Method, device, electronic device and storage medium for identifying abnormal coking sound signals in annular pipelines

Abstract

The present invention discloses a method and device for identifying abnormal coking acoustic signals in a circular pipeline. The method comprises the following steps: A. Collect waveform flow data of acoustic emission signals at different flow rates and different coking positions in an analog circular pipeline; B. Perform singular spectrum and high-order spectrum fusion analysis on the waveform flow data, and extract the intensity and phase characteristics in the frequency domain of the denoised high-order spectrum data; C. Use the intensity and phase characteristics in the frequency domain as input variables of the coking abnormal acoustic signal recognition model, and train the model; D. Identify the coking conditions of the circular pipeline at different flow rates through the trained model. The present invention applies the acoustic emission principle, extracts preprocessed characteristic parameters from the sample data of abnormal coking acoustic signals to train the model, so as to be able to identify abnormal coking acoustic signals under various actual working conditions in real time and accurately, and timely and effectively prevent the occurrence of abnormal shutdowns and accidents caused by coking.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial non-destructive testing, and particularly relates to a method, device, electronic device and storage medium for identifying abnormal acoustic signals of coking in annular pipelines. Background Art

[0002] The coking phenomenon widely exists in chemical processes such as fluid catalytic cracking (FCC) of crude oil, methanol to olefins (MTO), and natural gas to olefins (GTO). Catalyst coking refers to the formation of coke deposits in the active centers of the catalyst and in the catalyst pores, resulting in a decrease or loss of catalyst activity. Catalyst coking easily leads to coking on the reactor wall. Coke nuclei are formed on the catalyst and gradually coke on the wall, or unreacted macromolecular substances adhere to the wall and gradually coke. Once the coke chunks on the wall fall off, they are likely to block the pipe section or impact the internal components, causing damage to the internal components of the reactor and resulting in unplanned shutdowns and serious accidents. Inside the reactor, the part that forms a space similar to an annular pipeline (i.e., between the inner wall of the outer pipe and the outer wall of the inner pipe) with the internal components is more prone to coking. Currently, it is difficult to detect the coking situation inside the reactor during the production process.

[0003] Acoustic emission (AE) is a phenomenon in which local materials emit transient elastic stress waves due to the rapid release of energy. When the catalyst flows inside the reactor and rubs against and impacts the reactor wall, acoustic emission waves are formed. When coking occurs on the wall, the acoustic emission waves will be abnormal, that is, coking noise. Identifying the coking state of the wall through the acoustic emission signals monitored on the wall is the key and difficulty in monitoring the operating state of the reactor. Applying the acoustic emission principle to monitor the coking rate at key positions of the reactor and locate the coking position is of great significance for preventing accidents.

[0004] There are prior art solutions for identifying coking using the acoustic emission principle. For example, Chinese Patent Application CN102192955A discloses a method for detecting the coking amount of solid particles in a reactor. This method mainly solves the problem that in the prior art, it is impossible to accurately and real-time online measure the coke deposition amount of the catalyst in the reactor. By receiving the acoustic emission signals generated by the impact of solid particles at any position inside the reactor on the reactor wall, analyzing the power spectrum diagram after fast Fourier transform (FFT), and combining methods such as smoothing and noise reduction, the coking amount of the catalyst particles is quantitatively determined, and the technical solution for realizing real-time online monitoring of the processes such as catalyst coking and burning solves this problem well and can be used in industrial production such as fluid catalytic cracking industry (FCC), methanol to olefins process (MTO), and natural gas to olefins (GTO). This method only judges the coking degree through the power spectrum, and the collected acoustic wave signals are still interfered by other noises (noises not generated by coking), and the noise reduction effect of the data is poor, and the judgment accuracy will be greatly reduced.

[0005] Therefore, there is an urgent need for a method that applies the principle of acoustic emission technology to effectively identify abnormal acoustic signals of coking in annular pipelines through an algorithm model, and can greatly improve the recognition accuracy of coking conditions in annular pipelines with different flow rates and different diameters, so as to monitor the coking rate at key positions of the reactor, locate the coking position, and prevent accidents.

[0006] The information disclosed in this background art section is only intended to enhance the overall understanding of the present invention and should not be regarded as an admission or any form of implication that this information constitutes prior art already known to those of ordinary skill in the art. Summary of the Invention

[0007] The object of the present invention is to provide a method and device for identifying abnormal acoustic signals of coking in annular pipelines. By collecting waveform stream data of abnormal sound signals at the coking parts of a simulated annular pipeline, extracting singular spectrum data from the waveform stream data, then using the singular spectrum data as the input of a high-order spectrum program, and further using the data obtained by fusing and processing the singular spectrum and the high-order spectrum as the input of an algorithm model, the coking conditions of annular pipelines with different diameters under different flow rate states are identified, which can effectively improve the recognition accuracy.

[0008] To achieve the above object, according to the first aspect of the present invention, a method for identifying abnormal acoustic signals of coking in annular pipelines is provided, including the following steps: A. Collect waveform stream data of acoustic emission signals at different flow rates and different coking parts in a simulated annular pipeline; B. Perform fusion analysis and processing on the waveform stream data for singular spectrum and high-order spectrum, and extract the intensity and phase features in the frequency domain of the denoised high-order spectrum data; C. Use the intensity and phase features in the frequency domain as input variables of a coking abnormal acoustic signal recognition model, and train the model; D. Identify the coking conditions of the annular pipeline under different flow rates through the trained model.

[0009] Further, in the above technical solution, the simulated annular pipeline in step A may include multiple outer pipes with the same diameter and inner pipes with different diameters, and simulated coking materials are adhered to the inner wall of the outer pipe of the annular pipeline, the outer wall of the inner pipe, and between the inner wall of the outer pipe and the outer wall of the inner pipe respectively.

[0010] Further, in the above technical solution, the acquisition duration of the waveform stream data is 1 second.

[0011] Further, in the above technical solution, the singular spectrum and high-order spectrum fusion analysis and processing of the waveform stream data in step B may specifically include: performing singular value decomposition through a constructed trajectory matrix, reconstructing the separated singular matrix into multiple one-dimensional data components; removing the component with the lowest energy among the multiple one-dimensional data components and reconstructing the remaining data components into a one-dimensional data component as the original input signal of the high-order spectrum program; performing Fourier transform of the K-order cumulant on the input signal to obtain a K-order spectrum; and extracting the intensity and phase features in the frequency domain from the spectrogram of the K-order spectrum.

[0012] Further, in the above technical solution, the intensity feature in the frequency domain may be the maximum amplitude and the frequency corresponding to the maximum amplitude; the phase feature in the frequency domain may be the maximum phase and the frequency corresponding to the maximum phase.

[0013] Further, in the above technical solution, the order of the K-order spectrum is preferably three.

[0014] Further, in the above technical solution, the multiple one-dimensional data components can be expressed as:

[0015] SVD[X1 X2 X3 … X n-1 =X{i};

[0016] where SVD is the singular spectrum and X{i} is the signal after reconstruction of the singular spectrum component.

[0017] Further, in the above technical solution, the coking conditions in step D include but are not limited to: whether there is a coking phenomenon, the coking site, and the shape and size of the coking, etc.

[0018] Further, in the above technical solution, the coking abnormal sound signal recognition model can adopt a neural network model, preferably a BP neural network model.

[0019] According to the second aspect of the present invention, the present invention provides a coking abnormal sound signal recognition device for an annular pipeline, including: a waveform stream acquisition module for acquiring waveform stream data of acoustic emission signals at different flow rates and different coking sites in an analog annular pipeline; a feature parameter extraction module for performing singular spectrum and high-order spectrum fusion analysis and processing on the waveform stream data and extracting the intensity and phase features in the frequency domain of the denoised high-order spectrum data; a model training module for using the intensity and phase features in the frequency domain as input variables of the coking abnormal sound signal recognition model and training the model; and a coking recognition module for recognizing the coking conditions of the annular pipeline at different flow rates through the trained model.

[0020] Further, in the above technical solution, the feature parameter extraction module may specifically include: a singular value component construction sub-module, configured to perform singular value decomposition on a constructed trajectory matrix and reconstruct the separated singular matrix into multiple one-dimensional data components; a high-order spectrum input selection sub-module, configured to remove the component with the lowest energy among the multiple one-dimensional data components and reconstruct the remaining data components into a one-dimensional data component as the original input signal of the high-order spectrum program; a K-order spectrum acquisition sub-module, configured to perform Fourier transform on the input signal for the K-order cumulant to obtain a K-order spectrum; an intensity and phase feature extraction sub-module, configured to extract the intensity and phase features in the frequency domain from the spectrogram of the K-order spectrum.

[0021] Further, in the above technical solution, the intensity and phase features in the frequency domain may include the following four feature parameters: the maximum amplitude and the frequency corresponding to the maximum amplitude; the maximum phase and the frequency corresponding to the maximum phase.

[0022] According to the third aspect of the present invention, there is provided an electronic device for identifying abnormal coking sound signals in an annular pipeline, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and when the instructions are executed by the at least one processor, the at least one processor is caused to execute the method for identifying abnormal coking sound signals in an annular pipeline as described above.

[0023] According to the fourth aspect of the present invention, there is provided a non-transitory computer-readable storage medium storing computer-executable instructions for causing a computer to execute the method for identifying abnormal coking sound signals in an annular pipeline as described above.

[0024] Compared with the prior art, the present invention has one or more of the following beneficial effects:

[0025] 1) The method of the present invention can identify whether there is coking on the wall of the container (especially the wall of the annular pipeline), and can collect both the acoustic emission signals of the coking part and the acoustic emission signals of the non-coking part as sample data, which is convenient for subsequent comparison and identification.

[0026] 2) Acoustic emission signals belong to transient signals, and conform to the law that the longer the sampling length, the higher the frequency resolution. It is found by the inventor that in order to effectively improve the frequency resolution, compared with collecting waveform segments in the prior art, the present invention adopts a signal acquisition method of waveform stream.

[0027] 3) In order to suppress the mixed noise components (i.e., noise signals unrelated to the abnormal coking sound signal) and make the signal have higher recognition, the inventor found that the analysis and processing method of fusing singular spectrum and high-order spectrum can more effectively remove the noise with lower energy. By extracting the intensity and phase characteristic parameters in the frequency domain and suppressing the mixed noise components at the same time, the accuracy and recognition of the signal characteristics can be greatly improved, thereby improving the accuracy of coking part recognition;

[0028] 4) Since the sample data collects the signal in three cases respectively: the non-coking state, the inner wall of the outer pipe, the outer wall of the inner pipe, and the inner wall of the outer pipe and the outer wall of the inner pipe in the annular pipe space, it is possible to know whether there is a coking phenomenon by comparing the signal data of non-coking and coking; by setting different inner pipe diameters in the simulation experiment, annular pipes of different sizes can be simulated, and by adjusting the gas flow rate, the coking states in various situations such as different flow rates, different annular pipe sizes, and different coking patterns can be simulated.

[0029] The above description is only an overview of the technical solution of the present invention. In order to understand the technical means of the present invention more clearly and be able to implement it according to the content of the specification, and in order to make the above and other purposes, technical features and advantages of the present invention more understandable, one or more preferred embodiments are listed below and described in detail with the accompanying drawings as follows. Brief Description of the Drawings

[0030] Figure 1 It is a schematic flow chart of the method for identifying abnormal coking sound signals in an annular pipe in Embodiment 1 of the present invention.

[0031] Figure 2 It is a schematic structural diagram of the device for identifying abnormal coking sound signals in an annular pipe in Embodiment 2 of the present invention.

[0032] Figure 3 It is a schematic structural diagram of the electronic device for identifying abnormal coking sound signals in an annular pipe in Embodiment 3 of the present invention. Detailed Embodiments

[0033] The following will describe in detail the specific embodiments of the present invention with reference to the accompanying drawings, but it should be understood that the protection scope of the present invention is not limited by the specific embodiments.

[0034] Unless otherwise clearly stated, in the whole specification and claims, the term "comprising" or its variations such as "including" or "having" etc. will be understood to include the stated elements or components, without excluding other elements or other components.

[0035] In this text, for the convenience of description, spatial relative terms such as "below", "beneath", "under", "above", "over", "on" etc. may be used to describe the relationship between one element or feature and another element or feature in the drawings. It should be understood that spatial relative terms are intended to encompass different directions of an object in use or operation in addition to the directions depicted in the figures. For example, if the object in the figure is flipped, an element described as "below" or "under" other elements or features will be oriented "above" the element or feature. Thus, the exemplary term "below" can encompass both the below and above directions. The object can also have other orientations (rotated 90 degrees or other orientations) and the spatial relative terms used herein should be interpreted accordingly.

[0036] In this text, terms such as "first", "second" etc. are used to distinguish two different elements or parts, and are not used to define a specific position or relative relationship. In other words, in some embodiments, the terms "first", "second" etc. can also be interchanged with each other.

[0037] The method, system, electronic device and storage medium of the present invention will be described in more detail below in the manner of specific embodiments. It should be understood that the embodiments are only exemplary, and the present invention is not limited thereto.

[0038] First of all, it should be noted that: the method and device of the present invention can identify whether there is coking on the wall of a reactor (especially the wall of an annular pipeline). It can collect the acoustic emission signals of both the coking part and the non-coking part as sample data for subsequent comparison. Since the catalyst flows continuously, it will continuously rub and impact the reactor wall, forming a continuous acoustic emission signal. The signals of the coking and non-coking parts are mainly reflected in the difference in the frequency domain. The acoustic emission signal belongs to a transient signal, and it conforms to the law that the longer the sampling length, the higher the frequency resolution. In order to effectively improve the frequency resolution, the present invention adopts a waveform stream signal acquisition method. In order to suppress the mixed noise components (i.e., noise signals irrelevant to the abnormal coking acoustic signal), and at the same time make the signal highly distinguishable, the inventors found that fusing the singular spectrum and the high-order spectrum can more effectively remove the noise with lower energy, further extract the intensity and phase characteristics in the frequency domain, and at the same time suppress the mixed noise components, which can greatly improve the accuracy and distinguishability of the signal characteristics, thereby improving the accuracy of coking part identification.

[0039] Example 1

[0040] As Figure 1 shown, this embodiment provides a method for identifying abnormal coking acoustic signals in an annular pipeline, including the following steps:

[0041] Step S101, collect waveform stream data of acoustic emission signals at different flow rates and different coking parts in an analog annular pipeline.

[0042] The simulated annular pipeline is a space formed by sleeving an outer pipe and an inner pipe. In this embodiment, different-sized annular spaces are used for simulation. Multiple annular spaces are composed of multiple outer pipes with the same size and inner pipes with different sizes. In this embodiment, simulated coking materials can be adhesively bonded to the annular pipeline respectively. Specifically, this embodiment uses three groups of annular pipelines with different sizes, and the material can be plexiglass. The size of the outer pipe in the annular pipeline remains unchanged, and the size of the inner pipe increases in sequence, which are φ30, φ50, and φ70 in sequence, and the size of the outer pipe is uniformly φ90. Simulated coking materials with different sizes, thicknesses, and shapes are adhesively bonded to the inner wall of the outer pipe, the outer wall of the inner pipe, and both the inner wall of the outer pipe and the outer wall of the inner pipe of the annular pipeline respectively, so as to simulate the coking conditions under different states.

[0043] This embodiment uses a fan and a buffer tank to provide gas power, and adjusts the gas flow rate through a ball valve. The gas contains catalyst particles and is used to simulate the actual working conditions. According to the gas dynamics principle, when the annular pipeline is in normal working conditions, that is, when there is no coking material in the component, the detected acoustic signal has no abnormal flow sound. However, when there is coking material in the annular pipeline (that is, under different annular space sizes, different flow rates, and different coking sample states), different abnormal flow sound signals can be detected. The sound waves are conducted out through a waveguide rod, and these different abnormal flow sound signals can be collected by using an acoustic emission sensor.

[0044] Since the acoustic emission signal belongs to a transient signal, it conforms to the law that the longer the sampling length, the higher the frequency resolution. In order to improve the frequency resolution, in this embodiment, while keeping the sampling rate at 1M unchanged, the sampling length of the signal is selected to be increased, and the waveform stream signal acquisition method is adopted (the sampling rate is 1M, and the sampling length is generally up to 4k at most, that is, a signal is at most 4096μs long. In this embodiment, a waveform stream signal with a duration of 1s is extracted, that is, a waveform stream signal has 1M sampling points).

[0045] Step S102, perform singular spectrum and high-order spectrum fusion analysis processing on the waveform stream data collected in step S101 (this is actually a data preprocessing process that effectively improves the denoising effect), and extract the intensity and phase characteristics in the frequency domain of the high-order spectrum data after denoising. That is, while extracting the required characteristic parameters, the unwanted noise components mixed in are suppressed, and the accuracy and recognition rate of the signal characteristics are improved, thereby improving the accuracy of coking part identification.

[0046] Furthermore, the inventor has found through research that selecting the characteristic parameters of the present invention for identifying coking abnormal sound signals can be more accurate. And performing singular spectrum and high-order spectrum fusion analysis processing on the collected waveform stream data has a more obvious denoising effect. This fusion analysis means that the analysis processing result of the singular spectrum is further used as the input of the high-order spectrum program, and then the high-order spectrum analysis processing is performed. The following is a detailed description.

[0047] First, perform the analysis and processing of the singular spectrum:

[0048] The singular spectrum converts a one-dimensional signal into a multi-dimensional signal, dividing a single signal into multiple groups of signals classified according to energy magnitude. It starts from the dynamic energy reconstruction of the time series, is related to the empirical orthogonal function, is not restricted by the sine wave assumption, extracts as much useful information as possible from the data containing noise, extracts the components with significant oscillations, and selects several meaningful components for reconstruction, thereby reducing noise.

[0049] (1) Construct the trajectory matrix:

[0050] Select an appropriate window length L (2 ≤ L ≤ K / 2). Such a value can make the matrix as small as possible when it is full column rank, because if L is too long, not only the calculation speed will slow down, but also the sensitivity of the trend line will decrease. K can be directly taken as the full data length, converting the one-dimensional time series into a multi-dimensional series and defining it as the trajectory matrix X.

[0051]

[0052] (2) Singular value decomposition of the trajectory matrix X:

[0053] X = U∑V T Formula (1);

[0054] Among them, U is an L×L unitary matrix, and its column vectors are composed of the orthonormal set of the left singular vectors of the trajectory matrix X. V is a K×K unitary matrix, and its column vectors are composed of the orthonormal set of the right singular vectors of the trajectory matrix X. ∑ is an L×K rectangular diagonal matrix, which contains L singular values of the trajectory matrix X arranged from large to small.

[0055] Therefore, using singular value decomposition, the trajectory matrix X can be rewritten as:

[0056]

[0057] Among them, σ i is the i-th singular value, U i and V i are the i-th column vectors of matrix U and matrix V respectively. d is the rank of the trajectory matrix X, satisfying d ≤ L. And X i = σ i U i V i T is the i-th elementary matrix of the trajectory matrix X. {U i ,σ i ,V i} represents the i-th set of eigenvalues of the singular value decomposition.

[0058] (3) Reconstruct the separated singular matrix into multiple one-dimensional data components:

[0059] In this embodiment, the required signal energy components are separated according to the singular values of the singular spectrum. The abscissa of the separated singular values and singular matrix is the sampling point, and the ordinate is the amplitude. Since the abnormal coking sound signal exists in the separation of signals with larger energy, the singular matrix with larger energy (for example, in the range of [0.1, 0.2]) decomposed can be reconstructed into a new signal (i.e., removing the noise signal), and then the reconstructed one-dimensional data components are extracted and derived from it. For example, if the one-dimensional data components are four components, the last component with the lowest energy can be removed, and the other three components are reconstructed into a one-dimensional data component and substituted into the high-order spectrum (such as the third-order spectrum) program. The multiple one-dimensional data components are expressed as: SVD[X1 X2 X3 … X n-1 = X{i}, where SVD is the singular spectrum and X{i} is the signal after the reconstruction of the singular spectrum component.

[0060] Secondly, perform the analysis and processing of the high-order spectrum:

[0061] According to the order K of the high-order spectrum (K = 3 in this embodiment), after removing the component with the lowest energy from the multiple one-dimensional data components of the aforementioned singular spectrum, the remaining three are reconstructed into a one-dimensional data component as the original input signal of the third-order spectrum program. Perform the Fourier transform of the third-order cumulant on the input signal to obtain a third-order spectrum. The specific method is as follows:

[0062] The k-order spectrum S kx (ω1, ω2, …, ω k-1 ) of the one-dimensional real stationary sequence X{i} extracted from the aforementioned singular spectrum is defined as the Fourier transform of the k-order cumulant c k (τ1, τ2, …, τ k-1 ), that is,

[0063]

[0064] Generally, S kx (ω1, ω2, …, ω k-1 ) is a complex number, and the absolute summability of c kx (τ1, τ2, …, τ k-1 ) is a sufficient condition for the existence of the high-order spectrum.

[0065] When k = 3, the above formula (3) can be rewritten as:

[0066]

[0067] S in formula (4) 3x(ω1, ω2) is called the bispectrum or third-order spectrum. After combining with the ARMA model (autoregressive model, fitting effect), the expression is:

[0068]

[0069] where s(t) is the signal sequence; t is the length of the time sequence; a(t) is an independent and identically distributed sequence; the coefficients φ1, φ2, …, φ p and θ1, θ2, …, θ q are the AR model parameters and the MA model parameters respectively, p is the AR order; q is the MA order. The transfer function of the model (this is the prior art) is:

[0070]

[0071] The third-order spectrum is obtained by performing a Fourier transform on the third-order cumulant of the signal to obtain a third-order spectrum, which can effectively suppress Gaussian noise. Through the above data processing, it has a higher precision advantage. After reconstructing the component signals of the aforementioned singular spectrum to form a new one-dimensional signal, substituting it into the parametric bispectrum estimation program can extract the amplitude and phase information of the reconstructed signal, further removing noise to avoid the troubles caused by phase problems and noise. In this embodiment, a higher-precision signal purification is performed from the singular spectrum to the higher-order spectrum, and finally a three-dimensional graph is obtained through optimization processing. The X and Y axes of the three-dimensional graph represent frequency, and the Z axis represents amplitude. Further, in this embodiment, a two-dimensional curve slice can be taken at the peak of the three-dimensional higher-order spectrum, and then the bispectrum estimation formula based on the ARMA model is:

[0072]

[0073] where H * is orthogonal to H; has a finite variance; it is generated by non-Gaussian white noise, is the skewness, and a(e) is an independent and identically distributed non-Gaussian random process. Read the cut planar graph, with the X axis being the frequency and the Y axis being the amplitude. Extract the maximum amplitude and the frequency corresponding to the maximum amplitude as two of the characteristic parameters for the subsequent coking abnormal sound signal recognition model.

[0074] Similarly, the higher-order spectrum cumulant retains the phase information of the signal, and the phase spectrum of the signal can be obtained from the higher-order spectrum cumulant. The random signal x(i) is regarded as being generated by exciting the linear system H(ω) with white noise u(n). In practical applications, the bispectrum corresponding to the third-order cumulant is used to obtain the phase spectrum.

[0075]

[0076]

[0077] According to Formula (8) and Formula (9), the phase spectrum of the system can be deduced from the estimated ψ3 (where ψ(ω) is the cumulant generating function). Obtaining the phase spectrum of the signal from the high-order spectral cumulants is a prior art and will not be elaborated here.

[0078] At the same time, readings are also taken for the cut planar graph, with the X-axis being the frequency and the Y-axis being the phase. The maximum phase and the frequency corresponding to the maximum phase are extracted as the other two characteristic parameters for the subsequent recognition model of the abnormal coking sound signal. The maximum amplitude and the frequency corresponding to the maximum amplitude, the maximum phase and the frequency corresponding to the maximum phase, these four characteristic parameters can respectively characterize the intensity and phase characteristics in the frequency domain and jointly serve as the input of the algorithm model.

[0079] Step S103: Use the intensity and phase characteristics in the aforementioned frequency domain (which can be the four characteristic parameters in step S102) as the input variables of the recognition model of the abnormal coking sound signal, and train the model.

[0080] Specifically, the recognition model of the abnormal coking sound signal in this embodiment can adopt a neural network model, preferably a BP neural network model. Using the aforementioned four characteristic parameters as the input layer and the abnormal coking sound emission source as the output layer, with the number of hidden nodes set to six, a BP neural network with a single hidden layer and a topological structure of 4-6-1 is constructed. During the model training process, through continuous correction and update of the weights and biases, the error gradually converges.

[0081] Step S104: Use the trained model to recognize the coking situation of the annular pipeline under different flow rates. The coking situation includes but is not limited to: whether there is a coking phenomenon, the coking location, the shape and size of the coking, etc. Since the sample data collects signals in three cases respectively: the non-coking state, the inner wall of the outer pipe, the outer wall of the inner pipe, and both the inner wall of the outer pipe and the outer wall of the inner pipe in the annular pipeline space, it is possible to know whether there is a coking phenomenon by comparing the signal data of non-coking and coking. By setting different inner pipe diameters in the simulation experiment, annular pipelines with different sizes can be simulated, and by adjusting the gas flow rate, the coking states under different flow rates and different annular pipeline sizes can be simulated.

[0082] This embodiment applies the principle of acoustic emission, extracts characteristic parameters from the sample data of the abnormal sound signal of coking to train the model, so that the abnormal sound signal of coking under various actual working conditions can be identified in real time and accurately, and the abnormal shutdown and accidents caused by coking can be prevented in time and effectively. During the signal acquisition process, in order to effectively improve the frequency resolution, the waveform stream signal acquisition method is adopted in this embodiment. In order to suppress the mixed noise components (i.e., the noise signals unrelated to the abnormal sound signal of coking) and make the signal highly recognizable at the same time, the singular spectrum and high-order spectrum fusion analysis and processing of the present invention can more effectively remove the low-energy noise, and at the same time suppress the mixed noise components, which can greatly improve the accuracy and recognition of the signal characteristics, thereby improving the accuracy of the identification of the coking site. After the signal is preprocessed, the coking situation in the annular pipeline can be more effectively identified by extracting the intensity and phase characteristics in the frequency domain.

[0083] Example 2

[0084] like Figure 2 As shown, this embodiment is a virtual device embodiment corresponding to the method for identifying abnormal acoustic signals of coking in an annular pipeline in Embodiment 1.

[0085] The annular pipeline coking abnormal acoustic signal recognition device of this embodiment includes: a waveform flow acquisition module 201, a characteristic parameter extraction module 202, a model training module 203 and a coking recognition module 204. Among them, the waveform flow acquisition module 201 is used to collect waveform flow data of acoustic emission signals at different flow rates and different coking locations in the simulated annular pipeline; the characteristic parameter extraction module 202 is used to perform singular spectrum and high-order spectrum fusion analysis processing on the waveform flow data, and extract the intensity and phase characteristics in the frequency domain of the denoised high-order spectrum data; the model training module 203 is used to use the intensity and phase characteristics in the frequency domain as input variables of the coking abnormal acoustic signal recognition model to train the model; the coking recognition module 204 is used to identify the coking conditions of the annular pipeline at different flow rates through the trained model.

[0086] Furthermore, the characteristic parameter extraction module may specifically include: a singular value component construction submodule, a high-order spectrum input selection submodule, a K-order spectrum acquisition submodule, and an intensity and phase feature extraction submodule. Among them, the singular value component construction submodule is used to perform singular value decomposition through the constructed trajectory matrix, and reconstruct the separated singular matrix into multiple one-dimensional data components; the high-order spectrum input selection submodule is used to eliminate the component with the lowest energy among the multiple one-dimensional data components and reconstruct the remaining data components into a one-dimensional data component as the original input signal of the high-order spectrum program; the K-order spectrum acquisition submodule is used to perform a Fourier transform of the K-order cumulant on the input signal to obtain a K-order spectrum; the intensity and phase feature extraction submodule is used to extract the intensity and phase features in the frequency domain from the spectrum of the K-order spectrum.

[0087] Preferably but not limited thereto, the intensity and phase features in the frequency domain may include the following four feature parameters: the maximum amplitude and the frequency corresponding to the maximum amplitude; the maximum phase and the frequency corresponding to the maximum phase.

[0088] The device of this embodiment corresponds to the method of Embodiment 1 and can achieve the same technical effects.

[0089] Example 3

[0090] Figure 3 It is a schematic diagram of the hardware structure of the electronic device for identifying abnormal coking sound signals in the annular pipeline of this embodiment. This device (such as a terminal, a server, etc.) includes one or more processors 610 and a memory 620. Taking one processor 610 as an example, this device may further include: an input device 630 and an output device 640.

[0091] The processor 610, the memory 620, the input device 630, and the output device 640 may be connected through a bus or other means.

[0092] The memory 620, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs, non-transitory computer-executable programs, and modules. The processor 610 executes various functional applications and data processing of the electronic device by running the non-transitory software programs, instructions, and modules stored in the memory 620, that is, to implement the processing method of the above method embodiment.

[0093] The memory 620 may include a program storage area and a data storage area. Among them, the program storage area can store an operating system and application programs required for at least one function; the data storage area can store data, etc. In addition, the memory 620 may include a high-speed random access memory, and may also include a non-transitory memory, such as at least one magnetic disk storage device, a flash memory device, or other non-transitory solid-state storage devices. In some embodiments, the memory 620 may optionally include a memory remotely set relative to the processor 610, and these remote memories can be connected to the processing device through a network. Examples of the above network include but are not limited to the Internet, an enterprise intranet, a local area network, a mobile communication network, and combinations thereof.

[0094] The input device 630 can receive input digital or character information and generate a signal input. The output device 640 may include a display device such as a display screen.

[0095] The one or more modules are stored in the memory 620 and, when executed by the one or more processors 610, perform the following steps: A. Collect waveform stream data of acoustic emission signals at different flow rates and at different coking sites in the analog annular pipeline; B. Perform singular spectrum and high-order spectrum fusion analysis on the waveform stream data to extract the intensity and phase characteristics in the frequency domain of the denoised high-order spectrum data; C. Use the intensity and phase characteristics in the frequency domain as input variables of a coking abnormal acoustic signal recognition model and train the model; D. Identify the coking conditions of the annular pipeline at different flow rates through the trained model.

[0096] The above electronic device can execute the method provided by the embodiment of the present invention and has the corresponding functional modules and beneficial effects for executing the method. For technical details not described in detail in this embodiment, reference can be made to the methods provided in other embodiments of the present invention.

[0097] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0098] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the above technical solution, in essence, or the part that contributes to the related technology can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disc, etc., and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0099] Example 4

[0100] In this embodiment, a non-transitory computer-readable storage medium is provided. The non-transitory computer-readable storage medium stores computer-executable instructions for causing a computer to execute the following method for identifying abnormal acoustic signals of coking in an annular pipeline:

[0101] A. Collect waveform flow data of acoustic emission signals at different flow rates and different coking positions in the simulated annular pipeline; B. Perform singular spectrum and high-order spectrum fusion analysis on the waveform flow data to extract the intensity and phase characteristics in the frequency domain of the denoised high-order spectrum data; C. Use the intensity and phase characteristics in the frequency domain as input variables of the coking abnormal acoustic signal recognition model and train the model; D. Identify the coking situation of the annular pipeline at different flow rates through the trained model.

[0102] The foregoing description of specific exemplary embodiments of the present invention is for purposes of illustration and exemplification. These descriptions are not intended to limit the invention to the precise forms disclosed, and it is obvious that many changes and variations are possible in light of the above teachings. The purpose of selecting and describing the exemplary embodiments is to explain the specific principles of the present invention and its practical applications, so that those skilled in the art can implement and utilize various different exemplary embodiments of the present invention, as well as various different selections and changes. Any simple modifications, equivalent variations, and modifications made to the above exemplary embodiments shall fall within the protection scope of the present invention.

Claims

1. A method for identifying abnormal coking sound signals in a circular pipeline, characterized in that It includes the following steps: A. Collect the waveform flow data of acoustic emission signals at different flow rates and different coking sites in the simulated annular pipeline; B. Perform singular spectrum and high-order spectrum fusion analysis on the waveform flow data, and extract the intensity and phase characteristics in the frequency domain of the denoised high-order spectrum data; The specific steps of performing singular spectrum and high-order spectrum fusion analysis on the waveform flow data include: performing singular value decomposition through the constructed trajectory matrix, reconstructing the separated singular matrix into multiple one-dimensional data components; removing the component with the lowest energy among the multiple one-dimensional data components and reconstructing the remaining data components into a one-dimensional data component as the original input signal of the high-order spectrum program; performing Fourier transform of the K-order cumulant on the input signal to obtain a K-order spectrum; extracting the intensity and phase characteristics in the frequency domain from the spectrogram of the K-order spectrum; C. Use the intensity and phase characteristics in the frequency domain as input variables of the coking abnormal acoustic signal recognition model, and train the model; D. Identify the coking situation of the annular pipeline at different flow rates through the trained model.

2. The method for identifying abnormal coking sound signals of an annular pipeline according to claim 1, wherein The simulated annular pipeline in step A includes multiple outer pipes with the same diameter and inner pipes with different diameters, and simulated coking materials are bonded to the inner wall of the outer pipe, the outer wall of the inner pipe, and the inner wall of the outer pipe and the outer wall of the inner pipe respectively.

3. The abnormal coking sound signal recognition method for an annular pipeline according to claim 2, characterized in that The acquisition duration of the waveform flow data is 1 second.

4. The method for identifying abnormal coking sound signals of an annular pipeline according to claim 1, characterized in that, The intensity characteristic in the frequency domain is the maximum amplitude and the frequency corresponding to the maximum amplitude; the phase characteristic in the frequency domain is the maximum phase and the frequency corresponding to the maximum phase.

5. The method for identifying abnormal coking sound signals in an annular pipeline according to claim 1, characterized in that, The order of the K-order spectrum is three.

6. The method for identifying abnormal coking sound signals of an annular pipeline according to claim 5, characterized in that, The multiple one-dimensional data components are expressed as: ； where SVD is the singular spectrum and X{i} is the signal reconstructed from the singular spectrum component.

7. The method for identifying abnormal coking sound signals of an annular pipeline according to claim 1, wherein The coking situation in step D includes: whether there is a coking phenomenon, the coking site, and the shape and size of the coking.

8. The method for identifying abnormal coking sound signals of an annular pipeline according to claim 1, characterized in that, The coking abnormal acoustic signal recognition model uses a neural network model.

9. An abnormal coking sound signal recognition device for an annular pipeline, characterized in that, It includes: A waveform flow acquisition module for collecting the waveform flow data of acoustic emission signals at different flow rates and different coking sites in the simulated annular pipeline; A feature parameter extraction module for performing singular spectrum and high-order spectrum fusion analysis on the waveform flow data and extracting the intensity and phase characteristics in the frequency domain of the denoised high-order spectrum data; The feature parameter extraction module specifically includes: a singular value component construction sub-module for performing singular value decomposition through the constructed trajectory matrix and reconstructing the separated singular matrix into multiple one-dimensional data components; a high-order spectrum input selection sub-module for removing the component with the lowest energy among the multiple one-dimensional data components and reconstructing the remaining data components into a one-dimensional data component as the original input signal of the high-order spectrum program; a K-order spectrum acquisition sub-module for performing Fourier transform of the K-order cumulant on the input signal to obtain a K-order spectrum; an intensity and phase characteristic extraction sub-module for extracting the intensity and phase characteristics in the frequency domain from the spectrogram of the K-order spectrum; A model training module for using the intensity and phase characteristics in the frequency domain as input variables of the coking abnormal acoustic signal recognition model and training the model; The coking recognition module is used to recognize the coking situation of the annular pipeline at different flow rates through the trained model.

10. The annular pipeline coking abnormal sound signal recognition device according to claim 9, characterized in that, The intensity and phase features in the frequency domain include the following four feature parameters: the maximum amplitude and the frequency corresponding to the maximum amplitude; the maximum phase and the frequency corresponding to the maximum phase.

11. An electronic device for identifying abnormal coking sound signals in a circular pipeline, characterized in that, Comprising: At least one processor; And A memory communicatively connected to 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 execute the method for identifying abnormal sound signals of coking in an annular pipeline according to any one of claims 1 to 8.

12. A non-transitory computer-readable storage medium, characterized in that, The non-transitory computer-readable storage medium stores computer-executable instructions for causing the computer to execute the method for identifying abnormal sound signals of coking in an annular pipeline according to any one of claims 1 to 8.

Images