CN115345203A - Pipeline signal identification method and system based on VMD and multi-feature fusion - Google Patents
Pipeline signal identification method and system based on VMD and multi-feature fusion Download PDFInfo
- Publication number
- CN115345203A CN115345203A CN202211019804.7A CN202211019804A CN115345203A CN 115345203 A CN115345203 A CN 115345203A CN 202211019804 A CN202211019804 A CN 202211019804A CN 115345203 A CN115345203 A CN 115345203A
- Authority
- CN
- China
- Prior art keywords
- decomposition
- pipeline
- value
- wcc
- vmd
- 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.)
- Pending
Links
Images
Landscapes
- Image Analysis (AREA)
Abstract
A pipeline signal identification method and system based on VMD and multi-feature fusion relates to the technical field of pipeline signal identification and is used for solving the problem that pipeline leakage detection precision is affected due to poor pipeline signal feature extraction effect in the prior art. The technical points of the invention comprise: preprocessing a pipeline signal through a VMD algorithm, providing a multi-angle measuring distance algorithm-WCC algorithm, and determining a VMD decomposition layer number K value by using a weighted correlation coefficient and a cosine value index; determining characteristic modal components according to the similarity degree of the IMFs components and the original signals, and extracting characteristic parameters of the characteristic modes, wherein the characteristic parameters comprise entropy characteristics, waveform parameters and time-frequency domain characteristics; constructing a feature vector group based on multi-feature fusion by using the extracted feature parameters; and inputting the feature vector group into a pre-trained neural network to identify the working condition of the pipeline signal. The invention effectively improves the accuracy of signal feature extraction, thereby improving the pipeline signal identification precision.
Description
Technical Field
The invention relates to the technical field of pipeline signal identification, in particular to a pipeline signal identification method and system based on VMD and multi-feature fusion.
Background
Pipeline transportation has been widely used in various industries because of its unique advantages in the transportation of liquids, gases, slurries, etc. Pipeline transportation has become one of the five transportation modes in our country. Due to the fact that the pipeline has the phenomena of corrosion, aging, cathode protection failure, natural disasters, production and construction, theft of lawless persons and the like, the pipeline leakage event is difficult to stop. The leakage of the pipeline not only can influence the normal operation of pipeline transportation, cause the pollution of the environment and the waste of resources, but also can seriously influence the normal life of people, threaten the life of people and cause serious loss of the property of people. Therefore, the pipeline is monitored by adopting a proper pipeline leakage detection technology, leakage is prevented, leakage is found timely, the leakage is accurately positioned, and environmental pollution and economic loss can be effectively reduced.
The processing of the pipeline signal is a problem of nonlinear signal processing, and common nonlinear signal processing methods include a wavelet transform method, a singular value noise reduction method, an EMD decomposition method, a variational modal decomposition method and the like. Although the EMD and VMD methods are gradually applied to pipeline leakage detection in recent years, both methods have certain limitations. After the EMD algorithm adaptively decomposes a signal into the sum of a plurality of intrinsic mode functions, the problems of under-enveloping, over-enveloping, power-off effect and modal chaos are easily caused. The VMD can avoid the problems of EMD decomposition, but before decomposition, some parameters need to be preset, and the value of the parameters influences the decomposition effect of the VMD.
In the process of leakage detection, extracting the characteristic information of the signal plays a key role in improving the precision of pipeline leakage detection, and therefore, extracting effective characteristics from the pipeline signal has important significance in reducing the false alarm rate.
Disclosure of Invention
In view of the above problems, the present invention provides a method and a system for pipeline signal identification based on VMD and multi-feature fusion, so as to solve the problem in the prior art that the pipeline leakage detection accuracy is affected due to poor pipeline signal feature extraction effect.
According to an aspect of the invention, a method for identifying a pipeline signal based on VMD and multi-feature fusion is provided, which comprises the following steps:
step 2, carrying out variation modal decomposition on the pipeline signals in the data set to obtain a plurality of intrinsic modal components; determining the optimal decomposition layer number in the variational modal decomposition by utilizing a multi-angle measuring distance algorithm in the variational modal decomposition;
step 3, comparing the similarity degree of each eigenmode component and the original pipeline signal so as to determine a characteristic component;
step 4, calculating entropy characteristics, waveform characteristics and time-frequency domain characteristics of the characteristic components to obtain a characteristic vector group based on multi-characteristic fusion;
step 5, inputting the feature vector group into a probabilistic neural network for training to obtain a trained pipeline signal recognition model based on the probabilistic neural network;
and 6, inputting the pipeline signal to be identified into a trained pipeline signal identification model based on the probabilistic neural network for identification, and obtaining an identification result.
Further, the working condition types in the step 1 comprise three working conditions of normal, knocking and leakage.
Further, the specific process of determining the optimal number of decomposition layers in the variational modal decomposition by using the multi-angle distance measurement algorithm in the step 2 comprises: determining the decomposition layer number K in the variational modal decomposition by using the weighted correlation coefficient and the cosine value index, initializing the decomposition layer number K, and calculating the WCC value between adjacent modes corresponding to the decomposition layer number K, wherein the calculation formula of the WCC value is as follows:
WCC＝λCC+ηCD
wherein CC represents a correlation coefficient, CD represents a cosine distance, and lambda and eta are weighting coefficients;
adding 1 to the number K of decomposition layers when the WCC value is not greater than a preset maximum threshold value; repeating the process, namely calculating the WCC value between the adjacent modes corresponding to the decomposition layer number K + 1; and when the WCC value is larger than a preset maximum threshold value and the decomposition layer number is K + N at the moment, the optimal decomposition layer number is K + N-1,N to represent a natural number.
Further, the calculation formula of the weighting coefficients λ and η in step 2 is as follows:
wherein, the first and the second end of the pipe are connected with each other,
a coefficient of variation representing the coefficient of correlation,
S CC is the standard deviation of the correlation coefficient, mu CC Is the mean of the correlation coefficients;
a coefficient of variation representing a cosine distance,
S CD is the standard deviation of the cosine distance, mu CD Is the mean of the cosine distances.
Further, in step 4, the entropy characteristics adopt spread entropy, the waveform characteristics adopt margin factors, and the time-frequency domain characteristics adopt standard deviation.
According to another aspect of the present invention, there is provided a conduit signal identification system based on VMD and multi-feature fusion, the system comprising:
the signal acquisition module is configured to acquire pipeline signals of different working condition types and construct a data set;
the variable modal decomposition module is configured to perform variable modal decomposition on the pipeline signals in the data set to obtain a plurality of intrinsic modal components; determining the optimal decomposition layer number in the variational modal decomposition by utilizing a multi-angle measuring distance algorithm in the variational modal decomposition;
a characteristic component determination module configured to compare the degree of similarity of each eigenmode component to the original pipeline signal, thereby determining a characteristic component;
a feature vector determination module configured to calculate entropy features, waveform features, and time-frequency domain features of the feature components to obtain a feature vector group based on multi-feature fusion;
the recognition model training module is configured to input the feature vector group into a probabilistic neural network for training to obtain a trained pipeline signal recognition model based on the probabilistic neural network;
and the signal identification module is configured to input the pipeline signal to be identified into a trained pipeline signal identification model based on the probabilistic neural network for identification, and obtain an identification result.
Furthermore, the working condition types in the signal acquisition module comprise three working conditions of normal, knocking and leakage.
Further, the specific process of determining the optimal number of decomposition layers in the variational modal decomposition by using a multi-angle distance measurement algorithm in the variational modal decomposition module comprises the following steps: determining the decomposition layer number K in the variational modal decomposition by using the weighted correlation coefficient and the cosine value index, initializing the decomposition layer number K, and calculating the WCC value between adjacent modes corresponding to the decomposition layer number K, wherein the calculation formula of the WCC value is as follows:
WCC＝λCC+ηCD
wherein CC represents a correlation coefficient, CD represents a cosine distance, and lambda and eta are weighting coefficients;
adding 1 to the number K of decomposition layers when the WCC value is not greater than a preset maximum threshold value; repeating the process, namely calculating the WCC value between the adjacent modes corresponding to the decomposition layer number K + 1; and when the WCC value is larger than a preset maximum threshold value, the number of decomposition layers at the moment is set to be K + N, and the optimal number of decomposition layers is K + N-1,N to represent a natural number.
Further, the calculation formula of the weighting coefficients λ and η in the variation modal decomposition module is as follows:
wherein, the first and the second end of the pipe are connected with each other,
a coefficient of variation representing the coefficient of correlation,
S CC is the standard deviation of the correlation coefficient, mu CC Is the mean of the correlation coefficients;
a coefficient of variation representing a cosine distance,
S CD is the standard deviation of the cosine distance, μ CD Is the mean of the cosine distances.
Further, the entropy features in the feature vector determination module adopt spread entropy, the waveform features adopt margin factors, and the time-frequency domain features adopt standard deviation.
The beneficial technical effects of the invention are as follows:
the invention provides a pipeline signal identification method and system based on VMD and multi-feature fusion. Firstly, preprocessing a pipeline signal through a VMD algorithm, providing a multi-angle measuring distance algorithm-WCC (Weighting of Correlation and Cosine distance) algorithm in the process, determining a VMD optimal decomposition layer number K value by using a weighted Correlation coefficient and a Cosine value index, then determining a characteristic modal component according to the similarity degree of IMFs component and an original signal, extracting characteristic parameters of the characteristic modal, including entropy characteristics, waveform parameters and time-frequency domain characteristics, and respectively selecting a dispersion entropy, a margin factor and a standard deviation; and constructing a high-dimensional feature vector group based on multi-feature fusion by using the extracted feature parameters, and finally, inputting the feature vector group into the PNN to identify the working condition of the pipeline signal. The invention effectively improves the accuracy of signal feature extraction, thereby improving the pipeline signal identification precision.
Drawings
The present invention may be better understood by reference to the following description taken in conjunction with the accompanying drawings, which are incorporated in and form a part of this specification, and which are used to further illustrate preferred embodiments of the present invention and to explain the principles and advantages of the present invention.
FIG. 1 is a flow chart of a pipeline signal identification method based on VMD and multi-feature fusion according to an embodiment of the present invention;
FIG. 2 is a flowchart of determining an optimal decomposition level number K based on WCC value according to an embodiment of the present invention;
FIG. 3 is a diagram illustrating a network structure of a PNN according to an embodiment of the present invention;
fig. 4 is a schematic structural diagram of a conduit signal identification system based on VMD and multi-feature fusion according to an embodiment of the present invention.
Detailed Description
In order that those skilled in the art will better understand the disclosure, exemplary embodiments or examples of the disclosure are described below with reference to the accompanying drawings. It is obvious that the described embodiments or examples are only some, but not all embodiments or examples of the invention. All other embodiments or examples, which can be obtained by a person skilled in the art without making any creative effort based on the embodiments or examples in the present invention, shall fall within the protection scope of the present invention.
The embodiment of the invention provides a pipeline signal identification method based on VMD and multi-feature fusion, which comprises the following steps as shown in figure 1:
step 2, carrying out variation modal decomposition on the pipeline signals in the data set to obtain a plurality of intrinsic modal components; determining the optimal decomposition layer number in the variational modal decomposition by utilizing a multi-angle measuring distance algorithm in the variational modal decomposition;
step 3, comparing the similarity degree of each intrinsic mode component and the original pipeline signal, and determining a characteristic component;
step 4, calculating entropy characteristics, waveform characteristics and time-frequency domain characteristics of the characteristic components to obtain a characteristic vector group based on multi-characteristic fusion;
step 5, inputting the feature vector group into a probabilistic neural network for training to obtain a trained pipeline signal recognition model based on the probabilistic neural network;
and 6, inputting the pipeline signal to be recognized into a trained pipeline signal recognition model based on the probabilistic neural network for recognition, and obtaining a recognition result.
In this embodiment, preferably, the operating condition types include three operating conditions, namely normal operating conditions, knocking operating conditions and leakage operating conditions.
In this embodiment, preferably, the specific process of determining the optimal number of decomposition layers in the variational modal decomposition by using the multi-angle distance measurement algorithm in step 2 includes: determining the decomposition layer number K in the variational modal decomposition by using the weighted correlation coefficient and the cosine value index, initializing the decomposition layer number K, and calculating the WCC value between adjacent modes corresponding to the decomposition layer number K, wherein the calculation formula of the WCC value is as follows:
WCC＝λCC+ηCD
wherein CC represents a correlation coefficient, CD represents a cosine distance, and lambda and eta are weighting coefficients;
when the WCC value is not larger than a preset maximum threshold value, adding 1 to the decomposition layer number K, and repeating the process, namely calculating the WCC value between adjacent modes corresponding to the decomposition layer number K + 1; and when the WCC value is larger than a preset maximum threshold value and the decomposition layer number is K + N at the moment, the optimal decomposition layer number is K + N-1,N to represent a natural number.
In this embodiment, preferably, the calculation formula of the weighting coefficients λ and η in step 2 is as follows:
wherein, the first and the second end of the pipe are connected with each other,
a coefficient of variation representing the coefficient of correlation,
S CC is the standard deviation of the correlation coefficient, mu CC Is the mean of the correlation coefficients;
a coefficient of variation representing a cosine distance,
S CD is the standard deviation of the cosine distance, mu CD Is the mean of the cosine distances.
Another embodiment of the present invention provides a method for identifying a pipeline signal based on VMD and multi-feature fusion, comprising the following steps:
And 2, determining the decomposition layer number K in the VMD algorithm by using the WCC value.
According to the embodiment of the invention, a WCC algorithm is provided in the step 2, the K value of the VMD decomposition layer number is determined by using the weighted correlation coefficient and the cosine value index, different K values are set for VMD decomposition, the default value of the penalty factor alpha is 2000, and the WCC value between adjacent modes under different K values is calculated. The following formula is provided.
WCC＝λCC+ηCD
Wherein, CC is a correlation coefficient, CD is a cosine distance, and lambda and eta are weighting coefficients, as shown in the following formula:
wherein, the first and the second end of the pipe are connected with each other,
is the coefficient of variation of the correlation coefficient,
is the coefficient of variation of the cosine distance, S CC Is the standard deviation of the correlation coefficient, mu CC Is the mean of the correlation coefficients, S CD Is the standard deviation of the cosine distance, mu CD Is the mean of the cosine distances.
When the WCC value between adjacent modes is smaller, the similarity between the two modes is smaller; when the WCC value between adjacent modalities is larger, the similarity between the two modalities is larger, and the VMD decomposition at the moment is possible to have over-decomposition. According to the theory of correlation coefficients, the correlation coefficients are weak correlation between 0.1 and 0.3, medium correlation between 0.3 and 0.8 and strong correlation between 0.8 and 1; according to the correlation coefficient theory and a large number of experiments, when the WCC value is larger than 0.8, the VMD decomposition is over-decomposition, the optimal K value is K-1, and therefore the decomposition layer number K value of the VMD decomposition can be determined. The specific flow chart is shown in fig. 2.
Step 3, decomposing the pipeline signal through a VMD algorithm to obtain a plurality of IMFs, wherein IMFs components are more regular than original signals, and signal characteristics are more obvious; the characteristic components are determined by comparing the similarity of each modal component to the original signal.
And 4, calculating entropy characteristics, waveform characteristics and time-frequency domain characteristics of the characteristic components to form a characteristic vector group based on multi-characteristic fusion.
According to the embodiment of the present invention, when analyzing the characteristic parameters of the pipeline signals in step 4, the commonly used characteristic parameters can be generally classified into the following three categories: time-frequency domain features, entropy features, and waveform features.
(1) Time-frequency domain characteristics: the waveform characteristics of the pipe signal may be represented by analyzing time-frequency domain characteristics of the pipe signal. For example, the average value can measure the average distribution trend of the pipeline data; the peak value is obtained by utilizing the difference between the maximum value and the minimum value in the pipeline data, and the fluctuation condition of the signal can be reflected; the root mean square planting is to obtain the vibration energy of the pipeline data, and the value of the vibration energy influences the energy; the standard deviation can measure the discrete degree between pipeline signals, and the standard deviation is selected as a characteristic parameter in the embodiment of the invention.
(2) Entropy characteristics: the entropy characteristic parameters can measure the complexity of the pipeline data, the pipeline data has the characteristics of nonlinearity and complexity, and the randomness of the interior of the whole system can be measured by utilizing the entropy characteristic parameters, so the entropy characteristic parameters can be used as characteristic indexes to measure the complexity of the interior of the pipeline data, and the distributed entropy is selected as the characteristic parameters in the embodiment of the invention.
(3) Waveform characteristics: when three working condition signals of the pipeline signals are collected, the flow velocity in the pipeline is different, and the waveform difference of the three working condition signals is larger, so that the waveform characteristics of the pipeline signals can effectively reflect different working condition conditions, and the pipeline signals are dimensionless and cannot change due to the change of external physical quantities. The margin factors can reflect different impact characteristics of the signals caused by drastic changes, and the margin factors are selected as characteristic parameters in the embodiment of the invention.
And 5, carrying out model building on the PNN through the extracted feature vector group, and carrying out identification and classification.
According to the embodiment of the invention, the Probabilistic Neural Network (PNN) is developed on the premise of Bayes minimum risk standard, estimates the value density by using a Parzen window method, has a relatively simplified structure, has energy efficiency of classification, analysis and supervision, and is widely applied. The strong nonlinear classification function of the PNN network enables the PNN network to play an important role in classification, and the essence is that a network diagnosis system with strong adaptability and fault tolerance is formed by utilizing the sample mapping capability.
PNN networks are a type of supervised learning network that encompasses a contention layer and a radial base layer. The method comprises the following steps that a four-layer forward neural network consisting of a certain number of Gaussian functions is developed on the basis of a radial basis neural network and mainly comprises an input layer, a mode layer, a summation layer and an output layer, wherein the first layer is the input layer and is used for receiving feature vectors of training samples; the second layer is a mode layer, the distance between the input vector and the training sample is calculated, and the distance result represents the proximity between the output vector and the training sample; the third layer is a summation layer, is connected with the input vector of the second layer, and expresses the output of the neural network as a probability vector; the fourth layer is an output layer, which outputs signals of different states corresponding to each neuron, and the conventional framework thereof is shown in fig. 3.
Another embodiment of the present invention further provides a conduit signal identification system based on VMD and multi-feature fusion, as shown in fig. 4, the system includes:
the signal acquisition module 10 is configured to acquire pipeline signals of different working condition types and construct a data set;
a variational modal decomposition module 20 configured to perform a variational modal decomposition on the pipeline signals in the data set to obtain a plurality of intrinsic modal components; determining the optimal decomposition layer number in the variational modal decomposition by utilizing a multi-angle measuring distance algorithm in the variational modal decomposition;
a feature component determination module 30 configured to compare the degree of similarity of each eigenmode component to the original pipeline signal, thereby determining a feature component;
a feature vector determination module 40 configured to calculate entropy features, waveform features, and time-frequency domain features of the feature components, obtaining a feature vector group based on multi-feature fusion;
a recognition model training module 50 configured to input the feature vector set into a probabilistic neural network for training, so as to obtain a trained pipeline signal recognition model based on the probabilistic neural network;
and the signal identification module 60 is configured to input the pipeline signal to be identified into a trained pipeline signal identification model based on the probabilistic neural network for identification, and obtain an identification result.
In this embodiment, preferably, the working condition types in the signal obtaining module 10 include three working conditions, namely normal working conditions, knocking working conditions, and leakage working conditions.
In this embodiment, preferably, the specific process of determining the optimal decomposition level in the variational modal decomposition by using the multi-angle distance measurement algorithm in the variational modal decomposition module 20 includes: determining the decomposition layer number K in the variational modal decomposition by using the weighted correlation coefficient and the cosine value index, initializing the decomposition layer number K, and calculating the WCC value between adjacent modes corresponding to the decomposition layer number K, wherein the calculation formula of the WCC value is as follows:
WCC＝λCC+ηCD
wherein CC represents a correlation coefficient, CD represents a cosine distance, and lambda and eta are weighting coefficients;
when the WCC value is not larger than a preset maximum threshold value, adding 1 to the decomposition layer number K, and repeating the process, namely calculating the WCC value between adjacent modes corresponding to the decomposition layer number K + 1; and when the WCC value is larger than a preset maximum threshold value and the decomposition layer number is K + N at the moment, the optimal decomposition layer number is K + N-1,N to represent a natural number.
In this embodiment, preferably, the calculation formula of the weighting coefficients λ and η in the variation modal decomposition module 20 is as follows:
wherein the content of the first and second substances,
a coefficient of variation representing the coefficient of correlation,
S CC is the standard deviation of the correlation coefficient, mu CC Is the mean of the correlation coefficients;
a coefficient of variation representing a cosine distance,
S CD is the standard deviation of the cosine distance, mu CD Is the mean of the cosine distances.
In this embodiment, preferably, in the feature vector determining module 40, the entropy features adopt spread entropy, the waveform features adopt margin factors, and the time-frequency domain features adopt standard deviation.
The functions of the VMD and multi-feature fusion based pipeline signal identification system in this embodiment may be described by the aforementioned VMD and multi-feature fusion based pipeline signal identification method, so that a detailed portion in this embodiment may be referred to the above method embodiment, and further description is omitted here.
While the invention has been described with respect to a limited number of embodiments, those skilled in the art, having benefit of this description, will appreciate that other embodiments can be devised which do not depart from the scope of the invention as disclosed herein. The present invention has been disclosed in an illustrative rather than a restrictive sense, and the scope of the present invention is defined by the appended claims.
Claims (10)
1. A pipeline signal identification method based on VMD and multi-feature fusion is characterized by comprising the following steps:
step 1, collecting pipeline signals of different working condition types, and constructing a data set;
step 2, carrying out variation modal decomposition on the pipeline signals in the data set to obtain a plurality of intrinsic modal components; determining the optimal decomposition layer number in the variational modal decomposition by utilizing a multi-angle measuring distance algorithm in the variational modal decomposition;
step 3, comparing the similarity degree of each intrinsic mode component and the original pipeline signal, and determining a characteristic component;
step 4, calculating entropy characteristics, waveform characteristics and time-frequency domain characteristics of the characteristic components to obtain a characteristic vector group based on multi-characteristic fusion;
step 5, inputting the feature vector group into a probabilistic neural network for training to obtain a trained pipeline signal recognition model based on the probabilistic neural network;
and 6, inputting the pipeline signal to be recognized into a trained pipeline signal recognition model based on the probabilistic neural network for recognition, and obtaining a recognition result.
2. The method for identifying pipeline signals based on VMD and multi-feature fusion as claimed in claim 1, wherein the type of the working conditions in step 1 comprises three working conditions of normal, knocking and leakage.
3. The method for identifying a pipeline signal based on VMD and multi-feature fusion according to claim 2, wherein the specific process of determining the optimal number of decomposition layers in the variational modal decomposition by using the multi-angle distance measurement algorithm in step 2 comprises: determining the decomposition layer number K in the variational modal decomposition by using the weighted correlation coefficient and the cosine value index, initializing the decomposition layer number K, and calculating the WCC value between adjacent modes corresponding to the decomposition layer number K, wherein the calculation formula of the WCC value is as follows:
WCC＝λCC+ηCD
wherein CC represents a correlation coefficient, CD represents a cosine distance, and lambda and eta are weighting coefficients;
adding 1 to the number K of decomposition layers when the WCC value is not greater than a preset maximum threshold value; repeating the process, namely calculating the WCC value between the adjacent modes corresponding to the decomposition layer number K + 1; and when the WCC value is larger than a preset maximum threshold value and the decomposition layer number is K + N at the moment, the optimal decomposition layer number is K + N-1,N to represent a natural number.
4. The method for identifying the pipeline signal based on the VMD and the multi-feature fusion as claimed in claim 3, wherein the weighting coefficients λ and η in step 2 are calculated as follows:
wherein the content of the first and second substances,
a coefficient of variation representing the coefficient of correlation,
S CC is the standard deviation of the correlation coefficient, mu CC Is the mean of the correlation coefficients;
a coefficient of variation representing a cosine distance,
S CD is the standard deviation of the cosine distance, mu CD Is the mean of the cosine distances.
5. The method for identifying the pipeline signal based on the VMD and the multi-feature fusion according to claim 4, wherein the entropy feature in step 4 is a dispersion entropy, the waveform feature is a margin factor, and the time-frequency domain feature is a standard deviation.
6. A pipeline signal identification system based on VMD and multi-feature fusion is characterized by comprising:
the signal acquisition module is configured to acquire pipeline signals of different working condition types and construct a data set;
the variable modal decomposition module is configured to perform variable modal decomposition on the pipeline signals in the data set to obtain a plurality of intrinsic modal components; determining the optimal decomposition layer number in the variational modal decomposition by utilizing a multi-angle measuring distance algorithm in the variational modal decomposition;
a characteristic component determination module configured to compare the degree of similarity of each eigenmode component to the original pipeline signal, thereby determining a characteristic component;
a feature vector determination module configured to calculate entropy features, waveform features, and time-frequency domain features of the feature components to obtain a feature vector group based on multi-feature fusion;
the recognition model training module is configured to input the feature vector group into a probabilistic neural network for training to obtain a trained pipeline signal recognition model based on the probabilistic neural network;
and the signal identification module is configured to input the pipeline signal to be identified into a trained pipeline signal identification model based on the probabilistic neural network for identification, and obtain an identification result.
7. The VMD and multi-feature fusion based pipeline signal identification system of claim 6, wherein the type of operating conditions in the signal acquisition module comprises three operating conditions of normal, knocking, and leakage.
8. The system according to claim 7, wherein the specific process of determining the optimal number of decomposition layers in the variational modal decomposition using the multi-angle distance measurement algorithm in the variational modal decomposition module comprises: determining the decomposition layer number K in the variational modal decomposition by using the weighted correlation coefficient and the cosine value index, initializing the decomposition layer number K, and calculating the WCC value between adjacent modes corresponding to the decomposition layer number K, wherein the calculation formula of the WCC value is as follows:
WCC＝λCC+ηCD
wherein CC represents a correlation coefficient, CD represents a cosine distance, and lambda and eta are weighting coefficients;
adding 1 to the number K of decomposition layers when the WCC value is not greater than a preset maximum threshold value; repeating the process, namely calculating the WCC value between the adjacent modes corresponding to the decomposition layer number K + 1; and when the WCC value is larger than a preset maximum threshold value and the decomposition layer number is K + N at the moment, the optimal decomposition layer number is K + N-1,N to represent a natural number.
9. The VMD and multi-feature fusion based pipeline signal identification system of claim 8, wherein the weighting coefficients λ and η in the variational modal decomposition module are calculated as follows:
wherein the content of the first and second substances,
a coefficient of variation representing the coefficient of correlation,
S CC is the standard deviation of the correlation coefficient, mu CC Is the mean of the correlation coefficients;
a coefficient of variation representing a cosine distance,
S CD is a criterion of cosine distanceDifference, mu CD Is the mean of the cosine distances.
10. The system according to claim 9, wherein the entropy features in the feature vector determination module use a dispersion entropy, the waveform features use a margin factor, and the time-frequency domain features use a standard deviation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211019804.7A CN115345203A (en) | 2022-08-24 | 2022-08-24 | Pipeline signal identification method and system based on VMD and multi-feature fusion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211019804.7A CN115345203A (en) | 2022-08-24 | 2022-08-24 | Pipeline signal identification method and system based on VMD and multi-feature fusion |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115345203A true CN115345203A (en) | 2022-11-15 |
Family
ID=83954169
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211019804.7A Pending CN115345203A (en) | 2022-08-24 | 2022-08-24 | Pipeline signal identification method and system based on VMD and multi-feature fusion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115345203A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116415119A (en) * | 2023-04-26 | 2023-07-11 | 山东大学 | Entropy aliasing and feature enhancement-based gas abnormal signal detection method and system |
-
2022
- 2022-08-24 CN CN202211019804.7A patent/CN115345203A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116415119A (en) * | 2023-04-26 | 2023-07-11 | 山东大学 | Entropy aliasing and feature enhancement-based gas abnormal signal detection method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20220318635A1 (en) | Energy identification method for micro-energy device based on bp neural network | |
| CN111734961B (en) | Natural gas pipeline leakage detection method | |
| CN111664365B (en) | Oil and gas pipeline leakage detection method based on improved VMD and 1DCNN | |
| Yang et al. | Novel leakage detection by ensemble 1DCNN-VAPSO-SVM in oil and gas pipeline systems | |
| CN109974782B (en) | Equipment fault early warning method and system based on big data sensitive characteristic optimization selection | |
| KR100795227B1 (en) | Method and apparatus for analyzing signal pattern of sensor array | |
| CN110751108A (en) | Subway distributed vibration signal similarity determination method | |
| CN112039903B (en) | Network security situation assessment method based on deep self-coding neural network model | |
| CN111291860A (en) | Anomaly detection method based on convolutional neural network feature compression | |
| US20240133391A1 (en) | Prediction method for stall and surging of axial-flow compressor based on deep autoregressive network | |
| CN112036042A (en) | Power equipment abnormality detection method and system based on variational modal decomposition | |
| CN111881594B (en) | Non-stationary signal state monitoring method and system for nuclear power equipment | |
| CN113822201B (en) | Deep learning method for underwater object shape recognition based on flow field velocity component time course | |
| CN110674858A (en) | Traffic jam detection method based on space-time association and big data mining | |
| CN115345203A (en) | Pipeline signal identification method and system based on VMD and multi-feature fusion | |
| CN111191855B (en) | Water quality abnormal event identification and early warning method based on pipe network multi-element water quality time sequence data | |
| CN115563880A (en) | Enterprise power consumption abnormity detection method based on Isolated forest-variable point enhancement | |
| CN114547608A (en) | Network security situation assessment method based on noise reduction self-coding kernel density estimation | |
| CN114487129A (en) | Flexible material damage identification method based on acoustic emission technology | |
| CN113626929A (en) | Multi-stage multi-topology ship traffic complexity measuring method and system | |
| CN113361579B (en) | Underwater target detection and identification method, system, equipment and readable storage medium | |
| CN116758922A (en) | Voiceprint monitoring and diagnosing method for transformer | |
| CN113792774B (en) | Intelligent fusion sensing method for underwater targets | |
| Geng et al. | Dispersion Complex Network-Transition Entropy: A Novel Metric for Nonlinear Signal Processing | |
| Ma et al. | Sparse spatiotemporal feature learning for pipeline anomaly detection |
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 |