Electromagnetic water meter noise reduction algorithm based on combination of optimized mode decomposition and wavelet noise reduction algorithm

Abstract

In order to solve the problem that a small-flow signal under low-frequency measurement of an electromagnetic water meter is easily submerged by noise, the invention provides an improved litsea optimization algorithm for optimizing ICEEMDAN in combination with an improved wavelet threshold noise reduction algorithm. The method comprises the following steps: firstly, optimizing ICEEMDAN by improving an eagle optimization algorithm to decompose a traffic signal into optimal modal components, screening out IMF components with rich traffic information by permutation entropy, then carrying out noise reduction processing by combining a wavelet threshold algorithm of an improved adaptive threshold function, and finally reconstructing all components to obtain a noise reduction signal. The precision of the electromagnetic water meter can be effectively improved.

Description

Technical Field

[0001] The present invention relates to the field of electromagnetic water meter noise reduction, and specifically designs an electromagnetic water meter noise reduction algorithm based on optimized modal decomposition combined with wavelet noise reduction algorithm Background Art

[0002] As a commonly used flow detection instrument, electromagnetic water meter is widely used in the flow measurement of various conductive liquids in industrial environments. Electromagnetic water meter has many advantages such as high measurement accuracy, strong reliability, and not easily disturbed by fluid media. Electromagnetic water meter has many methods such as steady-state measurement, transient measurement, and residual magnetism measurement. Among them, steady-state measurement is the most widely used because it is not affected by differential interference, but due to power consumption requirements, most of them are low-frequency measurements. Under low-frequency measurement, the accuracy of electromagnetic water meter is easily disturbed by noise, especially under small flow. The noise interference is very large, which seriously affects the accuracy of electromagnetic water meter. At present, the traditional noise reduction algorithms for electromagnetic water meter include median filtering, Fourier transform noise reduction, Kalman filtering, etc. Affected by factors such as the material properties, manufacturing process, and installation method of electromagnetic water meter, electromagnetic water meter will generate more interference during operation. At the same time, there is a lot of interference noise in the working environment of electromagnetic water meter. Under the measurement of small flow, the flow signal is easily disturbed by these noises, resulting in errors in the final flow measurement and reducing the accuracy of water meter.

[0003] In summary, in order to improve the accuracy of electromagnetic water meters under low-frequency measurement, it is necessary to select appropriate signal processing methods to extract the flow signal components submerged in numerous noises, and accurately separate the flow signal and the noise signal to effectively improve the accuracy of the electromagnetic water meter. Summary of the invention

[0004] In order to solve the problem that small flow signals under low-frequency measurement of electromagnetic water meters are easily submerged by noise, this paper proposes an improved Eagle optimization algorithm to optimize ICEEMDAN combined with an improved wavelet threshold denoising algorithm. First, the improved Eagle optimization algorithm is used to optimize ICEEMDAN to decompose the flow signal into the optimal modal components, and then the IMF components with relatively rich flow information are screened out through permutation entropy. After that, the wavelet threshold algorithm combined with the improved adaptive threshold function is used for denoising, and finally all components are reconstructed to obtain the denoised signal. The present invention has a good denoising effect in processing electromagnetic water meter flow signals.

[0005] The specific invention of this article adopts the following scheme:

[0006] An electromagnetic water meter denoising algorithm based on optimized modal decomposition combined with wavelet threshold denoising algorithm, the method comprises the following steps:

[0007] (1) Improved Tianying optimization algorithm for ICEEMDAN optimization

[0008] (1.1) By using TENT chaotic mapping and adding adaptive weight factors, the Eagle optimization algorithm can balance the ability to expand and reduce exploration. In the early stage of iteration, the Eagle moves at a faster speed and quickly reaches the target value. In the later stage of iteration, the Eagle moves at a lower speed and is less likely to fall into the local optimum.

[0009] (1.2) The improved SkyEagle optimization algorithm is used to adjust the parameters of ICEEMDAN, including the noise amplitude weight and the number of noise additions. The envelope entropy is used as the objective function, and the component with the largest envelope entropy peak factor is selected as the optimal component.

[0010] (2) Improved wavelet threshold algorithm

[0011] (2.1) An adaptive threshold is added to the wavelet threshold algorithm to solve the problem that the traditional wavelet algorithm uses the same threshold at different decomposition levels, resulting in information loss or excessive noise. This paper improves the threshold formula so that it can adaptively adjust the threshold as the number of decomposition levels decreases layer by layer.

[0012] (2.2) The threshold function of the wavelet threshold algorithm is improved. The hard threshold function is better than the soft threshold method in terms of mean square error, but the signal will produce additional oscillations and jump points, and it does not have the smoothness of the original signal. The wavelet coefficients estimated by the soft threshold have good overall continuity, and the estimated signal will not produce additional oscillations. However, since the signal will be compressed, the information of the real signal will be lost. In order to solve the shortcomings of soft and hard thresholds and improve the noise reduction ability, the threshold function is improved so that it can solve the shortcomings of soft and hard threshold functions while improving the noise reduction level. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] The following drawings show specific examples of the technical solutions of the present invention, and are used to explain the content, principle and effect of the present invention.

[0014] Unless otherwise specified or defined, the same reference numerals in different drawings represent the same or similar technical features. Different reference numerals may be used to represent the same or similar technical features.

[0015] Figure 1 It is a specific flow chart of the algorithm of the present invention.

[0016] Figure 2 This is the effect diagram of the threshold function of the improved wavelet denoising algorithm Specific implementation plan

[0017] In order to facilitate the understanding of the present invention, specific embodiments of the present invention will be described in more detail below with reference to the accompanying drawings.

[0018] Unless otherwise specified or defined, all technical and scientific terms used herein have the same meaning as those generally understood by those skilled in the art. In combination with the technical solution of the present invention in a realistic scenario, all technical and scientific terms used herein may also have meanings corresponding to the technical solution for implementing the present invention. The "first, second..." used herein is merely used to distinguish the names and does not represent a specific quantity or order. The term "and / or" used herein includes any and all combinations of one or more related listed items.

[0019] Unless otherwise specified or defined, the “said” and “the” used in this document refer to the technical features or contents mentioned or described before the corresponding position, and the technical features or technical contents may be the same as or similar to the technical features or technical contents mentioned therein.

[0020] Undoubtedly, technical contents or technical features that are contrary to the purpose of the present invention or are obviously contradictory should be excluded.

[0021] like Figure 1 The present invention designs an electromagnetic water meter noise reduction algorithm based on adaptive modal decomposition combined with wavelet noise reduction algorithm. The specific process is as follows:

[0022] (1) Extraction of electromagnetic water meter flow signal

[0023] Aiming at the problem that the flow signal of electromagnetic water meter is easily submerged by noise when measuring small flow water, this paper proposes an electromagnetic water meter noise reduction algorithm based on improved and optimized ICEEMDAN combined with improved wavelet threshold algorithm, which effectively improves the noise reduction level of electromagnetic water meter and improves the accuracy of electromagnetic water meter.

[0024] The ICEEMDAN algorithm can adaptively adjust the noise level according to different signals, suppress the noise components in the signal, and retain the flow information in the signal, effectively improving the effect of signal decomposition. At the same time, the improved Tianying optimization algorithm is used to optimize the parameters of ICEEMDAN to improve the signal decomposition effect.

[0025] (1.1) Tianying optimization algorithm optimizes ICEEMDAN

[0026] Since the final effect of the ICEEMDAN method is determined by the amplitude weight and number of noise introduction, and the noise amplitude weight and noise addition number set based on experience are not ideal for the decomposition effect of electromagnetic water meter signals, the present invention uses the Sky Eagle optimization algorithm to optimize ICEEMDAN.

[0027] The fitness function has a crucial influence on the optimization effect of the Tianying optimization algorithm and determines the final optimization result. The present invention adopts envelope entropy as the objective function of the Tianying optimization algorithm. When the envelope entropy is minimized, the optimal noise amplitude weight and noise addition times are obtained.

[0028] Steps of optimizing ICEEMDAN with Tianying optimization algorithm

[0029] Step 1: Although the Tianying optimization algorithm has the advantages of fast convergence speed and high accuracy, it may have uneven population distribution due to the initialization population problem in the expansion stage as shown in Formula 1. In order to further obtain the optimal accuracy of the flow, the AO algorithm can be improved to obtain faster convergence speed and higher accuracy. This paper introduces Tent chaotic mapping to generate a better initialization population to improve population diversity and prevent the uneven population distribution problem that may occur during the optimization process.

[0030]

[0031] where x best (t) is the best individual; x1(t) is a random solution in the population at the tth iteration, rand is a random number between [0,1]; N is the number of groups; T is the maximum number of iterations, and t is the current number of iterations.

[0032] The expression of Tent chaos map is:

[0033]

[0034] Where x i is the optimal value.

[0035] Step 2: Because it is easy to fall into the local optimum during the algorithm optimization process, the present invention adds adaptive weights to balance the ability of expanding exploration and narrowing exploration. In the early stage of iteration, the eagle moves at a faster speed to quickly reach the target value, and in the later stage of iteration, the eagle moves at a lower speed, which is not easy to fall into the local optimum.

[0036] The improved and reduced exploration phase formula is shown in Formula 7, where g1 is a random number between -1 and 1:

[0037]

[0038] Levy(D) is the Levy flight distribution function

[0039] The adaptive weight formula is:

[0040]

[0041] By adding the adaptive weight factor ω, the algorithm allows the Eagle to move at a faster speed in the early iteration and quickly reach the target value. In the later iteration, the Eagle moves at a slower speed and is less likely to fall into the local optimum.

[0042] Step 3: Set the number of iterations of the Tianying optimization algorithm to output the optimal parameter combination of noise amplitude weight and minimum envelope entropy of noise addition times.

[0043] Step 4: Substitute the optimal parameter combination of noise amplitude weight and noise addition times into ICEEMDAN to decompose the mode and obtain the optimal mode combination.

[0044] (1.2) Multi-scale permutation entropy extraction of useful signals

[0045] Permutation Entropy is an average entropy parameter that measures the complexity of a one-dimensional time series. Permutation Entropy (PE) is an average entropy function that measures the complexity of a time series and detects dynamic mutations. It has a simple algorithm and fast calculation speed. The more useful signal components in a signal, the greater the permutation entropy value. The present invention uses permutation entropy to calculate the information value of each modal component and selects the IMF component containing rich useful signals. The noise reduction level of the final electromagnetic water meter is greatly improved.

[0046] Aiming at the problem that the flow signal of electromagnetic water meter is easily submerged by noise when measuring small flow water, the present invention proposes an improved ICEEMDAN algorithm. The noise reduction level of electromagnetic water meter is effectively improved, because the decomposition effect of ICEMFDAN is affected by the required parameter noise amplitude weight and the number of noise additions. Therefore, the improved Sky Eagle optimization algorithm is used to optimize the two parameters through the fitness function envelope entropy, and all IMF components are calculated by permutation entropy to obtain several IMF components containing rich flow information. This method can effectively enhance the characteristic information of flow signal.

[0047] (2) Flow signal noise reduction

[0048] Although the several IMF components obtained by ICEEMDAN contain rich traffic signal information, there is still a lot of noise, which needs to be further reduced. In view of the fact that the traditional wavelet threshold has only a fixed threshold in the noise reduction process, resulting in a large difference in the noise reduction level of different decomposition layers, and the traditional soft and hard threshold functions will respectively cause the signal to produce additional oscillations, jump points and signal loss, the present invention proposes an adaptive threshold function, so that a good noise reduction level can be achieved at different decomposition layers, and at the same time, an improved threshold function is used to solve the shortcomings of the traditional threshold function.

[0049] Improved denoising steps of wavelet threshold denoising algorithm

[0050] Step 1: Determine the selection of wavelet basis, decomposition scale, threshold and threshold function for the improved wavelet threshold algorithm. Compare different wavelet basis functions and obtain the optimal wavelet basis and optimal decomposition scale based on the signal-to-noise ratio of the final denoising result.

[0051] Step 2: Common threshold selection methods include fixed threshold estimation, extreme threshold estimation, etc. There may be a phenomenon that the noise signal is lost or the signal still has a lot of noise after noise reduction. In order to solve the problem of using the same threshold for different decomposition levels, the present invention improves the threshold formula so that it can adaptively adjust the threshold as the decomposition level decreases layer by layer. The corrected threshold is:

[0052]

[0053] Where σ is the noise standard deviation, N is the signal length, and k is the number of decomposition levels.

[0054] Step 3: After determining the threshold, the threshold function must also be determined. In the wavelet threshold algorithm, the hard threshold function and the soft threshold function are currently the most widely used. The hard threshold function is better than the soft threshold method in terms of mean square error, but the signal will produce additional oscillations and jump points, and it does not have the smoothness of the original signal. The wavelet coefficients estimated by the soft threshold have good overall continuity, and the estimated signal will not produce additional oscillations, but because it compresses the signal, the information of the real signal is lost. In order to solve the shortcomings of soft and hard thresholds and improve the noise reduction ability, the improved threshold function is as follows

[0055]

[0056] Where w j,k represents the threshold function, λ i Represents wavelet coefficients of different layers.

[0057] After the threshold correction, the new threshold function can obtain the optimal threshold under the current IMF component according to the correction function, while ensuring that the threshold function is The proof of continuity within the range is as follows.

[0058]

[0059] It can be seen from the formula that the threshold function is continuous at λi. Similarly, the improved threshold function is also continuous at -λi, as shown in Figure 1 As shown in the figure, the improved threshold function is nonlinear, which not only solves the problem of constant error in the soft threshold function but also solves the problem of local signal oscillation in the hard threshold function, and the denoising effect is closer to the real signal.

[0060] Step 4: Reconstruct all IMF components after noise reduction to obtain the final required traffic signal.

[0061] The present invention uses an improved wavelet threshold denoising algorithm to denoise the flow signal of the electromagnetic water meter. First, an adaptive threshold is set according to the flow signal. Secondly, a threshold function with better effect is designed so that the wavelet denoising algorithm can achieve good effect when processing the IMF component. Finally, the IMF component after denoising is reconstructed. This method can effectively reduce noise interference and improve the accuracy of the electromagnetic water meter.

Claims

1. An electromagnetic water meter denoising algorithm based on optimized modal decomposition combined with wavelet denoising algorithm, characterized in that It includes an electromagnetic water meter flow signal extraction algorithm with optimized modal decomposition and an improved wavelet denoising algorithm. The electromagnetic water meter flow signal extraction algorithm with optimized modal decomposition is mainly composed of improving the Sky Eagle optimization algorithm to optimize the noise amplitude weight and the number of noise additions of ICEEMDAN. The electromagnetic water meter flow signal extraction algorithm with optimized modal decomposition completes the selection of IMF components through permutation entropy. The improved wavelet denoising algorithm is mainly composed of setting an adaptive function and improving a threshold function.

2. Flow signal extraction by optimizing modal decomposition according to claim 1, characterized in that: The ICEEMDAN algorithm can adaptively adjust the noise level according to different signals, suppress the noise components in the signal, and retain the flow information in the signal, effectively improving the effect of signal decomposition. At the same time, the improved Tianying optimization algorithm is used to optimize the parameters of ICEEMDAN to improve the signal decomposition effect.

3. Flow signal extraction by optimizing modal decomposition according to claim 2, characterized in that: The present invention adopts envelope entropy as the objective function of the Sky Eagle optimization algorithm, and obtains the optimal noise amplitude weight and noise addition times when the envelope entropy is minimum.

4. Flow signal extraction by optimizing modal decomposition according to claim 2, characterized in that: The improved Sky Eagle optimization algorithm introduces Tent chaotic mapping to generate a better initialization population to improve population diversity and prevent the problem of uneven population distribution that may occur during the optimization process.

5. Flow signal extraction by optimizing modal decomposition according to claim 2, characterized in that: By adding the adaptive weight factor ω, the algorithm allows the Eagle to move at a faster speed in the early iteration and quickly reach the target value. In the later iteration, the Eagle moves at a slower speed and is less likely to fall into the local optimum.

6. Flow signal extraction by optimizing modal decomposition according to claim 2, characterized in that: Set the number of iterations of the Tianying optimization algorithm to output the best parameter combination of noise amplitude weight and minimum envelope entropy of noise addition times.

7. Flow signal extraction by optimizing modal decomposition according to claim 2, characterized in that: The optimal parameter combination of noise amplitude weight and noise addition times is substituted into ICEEMDAN to decompose the mode and obtain the optimal mode combination.

8. Flow signal extraction by optimizing modal decomposition according to claim 1, characterized in that: The permutation entropy is used to calculate the information value of each modal component and select the IMF component containing rich useful signals, which greatly improves the noise reduction level of the final electromagnetic water meter.

9. The improved wavelet denoising algorithm according to claim 1 is characterized in that: The improved wavelet threshold algorithm determines the selection of wavelet basis, decomposition scale, threshold and threshold function. By comparing different wavelet basis functions, the optimal wavelet basis and optimal decomposition scale are obtained according to the signal-to-noise ratio of the final denoising result.

10. The improved wavelet denoising algorithm according to claim 1 is characterized in that: The threshold formula is improved so that the threshold can be adaptively adjusted as the number of decomposition layers decreases.

11. The improved wavelet denoising algorithm according to claim 1 is characterized in that: The wavelet threshold function is improved. The improved threshold function is nonlinear, which not only solves the constant error problem of the soft threshold function but also solves the problem of local signal oscillation of the hard threshold function. The denoising effect is closer to the real signal.

12. Flow signal extraction by optimizing modal decomposition according to claim 1, characterized in that: All the IMF components after noise reduction are reconstructed to obtain the final required traffic signal.

Images