CN112801033A - AlexNet network-based construction disturbance and leakage identification method along long oil and gas pipeline - Google Patents
AlexNet network-based construction disturbance and leakage identification method along long oil and gas pipeline Download PDFInfo
- Publication number
- CN112801033A CN112801033A CN202110201069.0A CN202110201069A CN112801033A CN 112801033 A CN112801033 A CN 112801033A CN 202110201069 A CN202110201069 A CN 202110201069A CN 112801033 A CN112801033 A CN 112801033A
- Authority
- CN
- China
- Prior art keywords
- layer
- output
- signal
- leakage
- alexnet network
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/14—Fourier, Walsh or analogous domain transformations, e.g. Laplace, Hilbert, Karhunen-Loeve, transforms
- G06F17/141—Discrete Fourier transforms
- G06F17/142—Fast Fourier transforms, e.g. using a Cooley-Tukey type algorithm
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
- G06F18/24133—Distances to prototypes
- G06F18/24137—Distances to cluster centroïds
- G06F18/2414—Smoothing the distance, e.g. radial basis function networks [RBFN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
Abstract
The invention belongs to the field of pipeline state monitoring, and discloses a construction disturbance and leakage identification method along a long oil and gas pipeline based on an AlexNet network, which comprises the steps of collecting signals and establishing a database; signal processing; building an AlexNet network, and training and testing; and optimizing the AlexNet network, and identifying and the like. The method for identifying the construction disturbance and the leakage along the long oil and gas pipeline based on the AlexNet network has the advantages of high reliability, realization of comprehensive identification of hidden danger and danger of leakage of the long oil and gas pipeline, high identification rate, realization of combination of software and hardware, and capability of detecting all points on an optical fiber link without being influenced by current. The method is suitable for identifying the construction disturbance and leakage along the long-distance oil and gas pipeline.
Description
Technical Field
The invention belongs to the field of pipeline state monitoring, relates to construction disturbance and leakage identification along an oil and gas pipeline, and particularly relates to a construction disturbance and leakage identification method along a long oil and gas pipeline based on an AlexNet network.
Background
In recent years, the construction speed of pipeline transportation in China is rapidly increased, so that a safety early warning technology far behind the development of pipeline transportation is exposed, pipeline transportation accidents frequently happen in recent years due to the asynchronous development of the pipeline transportation technology and the safety early warning technology, and great loss is brought to the life and property safety of people.
The current methods applied in the field of leakage detection mainly include artificial neural networks, support vector machines, etc., and the following describes several conventional algorithms commonly applied to long-distance pipeline signal identification: (1) a pipeline leakage sound signal identification method based on a BP neural network comprises the following steps: the BP neural network is one of artificial neural networks, is a neural network based on error back propagation training, and has the advantage that any mapping relation can be approximated under the condition that the number of hidden layers and nodes is enough. The algorithm is essentially a gradient descent method, the convergence speed of the algorithm is low, the algorithm is relatively inefficient, a large number of characteristic values are required to select and improve the recognition rate, the requirement on case type selection is high, the dependence on samples is high, enough typical samples must be selected for learning, otherwise, the prediction capability is influenced, and the application and training errors are caused. (2) The pipeline abnormal vibration time identification method based on wavelet energy and wavelet information entropy comprises the following steps: the method comprises the steps of performing wavelet transformation on collected acoustic signals, solving signal energy on different scales by applying a wavelet energy spectrum, arranging the energy values in sequence to form a characteristic vector for analysis, identifying a characteristic frequency band of the signals for characteristic extraction, analyzing signal characteristics from a frequency domain and an energy domain, not describing signal characteristics on a time domain, not being suitable for decomposition of non-stationary signals, and having relatively higher difficulty in identifying signals with low signal-to-noise ratio and overlapping of useful signals and noise frequency bands under a strong noise background. (3) An optical fiber early warning signal identification and classification method of a Support Vector Machine (SVM) is applied; and establishing a decision tree, selecting and creating a smooth curved surface in multiple dimensions through the characteristic values, and dividing the sample data into two types, wherein the advantages are obvious in the field of two types of classification, but the performance is inferior in the field of multiple types of classification. (4) The signal identification method based on the Radial Basis Function (RBF) network comprises the following steps: the radial basis function algorithm of the radial basis function neural network, which is derived from multivariable difference values in numerical analysis, is a forward feedback-free neural network, and has high approximation precision, but high network complexity, huge structure and large computation amount.
In summary, in the pipeline leakage detection process, the neural network is adopted for identification, the selection requirements for the typicality, the number of samples and the characteristic value of the sample selection are high, and the defects that the prediction effect is poor if the training samples are few, the overfitting is easy to occur if the training samples are too many, the algorithm convergence speed is low and the like exist. In addition, from the perspective of signal coverage, the prior art has higher attention to leakage signal identification, but has less attention to disturbance signals such as comprehensive artificial invasion and excavation events, and the aspects of 'prevention in advance, timely perception, quick solution' and the like are required to be realized to ensure the safe transportation of pipelines, and the vision needs to be relaxed to grasp the danger early warning and the leakage identification, so that 'advance' and 'report' are really realized.
Disclosure of Invention
The invention aims to provide a method for identifying construction disturbance and leakage along a long oil and gas pipeline based on an AlexNet network, which aims to solve the problems that a traditional identification method needs a step of selecting a characteristic value, the identification rate is reduced due to improper characteristic value selection, and disturbance signals are subjected to event refinement, namely manual excavation and machine excavation, and are combined with leakage signals to realize comprehensive identification of hidden danger and danger of leakage of the long oil and gas pipeline.
In order to achieve the purpose, the technical method comprises the following steps:
a construction disturbance and leakage identification method along a long oil and gas pipeline based on an AlexNet network comprises the following steps:
s1, collecting construction disturbance signals during manual excavation along a long-distance oil and gas pipeline, construction disturbance signals during machine excavation, pipeline leakage signals, environmental background noise signals and soil vibration signals, establishing a field signal database, building a laboratory simulation field by analyzing relevant characteristics, simulating the state of the oil and gas pipeline under different working conditions, collecting experimental data of the laboratory simulation field, and establishing a laboratory signal database;
s2, performing signal processing on signals in the field signal database and the laboratory signal database to obtain a time domain diagram and a frequency domain diagram;
s3, building an AlexNet network, dividing the time domain graph and the frequency domain graph obtained in the step S2 into a training set and a testing set, training the AlexNet network by using the training set, testing the recognition rate of the AlexNet network by using the testing set, obtaining the recognition rate of each type of working condition, and storing the recognition result;
and S4, adjusting parameters of the AlexNet network according to the test result of the AlexNet network identification rate, optimizing the AlexNet network, identifying construction disturbance and leakage along the long oil and gas pipeline by using the optimized AlexNet network, and uploading the result to an upper computer for storage and image display.
As a limitation: in the step S1, signal acquisition is carried out by adopting a sensing method based on a phi-OTDR technology, a sensing optical cable is laid in the same ditch of an oil-gas transportation pipeline, when the oil-gas transportation pipeline is subjected to external force, the applied external force is contacted with soil to generate a vibration signal, the vibration signal is transmitted along the soil, is acquired by a distributed micro-vibration optical fiber sensor, is amplified by a relay amplifier and is transmitted to a sensing optical cable, and the sensing optical cable is disturbed by external signals and is transmitted to a light vibration signal detection device by a guide optical cable; s11, changing the diameter of a leakage point, the gas-liquid flow and the circumferential position of the leakage point, collecting a soil vibration signal at the moment, and analyzing the difference between the flow rate and pressure change rule caused by the transient operation and the leakage; compressing air in the pipeline by using a wind pressure machine, collecting a soil vibration signal at the moment, and continuously collecting a signal of subsequent rupture and leakage of the pipeline; in an environment with background noise, carrying out experimental simulation at different distances from a pipeline, simulating the influence on the pipeline under the operations of digging and passing a vehicle, and collecting a soil vibration signal at the moment; and S12, building a laboratory simulation field according to the collected signals.
As a further limitation: the signal processing in step S2 is: s21, framing processing: performing framing processing on signals in a field signal database and a laboratory signal database to obtain time domain signals; s22, fast Fourier transform: converting the time domain signal into a frequency domain signal through fast Fourier transform; s23, filtering by using a filter: filtering the frequency domain signal by using a filter, wherein the filter is a band-pass filter which consists of a high-resistance filter and a low-resistance filter and is optimized in performance by using an SVM (support vector machine) identification method; and S24, drawing the time domain signal into a time domain diagram, drawing the filtered frequency domain signal into a frequency domain diagram, and compressing the frequency domain diagram and the frequency domain diagram to be used as the input of the AlexNet network.
As a further limitation: in step S3, the AlexNet network has eight layers in total, and its specific structure is:
the first layer is a convolution layer, the time domain graph and the frequency domain graph in the step S2 are used as input, ReLU is used as an activation function, the number of cores is 96, the size of the cores is 11 × 11, the step is 4 × 4, the size of an output matrix is 55 × 96, and two groups of matrices 27 × 96 are output after overlapping pooling normalization;
the second layer is a convolution layer, the output of the first layer is used as input, the kernel number is 256, the kernel size is 5 × 5, the step is 1 × 1, the output matrix size is 27 × 256, and two groups of 13 × 256 matrixes are output after overlapping and pooling;
the third layer is a convolution layer, the output of the second layer is used as input, the number of cores is 384, the size of each core is 3 x 3, the step is 1 x 1, and the size of an output matrix is 13 x 384;
the fourth layer is a convolution layer, the output of the third layer is used as input, the number of cores is 384, the size of each core is 3 x 3, the step is 1 x 1, and the size of an output matrix is 13 x 384;
the fifth layer is a convolution layer, the output of the fourth layer is used as input, the number of cores is 256, the size of each core is 3 × 3, the step is 1 × 1, the size of an output matrix is 13 × 256, and the output matrixes are subjected to superposition pooling to output two groups of matrixes 6 × 256;
the sixth layer is a fully-connected layer, the output of the fifth layer is used as input, the size of the filter is 6 x 256, each filter performs convolution operation on input data to generate an operation result, and the operation result is output through a neuron; 4096 filters are adopted, and 4096 data are output through a ReLU activation function and a dropout operation;
the seventh layer is a full connection layer, 4096 data output by the sixth layer is fully connected with 4096 neurons of the seventh layer by taking the output of the sixth layer as input, and 4096 data are generated after being processed by the ReLU and Dropout;
the eighth layer is a full connection layer, the output of the seventh layer is used as input, 4096 data input by the seventh layer are fully connected with 1000 neurons of the eighth layer, and trained values are output.
Due to the adoption of the scheme, compared with the prior art, the invention has the beneficial effects that:
(1) according to the construction disturbance and leakage identification method along the long-distance oil and gas pipeline based on the AlexNet network, provided by the invention, a laboratory simulation field is built, laboratory environments simulated by different working condition environments are built on the basis of real monitored field data, real external construction disturbance, leakage and background noise conditions are simulated, a laboratory database is built and is mutually complemented with a field database, data information is enriched, the requirement of the AlexNet network on the number of learning objects is met, and the trained AlexNet network has better engineering applicability and high engineering reliability; in addition, events of the disturbance signals are divided into manual excavation and machine excavation, and leakage signals are combined to realize comprehensive long-distance oil and gas pipeline leakage hidden danger and danger identification;
(2) according to the method for identifying the construction disturbance and leakage along the long oil and gas pipeline based on the AlexNet network, the filter is designed based on the real construction disturbance and pipeline leakage signals along the long oil and gas pipeline acquired on site, the interference of some noises on the signal characteristics is greatly reduced, the performance of the band-pass filter is optimized through the identification method of the SVM, the filter with the highest identification rate under the condition of controlling other variables to be unchanged is selected preferably, and the identification accuracy is improved;
(3) according to the AlexNet network-based construction disturbance and leakage identification method along the long oil and gas pipeline, the traditional acoustic signal identification is converted into image identification, and the AlexNet network is adopted for image identification, so that the method has the characteristics of abundant characteristics, enhanced data, enhanced model generalization capability and reduced communication performance loss, not only solves the problem of insufficient storage capacity of hardware, but also improves the processing capability of a processor, further improves the overall performance of the whole network, and in addition, the characteristic value selection is not required, the problem of identification rate reduction caused by improper characteristic value selection is avoided, and the identification rate is improved;
(4) the invention provides a construction disturbance and leakage identification method along a long-distance oil and gas pipeline based on an AlexNet network, wherein a sensing method based on a phi-OTDR technology is adopted for signal acquisition, a time domain scanning mode is adopted for detecting vibration signals along the pipeline, when light pulses are transmitted to an optical fiber section acted by external vibration signals, the light signals of Rayleigh scattering back to a detector are changed, the vibration condition along the pipeline can be detected by detecting the change of scattered light signals, the positioning can be accurate, LabVIEW software is used for storing and displaying images of the acquired signals, the combination of software and hardware is realized, and all points on an optical fiber link can be detected without being influenced by current.
In conclusion, the method for identifying the construction disturbance and the leakage along the long oil and gas pipeline based on the AlexNet network has high reliability, realizes comprehensive identification of hidden danger and danger of leakage of the long oil and gas pipeline, has high identification rate, realizes combination of software and hardware, and can detect all points on an optical fiber link without being influenced by current.
The method is suitable for identifying the construction disturbance and leakage along the long-distance oil and gas pipeline.
Drawings
The invention is described in further detail below with reference to the figures and the embodiments.
FIG. 1 is a flow chart of an embodiment of the present invention;
FIG. 2 is a time domain diagram after compression according to an embodiment of the present invention;
FIG. 3 is a frequency domain diagram after compression according to an embodiment of the present invention.
Detailed Description
The present invention is further described with reference to the following examples, but it should be understood by those skilled in the art that the present invention is not limited to the following examples, and any modifications and equivalent changes based on the specific examples of the present invention are within the scope of the claims of the present invention.
Embodiment based on AlexNet network, construction disturbance and leakage identification method along long oil and gas pipeline
A construction disturbance and leakage identification method along a long oil and gas pipeline based on an AlexNet network is disclosed, a flow chart of which is shown in figure 1, and comprises the following steps:
s1, collecting construction disturbance signals when a long oil and gas pipeline is manually excavated within a range of 5 meters along the pipeline, construction disturbance signals when a machine is excavated within a range of 50 meters along the pipeline, pipeline leakage signals, environmental background noise signals and soil vibration signals by adopting a sensing method based on a phi-OTDR technology, establishing a field signal database, establishing a laboratory simulation field by analyzing relevant characteristics, simulating the states of the oil and gas pipeline under different working conditions, collecting experimental data of the laboratory simulation field, and establishing a laboratory signal database;
the method comprises the following steps that sensing optical cables are laid in the same ditch of an oil and gas transportation pipeline, when the oil and gas transportation pipeline is subjected to external force, the applied external force is contacted with soil to generate vibration signals, the vibration signals are transmitted along the soil, collected by a distributed micro-vibration optical fiber sensor, amplified by a relay amplifier and transmitted to the sensing optical cables, and the sensing optical cables are disturbed by external signals and transmitted to a light vibration signal detection device through a guide optical cable;
a method for building a laboratory simulation field comprises the following steps:
s11, changing the diameter of a leakage point, the gas-liquid flow and the circumferential position of the leakage point, collecting a soil vibration signal at the moment, and analyzing the difference between the flow rate and pressure change rule caused by the transient operation and the leakage; compressing air in the pipeline by using a wind pressure machine, collecting a soil vibration signal at the moment, and continuously collecting a signal of subsequent rupture and leakage of the pipeline; in an environment with background noise, carrying out experimental simulation at different distances from a pipeline, simulating the influence on the pipeline under the operations of digging and passing a vehicle, and collecting a soil vibration signal at the moment;
and S12, building a laboratory simulation field according to the collected signals.
S2, signal processing is carried out on the signals in the field signal database and the laboratory signal database, and the signal processing is as follows:
s21, framing processing: performing framing processing on signals in a field signal database and a laboratory signal database according to time, wherein the frame length is 1s, and obtaining time domain signals;
s22, fast Fourier transform: converting the time domain signal into a frequency domain signal through fast Fourier transform;
s23, filtering by using a filter: filtering the frequency domain signal by using a filter, wherein the filter is a band-pass filter which is composed of a high-resistance filter and a low-resistance filter and performs performance optimization through an SVM (support vector machine) identification method, and the accessible frequency range is 300-2500 HZ;
s24, plotting the time domain signal into a time domain graph, plotting the filtered frequency domain signal into a frequency domain graph, and performing pressure-compression to 227 × 3, where the compressed time domain graph is shown in fig. 2, and the compressed frequency domain graph is shown in fig. 3, and is used as the input of the AlexNet network.
S3, building an AlexNet network, dividing the time domain graph and the frequency domain graph obtained in the step S2 into a training set and a testing set, training the AlexNet network by using the training set, testing the recognition rate of the AlexNet network by using the testing set, obtaining the recognition rate of each type of working condition, and storing the recognition result;
the AlexNet network has eight layers in total, and the specific structure is as follows:
the first layer is a convolution layer, the compressed time domain graph and the frequency domain graph are used as input, the ReLU is used as an activation function, the number of kernels is 96, the size of each kernel is 11 × 11, the step is 4 × 4, the size of an output matrix is 55 × 96, and two groups of 27 × 96 matrixes are output after overlapping and pooling;
the second layer is a convolution layer, the output of the first layer is used as input, the kernel number is 256, the kernel size is 5 × 5, the step is 1 × 1, the output matrix size is 27 × 256, and two groups of 13 × 256 matrixes are output after overlapping and pooling;
the third layer is a convolution layer, the output of the second layer is used as input, the number of cores is 384, the size of each core is 3 x 3, the step is 1 x 1, and the size of an output matrix is 13 x 384;
the fourth layer is a convolution layer, the output of the third layer is used as input, the number of cores is 384, the size of each core is 3 x 3, the step is 1 x 1, and the size of an output matrix is 13 x 384;
the fifth layer is a convolution layer, the output of the fourth layer is used as input, the number of cores is 256, the size of each core is 3 × 3, the step is 1 × 1, the size of an output matrix is 13 × 256, and the output matrixes are subjected to superposition pooling to output two groups of matrixes 6 × 256;
the sixth layer is a fully-connected layer, the output of the fifth layer is used as input, the size of the filter is 6 x 256, each filter performs convolution operation on input data to generate an operation result, and the operation result is output through a neuron; 4096 filters are adopted, and 4096 data are output through a ReLU activation function and a dropout operation;
the seventh layer is a full connection layer, 4096 data output by the sixth layer is fully connected with 4096 neurons of the seventh layer by taking the output of the sixth layer as input, and 4096 data are generated after being processed by the ReLU and Dropout;
the eighth layer is a full connection layer, the output of the seventh layer is used as input, 4096 data input by the seventh layer are fully connected with 1000 neurons of the eighth layer, and trained values are output.
S4, adjusting parameters of the AlexNet network according to the test result of the AlexNet network identification rate, optimizing the AlexNet network, identifying construction disturbance and leakage along the long-distance oil and gas pipeline by using the optimized AlexNet network, uploading the result to an upper computer for storage and image display, and storing and displaying the acquired signal by the upper computer through LabVIEW software.
The average recognition rate obtained in this example under each condition is shown in table 1, the average recognition rate in this example under each condition is 94.26%, and the comparison with the recognition rate in the prior art is shown in table 2.
Table 1 average recognition rate of each operating condition obtained in this embodiment
TABLE 2 comparison of the recognition rates of the present example with those of the prior art
As can be seen from tables 1 and 2, the embodiment of the present invention identifies signals under different working conditions, and the identification rate is higher than that of the prior art.
Claims (4)
1. A construction disturbance and leakage identification method along a long oil and gas pipeline based on an AlexNet network is characterized by comprising the following steps:
s1, collecting construction disturbance signals during manual excavation along a long-distance oil and gas pipeline, construction disturbance signals during machine excavation, pipeline leakage signals, environmental background noise signals and soil vibration signals, establishing a field signal database, building a laboratory simulation field by analyzing relevant characteristics, simulating the state of the oil and gas pipeline under different working conditions, collecting experimental data of the laboratory simulation field, and establishing a laboratory signal database;
s2, performing signal processing on signals in the field signal database and the laboratory signal database to obtain a time domain diagram and a frequency domain diagram;
s3, building an AlexNet network, dividing the time domain graph and the frequency domain graph obtained in the step S2 into a training set and a testing set, training the AlexNet network by using the training set, testing the recognition rate of the AlexNet network by using the testing set, obtaining the recognition rate of each type of working condition, and storing the recognition result;
and S4, adjusting parameters of the AlexNet network according to the test result of the AlexNet network identification rate, optimizing the AlexNet network, identifying construction disturbance and leakage along the long oil and gas pipeline by using the optimized AlexNet network, and uploading the result to an upper computer for storage and image display.
2. The AlexNet network-based construction disturbance and leakage identification method along the long oil and gas pipeline according to claim 1, characterized in that in step S1, the signal acquisition is performed by a sensing method based on a phi-OTDR technology, the oil and gas pipeline is laid with a sensing optical cable in the same trench, when the oil and gas pipeline is subjected to an external force, the applied external force is contacted with soil to generate a vibration signal, the vibration signal is transmitted along the soil, acquired by a distributed micro-vibration optical fiber sensor, amplified by a relay amplifier and transmitted to the sensing optical cable, the sensing optical cable is disturbed by an external signal and transmitted to a light vibration signal detection device through a guide optical cable:
s11, changing the diameter of a leakage point, the gas-liquid flow and the circumferential position of the leakage point, collecting a soil vibration signal at the moment, and analyzing the difference between the flow rate and pressure change rule caused by the transient operation and the leakage; compressing air in the pipeline by using a wind pressure machine, collecting a soil vibration signal at the moment, and continuously collecting a signal of subsequent rupture and leakage of the pipeline; in an environment with background noise, carrying out experimental simulation at different distances from a pipeline, simulating the influence on the pipeline under the operations of digging and passing a vehicle, and collecting a soil vibration signal at the moment;
and S12, building a laboratory simulation field according to the collected signals.
3. The AlexNet network-based construction disturbance and leakage identification method along the long oil and gas pipeline according to claim 1 or 2, characterized in that the signal processing in the step S2 is as follows:
s21, framing processing: performing framing processing on signals in a field signal database and a laboratory signal database to obtain time domain signals;
s22, fast Fourier transform: converting the time domain signal into a frequency domain signal through fast Fourier transform;
s23, filtering by using a filter: filtering the frequency domain signal by using a filter, wherein the filter is a band-pass filter which consists of a high-resistance filter and a low-resistance filter and is optimized in performance by using an SVM (support vector machine) identification method;
and S24, drawing the time domain signal into a time domain diagram, drawing the filtered frequency domain signal into a frequency domain diagram, and compressing the frequency domain diagram and the frequency domain diagram to be used as the input of the AlexNet network.
4. The AlexNet network-based construction disturbance and leakage identification method along the long oil and gas pipeline according to claim 1, wherein the AlexNet network in the step S3 has eight layers, and the specific structure is as follows:
the first layer is a convolution layer, the time domain graph and the frequency domain graph in the step S2 are used as input, ReLU is used as an activation function, the number of cores is 96, the size of the cores is 11 × 11, the step is 4 × 4, the size of an output matrix is 55 × 96, and two groups of matrices 27 × 96 are output after overlapping pooling normalization;
the second layer is a convolution layer, the output of the first layer is used as input, the kernel number is 256, the kernel size is 5 × 5, the step is 1 × 1, the output matrix size is 27 × 256, and two groups of 13 × 256 matrixes are output after overlapping and pooling;
the third layer is a convolution layer, the output of the second layer is used as input, the number of cores is 384, the size of each core is 3 x 3, the step is 1 x 1, and the size of an output matrix is 13 x 384;
the fourth layer is a convolution layer, the output of the third layer is used as input, the number of cores is 384, the size of each core is 3 x 3, the step is 1 x 1, and the size of an output matrix is 13 x 384;
the fifth layer is a convolution layer, the output of the fourth layer is used as input, the number of cores is 256, the size of each core is 3 × 3, the step is 1 × 1, the size of an output matrix is 13 × 256, and the output matrixes are subjected to superposition pooling to output two groups of matrixes 6 × 256;
the sixth layer is a fully-connected layer, the output of the fifth layer is used as input, the size of the filter is 6 x 256, each filter performs convolution operation on input data to generate an operation result, and the operation result is output through a neuron; 4096 filters are adopted, and 4096 data are output through a ReLU activation function and a dropout operation;
the seventh layer is a full connection layer, 4096 data output by the sixth layer is fully connected with 4096 neurons of the seventh layer by taking the output of the sixth layer as input, and 4096 data are generated after being processed by the ReLU and Dropout;
the eighth layer is a full connection layer, the output of the seventh layer is used as input, 4096 data input by the seventh layer are fully connected with 1000 neurons of the eighth layer, and trained values are output.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110201069.0A CN112801033A (en) | 2021-02-23 | 2021-02-23 | AlexNet network-based construction disturbance and leakage identification method along long oil and gas pipeline |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110201069.0A CN112801033A (en) | 2021-02-23 | 2021-02-23 | AlexNet network-based construction disturbance and leakage identification method along long oil and gas pipeline |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112801033A true CN112801033A (en) | 2021-05-14 |
Family
ID=75815402
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110201069.0A Pending CN112801033A (en) | 2021-02-23 | 2021-02-23 | AlexNet network-based construction disturbance and leakage identification method along long oil and gas pipeline |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112801033A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113919395A (en) * | 2021-10-12 | 2022-01-11 | 大连理工大学 | Water supply pipe network leakage accident diagnosis method based on one-dimensional convolutional neural network |
| TWI814661B (en) * | 2022-12-09 | 2023-09-01 | 明志科技大學 | Method for discriminating pipeline leak and system thereof |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101183899A (en) * | 2007-12-19 | 2008-05-21 | 天津大学 | BP network based pipeline security identifying method for optical fiber pipeline leakage monitoring device |
| US20180341859A1 (en) * | 2017-05-24 | 2018-11-29 | Southwest Research Institute | Detection of Hazardous Leaks from Pipelines Using Optical Imaging and Neural Network |
| CN109374115A (en) * | 2018-12-07 | 2019-02-22 | 武汉理工光科股份有限公司 | Broken vibration monitoring recognition methods outside oil-gas pipeline based on φ-OTDR |
| CN110185939A (en) * | 2019-05-16 | 2019-08-30 | 西北工业大学 | Gas pipe leakage recognition methods based on convolutional neural networks |
| CN211123729U (en) * | 2020-01-20 | 2020-07-28 | 西安科技大学 | Oil gas long distance pipeline monitored control system along line |
| US20200240940A1 (en) * | 2019-01-30 | 2020-07-30 | Xi'an Jiaotong University | Method for detecting an air discharge decomposed product based on a virtual sensor array |
-
2021
- 2021-02-23 CN CN202110201069.0A patent/CN112801033A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101183899A (en) * | 2007-12-19 | 2008-05-21 | 天津大学 | BP network based pipeline security identifying method for optical fiber pipeline leakage monitoring device |
| US20180341859A1 (en) * | 2017-05-24 | 2018-11-29 | Southwest Research Institute | Detection of Hazardous Leaks from Pipelines Using Optical Imaging and Neural Network |
| CN109374115A (en) * | 2018-12-07 | 2019-02-22 | 武汉理工光科股份有限公司 | Broken vibration monitoring recognition methods outside oil-gas pipeline based on φ-OTDR |
| US20200240940A1 (en) * | 2019-01-30 | 2020-07-30 | Xi'an Jiaotong University | Method for detecting an air discharge decomposed product based on a virtual sensor array |
| CN110185939A (en) * | 2019-05-16 | 2019-08-30 | 西北工业大学 | Gas pipe leakage recognition methods based on convolutional neural networks |
| CN211123729U (en) * | 2020-01-20 | 2020-07-28 | 西安科技大学 | Oil gas long distance pipeline monitored control system along line |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113919395A (en) * | 2021-10-12 | 2022-01-11 | 大连理工大学 | Water supply pipe network leakage accident diagnosis method based on one-dimensional convolutional neural network |
| TWI814661B (en) * | 2022-12-09 | 2023-09-01 | 明志科技大學 | Method for discriminating pipeline leak and system thereof |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11562224B2 (en) | 1D-CNN-based distributed optical fiber sensing signal feature learning and classification method | |
| CN111638028B (en) | High-voltage parallel reactor mechanical state evaluation method based on vibration characteristics | |
| US20220092428A1 (en) | Prediction method for stall and surge of axial compressor based on deep learning | |
| CN110398647B (en) | Transformer state monitoring method | |
| CN110991295B (en) | Self-adaptive fault diagnosis method based on one-dimensional convolutional neural network | |
| CN112801033A (en) | AlexNet network-based construction disturbance and leakage identification method along long oil and gas pipeline | |
| CN111442827A (en) | Optical fiber passive online monitoring system and method for transformer winding vibration | |
| CN110823356B (en) | Distributed optical fiber intrusion detection method based on Mel frequency spectrum | |
| CN104616416A (en) | Multi-sensor information fusion-based wireless fire alarm system | |
| CN112668105B (en) | Helicopter transmission shaft abnormity judgment method based on SAE and Mahalanobis distance | |
| CN105626502A (en) | Plunger pump health assessment method based on wavelet packet and Laplacian Eigenmap | |
| CN110632662A (en) | Algorithm for automatically identifying microseism signals by using DCNN-inclusion network | |
| CN112052712B (en) | Power equipment state monitoring and fault identification method and system | |
| CN114169377A (en) | G-MSCNN-based fault diagnosis method for rolling bearing in noisy environment | |
| CN111881594B (en) | Non-stationary signal state monitoring method and system for nuclear power equipment | |
| CN116517860A (en) | Ventilator fault early warning system based on data analysis | |
| CN115824519A (en) | Valve leakage fault comprehensive diagnosis method based on multi-sensor information fusion | |
| CN1120366C (en) | Fault detecting and diagnosing method based on non-linear spectral analysis | |
| Ding et al. | Gear fault diagnosis based on VMD sample entropy and discrete hopfield neural network | |
| CN115931141A (en) | Temperature identification method of infrared temperature measurement map based on improved ANN algorithm | |
| WO2023004180A1 (en) | Identification of false transformer humming using machine learning | |
| CN115130748A (en) | Building electrical system fault diagnosis method | |
| Rui et al. | Research on fault diagnosis and state assessment of vacuum pump based on acoustic emission sensors | |
| CN110108489B (en) | Method for acquiring performance degradation trend of rolling bearing | |
| CN113686577A (en) | Bearing fault diagnosis method based on rapid nonlinear sparse spectrum |
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 |