CN112163570B - SVM (support vector machine) electrocardiosignal identification method based on improved Husky algorithm optimization - Google Patents

SVM (support vector machine) electrocardiosignal identification method based on improved Husky algorithm optimization Download PDF

Info

Publication number
CN112163570B
CN112163570B CN202011180394.5A CN202011180394A CN112163570B CN 112163570 B CN112163570 B CN 112163570B CN 202011180394 A CN202011180394 A CN 202011180394A CN 112163570 B CN112163570 B CN 112163570B
Authority
CN
China
Prior art keywords
wolf
gray
individual
fitness value
svm
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.)
Active
Application number
CN202011180394.5A
Other languages
Chinese (zh)
Other versions
CN112163570A (en
Inventor
刘继忠
李继发
徐文斌
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.)
Nanchang University
Original Assignee
Nanchang University
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 Nanchang University filed Critical Nanchang University
Priority to CN202011180394.5A priority Critical patent/CN112163570B/en
Publication of CN112163570A publication Critical patent/CN112163570A/en
Application granted granted Critical
Publication of CN112163570B publication Critical patent/CN112163570B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/02—Preprocessing
    • G06F2218/04—Denoising
    • 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
    • G06F18/253—Fusion techniques of extracted features
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/004—Artificial life, i.e. computing arrangements simulating life
    • G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/08—Feature extraction

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Artificial Intelligence (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Biophysics (AREA)
  • Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Signal Processing (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
  • Eye Examination Apparatus (AREA)

Abstract

The invention discloses an SVM electrocardiosignal recognition method based on improved wolf algorithm optimization, wherein an improved wolf algorithm (DIGWO) is used for optimizing parameters of a Support Vector Machine (SVM), so that the recognition rate of the SVM to an electrocardiosignal is improved. The method comprises the steps of extracting and fusing collected ECG signals into a characteristic space serving as an input vector through multi-scale sample entropy and wavelet energy ratio characteristics, introducing a dynamic double-subgroup strategy on the basis of a traditional wolf algorithm, improving from three aspects of subgroup division, convergence factors and proportion weight, then using the improved wolf algorithm for seeking the optimal combination of a penalty factor C and a kernel function parameter g of an SVM, and training and constructing a DIGWO-SVM classification model. The electrocardiosignal identification method provided by the invention improves the electrocardiosignal identification efficiency and the accuracy of the analysis result.

Description

SVM (support vector machine) electrocardiosignal identification method based on improved Husky algorithm optimization
Technical Field
The invention relates to the field of electrocardiosignal recognition, and particularly belongs to an SVM electrocardiosignal recognition method based on improved Huuler algorithm optimization.
Background
According to the World Health Organization (WHO), the number of deaths annually from non-infectious diseases in the world accounts for two thirds of the global deaths. Among them, the incidence of cardiovascular diseases is the first. About 1750 million people die worldwide each year from cardiovascular-related diseases. In recent years, the prevalence rate of cardiovascular diseases in China is continuously increased, the number of patients suffering from cardiovascular diseases is as high as 2.9 hundred million, the death rate of the cardiovascular diseases is far higher than that of other diseases, the number of deaths of the cardiovascular diseases accounts for more than two fifths of the total number of deaths of the diseases, and the cardiovascular diseases become diseases which pose the greatest threat to the health of people. Therefore, the method has very important significance and research value for the analysis and diagnosis of the cardiovascular diseases.
Electrocardiogram (ECG) is a condition that cardiac muscle cells generate potential response when the heart is stimulated by excitation, and is collected by a body surface electrode in a physiological collection device and presented in a waveform mode. The electrocardiogram reflects the current physiological state of the heart, and generally, the normal heart usually shows the regularity in physiological activities; when different types of pathological changes are generated in the heart, the beating of the heart can cause abnormal frequency or rhythm, and the electrocardiographic waveforms displayed by different symptoms can be different, so that the electrocardiogram is an important technical means and a main basis for detecting and diagnosing arrhythmia diseases. Because there are many types of electrocardiograms caused by lesions, and even if the same type of lesions exist, the electrocardiograms have a large difference. Early cardiovascular diagnosis was performed by manually analyzing the electrocardiogram with experts with abundant clinical experience. However, the medical experience of the expert has a great influence on the accuracy of the diagnosis, the diagnosis process is long, careless mistakes are caused, and therefore misdiagnosis is caused. In addition, the continuous increase of the cardiovascular disease diagnosis demand and the lack of a large number of experienced electrocardiograph experts bring great challenges to cardiovascular disease diagnosis.
With the development of computer informatization technology in recent years, technology for automatically identifying cardiac electrical signals provides reliable support for cardiovascular diagnosis, improves the identification efficiency of cardiac electrical signals, and has become a popular research in the field of automatic analysis of biological electrical signals. However, the current electrocardiogram recognition only plays a role of auxiliary diagnosis, and the final diagnosis result still needs experts to make a decision; the existing automatic electrocardiosignal identification system still has the problems of low identification accuracy and the like, so that the identification of the electrocardiosignals still has a large research space.
Disclosure of Invention
In order to solve the technical problems, the invention provides an SVM electrocardiosignal identification method based on improved Husky algorithm optimization. When the method is used for identifying and classifying the electrocardiosignals, the identification result can be quickly obtained, and the method has higher identification rate, and the aim of the invention is realized by the following technical scheme:
an SVM electrocardiosignal recognition method based on improved Husky algorithm optimization comprises the following steps:
step S1: acquiring ECG data;
step S2: after simple denoising processing is carried out on the ECG signal, feature extraction is carried out through wavelet transformation, and thus the entropy feature of the original ECG signal sample is obtained
Figure GDA0003197949550000011
Sum wavelet energy ratio characterization
Figure GDA0003197949550000012
And the combined and fused feature space is used as an input vector of a classification model;
step S3: entering an improved gray wolf algorithm DIGWO, calculating the fitness value of each gray wolf in the iteration process, comparing the fitness value with the fitness value of the corresponding gray wolf in the previous iteration, replacing position information if the fitness value is better than the fitness value, otherwise, stopping the iteration when the maximum iteration frequency is reached, and outputting the optimal combination of a parameter penalty factor C and a kernel function parameter g, wherein the improved gray wolf algorithm DIGWO is an improvement on the gray wolf algorithm from three aspects of subgroup division, a convergence factor and a proportion weight on the basis of the traditional gray wolf algorithm, and the improved gray wolf algorithm comprises the following specific flow steps:
step S31: initializing parameters, setting the number N of the wolf populations, the iteration times T, and randomly initializing the positions of the wolf populations;
step S32: traversing the current population of the wolfs, calculating the fitness value, finding out the optimal individual alpha wolf in the wolf population and the omega wolf with the worst fitness, and respectively calculating and keeping the coordinate X of the alpha wolfα,YαAnd the coordinate X of the worst position omega wolfω,Yω；
Step S33: respectively calculating the distance Disi _ abest between the wolf individual i and the optimal individual and the distance Disi _ worst between the wolf individual i and the worst individual; if Disti _ abest is less than or equal to (Disti _ worst)/2, dividing the wolf individual i into a better subgroup, and turning to the step S34; otherwise, divide the gray wolf into the bad subgroup, go to step S35;
step S34: the gray wolf in the better subgroup adopts a convergence factor a1 and an Euclidean weight to update the position of the gray wolf individual, calculates the fitness value of each gray wolf, compares the fitness value with the fitness value of the corresponding gray wolf in the previous iteration, replaces the position information if the fitness value is better, and otherwise does not move;
step S35: updating the individual positions of the gray wolves by adopting a convergence factor a2 and an Euclidean weight in the poor subgroup, calculating the fitness value of each gray wolve, comparing the fitness value with the fitness value of the corresponding gray wolve in the previous iteration, and replacing the position information if the fitness value is better than the fitness value of the corresponding gray wolve, otherwise, keeping the fitness value unchanged;
step S36: combining the better and worse sub-groups;
step S37: judging whether the DIGWO algorithm reaches the iteration times, if so, ending, and if not, returning to the step S32;
step S4: training the SVM by using the obtained parameters C and g, and establishing a DIGWO-SVM classification model;
step S5: classifying based on a trained DIGWOO-SVM classification system;
step S6: and taking the DIGWO-SVM classification result as an electrocardiosignal identification result.
Further, the calculation formula of the distance Disti _ abst between the wolf individual i and the optimal individual and the distance Disti _ worst between the wolf individual i and the worst individual is as follows (1):
Figure GDA0003197949550000021
in the formula, XiIs the position of the present generation of wolfαIs the alpha wolf position with the highest fitness value in the contemporary population, XωIs the lowest fitness gray wolf position in the contemporary population.
According to a further scheme, aiming at the difference between fitness values of a better subgroup and a poorer subgroup, different convergence factors are adopted, wherein the formula (2) is a convergence factor formula of the better subgroup, and the formula (3) is a convergence factor formula of the poorer subgroup:
Figure GDA0003197949550000022
Figure GDA0003197949550000023
the further scheme is that in the improved grey wolf algorithm, a position updating strategy based on Euclidean distance is provided, the position of the current grey wolf individual is updated, and the position information of the grey wolf individual is updated according to the following formulas (4) to (7):
the distance formula of each omega wolf and alpha, beta and delta wolf is as follows (4):
Figure GDA0003197949550000031
updating the position information of the alpha, the beta and the delta wolfs as the following formula (5):
Figure GDA0003197949550000032
in the formula C1＝2r11，C2＝2r12，C3＝2r13，A1＝2ar21-a，A2＝2ar22-a，A3＝2ar23A where a decreases linearly from 2 to 0 in an iterative process, a is taken as a1 if the current grey wolf individual belongs to the better subgroup, a is taken as a2 if the current grey wolf individual belongs to the worse subgroup; r11, r12, r13, r21, r22 and r23 are all values of [0,1 ]]A random vector in between;
x in the formula (4)α、Xβ、XδThe position vectors of alpha, beta and delta in the current population respectively, X (t) is the position of the tth generation wolf individual, Dα、Dβ、DδRespectively representing the distances between the current candidate alpha, beta, delta gray wolf and the t generation gray wolf individual, C1，C2，C3Is the wobble factor; x in the formula (5)1Denotes the iterated alpha wolf position, X2Denotes the beta wolf position, X3Denotes the delta wolf position, A1，A2，A3Are respective convergence factors;
location update strategy based on Euclidean distance, with Dα、Dβ、DδThe proportion of the reciprocal is used as a weight coefficient, the positions of the individuals of the other gray wolves except the alpha, the beta and the delta wolves in the current gray wolve are updated, and the calculation formula is as the following formula (7):
Figure GDA0003197949550000033
Figure GDA0003197949550000034
compared with the prior art, the invention has the following beneficial effects:
(1) the invention adopts a multi-scale sample entropy characteristic extraction method, combines the sample entropy characteristic and the wavelet energy ratio characteristic together to form a characteristic space, completely expresses the essence of electrocardiosignals, reduces the influence of noises such as electromyographic interference, power frequency interference and the like, and simultaneously reduces the dimensionality of the electrocardio characteristics.
(2) On the basis of the traditional gray wolf algorithm, the method improves three aspects of subgroup division, convergence factors and proportion weight by introducing a double subgroup strategy, is different from the traditional gray wolf algorithm, is easy to fall into local optimization, can obtain a global optimal solution, improves the solving precision to a certain extent, and has higher convergence speed.
(3) The method provided by the invention has higher accuracy, has a larger effect of improving the recognition rate of different heart beats, and has the advantages of simple method, low cost and the like.
Drawings
FIG. 1 is a flow chart of an SVM electrocardiosignal recognition method based on improved Husky algorithm optimization;
fig. 2 is a flow chart of an improved graying algorithm.
Detailed Description
The technical scheme of the invention is further explained with reference to the accompanying drawings 1-2.
The invention provides an SVM electrocardiosignal recognition method based on improved Husky algorithm optimization, a flow chart is shown as figure 1, and the specific implementation steps are as follows:
step S1: acquiring ECG data;
step S2: after simple denoising processing is carried out on the ECG signal, feature extraction is carried out through wavelet transformation, and thus the entropy feature of the original ECG signal sample is obtained
Figure GDA0003197949550000041
Sum wavelet energy ratio characterization
Figure GDA0003197949550000042
And make the combined feature spaceAn input vector of the classification model;
step S3: entering an improved wolf algorithm DIGWO, initializing positions of alpha, beta and delta wolfs and an objective function value of a wolf group, wherein the position of each wolf individual is composed of a penalty factor C to be optimized and a kernel function parameter g, calculating the fitness value of each wolf in an iteration process, comparing the fitness value with the fitness value of the corresponding wolf in the previous iteration, replacing position information if the fitness value is better than the fitness value, otherwise, keeping the fitness, finally judging whether the maximum iteration number is reached, stopping the iteration when the maximum iteration number is reached, and outputting the optimal combination of the parameters C and g;
step S4: training the SVM by using the obtained parameters C and g, and establishing a DIGWO-SVM classification model;
step S5: classifying based on a trained DIGWOO-SVM classification system;
step S6: and taking the DIGWO-SVM classification result as an electrocardiosignal identification result.
Specifically, the improved grayling algorithm DIGWOL is improved from three aspects of subgroup division, convergence factors and proportion weight on the basis of the traditional grayling algorithm.
Subgroup division: a dynamic subgroup-based mode is adopted, and the distance Disti _ best between the current grey wolf individual and the current optimal individual alpha wolf and the distance Disti _ best between the current worst individual are respectively calculated in an iteration process, wherein the grey wolf individual is divided into alpha, beta, delta and omega wolfs, the number of omega wolfs is large, the current worst individual is also one of omega, and the fitness value of the worst individual is the lowest) as shown in the following formula (1).
If Disti _ abest is less than or equal to (Disti _ worst)/2, dividing the Grey wolf individual into a better subgroup taking the current generation optimal individual as a geometric center, otherwise dividing the Grey wolf individual into a worse subgroup taking the current generation worse individual as a center, and then adopting different optimizing strategies.
Figure GDA0003197949550000043
In the formula, XiIs the position of the present generation of wolfαIs the alpha wolf position with the highest fitness value in the contemporary population, XωIs the lowest fitness gray wolf position in the contemporary population.
Convergence factor: and aiming at the difference of fitness values of the better subgroup and the poorer subgroup, different convergence factors are adopted. For a better subgroup with a higher fitness value, in order to make the wolf with the higher fitness value quickly close to the optimal solution, the reduction rate of a1 is large in the initial stage, and the local optimal a1 value is prevented from slowly decreasing in the later stage; for the poor subgroup with lower fitness value, the following formula (3) is adopted, in order to obtain a larger optimizing range for the wolf with lower fitness in the initial stage, the value of a2 is slowly reduced, and in the later stage, the convergence is completed by rapidly reducing.
Figure GDA0003197949550000051
Figure GDA0003197949550000052
Proportional weight: in order to reflect the importance of alpha, beta and delta wolfs in guiding omega wolfs to carry out position updating, a position updating strategy based on Euclidean distance is adopted. In the traditional gray wolf algorithm, the current gray wolf individual position is updated by formula (6), and in the improved gray wolf algorithm, the current gray wolf individual position is updated by formula (7), and the traditional gray wolf algorithm and the improved gray wolf algorithm update the current gray wolf individual position by the calculation method as follows:
Figure GDA0003197949550000053
Figure GDA0003197949550000054
in the formula C1＝2r11，C2＝2r12，C3＝2r13，A1＝2ar21-a，A2＝2ar22-a，A3＝2ar23A wherein a is in iterationLinearly decreasing from 2 to 0 in the process, if the current wolf individual belongs to the superior subgroup, a is taken as a1, if the current wolf individual belongs to the inferior subgroup, a is taken as a 2; r11, r12, r13, r21, r22 and r23 are all values of [0,1 ]]Random vector in between.
Figure GDA0003197949550000055
X in the formula (4)α、Xβ、XδThe position vectors of alpha, beta and delta in the current population respectively, X (t) is the position of the tth generation wolf individual, Dα、Dβ、DδRespectively representing the distances between the current candidate alpha, beta, delta gray wolf and the t generation gray wolf individual, C1，C2，C3Is the wobble factor; x in the formula (5)1Denotes the iterated alpha wolf position, X2Denotes the beta wolf position, X3Denotes the delta wolf position, A1，A2，A3Are the respective convergence factors.
The improved Husky wolf algorithm of the invention is based on the position updating strategy of Euclidean distance, and D is used after the step (5)α、Dβ、DδThe proportion of the reciprocal is used as a weight coefficient, and the positions of the individuals of the gray wolfs except for alpha, beta and delta wolfs in the current gray wolf are updated by the following formula (7):
Figure GDA0003197949550000056
Figure GDA0003197949550000061
Figure GDA0003197949550000062
the three aspects of improvement are carried out on the basis of the traditional gray wolf algorithm, and an algorithm flow chart of the gray wolf algorithm DIGWO based on double subgroup nonlinear convergence factors is provided and is shown in FIG. 2, wherein the flow steps are as follows:
step S31: initializing parameters, setting the number N of the wolf populations, the iteration times T, and randomly initializing the positions of the wolf populations;
step S32: and traversing the current population of the wolfs, calculating the fitness value, and finding out the optimal individual alpha wolf and the omega wolf with the worst fitness in the wolf population. Respectively calculating and retaining coordinate X of alpha wolfα,YαAnd the coordinate X of the worst position omega wolfω,Yω；
Step S33: respectively calculating the distance Disi _ abest between the wolf individual i and the optimal individual and the distance Disi _ worst between the wolf individual i and the worst individual; if Disti _ abest is less than or equal to (Disti _ worst)/2, dividing the wolf individual i into a better subgroup, and turning to the step S34; otherwise, divide the gray wolf into the bad subgroup, go to step S35;
step S34: the gray wolf in the better subgroup adopts a convergence factor a1 and an Euclidean weight to update the position of the gray wolf individual, calculates the fitness value of each gray wolf, compares the fitness value with the fitness value of the corresponding gray wolf in the previous iteration, replaces the position information if the fitness value is better, and otherwise does not move;
step S35: updating the individual positions of the gray wolves by adopting a convergence factor a2 and an Euclidean weight in the poor subgroup, calculating the fitness value of each gray wolve, comparing the fitness value with the fitness value of the corresponding gray wolve in the previous iteration, and replacing the position information if the fitness value is better than the fitness value of the corresponding gray wolve, otherwise, keeping the fitness value unchanged;
step S36: combining the better and worse sub-groups;
step S37: and judging whether the DIGWO algorithm reaches the iteration times, if so, ending the process, and if not, returning to the step S32.
TABLE 1 improved Grey wolf algorithm and conventional algorithm comparative test results
Model (model) GWO-SVM PSO-SVM CS-SVM DIGWO-SVM
Average recognition rate 98.28％ 97.38％ 98.92％ 99％
Highest recognition rate 99％ 98％ 99％ 99％
Lowest recognition rate 98.2％ 97％ 98.8％ 99％
Standard deviation of 0.253 0.273 0.103 0
The foregoing merely represents preferred embodiments of the invention, which are described in some detail and detail, and therefore should not be construed as limiting the scope of the invention. It should be noted that, for those skilled in the art, various changes, modifications and substitutions can be made without departing from the spirit of the present invention, and these are all within the scope of the present invention. Therefore, the protection scope of the present patent shall be subject to the appended claims.

Claims (4)

1. An SVM electrocardiosignal recognition method based on improved Husky algorithm optimization is characterized by comprising the following steps:
step S1: acquiring ECG data;
step S2: after simple denoising processing is carried out on the ECG signal, feature extraction is carried out through wavelet transformation, and thus the entropy feature of the original ECG signal sample is obtained
Figure FDA0003197949540000011
Sum wavelet energy ratio characterization
Figure FDA0003197949540000012
And the combined and fused feature space is used as an input vector of a classification model;
step S3: entering an improved gray wolf algorithm DIGWO, calculating the fitness value of each gray wolf in the iteration process, comparing the fitness value with the fitness value of the corresponding gray wolf in the previous iteration, replacing position information if the fitness value is better than the fitness value, otherwise, stopping the iteration when the maximum iteration frequency is reached, and outputting the optimal combination of a parameter penalty factor C and a kernel function parameter g, wherein the improved gray wolf algorithm DIGWO is an improvement on the gray wolf algorithm from three aspects of subgroup division, a convergence factor and a proportion weight on the basis of the traditional gray wolf algorithm, and the improved gray wolf algorithm comprises the following specific flow steps:
step S31: initializing parameters, setting the number N of the wolf populations, the iteration times T, and randomly initializing the positions of the wolf populations;
step S32: traversing the current population of the wolfs, calculating the fitness value, finding out the optimal individual alpha wolf in the wolf population and the omega wolf with the worst fitness, and respectively calculating and keeping the coordinate X of the alpha wolfα,YαAnd coordinates of worst position ω wolfXω,Yω；
Step S33: respectively calculating the distance Disi _ abest between the wolf individual i and the optimal individual and the distance Disi _ worst between the wolf individual i and the worst individual; if Disti _ abest is less than or equal to (Disti _ worst)/2, dividing the wolf individual i into a better subgroup, and turning to the step S34; otherwise, divide the gray wolf into the bad subgroup, go to step S35;
step S34: the gray wolf in the better subgroup adopts a convergence factor a1 and an Euclidean weight to update the position of the gray wolf individual, calculates the fitness value of each gray wolf, compares the fitness value with the fitness value of the corresponding gray wolf in the previous iteration, replaces the position information if the fitness value is better, and otherwise does not move;
step S35: updating the individual positions of the gray wolves by adopting a convergence factor a2 and an Euclidean weight in the poor subgroup, calculating the fitness value of each gray wolve, comparing the fitness value with the fitness value of the corresponding gray wolve in the previous iteration, and replacing the position information if the fitness value is better than the fitness value of the corresponding gray wolve, otherwise, keeping the fitness value unchanged;
step S36: combining the better and worse sub-groups;
step S37: judging whether the DIGWO algorithm reaches the iteration times, if so, ending, and if not, returning to the step S32;
step S4: training the SVM by using the obtained parameters C and g, and establishing a DIGWO-SVM classification model;
step S5: classifying based on a trained DIGWOO-SVM classification system;
step S6: and taking the DIGWO-SVM classification result as an electrocardiosignal identification result.
2. The SVM electrocardiosignal identification method based on the improved Grey wolf algorithm optimization as claimed in claim 1, wherein the calculation formula of the distance Disi _ abest between the Grey wolf individual i and the optimal individual and the distance Disi _ worst between the Grey wolf individual i and the worst individual is as follows (1):
Figure FDA0003197949540000013
in the formula, XiIs the position of the present generation of wolfαIs the alpha wolf position with the highest fitness value in the contemporary population, XωIs the lowest fitness gray wolf position in the contemporary population.
3. The SVM electrocardiosignal recognition method based on the improved grayling algorithm optimization as claimed in claim 1, wherein different convergence factors are adopted for the difference between fitness values of the better subgroup and the poorer subgroup, wherein the following formula (2) is a convergence factor formula of the better subgroup, and the following formula (3) is a convergence factor formula of the poorer subgroup:
Figure FDA0003197949540000021
Figure FDA0003197949540000022
4. the SVM electrocardiosignal recognition method based on the improved grey wolf algorithm optimization as claimed in claim 3, wherein in the improved grey wolf algorithm, a position updating strategy based on the euclidean distance is proposed, the position of the current grey wolf individual is updated, and the position information of the grey wolf individual is updated according to the following formula (4) to formula (7):
the distance formula of each omega wolf and alpha, beta and delta wolf is as follows (4):
Figure FDA0003197949540000023
updating the position information of the alpha, the beta and the delta wolfs as the following formula (5):
Figure FDA0003197949540000024
in the formula C1＝2r11，C2＝2r12，C3＝2r13，A1＝2ar21-a，A2＝2ar22-a，A3＝2ar23A, where a decreases linearly from 2 to 0 in an iterative process, a is taken as a1 if the current grey wolf individual belongs to the better subgroup, a is taken as a2 if the current grey wolf individual belongs to the worse subgroup; r11, r12, r13, r21, r22 and r23 are all values of [0,1 ]]A random vector in between;
x in the formula (4)α、Xβ、XδThe position vectors of alpha, beta and delta in the current population respectively, X (t) is the position of the tth generation wolf individual, Dα、Dβ、DδRespectively representing the distances between the current candidate alpha, beta, delta gray wolf and the t generation gray wolf individual, C1，C2，C3Is the wobble factor; x in the formula (5)1Denotes the iterated alpha wolf position, X2Denotes the beta wolf position, X3Denotes the delta wolf position, A1，A2，A3Are respective convergence factors;
location update strategy based on Euclidean distance, with Dα、Dβ、DδThe proportion of the reciprocal is used as a weight coefficient, the positions of the individuals of the other gray wolves except the alpha, the beta and the delta wolves in the current gray wolve are updated, and the calculation formula is as the following formula (7):
Figure FDA0003197949540000025
Figure FDA0003197949540000031
Figure FDA0003197949540000032
CN202011180394.5A 2020-10-29 2020-10-29 SVM (support vector machine) electrocardiosignal identification method based on improved Husky algorithm optimization Active CN112163570B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011180394.5A CN112163570B (en) 2020-10-29 2020-10-29 SVM (support vector machine) electrocardiosignal identification method based on improved Husky algorithm optimization

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011180394.5A CN112163570B (en) 2020-10-29 2020-10-29 SVM (support vector machine) electrocardiosignal identification method based on improved Husky algorithm optimization

Publications (2)

Publication Number Publication Date
CN112163570A CN112163570A (en) 2021-01-01
CN112163570B true CN112163570B (en) 2021-10-19

Family

ID=73865123

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011180394.5A Active CN112163570B (en) 2020-10-29 2020-10-29 SVM (support vector machine) electrocardiosignal identification method based on improved Husky algorithm optimization

Country Status (1)

Country Link
CN (1) CN112163570B (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112767432B (en) * 2021-02-24 2022-10-25 长沙民政职业技术学院 Nuclear intuition fuzzy clustering image segmentation method based on differential mutation grayish wolf optimization
CN113053348B (en) * 2021-03-12 2023-08-11 上海物骐微电子有限公司 Active noise control method and system based on wolf algorithm
CN114462627A (en) * 2022-03-16 2022-05-10 兰州理工大学 Method for diagnosing abnormity of top-blown smelting system based on Hui wolf algorithm and support vector machine

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105183973B (en) * 2015-09-01 2018-03-02 荆楚理工学院 A kind of grey wolf algorithm optimization method of variable weight
CN106355192B (en) * 2016-08-16 2019-12-31 温州大学 Support vector machine method based on chaos grey wolf optimization
CN108694390B (en) * 2018-05-15 2022-06-14 南京邮电大学 Modulation signal classification method for cuckoo search improved wolf optimization support vector machine

Also Published As

Publication number Publication date
CN112163570A (en) 2021-01-01

Similar Documents

Publication Publication Date Title
CN112163570B (en) SVM (support vector machine) electrocardiosignal identification method based on improved Husky algorithm optimization
He et al. Automatic cardiac arrhythmia classification using combination of deep residual network and bidirectional LSTM
Liu et al. Multiple-feature-branch convolutional neural network for myocardial infarction diagnosis using electrocardiogram
CN107837082B (en) Automatic electrocardiogram analysis method and device based on artificial intelligence self-learning
CN106214145B (en) Electrocardiogram classification method based on deep learning algorithm
Xia et al. A novel wearable electrocardiogram classification system using convolutional neural networks and active learning
CN111990989A (en) Electrocardiosignal identification method based on generation countermeasure and convolution cyclic network
WO2020006939A1 (en) Electrocardiogram generation and classification method based on generative adversarial network
CN110974214A (en) Automatic electrocardiogram classification method, system and equipment based on deep learning
CN111261289A (en) Heart disease detection method based on artificial intelligence model
CN112826513B (en) Fetal heart rate detection system based on deep learning and specificity correction on FECG
JP7487965B2 (en) Prediction method of electrocardiogram heart rate multi-type based on graph convolution
CN113499079A (en) Atrial fibrillation detection method in electrocardiogram
Ma et al. An effective data enhancement method for classification of ECG arrhythmia
CN112932498A (en) T wave morphology classification system with strong generalization capability based on deep learning
Tseng et al. Sliding large kernel of deep learning algorithm for mobile electrocardiogram diagnosis
CN114190950B (en) Electrocardiogram intelligent analysis method for noise-containing label and electrocardiograph
Schmidt et al. Nocturnal ventricular repolarization lability predicts cardiovascular mortality in the Sleep Heart Health Study
Han et al. Optimal DNN architecture search using Bayesian Optimization Hyperband for arrhythmia detection
Chen et al. Heart murmurs clustering using machine learning
CN111557660A (en) Arrhythmia identification method under sub-population deep learning framework
Senapati et al. Cardiac arrhythmia classification of ECG signal using morphology and heart beat rate
CN114692698A (en) One-dimensional electrocardiogram data classification method based on residual error network
CN110648761A (en) Method for predicting survival rate of dilated cardiomyopathy patient based on machine learning
WO2021031155A1 (en) Method and device for multi-scale characteristic extraction based on ecg

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