CN119832932A - Transformer acoustic signal processing method and system based on improved ICA algorithm - Google Patents
Transformer acoustic signal processing method and system based on improved ICA algorithm Download PDFInfo
- Publication number
- CN119832932A CN119832932A CN202411811745.6A CN202411811745A CN119832932A CN 119832932 A CN119832932 A CN 119832932A CN 202411811745 A CN202411811745 A CN 202411811745A CN 119832932 A CN119832932 A CN 119832932A
- Authority
- CN
- China
- Prior art keywords
- signal
- noise
- imf component
- transformer
- dimension
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Landscapes
- Complex Calculations (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
Abstract
The invention discloses a transformer acoustic signal processing method and system based on an improved ICA algorithm, wherein the method comprises the steps of inputting a noise-containing transformer acoustic signal to obtain a mixed signal matrix to be processed; and decomposing the underdetermined mixed to-be-processed audio signal matrix based on an EEMD algorithm with improved minimum lower limit frequency to obtain IMF components, performing preliminary denoising, estimating source signal dimensions based on a voiceprint dimension identification model of the GRU network, increasing dimension acoustic signal matrix dimensions, judging whether the dimension-increasing signal matrix is underdetermined, extracting residual errors if underdetermined, screening dimension-increasing signal groups based on a correlation maximum principle, and improving ICA algorithm component blind source signals based on N-MNE. The processing effect of the transformer acoustic signals can be effectively improved.
Description
Technical Field
The invention belongs to the technical field of power systems, and particularly relates to a transformer acoustic signal processing method and system based on an improved ICA algorithm.
Background
The transformer is a key device in a high-voltage and long-distance power transmission and transformation system in a transformer substation. Meanwhile, the transformer is also one of the main noise sources in the transformer substation. In the running process, acoustic signals generated by mechanical vibration of structures such as a transformer winding, an iron core, an oil tank and the like contain abundant state characteristics, faults of the transformer are closely related to voiceprints, and anomalies of the acoustic signals of the transformer body are highly related to self-die barriers. In addition, compared with the power failure detection such as the transition resistance method, voiceprint recognition has become a research hotspot for students at home and abroad in recent years due to its practical performance, convenience and economy. The development of the voiceprint detection and fault diagnosis of the transformer has great significance for guaranteeing the operation safety of the transformer, the microphone array is designed to collect the transformer acoustic signals, and the working condition detection and diagnosis of the transformer can be realized through the analysis of the acoustic signals. However, during the process of collecting the acoustic signal of the power transformer, a large amount of interference noise signals such as ambient noise, corona discharge, bird song and the like often exist, so that a pure transformer gas operation state acoustic signal cannot be obtained efficiently and quickly, and misjudgment is often caused by a fault state monitoring system.
Blind source separation (blind source separation, BSS) is a technique that recovers the source signal by observing the signal alone, with a priori knowledge of the passive signal and the aliasing coefficients. Due to the advancement and superiority, the method becomes a research hotspot for domestic and foreign scholars in recent years. When the number of microphone elements is less than the number of noise source signals, the BSS problem is called Underdetermined Blind Source Separation (UBSS), and UBSS has become a research hotspot in the international signal processing field due to its wider application. The FastICA algorithm is used as a common algorithm for solving the blind source separation problem, has good separation precision under the non-underdetermined working condition, but in the actual working condition, the types of noise sources are generally random and complex (up to tens of types), and the number of noise sensors is in the underdetermined condition. How to meet the reversibility constraint of the mixing process of the FastICA algorithm, the uniqueness constraint condition of a separation signal and the like in the blind source separation of the transformer voiceprint becomes a key problem for solving the underdetermined blind source separation of the transformer voiceprint.
Disclosure of Invention
The invention provides a transformer acoustic signal processing method and system based on an improved ICA algorithm, which are used for solving the technical problem of underdetermined blind source separation of a transformer voiceprint.
In a first aspect, the present invention provides
In a second aspect, the present invention provides
In a third aspect, an electronic device is provided that includes at least one processor and a memory communicatively coupled to the at least one processor, wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the steps of the improved ICA algorithm-based transformer acoustic signal processing method of any of the embodiments of the present invention.
In a fourth aspect, the present invention also provides a computer-readable storage medium having stored thereon a computer program, which when executed by a processor, causes the processor to perform the steps of the method for processing acoustic signals of a transformer based on the improved ICA algorithm of any of the embodiments of the present invention.
According to the transformer acoustic signal processing method and system based on the improved ICA algorithm, noise-containing transformer acoustic signals are input to obtain a mixed to-be-processed signal matrix, an EEMD (ensemble empirical mode decomposition) algorithm based on minimum lower limit frequency improvement is used for decomposing an underdetermined mixed to-be-processed audio signal matrix to obtain IMF (in-plane-with-view) components and performing preliminary denoising, a voiceprint dimension identification model based on a GRU (generalized unit) network is used for estimating source signal dimension, dimension of a dimension-increasing acoustic signal matrix is judged, whether the dimension-increasing signal matrix is underdetermined or not is judged, residual errors are extracted if underdetermined, dimension-increasing signal groups are selected based on a correlation maximum principle, and an ICA algorithm component blind source signal is improved based on N-MNE, so that the processing effect of the transformer acoustic signals can be effectively improved.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flowchart of a transformer acoustic signal processing method based on an improved ICA algorithm according to an embodiment of the present invention;
FIG. 2 is a block diagram of a transformer acoustic signal processing system based on an improved ICA algorithm according to an embodiment of the present invention;
fig. 3 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Referring to fig. 1, a flowchart of a transformer acoustic signal processing method based on an improved ICA algorithm of the present application is shown.
As shown in fig. 1, the transformer acoustic signal processing method based on the improved ICA algorithm specifically includes the following steps:
Step S101, constructing a mixed signal matrix to be processed according to the acquired noise-containing transformer signals.
And step S102, decomposing the mixed signal matrix to be processed by adopting an improved EEMD algorithm based on the minimum lower limit frequency to obtain at least one IMF component, and primarily denoising the at least one IMF component to obtain at least one target IMF component.
In the step, short-time Fourier transform power spectrum analysis is carried out on the noise-containing transformer acoustic signal collected by the ith microphone, and the minimum frequency in the noise-containing transformer acoustic signal is found out and used as the cut-off frequency f L of corresponding signal decomposition, and is used as the lower limit of signal decomposition.
Adding standard Gaussian white noise into the noise-containing transformer signal to obtain a mixed noise signal, wherein the expression is as follows:
Mi(t)＝mi(t)+nw(t),
Wherein M i (t) is the ith noise-containing transformer acoustic signal, n w (t) is standard white gaussian noise, and M i (t) is the ith mixed noise signal;
EEMD (ensemble empirical mode decomposition) is performed on the mixed noise signal to obtain an IMF (intrinsic mode) component and a residual error, wherein the expression is as follows:
Wherein, I i,j (t) is the J-th IMF component obtained by decomposing the I-th noise-containing transformer acoustic signal, r i (t) is the residual error obtained by decomposing the I-th noise-containing transformer acoustic signal EEMD, and J is the order number of the IMF component group;
Performing short-time Fourier transform power spectrum analysis on the decomposed I i,j (t) to obtain a target frequency f max of I i,j (t), comparing the target frequency f max with a cut-off frequency f L, if the target frequency f max is smaller than the cut-off frequency f L, ending the decomposition, otherwise, continuing the decomposition, and finally obtaining an N-order IMF component of the ith noise-containing transformer acoustic signal after EEMD decomposition;
And (3) averaging all IMF components, and obtaining the integral average of all IMF components of each order after the decomposition of the acoustic signal of the noise-containing transformer after the Gaussian white noise influence is counteracted, wherein the expression is as follows:
Wherein, C k,j is the integral average of the j-th IMF component of the K-th decomposition, C i,j is the j-th IMF component of the i-th noise-containing transformer acoustic signal, and M is the number of noise-containing transformer acoustic signals;
Calculating residual error r (t) obtained by EEMD decomposition of noise-containing transformer acoustic signals, wherein the expression is as follows:
and respectively calculating correlation coefficients of each IMF component and the corresponding noise-containing transformer acoustic signal, wherein the expression is as follows:
Wherein ρ k is a correlation coefficient value of a kth IMF component and a corresponding noise-containing transformer acoustic signal, C k (t) is a kth IMF component, C abs is an average value of the kth IMF component, x k (t) is a transformer acoustic signal, x abs is an average value of the transformer acoustic signal, and K' is an IMF component order;
Setting a threshold mu, reserving an IMF component with a correlation coefficient value larger than the threshold mu, obtaining at least one target IMF component, and calculating an expression of the threshold mu as follows:
In the formula, Is the average value of ρ k.
It should be noted that, after the mixed signal matrix to be processed is decomposed by adopting an improved EEMD algorithm based on a minimum lower limit frequency to obtain at least one IMF component, a voiceprint order recognition model based on a GRU network is constructed, dimension recognition is performed on the noise-containing transformer signal according to the voiceprint order recognition model, and whether the dimension of the noise-containing transformer signal is greater than the dimension of a certain IMF component is judged, wherein the expression for calculating the dimension of the noise-containing transformer signal is as follows:
Se＝GRU(mi(t)),
Wherein m i (t) is the ith noise-containing transformer acoustic signal, GRU is a gating cycle unit, and S e is the dimension of the noise-containing transformer signal;
if the dimension of the IMF component is larger than the dimension of the IMF component, decomposing the residual error corresponding to the IMF component based on an improved EEMD algorithm again;
And if the dimension of the IMF component is not larger than the dimension of the IMF component, performing preliminary denoising on the IMF component.
Specifically, through a transformer voiceprint simulation laboratory, simulating a site transformer noiseless fault voiceprint, a fault voiceprint containing a single interference source, a fault voiceprint containing two interference sources and more than two noise interference sources, and constructing a voiceprint dimension identification model database according to the fault voiceprint.
Because of the strong timing characteristics of the transformer acoustic signals, it is contemplated to use a recurrent neural network that is good at analyzing the timing relationship between discrete data sequences. The GRU network is a simplified long-short term memory (LSTM network, which combines an input gate and a forgetting gate in the LSTM network into an update gate to reduce parameter amount and alleviate overfitting, and the computing efficiency is improved while the prediction effect close to the LSTM is maintained, the specific process for constructing the voiceprint dimension identification model comprises the following steps:
And training a voiceprint dimension identification model based on the GRU network by using a voiceprint dimension identification model database, respectively constructing a voiceprint dimension identification model and a voiceprint dimension prediction model in a model training stage, setting training super-parameters including batch size, learning rate, iteration number and the like, and verifying the identification accuracy of the model and the prediction error of the prediction model.
The expression of the reset gate gating signal r t' of the GRU network is:
r′t＝Sig(WrXt+Urht-1),
Wherein, W r is a reset gate connection weight matrix, X t is the current moment input, h t-1 is the t-1 moment hidden gate output, U r is a connection weight matrix of the reset gate and the previous time step hidden state, the characteristics related to the reset gate are extracted from the previous time step hidden state, and Sig is a Sigmoid function;
The expression of the updated gate gating signal Z t for the GRU network is:
Zt＝Sig{WZ[ht-1,xt]},
wherein W Z is an updated gate connection weight matrix;
c t represents memory information at time t, and the expression is:
ct＝tanh(WcXt+Uc(rt′⊙ht-1)),
Wherein, W c and U c each represent a learnable weight parameter, and as such, as ";
Wherein h t is the hidden gate output at time t, and the expression is:
ht＝ztht-1+(1-zt)ct。
step S103, screening the previous S e order target IMF component with the maximum correlation with the noise-containing transformer signal based on the correlation maximum principle to form an ascending order signal group D u, wherein S e is the dimension of the noise-containing transformer signal.
In this step, the expression for calculating the correlation is:
wherein cov (·) is a covariance function, σ (·) is a standard deviation function, As a correlation coefficient, m (t) is a noisy transformer signal, C i,j (t) is an IMF component of the j-th order of the i-th noisy transformer acoustic signal;
The dimension-increasing signal group is equivalent to the original underdetermined voiceprint signal group obtained by linear transformation. Because a nonlinear process does not exist, the linear requirement of the FastICA algorithm cannot be destroyed by using the IMF signal group as input to perform blind source separation, and the signal group after dimension rising solves the reversibility constraint of the mixed signal and the uniqueness constraint of the separation signal, and solves the key problem in underdetermined blind source separation of the voiceprint signal of the transformer.
Further, the expression of the ascending order signal group D u is:
Where C 1 (t) is the first before IMF component with the greatest correlation to the noise-containing transformer signal, C 2 (t) is the second before IMF component with the greatest correlation to the noise-containing transformer signal, The first S e th IMF component with the largest correlation of the noise-containing transformer signals.
And step S104, performing blind source separation on the ascending order signal group D u based on a preset N-MNE-ICA model to obtain a blind source signal.
In this step, since the conventional negative entropy calculation method is too difficult, the accuracy of the empirical formula is not high enough, and thus an improvement is required, namely, an improved independent component analysis algorithm (N-MNE-ICA) based on Newton-maximum information negative entropy (Newton-Max Negative Entropy, N-MNE). And inputting the signal group after the dimension rising into an ICA algorithm module based on N-MNE improvement, so as to reduce the calculation complexity of the negative entropy, improve the convergence rate and increase the separation precision.
The maximum entropy principle assumes that the information does not make any unknown assumptions and considers the unknown events as equiprobable events. Substituting the maximum entropy principle into a negative entropy formula to obtain the maximum approximate negative entropy, wherein the expression of the maximum approximate negative entropy is as follows:
Wherein J max (y) is the maximum approximate negative entropy, N, E is the total number of samples and mathematical expectation, k i is a weight factor, is a normal number, G i is a non-quadratic function, y and v are variables with mean value of 0 and variance of 1;
To obtain the maximum approximate negative entropy value of y=b T x, the non-gaussian property of y is maximized by E { G i (y) }, the constraint is expressed as:
Where B is the convergence threshold function, B T is the transpose of the convergence threshold function B, χ is the input signal, J G is the input signal, G 'i is the derivative of the non-quadratic function, J G is the non-Gaussian measure function based on the input signal, and β' is the Lagrangian multiplier operator for constraint optimization.
The maximum entropy theory and the negative entropy formula have proved that an optimal solution is necessary, so that the constraint equation can be established. Assuming that the optimal solution is B 0, solving constraint conditions to obtain:
In the formula, Transpose of the optimal solution;
The tangent to the curve may be used to approximate the descriptive curve, and thus also approximate the root of the solution tangent, according to newton's limit theorem. If the accuracy requirement is not met, newton's method is continued until convergence. The general formula for newton's limit method is:
Where x n is the estimated value of the nth iteration, x n+1 is the estimated value of the n+1st iteration, f (x n) is the newton's limit function, and f ' (x n) is the derivative of the newton's limit function.
The iterative formula of the newton's limit method is:
Wherein B * is an iterative function, G' i is a second derivative of a non-quadratic function, and x is an input signal;
The application of the Newton limit method not only simplifies complex calculation such as solving probability density functions, but also reduces error accumulation possibly caused by high power calculation in an empirical formula. The method converts the original complex statistical mathematical structure into a relatively simple iterative root-finding structure, thereby improving the accuracy and efficiency of calculation.
In view of facilitating subsequent blind source separation calculations, a de-averaging process should be performed prior to the calculation. After the data is subjected to de-averaging, the data is distributed near the coordinate axis, and the function obtained through random initialization can approach the objective function more quickly. The expression for de-averaging is:
Wherein Z is the transformer voiceprint signal sample after the de-averaging, alpha is the characteristic vector of the transformer voiceprint signal sample X, and lambda is the characteristic value of X.
Setting an iteration function B *, wherein the expression is as follows:
B*＝E{ZG′i(BTZ)}-E{G″i(BTZ)}B,
defining a convergence threshold function B, and expressing as follows:
It should be noted that, in the blind source separation process of the ascending signal group D u based on the preset N-MNE-ICA model, the separation result is evaluated by using the signal similarity coefficient as an evaluation index, and the expression for calculating the signal similarity coefficient is as follows:
Where ζ ij is a signal similarity coefficient of a j-th signal in an i-th signal β packet in an α packet, ζ ij = [0,1], when ζ ij =1, it is considered that the i-th signal is identical to the j-th signal, and ζ ij =0 is opposite, α i (t) is the i-th signal in the α packet, β j (t) is the j-th signal in the β packet, ζ (α i,βj) is a signal similarity coefficient of α i (t) and β j (t), and N' is an upper limit of the acoustic signal time t.
In summary, the method of the application inputs noise-containing transformer acoustic signals to obtain a mixed to-be-processed signal matrix, the EEMD algorithm based on minimum lower limit frequency improvement decomposes the underdetermined mixed to-be-processed audio signal matrix to obtain IMF components and performs preliminary denoising, the voiceprint dimension identification model based on the GRU network estimates the source signal dimension, the dimension of the dimension-increasing acoustic signal matrix is increased, whether the dimension-increasing signal matrix is underdetermined or not is judged, if underdetermined, residual errors are extracted, if underdetermined, the dimension-increasing signal group is selected based on the correlation maximum principle, and the ICA algorithm component blind source signal is improved based on N-MNE, thereby effectively improving the processing effect of the transformer acoustic signals.
Referring to fig. 2, a block diagram of a transformer acoustic signal processing system based on the improved ICA algorithm of the present application is shown.
As shown in fig. 2, the transformer acoustic signal processing system 200 includes a construction module 210, a decomposition module 220, a screening module 230, and a separation module 240.
The system comprises a construction module 210, a decomposition module 220, a screening module 230, a separation module 240 and a blind source separation module, wherein the construction module 210 is configured to construct a mixed signal matrix to be processed according to an acquired noise-containing transformer signal, the decomposition module 220 is configured to decompose the mixed signal matrix by adopting an improved EEMD algorithm based on a minimum lower limit frequency to obtain at least one IMF component, the at least one IMF component is subjected to preliminary denoising to obtain at least one target IMF component, the screening module 230 is configured to screen a front S e -order target IMF component with the maximum correlation with the noise-containing transformer signal based on a correlation maximum principle to form an ascending order signal group D u, S e is the dimension of the noise-containing transformer signal, and the separation module 240 is configured to perform blind source separation on the ascending order signal group D u based on a preset N-MNE-ICA model to obtain a blind source signal.
It should be understood that the modules depicted in fig. 2 correspond to the various steps in the method described with reference to fig. 1. Thus, the operations and features described above for the method and the corresponding technical effects are equally applicable to the modules in fig. 2, and are not described here again.
In other embodiments, embodiments of the present invention further provide a computer readable storage medium having stored thereon a computer program, which when executed by a processor, causes the processor to perform the method for processing acoustic signals of a transformer based on the improved ICA algorithm in any of the method embodiments described above;
As one embodiment, the computer-readable storage medium of the present invention stores computer-executable instructions configured to:
Constructing a mixed signal matrix to be processed according to the acquired noise-containing transformer signals;
Decomposing the mixed signal matrix to be processed by adopting an improved EEMD algorithm based on the minimum lower limit frequency to obtain at least one IMF component, and primarily denoising the at least one IMF component to obtain at least one target IMF component;
Screening a front S e -order target IMF component with the maximum correlation with the noise-containing transformer signal based on a correlation maximum principle to form an ascending order signal group D u, wherein S e is the dimension of the noise-containing transformer signal;
And performing blind source separation on the ascending order signal group D u based on a preset N-MNE-ICA model to obtain a blind source signal.
The computer-readable storage medium may include a storage program area that may store an operating system, an application program required for at least one function, and a storage data area that may store data created according to the use of the transformer acoustic signal processing system based on the modified ICA algorithm, and the like. In addition, the computer-readable storage medium may include high-speed random access memory, and may also include memory, such as at least one magnetic disk storage device, flash memory device, or other non-volatile solid-state storage device. In some embodiments, the computer readable storage medium optionally includes memory remotely located with respect to the processor, which may be connected to the improved ICA algorithm based transformer acoustic signal processing system via a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
Fig. 3 is a schematic structural diagram of an electronic device according to an embodiment of the present invention, and as shown in fig. 3, the device includes a processor 310 and a memory 320. The electronic device may further comprise input means 330 and output means 340. The processor 310, memory 320, input device 330, and output device 340 may be connected by a bus or other means, for example in fig. 3. Memory 320 is the computer-readable storage medium described above. The processor 310 performs various functional applications of the server and data processing, i.e., implements the transformer acoustic signal processing method based on the improved ICA algorithm of the above-described method embodiments, by running nonvolatile software programs, instructions, and modules stored in the memory 320. The input device 330 may receive input numeric or character information and generate key signal inputs related to user settings and function control of the improved ICA algorithm-based transformer acoustic signal processing system. The output device 340 may include a display device such as a display screen.
The electronic equipment can execute the method provided by the embodiment of the invention, and has the corresponding functional modules and beneficial effects of the execution method. Technical details not described in detail in this embodiment may be found in the methods provided in the embodiments of the present invention.
As an embodiment, the electronic device is applied to a transformer acoustic signal processing system based on an improved ICA algorithm, and is used for a client, and comprises at least one processor, and a memory communicatively connected with the at least one processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to:
Constructing a mixed signal matrix to be processed according to the acquired noise-containing transformer signals;
Decomposing the mixed signal matrix to be processed by adopting an improved EEMD algorithm based on the minimum lower limit frequency to obtain at least one IMF component, and primarily denoising the at least one IMF component to obtain at least one target IMF component;
Screening a front S e -order target IMF component with the maximum correlation with the noise-containing transformer signal based on a correlation maximum principle to form an ascending order signal group D u, wherein S e is the dimension of the noise-containing transformer signal;
And performing blind source separation on the ascending order signal group D u based on a preset N-MNE-ICA model to obtain a blind source signal.
From the above description of the embodiments, it will be apparent to those skilled in the art that the embodiments may be implemented by means of software plus necessary general hardware platforms, or of course may be implemented by means of hardware. Based on such understanding, the foregoing technical solutions may be embodied essentially or in part in the form of a software product, which may be stored in a computer-readable storage medium, such as a ROM/RAM, a magnetic disk, an optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, a server, or a network device, etc.) to perform the various embodiments or methods of some parts of the embodiments.
It should be noted that the above-mentioned embodiments are merely for illustrating the technical solution of the present invention, and not for limiting the same, and although the present invention has been described in detail with reference to the above-mentioned embodiments, it should be understood by those skilled in the art that the technical solution described in the above-mentioned embodiments may be modified or some technical features may be equivalently replaced, and these modifications or substitutions do not make the essence of the corresponding technical solution deviate from the spirit and scope of the technical solution of the embodiments of the present invention.
Claims (9)
1. A method for processing an acoustic signal of a transformer based on an improved ICA algorithm, comprising:
Constructing a mixed signal matrix to be processed according to the acquired noise-containing transformer signals;
Decomposing the mixed signal matrix to be processed by adopting an improved EEMD algorithm based on the minimum lower limit frequency to obtain at least one IMF component, and primarily denoising the at least one IMF component to obtain at least one target IMF component;
Screening a front S e -order target IMF component with the maximum correlation with the noise-containing transformer signal based on a correlation maximum principle to form an ascending order signal group D u, wherein S e is the dimension of the noise-containing transformer signal;
And performing blind source separation on the ascending order signal group D u based on a preset N-MNE-ICA model to obtain a blind source signal.
2. The method of claim 1, wherein decomposing the mixed signal matrix with the modified EEMD algorithm based on the minimum lower limit frequency to obtain at least one IMF component comprises:
Adding standard Gaussian white noise into the noise-containing transformer signal to obtain a mixed noise signal, wherein the expression is as follows:
Mi(t)＝mi(t)+nw(t),
Wherein M i (t) is the ith noise-containing transformer acoustic signal, n w (t) is standard white gaussian noise, and M i (t) is the ith mixed noise signal;
EEMD (ensemble empirical mode decomposition) is performed on the mixed noise signal to obtain an IMF (intrinsic mode) component and a residual error, wherein the expression is as follows:
Wherein, I i,j (t) is the J-th IMF component obtained by decomposing the I-th noise-containing transformer acoustic signal, r i (t) is the residual error obtained by decomposing the I-th noise-containing transformer acoustic signal EEMD, and J is the order number of the IMF component group;
Performing short-time Fourier transform power spectrum analysis on the decomposed I i,j (t) to obtain a target frequency f max of I i,j (t), comparing the target frequency f max with a cut-off frequency f L, if the target frequency f max is smaller than the cut-off frequency f L, ending the decomposition, otherwise, continuing the decomposition, and finally obtaining an N-order IMF component of the ith noise-containing transformer acoustic signal after EEMD decomposition;
And (3) averaging all IMF components, and obtaining the integral average of all IMF components of each order after the decomposition of the acoustic signal of the noise-containing transformer after the Gaussian white noise influence is counteracted, wherein the expression is as follows:
Wherein, C k,j is the integral average of the j-th IMF component of the K-th decomposition, C i,j is the j-th IMF component of the i-th noise-containing transformer acoustic signal, and M is the number of noise-containing transformer acoustic signals;
Calculating residual error r (t) obtained by EEMD decomposition of noise-containing transformer acoustic signals, wherein the expression is as follows:
3. The method of claim 2, wherein said preliminary denoising of said at least one IMF component to obtain at least one target IMF component comprises:
and respectively calculating correlation coefficients of each IMF component and the corresponding noise-containing transformer acoustic signal, wherein the expression is as follows:
Wherein ρx is a correlation coefficient value of the kth IMF component and the corresponding noise-containing transformer acoustic signal, C k (t) is the kth IMF component, C abs is an average value of the kth IMF component, x k (t) is the transformer acoustic signal, x abs is an average value of the transformer acoustic signal, and K' is an IMF component order;
Setting a threshold mu, reserving an IMF component with a correlation coefficient value larger than the threshold mu, obtaining at least one target IMF component, and calculating an expression of the threshold mu as follows:
In the formula, Is the average value of ρ k.
4. The method of claim 1, wherein after decomposing the mixed signal matrix to be processed using an improved EEMD algorithm based on a minimum lower limit frequency to obtain at least one IMF component, the method further comprises:
establishing a voiceprint order identification model based on a GRU network, carrying out dimension identification on the noise-containing transformer signal according to the voiceprint order identification model, and judging whether the dimension of the noise-containing transformer signal is larger than the dimension of a certain IMF component, wherein the expression for calculating the dimension of the noise-containing transformer signal is as follows:
Se＝GRU(mi(t)),
Wherein m i (t) is the ith noise-containing transformer acoustic signal, GRU is a gating cycle unit, and S e is the dimension of the noise-containing transformer signal;
if the dimension of the IMF component is larger than the dimension of the IMF component, decomposing the residual error corresponding to the IMF component based on an improved EEMD algorithm again;
And if the dimension of the IMF component is not larger than the dimension of the IMF component, performing preliminary denoising on the IMF component.
5. The method for processing acoustic signals of a transformer based on the improved ICA algorithm according to claim 1, wherein the expression of the set of up-scaled signals D u is:
Where C 1 (t) is the first before IMF component with the greatest correlation to the noise-containing transformer signal, C 2 (t) is the second before IMF component with the greatest correlation to the noise-containing transformer signal, The first S e th IMF component with the largest correlation of the noise-containing transformer signals.
6. The method for processing acoustic signals of a transformer based on an improved ICA algorithm according to claim 1, wherein in the process of blind source separation of the ascending signal group D u based on a preset N-MNE-ICA model, the separation result is evaluated by using a signal similarity coefficient as an evaluation index, and the expression for calculating the signal similarity coefficient is as follows:
Where ζ ij is a signal similarity coefficient of a j-th signal in an i-th signal β packet in an α packet, ζ ij = [0,1], when ζ ij =1, it is considered that the i-th signal is identical to the j-th signal, and ζ ij =0 is opposite, α i (t) is the i-th signal in the α packet, β j (t) is the j-th signal in the β packet, ζ (α i,βj) is a signal similarity coefficient of α i (t) and β j (t), and N' is an upper limit of the acoustic signal time t.
7. A transformer acoustic signal processing system based on an improved ICA algorithm, comprising:
the construction module is configured to construct a mixed signal matrix to be processed according to the acquired noise-containing transformer signals;
the decomposition module is configured to decompose the mixed signal matrix to be processed by adopting an improved EEMD algorithm based on the minimum lower limit frequency to obtain at least one IMF component, and perform preliminary denoising on the at least one IMF component to obtain at least one target IMF component;
The screening module is configured to screen a front S e -order target IMF component with the maximum correlation with the noise-containing transformer signal based on the correlation maximum principle to form an ascending order signal group D u, wherein S e is the dimension of the noise-containing transformer signal;
and the separation module is configured to perform blind source separation on the ascending order signal group D u based on a preset N-MNE-ICA model to obtain a blind source signal.
8. An electronic device comprising at least one processor and a memory communicatively coupled to the at least one processor, wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-6.
9. A computer readable storage medium, on which a computer program is stored, characterized in that the program, when being executed by a processor, implements the method of any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202411811745.6A CN119832932B (en) | 2024-12-10 | 2024-12-10 | Transformer acoustic signal processing method and system based on improved ICA algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202411811745.6A CN119832932B (en) | 2024-12-10 | 2024-12-10 | Transformer acoustic signal processing method and system based on improved ICA algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN119832932A true CN119832932A (en) | 2025-04-15 |
| CN119832932B CN119832932B (en) | 2025-11-28 |
Family
ID=95291285
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202411811745.6A Active CN119832932B (en) | 2024-12-10 | 2024-12-10 | Transformer acoustic signal processing method and system based on improved ICA algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN119832932B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2007295085A (en) * | 2006-04-21 | 2007-11-08 | Kobe Steel Ltd | Sound source separation apparatus, and sound source separation method |
| CN104375973A (en) * | 2014-11-24 | 2015-02-25 | 沈阳建筑大学 | Blind source signal denoising method based on ensemble empirical mode decomposition |
-
2024
- 2024-12-10 CN CN202411811745.6A patent/CN119832932B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2007295085A (en) * | 2006-04-21 | 2007-11-08 | Kobe Steel Ltd | Sound source separation apparatus, and sound source separation method |
| CN104375973A (en) * | 2014-11-24 | 2015-02-25 | 沈阳建筑大学 | Blind source signal denoising method based on ensemble empirical mode decomposition |
Non-Patent Citations (1)
| Title |
|---|
| 王广真等: "基于重复模式提取与高斯混合模型的变压器故障声纹识别", 广东电力, vol. 36, no. 1, 31 January 2023 (2023-01-31), pages 126 - 134 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN119832932B (en) | 2025-11-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111193256A (en) | Power load prediction method based on variational modal decomposition and gated cyclic unit | |
| CN114363195A (en) | Network flow prediction early warning method for time and spectrum residual convolution network | |
| CN113836823A (en) | Load combination prediction method based on load decomposition and optimized bidirectional long-short term memory network | |
| CN117407797B (en) | Equipment fault diagnosis method and model construction method based on incremental learning | |
| CN109145516B (en) | A fault identification method for analog circuits based on an improved extreme learning machine | |
| CN112986950A (en) | Single-pulse laser radar echo feature extraction method based on deep learning | |
| CN118709095B (en) | Partial discharge type identification method and system based on multi-feature extraction and fusion | |
| Ni et al. | A hybrid method for short-term sensor data forecasting in Internet of Things | |
| CN117216659A (en) | Atmospheric particle source analysis method and system based on single-particle aerosol mass spectrometry | |
| CN114861792B (en) | Complex power grid key node identification method based on deep reinforcement learning | |
| CN116665483A (en) | A Novel Method for Predicting Remaining Parking Spaces | |
| CN117875365A (en) | A short-term power load forecasting method based on SMBO-BiGRU-Attention | |
| CN119535124A (en) | Cable partial discharge identification method and equipment based on improved decision tree-rotation forest | |
| CN113485986B (en) | Electric power data restoration method | |
| CN119598387A (en) | Photovoltaic power generation prediction method and device based on dynamic decomposition and digital-analog fusion | |
| CN117780735A (en) | Fault diagnosis method and device and electronic equipment | |
| CN117952646A (en) | A TCN-LSTM combined electricity price prediction method based on wavelet packet decomposition | |
| CN119832932B (en) | Transformer acoustic signal processing method and system based on improved ICA algorithm | |
| Kumar et al. | An intuitive explanation of graph signal processing-based electrical load disaggregation | |
| CN114397524A (en) | Grid equipment fault identification method and system based on GA and SE optimization VMD parameters | |
| CN113595242A (en) | Non-invasive load identification method based on deep CNN-HMM | |
| CN113782057A (en) | Transformer status real-time monitoring method and system based on voiceprint analysis | |
| CN118820995A (en) | A photovoltaic power generation system current transformer state analysis and prediction method and system | |
| CN119271971A (en) | Ultra-short-term wind power forecasting method based on quadratic decomposition and TCN-ECA-BiGRU | |
| CN118941854A (en) | Power quality disturbance classification method based on Markov switching field and particle swarm optimization |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |