A signal denoising method and device based on DE optimization of wavelet parameters

Abstract

The application discloses a signal denoising method and device based on DE optimization of wavelet parameters. The method comprises the following steps: selecting an initial wavelet base function, performing wavelet decomposition on a noisy signal according to an initial decomposition layer number and initial layer thresholds at each decomposition scale to obtain detail coefficients at each decomposition scale and approximation coefficients at a maximum decomposition scale; taking the initial wavelet base function, the initial decomposition layer number and the initial layer thresholds at each decomposition scale as position parameters of an initial population in a DE algorithm, and obtaining optimal wavelet base functions, optimal decomposition layer numbers and optimal layer thresholds at each decomposition scale based on the DE algorithm; performing denoising processing on each detail coefficient according to the optimal wavelet base functions, the optimal decomposition layer numbers and the optimal layer thresholds at each decomposition scale to obtain target detail coefficients, and combining the target detail coefficients and the approximation coefficients to perform signal reconstruction to obtain a target signal. The application can optimize the decomposition layer number and the layer threshold by using the DE algorithm, and improve the denoising effect.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of signal processing, in particular to a signal denoising method and device based on DE optimized wavelet parameters. BACKGROUND

[0002] With the development of information technology, people have higher and higher requirements for the communication quality of wireless communication. Wireless communication is signal transmission through the atmosphere channel, which is easily affected by external interference during transmission, resulting in a decline in communication quality, and thus the received noisy signal needs to be denoised.

[0003] In 1994, Donoho and Johnstone proposed the idea of wavelet threshold denoising, which is widely used in signal denoising due to its multi-resolution, low entropy and decorrelation. The wavelet threshold denoising method can be used to identify noise in the signal and suppress high-frequency noise to reconstruct the signal. Although Donoho et al. have theoretically proved and found the optimal universal threshold, in the wavelet transform, the amplitude of the noise wavelet coefficient decreases with the increase of the scale, while the amplitude of the signal wavelet transform coefficient does not decrease with the increase of the scale. Therefore, the effect of the universal threshold in practical application is not very ideal. In view of this, when applying the wavelet threshold denoising method, a hierarchical threshold can be used, that is, different thresholds are taken at each decomposition scale to overcome the technical defects of the universal threshold. The size of each hierarchical threshold is the core of the thresholding process. If the threshold is too large, part of the noise cannot be filtered out, and the denoising effect is poor. If the threshold is too small, the effective signal will be filtered out, resulting in signal distortion. Therefore, how to accurately determine the hierarchical threshold and the decomposition scale of the signal in the denoising process according to different signal characteristics is currently a problem to be considered. SUMMARY

[0004] In order to overcome the defects of the prior art, the present application provides a signal denoising method and device based on DE optimized wavelet parameters, which can optimize the decomposition layer number and hierarchical threshold using the DE algorithm to improve the denoising effect.

[0005] To solve the above technical problems, in a first aspect, an embodiment of the present application provides a signal denoising method based on DE optimized wavelet parameters, comprising:

[0006] Selecting an initial wavelet basis function, wavelet decomposing the noisy signal according to an initial decomposition layer number and an initial hierarchical threshold at each decomposition scale to obtain detail coefficients at each decomposition scale and approximation coefficients at the maximum decomposition scale;

[0007] Taking the initial wavelet basis function, the initial decomposition layer number and the initial hierarchical threshold at each decomposition scale as the position parameters of the initial population in the DE algorithm, and obtaining the optimal wavelet basis function, the optimal decomposition layer number and the optimal hierarchical threshold at each decomposition scale based on the DE algorithm.

[0008] According to the optimal wavelet basis function, the optimal decomposition layer number and the optimal layered threshold value under each decomposition scale, each of the detail coefficients is denoised to obtain a target detail coefficient, and a target signal is obtained by combining the target detail coefficient and the approximation coefficient.

[0009] Further, the optimal wavelet basis function, the optimal decomposition layer number and the optimal layered threshold value under each decomposition scale are obtained based on the DE algorithm, and specifically:

[0010] Parameter initialization: setting the population size, the maximum number of iterations, the fitness function, and defining each individual in the population to represent the layered threshold value under each decomposition scale;

[0011] According to the fitness function, the fitness value of each individual in the population is calculated, and when the current number of iterations reaches the maximum number of iterations, the population with the minimum fitness value is selected as the optimal wavelet basis function, the optimal decomposition layer number and the optimal layered threshold value under each decomposition scale according to all the fitness values;

[0012] When the current number of iterations does not reach the maximum number of iterations, the mutation individuals and the original individuals in the population are subjected to a crossover operation, and the mutation individuals and / or the original individuals with a fitness value less than a preset fitness value are used as the next generation population to perform a differential evolution operation in a loop until the optimal wavelet basis function, the optimal decomposition layer number and the optimal layered threshold value under each decomposition scale are obtained when the maximum number of iterations is reached or the fitness function requirement is met.

[0013] Further, before the crossover operation of the mutation individuals and the original individuals in the population, the method further comprises:

[0014] Based on a predefined differential strategy, the original individuals in the population are subjected to a mutation operation to obtain the mutation individuals.

[0015] Further, the crossover operation of the mutation individuals and the original individuals in the population is specifically:

[0016] The mutation individuals and the original individuals in the population are subjected to a crossover operation through a binomial crossover model.

[0017] Further, the fitness function is:

[0018]

[0019] wherein MAX is the maximum amplitude of the input signal, RMSE is the root mean square error of the noisy signal, N is the length of the noisy signal, and w(k) is the wavelet coefficient of the input signal. wavelet coefficients after threshold processing, λ represents the hierarchical threshold at each decomposition scale.

[0020] In a second aspect, an embodiment of the present application provides a signal denoising device based on DE-optimized wavelet parameters, comprising:

[0021] a wavelet decomposition module configured to select an initial wavelet basis function, perform wavelet decomposition on a noisy signal according to an initial decomposition level and initial hierarchical thresholds at each decomposition scale, and obtain detail coefficients at each decomposition scale and approximation coefficients at a maximum decomposition scale;

[0022] a parameter optimization module configured to take the initial wavelet basis function, the initial decomposition level and the initial hierarchical thresholds at each decomposition scale as position parameters of an initial population in a DE algorithm, and obtain an optimal wavelet basis function, an optimal decomposition level and optimal hierarchical thresholds at each decomposition scale based on the DE algorithm;

[0023] a signal denoising module configured to perform denoising processing on each of the detail coefficients according to the optimal wavelet basis function, the optimal decomposition level and the optimal hierarchical thresholds at each decomposition scale, obtain target detail coefficients, and reconstruct a target signal by combining the target detail coefficients and the approximation coefficients.

[0024] Further, the DE algorithm-based obtaining of the optimal wavelet basis function, the optimal decomposition level and the optimal hierarchical thresholds at each decomposition scale is specifically:

[0025] parameter initialization: setting a population size, a maximum iteration number, a fitness function, and defining each individual in the population to represent hierarchical thresholds at each decomposition scale;

[0026] calculating fitness values of each individual in the population according to the fitness function, and selecting a population with the smallest fitness value as the optimal wavelet basis function, the optimal decomposition level and the optimal hierarchical thresholds at each decomposition scale when the current iteration number reaches the maximum iteration number;

[0027] when the current iteration number does not reach the maximum iteration number, performing a crossover operation on a mutated individual and an original individual in the population, taking the mutated individual and / or the original individual with a fitness value less than a preset fitness value as a next generation population, and performing a differential evolution operation in a loop until the optimal wavelet basis function, the optimal decomposition level and the optimal hierarchical thresholds at each decomposition scale are obtained when the maximum iteration number is reached or the fitness function requirement is met.

[0028] Further, before the crossover operation is performed on the mutation individuals and the original individuals in the population, the method further comprises:

[0029] The original individuals in the population are subjected to a mutation operation based on a predefined difference strategy, to obtain the mutation individuals.

[0030] Further, the crossover operation is performed on the mutation individuals and the original individuals in the population, specifically as follows:

[0031] The mutation individuals and the original individuals in the population are subjected to a crossover operation through a binomial crossover model.

[0032] Further, the fitness function is as follows:

[0033]

[0034] wherein MAX is the maximum amplitude of the input signal, RMSE is the root mean square error of the noisy signal, N is the length of the noisy signal, and w(k) is the wavelet coefficient of the input signal, is the wavelet coefficient after threshold processing, λ represents the hierarchical threshold at each decomposition scale.

[0035] The embodiments of the present application have the following beneficial effects:

[0036] By selecting an initial wavelet basis function, according to an initial decomposition layer number and initial hierarchical thresholds at each decomposition scale, the noisy signal is subjected to wavelet decomposition to obtain detail coefficients at each decomposition scale and approximation coefficients at a maximum decomposition scale, the initial wavelet basis function, the initial decomposition layer number and the initial hierarchical thresholds at each decomposition scale are taken as position parameters of an initial population in the DE algorithm, the optimal wavelet basis function, the optimal decomposition layer number and the optimal hierarchical thresholds at each decomposition scale are obtained based on the DE algorithm, the each detail coefficient is subjected to denoising processing according to the optimal wavelet basis function, the optimal decomposition layer number and the optimal hierarchical thresholds at each decomposition scale, to obtain a target detail coefficient, and the target signal is obtained by signal reconstruction combining the target detail coefficient and the approximation coefficient, to complete signal denoising. Compared with the prior art, the embodiments of the present application obtain the optimal wavelet basis function, the optimal decomposition layer number and the optimal hierarchical thresholds at each decomposition scale based on the DE algorithm, and utilize the optimal wavelet basis function, the optimal decomposition layer number and the optimal hierarchical thresholds at each decomposition scale to perform denoising processing on the noisy signal based on the discrete wavelet transform, so that the decomposition layer number and the hierarchical thresholds can be optimized by the DE algorithm, and the denoising effect is improved. BRIEF DESCRIPTION OF DRAWINGS

[0037] Figure 1 FIG. 1 is a flowchart of a signal denoising method based on DE optimized wavelet parameters according to an embodiment of the present application;

[0038] Figure 2 A flowchart of a signal denoising method based on DE optimization of wavelet parameters for the application exemplified in the first embodiment of the present application;

[0039] Figure 3 A flowchart of obtaining optimal decomposition levels and optimal hierarchical thresholds under each decomposition scale based on DE algorithm for the application exemplified in the first embodiment of the present application;

[0040] Figure 4 A signal diagram of a noisy signal for the application exemplified in the first embodiment of the present application;

[0041] Figure 5 A signal diagram of a target signal for the application exemplified in the first embodiment of the present application;

[0042] Figure 6 A structural diagram of a signal denoising device based on DE optimization of wavelet parameters for the second embodiment of the present application. DETAILED DESCRIPTION

[0043] The technical solutions in the present application will be described clearly and completely below with reference to the drawings in the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but 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 fall within the scope of protection of the present application.

[0044] It should be noted that the step numbers in the text are only for the convenience of explaining the specific embodiments, and do not serve as the function of limiting the execution sequence of the steps. The method provided in the present embodiment can be executed by the related terminal device, and the processor is taken as an example for description hereinafter.

[0045] As shown in Figure 1 The first embodiment provides a signal denoising method based on DE optimization of wavelet parameters, including steps S1-S3:

[0046] S1, selecting an initial wavelet basis function, and performing wavelet decomposition on a noisy signal according to an initial decomposition level and initial hierarchical thresholds under each decomposition scale to obtain detail coefficients under each decomposition scale and approximation coefficients under the maximum decomposition scale;

[0047] S2, taking the initial wavelet basis function, the initial decomposition level and the initial hierarchical thresholds under each decomposition scale as the position parameters of the initial population in the DE algorithm, and obtaining optimal wavelet basis function, optimal decomposition level and optimal hierarchical thresholds under each decomposition scale based on the DE algorithm;

[0048] S3. Based on the optimal wavelet basis function, the optimal number of decomposition levels, and the optimal layering threshold at each decomposition scale, denoise each detail coefficient to obtain the target detail coefficients. Then, combine the target detail coefficients and approximation coefficients to reconstruct the signal and obtain the target signal.

[0049] Differential Evolution (DE) solves optimization problems through cooperation and competition among individuals within a population, exhibiting strong global convergence capabilities. The process of DE is similar to other evolutionary algorithms, involving mutation, crossover, and selection operations. However, compared to other algorithms, DE is more stable, converges faster, and has lower complexity.

[0050] like Figure 2 As shown, as an example, in step S1, an initial wavelet basis function is selected from db1, sym1, and coif1. Other types of wavelet basis functions can also be selected as the initial wavelet basis function. Based on the initial decomposition level and the initial layer threshold at each decomposition scale, the noisy signal is decomposed by wavelet to obtain the detail coefficients at each decomposition scale and the approximation coefficients at the maximum decomposition scale.

[0051] In step S2, the initial wavelet basis function, the initial number of decomposition layers, and the initial stratification threshold at each decomposition scale are used as the position parameters of the initial population in the DE algorithm. The optimal wavelet basis function and the optimal number of decomposition layers are obtained based on the DE algorithm. Then, the stratification threshold is optimized based on the DE algorithm to obtain the optimal stratification threshold at each decomposition scale.

[0052] In step S3, the optimal wavelet basis function is selected, and the detail coefficients at each decomposition scale are denoised according to the optimal number of decomposition levels and the optimal layer threshold at each decomposition scale to obtain the target detail coefficients. The target detail coefficients and approximation coefficients are then combined to reconstruct the signal and obtain the target signal, thus completing the signal denoising.

[0053] This embodiment obtains the optimal wavelet basis function, the optimal number of decomposition levels, and the optimal stratification thresholds at each decomposition scale based on the DE algorithm. Based on discrete wavelet transform, it uses the optimal wavelet basis function, the optimal number of decomposition levels, and the optimal stratification thresholds at each decomposition scale to denoise the noisy signal. Thus, the DE algorithm can be used to optimize the number of decomposition levels and stratification thresholds, thereby improving the denoising effect.

[0054] In the preferred embodiment, the DE algorithm-based optimal wavelet basis function, optimal decomposition layer number and optimal hierarchical threshold at each decomposition scale are obtained, specifically: parameter initialization: setting the population size, maximum iteration number, fitness function, and defining each individual in the population representing the hierarchical threshold at each decomposition scale; calculating the fitness value of each individual in the population according to the fitness function, and when the current iteration number reaches the maximum iteration number, selecting the population with the smallest fitness value as the optimal wavelet basis function, optimal decomposition layer number and optimal hierarchical threshold at each decomposition scale according to all fitness values; when the current iteration number does not reach the maximum iteration number, performing crossover operation on the mutated individuals and original individuals in the population, and taking the mutated individuals and / or original individuals with fitness values less than the preset fitness value as the next generation population, to perform differential evolution operation in a loop until the maximum iteration number is reached or the fitness function requirement is met, to obtain the optimal wavelet basis function, optimal decomposition layer number and optimal hierarchical threshold at each decomposition scale.

[0055] As shown in Figure 3 , as an example, the DE algorithm-based optimal wavelet basis function, optimal decomposition layer number and optimal hierarchical threshold at each decomposition scale are obtained, specifically including the steps of:

[0056] S2-1, parameter initialization: setting the population size NP, maximum iteration number G, scaling factor F, crossover operator CR, fitness function f(x), and defining each individual X i in the population representing the hierarchical threshold at each scale, and the population dimension is D;

[0057] S2-2, calculating the individual fitness value, judging whether the iteration termination condition is met, if the current iteration number reaches the maximum iteration number, determining that the iteration termination condition is met, and taking the population with the smallest current fitness value as the optimal wavelet basis function, optimal decomposition layer number and optimal hierarchical threshold at each decomposition scale, otherwise, continuing to step S2-3;

[0058] S2-3, performing mutation operation on the NP individuals in the population to generate differential individuals and mutated individuals;

[0059] S2-4, in order to improve the population diversity, performing crossover operation on the mutated individuals generated in S2-3 and the original individuals in the population through the binomial crossover model;

[0060] S2-5, arranging the individuals according to the order from small to large fitness value, selecting the individuals with smaller fitness value into the next generation of population to ensure the evolution of the population to the optimal solution;

[0061] S2-6, continue to loop the differential evolution operation by using the optimal learning search equation until the iteration termination condition is met, that is, when the current iteration number reaches the maximum iteration number, the population with the minimum fitness value is taken as the optimal wavelet basis function, the optimal decomposition layer number and the optimal hierarchical threshold under each decomposition scale.

[0062] The implementation process of initializing the population size NP in step S2-1 is as follows:

[0063] S2-11, initialize the wavelet basis function, and set the wavelet basis function to db1, sym1 and coif2, that is:

[0064] r1 = round(1 + 1.9 * r(1)) (1);

[0065] Wherein, r1 is the wavelet basis function, and round() is the rounding operation;

[0066] S2-12, initialize the decomposition layer number, and set the decomposition layer number in the range of 3 to 8 layers, that is:

[0067] r2 = round(3 + 5.9 * r(3)) (2);

[0068] Wherein, r2 is the decomposition layer number;

[0069] S2-13, initialize the hierarchical threshold, that is:

[0070] rp = r(3:2+r2) (3);

[0071] Wherein, rp is the hierarchical threshold.

[0072] In the preferred embodiment, before the crossover operation of the mutation individuals and the original individuals in the population, the method further comprises: based on the pre-defined differential strategy, performing mutation operation on the original individuals in the population to obtain the mutation individuals.

[0073] For example, based on the pre-defined differential strategy, the mutation operation is performed on the original individuals in the population, that is:

[0074]

[0075]

[0076] Wherein, are three different vectors randomly selected from the population, wherein r1, r2, i ∈ NP; is a differential vector, F is a scaling factor of the DE algorithm, generally between [0, 1], which controls the amplification effect of the differential vector; V i t represents the mutation vector.

[0077] In a preferred embodiment, the crossover operation between the mutated individuals and the original individuals in the population specifically involves performing a crossover operation between the mutated individuals and the original individuals in the population using a binomial crossover model.

[0078] As an example, to improve population diversity, the binomial crossover model is operated as follows:

[0079]

[0080] in, It is the experimental vector The j-th gene segment, It is a mutation vector The j-th gene segment, randb is a random number between [0,1]; CR is the crossover operator, which controls the binomial crossover process, and is usually controlled between [0,1]. The larger the CR value, the greater the probability of hybridization; j rand It is a randomly generated number within [1, D], which can be guaranteed The jth rand The position is from This is obtained by ensuring that the experimental vector and the target vector mutated vector are different, thus avoiding invalid hybridization.

[0081] In a preferred embodiment, the fitness function is:

[0082]

[0083] Where MAX is the maximum amplitude of the input signal, and RMSE is the root mean square error of the noisy signal. N is the length of the noisy signal, and w(k) is the wavelet coefficient of the input signal. These are the wavelet coefficients after thresholding. λ represents the stratification threshold at each decomposition scale.

[0084] It is understandable that the signal length is the same as the number of signal sampling points.

[0085] As an example, the optimal target vector or experimental vector is selected through the fitness function f(x) as the individual to enter the next generation of the population, ensuring that the population evolves towards the optimal solution. The process is as follows:

[0086]

[0087] After completing this step, the differential evolution operation continues to be performed repeatedly until the number of iterations or the fitness function requirement is met, at which point the algorithm terminates.

[0088] Optionally, the fitness function can be calculated using the following formula:

[0089]

[0090] wherein SNR represents the signal-to-noise ratio of the noisy signal, δ is an additional factor, mean() is a signal mean value operation, U is the voltage amplitude of the noisy signal, and R is the resistance.

[0091] In step S3, the noisy signal is decomposed by wavelet using the decomposition scale of the individual selected in step S2, i.e., the optimal decomposition level, and the discrete wavelet can be expressed as:

[0092]

[0093] wherein a is a scale parameter, and b is a translation parameter, is discretized. At this time, j is an integer value, and a0 is a fixed scaling step greater than 1. For b, take wherein b0>0 and is related to the specific form of the wavelet ψ(t), and k is an integer.

[0094] The discretized wavelet coefficient can be expressed as:

[0095]

[0096] wherein s(n) is the noisy signal.

[0097] The wavelet threshold denoising utilizes the characteristic that noise is mainly distributed in the high frequency part, and through wavelet transform, the received signal is decomposed into a useful signal and a noisy signal. Then, a suitable wavelet basis function is selected to decompose the noisy signal to obtain the approximation coefficient and the detail coefficient. The detail coefficient obtained by wavelet decomposition is processed by threshold to realize the denoising of the signal, X i i.e., the optimal threshold under different decomposition scales. The detail coefficient processed by the threshold can form new approximation coefficients and detail coefficients, and the wavelet coefficient after the optimal decomposition scale is used to reconstruct the denoised signal. The reconstruction process is as follows:

[0098]

[0099] wherein C is a constant irrelevant to the signal.

[0100] Optionally, the wavelet basis of the wavelet reconstruction is determined by algorithm optimization.

[0101] The signal-to-noise ratio can effectively reflect the denoising effect of the target signal and the deviation from the noisy signal, and the larger the signal-to-noise ratio, the better the denoising effect.

[0102] In order to verify the effectiveness of the present application, the signal-to-noise ratio is used as the evaluation index of the denoising performance, which can be calculated by the ratio of signal power to noise power, and the calculation formula is as follows:

[0103]

[0104] The signal-to-noise ratios of the noisy signal and the target signal are shown in Table 1:

[0105] Table 1 Comparison of signal-to-noise ratios before and after denoising (dB)

[0106] Before denoising After denoising 23.543 31.109

[0107] The signal schematic diagrams of the noisy signal and the target signal are shown in Figure 4 , 5 It can be seen from Figure 4 , 5 that the signal denoising method based on DE optimized wavelet parameters provided in the embodiment can effectively improve the denoising effect.

[0108] Based on the same technical concept as the first embodiment, the second embodiment provides a signal denoising device based on DE optimized wavelet parameters as shown in Figure 6 , which comprises: a wavelet decomposition module, configured to select an initial wavelet basis function, perform wavelet decomposition on a noisy signal according to an initial decomposition layer number and initial hierarchical thresholds at each decomposition scale, to obtain detail coefficients at each decomposition scale and approximation coefficients at the maximum decomposition scale; a parameter optimization module, configured to take the initial wavelet basis function, the initial decomposition layer number and the initial hierarchical thresholds at each decomposition scale as the position parameters of the initial population in the DE algorithm, and obtain an optimal wavelet basis function, an optimal decomposition layer number and optimal hierarchical thresholds at each decomposition scale based on the DE algorithm; and a signal denoising module, configured to perform denoising processing on each detail coefficient according to the optimal wavelet basis function, the optimal decomposition layer number and the optimal hierarchical thresholds at each decomposition scale to obtain target detail coefficients, and perform signal reconstruction by combining the target detail coefficients and the approximation coefficients to obtain a target signal.

[0109] In the preferred embodiments, the DE algorithm-based method obtains the optimal wavelet basis function, the optimal decomposition layer number and the optimal hierarchical threshold at each decomposition scale, and specifically comprises: parameter initialization: setting the population size, the maximum iteration number, the fitness function, and defining each individual in the population to represent the hierarchical threshold at each decomposition scale; calculating the fitness value of each individual in the population according to the fitness function, and when the current iteration number reaches the maximum iteration number, selecting the population with the minimum fitness value as the optimal wavelet basis function, the optimal decomposition layer number and the optimal hierarchical threshold at each decomposition scale according to all the fitness values; when the current iteration number does not reach the maximum iteration number, performing a crossover operation on the mutated individuals and the original individuals in the population, and taking the mutated individuals and / or the original individuals with a fitness value less than a preset fitness value as the next generation population, so as to perform the differential evolution operation in a loop until the optimal wavelet basis function, the optimal decomposition layer number and the optimal hierarchical threshold at each decomposition scale are obtained when the maximum iteration number is reached or the fitness function requirement is met.

[0110] In the preferred embodiments, before the crossover operation on the mutated individuals and the original individuals in the population, the method further comprises: performing a mutation operation on the original individuals in the population based on a predefined differential strategy to obtain the mutated individuals.

[0111] In the preferred embodiments, the crossover operation on the mutated individuals and the original individuals in the population specifically comprises: performing a crossover operation on the mutated individuals and the original individuals in the population through a binomial crossover model.

[0112] In the preferred embodiments, the fitness function is:

[0113]

[0114] wherein, MAX is the maximum amplitude of the input signal, RMSE is the root mean square error of the noisy signal, N is the length of the noisy signal, w(k) is the wavelet coefficient of the input signal, is the wavelet coefficient after threshold processing, λ represents the hierarchical threshold at each decomposition scale.

[0115] In summary, the embodiments of the present application have the following beneficial effects:

[0116] By selecting an initial wavelet base function, according to the initial decomposition layer and the initial hierarchical threshold at each decomposition scale, the noisy signal is decomposed by wavelet to obtain the detail coefficient at each decomposition scale and the approximate coefficient at the maximum decomposition scale, the initial wavelet base function, the initial decomposition layer and the initial hierarchical threshold at each decomposition scale are taken as the position parameters of the initial population in the DE algorithm, the optimal wavelet base function, the optimal decomposition layer and the optimal hierarchical threshold at each decomposition scale are obtained based on the DE algorithm, the optimal wavelet base function, the optimal decomposition layer and the optimal hierarchical threshold at each decomposition scale are used for denoising processing of each detail coefficient to obtain the target detail coefficient, and the target signal is obtained by combining the target detail coefficient and the approximate coefficient, and the signal denoising is completed. The embodiments of the present application obtain the optimal wavelet base function, the optimal decomposition layer and the optimal hierarchical threshold at each decomposition scale based on the DE algorithm, and use the optimal wavelet base function, the optimal decomposition layer and the optimal hierarchical threshold at each decomposition scale to denoise the noisy signal based on the discrete wavelet transform, so that the decomposition layer and the hierarchical threshold can be optimized by using the DE algorithm, and the denoising effect is improved.

[0117] The above is the preferred embodiment of the present application, it should be noted that for those skilled in the art, without departing from the principles of the present application, a number of improvements and refinements can be made, these improvements and refinements are also considered to be within the scope of the present application.

[0118] Those skilled in the art can understand that all or part of the processes in the above embodiments can be completed by a computer program instructing related hardware, the program can be stored in a computer readable storage medium, and the program can include the processes of the above embodiments when executed. The storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM).

Claims

1. A signal denoising method based on DE-optimized wavelet parameters, characterized in that, include: Select an initial wavelet basis function, and perform wavelet decomposition on the noisy signal according to the initial decomposition level and the initial layer threshold at each decomposition scale to obtain the detail coefficients at each decomposition scale and the approximation coefficients at the maximum decomposition scale. The initial wavelet basis function, the initial number of decomposition layers, and the initial stratification thresholds at each decomposition scale are used as the position parameters of the initial population in the DE algorithm. Based on the DE algorithm, the optimal wavelet basis function, the optimal number of decomposition layers, and the optimal stratification thresholds at each decomposition scale are obtained. The method for obtaining the optimal wavelet basis function, optimal decomposition level, and optimal stratification threshold at each decomposition scale based on the DE algorithm specifically involves: parameter initialization: setting the population size, maximum number of iterations, fitness function, and defining each individual in the population as representing the stratification threshold at each decomposition scale; calculating the fitness value of each individual in the population according to the fitness function, and when the current iteration count reaches the maximum iteration count, selecting the population with the smallest fitness value as the optimal wavelet basis function, the optimal number of decomposition levels, and the optimal stratification threshold at each decomposition scale; when the current iteration count has not reached the maximum iteration count, performing a crossover operation on the mutant individuals and original individuals in the population, and using mutant individuals and / or original individuals with fitness values ​​less than a preset fitness value as the next generation population, and performing differential evolution operations in a loop until the maximum number of iterations is reached or the fitness function requirements are met, thus obtaining the optimal wavelet basis function, the optimal number of decomposition levels, and the optimal stratification threshold at each decomposition scale; Based on the optimal wavelet basis function, the optimal number of decomposition layers, and the optimal layering threshold at each decomposition scale, the detail coefficients are denoised to obtain the target detail coefficients. The target detail coefficients and the approximation coefficients are then combined to reconstruct the signal and obtain the target signal.

2. The signal denoising method based on DE-optimized wavelet parameters as described in claim 1, characterized in that, Before performing the crossover operation on the mutated individuals and the original individuals in the population, the following is also included: Based on a predefined differential strategy, mutation operations are performed on the original individuals in the population to obtain the mutated individuals.

3. The signal denoising method based on DE-optimized wavelet parameters as described in claim 1, characterized in that, The crossover operation between the mutated individuals and the original individuals in the population specifically involves: The binary crossover model is used to perform crossover operations on the variant individuals and the original individuals in the population.

4. The signal denoising method based on DE-optimized wavelet parameters as described in claim 1, characterized in that, The fitness function is: ； Where MAX is the maximum amplitude of the input signal, and RMSE is the root mean square error of the noisy signal. N is the length of the noisy signal. The wavelet coefficients of the input signal, These are the wavelet coefficients after thresholding. λ represents the stratification threshold at each decomposition scale.

5. A signal denoising device based on DE-optimized wavelet parameters, characterized in that, include: The wavelet decomposition module is used to select the initial wavelet basis function, and to perform wavelet decomposition on the noisy signal according to the initial decomposition level and the initial layer threshold at each decomposition scale, so as to obtain the detail coefficients at each decomposition scale and the approximation coefficients at the maximum decomposition scale. The parameter optimization module is used to take the initial wavelet basis function, the initial number of decomposition layers, and the initial stratification thresholds at each decomposition scale as the position parameters of the initial population in the DE algorithm, and to obtain the optimal wavelet basis function, the optimal number of decomposition layers, and the optimal stratification thresholds at each decomposition scale based on the DE algorithm. The method for obtaining the optimal wavelet basis function, optimal decomposition level, and optimal stratification threshold at each decomposition scale based on the DE algorithm specifically involves: parameter initialization: setting the population size, maximum number of iterations, fitness function, and defining each individual in the population as representing the stratification threshold at each decomposition scale; calculating the fitness value of each individual in the population according to the fitness function, and when the current iteration count reaches the maximum iteration count, selecting the population with the smallest fitness value as the optimal wavelet basis function, the optimal number of decomposition levels, and the optimal stratification threshold at each decomposition scale; when the current iteration count has not reached the maximum iteration count, performing a crossover operation on the mutant individuals and original individuals in the population, and using mutant individuals and / or original individuals with fitness values ​​less than a preset fitness value as the next generation population, and performing differential evolution operations in a loop until the maximum number of iterations is reached or the fitness function requirements are met, thus obtaining the optimal wavelet basis function, the optimal number of decomposition levels, and the optimal stratification threshold at each decomposition scale; The signal denoising module is used to denoise each of the detail coefficients according to the optimal wavelet basis function, the optimal number of decomposition layers, and the optimal layering threshold at each decomposition scale to obtain the target detail coefficients, and to reconstruct the signal by combining the target detail coefficients and the approximation coefficients to obtain the target signal.

6. The signal denoising device based on DE-optimized wavelet parameters as described in claim 5, characterized in that, Before performing the crossover operation on the mutated individuals and the original individuals in the population, the following is also included: Based on a predefined differential strategy, mutation operations are performed on the original individuals in the population to obtain the mutated individuals.

7. The signal denoising device based on DE-optimized wavelet parameters as described in claim 5, characterized in that, The crossover operation between the mutated individuals and the original individuals in the population specifically involves: The binary crossover model is used to perform crossover operations on the variant individuals and the original individuals in the population.

8. The signal denoising device based on DE-optimized wavelet parameters as described in claim 5, characterized in that, The fitness function is: ； Where MAX is the maximum amplitude of the input signal, and RMSE is the root mean square error of the noisy signal. N is the length of the noisy signal. The wavelet coefficients of the input signal, These are the wavelet coefficients after thresholding. λ represents the stratification threshold at each decomposition scale.

Images