Neural network-based signal denoising method, device and equipment, and storage medium

Abstract

The application provides a neural network-based signal denoising method, device, equipment and storage medium. The method comprises the following steps: determining an extended signal of a data signal in a preset time period based on a preset neural network model according to the data signal in the preset time period; the extended signal comprises at least two signal points before the preset time period and at least two signal points after the preset time period, there is an extreme point in the at least two signal points before the preset time period, and there is an extreme point in the at least two signal points after the preset time period; combining the data signal and the extended signal to obtain a target signal; performing ensemble empirical mode decomposition on the target signal, and performing denoising on the decomposed target signal to obtain a denoised signal of the target signal; and obtaining a denoised signal of the data signal according to the denoised signal of the target signal. The neural network model is used to predict the signal points outside the data signal, so that the denoising precision is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to big data technology, and in particular to a signal denoising method and device based on a neural network, a signal denoising device based on a neural network, a signal denoising equipment based on a neural network, and a storage medium. BACKGROUND

[0002] In the current era of big data, the amount of data generated by users in transactions is massive, and these data all have more or less noise interference. In the face of such a huge amount of data and noise signals, how to carry out effective analysis has always been an important exploration direction, and the first step of data analysis is signal denoising.

[0003] Most of the current denoising methods are improvements made for modal aliasing phenomena, for example, using an EEMD (Ensemble Empirical Mode Decomposition, ensemble empirical mode decomposition) algorithm to alleviate modal aliasing phenomena for denoising. However, using EEMD requires pre-determining the signals extended at the front and back of the signal to be denoised, and currently the front and back endpoints are extended in an artificial or empirical manner. Using an artificial or empirical manner requires the operator to be extremely familiar with the signal, and once a strange signal is encountered, the extended signal will have a large difference from the true signal, resulting in a severely distorted denoised signal, affecting the precision and efficiency of signal denoising. SUMMARY

[0004] The present application provides a signal denoising method and device based on a neural network, and equipment and a storage medium, to improve the precision and efficiency of signal denoising.

[0005] In a first aspect, the present application provides a signal denoising method based on a neural network, comprising:

[0006] According to the data signal in a preset time period, based on a preset neural network model, an extended signal of the data signal is determined; wherein the extended signal includes at least two signal points before the preset time period and at least two signal points after the preset time period, there is an extreme point in the at least two signal points before the preset time period, and there is an extreme point in the at least two signal points after the preset time period;

[0007] The data signal and the extended signal are combined to obtain a target signal;

[0008] The target signal is subjected to ensemble empirical mode decomposition, and the decomposed target signal is denoised to obtain a denoised signal of the target signal;

[0009] According to the denoised signal of the target signal, a denoised signal of the data signal is obtained.

[0010] In a second aspect, the present application provides a neural network-based signal denoising device, comprising:

[0011] an extended signal determination module configured to determine an extended signal of the data signal based on a preset neural network model according to a data signal in a preset time period, wherein the extended signal comprises at least two signal points before the preset time period and at least two signal points after the preset time period, there is an extreme point in the at least two signal points before the preset time period, and there is an extreme point in the at least two signal points after the preset time period;

[0012] a target signal obtaining module configured to combine the data signal and the extended signal to obtain a target signal;

[0013] a target signal denoising module configured to perform ensemble empirical mode decomposition on the target signal, and perform denoising on the decomposed target signal to obtain a denoised signal of the target signal;

[0014] a denoised signal obtaining module configured to obtain a denoised signal of the data signal according to the denoised signal of the target signal.

[0015] In a third aspect, the present application provides an electronic device, comprising a processor and a memory in communication connection with the processor;

[0016] the memory stores computer execution instructions;

[0017] the processor executes the computer execution instructions stored in the memory to implement the neural network-based signal denoising method according to the first aspect of the present application.

[0018] In a fourth aspect, the present application provides a computer readable storage medium, wherein the computer readable storage medium stores computer execution instructions, and the computer execution instructions are executed by a processor to implement the neural network-based signal denoising method according to the first aspect of the present application.

[0019] In a fifth aspect, the present application provides a computer program product, comprising a computer program, and the computer program is executed by a processor to implement the neural network-based signal denoising method according to the first aspect of the present application.

[0020] The application provides a neural network-based signal denoising method and device, a neural network-based signal denoising equipment and a storage medium. BRIEF DESCRIPTION OF DRAWINGS

[0021] The accompanying drawings, which are incorporated herein and constitute part of the specification, illustrate embodiments consistent with the application and, together with the description, serve to explain the principles of the application.

[0022] Figure 1 A flowchart of a neural network-based signal denoising method provided by an embodiment of the application is shown in FIG. 1.

[0023] Figure 2 A neural network structure diagram of an RBF model provided by an embodiment of the application is shown in FIG. 2.

[0024] Figure 3 A schematic diagram of a spliced target signal provided by an embodiment of the application is shown in FIG. 3.

[0025] Figure 4 A flowchart of a neural network-based signal denoising method provided by an embodiment of the application is shown in FIG. 1.

[0026] Figure 5 A schematic diagram of a signal point provided by an embodiment of the application is shown in FIG. 4.

[0027] Figure 6 A structure block diagram of a neural network-based signal denoising device provided by an embodiment of the application is shown in FIG. 5.

[0028] Figure 7 A structure block diagram of a neural network-based signal denoising device provided by an embodiment of the application is shown in FIG. 5.

[0029] Figure 8 A structure block diagram of an electronic device provided by an embodiment of the application is shown in FIG. 6.

[0030] Figure 9 A structure block diagram of an electronic device provided by an embodiment of the application is shown in FIG. 6.

[0031] The specific embodiments of the present application have been shown by the above drawings, and will be described in more detail hereinafter. These drawings and detailed description are not intended to limit the scope of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0032] In order to make the purpose, technical solutions and advantages of the present application clearer, the embodiments of the present application will be described in more detail below with reference to the drawings.

[0033] It should be clear that the described embodiments are only a part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0034] The following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all the embodiments consistent with the present application. Instead, they are only examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.

[0035] In the description of the present application, it should be understood that the terms "first", "second", "third" and the like are only used to distinguish similar objects, and do not necessarily describe a specific order or sequence, nor can they be understood as indicating or implying relative importance. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances. In addition, in the description of the present application, "a plurality of" means two or more, unless otherwise indicated. "And / or", the association between the associated objects, means that there can be three relationships, for example, A and / or B, which can represent the three cases of A alone, A and B together, and B alone. The character " / " generally represents a "or" relationship between the associated objects before and after.

[0036] It should be noted that due to the limitation of the length of the specification, all optional embodiments cannot be enumerated in the present application. Those skilled in the art should be able to think of any combination of technical features as long as the technical features are not contradictory, which can constitute an optional embodiment. Each embodiment will be described in detail below.

[0037] In the current era of big data, the amount of user transaction data owned by each financial institution is massive, and almost all of these data are disturbed by noise. Therefore, signal denoising of these data is an important task for data analysis. The data for signal denoising can be the production and operation parameters of the financial institution, for example, the production and operation parameters can be the transaction volume.

[0038] Most of the current denoising methods are improved for modal aliasing phenomenon, and the EEMD algorithm can greatly alleviate the modal aliasing phenomenon, but the analysis of the end effect is not sufficient at present, and most of the data signal endpoints are extended by artificial or empirical methods.

[0039] EMD (Empirical Mode Decomposition, EMD) is a data processing method that decomposes signals according to their own time scale characteristics without pre-setting any basis function. It can decompose complex signals into a finite number of IMFs (Intrinsic Mode Function), which contain local characteristic signals of the original signal, but their time scales can be different. For data with obvious noise signals, the first IMF after EMD decomposition contains the most noise. The two biggest problems of EMD are modal aliasing and end effect. Modal aliasing refers to the existence of data similar to intermittent signals, and end effect refers to the uncertainty of the extreme value outside the two endpoints of the signal. EEMD is a denoising algorithm that adds adaptive white noise to solve the modal aliasing phenomenon. Because white noise has the same power spectral density in each frequency band, the mean value of white noise is 0 after multiple decomposition and summation, which can effectively reduce the modal aliasing phenomenon caused by intermittent signals.

[0040] However, extending the two endpoints of the signal by artificial or empirical methods requires the operator to be very familiar with the signal. Once a strange signal is encountered, the extended signal will be significantly different from the true signal. And when EEMD is used for denoising, it relies heavily on the original signal. If the signal used for denoising has a large error from the true signal, there will be a lot of true signal in each intrinsic mode function obtained by decomposition, resulting in serious distortion of the denoised signal and affecting the accuracy and efficiency of signal denoising.

[0041] The signal denoising method, device and equipment based on neural network and storage medium provided by the present application aim to solve the above technical problems of the prior art.

[0042] The technical solutions of the present application and how the technical solutions of the present application solve the above technical problems will be described in detail below with specific examples. The following specific examples can be combined with each other, and the same or similar concepts or processes may not be described again in some examples. The embodiments of the present application will be described below with reference to the accompanying drawings.

[0043] Figure 1 is a flowchart of a signal denoising method based on neural network according to an embodiment of the present application, which can be executed by a signal denoising device based on neural network. As shown in the figure, the method comprises the following steps:Figure 1 As shown, the method comprises the following steps:

[0044] S101, according to the data signal in the preset time period, based on the preset neural network model, determine the extension signal of the data signal; wherein, the extension signal includes at least two signal points before the preset time period and at least two signal points after the preset time period, there is an extreme point in the at least two signal points before the preset time period, and there is an extreme point in the at least two signal points after the preset time period.

[0045] Exemplarily, the data signal can be the signal of the data generated in the user transaction process, for example, the data generated in the user transaction process can be the transaction volume, and the data signal can represent the change of the number of transaction volume in a period of time. The data signal in the preset time period is obtained, for example, the signal of the transaction volume in a day can be obtained. The change of the transaction volume can be indicated by the data signal, for example, the data signal of the transaction volume can be represented by a waveform diagram.

[0046] The neural network model is constructed and trained in advance, and in the embodiment, the neural network model can be an RBF (Radial Basis Function) model. Figure 2 The neural network structure diagram of the RBF model in the embodiment of the application is shown in the figure. Figure 2 As shown, the RBF model can include an input layer, a hidden layer and an output layer. The input layer does not process the data signal, and only plays a role in transmitting data. When a group of data signals are transmitted into the network, the input layer first transmits the data signals to the hidden layer, the hidden layer receives the data signals and projects the data signals to a new space, and adjusts the parameters of the preset activation function, and the output layer continuously adjusts the weight value in this new space to obtain the output value. Therefore, the core of the radial basis function neural network lies in the non-linear mapping of the data signal by the hidden layer and the weight value adjustment of the output layer.

[0047] In the training of the pre-constructed network model, the training sample of the data signal to be trained can be pre-acquired, for example, the training sample of the transaction volume in a period of time can be pre-acquired. The signal points outside the front and rear endpoints of the transaction volume data are predicted by using the training sample, and the goal is to predict a period of signal before and after the training sample signal, which includes at least one extreme point, and the extreme point is an inflection point. For example, given a set of training samples of transaction volume signals [S1, S2, …, Sn], S1 to Sn are signal points. When determining whether the neural network model can predict the signal points after a period of time, according to the training sample, the input for training can be [S1, S2, S3, S4, S5], and the corresponding output for training can be S6; the input for training can be [S2, S3, S4, S5, S6], and the corresponding output for training can be S7; the input for training can be [S3, S4, S5, S6, S7], and the corresponding output for training can be S8; the input for training can be [Sn-5, Sn-4, Sn-3, Sn-2, Sn-1], and the corresponding output for training can be Sn.

[0048] The input for training is sent to the pre-constructed neural network model, and it is determined whether the output result is the corresponding output for training. If yes, it is determined that the neural network model can obtain the signal points after a period of time of data signals. The neural network model for outputting the signal points after a period of time of data signals can be determined as a rear-end prediction neural network model.

[0049] When determining whether the neural network model can predict the signal points before a period of time, according to the training sample, the input for training can be [Sn, Sn-1, Sn-2, Sn-3, Sn-4], and the corresponding output for training can be Sn-5; the input for training can be [Sn-1, Sn-2, Sn-3, Sn-4, Sn-5], and the corresponding output for training can be Sn-6; the input for training can be [S6, S5, S4, S3, S2], and the corresponding output for training can be S1.

[0050] The input for training is sent to the pre-constructed neural network model, and it is determined whether the output result is the corresponding output for training. If yes, it is determined that the neural network model can obtain the signal points before a period of time of data signals. The neural network model for outputting the signal points before a period of time of data signals can be determined as a front-end prediction neural network model.

[0051] The preset neural network model can be two, one for predicting a signal point before a period of time, and one for predicting a signal point after a period of time. The data signal in the preset period of time is input into the pre-trained neural network model, and each time the model can output a signal point before the preset period of time, or output a signal point after the preset period of time. After each output of a signal point before the preset period of time, it is judged whether there is a maximum value or a minimum value in the signal point before the preset period of time, if there is, the prediction of the signal point before the preset period of time is stopped, if not, the prediction of the signal point before the preset period of time is continued; after each output of a signal point after the preset period of time, it is judged whether there is a maximum value or a minimum value in the signal point after the preset period of time, if there is, the prediction of the signal point after the preset period of time is stopped, if not, the prediction of the signal point after the preset period of time is continued.

[0052] The predicted signal points before and after the data signal in the preset period of time are determined as the extension signal of the data signal, that is, the extension signal is an extension of the front and rear ends of the data signal.

[0053] S102, the data signal and the extension signal are combined to obtain a target signal.

[0054] Exemplarily, after obtaining the extension signal of the data signal, the data signal and the extension signal can be combined into a complete signal, and the combined signal is the target signal. That is, the time scale of the target signal is greater than the time scale of the data signal.

[0055] In this embodiment, the combination of the data signal and the extension signal includes: splicing the part of the extension signal before the data signal with the data signal, and splicing the part of the extension signal after the data signal with the data signal.

[0056] Specifically, the extension signal can include two parts, which are the part before the data signal in the preset period of time and the part after the data signal in the preset period of time. Splicing the part before the data signal with the data signal can be splicing the part before the data signal with the front end of the data signal. Splicing the part after the data signal with the data signal can be splicing the part after the data signal with the rear end of the data signal. Figure 3 A schematic diagram of the spliced target signal is shown. Figure 3 The solid line part is the data signal in the preset period of time, and the dashed line part is the extension signal.

[0057] The beneficial effect of the arrangement is that the target signal after extension of the data signal can be obtained by splicing the extension signal and the data signal, which facilitates denoising of the extended data signal according to the EEMD algorithm, fully utilizes the EEMD algorithm, and improves the denoising precision of the EEMD algorithm.

[0058] In S103, the target signal is subjected to ensemble empirical mode decomposition, and the decomposed target signal is denoised to obtain a denoised signal of the target signal.

[0059] For example, the target signal is denoised by using a preset EEMD algorithm, which can decompose the target signal and denoise the decomposed target signal. For example, the three decomposed local signals can be denoised respectively. Each local signal is denoised to obtain a corresponding denoised signal, and the denoised signals corresponding to the local signals are spliced to obtain the denoised signal of the target signal.

[0060] In this embodiment, the target signal is subjected to ensemble empirical mode decomposition, and the decomposed target signal is denoised to obtain a denoised signal of the target signal, including: the ensemble empirical mode decomposition of each signal point in the target signal is performed to obtain an intrinsic mode component of each signal point; the intrinsic mode component of each signal point is denoised according to a preset denoising algorithm; and the denoised intrinsic mode component of each signal point is spliced into the denoised signal of the target signal according to the time arrangement order of each signal point in the data signal.

[0061] Specifically, the data signal can be denoted as S, S=[S1,S2,…,Sn], and the target signal can be denoted as S', S'=[S-4,S-3,S-2,S-1,S0,S1,S2…Sn,Sn+1,Sn+2,Sn+3,Sn+4,Sn+5], wherein S1, S2, …, Sn are data signals of a preset time period, S-4, S-3, S-2, S-1, S0, and Sn+1, Sn+2, Sn+3, Sn+4, Sn+5 are extension signals, and S-3 and Sn+4 in the extension signals are extreme points. Each element in S' is decomposed by using EEMD, i.e., each signal point is decomposed to obtain an intrinsic mode component of each signal point at different time scales, which can include IMF1, IMF2, …, IMFn. Generally, for a signal containing random noise, the first intrinsic mode component IMF1 obtained by decomposition contains a large amount of noise, which can be directly filtered as a pure noise signal. For the intrinsic mode components IMF2 and thereafter, a preset denoising algorithm can be used to judge the noise amount. That is, the preset denoising algorithm can be used to judge the noise amount of each intrinsic mode component of each signal point.

[0062] Assuming that the IMF2 and the subsequent intrinsic mode components contain more real signals, these components are not suitable for direct filtering, and the preset denoising algorithm can be sequentially used for denoising. The denoised IMF2 is determined as IMF2', the denoised IMF3 is determined as IMF3', and the denoised IMFn is determined as IMFn'. The denoised signal of each element in S' is the sum of the denoised intrinsic mode components in order, that is, IMF2'+IMF3'+…+IMFn', and the denoised signal of the target signal is obtained by arranging the denoised signal of each element in S' according to the time arrangement order of each element in S', that is, the denoised signal of the target signal is obtained.

[0063] The beneficial effect of such a setting is that the EEMD algorithm is used to alleviate the modal aliasing phenomenon, and the extension signal in the target signal is predicted by the neural network, avoiding the analysis error of the EEMD on the end effect, and effectively improving the accuracy of signal denoising.

[0064] S104, obtaining the denoised signal of the data signal according to the denoised signal of the target signal.

[0065] Exemplarily, the target signal includes the denoised signal of the data signal, and also includes the denoised signal of the extension signal predicted by the neural network, so that the denoised signal of the extension signal can be cut from the target signal to obtain the denoised signal of the data signal.

[0066] In this embodiment, obtaining the denoised signal of the data signal according to the denoised signal of the target signal includes discarding the denoised signal corresponding to the extension signal in the denoised signal of the target signal to obtain the denoised signal of the data signal.

[0067] Specifically, the denoised signal corresponding to the extension signal in the target signal can be determined, for example, the denoised signal corresponding to the dashed portion in the target signal. Figure 3 The denoised signal of the target signal is cut to the interval of the preset time period of the data signal, that is, the denoised signal corresponding to the extension signal in the denoised signal of the target signal is discarded, for example, the denoised signal corresponding to the dashed portion in the target signal is discarded. Figure 3 Thus, the denoised signal of the data signal is obtained. If the data signal is a transaction volume signal, the final transaction volume denoising result can be obtained.

[0068] The beneficial effect of such a setting is that by discarding the denoised signal of the extension signal, only the denoised signal of the data signal can be obtained, the accuracy of signal denoising is improved, and the user can obtain real data.

[0069] The embodiment of the application provides a signal denoising method based on a neural network, which comprises the following steps: determining an extended signal before and after a data signal to be denoised through a preset neural network model, combining the extended signal with the data signal to obtain a target signal, denoising the target signal according to a preset EEMD to obtain a denoised signal of the target signal, and cutting the denoised signal of the target signal to obtain a denoised signal with a time interval consistent with that of the data signal, which is used as the denoised signal of the data signal. The embodiment solves the problem of incorrect determination of the extended signal caused by manual determination of the extended signal, effectively saves manpower and time by using the neural network model, and alleviates the modal aliasing phenomenon and improves the precision and efficiency of signal denoising by using the EEMD algorithm.

[0070] Figure 4 A flowchart of a signal denoising method based on a neural network is provided in the embodiment of the application, which is an optional embodiment based on the above-mentioned embodiment.

[0071] In the embodiment, the determination of the extended signal of the data signal based on the preset neural network model according to the data signal in the preset time period can be refined as follows: obtaining the data signal in the preset time period, and determining a front-end signal point and a rear-end signal point in the data signal; the front-end signal point is a signal point in the data signal at a preset first time point, and the rear-end signal point is a signal point in the data signal at a preset second time point; inputting the data signal into the preset neural network model, outputting a signal point before the front-end signal point as a first front-end extended signal point, and outputting a signal point after the rear-end signal point as a first rear-end extended signal point; inputting the data signal, the first front-end extended signal point and the first rear-end extended signal point into the preset neural network model, outputting a signal point before the first front-end extended signal point as a second front-end extended signal point, and outputting a signal point after the first rear-end extended signal point as a second rear-end extended signal point; if the first front-end extended signal point is an extreme point in the front-end signal point and the second front-end extended signal point, and the first rear-end extended signal point is an extreme point in the rear-end signal point and the second rear-end extended signal point, then the first front-end extended signal point, the second front-end extended signal point, the first rear-end extended signal point and the second rear-end extended signal point are determined as the extended signal of the data signal.

[0072] As Figure 4 shown, the method comprises the following steps:

[0073] S401, obtaining a data signal in a preset time period, and determining a front-end signal point and a rear-end signal point in the data signal; wherein the front-end signal point is a signal point in the data signal at a preset first time point, and the rear-end signal point is a signal point in the data signal at a preset second time point.

[0074] Exemplarily, a time period is preset, for example, the preset time period can be 0 point to 24 point. Data signals generated by the user in the preset time period are acquired, and the data signals can be a waveform diagram connected in time sequence by a plurality of signal points. The front-end signal point and the rear-end signal point are determined from the data signals. A first time point and a second time point can be preset, the signal point located at the preset first time point is determined as the front-end signal point, and the signal point located at the preset second time point is determined as the rear-end signal point. For example, the first signal point in the data signal can be determined as the front-end signal point, and the last signal point in the data signal can be determined as the rear-end signal point.

[0075] S402, input the data signal into the preset neural network model, output a signal point before the front-end signal point, which is a first front-end extended signal point, and output a signal point after the rear-end signal point, which is a first rear-end extended signal point.

[0076] Exemplarily, two neural network models can be trained in advance, which are a front-end prediction neural network model and a rear-end prediction neural network model. The front-end prediction neural network model is used to predict the signal point before the front-end signal point, and the rear-end prediction neural network model is used to predict the signal point after the rear-end signal point. The front-end prediction neural network model and the rear-end prediction neural network model can have consistent model structures and different model parameters.

[0077] The data signal can be input into the neural network model, which can be input into the front-end prediction neural network model and the rear-end prediction neural network model respectively. All data signals can be input into the model, or a preset number of signal points in the data signal can be input into the model. For example, the first ten signal points in the data signal can be input into the front-end prediction neural network model, and the last ten signal points in the data signal can be input into the rear-end prediction neural network model.

[0078] The front-end prediction neural network model can predict a signal point before the front-end signal point, for example, the front-end neural network model can predict a signal point at a time of 10 minutes before the first signal point in the input data. The rear-end prediction neural network model can predict a signal point after the rear-end signal point, for example, the rear-end neural network model can predict a signal point at a time of 10 minutes after the last signal point in the input data. When training the neural network model, the time interval between the signal points output by the neural network model can be preset, for example, the time interval can be 10 minutes.

[0079] The signal point located before the front-end signal point and adjacent to the front-end signal point is determined as a first front-end extended signal point; and the signal point located after the rear-end signal point and adjacent to the rear-end signal point is determined as a first rear-end extended signal point.

[0080] S403, inputting the data signal, the first front-end extended signal point and the first rear-end extended signal point into a preset neural network model, outputting one signal point before the first front-end extended signal point as a second front-end extended signal point, and outputting one signal point after the first rear-end extended signal point as a second rear-end extended signal point.

[0081] Exemplarily, after obtaining the first front-end extended signal point, the first front-end extended signal point and the data signal are input into a front-end prediction neural network model, or a preset number of signal points including the first front-end extended signal point are input into the front-end prediction neural network model. For example, the preset number is 10 signal points, and then the first front-end extended signal point and the first nine signal points in the data signal are input into the front-end prediction neural network model. The front-end prediction neural network model can output one signal point before the first front-end extended signal point, and the signal point located before the first front-end extended signal point and adjacent to the first front-end extended signal point is determined as the second front-end extended signal point.

[0082] After obtaining the first rear-end extended signal point, the first rear-end extended signal point and the data signal are input into a rear-end prediction neural network model, or a preset number of signal points including the first rear-end extended signal point are input into the rear-end prediction neural network model. For example, the preset number is 10 signal points, and then the first rear-end extended signal point and the last nine signal points in the data signal are input into the rear-end prediction neural network model. The rear-end prediction neural network model can output one signal point after the first rear-end extended signal point, and the signal point located after the first rear-end extended signal point and adjacent to the first rear-end extended signal point is determined as the second rear-end extended signal point.

[0083] S404, if the first front-end extended signal point is an extreme point in the front-end signal point and the second front-end extended signal point, and the first rear-end extended signal point is an extreme point in the rear-end signal point and the second rear-end extended signal point, then the first front-end extended signal point, the second front-end extended signal point, the first rear-end extended signal point and the second rear-end extended signal point are determined as extended signals of the data signal.

[0084] Exemplarily, the EEMD algorithm has an end effect, and therefore, it is necessary to determine a signal point in which an extreme value point exists in the extended signal. After obtaining the first front-end extended signal point and the second front-end extended signal point, it is determined whether an extreme value point exists in the signal points before the front-end signal point, that is, it is determined whether the first front-end extended signal point is an extreme value point in the front-end signal point and the second front-end extended signal point, that is, it is determined whether the first front-end extended signal point is an inflection point. If yes, the prediction of the signal points before the front-end signal point is stopped, and the waveform graph connected by the front-end signal point, the first front-end extended signal point and the second front-end extended signal point is taken as the extended signal located before the data signal.

[0085] After obtaining the first back-end extended signal point and the second back-end extended signal point, it is determined whether an extreme value point exists in the signal points after the back-end signal point, that is, it is determined whether the first back-end extended signal point is an extreme value point in the back-end signal point and the second back-end extended signal point, that is, it is determined whether the first back-end extended signal point is an inflection point. If yes, the prediction of the signal points after the back-end signal point is stopped, and the waveform graph connected by the back-end signal point, the first back-end extended signal point and the second back-end extended signal point is taken as the extended signal located after the data signal. Figure 5 A signal point diagram is shown in the embodiment. Figure 5 In the diagram, A is the front-end signal point, B is the back-end signal point, C is the first front-end extended signal point, D is the second front-end extended signal point, E is the first back-end extended signal point, and F is the second back-end extended signal point. C and E are extreme value points.

[0086] In the embodiment, before outputting the first front-end extended signal point, the signal point after the second front-end extended signal point is obtained, and if the first front-end extended signal point is not an extreme value point in the front-end signal point and the second front-end extended signal point, the first front-end extended signal point, the second front-end extended signal point and the data signal are input into a preset neural network model to output the signal point before the second front-end extended signal point, which is the third front-end extended signal point; if the second front-end extended signal point is an extreme value point in the front-end signal point, the first front-end extended signal point and the third front-end extended signal point, the first front-end extended signal point, the second front-end extended signal point and the third front-end extended signal point are determined as the extended signal of the data signal.

[0087] Specifically, if the first front-end extension signal point is not an extreme point among the front-end signal point and the second front-end extension signal point, it is still necessary to continue to extend forward. The first front-end extension signal point, the second front-end extension signal point and the data signal can be input into a preset neural network model, or the first front-end extension signal point, the second front-end extension signal point and a preset part of the data signal can be input into the front-end prediction neural network model. For example, ten signal points are required when the neural network model is preset, and the first front-end extension signal point, the second front-end extension signal point and the first eight signal points in the data signal can be input into the front-end prediction neural network model. The signal point before the second front-end extension signal point is output is determined as the third front-end extension signal point.

[0088] After obtaining the third front-end extension signal point, it is determined whether there is an extreme point before the front-end signal point, that is, whether the second front-end extension signal point is an extreme point among the front-end signal point, the first front-end extension signal point and the third front-end extension signal point. If yes, the extension forward is stopped, and the extension signal before the data signal is composed of the first front-end extension signal point, the second front-end extension signal point and the third front-end extension signal point, that is, the waveform diagram of the signal composed of the front-end signal point, the first front-end extension signal point, the second front-end extension signal point and the third front-end extension signal point is output; if not, the extension forward is continued until there is an extreme point.

[0089] The beneficial effect of such arrangement is that the neural network model is used to extend the front-end point of the data signal, and the extreme point is determined and obtained, which facilitates the EEMD decomposition of the extended signal data, improves the authenticity and reliability of the extended signal by using the neural network to extend the signal, and improves the determination efficiency of the extended signal.

[0090] In the embodiment, the signal point after the first back-end extension signal point is output is the second back-end extension signal point, and further comprising: if the first back-end extension signal point is not an extreme point among the back-end signal point and the second back-end extension signal point, the first back-end extension signal point, the second back-end extension signal point and the data signal are input into a preset neural network model, and the signal point after the second back-end extension signal point is output as the third back-end extension signal point; if the second back-end extension signal point is an extreme point among the back-end signal point, the first back-end extension signal point and the third back-end extension signal point, the first back-end extension signal point, the second back-end extension signal point and the third back-end extension signal point are determined as the extension signal of the data signal.

[0091] Specifically, if the first rear-end extension signal point is not an extreme point among the rear-end signal points and the second rear-end extension signal point, it is still necessary to continue to extend backward. The first rear-end extension signal point, the second rear-end extension signal point and the data signal can be input into a preset neural network model, or the first rear-end extension signal point, the second rear-end extension signal point and a preset part of the data signal can be input into the rear-end prediction neural network model. For example, ten signal points are required when the neural network model is preset, and the first rear-end extension signal point, the second rear-end extension signal point and the last eight signal points in the data signal can be input into the rear-end prediction neural network model. The signal point after the output of the second rear-end extension signal point is determined as the third rear-end extension signal point.

[0092] After obtaining the third rear-end extension signal point, it is determined whether there is an extreme point after the rear-end signal point, that is, whether the second rear-end extension signal point is an extreme point among the rear-end signal points, the first rear-end extension signal point and the third rear-end extension signal point. If yes, the extension backward is stopped, and the extension signal after the data signal is composed of the first rear-end extension signal point, the second rear-end extension signal point and the third rear-end extension signal point, that is, the waveform diagram of the signal composed of the rear-end signal point, the first rear-end extension signal point, the second rear-end extension signal point and the third rear-end extension signal point; if not, the extension backward is continued until the extreme point appears.

[0093] The beneficial effect of such arrangement is that the neural network model is used to extend the rear-end point of the data signal, and the extreme point is determined and obtained, which facilitates the EEMD decomposition of the extended signal data, improves the authenticity and reliability of the extended signal by using the neural network to extend the signal, and improves the determination efficiency of the extended signal.

[0094] S405, the data signal and the extension signal are combined to obtain a target signal.

[0095] By way of example, this step can refer to the above step S102, and will not be described in detail.

[0096] S406, the target signal is subjected to ensemble empirical mode decomposition, and the decomposed target signal is denoised to obtain a denoised signal of the target signal.

[0097] By way of example, this step can refer to the above step S103, and will not be described in detail.

[0098] S407, the denoised signal of the data signal is obtained according to the denoised signal of the target signal.

[0099] By way of example, this step can refer to the above step S104, and will not be described in detail.

[0100] This application provides a neural network-based signal denoising method. Using a preset neural network model, it determines the extended signals at the beginning and end of the data signal to be denoised, and combines the extended signals with the data signal to form a target signal. The target signal is then denoised according to a preset EEMD (Extended Emotion Modulation Determination) to obtain a denoised target signal. This denoised target signal is then trimmed to obtain a denoised signal with the same time interval as the data signal, which is then used as the denoised data signal. This method solves the problem of errors in extended signal determination caused by manual determination in existing technologies. By using a neural network model, it effectively saves manpower and time, and by employing the EEMD algorithm, it alleviates mode aliasing, improving the accuracy and efficiency of signal denoising.

[0101] Figure 6 This is a structural block diagram of a signal denoising device based on a neural network, provided in an embodiment of this application. For ease of explanation, only the parts relevant to the embodiments of this disclosure are shown. (Refer to...) Figure 6 The device includes: an extended signal determination module 601, a target signal acquisition module 602, a target signal denoising module 603, and a denoised signal acquisition module 604.

[0102] The extended signal determination module 601 is used to determine the extended signal of the data signal based on the data signal within a preset time period and a preset neural network model; wherein the extended signal includes at least two signal points before the preset time period and at least two signal points after the preset time period, wherein there are extreme points among the at least two signal points before the preset time period and the at least two signal points after the preset time period.

[0103] The target signal acquisition module 602 is used to combine the data signal and the extended signal to obtain the target signal;

[0104] The target signal denoising module 603 is used to perform ensemble empirical mode decomposition on the target signal and denoise the decomposed target signal to obtain the denoised signal of the target signal.

[0105] The denoising signal acquisition module 604 is used to obtain the denoised signal of the data signal based on the denoised signal of the target signal.

[0106] Figure 7 This application provides a structural block diagram of a neural network-based signal denoising device. Figure 6 Based on the illustrated embodiments, as Figure 7 As shown, the extended signal determination module 601 includes an endpoint determination unit 6011, a first extended signal point determination unit 6012, a second extended signal point determination unit 6013, and an extreme point judgment unit 6014.

[0107] The endpoint determination unit 6011 is configured to acquire a data signal in a preset time period, and determine a front-end signal point and a rear-end signal point in the data signal; the front-end signal point is a signal point in the data signal at a preset first time point, and the rear-end signal point is a signal point in the data signal at a preset second time point;

[0108] The first extended signal point determination unit 6012 is configured to input the data signal into a preset neural network model, output a signal point before the front-end signal point as a first front-end extended signal point, and output a signal point after the rear-end signal point as a first rear-end extended signal point.

[0109] The second extended signal point determination unit 6013 is configured to input the data signal, the first front-end extended signal point and the first rear-end extended signal point into a preset neural network model, output a signal point before the first front-end extended signal point as a second front-end extended signal point, and output a signal point after the first rear-end extended signal point as a second rear-end extended signal point.

[0110] The extreme point judgment unit 6014 is configured to determine the first front-end extended signal point, the second front-end extended signal point, the first rear-end extended signal point and the second rear-end extended signal point as the extended signals of the data signal, if the first front-end extended signal point is an extreme point in the front-end signal point and the second front-end extended signal point, and the first rear-end extended signal point is an extreme point in the rear-end signal point and the second rear-end extended signal point.

[0111] In one example, the extended signal determination module 601 further comprises:

[0112] The third front-end extended signal point determination unit is configured to, after outputting a signal point before the first front-end extended signal point as a second front-end extended signal point, if the first front-end extended signal point is not an extreme point in the front-end signal point and the second front-end extended signal point, input the first front-end extended signal point, the second front-end extended signal point and the data signal into a preset neural network model, output a signal point before the second front-end extended signal point as a third front-end extended signal point.

[0113] If the second front-end extended signal point is an extreme point in the front-end signal point, the first front-end extended signal point and the third front-end extended signal point, the first front-end extended signal point, the second front-end extended signal point and the third front-end extended signal point are determined as the extended signals of the data signal.

[0114] In one example, the extension signal determination module 601 further includes:

[0115] The third rear-end extension signal point determination unit is configured to, after outputting the first rear-end extension signal point, input the first rear-end extension signal point, the second rear-end extension signal point and the data signal into a preset neural network model to output a signal point after the second rear-end extension signal point as a third rear-end extension signal point, if the first rear-end extension signal point is not an extreme point among the rear-end signal point and the second rear-end extension signal point.

[0116] If the second rear-end extension signal point is an extreme point among the rear-end signal point, the first rear-end extension signal point and the third rear-end extension signal point, the first rear-end extension signal point, the second rear-end extension signal point and the third rear-end extension signal point are determined as the extension signal of the data signal.

[0117] In one example, the target signal obtaining module 602 is specifically configured to:

[0118] Splice the part of the extension signal before the data signal with the data signal, and splice the part of the extension signal after the data signal with the data signal.

[0119] In one example, the target signal denoising module 603 is specifically configured to:

[0120] Perform ensemble empirical mode decomposition on each signal point in the target signal to obtain an intrinsic mode component of each signal point.

[0121] According to a preset denoising algorithm, the intrinsic mode component of each signal point is denoised.

[0122] According to the time arrangement order of each signal point in the data signal, the denoised intrinsic mode component of each signal point is spliced into a denoised signal of the target signal.

[0123] In one example, the denoised signal obtaining module 604 is specifically configured to:

[0124] Discard the denoised signal corresponding to the extension signal in the denoised signal of the target signal to obtain a denoised signal of the data signal.

[0125] Figure 8 A structural block diagram of an electronic device provided by the embodiments of the present application is shown in FIG. 1. Figure 8 As shown in FIG. 1, the electronic device includes a memory 81 and a processor 82; the memory 81 is configured to store a memory of instructions executable by the processor 82.

[0126] The processor 82 is configured to perform the methods provided in the above embodiments.

[0127] The electronic device also includes a receiver 83 and a transmitter 84. The receiver 83 is used to receive instructions and data sent by other devices, and the transmitter 84 is used to send instructions and data to external devices.

[0128] Figure 9 This is a structural block diagram of an electronic device according to an exemplary embodiment. The device may be a mobile phone, computer, digital broadcasting terminal, messaging device, game console, tablet device, medical device, fitness equipment, personal digital assistant, vehicle, etc.

[0129] Device 900 may include one or more of the following components: processing component 902, memory 904, power supply component 906, multimedia component 908, audio component 910, input / output (I / O) interface 912, sensor component 914, and communication component 916.

[0130] Processing component 902 typically controls the overall operation of device 900, such as operations associated with display, telephone calls, data communication, camera operation, and recording. Processing component 902 may include one or more processors 920 to execute instructions to perform all or part of the steps of the methods described above. Furthermore, processing component 902 may include one or more modules to facilitate interaction between processing component 902 and other components. For example, processing component 902 may include a multimedia module to facilitate interaction between multimedia component 908 and processing component 902.

[0131] Memory 904 is configured to store various types of data to support the operation of device 900. Examples of this data include instructions for any application or method operating on device 900, contact data, phonebook data, messages, pictures, videos, etc. Memory 904 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0132] Power supply component 906 provides power to various components of device 900. Power supply component 906 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to device 900.

[0133] The multimedia component 908 includes a screen providing an output interface between the device 900 and a user. In some embodiments, the screen can include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen can be implemented as a touch screen to receive input signals from a user. The touch panel includes one or more touch sensors to sense touch, swiping, and gestures on the touch panel. The touch sensors can not only sense a boundary of a touching or swiping action, but also detect duration and pressure related to the touching or swiping action. In some embodiments, the multimedia component 908 includes a front camera and / or a rear camera. When the device 900 is in an operation mode, such as a shooting mode or a video mode, the front camera and / or the rear camera can receive external multimedia data. Each of the front and rear camera can be a fixed optical lens system or have a focal length and optical zooming capability.

[0134] The audio component 910 is configured to output and / or input audio signals. For example, the audio component 910 includes a microphone (MIC) to receive an external audio signal when the device 900 is in an operation mode, such as a call mode, a recording mode, and a voice recognition mode. The received audio signal can be further stored in the memory 904 or transmitted via the communication component 916. In some embodiments, the audio component 910 also includes a speaker to output audio signals.

[0135] The I / O interface 912 provides an interface between the processing component 902 and peripheral interface modules, which can be a keypad, a click wheel, buttons, and the like. The buttons can include, but are not limited to, a home button, a volume button, a start button, and a lock button.

[0136] The sensor component 914 includes one or more sensors to provide various state assessments for the device 900. For example, the sensor component 914 can detect an open / closed state of the device 900, relative positioning of components, such as a display and a keypad of the device 900, a change in position of the device 900 or a component of the device 900, presence or absence of user contact with the device 900, a change in orientation of the device 900 or acceleration / deceleration of the device 900, and a temperature change of the device 900. The sensor component 914 can include a proximity sensor configured to detect presence of a nearby object without any physical touch. The sensor component 914 can further include a light sensor, such as a CMOS or CCD image sensor, for use in an imaging application. In some embodiments, the sensor component 914 can further include an acceleration sensor, a gyroscope sensor, a magnetic sensor, a pressure sensor, or a temperature sensor.

[0137] The communication component 916 is configured to facilitate wired or wireless communication between the device 900 and other devices. The device 900 can access a wireless network based on a communication standard, such as WiFi, 2G, or 3G, or a combination thereof. In an exemplary embodiment, the communication component 916 receives a broadcast signal or broadcast related information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component 916 also includes a Near Field Communication (NFC) module to facilitate short-range communication. For example, the NFC module can be implemented based on Radio Frequency Identification (RFID) technology, Infrared Data Association (IrDA) technology, Ultra-WideBand (UWB) technology, Bluetooth (BT) technology and other technologies.

[0138] In an exemplary embodiment, the device 900 can be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, micro-controllers, microprocessors or other electronic elements, for executing the above-described methods.

[0139] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions, such as the memory 904 including instructions, is also provided, which can be executed by the processor 920 of the device 900 to complete the above-described methods. For example, the non-transitory computer-readable storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disc, and an optical data storage device, etc.

[0140] A non-transitory computer-readable storage medium, when instructions in the storage medium are executed by a processor of a terminal device, enables the terminal device to perform the above-described neural network-based signal denoising method of the terminal device.

[0141] The present application also discloses a computer program product, including a computer program, which, when executed by a processor, implements the method as described in the embodiments.

[0142] The various embodiments of the systems and techniques described above can be implemented in digital electronic circuitry, integrated circuitry, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system on a chip systems (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include implementation in one or more computer programs that are executable and / or interpretable on a programmable system including at least one programmable processor, which can be special or general purpose, coupled to receive data and instructions from, and to transmit data and instructions to, a storage system, at least one input device, and at least one output device.

[0143] Program code for carrying out methods of the present application can be written in any combination of one or more programming languages. This program code can be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the program code, when executed by the processor or controller, produces a means for implementing the functions / acts specified in the flowcharts and / or block diagrams. The program code can be executed entirely on a machine, partially on a machine, partially on a machine as a stand-alone software package, or entirely on a remote machine or electronic device.

[0144] In the context of the present application, a machine-readable medium can be a tangible medium that contains or stores a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include but is not limited to an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of the machine-readable storage medium will include one or more lines of electrical connections, portable computer disks, hard disk drives, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or Flash memory), optical fibers, portable compact disc read-only memories (CD-ROMs), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0145] To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the computer. Other kinds of devices can be used to provide for interaction with a user as well; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form, including acoustic, speech, or tactile input.

[0146] The systems and techniques described here can be implemented in a computing system that includes a back end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front end component (e.g., a client computer having a graphical user interface or a Web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back end, middleware, or front end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), and the Internet.

[0147] The computer system can include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server can be a cloud server, also known as cloud computing server or cloud host, which is a host product in the cloud computing service system, and solves the defects of large management difficulty and weak business scalability in traditional physical host and VPS (Virtual Private Server, or VPS for short) services. The server can also be a distributed system server or a server combined with a blockchain. It should be understood that the various forms of procedures shown above can be used to reorder, add, or delete steps. For example, the steps described in this application can be executed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this application can be achieved, and this document does not limit this.

[0148] Other embodiments of the application will be apparent to those skilled in the art from consideration of the specification and practice of the application disclosed herein. It is intended that the specification and examples be considered as exemplary only, with the true scope and spirit of the application being indicated by the following claims.

[0149] It is to be understood that the application is not limited to the precise construction herein disclosed and shown in the drawings, and that various modifications and changes can be made by those skilled in the art without departing from the scope of the application. The scope of the application is limited only by the claims that follow.

Claims

1. A signal denoising method based on neural networks, characterized in that, include: Based on the data signal within a preset time period and a preset neural network model, the extended signal of the data signal is determined. Take data signals within a preset time period and determine the front-end signal point and the back-end signal point in the data signal; wherein, the front-end signal point is the signal point in the data signal located at a preset first time point, and the back-end signal point is the signal point in the data signal located at a preset second time point; The data signal is input into a preset neural network model, and a signal point before the front-end signal point is output, which is the first front-end extension signal point, and a signal point after the rear-end signal point is output, which is the first rear-end extension signal point. The data signal, the first front-end extension signal point, and the first back-end extension signal point are input into a preset neural network model, and a signal point before the first front-end extension signal point is output as the second front-end extension signal point, and a signal point after the first back-end extension signal point is output as the second back-end extension signal point. If the first front-end extended signal point is an extreme point among the front-end signal point and the second front-end extended signal point, and the first rear-end extended signal point is an extreme point among the rear-end signal point and the second rear-end extended signal point, then the first front-end extended signal point, the second front-end extended signal point, the first rear-end extended signal point, and the second rear-end extended signal point are determined to be extended signals of the data signal. The data signal and the extended signal are combined to obtain the target signal; The target signal is subjected to ensemble empirical mode decomposition, and the decomposed target signal is then denoised to obtain the denoised target signal. The denoised signal of the data signal is obtained based on the denoised signal of the target signal.

2. The method according to claim 1, characterized in that, The signal point preceding the first front-end extension signal point, and following the second front-end extension signal point, also includes: If the first front-end extended signal point is not an extreme point among the front-end signal point and the second front-end extended signal point, then the first front-end extended signal point, the second front-end extended signal point and the data signal are input into a preset neural network model, and the signal point before the second front-end extended signal point is output as the third front-end extended signal point. If the second front-end extended signal point is an extreme point among the front-end signal point, the first front-end extended signal point, and the third front-end extended signal point, then the first front-end extended signal point, the second front-end extended signal point, and the third front-end extended signal point are determined to be extended signals of the data signal.

3. The method according to claim 1, characterized in that, After the output of the first rear-end extension signal point, and after the second rear-end extension signal point, the following is also included: If the first back-end extension signal point is not an extreme point among the back-end signal point and the second back-end extension signal point, then the first back-end extension signal point, the second back-end extension signal point and the data signal are input into a preset neural network model, and a signal point after the second back-end extension signal point is output as the third back-end extension signal point. If the second back-end extended signal point is an extreme point among the back-end signal point, the first back-end extended signal point, and the third back-end extended signal point, then the first back-end extended signal point, the second back-end extended signal point, and the third back-end extended signal point are determined to be extended signals of the data signal.

4. The method according to claim 1, characterized in that, Combining the data signal and the extended signal includes: The portion of the extended signal preceding the data signal is concatenated with the data signal, and the portion of the extended signal following the data signal is concatenated with the data signal.

5. The method according to claim 1, characterized in that, The target signal is subjected to ensemble empirical mode decomposition, and the decomposed target signal is then denoised to obtain the denoised target signal, including: Perform ensemble empirical mode decomposition on each signal point in the target signal to obtain the intrinsic mode components of each signal point; According to the preset denoising algorithm, the intrinsic mode components of each signal point are denoised; Based on the time arrangement order of each signal point in the data signal, the denoised intrinsic mode components of each signal point are spliced ​​together to form the denoised signal of the target signal.

6. The method according to claim 1, characterized in that, The process of obtaining the denoised signal of the data signal based on the denoised signal of the target signal includes: The denoised signal corresponding to the extended signal is discarded from the denoised signal of the target signal to obtain the denoised signal of the data signal.

7. A signal denoising device based on a neural network, characterized in that, include: An extended signal determination module takes data signals within a preset time period and determines the front-end signal point and the back-end signal point in the data signals; wherein, the front-end signal point is a signal point in the data signals located at a preset first time point, and the back-end signal point is a signal point in the data signals located at a preset second time point; the data signals are input into a preset neural network model, and a signal point before the front-end signal point is output as a first front-end extended signal point, and a signal point after the back-end signal point is output as a first back-end extended signal point; the data signals, the first front-end extended signal point, and the first back-end extended signal point are input... The signal is fed into a preset neural network model, and a signal point before the first front-end extended signal point is output as the second front-end extended signal point, and a signal point after the first back-end extended signal point is output as the second back-end extended signal point; if the first front-end extended signal point is an extreme point among the front-end signal point and the second front-end extended signal point, and the first back-end extended signal point is an extreme point among the back-end signal point and the second back-end extended signal point, then the first front-end extended signal point, the second front-end extended signal point, the first back-end extended signal point, and the second back-end extended signal point are determined to be extended signals of the data signal; A target signal acquisition module is used to combine the data signal and the extended signal to obtain a target signal; The target signal denoising module is used to perform ensemble empirical mode decomposition on the target signal and denoise the decomposed target signal to obtain the denoised signal of the target signal. The denoising signal acquisition module is used to obtain the denoised signal of the data signal based on the denoised signal of the target signal.

8. An electronic device, characterized in that, include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory to implement the neural network-based signal denoising method as described in any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the neural network-based signal denoising method as described in any one of claims 1-6.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, is used to implement the neural network-based signal denoising method according to any one of claims 1 to 6.

Images