DAS pipeline flow monitoring method based on lightweight CNN

Abstract

The invention discloses a DAS pipeline flow monitoring method based on a lightweight CNN, and belongs to the technical field of pipeline flow monitoring, and the method comprises the steps: collecting pipeline DAS data through a DAS system, recording a pipeline flow value, and constructing a sample data set; constructing a lightweight pipeline flow monitoring model; training the pipeline flow monitoring model by using the sample data set; and realizing pipeline flow monitoring based on the trained pipeline flow monitoring model. According to the technical scheme, through innovative network architecture design, an advanced training strategy and an efficient post-processing algorithm, the trained model is used for monitoring the pipeline flow, and the pipeline flow value can be rapidly and accurately predicted.

Description

Technical Field

[0001] The present invention relates to the technical field of pipeline flow monitoring, and particularly to a DAS pipeline flow monitoring method based on lightweight CNN. Background Art

[0002] Distributed Acoustic Sensing (DAS) technology utilizes the Rayleigh Backward Scattering (RBS) light in optical fibers to continuously collect vibration data along the optical fiber with a spatial resolution of 1 meter to 10 meters. Its sensing distance can exceed 10 kilometers and has significant advantages such as high temperature resistance, electromagnetic interference resistance, and convenient deployment. In practical applications, the intensity and phase of the received Rayleigh scattered light will change due to the influence of vibration signals along the optical fiber. Through various modulation and demodulation techniques, the intensity and phase information contained in the RBS can be accurately extracted, especially the phase information, which shows a linear correlation with the vibration at each point of the optical fiber. Since the pulsed light emission mode is adopted, the vibration characteristics of each point along the optical fiber can be accurately obtained through the correspondence between the reception time and position.

[0003] DAS technology can monitor the flow position of downhole fluids and production intervals in real time, which is crucial for understanding the production status of oil and gas wells in real time. Through real-time data, engineers can adjust production strategies in a timely manner and optimize stimulation measures. DAS monitoring can reflect the flow distribution of injection or outflow intervals. This helps to identify high-yield and low-yield intervals, so as to conduct fracturing design and adjustment more targeted and improve oil and gas recovery. By establishing a flow prediction model, it is used to optimize fracturing design, such as the distribution of proppants, the injection volume and injection pressure of fracturing fluid, etc., so as to improve the productivity of oil and gas wells. At present, oil and gas are mainly transported by pipelines, which has less loss and low cost. However, the harsh environments of high temperature, high pressure and corrosion in oil wells are still a major pain point and difficulty in the industry.

[0004] Traditional monitoring methods (such as mechanical and electronic flow meters) are expensive, have low reliability, and carry the risk of downhole intervention operations. In contrast, DAS is resistant to high temperatures and pressures, has low costs, and is convenient to deploy, effectively addressing these issues. However, it is very difficult to extract flow information from DAS data. Using deep learning (DL) technology to process DAS data has become a cutting-edge technological trend, showing extensive application potential in fields such as noise suppression, feature extraction, pattern recognition, and event detection. Lightweight convolutional neural networks (CNNs) are a type of convolutional neural network architecture specifically designed to operate efficiently in environments with limited computing resources. Conducting research on DAS pipeline flow monitoring based on lightweight CNNs can not only deepen the integrated application of DL and DAS technologies and reduce computing resource consumption but also improve the oil and gas resource extraction efficiency by optimizing the flow monitoring accuracy. This research result has important practical significance for guiding on-site production practices and promoting the intelligent transformation of the oil and gas industry.

[0005] Currently, the methods for obtaining pipeline flow rates from DAS data are mainly based on knowledge related to fluid mechanics and can be divided into flow monitoring based on the fluid-induced vibration (FIV) effect, energy method flow monitoring, and sonic velocity method flow monitoring based on the Doppler effect.

[0006] (1) Flow monitoring based on the FIV effect: The FIV effect is the vibration effect generated by the interaction between the steadily developing turbulent flow in the pipe and the pipe wall. When the FIV phenomenon occurs, the microstructure optical fiber attached to the pipe wall will vibrate accordingly, causing a corresponding change in the phase of its RBS light. The perception of the in-pipe flow rate can be achieved through the DAS's recovery of the phase.

[0007] (2) Energy method flow monitoring: Inside the pipeline, when the fluid is in a moving state, it will exert a certain impact on the pipe wall. This impact not only transfers energy but also converts most of the kinetic energy into pressure acting on the pipe wall. By using optical fibers for flow detection, a fitting relationship between the pipe wall pressure and the change in the optical phase can be established, and then the corresponding relationship between the flow velocity and the change in the optical phase can be determined. Subsequently, through phase demodulation technology, the flow rate value can be demodulated from the change in the optical phase, thereby realizing the DAS system's monitoring of the flow rate. However, in practical applications, the flow rate value is related to the material, size, and liquid type of the pipeline, and this method has poor robustness.

[0008] (3) Flow monitoring by the sound velocity method based on the Doppler effect: The Doppler effect is based on wave phenomena such as sound waves and light waves. Its essence is that the relative motion between the wave source and the observer causes changes in the observed wavelength and frequency. Since the fluid in the pipeline is a moving object, the Doppler effect occurs between the sound source and the moving fluid. When the fluid flows, it continuously generates sound waves at different positions with the pipe wall and impurities. The DAS system can be used to measure the sound waves propagating in the time domain and spatial domain, and then solve the flow rate in the pipeline.

[0009] For the above-mentioned pipeline flow monitoring methods, their underlying principles are all based on fluid mechanics, and there are mainly the following problems: (1) In complex flow states (such as multiphase flow or non-uniform flow), the manifestation of the FIV effect is more complex, and the vibration signal is difficult to analyze, thus affecting the stability and accuracy of flow measurement. (2) In practical applications of the energy method for flow monitoring, the flow value is related to the material, size of the pipeline, and type of liquid, and this method has poor robustness. (3) Since the acoustic signals generated by the Doppler effect are usually weak and complex, complex signal processing algorithms (such as filtering, noise reduction, or pattern recognition) need to be used to extract effective information, and its computational complexity is high, which poses high requirements for data processing capabilities and computing resources. Therefore, its application scenarios are limited. Summary of the Invention

[0010] The present invention provides a DAS pipeline flow monitoring method based on a lightweight CNN to solve the technical problems of poor stability and accuracy of flow measurement, poor robustness, and high computational complexity existing in the prior art.

[0011] To solve the above technical problems, the present invention provides the following technical solutions:

[0012] On the one hand, the present invention provides a DAS pipeline flow monitoring method based on a lightweight CNN. The DAS pipeline flow monitoring method based on a lightweight CNN includes:

[0013] Using a DAS system to collect pipeline DAS data and record the pipeline flow value to construct a data set;

[0014] Constructing a lightweight pipeline flow monitoring model;

[0015] Training the pipeline flow monitoring model using the data set;

[0016] Realizing pipeline flow monitoring based on the trained pipeline flow monitoring model.

[0017] Further, using a DAS system to collect pipeline DAS data and record the pipeline flow value to construct a data set includes:

[0018] Use the DAS system to collect pipeline DAS data every preset time interval and synchronously record the pipeline flow rate value;

[0019] Perform band-pass filtering on the pipeline DAS data, and use the filtered pipeline DAS data to draw a waterfall plot;

[0020] Classify the category of the pipeline flow rate value according to the preset flow rate category division standard;

[0021] Use the waterfall plot as sample data and the category of the pipeline flow rate value as sample labels to construct a data set.

[0022] Further, the flow rate category division standard is:

[0023] If the flow rate value is in the range of [3, 6), then determine its category as low;

[0024] If the flow rate value is in the range of [6, 9), then determine its category as medium;

[0025] If the flow rate value is in the range of [9, 10.438], then determine its category as high.

[0026] Further, the input of the pipeline flow rate monitoring model is the waterfall plot corresponding to the pipeline DAS data, and the output of the pipeline flow rate monitoring model is the category of the pipeline flow rate value.

[0027] Further, the pipeline flow rate monitoring model includes: a 7×7 convolution layer, a batch normalization layer, a rectified linear unit, a max pooling layer, 6 stacked residual blocks, an average pooling layer, a flattening unit, and a fully connected layer.

[0028] Further, the output sizes of the 6 residual blocks are successively: 1×64×56×56, 1×64×56×56, 1×128×28×28, 1×128×28×28, 1×256×14×14, 1×256×14×14; each residual block uses two layers of depthwise separable convolution and a convolutional block attention module to extract features.

[0029] Further, the process of the pipeline flow rate monitoring model processing the input waterfall plot to obtain the category of the pipeline flow rate value is:

[0030] The pipeline flow rate monitoring model uses a 7×7 convolution layer to perform preliminary feature extraction on the input data to obtain a primary feature map; the primary feature map passes through batch normalization and rectified linear unit activation in sequence, and is downsampled through max pooling to obtain a low-level feature map; the low-level feature map enters the stacked residual blocks to extract high-level features to obtain a high-level feature map; the high-level feature map undergoes average pooling and flattening processing to obtain a feature tensor; the feature tensor passes through a fully connected layer and a Softmax activation function in sequence to obtain the output result.

[0031] Further, when training the pipeline flow monitoring model, the stochastic gradient descent optimization method is adopted; the focal loss is used as the loss function to calculate the error of the model, and the gamma value is set to 2; the step-type learning rate is adopted, the initial value is set to 0.001, and it is reduced to 1 / 10 of the original every 10 rounds; the batch size is 4; the number of training rounds is 50.

[0032] Further, based on the trained pipeline flow monitoring model, pipeline flow monitoring is realized, including:

[0033] Use the DAS system to collect the DAS data of the pipeline to be measured, perform band-pass filtering on the collected DAS data of the pipeline to be measured, and use the filtered DAS data to draw a waterfall plot;

[0034] Input the waterfall plot corresponding to the DAS data of the pipeline to be measured into the trained pipeline flow monitoring model, and use the trained pipeline flow monitoring model to obtain the probabilities that the pipeline flow to be measured belongs to different flow categories;

[0035] Based on the probabilities that the pipeline flow to be measured belongs to different flow categories, weighting is performed to obtain the pipeline flow value.

[0036] Further, the formula for obtaining the pipeline flow value by weighting based on the probabilities that the pipeline flow to be measured belongs to different flow categories is expressed as:

[0037] F = F H ·P H + F M ·P M + F L ·P L

[0038] Among them, F represents the calculated pipeline flow value; P H represents the probability that the pipeline flow to be measured belongs to the high category obtained by the trained pipeline flow monitoring model; P M represents the probability that the pipeline flow to be measured belongs to the medium category obtained by the trained pipeline flow monitoring model; P L represents the probability that the pipeline flow to be measured belongs to the low category obtained by the trained pipeline flow monitoring model; F H represents the average value of the pipeline flow values in the high category; F M represents the average value of the pipeline flow values in the medium category; F L represents the average value of the pipeline flow values in the low category.

[0039] On the other hand, the present invention also provides an electronic device, which includes a processor and a memory; wherein, at least one instruction is stored in the memory, and the instruction is loaded and executed by the processor to implement the above method.

[0040] In another aspect, the present invention also provides a computer-readable storage medium, in which at least one instruction is stored, and the instruction is loaded and executed by a processor to implement the above method.

[0041] The beneficial effects brought by the technical solution provided by the present invention at least include:

[0042] The present invention uses a DAS system to collect DAS data of a pipeline and record the pipeline flow value to construct a data set; constructs a lightweight pipeline flow monitoring model; trains the model; and realizes pipeline flow monitoring based on the trained model. Thus, through an innovative network architecture design, an advanced training strategy, and an efficient post-processing algorithm, the trained model is used to monitor the pipeline flow, and the pipeline flow value can be accurately predicted at an extremely fast speed. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention, and those of ordinary skill in the art can obtain other drawings without creative efforts based on these drawings.

[0044] Figure 1 is a schematic execution flowchart of the DAS pipeline flow monitoring method based on a lightweight CNN provided by an embodiment of the present invention;

[0045] Figure 2 is a schematic implementation principle diagram of the DAS pipeline flow monitoring method based on a lightweight CNN provided by an embodiment of the present invention;

[0046] Figure 3 is a structural diagram of the flow monitoring model LAMNet provided by an embodiment of the present invention;

[0047] Figure 4 is a data distribution diagram provided by an embodiment of the present invention; among them, (a) is the distribution diagram of the original data; (b) is the distribution diagram of the data after feature extraction by pre-trained LAMNet;

[0048] Figure 5 is a system block diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0049] To make the objectives, technical solutions, and advantages of the present invention clearer, the following will further describe the embodiments of the present invention in detail with reference to the drawings.

[0050] First of all, it should be noted that in the embodiments of the present invention, words such as "exemplarily" and "for example" are used to give examples, illustrations or explanations. Any embodiment or design solution described as an "example" in the present invention should not be construed as being more preferred or more advantageous than other embodiments or design solutions. Specifically, the use of the word "exemplarily" is intended to present concepts in a specific way. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one of the two.

[0051] First Embodiment

[0052] This embodiment provides a DAS pipeline flow monitoring method based on a lightweight CNN. This method can be implemented by an electronic device, and the execution process of this method is as Figure 1 shown, and its implementation principle is as Figure 2 shown.

[0053] The DAS pipeline flow monitoring method based on the lightweight CNN includes the following steps:

[0054] S1. Use the DAS system to collect pipeline DAS data and record the pipeline flow value, and construct a data set;

[0055] Specifically, in this embodiment, the implementation process of the above S1 is as follows:

[0056] S11. Use the DAS system to collect pipeline DAS data every 1 minute and synchronously record the pipeline flow value;

[0057] S12. Perform band-pass filtering on the collected pipeline DAS data and draw a waterfall plot;

[0058] S13. Divide the category of the pipeline flow value according to the preset flow category division standard; among them, the flow category division standard is used to set different categories for the data according to the size of the flow value. Specifically: if the flow value is in the range of [3, 6), it is determined that its category is low; if the flow value is in the range of [6, 9), it is determined that its category is medium; if the flow value is in the range of [9, 10.438], it is determined that its category is high;

[0059] S14. Use the waterfall plot as sample data and the category of the pipeline flow value as a label to construct a data set.

[0060] S2. Construct a lightweight pipeline flow monitoring model;

[0061] Among them, the input of the pipeline flow monitoring model (Light-weight Neural Network with AttentionModule, LAMNet) is the waterfall plot corresponding to the pipeline DAS data, and the output is the category of the pipeline flow value.

[0062] Specifically, as Figure 3 shown, the LAMNet model includes: a 7×7 convolution layer, a batch normalization layer, a rectified linear unit, a max pooling layer, 6 stacked residual blocks, an average pooling layer, a flattening unit, and a fully connected layer. Among them, the output sizes of the 6 residual blocks are: 1×64×56×56, 1×64×56×56, 1×128×28×28, 1×128×28×28, 1×256×14×14, 1×256×14×14; each residual block uses two layers of depthwise separable convolution and a convolutional block attention module to extract features. Based on this, the process of the LAMNet model for traffic class prediction is as follows:

[0063] The LAMNet model uses a 7×7 convolution layer to initially extract features from the input data (with a size of 1×3×224×224) to obtain a feature map with a size of 1×64×112×112. Then, it successively passes through batch normalization (BatchNormalization, BN), activation by a rectified linear unit (Rectified Linear Unit, ReLU), and uses max pooling (MaxPool) for downsampling to obtain a low-level feature map with a size of 1×64×56×56. After that, it enters the residual blocks to extract high-level features. In this embodiment, in the LAMNet model, a total of 6 residual blocks are stacked, and their output sizes are 1×64×56×56, 1×64×56×56, 1×128×28×28, 1×128×28×28, 1×256×14×14, 1×256×14×14. For each residual block, two layers of depthwise separable convolution (Depthwise Separable Convolution, DSC) and a convolutional block attention module (Convolutional Block Attention Module, CBAM) are used to extract features. Among them, DSC can significantly reduce the number of parameters compared with the standard convolution, and the CBAM module can double-enhance the features in terms of space and channels. After feature extraction by the stacked residual blocks, average pooling (AvgPool) and flattening (Flatten) are used to obtain a tensor of 1×256. Finally, the network output result is obtained through a fully connected (Fully Connected, FC) layer.

[0064] S3. Use the dataset to train the pipeline traffic monitoring model;

[0065] Among them, a step learning rate (SLR) is introduced in the LAMNet training module of this embodiment to quickly explore the parameter space in the early stage of training and approach the optimal solution in the later stage of training. The focal loss is used as the loss function to handle difficult-to-separate samples.

[0066] Specifically, in this embodiment, the implementation process of the above S3 is as follows:

[0067] S31, divide the dataset into a training set, a validation set, and a test set according to a ratio of 6:2:2;

[0068] S32, perform training on the GPU image processing unit; train on the training set, observe the effect of the pre-trained LAMNet on the validation set, and adjust the hyperparameters; adopt the stochastic gradient descent optimization method; use the focal loss as the loss function to calculate the error of the network, and set the gamma value to 2; adopt the step learning rate, with the initial value set to 0.001, and reduce it to 1 / 10 of the original every 10 rounds; the batch size is 4; the number of training rounds is 50;

[0069] S33, input the data of the test set into the pre-trained LAMNet, and output a 1×3 tensor; use the Softmax activation function for this tensor to obtain the probabilities that the traffic belongs to different classes, and take the maximum value to obtain the predicted traffic class.

[0070] S4, implement pipeline flow monitoring based on the trained pipeline flow monitoring model;

[0071] Among them, this embodiment uses the pre-trained LAMNet to output the traffic class, and after processing by a post-processing algorithm (weighted based on probability), the traffic value is obtained and the prediction error is further reduced.

[0072] Specifically, in this embodiment, the implementation process of the above S4 is as follows:

[0073] S41, use the DAS system to collect the DAS data of the pipeline to be measured, perform band-pass filtering on the collected DAS data of the pipeline to be measured, and use the filtered DAS data to draw a waterfall plot;

[0074] S42, input the waterfall plot corresponding to the DAS data of the pipeline to be measured into the trained pipeline flow monitoring model, and use the trained pipeline flow monitoring model to obtain the probabilities that the pipeline flow to be measured belongs to different flow classes;

[0075] S43, perform weighting based on the probabilities that the pipeline flow to be measured belongs to different flow classes to obtain the pipeline flow value.

[0076] Among them, it should be noted that during model training, there is only one true value for the category, so only one result is output. However, during model inference, the three results (probabilities of different traffic categories) of the Softmax activation function are all valuable, which reflect the probability distribution of the data. Therefore, weighting based on probability is closer to the true value of the traffic. Let the average values of high, medium, and low category traffic be F H 、F M 、F L , and the corresponding probability values be P H 、P M 、P L , then the final predicted traffic value F can be expressed by the following formula:

[0077] F = F H ·P H + F M ·P M + F L ·P L #(1)

[0078] Among them, the traffic mean value of each category is calculated through the data set. Specifically, it is the total traffic value of the traffic data of a certain category divided by the number of data. The specific traffic mean value calculation formula is:

[0079]

[0080] In the formula, C represents the traffic category, N c represents the number of samples in category C, and f i represents the traffic value of the i-th sample in category C.

[0081] In summary, this embodiment provides a DAS pipeline traffic monitoring method based on a lightweight CNN. Based on the pipeline traffic data collected by using the DAS system at an oil field site, a lightweight CNN - LAMNet is built and trained. Through innovative network architecture design, advanced training strategies, and efficient post - processing algorithms, the trained LAMNet is used to monitor the pipeline traffic, and the pipeline traffic value can be accurately predicted at an extremely fast speed. Moreover, the LAMNet model consumes less computing resources (the number of model parameters is 0.395M, the model size is 1.568MB, and the single - image inference time on the commercial GPU RTX3090 is only 7 milliseconds), can fully extract the features of the DAS waterfall diagram, and through the proposed post - processing algorithm, high - precision traffic monitoring is achieved. Figure 4(a) and (b) in it respectively show the distributions of the original data and the distributions after feature extraction by the pre-trained LAMNet. It can be seen that there is a large amount of noise in the original data and it is difficult to distinguish between different traffic categories. After feature extraction by the pre-trained LAMNet, the three traffic categories are clearly divided. Generally speaking, LAMNet has good feature extraction effect on the DAS pipeline traffic data and high discrimination.

[0082] Moreover, it has been verified that by adopting the probability-based weighted post-processing algorithm, the mean absolute error and mean relative error of traffic monitoring can be significantly reduced. As shown in Table 1 and Table 2, this post-processing algorithm is effective in any traffic interval.

[0083] Table 1 Comparison of mean absolute error (MAE) with and without post-processing

[0084]

[0085]

[0086] Table 2 Comparison of mean relative error (MRE) with and without post-processing

[0087] Flow rate range (m3 / h) MRE with post-treatment MRE without post-treatment MRE decrease value Low: 3 - 6 25.28％ 33.32％ 8.04％ Low-medium: 3 - 9 18.83％ 23.84％ 5.01％ Low-medium-high: 3 - 10.438 15.86％ 21.46％ 5.60％ Medium: 6 - 9 11.69％ 13.32％ 1.63％ Medium-high: 6 - 10.438 9.28％ 13.18％ 3.90％ High: 9 - 10.438 5.81％ 12.94％ 7.13％

[0088] Second Embodiment

[0089] This embodiment provides an electronic device, as Figure 5 shown, the electronic device includes: a processor and a memory; wherein, the processor and the memory can be connected through a communication bus; at least one instruction is stored in the memory, and the instruction is loaded and executed by the processor to implement the method of the above first embodiment. In addition, the electronic device may further include a transceiver, the processor and the transceiver can be connected through a communication bus, and the transceiver is used for communicating with other devices.

[0090] Next, in combination with Figure 5 specific introductions will be made to the respective components of this electronic device:

[0091] Among them, the processor is the control center of the electronic device. The electronic device may include multiple processors, and each of these processors may be a single-core processor (single-CPU) or a multi-core processor (multi-CPU). The processor here can be a single processor or a collective term for multiple processing elements. For example, the processor is one or more central processing units (CPUs), or it can also be other general-purpose processors, application specific integrated circuits (ASICs), or one or more integrated circuits configured to implement the embodiments of the present invention. For example: one or more digital signal processors (DSPs), or one or more field programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc. The processor can execute various functions of the electronic device by running or executing software programs stored in the memory and calling data stored in the memory.

[0092] In a specific implementation, as an embodiment, the processor may include one or more CPUs. For example Figure 5 CPU0 and CPU1 shown in [figure reference], of course, this is only an exemplary illustration.

[0093] The memory is used to store the software program for implementing the solution of the present invention and is controlled by the processor for execution. The specific implementation method can refer to the above method embodiments and will not be elaborated here.

[0094] Optionally, the memory may be a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM) or other types of dynamic storage devices that can store information and instructions, or may also be an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM), or other optical disc storage, optical disc storage (including compact discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), magnetic disk storage media, or other magnetic storage devices, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto. The memory may be integrated with the processor or exist independently and be coupled to the processor through the interface circuit ( Figure 5 not shown) of the electronic device. The embodiments of the present invention do not make specific limitations in this regard.

[0095] The transceiver may include a receiver and a transmitter ( Figure 5 not shown separately). Among them, the receiver is used to implement the receiving function, and the transmitter is used to implement the sending function. The transceiver may be integrated with the processor or exist independently and be coupled to the processor through the interface circuit ( Figure 5 not shown) of the electronic device. The embodiments of the present invention do not make specific limitations in this regard.

[0096] In addition, it should be noted that Figure 5 the structure of the electronic device shown in does not constitute a limitation on the device. The actual device may include more or fewer components than shown in the figure, or combine certain components, or have a different component layout. In addition, the technical effects achieved by the electronic device when executing the method of the first embodiment above may refer to the technical effects described in the first embodiment above. Therefore, they will not be elaborated here.

[0097] Third Embodiment

[0098] This embodiment provides a computer-readable storage medium storing at least one instruction, which is loaded and executed by a processor to implement the method of the first embodiment above. Among them, the computer-readable storage medium may be a ROM, a random access memory, a CD-ROM, a magnetic tape, a floppy disk, and an optical data storage device, etc. The instructions stored therein can be loaded and executed by the processor in the terminal to implement the above method.

[0099] In addition, it should be noted that the present invention can be provided as a method, apparatus, or computer program product. Therefore, the embodiments of the present invention can take the form of all or part of a hardware embodiment, all or part of a software embodiment, or an embodiment combining software and hardware aspects. Moreover, when implemented in software, the embodiments of the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present invention are fully or partially generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by wired (such as infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or a data center containing one or more collections of available media. The available media can be magnetic media (such as floppy disks, hard disks, magnetic tapes), optical media (such as DVDs), or semiconductor media. The semiconductor media can be a solid-state drive.

[0100] The embodiments of the present invention are described with reference to the flowcharts and / or block diagrams of methods, terminal devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, and the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, an embedded processor, or other programmable data processing terminal device to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing terminal device generate a device for implementing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0101] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing terminal device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including an instruction device that implements the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1The functions specified in one or more boxes. These computer program instructions can also be loaded onto a computer or other programmable data processing terminal device, so that a series of operation steps are executed on the computer or other programmable terminal device to generate a computer-implemented process. Thus, the instructions executed on the computer or other programmable terminal device provide steps for implementing the functions specified in one or more processes and / or boxes. Figure 1 One process or more processes and / or boxes Figure 1 The steps of the functions specified in one box or more boxes.

[0102] It should also be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. The term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or terminal device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or terminal device. Without further limitation, an element defined by the statement "comprising one..." does not exclude the existence of additional identical elements in the process, method, article or terminal device comprising the element. In addition, the term "and / or" is only a description of the association relationship of associated objects, indicating that three relationships can exist. For example, A and / or B can mean: A exists alone, A and B exist simultaneously, and B exists alone. Among them, A and B can be singular or plural. In addition, the character " / " in this article generally represents an "or" relationship between the associated objects before and after, but it may also represent an "and / or" relationship, which can be understood specifically with reference to the context. "At least one" means one or more, and "a plurality" means two or more. "At least one of the following (items)" or similar expressions refer to any combination of these items, including any combination of single (item) or plural (items). For example, at least one of a, b or c can mean: a, b, c, a - b, a - c, b - c, or a - b - c, where a, b, c can be single or multiple.

[0103] In addition, it can be understood that in various embodiments of the present invention, the magnitudes of the serial numbers of the above processes do not mean the sequence of execution. The execution sequence of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention.

[0104] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods for each specific application to implement the described functions, but such implementation should not be considered to exceed the scope of the present invention.

[0105] In several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of functional modules / units is only a logical functional division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces. The indirect couplings or communication connections of devices or units can be in electrical, mechanical, or other forms. The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment. In addition, in each embodiment of the present invention, the functional units can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit.

[0106] If the method is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in each embodiment of the present invention. The foregoing storage medium includes various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs.

[0107] Finally, it should be noted that the above description is only the preferred embodiment of the present invention. It should be pointed out that although the preferred embodiments of the present invention have been described, for those of ordinary skill in the art, once the basic creative concept of the present invention is known, several improvements and refinements can be made without departing from the principle described in the present invention. These improvements and refinements should also be regarded as the protection scope of the present invention. Therefore, the appended claims are intended to be construed as including the preferred embodiments and all changes and modifications falling within the scope of the embodiments of the present invention.

Claims

1. A DAS pipeline flow monitoring method based on lightweight CNN, characterized in that: include: Use the DAS system to collect pipeline DAS data and record pipeline flow values ​​to build a data set; Build a lightweight pipeline flow monitoring model; Using the data set to train the pipeline flow monitoring model; Pipeline flow monitoring is realized based on the trained pipeline flow monitoring model.

2. The DAS pipeline flow monitoring method based on lightweight CNN as claimed in claim 1 is characterized in that: Use the DAS system to collect pipeline DAS data and record pipeline flow values ​​to build a data set, including: Use the DAS system to collect pipeline DAS data at preset intervals and simultaneously record pipeline flow values; Perform bandpass filtering on the pipeline DAS data, and use the filtered pipeline DAS data to draw a waterfall chart; Classify the pipeline flow value according to the preset flow classification standard; The data set is constructed using the waterfall chart as sample data and the categories of pipeline flow values ​​as sample labels.

3. The DAS pipeline flow monitoring method based on lightweight CNN as claimed in claim 2 is characterized in that: The traffic classification standards are as follows: If the flow value is within the interval [3,6), its category is judged to be low; If the flow value is within the interval [6,9), its category is determined to be medium; If the flow value is within the interval [9,10.438], its category is determined to be high.

4. The DAS pipeline flow monitoring method based on lightweight CNN as claimed in claim 3 is characterized in that: The input of the pipeline flow monitoring model is a waterfall chart corresponding to the pipeline DAS data, and the output of the pipeline flow monitoring model is a pipeline flow value category.

5. The DAS pipeline flow monitoring method based on lightweight CNN as claimed in claim 4 is characterized in that: The pipeline flow monitoring model includes: a layer of 7×7 convolution, a batch normalization layer, a rectified linear unit, a maximum pooling layer, 6 stacked residual blocks, an average pooling layer, a flattening unit and a fully connected layer.

6. The DAS pipeline flow monitoring method based on lightweight CNN as claimed in claim 5 is characterized in that: The output sizes of the six residual blocks are 1×64×56×56, 1×64×56×56, 1×128×28×28, 1×128×28×28, 1×256×14×14, and 1×256×14×14, respectively; each residual block uses two layers of depthwise separable convolution and one convolutional block attention module to extract features.

7. The DAS pipeline flow monitoring method based on lightweight CNN as claimed in claim 6 is characterized in that: The pipeline flow monitoring model processes the input waterfall chart to obtain the pipeline flow value category as follows: The pipeline flow monitoring model uses a layer of 7×7 convolution to perform preliminary feature extraction on the input data to obtain a primary feature map; the primary feature map is sequentially batch normalized and activated by rectified linear units, and then downsampled by maximum pooling to obtain a low-level feature map; The low-level feature map enters the stacked residual block to extract high-level features and obtain a high-level feature map. After the high-level feature map is average pooled and flattened, a feature tensor is obtained. The feature tensor passes through a fully connected layer and a Softmax activation function in turn to obtain the output result.

8. The DAS pipeline flow monitoring method based on lightweight CNN as claimed in claim 1 is characterized in that: When training the pipeline flow monitoring model, the stochastic gradient descent optimization method is used; the focus loss is used as the loss function to calculate the model error, and the gamma value is set to 2; a step-type learning rate is used, the initial value is set to 0.001, and it is reduced to 1 / 10 of the original value every 10 rounds; the batch size is 4; The number of training rounds is 50.

9. The DAS pipeline flow monitoring method based on lightweight CNN as claimed in claim 4 is characterized in that: Pipeline flow monitoring is implemented based on the trained pipeline flow monitoring model, including: Use the DAS system to collect DAS data of the pipeline to be tested, perform bandpass filtering on the collected DAS data of the pipeline to be tested, and use the filtered DAS data to draw a waterfall chart; Input the waterfall chart corresponding to the DAS data of the pipeline to be tested into the trained pipeline flow monitoring model, and use the trained pipeline flow monitoring model to obtain the probability that the flow of the pipeline to be tested belongs to different flow categories; The pipeline flow value is obtained by weighting based on the probability that the pipeline flow to be measured belongs to different flow categories.

10. The DAS pipeline flow monitoring method based on lightweight CNN as claimed in claim 9, characterized in that: Based on the probability that the measured pipeline flow belongs to different flow categories, the formula for obtaining the pipeline flow value is expressed as: F=F H ·P H +F M ·P M +F L ·P L Where F represents the calculated pipeline flow value; P H represents the probability that the pipeline flow to be measured belongs to the high category obtained by the trained pipeline flow monitoring model; P M represents the probability that the pipeline flow to be measured belongs to the middle category obtained by the trained pipeline flow monitoring model; P L represents the probability that the pipeline flow to be measured belongs to the low category obtained by the trained pipeline flow monitoring model; F H Indicates the average value of the high category pipeline flow rate; F M Indicates the average value of the pipeline flow rate in the middle category; F L Indicates the average value of the pipe flow rate in the low category.

Images