技术领域technical field
本发明涉及信号与信息处理和神经生物学交叉领域，特别是涉及基于灰建模的脑电(electroencephalogram,EEG)和GP算法结合的方法。它提出和设计一种将灰建模和GP算法进行结合(GM-GP)的算法。由于对EEG信号进行关联维数计算工作时，计算效率较低，为此需要一种在少数据下依然保持鲁棒的算法，而灰色系统理论正是一种对少数据进行信息提取的理论，本发明是将二者进行结合的一种方法。The invention relates to the cross field of signal and information processing and neurobiology, in particular to a method combining gray modeling-based electroencephalogram (electroencephalogram, EEG) and GP algorithm. It proposes and designs an algorithm combining gray modeling and GP algorithm (GM-GP). Due to the low computational efficiency when performing correlation dimension calculations on EEG signals, an algorithm that remains robust with less data is needed, and the gray system theory is a theory for information extraction with less data. The present invention is a method of combining the two.
背景技术Background technique
EEG信号,作为一种非线性和复杂性信号，一般认为它是由非线性动力学过程产生的。相比于传统的线性分析方法，非线性动力学分析方法对EEG等非线性信号分析而言效果更加理想。关联维数，是一种描述了非线性系统的自由度信息的参数，可反映动力学过程的复杂性。EEG信号的关联维数越大，说明大脑的功能越复杂，反之，大脑功能活动越简单。EEG signal, as a kind of nonlinear and complex signal, is generally considered to be produced by nonlinear dynamic process. Compared with the traditional linear analysis method, the nonlinear dynamic analysis method is more ideal for nonlinear signal analysis such as EEG. The correlation dimension is a parameter that describes the degree of freedom information of the nonlinear system, which can reflect the complexity of the dynamic process. The greater the correlation dimension of the EEG signal, the more complex the brain's function, and vice versa, the simpler the brain's functional activities.
因此利用非线性动力学系统对EEG信号进行分析，计算不同生理状态下EEG信号的关联维数，可以检测到不同睡眠阶段的EEG非线性关联维数的变化，进而对睡眠状态进行评估。但是该方法计算速度慢，不能实现对睡眠状态的实时监测。Therefore, the nonlinear dynamic system is used to analyze the EEG signal and calculate the correlation dimension of the EEG signal in different physiological states, and the change of the nonlinear correlation dimension of the EEG in different sleep stages can be detected, and then the sleep state can be evaluated. However, this method has a slow calculation speed and cannot realize real-time monitoring of the sleeping state.
灰色系统理论(Grey System Theory)是一种对少数据信息提取的方法，因此它适用于对混沌信号进行少数据建模，进而计算它的关联维数，该方法为充分利用EEG信号的非线性特性的特点，本专利将灰建模和关联维数结合，提出一种GM-GP算法，它克服了计算时间长的不足，有效降低了计算时间。Gray System Theory (Grey System Theory) is a method for extracting less data information, so it is suitable for modeling chaotic signals with less data, and then calculating its correlation dimension. This method is to make full use of the nonlinearity of EEG signals The characteristics of the characteristics, this patent combines the gray modeling and the correlation dimension to propose a GM-GP algorithm, which overcomes the shortage of long calculation time and effectively reduces the calculation time.
基于灰色理论和关联维数，本发明提出了一种基于GM-GP算法的睡眠状态监测方法Based on gray theory and correlation dimension, the present invention proposes a sleep state monitoring method based on GM-GP algorithm
发明内容Contents of the invention
本发明提出了一种结合灰建模与关联维数的GM-GP算法的睡眠状态监测方法，它是采用灰建模分别提取EEG参数特征，然后对该参数特征序列建立动力学系统，对这一非线性动力系统进行关联维数计算，利用关联维数监测睡眠状态的方法，基本方案如下：The present invention proposes a sleep state monitoring method combining gray modeling and the GM-GP algorithm of the correlation dimension, which uses gray modeling to extract EEG parameter features respectively, and then establishes a dynamic system for the parameter feature sequence, which is A non-linear dynamical system carries out correlation dimension calculation, utilizes the method of correlation dimension monitoring sleep state, basic scheme is as follows:
1.对原始EEG进行预处理，采用坐标平移的方法对EEG信号进行数据提升，为灰色建模做准备；1. Preprocess the original EEG, and use the method of coordinate translation to upgrade the data of the EEG signal to prepare for gray modeling;
2.将原始数据进行累加生成，建立GM(1,1)灰色模型；2. The original data is accumulated and generated, and the GM(1,1) gray model is established;
3.利用GM(1,1)模型提取特征，求取相应的参数：3. Use the GM(1,1) model to extract features and find the corresponding parameters:
发展系数a和灰作用量b，以上是灰建模过程。The development coefficient a and the gray action b, the above is the gray modeling process.
下面仅利用参数b建立相应的多维动力系统The following only uses the parameter b to establish the corresponding multi-dimensional dynamical system
将得到的参数特征b序列作为多维动力系统的输出，确定所要建立的动力系统的维数m和延迟时间τ，对得到的特征参数b进行相空间重构，利用重构的相空间确定对应的动力系统；Take the obtained parameter characteristic b sequence as the output of the multidimensional dynamical system, determine the dimension m and delay time τ of the dynamical system to be established, perform phase space reconstruction on the obtained characteristic parameter b, and use the reconstructed phase space to determine the corresponding power system;
4.采用GP算法，计算系统的关联维数；4. Use the GP algorithm to calculate the correlation dimension of the system;
5.利用得到关联维数，监测睡眠状态。5. Use the obtained correlation dimension to monitor the sleep state.
本发明的有益效果是，在最后EEG信号进行状态监测过程中，由于采用灰建模的方法提取少数据量的特征，利用GP算法对关联维数进行计算效率得到极大提高；通过将GP算法和基于灰建模和GP算法结合的方法进行对比，本发明获得了高的计算效率。The beneficial effect of the present invention is that, in the state monitoring process of the last EEG signal, since the gray modeling method is used to extract the features with less data volume, the calculation efficiency of the correlation dimension is greatly improved by using the GP algorithm; Compared with the method based on the combination of gray modeling and GP algorithm, the present invention obtains high calculation efficiency.
附图说明Description of drawings
图1基于GM-GP算法的计算流程Figure 1 Calculation process based on GM-GP algorithm
图2本方案计算的关联维数和GP算法关联维数状态检测对比图Figure 2 Comparison of correlation dimension calculated by this scheme and correlation dimension state detection of GP algorithm
图3利用不同点灰建模和GP算法相结合和GP算法相关性分析图Figure 3 The combination of different point gray modeling and GP algorithm and the correlation analysis diagram of GP algorithm
具体实施方式detailed description
下面结合附图，对本发明的具体实施方式作详细说明。The specific implementation manner of the present invention will be described in detail below in conjunction with the accompanying drawings.
本发明提出的GM-GP算法的流程如图1所示，下面结合附图，对本发明的具体实施方式作详细说明。The flow of the GM-GP algorithm proposed by the present invention is shown in FIG. 1 , and the specific implementation of the present invention will be described in detail below in conjunction with the accompanying drawings.
1.对原始EEG进行预处理，包括：对EEG数据进行提升，以便于对数据进行灰色建模；将原始数据进行累加生成，构成累加生成序列；1. Preprocessing the original EEG, including: upgrading the EEG data to facilitate gray modeling of the data; accumulating and generating the original data to form an accumulative generation sequence;
设X(0)＝(x(0)(1),x(0)(2),...,x(0)(n))，其一次累加生成序列为X(1)＝(x(1)(1),x(1)(2),...,x(1)(n)),即二者满足关系：Let X(0) ＝(x(0) (1),x(0) (2),...,x(0) (n)), the sequence generated by one accumulation is X(1) ＝(x( 1) (1), x(1) (2),..., x(1) (n)), that is, the two satisfy the relationship:
其中，X(0)＝(x(0)(1),x(0)(2),...,x(0)(n))为EEG信号的时间序列。X(1)为时间序列的1阶累加生成序列。它满足灰预测方程Wherein, X(0) = (x(0) (1), x(0) (2), . . . , x(0) (n)) is the time series of EEG signals. X(1) is the first-order cumulative generation sequence of the time series. It satisfies the gray prediction equation
2.构建灰色模型GM(1,1)，将称为GM(1,1)模型。2. Construct the gray model GM(1,1), the Called the GM(1,1) model.
该模型的最小二乘估计参数满足其中：The least squares estimation parameters of the model satisfy in:
将(1)代入(2)可解得Substituting (1) into (2) can be solved
将(4)和(5)代入(2)(3)，即可解出发展系数a和灰作用量b。Substituting (4) and (5) into (2) (3), the development coefficient a and gray action b can be solved.
3.确定拟建立的动力系统的维数m和延迟时间τ。首先固定延迟时间τ，求取稳定使关联维数的延迟时间τ；然后固定入维数m，再次求得使关联维数稳定的延迟时间τ。使嵌入维数达到动态稳定的m和τ，即为选取的参数；3. Determine the dimension m and delay time τ of the dynamical system to be established. First, fix the delay time τ, and obtain the delay time τ that stabilizes the correlation dimension; then fix the input dimension m, and obtain the delay time τ that stabilizes the correlation dimension again. The m and τ that make the embedding dimension reach dynamic stability are the selected parameters;
4.对特征参数b进行相空间重构：4. Perform phase space reconstruction on the characteristic parameter b:
其中N＝n-(m-1)*τwhere N=n-(m-1)*τ
3.3.
这是一个m维向量构成的空间，N个列向量yi＝(bi,bi+τ,bi+2τ,…bi+(m-1)τ)This is a space composed of m-dimensional vectors, N column vectors yi =(bi ,bi+τ ,bi+2τ ,…bi+(m-1)τ )
这是要建立的m维动力系统的输出。对于相空间中的任意两个向量yi，yj，以它们之间的最大分量差作为距离，即This is the output of the m-dimensional dynamical system to be built. For any two vectors yi, yj in the phase space, the maximum component difference between them is taken as the distance, that is
5计算关联维数：5 Calculate the correlation dimension:
利用重构的相空间，计算输出累积分布积分。根据第3步得到的嵌入维数m和延迟时间τ以及特征参数b的相空间，对于给定半径，计算关联解放C(r)，如下式所示：Using the reconstructed phase space, the output cumulative distribution integral is computed. According to the phase space of embedding dimension m, delay time τ and characteristic parameter b obtained in step 3, for a given radius, calculate the associated liberation C(r), as shown in the following formula:
其中H(x)为Heaviside单位函数where H(x) is the Heaviside unit function
关联积分C(r)在r→0时与r存在以下关系The correlation integral C(r) has the following relationship with r when r→0
则关联维数为Then the correlation dimension is
6.利用系统活动程度与关联维数的关系，监测睡眠状态。6. Using the relationship between system activity level and correlation dimension to monitor sleep state.
把本发明的算法和经典的GP算法进行对比。实验结果，如图2所示，本发明的算法和经典GP算法有相似的趋势。睡眠阶段1和快速眼动阶段(REM)，关联维数值较高，表明在这两个阶段，人脑处于激活状态。Compare the algorithm of the present invention with the classical GP algorithm. Experimental results, as shown in Figure 2, the algorithm of the present invention has a similar trend to the classical GP algorithm. Sleep stage 1 and rapid eye movement (REM) have higher correlation dimension values, indicating that the human brain is active during these two stages.
在睡眠监测过程中，直接GP计算是相当费时的，通过灰建模和GP算法相结合，可以大大缩短技术时间。在算例中，实验数据来源于MIT-BIH数据库。本方案通过对不同睡眠阶段的关联维数进行显著性分析，实验结果表1和表2所示(被试1)，证实了本方案的有效性。同时通过对GM-GP算法和GP算法的计算进行对比，实验结果表3表示，结果表明，本发明算法可显著提高计算效率。In the process of sleep monitoring, direct GP calculation is quite time-consuming, and the combination of gray modeling and GP algorithm can greatly shorten the technical time. In the calculation example, the experimental data comes from the MIT-BIH database. This program through the significant analysis of the correlation dimension of different sleep stages, the experimental results shown in Table 1 and Table 2 (subject 1), confirmed the effectiveness of the program. At the same time, by comparing the calculations of the GM-GP algorithm and the GP algorithm, the experimental results are shown in Table 3, and the results show that the algorithm of the present invention can significantly improve the calculation efficiency.
表1Table 1
表2Table 2
表3table 3
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611153111.1ACN106682406B (en) | 2016-12-12 | 2016-12-12 | Sleep state monitoring method based on GM-GP algorithm |
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611153111.1ACN106682406B (en) | 2016-12-12 | 2016-12-12 | Sleep state monitoring method based on GM-GP algorithm |
| Publication Number | Publication Date |
|---|---|
| CN106682406Atrue CN106682406A (en) | 2017-05-17 |
| CN106682406B CN106682406B (en) | 2019-02-01 |
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611153111.1AActiveCN106682406B (en) | 2016-12-12 | 2016-12-12 | Sleep state monitoring method based on GM-GP algorithm |
| Country | Link |
|---|---|
| CN (1) | CN106682406B (en) |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109498001A (en)* | 2018-12-25 | 2019-03-22 | 深圳和而泰数据资源与云技术有限公司 | Sleep quality appraisal procedure and device |
| CN112614539A (en)* | 2020-12-31 | 2021-04-06 | 西安邮电大学 | Motor imagery detection method based on TEO-MIC algorithm |
| CN119377900A (en)* | 2024-12-30 | 2025-01-28 | 首都医科大学附属北京天坛医院 | A sleep state monitoring method based on GM-GP algorithm |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101822534A (en)* | 2010-04-02 | 2010-09-08 | 浙江大学 | Pulse wave observing method based on phase space reconstruction |
| CN102274022A (en)* | 2011-05-10 | 2011-12-14 | 浙江大学 | Sleep state monitoring method based on electroencephalogram signals |
| US20120296182A1 (en)* | 2010-03-10 | 2012-11-22 | Universidad De Valladolid | Method and apparatus for monitoring sleep apnea severity |
| CN106020453A (en)* | 2016-05-11 | 2016-10-12 | 西北工业大学 | Brain-computer-interface method based on grey system theory |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120296182A1 (en)* | 2010-03-10 | 2012-11-22 | Universidad De Valladolid | Method and apparatus for monitoring sleep apnea severity |
| CN101822534A (en)* | 2010-04-02 | 2010-09-08 | 浙江大学 | Pulse wave observing method based on phase space reconstruction |
| CN102274022A (en)* | 2011-05-10 | 2011-12-14 | 浙江大学 | Sleep state monitoring method based on electroencephalogram signals |
| CN106020453A (en)* | 2016-05-11 | 2016-10-12 | 西北工业大学 | Brain-computer-interface method based on grey system theory |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109498001A (en)* | 2018-12-25 | 2019-03-22 | 深圳和而泰数据资源与云技术有限公司 | Sleep quality appraisal procedure and device |
| CN112614539A (en)* | 2020-12-31 | 2021-04-06 | 西安邮电大学 | Motor imagery detection method based on TEO-MIC algorithm |
| CN112614539B (en)* | 2020-12-31 | 2023-04-11 | 西安邮电大学 | Motor imagery detection method based on TEO-MIC algorithm |
| CN119377900A (en)* | 2024-12-30 | 2025-01-28 | 首都医科大学附属北京天坛医院 | A sleep state monitoring method based on GM-GP algorithm |
| CN119377900B (en)* | 2024-12-30 | 2025-03-28 | 首都医科大学附属北京天坛医院 | A sleep state monitoring method based on GM-GP algorithm |
| Publication number | Publication date |
|---|---|
| CN106682406B (en) | 2019-02-01 |
| Publication | Publication Date | Title |
|---|---|---|
| Wen et al. | Degradation modeling and RUL prediction using Wiener process subject to multiple change points and unit heterogeneity | |
| CN110515456B (en) | Electroencephalogram signal emotion distinguishing method and device based on attention mechanism | |
| Zhang et al. | Spatiotemporal sparse Bayesian learning with applications to compressed sensing of multichannel physiological signals | |
| Liu et al. | Energy efficient telemonitoring of physiological signals via compressed sensing: A fast algorithm and power consumption evaluation | |
| Pereira Coutinho et al. | Novel fiducial and non‐fiducial approaches to electrocardiogram‐based biometric systems | |
| CN114424940B (en) | Emotion recognition method and system based on multimodal spatiotemporal feature fusion | |
| CN106682406A (en) | Sleep state monitoring method based on GM-GP algorithm | |
| CN110558971A (en) | Method for generating countermeasure network electrocardiogram abnormity detection based on single target and multiple targets | |
| Wang et al. | Image generation and recognition technology based on attention residual GAN | |
| Izonin et al. | A non-linear SVR-based cascade model for improving prediction accuracy of biomedical data analysis | |
| Yucelbas et al. | Effect of EEG time domain features on the classification of sleep stages | |
| CN120145147A (en) | A power equipment fault diagnosis system and related equipment based on deep learning | |
| Mandal et al. | Detection of shockable arrhythmia from electrocardiogram signal using recurrence quantification analysis based deep convolutional neural networks | |
| CN118760926B (en) | A method for EEG speech imagery recognition based on dynamic hypergraph learning | |
| JP7484054B2 (en) | Learning device, feature quantity calculation program generation method, similarity calculator, similarity calculation method, learning program, and similarity calculation program | |
| Li et al. | Classification of tongue coating using Gabor and Tamura features on unbalanced data set | |
| CN104202019B (en) | Kalman filtering method with unknown process noise covariance matrix recursive estimation | |
| Nema et al. | Complexity and dynamics of switched human balance control during quiet standing | |
| Arizmendi et al. | Diagnosis of patients with chronic heart failure implementing wavelet transform and machine learning techniques | |
| Tran et al. | Quantum LSTM Model for Estimation of Energy Expenditure in Human Aging using Wearable IoT Healthcare Technology | |
| Startz | Bayesian heteroskedasticity-Robust standard errors | |
| Kallas et al. | Modeling electrocardiogram using Yule-Walker equations and kernel machines | |
| Yang et al. | Detecting synaptic connections in neural systems using compressive sensing | |
| CN104331836A (en) | Crop decision support method and device | |
| CN119475100B (en) | Health state prediction method based on multi-vital sign time sequence data |
| 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 |