psychological burden of the patient is large in the detection process.
The development of machine learning techniques provides new approaches and methods for the detection of hand tremor. Empirical Mode Decomposition (EMD), which is essentially a smoothing method for nonlinear and non-stationary time series or signals, and is also a method capable of successively screening, processing or mining data; singular Value Decomposition (SVD) is a feature extraction method widely used in the fields of data mining and machine learning; support Vector Machines (SVMs) are supervised learning models and their associated learning algorithms that analyze data using classification and regression analysis. Therefore, how to combine the above methods and effectively utilize them, so that the hand tremor detection is more objective, rapid and accurate, is a problem to be solved.
Disclosure of Invention
In view of the above problems, the invention provides a hand tremor detection method and system based on EMD-SVD feature extraction, which is used for solving the problem that the existing hand tremor detection system cannot rapidly and accurately classify the acquired hand data effectively.
According to one aspect of the invention, a hand tremor detection method based on EMD-SVD feature extraction is provided, and the method comprises the following steps:
step one, collecting fingertip acceleration data and preprocessing the fingertip acceleration data to obtain an acceleration time sequence; the fingertip acceleration data comprises three types of acceleration data corresponding to no tremor, mild tremor and severe tremor of hands;
decomposing the acceleration time sequence based on an EMD method to obtain a characteristic matrix consisting of a plurality of intrinsic mode functions;
thirdly, reducing the dimension of the feature matrix based on an SVD method, and extracting a singular value of each eigenmode function as fingertip feature data;
dividing the fingertip characteristic data into training set data and test set data, training a multi-classification model based on an SVM algorithm according to the training set data, and obtaining a trained multi-classification model;
step five, inputting the fingertip characteristic data to be classified, which is obtained after preprocessing the fingertip acceleration data to be classified, processing the step two and processing the step three, into a trained multi-classification model to obtain a classification result; the classification result comprises three results of non-tremor, mild tremor and severe tremor of hands.
And further, in the fourth step, the test set data is input into the multi-classification model based on the SVM algorithm obtained by training for classification, the test set is used for detecting the classification effect of the multi-classification model, the influence of different amounts of fingertip characteristic data on the multi-classification model is compared, and the multi-classification model with the best classification effect and the fingertip characteristic data are stored and serve as the finally trained multi-classification model.
Further, the specific steps of the first step include: collecting fingertip acceleration data by using a sensor, wherein the three-axis acceleration is a when the sensor is staticx0,ay0,az0The three-axis acceleration in the acceleration data of each fingertip is ax,ay,azFusing the three-axis acceleration in each fingertip acceleration data to obtain a fused acceleration time sequence which changes along with time; wherein, the formula of fusion is:
in the formula, a represents fingertip acceleration data after fusion.
Further, the specific steps of the second step include: firstly, finding all maximum value points and minimum value points of an acceleration time sequence; then, respectively fitting a maximum value and a minimum value by utilizing cubic spline interpolation to obtain an upper envelope line and a lower envelope line of the acceleration time sequence; calculating the average value of the upper envelope and the lower envelope; then, the acceleration time sequence is subtracted from the average value to obtain a difference value; checking whether the difference value meets two basic conditions of the intrinsic mode functions, wherein if the difference value meets the two basic conditions of the intrinsic mode functions, the difference value is the first intrinsic mode function; if not, taking the difference value as a new acceleration time sequence, and repeating the steps until a first intrinsic mode function meeting the basic conditions is screened out; after obtaining a first intrinsic mode function, subtracting the original acceleration time sequence from the first intrinsic mode function to obtain a residual component; and taking the residual component as a new acceleration time sequence, and repeating the steps to obtain a plurality of residual intrinsic mode functions and monotonous residual components, wherein the intrinsic mode functions form a characteristic matrix.
Further, in the fourth step, three binomial classifiers are established through a one-vs-rest strategy in an SVM algorithm, and each binomial classifier classifies one of the classes and the remaining two classes; when the multi-classification model classifies the fingertip characteristic data to be classified in the sixth step, the three binomial classifiers classify the fingertip characteristic data to be classified respectively, the classification result of each binomial classifier is the prediction probability, and the class with the highest prediction probability in the three binomial classifiers is the final classification result.
According to another aspect of the present invention, a hand tremor detection system based on EMD-SVD feature extraction is provided, the system comprising:
the fingertip acceleration data acquisition module is used for acquiring fingertip acceleration data and preprocessing the fingertip acceleration data to obtain an acceleration time sequence; the fingertip acceleration data comprises three types of acceleration data corresponding to no tremor, mild tremor and severe tremor of hands;
the signal decomposition module is used for decomposing the acceleration time sequence based on an EMD method to obtain a characteristic matrix consisting of a plurality of intrinsic mode functions;
the feature extraction module is used for reducing the dimension of the feature matrix based on an SVD method and extracting a singular value of each eigenmode function as fingertip feature data;
the model training module is used for dividing the fingertip characteristic data into training set data and test set data, training a multi-classification model based on an SVM algorithm according to the training set data, inputting the test set data into the multi-classification model based on the SVM algorithm obtained through training for classification in the training process, detecting the classification effect of the multi-classification model by using the test set, comparing the influence of different amounts of fingertip characteristic data on the multi-classification model, and storing the multi-classification model with the best classification effect and the fingertip characteristic data to serve as the finally trained multi-classification model;
the classification detection module is used for preprocessing the fingertip acceleration data to be classified, inputting the fingertip characteristic data to be classified, which is acquired after the fingertip acceleration data is processed by the signal decomposition module and the characteristic extraction module, into the trained multi-classification model, and acquiring a classification result; the classification result comprises three results of non-tremor, mild tremor and severe tremor of hands.
Further, the fingertip data acquisition module comprises a data acquisition unit and a data processing unit, wherein the data acquisition unit is an MPU6050 sensor, and the data processing unit is an STM32F103 microcontroller; the three-axis acceleration of the sensor is a when the sensor is staticx0,ay0,az0The three-axis acceleration in the acceleration data of each fingertip is ax,ay,azFusing the three-axis acceleration in each fingertip acceleration data to obtain a fused acceleration time sequence which changes along with time; wherein, the formula of fusion is:
in the formula, a represents fingertip acceleration data after fusion.
Further, the specific step of decomposing the acceleration time sequence based on the EMD method in the signal decomposition module to obtain a feature matrix composed of a plurality of eigenmode functions includes: firstly, finding all maximum value points and minimum value points of an acceleration time sequence; then, respectively fitting a maximum value and a minimum value by utilizing cubic spline interpolation to obtain an upper envelope line and a lower envelope line of the acceleration time sequence; calculating the average value of the upper envelope and the lower envelope; then, the acceleration time sequence is subtracted from the average value to obtain a difference value; checking whether the difference value meets two basic conditions of the intrinsic mode functions, wherein if the difference value meets the two basic conditions of the intrinsic mode functions, the difference value is the first intrinsic mode function; if not, taking the difference value as a new acceleration time sequence, and repeating the steps until a first intrinsic mode function meeting the basic conditions is screened out; after obtaining a first intrinsic mode function, subtracting the original acceleration time sequence from the first intrinsic mode function to obtain a residual component; and taking the residual component as a new acceleration time sequence, and repeating the steps to obtain a plurality of residual intrinsic mode functions and monotonous residual components, wherein the intrinsic mode functions form a characteristic matrix.
Further, the method for constructing the multi-classification model in the model training module comprises the following steps: and establishing three binomial classifiers through a one-vs-rest strategy in the SVM algorithm, wherein each binomial classifier classifies one class and the remaining two classes.
Further, when the multi-classification model in the classification detection module classifies the fingertip characteristic data to be classified, three binomial classifiers are used for classifying the fingertip characteristic data to be classified respectively, the classification result of each binomial classifier is the prediction probability, and the class with the highest prediction probability in the three binomial classifiers is the final classification result.
The beneficial technical effects of the invention are as follows:
the basis function of the EMD is obtained by decomposing time sequence data, and is based on the time scale local characteristic of the time sequence, so that the EMD method is posterior and visual, and has self-adaptability; for unknown signals, the decomposition can be directly started without the need of preliminary analysis and research; the method can automatically perform layering according to certain fixed modes without manual setting and intervention; SVD can simplify data sets and data characteristics, and can eliminate data noise, thereby improving the algorithm prediction result. The feature extraction of the tremor data by the EMD-SVD method can extract more abundant features in the original signal than other methods.
The hand tremor evaluation method integrates hand tremor data acquisition and analysis, analyzes hand tremor data by using a machine learning related method, gives a prediction result, eliminates subjective factors, and enables the hand tremor evaluation method to be more objective and accurate; the hand tremor detection device is convenient and simple to use, easy to operate, capable of achieving rapid detection, suitable for popularization among a large number of hand tremor patients, and capable of providing powerful auxiliary means for hand tremor detection.
Drawings
The present invention may be better understood by reference to the following description taken in conjunction with the accompanying drawings, which are incorporated in and form a part of this specification, and which are used to further illustrate preferred embodiments of the present invention and to explain the principles and advantages of the present invention.
FIG. 1 is an overall flow chart of the hand tremor detection method based on EMD-SVD feature extraction.
FIG. 2 is a diagram of an apparatus for acquiring fingertip tremor data in an embodiment of the present invention.
Fig. 3 is an EMD decomposition flow chart according to an embodiment of the present invention.
FIG. 4 is a diagram illustrating decomposition results of an SVD algorithm according to an embodiment of the present invention; wherein, the graphs (a), (b) and (c) represent non-tremor, mild tremor and severe tremor data, respectively.
FIG. 5 is a graph of EMD decomposition results of patient fingertip tremor information in an embodiment of the present invention.
FIG. 6 is a diagram of the EMD singular value changes of three types of data in the embodiment of the present invention.
Fig. 7 is a flow chart of a one-vs-rest policy in the embodiment of the present invention.
FIG. 8 is a classification result with increasing feature numbers in an embodiment of the present invention.
FIG. 9 is a result of a single feature classification in an embodiment of the invention.
FIG. 10 is a schematic structural diagram of a hand tremor detection system based on EMD-SVD feature extraction.
Detailed Description
In order that those skilled in the art will better understand the disclosure, exemplary embodiments or examples of the disclosure are described below with reference to the accompanying drawings. It is obvious that the described embodiments or examples are only some, but not all embodiments or examples of the invention. All other embodiments or examples obtained by a person of ordinary skill in the art based on the embodiments or examples of the present invention without any creative effort shall fall within the protection scope of the present invention.
The invention provides a hand tremor detection method and system based on EMD-SVD feature extraction and SVM classification, wherein a hand tremor detection device is used for collecting a time sequence of hand tremor data, signal processing is carried out based on an EMD method to obtain an intrinsic mode function in a complex sequence, a SVD method is used for extracting singular values in an intrinsic mode function matrix to obtain effective features of the hand tremor data, and finally a multi-classification strategy is constructed through an SVM classifier to classify the features, so that the purpose of hand tremor detection is achieved.
The flow chart of the method of the invention is shown in figure 1, and mainly comprises the following steps:
the method comprises the following steps: data were collected by a hand tremor detection device.
According to the embodiment of the invention, the hand tremor detection device mainly comprises a data acquisition unit and a data processing unit. Wherein, MPU6050 is selected as the sensor to the data acquisition unit, and STM32F103 is selected as microcontroller to the data processing unit, and the serial ports standard is RS-232.
In this embodiment, tremor data at the tip of the finger of the subject is collected using the apparatus shown in FIG. 2. The tester wears the finger sleeve on the index finger of the left hand or the right hand, and the hand tremor detection device main body part is tied on the corresponding wrist, and the acquisition of the fingertip tremor data is carried out according to different motion states. Collected fingertip tremor data were manually classified into three types of data, no tremor, mild tremor, and severe tremor, for use as a training set and a test set, and the corresponding tester names were saved.
The data processing unit adopts the following processing method for the collected fingertip acceleration data: recording the three-axis acceleration of the sensor at rest as ax0,ay0,az0Fusing the three-axis acceleration in the data, and setting the three-axis addition of each fingertip data in the training data setSpeed is respectively ax,ay,azThen the fusion formula is:
wherein a is the fused acceleration; the fused acceleration sequence over time is denoted x (t).
Step two: hand tremor signal processing based on EMD.
The EMD (empirical mode decomposition) method can adaptively decompose hand tremor signals into intrinsic mode function forms which are different from each other but have narrow-band characteristics. The EMD method needs to obtain all local maximum points and local minimum points in an original time sequence, firstly, a cubic spline interpolation function is used for fitting all the local maximum points to form an upper envelope line of the original time sequence X (t), then, a cubic spline interpolation function is used for fitting all the local minimum points to form a lower envelope line of the original time sequence X (t), an instant balance position is determined through an average value m (t) of each corresponding point in the upper envelope line and the lower envelope line, then, the original time sequence and the envelope average value are subjected to subtraction, namely h (t) ═ X (t) — m (t), a gentle trend component is continuously removed, and the eigenmode function of a signal is further extracted. The continuous sequence screening process is realized by an EMD method, the components under different frequencies are effectively extracted in sequence from high frequency to low frequency according to the sequence, and the flow chart of EMD on fingertip data decomposition is shown in figure 3. The specific decomposition steps can be described as:
(1) firstly, finding all maximum value points and minimum value points of a time sequence X (t);
(2) respectively fitting a maximum value and a minimum value by using cubic spline interpolation, solving by using a natural boundary condition by using the cubic spline interpolation to obtain an upper envelope X of the time sequencemax(t) and the lower envelope Xmin(t)：
Si(x)＝ai+bix+cix2+dix3
Wherein S isi(x) Is each segmented inter-cell [ x ]i,xi+1]Fitting equation of (a)i,bi,ci,diIs the coefficient to be solved;
(3) calculate the average m (t) of the upper and lower envelopes:
(4) the difference is made between the original time series X (t) and the average m (t), resulting in h (t):
h(t)＝X(t)-m(t)
(5) checking whether the difference h (t) satisfies two basic conditions of the intrinsic mode function, if h (t) satisfies the conditions, c1(t) h (t), i.e. the difference h (t) is the first selected eigenmode function; if h (t) does not satisfy two basic conditions, repeating the steps (1) - (4) by taking h (t) as a new time sequence until an intrinsic mode function c meeting the basic conditions is screened out1(t)；
(6) Obtaining a first eigenmode function c1(t) thereafter, the original time series X (t) is brought into contact with a first eigenmode function c1(t) differencing to obtain a residual component r1(t)：
r1(t)＝X(t)-c1(t)
(7) Residual component r1(t) repeating steps (1) - (6) as new original signal to obtain residual intrinsic mode functions c2(t),c3(t),...,cn(t) and a monotonic residual component rn(t), the original time series x (t) can then be represented as:
since excessive screening is prone to errors, deviating from the original signal, so that the termination condition cannot trigger the trapping of the dead loop, a termination criterion SD is additionally introduced:
the acceleration time sequence x (t) is decomposed by the EMD method, SD is set to 0.28 in the termination condition, and after testing, all intrinsic mode functions of the data can be screened out on the premise of maintaining the linear characteristic and stability of the intrinsic mode functions, and the program does not enter into a dead loop, and the result of EMD decomposition is shown in fig. 4.
The three types of fingertip tremor data are decomposed by an EMD method to obtain six intrinsic mode functions and a residual error. C from the first eigenmode to the sixth eigenmode1(t),c2(t),...,c6(t), the original signal can be approximated as c1(t),c2(t),...,c6And (t) superposition, wherein the oscillation frequency of the decomposed eigenmode function is gradually reduced, the time scale is gradually increased, the information of the original signal contained in each eigenmode is also gradually reduced, the related characteristic values are further analyzed and extracted, and the characteristic matrix is represented as follows:
[c1(t),c2(t),...,c6(t)]T
step three: an SVD-based feature extraction method.
SVD (singular value decomposition) can remove some features with small occupation ratio, reduce redundancy of data, improve generalization capability of a prediction model to a certain extent, reduce calculated amount during model training, and improve training and prediction speed of the model.
For matrix A, A ∈ Rn×nλ is a characteristic value of A, w1,w2…wnFor a linearly independent feature vector, a can be decomposed as:
A＝WΣW-1
wherein W ═ W1,w2,…wn) Is a feature vector matrix; Σ ∈ Rn×nAnd is a singular value matrix.
For non-square matrix A, A ∈ Rm×nThen a can be decomposed into:
A＝UΣVT
wherein U is E.Rm×mIs AATA feature vector matrix of (a); v is an element of Rn×nIs ATA, a feature vector matrix; Σ ∈ Rm×nAnd is a singular value matrix.
In the singular value matrix, elements on a diagonal line are singular values of A, right singular value vectors corresponding to the first k singular values compress data from columns, only the first k characteristics are selected from a plurality of characteristics representing the data on the data, and k bases with the largest singular values are extracted as new coordinates, namely the characteristics are screened.
Simplifying data set and data characteristics by SVD method, eliminating data noise, and fitting characteristic matrix [ c ]1(t),c2(t),...,c6(t)]TSVD dimension reduction is carried out, singular values of each mode are extracted, left and right singular vectors and singular values after dimension reduction of fingertip data are shown as shaded parts in figure 5, SVD dimension reduction is carried out on the characteristic matrix, the singular values of each mode are extracted, and the change situation of the singular values of the intrinsic mode functions of the three types of data is shown in figure 6.
In the embodiment, six fingertip tremor features are extracted by using the EMD-SVD method, the first five of the six features show high correlation, all of which are above 0.7, the last one is not high, and is only 0.3, and no significant difference is shown, so that the feature selection is not considered.
Step four: and constructing a multi-classification model based on the SVM.
The SVM adopts the kernel function to map to a high-dimensional space for nonlinear samples, so that the linear divisibility is realized, and the popularization capability is good. And constructing a multi-classification model by using a multi-classification model construction strategy of the SVM to realize the detection of hand tremor. The SVM is a binary model, given a D { (x) for a binary problem1,y1),(x2,y2),…,(xm,ym) In which y isiAnd E { -1, +1} is a training set of the samples, and a hyperplane is expected to be found in a sample space, so that positive and negative samples in the training set can be divided, and the aim of classifying data is fulfilled. For better partitioning of the data set, a most robust one should be preferentially selected in the sample spaceThe good hyperplane avoids the classification error of the classification model caused by the local disturbance of the sample or the noise, and simultaneously improves the generalization capability of the model.
In the embodiment, the detection of three levels of fingertip tremor is realized, namely, a three-classification problem is realized through a one-vs-rest strategy in the SVM. The strategy is as follows: establishing 3 binomial classifiers, wherein each binomial classifier classifies one class and the rest two classes, namely a target class and other classes; when prediction is performed, the 3 second classifiers classify the classes to be detected respectively, the classification result of the classifier is the prediction probability, and the class with the highest probability in the three classifiers is the final result of the three classifications. A one-vs-rest policy flow diagram is shown in fig. 7.
Step five: and (5) training and testing the model.
Training the multi-classification model by using different quantities of feature data in a training set, checking the classification effect of the classification model by using test set data, comparing the influence of different quantities of features on the classification model, and storing the model with the best classification effect to obtain the fingertip tremor detection system.
In the embodiment, fingertip tremor is detected by using an EMD-SVD feature extraction and SVM classification method, a 5-fold cross-validation method is adopted for analysis, and finally the average value of five results is used as an evaluation index. Through experiments, the classification results of the continuously increased feature quantity are shown in fig. 8. As can be seen from FIG. 8, there is c1When the features are added or subtracted, the number of the features is not a main factor influencing the classification result, and thus the c pair1～c5Each feature is analyzed, the classification result is shown in fig. 9, the classification effect of the first feature classifier is better, and the performance of classifiers trained by other features is gradually reduced.
Another embodiment of the present invention provides a hand tremor detection system based on EMD-SVD feature extraction, as shown in fig. 10, the system includes:
the fingertip data acquisition module 110 is used for acquiring fingertip acceleration data and preprocessing the fingertip acceleration data to obtain an acceleration time sequence; the fingertip acceleration data comprises three types of acceleration data corresponding to no tremor, mild tremor and severe tremor of hands; the signal decomposition module 120 is configured to decompose the acceleration time sequence based on an EMD method to obtain a feature matrix composed of a plurality of eigenmode functions; the feature extraction module 130 is configured to perform dimension reduction on the feature matrix based on an SVD method, and extract a singular value of each eigenmode function as fingertip feature data; the model training module 140 is configured to divide the fingertip feature data into training set data and test set data, train a multi-classification model based on an SVM algorithm according to the training set data, input the test set data into the trained multi-classification model based on the SVM algorithm for classification in the training process, examine the classification effect of the multi-classification model by using the test set, compare the influence of different amounts of fingertip feature data on the multi-classification model, and store the multi-classification model with the best classification effect and the fingertip feature data as a finally trained multi-classification model; the classification detection module 150 is configured to input the fingertip characteristic data to be classified, which is obtained after the fingertip acceleration data to be classified is preprocessed and is processed by the signal decomposition module 120 and the characteristic extraction module 130, into the trained multi-classification model, so as to obtain a classification result; the classification results include three results of non-tremor, mild tremor and severe tremor in the hand.
The fingertip data acquisition module 110 includes a data acquisition unit 1110 and a data processing unit 1120, wherein the data acquisition unit 1110 is an MPU6050 sensor, and the data processing unit 1120 is an STM32F103 microcontroller; the three-axis acceleration of the sensor is a when the sensor is staticx0,ay0,az0The three-axis acceleration in the acceleration data of each fingertip is ax,ay,azFusing the three-axis acceleration in each fingertip acceleration data to obtain a fused acceleration time sequence which changes along with time; wherein, the formula of fusion is:
in the formula, a represents fingertip acceleration data after fusion.
The specific steps of decomposing the acceleration time sequence based on the EMD method in the signal decomposition module 120 to obtain the feature matrix composed of a plurality of eigenmode functions include: firstly, finding all maximum value points and minimum value points of an acceleration time sequence; then, respectively fitting a maximum value and a minimum value by utilizing cubic spline interpolation to obtain an upper envelope line and a lower envelope line of the acceleration time sequence; calculating the average value of the upper envelope and the lower envelope; then, the acceleration time sequence is subtracted from the average value to obtain a difference value; checking whether the difference value meets two basic conditions of the intrinsic mode functions, wherein if the difference value meets the two basic conditions of the intrinsic mode functions, the difference value is the first intrinsic mode function; if not, taking the difference value as a new acceleration time sequence, and repeating the steps until a first intrinsic mode function meeting the basic conditions is screened out; after obtaining a first intrinsic mode function, subtracting the original acceleration time sequence from the first intrinsic mode function to obtain a residual component; and taking the residual component as a new acceleration time sequence, and repeating the steps to obtain a plurality of residual intrinsic mode functions and monotonous residual components, wherein the intrinsic mode functions form a characteristic matrix.
The multi-classification model in the model training module 140 is formed by: and establishing three binomial classifiers through a one-vs-rest strategy in the SVM algorithm, wherein each binomial classifier classifies one class and the remaining two classes. When the multi-classification model in the classification detection module 150 classifies the fingertip characteristic data to be classified, three binomial classifiers are used for classifying the fingertip characteristic data to be classified respectively, the classification result of each binomial classifier is the prediction probability, and the class with the highest prediction probability in the three binomial classifiers is the final classification result.
The function of the hand tremor detection system based on the EMD-SVD feature extraction in this embodiment can be described by the hand tremor detection method based on the EMD-SVD feature extraction, so that the detailed description of this embodiment is omitted, and reference may be made to the above method embodiments, and further description is omitted here.
While the invention has been described with respect to a limited number of embodiments, those skilled in the art, having benefit of this description, will appreciate that other embodiments can be devised which do not depart from the scope of the invention as described herein. The present invention has been disclosed in an illustrative rather than a restrictive sense, and the scope of the present invention is defined by the appended claims.
Claims (10)
1. A hand tremor detection method based on EMD-SVD feature extraction is characterized by comprising the following steps:
step one, collecting fingertip acceleration data and preprocessing the fingertip acceleration data to obtain an acceleration time sequence; the fingertip acceleration data comprises three types of acceleration data corresponding to no tremor, mild tremor and severe tremor of hands;
decomposing the acceleration time sequence based on an EMD method to obtain a characteristic matrix consisting of a plurality of intrinsic mode functions;
thirdly, reducing the dimension of the feature matrix based on an SVD method, and extracting a singular value of each eigenmode function as fingertip feature data;
dividing the fingertip characteristic data into training set data and test set data, training a multi-classification model based on an SVM algorithm according to the training set data, and obtaining a trained multi-classification model;
step five, inputting the fingertip characteristic data to be classified, which is obtained after preprocessing the fingertip acceleration data to be classified, processing the step two and processing the step three, into a trained multi-classification model to obtain a classification result; the classification result comprises three results of non-tremor, mild tremor and severe tremor of hands.
2. The hand tremor detection method of claim 1, which is based on EMD-SVD feature extraction, and is characterized in that in step four, test set data is input into a trained multi-classification model based on SVM algorithm for classification, the test set is used to test the classification effect of the multi-classification model, the influence of different amounts of fingertip feature data on the multi-classification model is compared, and the multi-classification model with the best classification effect and the fingertip feature data are stored as the final trained multi-classification model.
3. The hand tremor detection method based on EMD-SVD feature extraction of claim 2, wherein the specific steps of step one include: collecting fingertip acceleration data by using a sensor, wherein the three-axis acceleration is a when the sensor is staticx0,ay0,az0The three-axis acceleration in the acceleration data of each fingertip is ax,ay,azFusing the three-axis acceleration in each fingertip acceleration data to obtain a fused acceleration time sequence which changes along with time; wherein, the formula of fusion is:
in the formula, a represents fingertip acceleration data after fusion.
4. The hand tremor detection method based on EMD-SVD feature extraction of claim 3, wherein the specific steps of step two include: firstly, finding all maximum value points and minimum value points of an acceleration time sequence; then, respectively fitting a maximum value and a minimum value by utilizing cubic spline interpolation to obtain an upper envelope line and a lower envelope line of the acceleration time sequence; calculating the average value of the upper envelope and the lower envelope; then, the acceleration time sequence is subtracted from the average value to obtain a difference value; checking whether the difference value meets two basic conditions of the intrinsic mode functions, wherein if the difference value meets the two basic conditions of the intrinsic mode functions, the difference value is the first intrinsic mode function; if not, taking the difference value as a new acceleration time sequence, and repeating the steps until a first intrinsic mode function meeting the basic conditions is screened out; after obtaining a first intrinsic mode function, subtracting the original acceleration time sequence from the first intrinsic mode function to obtain a residual component; and taking the residual component as a new acceleration time sequence, and repeating the steps to obtain a plurality of residual intrinsic mode functions and monotonous residual components, wherein the intrinsic mode functions form a characteristic matrix.
5. The hand tremor detection method based on EMD-SVD feature extraction of claim 4, wherein in step four, three binomial classifiers are established through one-vs-rest strategy in SVM algorithm, each binomial classifier classifies one of the categories and the remaining two categories; when the multi-classification model classifies the fingertip characteristic data to be classified in the sixth step, the three binomial classifiers classify the fingertip characteristic data to be classified respectively, the classification result of each binomial classifier is the prediction probability, and the class with the highest prediction probability in the three binomial classifiers is the final classification result.
6. A hand tremor detection system based on EMD-SVD feature extraction, comprising:
the fingertip acceleration data acquisition module is used for acquiring fingertip acceleration data and preprocessing the fingertip acceleration data to obtain an acceleration time sequence; the fingertip acceleration data comprises three types of acceleration data corresponding to no tremor, mild tremor and severe tremor of hands;
the signal decomposition module is used for decomposing the acceleration time sequence based on an EMD method to obtain a characteristic matrix consisting of a plurality of intrinsic mode functions;
the feature extraction module is used for reducing the dimension of the feature matrix based on an SVD method and extracting a singular value of each eigenmode function as fingertip feature data;
the model training module is used for dividing the fingertip characteristic data into training set data and test set data, training a multi-classification model based on an SVM algorithm according to the training set data, inputting the test set data into the multi-classification model based on the SVM algorithm obtained through training for classification in the training process, detecting the classification effect of the multi-classification model by using the test set, comparing the influence of different amounts of fingertip characteristic data on the multi-classification model, and storing the multi-classification model with the best classification effect and the fingertip characteristic data to serve as the finally trained multi-classification model;
the classification detection module is used for preprocessing the fingertip acceleration data to be classified, inputting the fingertip characteristic data to be classified, which is acquired after the fingertip acceleration data is processed by the signal decomposition module and the characteristic extraction module, into the trained multi-classification model, and acquiring a classification result; the classification result comprises three results of non-tremor, mild tremor and severe tremor of hands.
7. The hand tremor detection system of claim 6, wherein the fingertip data acquisition module comprises a data acquisition unit and a data processing unit, wherein the data acquisition unit is an MPU6050 sensor, and the data processing unit is an STM32F103 microcontroller; the three-axis acceleration of the sensor is a when the sensor is staticx0,ay0,az0The three-axis acceleration in the acceleration data of each fingertip is ax,ay,azFusing the three-axis acceleration in each fingertip acceleration data to obtain a fused acceleration time sequence which changes along with time; wherein, the formula of fusion is:
in the formula, a represents fingertip acceleration data after fusion.
8. The hand tremor detection system of claim 7, wherein the signal decomposition module decomposes the acceleration time series based on the EMD method, and the specific step of obtaining the feature matrix composed of a plurality of eigenmode functions includes: firstly, finding all maximum value points and minimum value points of an acceleration time sequence; then, respectively fitting a maximum value and a minimum value by utilizing cubic spline interpolation to obtain an upper envelope line and a lower envelope line of the acceleration time sequence; calculating the average value of the upper envelope and the lower envelope; then, the acceleration time sequence is subtracted from the average value to obtain a difference value; checking whether the difference value meets two basic conditions of the intrinsic mode functions, wherein if the difference value meets the two basic conditions of the intrinsic mode functions, the difference value is the first intrinsic mode function; if not, taking the difference value as a new acceleration time sequence, and repeating the steps until a first intrinsic mode function meeting the basic conditions is screened out; after obtaining a first intrinsic mode function, subtracting the original acceleration time sequence from the first intrinsic mode function to obtain a residual component; and taking the residual component as a new acceleration time sequence, and repeating the steps to obtain a plurality of residual intrinsic mode functions and monotonous residual components, wherein the intrinsic mode functions form a characteristic matrix.
9. The hand tremor detection system of claim 8, wherein the multi-classification model based on SVM algorithm in the model training module is formed by: and establishing three binomial classifiers through a one-vs-rest strategy in the SVM algorithm, wherein each binomial classifier classifies one class and the remaining two classes.
10. The hand tremor detection system of claim 9, wherein when the multi-classification model based on the SVM algorithm in the classification detection module classifies the fingertip feature data to be classified, three binomial classifiers are used to classify the fingertip feature data to be classified, the classification result of each binomial classifier is the prediction probability, and the class with the highest prediction probability among the three binomial classifiers is the final classification result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110959839.8A CN113705649B (en) | 2021-08-20 | 2021-08-20 | EMD-SVD feature extraction-based hand tremor detection method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110959839.8A CN113705649B (en) | 2021-08-20 | 2021-08-20 | EMD-SVD feature extraction-based hand tremor detection method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113705649A true CN113705649A (en) | 2021-11-26 |
| CN113705649B CN113705649B (en) | 2024-01-12 |
Family
ID=78654109
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110959839.8A Active CN113705649B (en) | 2021-08-20 | 2021-08-20 | EMD-SVD feature extraction-based hand tremor detection method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113705649B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104776908A (en) * | 2015-04-17 | 2015-07-15 | 南京理工大学 | EMD generalized energy-based wheeltrack vibration signal fault feature extraction method |
| WO2019036749A1 (en) * | 2017-08-24 | 2019-02-28 | Adams Warwick Russell | A system for detecting early parkinson's disease and other neurological diseases and movement disorders |
| CN112075940A (en) * | 2020-09-21 | 2020-12-15 | 哈尔滨工业大学 | Tremor detection system based on bidirectional long-time and short-time memory neural network |
| CN112580588A (en) * | 2020-12-29 | 2021-03-30 | 西北工业大学 | Intelligent flutter signal identification method based on empirical mode decomposition |
| CN113100756A (en) * | 2021-04-15 | 2021-07-13 | 重庆邮电大学 | Stacking-based Parkinson tremor detection method |
-
2021
- 2021-08-20 CN CN202110959839.8A patent/CN113705649B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104776908A (en) * | 2015-04-17 | 2015-07-15 | 南京理工大学 | EMD generalized energy-based wheeltrack vibration signal fault feature extraction method |
| WO2019036749A1 (en) * | 2017-08-24 | 2019-02-28 | Adams Warwick Russell | A system for detecting early parkinson's disease and other neurological diseases and movement disorders |
| CN112075940A (en) * | 2020-09-21 | 2020-12-15 | 哈尔滨工业大学 | Tremor detection system based on bidirectional long-time and short-time memory neural network |
| CN112580588A (en) * | 2020-12-29 | 2021-03-30 | 西北工业大学 | Intelligent flutter signal identification method based on empirical mode decomposition |
| CN113100756A (en) * | 2021-04-15 | 2021-07-13 | 重庆邮电大学 | Stacking-based Parkinson tremor detection method |
Non-Patent Citations (2)
| Title |
|---|
| 张亢;石阳春;唐明珠;吴家腾;: "局部波动特征分解及其在滚动轴承故障诊断中的应用研究", 振动与冲击, no. 01, pages 97 - 103 * |
| 马小刚;王永平;杜历;: "一种求解信号包络曲线端点值的新方法", 噪声与振动控制, no. 04, pages 20 - 23 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113705649B (en) | 2024-01-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111209885A (en) | Gesture information processing method and device, electronic equipment and storage medium | |
| EP3692903A1 (en) | Artificial intelligence self-learning-based automatic electrocardiography analysis method and apparatus | |
| CN107951485A (en) | Ambulatory ECG analysis method and apparatus based on artificial intelligence self study | |
| CN113729707A (en) | FECNN-LSTM-based emotion recognition method based on multi-mode fusion of eye movement and PPG | |
| CN107016233A (en) | The association analysis method and system of motor behavior and cognitive ability | |
| CN108511055A (en) | Ventricular premature beat identifying system and method based on Multiple Classifier Fusion and diagnostic rule | |
| CN113128585B (en) | Deep neural network based multi-size convolution kernel method for realizing electrocardiographic abnormality detection and classification | |
| CN108681685A (en) | A kind of body work intension recognizing method based on human body surface myoelectric signal | |
| CN112200162A (en) | Non-contact heart rate measuring method, system and device based on end-to-end network | |
| CN108962379B (en) | Mobile phone auxiliary detection system for cranial nerve system diseases | |
| CN110604578A (en) | Human hand and hand motion recognition method based on SEMG | |
| CN112869743B (en) | Exercise initiation intention neural analysis method considering cognitive distraction | |
| Okreghe et al. | A deep neural network-based spike sorting with improved channel selection and artefact removal | |
| CN109009148A (en) | A kind of gait function appraisal procedure | |
| CN113988135A (en) | Electromyographic signal gesture recognition method based on double-branch multi-stream network | |
| CN113095379A (en) | Human motion state identification method based on wearable six-axis sensing data | |
| CN116484290A (en) | Depression recognition model construction method based on Stacking integration | |
| CN113705649A (en) | Hand tremor detection method and system based on EMD-SVD feature extraction | |
| CN115736920A (en) | Depression state identification method and system based on bimodal fusion | |
| CN112914585B (en) | Single-lead electrocardiogram-oriented classification method, system and device | |
| Sanamdikar et al. | Classification of ECG Signal for Cardiac Arrhythmia Detection Using GAN Method | |
| Chen et al. | Learned Kernels for Interpretable and Efficient PPG Signal Quality Assessment and Artifact Segmentation | |
| Vijayvargiya et al. | Implementation of machine learning algorithms for automated human gait activity recognition using sEMG signals | |
| Xue et al. | SEMG based intention identification of complex hand motion using nonlinear time series analysis | |
| Seddiqi et al. | Recognition of turkish sign language (TID) using sEMG sensor |
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 |