CN116011307A - Signal analysis method and system based on acquisition and recognition of noise panoramic distribution model - Google Patents
Signal analysis method and system based on acquisition and recognition of noise panoramic distribution model Download PDFInfo
- Publication number
- CN116011307A CN116011307A CN202111226095.5A CN202111226095A CN116011307A CN 116011307 A CN116011307 A CN 116011307A CN 202111226095 A CN202111226095 A CN 202111226095A CN 116011307 A CN116011307 A CN 116011307A
- Authority
- CN
- China
- Prior art keywords
- noise
- measurement
- sample
- model
- panorama
- 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.)
- Pending
Links
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation
Landscapes
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
Abstract
The invention belongs to the technical field of signal analysis, and particularly relates to a signal analysis method and a system based on acquisition and recognition of a noise panoramic distribution model, wherein the method comprises the following steps of S1, presetting sample measurement time length; s2, repeated continuous measurement is carried out under a measurement environment of rich conditions, and a plurality of measurement results are obtained; s3, processing measurement results of the reference sample and the test sample to respectively form training data of the reference sample and the test sample; s4, taking the observability of noise as a convergence target, and performing artificial intelligent model training; s5: the measurement results are input into a trained artificial intelligence model and output as the specific type of sample to be identified. The invention is different from the noise processing scheme in the prior art, and provides a signal analysis method based on acquiring and identifying a noise panoramic distribution model from the perspective of completely different technologies so as to solve the problem of noise reduction which is difficult to process in the prior art.
Description
Technical Field
The invention belongs to the technical field of signal analysis, and particularly relates to a signal analysis method and system based on acquisition and recognition of a noise panoramic distribution model.
Background
For applications and demands in many practical fields, there are already a great variety of sample measurement means in the prior art. In theory, all types of sample measurements, even though they appear to be instantaneously completed, are essentially continuous measurement processes involving integration of the measurement values over the time domain. Different measurement time lengths are preset for the continuous measurement process according to the type of the sample to be measured so as to accumulate the signal strength and enable the measurement result to meet the actual observation requirement.
However, due to factors such as the measurement environment, the precision of the equipment, and the properties of the sample itself, the sample measurement result is necessarily mixed data of mixed signals and noise. In a persistent sample measurement process, the noise affected by various aspects is almost impossible to be ideal random noise, but is nonlinear noise, even with quite complex form and content. In this case, the intention is to completely solve the noise problem only to remove the noise source or to shield the interference, but this solution is certainly not practical in actual sample measurement. For complex noise, it is also difficult to design a denoising scheme by using one or several common mathematical models, which are conventional in the prior engineering technology.
In addition, for signals with extremely weak self-strength and complex characteristics, the signals are extremely likely to be submerged by noise. Conventional mathematical noise reduction methods have difficulty in processing such measurement results because it is difficult to build a reasonable mathematical model to simulate and remove the noise mixed in the measurement results. Under stable measurement conditions, even though the signal intensity can be enhanced in an integrated manner by continuous measurement, the complexity of the signal itself still affects its extraction effect; on the other hand, persistence measurement inevitably causes noise to be integrated as well, and thus, the situation that the signal is submerged in the noise cannot be improved, and the signal and the noise are difficult to be stripped all the time.
Disclosure of Invention
The invention mainly aims to provide a signal analysis method based on a noise panoramic distribution model, which aims to solve the technical problem that complex signals under the condition of ultralow signal-to-noise ratio are difficult to analyze in the prior art.
In order to achieve the above purpose, the technical scheme of the application is as follows:
a signal analysis method based on acquisition and recognition of a noise panoramic distribution model comprises the following steps:
s1: presetting a sample measurement time length according to actual sample measurement requirements;
S2: taking the sample measurement duration preset in the step S1 as the reference sample and the test sample, respectively carrying out repeated continuous measurement to obtain a plurality of measurement results; each measurement result comprises a signal and a noise panorama or at least a part of the noise panorama formed under the multi-element noise observation dimension; wherein each persistence measurement is performed under a measurement environment of rich conditions; the rich condition refers to: measurement conditions that are natural, include true complex noise factors, and that are not related to noise suppression, without the aim of maintaining consistency of external conditions.
S3: processing measurement results of the reference sample and the test sample to respectively form training data of the reference sample and the test sample;
s4: based on training data of the reference sample and the test sample, taking observability of noise as a convergence target, and training an artificial intelligent model to enable the model to identify signals and noise from measurement results and distinguish the reference sample and the test sample;
s5: and inputting the measurement result of the sample to be identified into a trained artificial intelligent model, wherein the output result of the artificial intelligent model is the specific type of the sample to be identified.
Alternatively, in step S2, a rich condition measurement environment is created by introducing a slight disturbance multiple times during each continuous measurement of the reference sample and the test sample. Thus, in a sample persistence measurement process, a multiple observation dimension of noise is created, forming a noise panorama or at least a partial noise panorama in the measurement result.
Further, in step S2, during each persistence measurement of the reference sample and the test sample, a plurality of slight perturbations are introduced in a manner selected from continuous introduction, introduction at fixed intervals, or introduction at random intervals.
In each continuous measurement process, the whole measurement results of the reference sample and the test sample and the signals in the measurement results respectively show stable statistical characteristics; at the same time, as the noise panorama, or at least part of the noise panorama, is formed, the statistical distribution pattern of the noise presentation will also tend to stabilize.
Further, the form of the slight disturbance may be selected from but not limited to a spatial disturbance, a physical disturbance, and an environmental disturbance; spatial perturbations include, but are not limited to: slightly displacing the measuring site and slightly rotating the measuring site; physical perturbations include, but are not limited to: agitating the fluid sample by a vibration measurement device or sample; environmental perturbations include, but are not limited to: changing the environment temperature, the environment humidity, the electromagnetic field intensity and the air pressure intensity.
Further, in step S3, the step of processing the measurement results of the reference sample and the test sample to form training data of the reference sample and the test sample, respectively, includes:
S31, carrying out normalization processing on measurement results of a reference sample and a test sample;
s32, establishing a posterior probability model frame based on the normalization result of the step S31;
and (3) respectively forming training data meeting the requirements after processing the measurement results of the reference sample and the test sample in the steps S31-S32 for subsequent artificial intelligent model training.
Further, in step S4, the artificial intelligence model may select, but is not limited to, an artificial neural network, a perceptron, a support vector machine, a bayesian classifier, a bayesian network, a random forest model, or a clustering model.
Further, in step S4, during the training process of the artificial intelligence model, the model performs a great deal of empirical learning, generalization and convergence on the features capable of identifying signals and noise and the features capable of distinguishing the reference sample from the test sample, which are included in the training data, in an iterative manner.
The characteristics capable of realizing signal and noise identification comprise a statistical distribution mode which is presented by a measurement result of persistence measurement and accords with a real mathematical statistical rule of the signal; features capable of realizing noise identification, including noise panorama formed in the persistence measurement process or statistical distribution pattern presented by at least part of the noise panorama and approaching to the real mathematical statistical law of noise; the characteristic of distinguishing the reference sample from the test sample can be realized, and the characteristic comprises a statistical distribution mode respectively presented by the continuous measurement results of the two types of samples.
The invention also provides a signal analysis system based on the obtained and identified noise panoramic distribution model, which comprises a setting module, a measuring module, a processing module, a training module and an analysis module;
the method comprises the steps that according to actual sample measurement requirements, a setting module presets sample measurement time length;
taking the sample measurement duration preset by the setting module as the reference sample and the test sample, respectively carrying out repeated continuous measurement by the measuring module to obtain a plurality of measurement results; each measurement result comprises a signal and a noise panorama or at least a part of the noise panorama formed under the multi-element noise observation dimension; wherein each persistence measurement is performed under a measurement environment of rich conditions; the rich condition refers to: measurement conditions that do not relate to noise suppression, are natural, include real complex noise factors, and are not aimed at maintaining consistency of external conditions;
the processing module processes the measurement results of the reference sample and the test sample to respectively form training data of the reference sample and the test sample;
based on training data of the reference sample and the test sample, the training module takes observability of noise as a convergence target, carries out artificial intelligent model training, enables the model to identify signals and noise from measurement results, and distinguishes the reference sample and the test sample;
For the measurement result of the sample to be identified, the analysis module inputs the measurement result into a trained artificial intelligent model, and the output result of the artificial intelligent model is the specific type of the sample to be identified.
Optionally, the measurement module includes a perturbation mechanism; the perturbation mechanism creates a rich condition measurement environment by introducing a slight perturbation multiple times during each continuous measurement of the reference and test samples by the measurement module. Thus, in a sample persistence measurement process, a multiple observation dimension of noise is created, forming a noise panorama or at least a partial noise panorama in the measurement result.
Further, the perturbation mechanism introduces a plurality of slight perturbations during each persistence measurement of the reference sample and the test sample in a manner selected from continuous introduction, introduction at fixed intervals, or introduction at random intervals.
In each continuous measurement process of the measurement module, the whole measurement results of the reference sample and the test sample and signals in the measurement results respectively show stable statistical characteristics; at the same time, as the noise panorama, or at least part of the noise panorama, is formed, the statistical distribution pattern of the noise presentation will also tend to stabilize.
Further, the form of the slight disturbance introduced by the perturbation mechanism can be selected from but not limited to spatial perturbation, physical perturbation and environmental perturbation; spatial perturbations include, but are not limited to: slightly displacing the measuring site and slightly rotating the measuring site; physical perturbations include, but are not limited to: agitating the fluid sample by a vibration measurement device or sample; environmental perturbations include, but are not limited to: changing the environment temperature, the environment humidity, the electromagnetic field intensity and the air pressure intensity.
Further, the processing module comprises a normalization module and a posterior probability module; the normalization module performs normalization processing on measurement results of the reference sample and the test sample; the posterior probability module establishes a posterior probability model frame based on the normalization result, and respectively forms training data of the reference sample and the test sample which meet the requirements for subsequent artificial intelligent model training.
Further, the artificial intelligence model may select, but is not limited to, an artificial neural network, a perceptron, a support vector machine, a bayesian classifier, a bayesian network, a random forest model, or a clustering model.
Further, in the training process of the artificial intelligence model, the model carries out a great deal of empirical learning, induction and convergence on the characteristics capable of realizing signal and noise identification and the characteristics capable of realizing distinguishing between the reference sample and the test sample contained in the training data in an iterative mode.
The characteristics capable of realizing signal and noise identification comprise a statistical distribution mode which is presented by a measurement result of persistence measurement and accords with a real mathematical statistical rule of the signal; features capable of realizing noise identification, including noise panorama formed in the persistence measurement process or statistical distribution pattern presented by at least part of the noise panorama and approaching to the real mathematical statistical law of noise; the characteristic of distinguishing the reference sample from the test sample can be realized, and the characteristic comprises a statistical distribution mode respectively presented by the continuous measurement results of the two types of samples.
The beneficial effects of this application are:
1. the invention is different from the noise processing scheme in the prior art, and provides a signal analysis method based on acquiring and identifying a noise panoramic distribution model from the perspective of completely different technologies so as to solve the problem of noise reduction which is difficult to process in the prior art.
The signal analysis method provided by the invention is based on a mathematical statistics principle, does not directly strip signals from noise, can still effectively distinguish noise from signals, and can realize successful identification of a plurality of independent signals based on different measurement samples, thereby carrying out practical applications such as sample detection, material classification and the like. In addition, the invention utilizes artificial intelligence technology to perform mixed modeling on noise and signals submerged in the noise. Even if the noise does not have mathematical assumption, the trained artificial intelligent model can deeply mine the hidden mathematical statistics rule in the measurement result, and accurately acquire the mathematical distribution model of the signal and the noise.
2. The invention integrates the signals through continuous sample measurement, improves the observability of the signals, ensures that the data distribution form of the relatively stable signals can be clearly presented, and is beneficial to the subsequent signal extraction and analysis. Meanwhile, the invention does not carry out consistency setting on the measurement conditions, but actively builds a dynamic rich condition measurement environment by introducing slight disturbance. The noise in multiple observation dimensions will accumulate to form a noise panorama or at least a partial noise panorama, i.e. the accumulation of noise "samples" will cover nearly all the possibilities of the noise itself. At the same time, the distribution model of noise will also tend to its true distribution form.
The invention can find the mathematical statistical rule of noise from the mixed data distribution form of the sample measurement result, and realize the distinction of noise and signals and the identification of different types of signals from the angle of a data distribution model. Compared with the noise removal and signal extraction directly performed in the prior engineering technology, the method and the device for the noise removal and signal extraction depth mining of the mathematical statistics rule of the noise and the signal can avoid the false elimination of the noise removal operation on the signal and the influence of the noise removal step commonly used in the prior art on the signal. It follows that the present invention provides an effective solution to the problem that the prior art fails to address, remove noise itself from the confounding sample measurements, or extract signals themselves.
3. Noise is an unavoidable influencing factor in sample measurement, and noise with extremely high complexity even has a dynamic change characteristic. Thus, even with the most excellent measurement conditions at the present stage, the persistent sample measurement will exhibit fluctuations that are infinitely close to the real signal, but such fluctuations can only be dynamic changes that are "statistically stable" in the vicinity of the real signal, and no pre-assumption can be made of specific measurement values in the dynamic changes.
However, the accumulation of the measurement results by the persistent sample measurement gradually eliminates the influence of the uncertainty factor in the dynamic change, so that the whole measurement result tends to be a stable data distribution model. This stable data distribution model represents a macroscopic set of interactions of all components during the measurement. That is, in addition to the actual signals, interference factors that may cause noise, such as environmental complexity, equipment accuracy, and inherent influence of sampling means, are all incorporated into the overall distribution model of the measurement results. Thus, the overall distribution model of the persistent sample measurement can adequately reflect its own characteristics. And the distribution model of the measurement result tends to be stable, and the noise panorama or at least part of the noise panorama formed by integration under multiple observation dimensions also presents a special mathematical statistical rule and approaches to the real distribution model of the noise. The invention realizes the distinction between noise and signals by identifying the complete noise mathematical model, and the identification scheme can obtain more comprehensive and accurate identification results.
4. The continuous sample measurement performed in the rich condition measurement environment enables the noise in the multi-element observation dimension to be accumulated to form a noise panorama or at least a part of the noise panorama, and the data distribution model of the noise tends to reflect the actual distribution model in the noise theory. In this case, the present invention adopts artificial intelligence technology to discover the noise distribution model. The trained artificial intelligent model can find out real characteristics meeting the analysis requirements of experimenters or meeting the analysis purposes of the experimenters in high background noise data, can provide more efficient mathematical operation, and can output highly empirical and more accurate analysis results in real time.
5. According to the invention, a rich condition measurement environment is created through different perturbation introduction means, so that the noise observation dimension in the continuous sample measurement process is increased, the measurement result of continuous measurement can show a complete noise panorama, or at least a part of the noise panorama with enough precision can be provided for subsequent signal analysis. The practical difficulties of different perturbation introduction means are different, and different effects can be caused in terms of increasing the noise observation dimension. In the practical application of the technical scheme, due to comprehensive consideration of various factors such as the characteristics of a sample, a sample measurement means, a measurement precision requirement and the like, an experimenter can completely select from perturbation introduction means provided by the invention according to the practical requirements. The diversified perturbation introduction means disclosed by the invention provides wide choices for experimenters, reduces the application difficulty of the invention to a certain extent, and ensures that the technical scheme has more popularization and application values.
Drawings
FIG. 1 is a schematic flow chart of a signal analysis method based on obtaining and identifying a noise panoramic distribution model;
FIG. 2 is a schematic flow chart of step 3 in the signal analysis method shown in FIG. 1 of the specification;
FIG. 3 is a schematic diagram of forming a noise panorama or at least a portion thereof during a continuous measurement;
FIG. 4 is a system configuration diagram of a signal analysis system based on obtaining and identifying a noise panoramic distribution model according to the present invention
Detailed Description
The purpose of the signal processing is to extract useful information from the sample measurements, such as content of research value or a difference feature from other signals. Limited by the uncertainty in the sample measurement process, these "content of research value" and "difference features" often cannot be represented by independent values, but rather are represented by the overall statistical distribution of the data of the signal.
The measurement results obtained during the actual sample measurement process, in which noise is necessarily mixed in addition to the signal reflecting the actual characteristics. In the signal processing schemes disclosed in the prior art, the influence of noise is either eliminated from the measurement results or signals are extracted from the measurement results. However, when the noise mixed in the measurement result cannot be simulated by a known mathematical model, the noise is eliminated or the signal is extracted with great difficulty.
The technical purpose of the invention is that: based on the data statistics principle, an 'unknown' distribution model of the signals and the noise is found out from the measurement result, so that the signals and the noise can be effectively distinguished. In addition, when the measurement results are derived from different measurement samples, accurate identification of the sample types is achieved by finding out differentiated data distribution models.
For noise in the measurement, the prior art generally considers the "ideal" mathematical statistics of noise to conform to a gaussian distribution. However, the actual sample measurement process often fails to account for "ideal" noise conditions. In addition, even if the sample measurement conditions are optimized as much as possible by means of improving the equipment accuracy, improving the material purity, etc., it is possible that the measurement results having ideal analysis conditions are not obtained. That is, the target signal to be analyzed is submerged in noise due to weak intensity, or the characteristics of the signal are extremely complex and difficult to analyze. Mining of data distribution models for such measurements often presents great difficulty, even without being able to assume the distribution model of noise therein at all.
Under the condition, the invention realizes the sample persistence measurement in the multidimensional perturbation environment by actively creating diversified measurement conditions, thereby realizing the omnibearing observation of noise in the sample persistence measurement process, namely constructing the noise panorama or at least partial noise panorama capable of showing the complete data statistical distribution model, and the data statistical distribution model can be infinitely close to the real distribution of the noise.
Specifically, for persistence measurements, the measurement results correspond to the integral of all transient measurement results over the measurement period. Due to the complexity of the noise environment, there is a difference in noise for each transient measurement. From the sample and sampling point of view, explaining this, the noise in each transient measurement result is equivalent to one random sampling in the noise sample population, and the random sampling result cannot reflect the real characteristics thereof. However, on the premise that the noise has a specific data statistics rule and accords with a specific data distribution model, when all transient measurement results are integrated into a persistence measurement result, the noise sampling range is expanded to be close to the overall noise sample, and the overall noise data statistics rule reflected by the persistence measurement result tends to reflect the real situation. From the statistical point of view, it is completely theoretically possible to show a statistical distribution model of noise through the noise panoramic construction in the perturbation environment.
Under the condition that at least part of the panorama of the noise is obtained and the noise presents a clear and stable distribution model, the invention adopts the artificial intelligence technology to deeply explore the statistical rule of the noise. Artificial intelligence techniques are an effective means for various data analysis and solving empirical data processing. For example, the artificial intelligence deep learning model can simulate the learning process of human beings and rapidly summarize the empirical data processing method of the human beings, thereby realizing the signal recognition and judgment actions. In the invention, the accuracy of the output empirical analysis result of the artificial intelligent model trained by big data can be ensured, so that the mathematical distribution model of noise can be effectively identified, and the specific analysis works such as subsequent noise separation, signal classification and the like can be carried out accordingly.
The present invention will be described in further detail with reference to the following examples in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
Example 1
A signal analysis method based on acquisition and recognition of a noise panoramic distribution model, the flow diagram of which is shown in fig. 1 of the specification, the method comprises the following steps:
s1: presetting a sample measurement time length according to actual sample measurement requirements;
in the embodiment of the present invention, the sample measurement is not a measurement that is transient, but a measurement that lasts for a period of time, and the duration of time needs to be preset before the sample measurement. For example, the measurement duration of the preset spectral measurement is 3s, and the measurement duration of the preset electroencephalogram measurement is 30s. The preset measurement duration is not explicitly specified, but rather depends to a great extent on the specific sample type, measurement means and actual measurement requirements. Specifically, for example, some signal measurements require a period of time to accumulate the measurements, resulting in enhanced results; but some optical measurements cannot last too long to avoid thermal damage to the sample. Therefore, when the sample persistence measurement has a limited condition or an empirical scheme, a person skilled in the art of corresponding sample measurement can reasonably preset the measurement duration according to the actual requirement.
S2: taking the sample measurement duration preset in the step S1 as the reference sample and the test sample, respectively carrying out repeated continuous measurement to obtain a plurality of measurement results; each measurement result comprises a signal and a noise panorama or at least a part of the noise panorama formed under the multi-element noise observation dimension; wherein each persistence measurement is performed under a measurement environment of rich conditions;
in the embodiment of the present invention, step S1 presets a measurement duration of the sample persistence measurement. For example, a measurement duration of 3s for spectral measurement and a measurement duration of 30s for electroencephalogram measurement is specified; in step 2, the persistence measurement of the corresponding sample is performed with the preset measurement duration.
In the technical field of signal acquisition and analysis, the method for maintaining the consistency of external conditions in the sample measurement process is a conventional means for reducing noise fluctuation and forming a good signal-to-noise ratio. However, in the embodiment of the present invention, the consistency setting of the external condition is not involved, but repeated measurement of the reference sample and the test sample is performed in the rich condition measurement environment. Wherein, the rich condition means: measurement conditions that are natural, include true complex noise factors, and that are not related to noise suppression, without the aim of maintaining consistency of external conditions.
In the embodiment of the invention, as shown in fig. 3 of the specification, a single noise observation dimension cannot acquire a real and comprehensive noise observation result, i.e. the noise cannot present a complete data statistics rule conforming to the real distribution characteristic of the noise. However, the invention enables the noise observation dimension in the continuous measurement process to be locally and comprehensively trend through the active nutrient enrichment condition measurement environment, which is beneficial to forming a noise panorama or at least part of the noise panorama. While constructing the noise panorama, the data statistics of the noise will tend to its true mathematical statistics. At the same time, the signal in the measurement will remain statistically unchanged all the time, subject to the properties of the sample itself.
In the embodiment of the invention, the noise panorama refers to: the distribution model of the noise can comprehensively reflect the theoretical real distribution model; partial noise panorama refers to: the distribution model of noise cannot fully reflect its theoretical true distribution model, but the distribution model already has a degree of accuracy that can be used for subsequent signal analysis.
In addition, repeated sample measurements are also considered to be an effective way to reduce random errors. However, in the embodiment of the present invention, the purpose of repeatedly performing the sample persistence measurement is mainly to: a sufficient number of measurements are obtained for artificial intelligence model training.
S3: processing measurement results of the reference sample and the test sample to respectively form training data of the reference sample and the test sample;
s4: based on training data of the reference sample and the test sample, taking observability of noise as a convergence target, and training an artificial intelligent model to enable the model to identify signals and noise from measurement results and distinguish the reference sample and the test sample;
in the embodiment of the invention, training data of the reference sample and the test sample are randomly distributed as learning data and detection data according to preset proportions respectively. Training the artificial intelligent model by adopting learning data, inputting detection data into the trained artificial intelligent model, calculating to obtain a signal identification result, continuing training by adopting the learning data if the signal identification accuracy is lower than a preset threshold, and finishing training by taking the artificial intelligent model if the signal identification accuracy is higher than the preset threshold.
S5: and inputting the measurement result of the sample to be identified into a trained artificial intelligent model, wherein the output result of the artificial intelligent model is the specific type of the sample to be identified.
In the embodiment of the invention, the reference sample and the test sample are taken as two known samples, a plurality of measurement results of the two known samples are respectively processed to form training data, and the trained artificial intelligent model can realize effective distinction of the two known samples. When the sample to be identified is one of two known samples, the artificial intelligent model can accurately identify the specific type of the sample to be identified.
Alternatively, in step S2, a rich condition measurement environment is created by introducing a slight disturbance multiple times during each continuous measurement of the reference sample and the test sample. Thus, in a sample persistence measurement process, a multiple observation dimension of noise is created, forming a noise panorama or at least a partial noise panorama in the measurement result.
In step S2, during each persistence measurement of the reference sample and the test sample, a plurality of slight perturbations are introduced in a manner selected from continuous introduction, introduction at fixed intervals, or introduction at random intervals.
In each continuous measurement process, the whole measurement results of the reference sample and the test sample and the signals in the measurement results respectively show stable statistical characteristics; at the same time, as the noise panorama, or at least part of the noise panorama, is formed, the statistical distribution pattern of the noise presentation will also tend to stabilize.
Further, the form of the slight disturbance may be selected from but not limited to a spatial disturbance, a physical disturbance, and an environmental disturbance; spatial perturbations include, but are not limited to: slightly displacing the measuring site and slightly rotating the measuring site; physical perturbations include, but are not limited to: agitating the fluid sample by a vibration measurement device or sample; environmental perturbations include, but are not limited to: changing the environment temperature, the environment humidity, the electromagnetic field intensity and the air pressure intensity.
Referring to fig. 2 of the specification, in step S3, the step of processing measurement results of the reference sample and the test sample to form training data of the reference sample and the test sample, respectively, includes:
s31, carrying out normalization processing on measurement results of a reference sample and a test sample;
s32, establishing a posterior probability model frame based on the normalization result of the step S31;
and (3) respectively forming training data meeting the requirements after processing the measurement results of the reference sample and the test sample in the steps S31-S32 for subsequent artificial intelligent model training.
Specifically, in the embodiment of the present invention, the measurement results of the reference sample and the test sample are regarded as measurement values obtained by measuring the measurement target constituted by the complex system.
Defining a measured density function as
S is the dimension of the measurement space; v is the measurement environment; then in the measurement target, the number of systems is N, which is defined by the formula (1):
defining B (V) as a measurement function, then measuring
The method comprises the following steps:
wherein, the liquid crystal display device comprises a liquid crystal display device,
the formula (3) is a normalization condition. In the embodiment of the present invention, in order to make the measurement results of the reference sample and the test sample satisfy the normalization condition of the formula (3), the normalization processing is performed on the measurement results of the reference sample and the test sample in step S31.
Since the persistence measurements of the reference and test samples are repeated, the repetition is expressed in a discrete manner, and formula (2) is rewritten as an ensemble:
defining H as the tie-down density function, there are:
<V>＝H<n> (5)
the statistical fluctuation of the complex system is as follows:
wherein δs is the measured information entropy and δp is the measured environmental change amount, corresponding to the repeated measurement.
δp is taken as the statistical space of the noise panorama and δs is taken as the statistical space of the signal. So according to the bayesian formula, there are:
in formula (7), it is defined that
Equation (8), equation (8) is a posterior probability condition. In the embodiment of the present invention, in order to make the measurement results of the reference sample and the test sample satisfy the posterior probability condition of the formula (8), step S32 is adopted, and a posterior probability model frame is established based on the normalization result obtained in step S31.
Estimation of statistical fluctuation δn for complex system * The method comprises the following steps:
δn * ＝argmax δn P(H<n>|δn)P(δn) (9)
in the embodiment of the present invention, the measurement results processed in steps S31 to S32 can satisfy the normalization condition of the formula (3) and the posterior probability condition of the formula (8). The measurement results satisfying the above two conditions can be used to realize the estimation of the statistical fluctuation of the complex system in the expression (9). The measurement results satisfying the two conditions are used as training data in the subsequent artificial intelligence model training step.
In step S4, the artificial intelligence model may select, but is not limited to, an artificial neural network, a perceptron, a support vector machine, a bayesian classifier, a bayesian network, a random forest model, or a clustering model.
In the embodiment of the present invention, the estimation of the statistical fluctuation of the complex system in the above formula (9) will be implemented by an artificial intelligence model.
In step S4, during the training process of the artificial intelligence model, the model performs a great deal of empirical learning, generalization and convergence on the features capable of identifying signals and noise and the features capable of distinguishing the reference sample from the test sample, which are included in the training data, in an iterative manner.
Specifically, the characteristics capable of realizing signal identification comprise a statistical distribution mode which is presented by a measurement result of persistence measurement and accords with a true mathematical statistical rule of the signal; features capable of realizing noise identification, including noise panorama formed in the persistence measurement process or statistical distribution pattern presented by at least part of the noise panorama and approaching to the real mathematical statistical law of noise; the characteristic of distinguishing the reference sample from the test sample can be realized, and the characteristic comprises a statistical distribution mode respectively presented by the continuous measurement results of the two types of samples.
Example 2
As shown in figure 4 of the specification, a signal analysis system based on acquiring and identifying a noise panoramic distribution model comprises a setting module 1, a measuring module 2, a processing module 3, a training module 4 and an analysis module 5;
according to the actual sample measurement requirement, the setting module 1 presets the sample measurement time length;
in the embodiment of the present invention, the sample measurement is not a measurement that is transient and is performed for a period of time, and the setting module 1 needs to be used to preset the duration of time before the sample measurement. For example, the measurement duration of the preset spectral measurement is 3s, and the measurement duration of the preset electroencephalogram measurement is 30s. The preset measurement duration is not explicitly specified, but rather depends to a great extent on the specific sample type, measurement means and actual measurement requirements. Specifically, for example, some signal measurements require a period of time to accumulate the measurements, resulting in enhanced results; but some optical measurements cannot last too long to avoid thermal damage to the sample. Therefore, when the sample persistence measurement has a limited condition or an empirical scheme, a person skilled in the art of corresponding sample measurement can reasonably preset the measurement duration according to the actual requirement.
Taking the sample measurement duration preset by the setting module 1 as the reference, the measurement module 2 respectively performs repeated continuous measurement on the reference sample and the test sample to obtain a plurality of measurement results; each measurement result comprises a signal and a noise panorama or at least a part of the noise panorama formed under the multi-element noise observation dimension; wherein each persistence measurement is performed under a measurement environment of rich conditions; the rich condition refers to: measurement conditions that do not relate to noise suppression, are natural, include real complex noise factors, and are not aimed at maintaining consistency of external conditions;
in the embodiment of the invention, the setting module 1 presets the measurement duration of the sample persistence measurement. For example, a measurement duration of 3s for spectral measurement and a measurement duration of 30s for electroencephalogram measurement is specified; the measurement module 2 will perform the persistence measurement of the corresponding sample for the preset measurement duration.
In the technical field of signal acquisition and analysis, the method for maintaining the consistency of external conditions in the sample measurement process is a conventional means for reducing noise fluctuation and forming a good signal-to-noise ratio. However, in the embodiment of the present invention, the consistency setting of the external condition is not involved, but repeated measurement of the reference sample and the test sample is performed in the rich condition measurement environment. Wherein, the rich condition means: measurement conditions that are natural, include true complex noise factors, and that are not related to noise suppression, without the aim of maintaining consistency of external conditions.
In the embodiment of the invention, as shown in fig. 3 of the specification, a single noise observation dimension cannot acquire a real and comprehensive noise observation result, i.e. the noise cannot present a complete data statistics rule conforming to the real distribution characteristic of the noise. However, the invention enables the noise observation dimension in the continuous measurement process to be locally and comprehensively trend through the active nutrient enrichment condition measurement environment, which is beneficial to forming a noise panorama or at least part of the noise panorama. While constructing the noise panorama, the data statistics of the noise will tend to its true mathematical statistics. At the same time, the signal in the measurement will remain statistically unchanged all the time, subject to the properties of the sample itself.
In the embodiment of the invention, the noise panorama refers to: the distribution model of the noise can comprehensively reflect the theoretical real distribution model; partial noise panorama refers to: the distribution model of noise cannot fully reflect its theoretical true distribution model, but the distribution model already has a degree of accuracy that can be used for subsequent signal analysis.
In addition, repeated sample measurements are also considered to be an effective way to reduce random errors. However, in the embodiment of the present invention, the purpose of repeatedly performing the sample persistence measurement is mainly to: a sufficient number of measurements are obtained for artificial intelligence model training.
The processing module 3 processes the measurement results of the reference sample and the test sample to form training data of the reference sample and the test sample respectively;
based on training data of the reference sample and the test sample, the training module 4 presents observability of noise as a convergence target, performs artificial intelligent model training, enables the model to identify signals and noise from measurement results, and distinguishes the reference sample and the test sample;
in the embodiment of the invention, training data of the reference sample and the test sample are randomly distributed as learning data and detection data according to preset proportions respectively. Training the artificial intelligent model by adopting learning data, inputting detection data into the trained artificial intelligent model, calculating to obtain a signal identification result, continuing training by adopting the learning data if the signal identification accuracy is lower than a preset threshold, and finishing training by taking the artificial intelligent model if the signal identification accuracy is higher than the preset threshold.
For the measurement result of the sample to be identified, the analysis module 5 inputs the measurement result into a trained artificial intelligent model, and the output result of the artificial intelligent model is the specific type of the sample to be identified.
In the embodiment of the invention, the reference sample and the test sample are taken as two known samples, a plurality of measurement results of the two known samples are respectively processed to form training data, and the trained artificial intelligent model can realize effective distinction of the two known samples. When the sample to be identified is one of two known samples, the artificial intelligent model can accurately identify the specific type of the sample to be identified.
Optionally, the measurement module 2 comprises a perturbation mechanism 21; the perturbation mechanism 21 creates a rich condition measurement environment by introducing a slight perturbation multiple times during each continuous measurement of the reference and test samples by the measurement module 2. Thus, in a sample persistence measurement process, a multiple observation dimension of noise is created, forming a noise panorama or at least a partial noise panorama in the measurement result.
The manner in which the perturbation mechanism 21 introduces a plurality of slight perturbations during each duration measurement of the reference sample and the test sample is selected from continuous introduction, introduction at regular intervals, or introduction at random intervals.
In each continuous measurement process of the measurement module 2, the overall measurement results of the reference sample and the test sample and the signals in the measurement results will respectively show stable statistical characteristics; at the same time, as the noise panorama, or at least part of the noise panorama, is formed, the statistical distribution pattern of the noise presentation will also tend to stabilize.
Further, the form of the slight disturbance introduced by the perturbation mechanism 21 may be selected from, but not limited to, spatial perturbation, physical perturbation, and environmental perturbation; spatial perturbations include, but are not limited to: slightly displacing the measuring site and slightly rotating the measuring site; physical perturbations include, but are not limited to: agitating the fluid sample by a vibration measurement device or sample; environmental perturbations include, but are not limited to: changing the environment temperature, the environment humidity, the electromagnetic field intensity and the air pressure intensity.
The processing module 3 comprises a normalization module 31 and a posterior probability module 32;
the normalization module 31 performs normalization processing on measurement results of the reference sample and the test sample; the posterior probability module 32 establishes a posterior probability model framework based on the normalization result, and respectively forms training data of the reference sample and the test sample which meet the requirements for subsequent artificial intelligence model training.
Specifically, in the embodiment of the present invention, the measurement results of the reference sample and the test sample are regarded as measurement values obtained by measuring the measurement target constituted by the complex system.
Defining a measured density function as
S is the dimension of the measurement space; v is the measurement environment; then in the measurement target, the number of systems is N, which is defined by the formula (1):
defining B (V) as a measurement function, then measuring
The method comprises the following steps:
wherein, the liquid crystal display device comprises a liquid crystal display device,
the formula (3) is a normalization condition. In the embodiment of the present invention, in order to make the measurement results of the reference sample and the test sample satisfy the normalization condition of the formula (3), the normalization module 31 performs normalization processing on the measurement results of the reference sample and the test sample.
Since the persistence measurements of the reference and test samples are repeated, the repetition is expressed in a discrete manner, and formula (2) is rewritten as an ensemble:
Defining H as the tie-down density function, there are:
<V>＝H<n> (5)
the statistical fluctuation of the complex system is as follows:
wherein δs is the measured information entropy and δp is the measured environmental change amount, corresponding to the repeated measurement.
δp is taken as the statistical space of the noise panorama and δs is taken as the statistical space of the signal. So according to the bayesian formula, there are:
in formula (7), it is defined that
Equation (8), equation (8) is a posterior probability condition. In the embodiment of the present invention, in order to make the measurement results of the reference sample and the test sample satisfy the posterior probability condition of equation (8), the posterior probability module 32 establishes a posterior probability model frame based on the normalization result output by the normalization module 31.
Estimation of statistical fluctuation δn for complex system * The method comprises the following steps:
δn * ＝argmax δn P(H<n>|δn)P(δn) (9)
in the embodiment of the present invention, the measurement result processed by the normalization module 31 and the posterior probability module 32 can satisfy the normalization condition of the formula (3) and the posterior probability condition of the formula (8). The measurement results satisfying the above two conditions can be used to realize the estimation of the statistical fluctuation of the complex system in the expression (9). The measurement results satisfying the two conditions are used as training data in the subsequent artificial intelligence model training step.
The artificial intelligence model may select, but is not limited to, an artificial neural network, a perceptron, a support vector machine, a bayesian classifier, a bayesian network, a random forest model, or a clustering model.
In the embodiment of the present invention, the estimation of the statistical fluctuation of the complex system in the above formula (9) will be implemented by an artificial intelligence model.
In the training process of the artificial intelligent model, the model carries out a great deal of empirical learning, induction and convergence on the characteristics capable of realizing signal and noise identification and the characteristics capable of realizing distinguishing between a reference sample and a test sample contained in training data in an iterative mode.
The characteristics capable of realizing signal and noise identification comprise a statistical distribution mode which is presented by a measurement result of persistence measurement and accords with a real mathematical statistical rule of the signal; features capable of realizing noise identification, including noise panorama formed in the persistence measurement process or statistical distribution pattern presented by at least part of the noise panorama and approaching to the real mathematical statistical law of noise; the characteristic of distinguishing the reference sample from the test sample can be realized, and the characteristic comprises a statistical distribution mode respectively presented by the continuous measurement results of the two types of samples.
The above embodiments are only for illustrating the technical solution of the present invention, and are not limiting. Variations and substitutions that would be apparent to one of ordinary skill in the art are within the scope of the present disclosure, and are intended to be included within the scope of the present disclosure. Therefore, the protection scope of the present invention should be defined by the claims.
Claims (10)
1. A signal analysis method based on acquisition and recognition of a noise panoramic distribution model is characterized by comprising the following steps of: the method comprises the following steps:
s1: presetting a sample measurement time length according to actual sample measurement requirements;
s2: taking the sample measurement duration preset in the step S1 as the reference sample and the test sample, respectively carrying out repeated continuous measurement to obtain a plurality of measurement results; each measurement result comprises a signal and a noise panorama or at least a part of the noise panorama formed under the multi-element noise observation dimension; wherein each persistence measurement is performed under a measurement environment of rich conditions; the rich condition refers to: measurement conditions that do not relate to noise suppression, are natural, include real complex noise factors, and are not aimed at maintaining consistency of external conditions;
s3: processing measurement results of the reference sample and the test sample to respectively form training data of the reference sample and the test sample;
S4: based on training data of the reference sample and the test sample, taking observability of noise as a convergence target, and training an artificial intelligent model to enable the model to identify signals and noise from measurement results and distinguish the reference sample and the test sample;
s5: and inputting the measurement result of the sample to be identified into a trained artificial intelligent model, wherein the output result of the artificial intelligent model is the specific type of the sample to be identified.
2. The signal analysis method based on acquiring and identifying a noise panorama distribution model according to claim 1, wherein: in step S2, during each continuous measurement of the reference sample and the test sample, creating a rich condition measurement environment by introducing a slight disturbance a plurality of times, forming a noise panorama or at least a partial noise panorama in the measurement result;
in step S2, during each persistence measurement of the reference sample and the test sample, a plurality of slight perturbations are introduced in a manner selected from continuous introduction, introduction at fixed intervals, or introduction at random intervals.
3. The signal analysis method based on acquiring and identifying a noise panorama distribution model according to claim 2, wherein: the form of the slight disturbance may be selected from but not limited to spatial disturbance, physical disturbance and environmental disturbance; the spatial perturbation includes: slightly displacing the measuring site and slightly rotating the measuring site; the physical perturbation includes: agitating the fluid sample by a vibration measurement device or sample; environmental perturbations include, but are not limited to: changing the environment temperature, the environment humidity, the electromagnetic field intensity and the air pressure intensity.
4. The signal analysis method based on acquiring and identifying a noise panorama distribution model according to claim 1, wherein: in step S3, the step of processing the measurement results of the reference sample and the test sample to form training data of the reference sample and the test sample, respectively, includes:
s31, carrying out normalization processing on measurement results of a reference sample and a test sample;
s32, establishing a posterior probability model frame based on the normalization result of the step S31;
and (3) respectively forming training data meeting the requirements after processing the measurement results of the reference sample and the test sample in the steps S31-S32 for subsequent artificial intelligent model training.
5. The signal analysis method based on acquiring and identifying a noise panorama distribution model according to claim 1, wherein: in step S4, the artificial intelligence model may select, but is not limited to, an artificial neural network, a perceptron, a support vector machine, a bayesian classifier, a bayesian network, a random forest model, or a clustering model.
6. The signal analysis method based on acquiring and identifying a noise panorama distribution model according to claim 1, wherein: in step S4, during the training process of the artificial intelligence model, the model performs a great amount of empirical learning, induction and convergence on the features capable of identifying signals and noise and the features capable of distinguishing the reference sample from the test sample, which are included in the training data, in an iterative manner;
The characteristics capable of realizing signal and noise identification comprise a statistical distribution mode which is presented by a measurement result of persistence measurement and accords with a real mathematical statistical rule of the signal; features capable of realizing noise identification, including noise panorama formed in the persistence measurement process or statistical distribution pattern presented by at least part of the noise panorama and approaching to the real mathematical statistical law of noise; the characteristic of distinguishing the reference sample from the test sample can be realized, and the characteristic comprises a statistical distribution mode respectively presented by the continuous measurement results of the two types of samples.
7. A signal analysis system based on acquiring and identifying a noise panoramic distribution model, characterized in that: the system comprises a setting module, a measuring module, a processing module, a training module and an analyzing module;
the method comprises the steps that according to actual sample measurement requirements, a setting module presets sample measurement time length;
taking the sample measurement duration preset by the setting module as the reference sample and the test sample, respectively carrying out repeated continuous measurement by the measuring module to obtain a plurality of measurement results; each measurement result comprises a signal and a noise panorama or at least a part of the noise panorama formed under the multi-element noise observation dimension; wherein each persistence measurement is performed under a measurement environment of rich conditions; the rich condition refers to: measurement conditions that do not relate to noise suppression, are natural, include real complex noise factors, and are not aimed at maintaining consistency of external conditions;
The processing module processes the measurement results of the reference sample and the test sample to respectively form training data of the reference sample and the test sample;
based on training data of the reference sample and the test sample, the training module takes observability of noise as a convergence target, carries out artificial intelligent model training, enables the model to identify signals and noise from measurement results, and distinguishes the reference sample and the test sample;
for the measurement result of the sample to be identified, the analysis module inputs the measurement result into a trained artificial intelligent model, and the output result of the artificial intelligent model is the specific type of the sample to be identified.
8. The signal analysis system based on acquiring and identifying a noise panorama distribution model according to claim 7, wherein: the measurement module comprises a perturbation mechanism; in the process of continuously measuring the reference sample and the test sample each time, the perturbation mechanism creates a rich condition measuring environment by introducing slight perturbation for a plurality of times, and in the process of continuously measuring the sample, a plurality of observation dimensions of noise are created to form a noise panorama or at least a part of noise panorama in a measuring result;
the manner of introducing the plurality of slight perturbations by the perturbation mechanism during each continuous measurement of the reference sample and the test sample is selected from continuous introduction, introduction at fixed intervals, or introduction at random intervals;
The form of the slight disturbance introduced by the perturbation mechanism comprises spatial perturbation, physical perturbation and environmental perturbation; the spatial perturbation includes: slightly displacing the measuring site and slightly rotating the measuring site; the physical perturbation includes: agitating the fluid sample by a vibration measurement device or sample; the environmental perturbation includes: changing the environment temperature, the environment humidity, the electromagnetic field intensity and the air pressure intensity.
9. The signal analysis system based on acquiring and identifying a noise panorama distribution model according to claim 7, wherein: the processing module comprises a normalization module and a posterior probability module; the normalization module performs normalization processing on measurement results of the reference sample and the test sample; the posterior probability module establishes a posterior probability model frame based on the normalization result, and respectively forms training data of the reference sample and the test sample which meet the requirements for subsequent artificial intelligent model training.
10. The signal analysis system based on acquiring and identifying a noise panorama distribution model according to claim 7, wherein: the artificial intelligent model comprises an artificial neural network, a perceptron, a support vector machine, a Bayesian classifier, a Bayesian network, a random forest model or a clustering model;
In the training process of the artificial intelligent model, the model carries out a great deal of empirical learning, induction and convergence on the characteristics capable of realizing signal and noise identification and the characteristics capable of realizing distinguishing between a reference sample and a test sample contained in training data in an iterative mode.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111226095.5A CN116011307A (en) | 2021-10-21 | 2021-10-21 | Signal analysis method and system based on acquisition and recognition of noise panoramic distribution model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111226095.5A CN116011307A (en) | 2021-10-21 | 2021-10-21 | Signal analysis method and system based on acquisition and recognition of noise panoramic distribution model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116011307A true CN116011307A (en) | 2023-04-25 |
Family
ID=86021709
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111226095.5A Pending CN116011307A (en) | 2021-10-21 | 2021-10-21 | Signal analysis method and system based on acquisition and recognition of noise panoramic distribution model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116011307A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116844572A (en) * | 2023-09-01 | 2023-10-03 | 北京圣传创世科技发展有限公司 | Urban noise map construction method based on clustering and machine learning |
-
2021
- 2021-10-21 CN CN202111226095.5A patent/CN116011307A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116844572A (en) * | 2023-09-01 | 2023-10-03 | 北京圣传创世科技发展有限公司 | Urban noise map construction method based on clustering and machine learning |
| CN116844572B (en) * | 2023-09-01 | 2024-03-15 | 装备智能计算芯片及系统应用北京市工程研究中心有限公司 | Urban noise map construction method based on clustering and machine learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110991406B (en) | RSVP electroencephalogram characteristic-based small target detection method and system | |
| CN112450947B (en) | Dynamic brain network analysis method for emotional arousal degree | |
| CN114564982B (en) | Automatic identification method for radar signal modulation type | |
| CN108304816B (en) | Identity recognition method and device, storage medium and electronic equipment | |
| CN104038792B (en) | For the video content analysis method and apparatus of IPTV supervision | |
| CN108764084A (en) | Video classification methods based on spatial domain sorter network and the time domain network integration | |
| Wang et al. | Morphological analysis of dendrites and spines by hybridization of ridge detection with twin support vector machine | |
| CN110619264B (en) | Method and device for identifying microseism effective signals based on UNet++, and method and device for identifying microseism effective signals based on UNet++ | |
| CN112545532A (en) | Data enhancement method and system for classification and identification of electroencephalogram signals | |
| CN116011307A (en) | Signal analysis method and system based on acquisition and recognition of noise panoramic distribution model | |
| CN114391846A (en) | Emotion recognition method and system based on filtering type feature selection | |
| CN111462173B (en) | Visual tracking method based on twin network discrimination feature learning | |
| US6337927B1 (en) | Approximated invariant method for pattern detection | |
| Gale et al. | Automatic detection of wireless transmissions | |
| CN112052730A (en) | 3D dynamic portrait recognition monitoring device and method | |
| CN114419379A (en) | System and method for improving fairness of deep learning model based on antagonistic disturbance | |
| US9646194B2 (en) | Methods and systems for analysis of fibers and branching structures within an image of a sample | |
| WO2022116508A1 (en) | Signal analysis method and system based on model for acquiring and identifying noise panoramic distribution | |
| CN111797690A (en) | Optical fiber perimeter intrusion identification method and device based on wavelet neural network grating array | |
| CN116340825A (en) | Method for classifying cross-tested RSVP (respiratory tract protocol) electroencephalogram signals based on transfer learning | |
| CN110728189A (en) | Lp/q-mixed norm-based spatial filtering method and device | |
| Fu et al. | Relative speech emotion recognition based artificial neural network | |
| CN113729711A (en) | Electroencephalogram signal analysis method, device, equipment and storage medium | |
| CN106503649A (en) | A kind of short-wave signal detection recognition method based on computer vision | |
| Wijetunge et al. | Unsupervised learning for exploring MALDI imaging mass spectrometry ‘omics’ data |
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 |