Nuclear power circulating water pump impeller vibration signal denoising method and system

Abstract

A nuclear power circulating water pump impeller vibration signal denoising method comprises the steps that a vibration signal of a nuclear power circulating water pump impeller is collected; the improved sea gull algorithm adopts Logistic chaotic mapping to initialize a sea gull population, a tanh function is utilized to design a non-linear decreasing inertia weight, a Levy flight mechanism is added in a sea gull attack stage, a correlation coefficient and permutation entropy are utilized to construct a fitness function, and the improved sea gull algorithm is utilized to optimize variational mode decomposition parameters. According to the variational mode decomposition parameters optimized by the improved sea gull algorithm, variational mode decomposition is carried out on the vibration signals to obtain intrinsic mode components, and evaluation scores of the intrinsic mode components are calculated through the constructed screening indexes; and determining the noisy degree of each intrinsic mode component according to the evaluation score, performing corresponding denoising, reconstructing a signal by using the denoised intrinsic mode component, and setting an evaluation index to evaluate a result.

Description

Technical Field

[0001] The invention belongs to the technical field of nuclear power circulating water pump impeller vibration, and in particular to a nuclear power circulating water pump impeller vibration signal denoising method and system. Background Art

[0002] The circulating water pump is the main pump of the three-circuit circulating water system of a nuclear power plant, providing cooling water for the nuclear power plant's steam turbine condenser and auxiliary cooling system. The unstable flow of the fluid in the circulating water pump has strong randomness and pulsation, and there are multiple vortices. This unsteady fluid pressure pulsation will act on the impeller inside the casing, causing vibration excitation to it, thereby generating the impeller passing frequency.

[0003] As a key component of the circulating water pump, impeller failure can cause major events such as unit power reduction, shutdown and reactor shutdown. Therefore, it is necessary to diagnose the type and degree of fault in the early stage of impeller failure. Vibration signal acquisition is the main means of monitoring the state of the circulating water pump impeller. The vibration sensor indirectly extracts the impeller characteristic frequency by collecting the vibration signal of the casing. However, when collecting the impeller vibration signal, it is easily interfered by the water flow impact noise, and the characteristic frequency is easily buried in the noise. The pass frequency is an important characteristic frequency for identifying whether the impeller has a fault and its fault type. Therefore, it is necessary to use feature enhancement and denoising methods to identify the weak pass frequency.

[0004] The variational modal decomposition method is a signal decomposition method proposed by KD et al. in 2014, and has been widely used in signal processing, fault diagnosis and other fields. However, the traditional variational modal decomposition method requires manual setting of parameters, which makes it difficult to effectively decompose the signal, resulting in poor denoising effect. Therefore, it is necessary to realize the adaptive optimization of the key parameters of variational modal decomposition, thereby improving the denoising effect of the vibration signal, identifying the passing frequency of the impeller, and providing high-quality monitoring data for subsequent impeller fault diagnosis.

[0005] The above information disclosed in this Background section is only for enhancement of understanding of the background of the invention and therefore it may contain information that does not form the prior art that is already known to a person of ordinary skill in the art. Summary of the invention

[0006] The present invention proposes a method and system for denoising vibration signals of an impeller of a nuclear power circulating water pump, which improve the denoising effect of vibration signals and identify the passing frequency of the impeller.

[0007] The method for denoising the vibration signal of the impeller of a nuclear power circulating water pump comprises the following steps:

[0008] The first step is to collect the vibration signal of the impeller of the nuclear power circulating water pump;

[0009] In the second step, the improved seagull algorithm uses Logistic chaotic mapping to initialize the seagull population, uses the tanh function to design the nonlinear decreasing inertia weight, adds the Levy flight mechanism in the seagull attack phase, constructs the fitness function using the correlation coefficient and permutation entropy, and uses the improved seagull algorithm to optimize the variational mode decomposition parameters;

[0010] The third step is to perform variational modal decomposition on the vibration signal according to the variational modal decomposition parameters optimized by the improved Seagull algorithm to obtain the intrinsic modal components, and calculate the evaluation score of each intrinsic modal component through the constructed screening index;

[0011] The fourth step is to determine the noise level of each eigenmode component according to the evaluation score, perform corresponding denoising, reconstruct the signal using the denoised eigenmode components, and set evaluation indicators to evaluate the results.

[0012] In the method described, in the first step, collecting the vibration signal of the impeller of the nuclear power circulating water pump includes arranging three acceleration sensors on the impeller housing of the circulating water pump test bench to collect the vibration signals of the impeller in the x-axis, y-axis and z-axis directions.

[0013] In the method, in the second step,

[0014] The Logistic chaos initialization method is used to initialize the seagull population. The Logistic chaos mapping formula is:

[0015] z k+1 =μz k (1-z k ),

[0016] Where k is the number of iterations, z k represents the current value of the chaotic sequence, z k+1 represents the value of the next iterative chaotic sequence, μ is an adjustable parameter, and μ∈[0,4],

[0017] The nonlinear inertia weight of the seagull algorithm migration phase is designed using the tanh function:

[0018]

[0019] Among them, the value A is an additional variable that controls the position update of the individual seagull, that is, the designed nonlinear inertia weight, f c is the coefficient that controls the frequency of A value decrease, k is the number of iterations, K MAx is the maximum number of iterations, γ is the parameter that controls the slope of the tanh function,

[0020] Introducing the Levy flight mechanism into the seagull’s prey attack phase to enhance the algorithm’s ability to escape from the local optimal solution:

[0021] P s (k)=(D s (k)×x×y×z×Levi(β))+P bs (k)

[0022] Among them, P s (k) is the new position of the seagull after attacking the prey, P bs (k) is the optimal individual position of the seagull group, D s (k) is the relative position of the current individual of the seagull and the optimal individual, x, y, z are the spatial parameters of the seagull in the attacking prey stage, β is the random parameter of Levy flight, and β∈[0,2],

[0023] The Levy flight mechanism is expressed as:

[0024]

[0025] Among them, u and v are parameters that obey the random normal distribution, which are specifically expressed as:

[0026]

[0027] in, It means that μ has a mean of 0 and a variance of The normal distribution of μ is the standard deviation of the parameter μ, and the calculation formula is as follows:

[0028]

[0029] Where β is the random parameter of the Levy flight, and β∈[0,2], Γ(x)=(x-1)!.

[0030] In the method described, in the second step, the penalty factor α and the modal number K of the variational modal decomposition are optimized using the improved seagull algorithm, and the fitness function constructed is:

[0031]

[0032] Where x is the original signal, X is the reconstructed signal, r(X, x) is the correlation coefficient between the reconstructed signal X and the original signal x, P k is the permutation entropy of the kth eigenmode component, β1 is the weight coefficient,

[0033] The calculation formula for the permutation entropy of the kth eigenmode component is as follows:

[0034]

[0035] Where M is the length of the sequence, IMF k is the kth eigenmode component, and the eigenmode component time series is reconstructed in phase space, Pj (·) represents each row of the reconstruction matrix, that is, the probability of each sequence appearing.

[0036] In the method described, in the second step, the iterative process of optimizing the variational mode decomposition parameters by the improved Seagull algorithm is as follows:

[0037] Step 1: Initialize the size of the seagull population, set the problem-solving dimension, the number of algorithm iterations, and give the search range of the parameter K to be optimized [K min , K max ], given the search range of the parameter α to be optimized [α min , α max ],

[0038] Step 2: Set the initialization parameter combination and calculate the fitness value of each seagull individual.

[0039] Step 3: Update the current individual position of the seagull and the optimal individual position of the group. If the fitness value of the current position of the seagull is less than the fitness value of the last iteration, update the optimal individual position and optimal fitness value of the seagull group. Otherwise, the optimal individual position and optimal fitness value remain unchanged.

[0040] Step 4: Repeat steps 2 and 3 until the optimal variational mode decomposition parameter combination is obtained, or the current number of iterations is greater than the set number of iterations.

[0041] In the method, in the third step,

[0042] The correlation coefficient and permutation entropy are integrated to construct the screening index of the eigenmode components, that is, the evaluation score calculation formula of each eigenmode component is:

[0043] I=β2r(IMF k ,x)+(1-β2)P k ,

[0044] Among them, x is the original signal, r(IMF k , x) is the correlation coefficient between the kth IMF component and the original signal x, β2 is the weight coefficient,

[0045] The evaluation score of each eigenmode component is calculated based on the screening index.

[0046] In the method described above, in the fourth step, the noise level of each intrinsic mode component is determined, and the components are divided into useful components, low-noise components, and high-noise components. The IMF component with the highest evaluation score is defined as the useful component, and the evaluation score I of each IMF component is compared with the evaluation score I of the maximum IMF component. max For comparison, if The corresponding modal components are defined as high-noise components, and the rest are defined as low-noise components.

[0047] Perform wavelet threshold denoising on the low-noise component, discard the high-noise component, reconstruct the denoised low-noise component and the useful component to obtain the denoising result. The wavelet threshold formula is:

[0048]

[0049] Where n is the signal length, j = 1, 2, ... is the current layer number, λ j is the wavelet threshold of the jth layer, σ is the signal variance,

[0050] Setting evaluation indicators to evaluate denoising effects includes:

[0051] For simulated signals, the signal-to-noise ratio is used to evaluate the denoising effect:

[0052]

[0053] Where n is the signal length, x is the original signal, X is the denoised signal, and t is the time.

[0054] For the simulated signal, the root mean square error is used to evaluate the denoising effect:

[0055]

[0056] Among them, RMSE is the root mean square error, n is the signal length, x is the original signal, X is the denoised signal, t is the time,

[0057] For the measured vibration signal of the impeller, the denoising effect is evaluated using the noise reduction rate:

[0058]

[0059] Where NRR is the noise reduction rate, σ x is the standard deviation of the original signal, σ X is the standard deviation of the denoised signal.

[0060] A nuclear power circulating water pump impeller vibration signal denoising system includes:

[0061] The acceleration sensor in the x-axis direction is arranged on the circumference of the impeller housing.

[0062] The acceleration sensor in the y-axis direction is arranged on the circumference of the impeller housing and is arranged at 90° to the acceleration sensor in the x-axis direction.

[0063] The z-axis acceleration sensor is installed on the top of the impeller housing.

[0064] A signal acquisition and processing unit is electrically connected to the acceleration sensor in the x-axis direction, the acceleration sensor in the y-axis direction, and the acceleration sensor in the z-axis direction to execute the nuclear power circulating water pump impeller vibration signal denoising method.

[0065] Beneficial Effects

[0066] The present invention constructs a fitness function that combines the correlation coefficient, energy ratio and permutation entropy of each eigenmodal component with the original signal after variational modal decomposition, performs adaptive optimization of key parameters of variational modal decomposition through the improved Seagull algorithm, and then performs variational modal decomposition on the original signal according to the optimized parameters, and performs corresponding denoising steps on the eigenmodal components, which not only effectively improves the adaptability of the algorithm, but also retains useful information as much as possible while removing noise, and has a better noise reduction effect. The problem that the difficulty in selecting VMD parameters leads to the difficulty in effectively decomposing the signal and the poor denoising effect is solved, the denoising effect of the impeller vibration signal is improved, and the passing frequency of the impeller is identified. In addition, the present invention can also be used as a pre-processing part of mechanical equipment fault diagnosis, expanding its practical value in engineering applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 It is a schematic diagram of steps of an implementation method of a nuclear power circulating water pump impeller vibration signal denoising method in the present invention;

[0068] Figure 2 It is a layout diagram of acceleration sensors for collecting vibration signals of circulating pump impellers in one embodiment of a nuclear power circulating water pump impeller vibration signal denoising system of the present invention;

[0069] Figure 3 It is a graph showing the change of the fitness curve of an implementation method of a nuclear power circulating water pump impeller vibration signal denoising method in the present invention with the number of iteration rounds;

[0070] Figure 4 It is a graph showing the variation of key parameters of variational mode decomposition with the number of iterations of an implementation of a method for denoising vibration signals of an impeller of a nuclear power circulating water pump in the present invention;

[0071] Figure 5 It is a permutation entropy diagram of each eigenmode component after variational mode decomposition in one implementation of a nuclear power circulating water pump impeller vibration signal denoising method in the present invention;

[0072] Figure 6 It is a time domain diagram of each eigenmode component after variational mode decomposition of an implementation method of a nuclear power circulating water pump impeller vibration signal denoising method in the present invention;

[0073] Figure 7(a) to Figure 7(c) It is a comparison diagram of the denoising effect of an implementation method of a nuclear power circulating water pump impeller vibration signal denoising method in the present invention, wherein Figure 7(a), Figure 7(b), and Figure 7(c) are respectively the envelope spectrum of the original signal, the envelope spectrum obtained by the method of the present invention, and the envelope spectrum obtained by the wavelet threshold denoising method.

[0074] By reading the detailed description of the preferred specific embodiments below, various other advantages and benefits of the present invention will become clear to those of ordinary skill in the art. The drawings in the specification are only for the purpose of illustrating the preferred embodiments and are not considered to be limitations of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings without creative work. Moreover, the same reference numerals are used to represent the same components throughout the drawings. DETAILED DESCRIPTION

[0075] The specific embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although the specific embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present invention and to enable the scope of the present invention to be fully communicated to those skilled in the art.

[0076] It should be noted that certain words are used in the specification and claims to refer to specific components. Those skilled in the art should understand that technicians may use different nouns to refer to the same component. This specification and claims do not use the difference in nouns as a way to distinguish components, but use the difference in the functions of the components as the criterion for distinction. As mentioned throughout the specification and claims, "including" or "comprising" is an open term, so it should be interpreted as "including but not limited to". The subsequent description of the specification is a preferred embodiment of the present invention, but the description is based on the general principles of the specification and is not intended to limit the scope of the present invention. The scope of protection of the present invention shall be determined by the attached claims.

[0077] To facilitate understanding of the embodiments of the present invention, further explanation will be given below by taking specific embodiments as examples in conjunction with the accompanying drawings, and the accompanying drawings do not constitute a limitation on the embodiments of the present invention.

[0078] like Figure 1 to Figure 7(c) As shown, the method for denoising the vibration signal of the impeller of a nuclear power circulating water pump includes the following steps:

[0079] The first step is to collect the vibration signal of the impeller of the nuclear power circulating water pump;

[0080] In the second step, the improved seagull algorithm uses Logistic chaotic mapping to initialize the seagull population, uses the tanh function to design the nonlinear decreasing inertia weight, adds the Levy flight mechanism in the seagull attack phase, constructs the fitness function using the correlation coefficient and permutation entropy, and uses the improved seagull algorithm to optimize the variational mode decomposition parameters;

[0081] The third step is to perform variational modal decomposition on the vibration signal according to the variational modal decomposition parameters optimized by the improved Seagull algorithm to obtain the intrinsic modal components, and calculate the evaluation score of each intrinsic modal component through the constructed screening index;

[0082] The fourth step is to determine the noise level of each eigenmode component according to the evaluation score, perform corresponding denoising, reconstruct the signal using the denoised eigenmode components, and set evaluation indicators to evaluate the results.

[0083] In a preferred embodiment of the method, in the first step, collecting the vibration signal of the nuclear power circulating water pump impeller includes arranging three acceleration sensors on the impeller housing of the circulating water pump test bench to collect the vibration signals of the impeller in the x-axis, y-axis and z-axis directions.

[0084] In a preferred embodiment of the method, in the second step,

[0085] The Logistic chaos initialization method is used to initialize the seagull population. The Logistic chaos mapping formula is:

[0086] z k+1 =μz k (1-z k ),

[0087] Where k is the number of iterations, z k represents the current value of the chaotic sequence, z k+1 represents the value of the next iterative chaotic sequence, μ is an adjustable parameter, and μ∈[0,4],

[0088] The nonlinear inertia weight of the seagull algorithm migration phase is designed using the tanh function:

[0089]

[0090] Among them, the value A is an additional variable that controls the position update of the individual seagull, that is, the designed nonlinear inertia weight, f c is the coefficient that controls the frequency of A value decrease, k is the number of iterations, K MAx is the maximum number of iterations, γ is the parameter that controls the slope of the tanh function,

[0091] Introducing the Levy flight mechanism into the seagull’s prey attack phase to enhance the algorithm’s ability to escape from the local optimal solution:

[0092] P s (k)=(D s (k)×x×y×z×Levi(β))+P bs (k)

[0093] Among them, P s(k) is the new position of the seagull after attacking the prey, P bs (k) is the optimal individual position of the seagull group, x, y, z are the spatial parameters of the seagulls in the attack stage, β is the random parameter of Lévy flight, and β∈[0,2],

[0094] The flight mechanism of Naiwei is described as follows:

[0095]

[0096] Among them, u and v are parameters that obey the random normal distribution, which are specifically expressed as:

[0097]

[0098] in, It means that μ has a mean of 0 and a variance of The normal distribution of μ is the standard deviation of the parameter μ, and the calculation formula is as follows:

[0099]

[0100] Where β is the random parameter of the Levy flight, and β∈[0,2], Γ(x)=(x-1)!.

[0101] In a preferred embodiment of the method, in the second step, the penalty factor α and the modal number K of the variational modal decomposition are optimized using the improved seagull algorithm, and the fitness function constructed is:

[0102]

[0103] Where x is the original signal, for example, the original signal is the collected vibration signal, X is the reconstructed signal, r(X, x) is the correlation coefficient between the reconstructed signal X and the original signal x, P k is the permutation entropy of the kth eigenmode component, β1 is the weight coefficient,

[0104] The calculation formula for the permutation entropy of the kth eigenmode component is as follows:

[0105]

[0106] Where M is the length of the sequence, IMF k is the kth eigenmode component, and the eigenmode component time series is reconstructed in phase space, P j (·) represents each row of the reconstruction matrix, that is, the probability of each sequence appearing.

[0107] In a preferred embodiment of the method, in the second step, the iterative process of optimizing the variational mode decomposition parameters by the improved Seagull algorithm is as follows:

[0108] Step 1: Initialize the size of the seagull population, set the problem-solving dimension, the number of algorithm iterations, and give the search range of the parameter K to be optimized [K min , K max ], given the search range of the parameter α to be optimized [α min , α max ],

[0109] Step 2: Set the initialization parameter combination and calculate the fitness value of each seagull individual.

[0110] Step 3: Update the current individual position of the seagull and the optimal individual position of the group. If the fitness value of the current position of the seagull is less than the fitness value of the last iteration, update the optimal individual position and optimal fitness value of the seagull group. Otherwise, the optimal individual position and optimal fitness value remain unchanged.

[0111] Step 4: Repeat steps 2 and 3 until the optimal variational mode decomposition parameter combination is obtained, or the current number of iterations is greater than the set number of iterations.

[0112] In a preferred embodiment of the method, in the third step,

[0113] The correlation coefficient and permutation entropy are integrated to construct the screening index of the eigenmode components, that is, the evaluation score calculation formula of each eigenmode component is:

[0114] I=β2r(IMF k ,x)+(1-β2)P k ,

[0115] Among them, x is the original signal, r(IMF k , x) is the correlation coefficient between the kth IMF component and the original signal x, β2 is the weight coefficient,

[0116] The evaluation score of each eigenmode component is calculated based on the screening index.

[0117] In the preferred embodiment of the method, in the fourth step, the noise level of each eigenmode component is determined, and the components are divided into useful components, low-noise components, and high-noise components. The IMF component with the highest evaluation score is defined as the useful component, and the evaluation score I of each IMF component is compared with the evaluation score I of the maximum IMF component. max For comparison, if The corresponding modal components are defined as high-noise components, and the rest are defined as low-noise components.

[0118] Perform wavelet threshold denoising on the low-noise component, discard the high-noise component, reconstruct the denoised low-noise component and the useful component to obtain the denoising result. The wavelet threshold formula is:

[0119]

[0120] Where n is the signal length, j = 1, 2, ... is the current layer number, λ j is the wavelet threshold of the jth layer, σ is the signal variance,

[0121] Setting evaluation indicators to evaluate denoising effects includes:

[0122] For simulated signals, the signal-to-noise ratio is used to evaluate the denoising effect:

[0123]

[0124] Where n is the signal length, x is the original signal, X is the denoised signal, and t is the time.

[0125] For the simulated signal, the root mean square error is used to evaluate the denoising effect:

[0126]

[0127] Among them, RMSE is the root mean square error, n is the signal length, x is the original signal, X is the denoised signal, t is the time,

[0128] For the measured vibration signal of the impeller, the denoising effect is evaluated using the noise reduction rate:

[0129]

[0130] Where NRR is the noise reduction rate, σ x is the standard deviation of the original signal, σ X is the standard deviation of the denoised signal.

[0131] A nuclear power circulating water pump impeller vibration signal denoising system includes:

[0132] The acceleration sensor in the x-axis direction is arranged on the circumference of the impeller housing.

[0133] The acceleration sensor in the y-axis direction is arranged on the circumference of the impeller housing and is arranged at 90° to the acceleration sensor in the x-axis direction.

[0134] The z-axis acceleration sensor is installed on the top of the impeller housing.

[0135] A signal acquisition and processing unit is electrically connected to the acceleration sensor in the x-axis direction, the acceleration sensor in the y-axis direction, and the acceleration sensor in the z-axis direction to execute the nuclear power circulating water pump impeller vibration signal denoising method as described above.

[0136] In one embodiment, in the second step, an improved seagull algorithm is proposed, the logistic chaotic map is used to initialize the seagull population, the tanh function is used to design the nonlinear decreasing inertia weight, the Levy flight mechanism is added in the seagull attack phase, and the fitness function is constructed by using the correlation coefficient and the permutation entropy, and the improved seagull algorithm is used to optimize the key parameters of the variational mode decomposition, including:

[0137] The Logistic chaos initialization method is used to initialize the seagull population. The Logistic chaos mapping formula is:

[0138] z k+1 =μz k (1-z k )

[0139] Where k is the number of iterations, z k represents the current value of the chaotic sequence, z k+1 Represents the value of the next iterative chaotic sequence, μ is an adjustable parameter with a value of 4.

[0140] The nonlinear inertia weight of the seagull algorithm migration phase is designed using the tanh function:

[0141]

[0142] Among them, the value A is an additional variable that controls the position update of the individual seagull, that is, the designed nonlinear inertia weight, f c is the coefficient that controls the frequency of A value decrease, the value is 1, k is the number of iterations, K MAX is the maximum number of iterations, which is 15, and γ is the parameter that controls the slope of the tanh function, which is 1.

[0143] Introducing the Levy flight mechanism into the seagull’s prey attack phase to enhance the algorithm’s ability to escape from the local optimal solution:

[0144] P s (k)=(D s (k)×x×y×z×Levi(β))+P bs (k)

[0145] Among them, P s (k) is the new position of the seagull after attacking the prey, D s (k) is the relative position of the current seagull individual and the optimal individual, P bs (k) is the optimal individual position of the seagull group, x, y, z are the spatial parameters of the seagulls during the attack phase, and β is the random parameter of the Lévy flight, which is 1.5.

[0146] The Levy flight mechanism is expressed as:

[0147]

[0148] Among them, β is the random parameter of Levy flight, u and v are parameters that obey random normal distribution, which are specifically expressed as:

[0149]

[0150] in, It means that μ has a mean of 0 and a variance of The normal distribution of μ is the standard deviation of the parameter μ, and the calculation formula is as follows:

[0151]

[0152] Among them, β is the random parameter of Levy flight, which takes the value of 1.5, and Γ(x)=(x-1)! .

[0153] In this embodiment, in the second step, the energy and information loss of the signal after variational mode decomposition and the randomness of each intrinsic mode component are fully considered, and a new fitness function is constructed. The penalty factor α and the mode number K of variational mode decomposition are optimized by using the improved Seagull algorithm. The constructed new fitness function is:

[0154]

[0155] Where x is the original signal, X is the reconstructed signal, r(X, x) is the correlation coefficient between the reconstructed signal X and the original signal x, P k is the permutation entropy of the kth eigenmode component, β1 is the weight coefficient, and its value is 0.6.

[0156] The calculation formula for the permutation entropy of the kth eigenmode component is as follows:

[0157]

[0158] Where M is the length of the sequence, IMF k is the kth eigenmode component, and the eigenmode component time series is reconstructed in phase space, P j (·) represents each row of the reconstruction matrix, that is, the probability of each sequence appearing.

[0159] In this embodiment, in the second step, the iterative process of improving the seagull algorithm to optimize the variational mode decomposition parameters is as follows:

[0160] Step 1: Initialize the seagull population size n = 15, set the problem solving dimension d = 2, the algorithm iteration number N = 15, and give the search range of the parameter K to be optimized [K min , K max ], given the search range of the parameter α to be optimized [α min, α max ].

[0161] Step 2: Set the initialization parameter combination and calculate the fitness value of each seagull individual.

[0162] Step 3: Update the current individual position of the seagull and the optimal individual position of the group. If the fitness value of the current position of the seagull is less than the fitness value of the last iteration, update the optimal individual position and optimal fitness value of the seagull group. Otherwise, the optimal individual position and optimal fitness value remain unchanged.

[0163] Step 4: Repeat steps 2 and 3 until the optimal variational mode decomposition parameter combination is obtained, or the current number of iterations is greater than the set number of iterations.

[0164] In this embodiment, in the third step, the original signal is subjected to variational mode decomposition according to the parameters optimized by the improved Seagull algorithm to obtain a series of intrinsic mode components, and the evaluation score of each intrinsic mode component is calculated by the constructed screening index, including:

[0165] The correlation coefficient and permutation entropy are integrated to construct the screening index of the eigenmode components, that is, the evaluation score calculation formula of each eigenmode component is:

[0166] I=β2r(IMF k ,x)+(1-β2)P k ,

[0167] Among them, x is the original signal, r(IMF k , x) is the correlation coefficient between the kth IMF component and the original signal x, β2 is the weight coefficient,

[0168] Based on the screening index constructed above, the evaluation score of each eigenmode component is calculated.

[0169] In this embodiment, in the fourth step, the noise level of each eigenmode component is determined according to the evaluation score, a corresponding denoising step is performed, the denoised eigenmode component is used to reconstruct the signal, and an evaluation index is set to evaluate the denoising result, including:

[0170] Determine the noise level of each eigenmode component and distinguish them into useful components, low-noise components, and high-noise components. The specific distinction method is as follows: (1) Define the IMF component with the highest evaluation score as the useful component. (2) Compare the evaluation score I of each IMF component with the evaluation score I of the maximum IMF component. max For comparison. The corresponding modal components are defined as high-noise components, and the rest are defined as low-noise components.

[0171] Perform wavelet threshold denoising on the low-noise component, discard the high-noise component, reconstruct the denoised low-noise component and the useful component to obtain the denoising result. The wavelet threshold formula is:

[0172]

[0173] Where n is the signal length, j = 1, 2, ... is the current layer number, λ j is the wavelet threshold of the jth layer, and σ is the signal variance.

[0174] Set evaluation indicators to evaluate the denoising effect, including:

[0175] For simulated signals, the signal-to-noise ratio is used to evaluate the denoising effect:

[0176]

[0177] Where n is the signal length, x is the original signal, X is the denoised signal, and t is the time.

[0178] For the simulated signal, the root mean square error is also used to evaluate the denoising effect:

[0179]

[0180] Among them, RMSE is the root mean square error, n is the signal length, x is the original signal, X is the denoised signal, t is the time,

[0181] For the measured vibration signal of the impeller, the denoising effect is evaluated using the noise reduction rate:

[0182]

[0183] Where NRR is the noise reduction rate, σ x is the standard deviation of the original signal, σ X is the standard deviation of the denoised signal.

[0184] To further illustrate the method of the present invention, an enhanced nuclear power circulating water pump impeller vibration signal denoising method of the present invention is explained below with a specific implementation method and in combination with the accompanying drawings, which is not intended to limit the present invention.

[0185] Figure 1 It is a schematic diagram of the steps of an implementation method of an enhanced nuclear power circulating water pump impeller vibration signal denoising method in the present invention. Figure 2 The present invention is a layout diagram of acceleration sensors for collecting vibration signals of a circulating pump impeller, which is an implementation of an enhanced nuclear power circulating water pump impeller vibration signal denoising method in the present invention. Figure 3 It is a graph showing the change of the fitness curve of an implementation of an enhanced nuclear power circulating water pump impeller vibration signal denoising method in the present invention with the number of iteration rounds. Figure 4It is a graph showing the variation of key parameters of variational modal decomposition with the number of iterations of an implementation of an enhanced nuclear power circulating water pump impeller vibration signal denoising method in the present invention, wherein the key parameters are the penalty factor α and the modal number K. Figure 5 It is a permutation entropy diagram of each eigenmode component after variational mode decomposition in one implementation of an enhanced nuclear power circulating water pump impeller vibration signal denoising method in the present invention. Figure 6 It is a time domain diagram of each eigenmode component after variational mode decomposition of an implementation method of an enhanced nuclear power circulating water pump impeller vibration signal denoising in the present invention. Figure 7(a) to Figure 7(c) It is a comparison diagram of the denoising effect of an embodiment of an enhanced nuclear power circulating water pump impeller vibration signal denoising method in the present invention, wherein Figure 7(a), Figure 7(b), and Figure 7(c) are respectively the envelope spectrum of the original signal, the envelope spectrum obtained by the method of the present invention, and the envelope spectrum obtained by the wavelet threshold denoising method, and the impeller passing frequency is marked in the elliptical frame.

[0186] According to an enhanced nuclear power circulating water pump impeller vibration signal denoising method of the present invention, an exemplary implementation is provided as follows:

[0187] In the first step, three acceleration sensors are arranged on the impeller casing of the nuclear power circulating water pump test bench to collect vibration signals of the impeller in the x-axis, y-axis and z-axis directions. The circulating water pump speed is 341r / min, the number of impeller blades is 4, the shaft rotation frequency is 5.68Hz, and the passing frequency of the impeller is 22.72Hz.

[0188] In the second step, if Figure 4 ,By improving the Seagull optimization algorithm to optimize the key parameters of variational mode decomposition, the value of the penalty factor α gradually stabilizes after the 10th round and takes a value of 12840, and the value of the modal number K gradually stabilizes after 10 rounds and takes a value of 6.

[0189] In the third step, as shown in Table 1, the evaluation scores of each eigenmode component are obtained by calculation.

[0190] Table 1 Evaluation scores of each eigenmode component

[0191]

[0192] In the fourth step, as shown in Table 1, among the six eigenmode components after variational mode decomposition, the third eigenmode component has the highest evaluation score and is not subjected to denoising. The evaluation scores of the remaining eigenmode components are all greater than half of the highest evaluation score and are subjected to wavelet denoising. The third eigenmode component and the remaining eigenmode components after denoising are reconstructed to obtain the denoised signal.

[0193] The denoising results of the present invention are compared with the denoising results of direct wavelet threshold processing, which verifies the effectiveness of the method of the present invention. Figure 7(a) to Figure 7(c) The figure shows a comparison of denoising effects. From the comparison of Figure 7(a) and Figure 7(c), it can be seen that the signal after direct wavelet threshold denoising cannot identify the impeller passing frequency, and the spectrum has many interference components, and the denoising effect is very poor. From the comparison of Figure 7(a) and Figure 7(b), it can be seen that the method of the present invention can not only clearly distinguish the impeller passing frequency, but also enhance the rotation frequency and its multiple amplitude to a certain extent. The spectrum is relatively clean and has fewer interference components. Table 2 is a comparison of the denoising rate of the method of the present invention and that after direct wavelet threshold processing. From Table 2, it can be seen that the denoising effect of the method of the present invention is better, and the denoising rate is significantly improved compared with that of direct wavelet threshold processing, thus verifying the effectiveness of the method proposed in the present invention.

[0194] Table 2 Comparison of denoising performance of the two methods

[0195]

[0196] The present disclosure also provides an enhanced nuclear power circulating water pump impeller vibration signal denoising system, comprising at least three acceleration sensors and a signal acquisition and processing module. The acceleration sensor in the z-axis direction is installed on the top of the impeller housing, and the acceleration sensors in the x-axis and y-axis directions are arranged on the circumference of the impeller housing at 90°. The acceleration sensors in the three directions are electrically connected to the signal acquisition and processing unit respectively.

[0197] Although the embodiments of the present invention are described above in conjunction with the accompanying drawings, the present invention is not limited to the above specific embodiments and application fields, and the above specific embodiments are only illustrative and instructive, rather than restrictive. A person of ordinary skill in the art can also make many forms under the guidance of this specification and without departing from the scope of protection of the claims of the present invention, all of which belong to the protection of the present invention.

Claims

1. A method for denoising vibration signals of a nuclear power circulating water pump impeller, characterized in that: It includes the following steps: The first step is to collect the vibration signal of the impeller of the nuclear power circulating water pump; In the second step, the improved seagull algorithm uses Logistic chaotic mapping to initialize the seagull population, uses the tanh function to design the nonlinear decreasing inertia weight, adds the Levy flight mechanism in the seagull attack phase, constructs the fitness function using the correlation coefficient and permutation entropy, and uses the improved seagull algorithm to optimize the variational mode decomposition parameters; The third step is to perform variational modal decomposition on the vibration signal according to the variational modal decomposition parameters optimized by the improved Seagull algorithm to obtain the intrinsic modal components, and calculate the evaluation score of each intrinsic modal component through the constructed screening index; The fourth step is to determine the noise level of each eigenmode component according to the evaluation score, perform corresponding denoising, reconstruct the signal using the denoised eigenmode components, and set evaluation indicators to evaluate the results.

2. The method according to claim 1, characterized in that Preferably, in the first step, collecting the vibration signal of the nuclear power circulating water pump impeller includes arranging three acceleration sensors on the impeller housing of the circulating water pump test bench to collect the vibration signals of the impeller in the x-axis, y-axis and z-axis directions.

3. The method according to claim 1, characterized in that In the second step, The Logistic chaos initialization method is used to initialize the seagull population. The Logistic chaos mapping formula is: z k+1 =μz k (1-z k ), Where k is the number of iterations, z k represents the current value of the chaotic sequence, z k+1 represents the value of the next iterative chaotic sequence, μ is an adjustable parameter, and μ∈[0,4], The nonlinear inertia weight of the seagull algorithm migration phase is designed using the tanh function: Among them, the value A is an additional variable that controls the position update of the individual seagull, that is, the designed nonlinear inertia weight, f c is the coefficient that controls the frequency of A value decrease, k is the number of iterations, K MAX is the maximum number of iterations, γ is the parameter that controls the slope of the tanh function, Introducing the Levy flight mechanism into the seagull’s prey attack phase to enhance the algorithm’s ability to escape from the local optimal solution: P.S s (k)(D s (k)×x×y×z×Levi(β))+P bs (k) Among them, P s (k) is the new position of the seagull after attacking the prey, D s (k) is the relative position of the current seagull individual and the optimal individual, P bs (k) is the optimal individual position of the seagull group, x, y, z are the spatial parameters of the seagulls in the attack stage, β is the random parameter of Lévy flight, and β∈[0,2], The Levy flight mechanism is expressed as: Among them, β is the random parameter of Levy flight, u and v are parameters that obey random normal distribution, which are specifically expressed as: in, It means that μ has a mean of 0 and a variance of The normal distribution of μ is the standard deviation of the parameter μ, and the calculation formula is as follows: Where β is the random parameter of the Levy flight, and β∈[0,2], Γ(x)=(x-1)!.

4. The method according to claim 1, characterized in that: In the second step, the penalty factor α and the modal number K of the variational mode decomposition are optimized using the improved Seagull algorithm, and the constructed fitness function is: Where x is the original signal, X is the reconstructed signal, r(X, x) is the correlation coefficient between the reconstructed signal X and the original signal x, P k is the permutation entropy of the kth eigenmode component, β1 is the weight coefficient, The calculation formula for the permutation entropy of the kth eigenmode component is as follows: Where M is the length of the sequence, IMF k is the kth eigenmode component, and the eigenmode component time series is reconstructed in phase space, P j (·) represents each row of the reconstruction matrix, that is, the probability of each sequence appearing.

5. The method according to claim 1, characterized in that In the second step, the iterative process of the improved Seagull algorithm to optimize the variational mode decomposition parameters is as follows: Step 1: Initialize the size of the seagull population, set the problem solving dimension, the number of algorithm iterations, and give the search range of the parameter K to be optimized [K min , K max ], given the search range of the parameter α to be optimized [α min , α max ], Step 2: Set the initialization parameter combination and calculate the fitness value of each seagull individual. Step 3: Update the current individual position of the seagull and the optimal individual position of the group. If the fitness value of the current position of the seagull is less than the fitness value of the last iteration, update the optimal individual position and optimal fitness value of the seagull group. Otherwise, the optimal individual position and optimal fitness value remain unchanged. Step 4: Repeat steps 2 and 3 until the optimal variational mode decomposition parameter combination is obtained, or the current number of iterations is greater than the set number of iterations.

6. The method according to claim 1, characterized in that In the third step, The correlation coefficient and permutation entropy are integrated to construct the screening index of the eigenmode components, that is, the evaluation score calculation formula of each eigenmode component is: I=β2r(IMF k ,x)+(1-β2)P k , Where x is the original signal; r(IMF k , x) is the correlation coefficient between the kth IMF component and the original signal x, β2 is the weight coefficient, The evaluation score of each eigenmode component is calculated based on the screening index.

7. A nuclear power circulating water pump impeller vibration signal denoising system, characterized in that: These include, The acceleration sensor in the x-axis direction is arranged on the circumference of the impeller housing. The acceleration sensor in the y-axis direction is arranged on the circumference of the impeller housing and is arranged at 90° to the acceleration sensor in the x-axis direction. The z-axis acceleration sensor is installed on the top of the impeller housing. A signal acquisition and processing unit is electrically connected to an acceleration sensor in the x-axis direction, an acceleration sensor in the y-axis direction, and an acceleration sensor in the z-axis direction to execute the nuclear power circulating water pump impeller vibration signal denoising method as described in any one of claims 1-6.

Images