Fluid identification method, device, electronic equipment and storage medium for marine riser

Abstract

The present application provides a method, device, electronic device, and storage medium for identifying fluid in a marine riser; applied to the field of marine platform equipment safety, the method comprises: obtaining liquid holdup data of at least one target location in the marine riser; preprocessing the liquid holdup data to obtain a first sample data set; the first sample data set includes the liquid holdup data after frame processing; training a preset long short-term memory network model based on the first sample data set to obtain a first prediction model; obtaining a second prediction model; the second prediction model is obtained by training a preset convolutional neural network model using a second sample data set corresponding to a liquid holdup time history curve; identifying the fluid to be tested in the marine riser based on the first prediction model and the second prediction model to obtain a fluid identification result. In this way, flow changes in the pipeline and the flow state can be accurately and real-time predicted, and potential risks can be warned in advance.

Description

Technical Field

[0001] The present application relates to the field of marine platform equipment safety technology, and in particular to a method, device, electronic equipment, and storage medium for identifying fluid in a marine riser. Background Art

[0002] In the process of offshore oil production, flexible risers, as key equipment connecting the seabed wellhead and the offshore platform, undertake the important task of transporting mixed oil and gas. The flow patterns in the pipeline are complex and changeable, including slug flow, annular flow and other types. These different flow patterns have a significant impact on production efficiency, equipment life and safety. For example, slug flow may cause pressure fluctuations and increased equipment wear in the pipeline, and even cause safety accidents; while annular flow will affect the efficiency of crude oil transportation under certain working conditions. Traditional fluid identification methods mainly rely on simple threshold judgments and are difficult to cope with complex flow changes. Therefore, current fluid identification methods are unable to accurately and in real time predict flow changes in the pipeline and identify flow patterns, and it is difficult to warn of potential risks in advance. Summary of the Invention

[0003] Embodiments of the present application provide a method, device, electronic device, and storage medium for identifying fluid in a marine riser to solve one or more problems existing in the related art.

[0004] According to a first aspect of the present application, a method for identifying fluid in a marine riser is provided, comprising: obtaining liquid holdup data of at least one target location in the marine riser; preprocessing the liquid holdup data to obtain a first sample data set; the first sample data set includes the liquid holdup data after frame processing; training a preset long short-term memory network model based on the first sample data set to obtain a first prediction model; the first prediction model is used to predict a liquid holdup time history curve of the fluid in the marine riser; obtaining a second prediction model; the second prediction model is obtained by training a preset convolutional neural network model using a second sample data set corresponding to the liquid holdup time history curve; the second sample data set includes liquid holdup time history curves of known manifold categories; identifying the fluid to be tested in the marine riser to be tested based on the first prediction model and the second prediction model to obtain a fluid identification result.

[0005] According to one embodiment of the present application, obtaining liquid holdup data at at least one target location in a marine riser includes: the target location includes at least one of the following: a riser inlet, a middle portion of a descending section, a dome point, a sag point, and a middle portion of a hanging section; and collecting liquid holdup data of the fluid in the marine riser using a conductive probe at the target location.

[0006] According to one embodiment of the present application, the preprocessing of the liquid holdup data to obtain a first sample data set includes: performing data processing on the liquid holdup data through a sliding window algorithm, and normalizing the data obtained from the data processing to obtain time series data; performing frame processing on the time series data to obtain frame-processed data; performing variational mode decomposition and channel attention mechanism processing on the frame-processed data to obtain feature data; the variational mode decomposition is used to decompose the frame-processed data into multiple modes; the channel attention mechanism processing is used to assign different weights to the data of the same mode obtained by decomposition; and constructing the first sample data set based on the feature data.

[0007] According to one embodiment of the present application, the method of training a preset long short-term memory network model based on the first sample data set to obtain a first prediction model includes: dividing the first sample data set into a first training data set and a first verification data set based on a preset first ratio; inputting the first training data set into a preset long short-term memory network model, training the preset long short-term memory network model, and obtaining a trained long short-term memory network model; determining a first prediction result of the trained long short-term memory network model based on the first verification data set; and determining the trained long short-term memory network model as the first prediction model in response to the first prediction result satisfying a preset prediction condition.

[0008] According to one embodiment of the present application, in response to the first prediction result satisfying a preset prediction condition, the trained long short-term memory network model is determined as the first prediction model, including: determining the coefficient of determination, root mean square error and mean absolute error corresponding to the first prediction result; performing a goodness of fit evaluation on the first prediction result based on the coefficient of determination, root mean square error and mean absolute error to obtain an evaluation result; in response to the evaluation result satisfying the preset prediction condition, the trained long short-term memory network model is determined as the first prediction model.

[0009] According to one embodiment of the present application, the training of a preset convolutional neural network model using a second sample data set corresponding to a liquid holdup time history curve includes: dividing the two sample data sets into a second training data set and a second validation data set based on a preset second ratio; inputting the second training data set into a preset convolutional neural network model, training the preset convolutional neural network model to obtain a trained convolutional neural network model; determining the classification accuracy of the trained convolutional neural network model based on the second validation data set; and determining the trained convolutional neural network model as the second prediction model in response to the classification accuracy satisfying a preset classification condition.

[0010] According to one embodiment of the present application, the identifying the fluid to be measured in the marine riser to be measured based on the first prediction model and the second prediction model to obtain a fluid identification result includes: performing flow prediction on the fluid to be measured in the marine riser to be measured based on the first prediction model to obtain a liquid holdup time history curve of the fluid to be measured; performing manifold identification on the liquid holdup time history curve of the fluid to be measured based on the second prediction model to obtain a manifold category of the fluid to be measured; and determining the fluid identification result based on the liquid holdup time history curve and the manifold category.

[0011] According to a second aspect of the present application, a fluid identification device for a marine riser is provided, comprising: a first acquisition module for acquiring liquid holdup data of at least one target location in the marine riser; a preprocessing module for preprocessing the liquid holdup data to obtain a first sample data set; the first sample data set includes the liquid holdup data after frame processing; a training module for training a preset long short-term memory network model based on the first sample data set to obtain a first prediction model; the first prediction model is used to predict a liquid holdup time history curve of the fluid in the marine riser; a second acquisition module for acquiring a second prediction model; the second prediction model is obtained by training a preset convolutional neural network model using a second sample data set corresponding to the liquid holdup time history curve; the second sample data set includes liquid holdup time history curves of known manifold categories; and an identification module for identifying the fluid to be tested in the marine riser to be tested based on the first prediction model and the second prediction model to obtain a fluid identification result.

[0012] According to a third aspect of the present application, an electronic device is provided, including:

[0013] at least one processor; and

[0014] a memory communicatively connected to the at least one processor; wherein,

[0015] The memory stores instructions that can be executed by the at least one processor. The instructions are executed by the at least one processor to enable the at least one processor to perform the method described in this application.

[0016] According to a fourth aspect of the present application, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to enable the computer to execute the method described in the present application.

[0017] The method of an embodiment of the present application includes obtaining liquid holdup data of at least one target location in a marine riser; preprocessing the liquid holdup data to obtain a first sample data set; the first sample data set includes liquid holdup data after frame processing; training a preset long short-term memory network model based on the first sample data set to obtain a first prediction model; the first prediction model is used to predict the liquid holdup time history curve of the fluid in the marine riser; obtaining a second prediction model; the second prediction model is obtained by training a preset convolutional neural network model using a second sample data set corresponding to the liquid holdup time history curve; the second sample data set includes liquid holdup time history curves of known manifold categories; based on the first prediction model and the second prediction model, identifying the fluid to be tested in the marine riser to be tested to obtain a fluid identification result. In this way, flow changes in the pipeline can be accurately and in real time predicted and manifold states can be identified, and potential risks can be warned in advance.

[0018] It should be understood that the teachings of this application do not necessarily achieve all of the beneficial effects described above, but that specific technical solutions can achieve specific technical effects, and other embodiments of this application can also achieve beneficial effects not mentioned above. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The above and other objects, features and advantages of the exemplary embodiments of the present application will become readily understood by reading the detailed description below with reference to the accompanying drawings. In the accompanying drawings, several embodiments of the present application are shown in an illustrative and non-limiting manner, in which:

[0020] In the drawings, the same or corresponding reference numerals denote the same or corresponding parts.

[0021] Figure 1 The schematic diagram shows the processing flow of the fluid identification method of the marine riser provided in the embodiment of the present application. Figure 1 ;

[0022] Figure 2 The schematic diagram shows the processing flow of the fluid identification method of the marine riser provided in the embodiment of the present application. Figure 2 ;

[0023] Figure 3 The schematic diagram shows the processing flow of the fluid identification method of the marine riser provided in the embodiment of the present application. Figure 3 ;

[0024] Figure 4 The schematic diagram shows the processing flow of the fluid identification method of the marine riser provided in the embodiment of the present application. Figure 4 ;

[0025] Figure 5 The application scenario of the fluid identification method of the marine riser provided in the embodiment of the present application is shown Figure 1 ;

[0026] Figure 6 The application scenario of the fluid identification method of the marine riser provided in the embodiment of the present application is shown Figure 2 ;

[0027] Figure 7 The application scenario of the fluid identification method of the marine riser provided in the embodiment of the present application is shown Figure 3 ;

[0028] Figure 8 The application scenario of the fluid identification method of the marine riser provided in the embodiment of the present application is shown Figure 4 ;

[0029] Figure 9 An optional schematic diagram of a fluid identification device for a marine riser provided in an embodiment of the present application is shown;

[0030] Figure 10 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0031] In order to make the purpose, features, and advantages of this application more obvious and easy to understand, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without making creative efforts shall fall within the scope of protection of this application.

[0032] In the following description, reference is made to “some embodiments”, which describes a subset of all possible embodiments, but it will be understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0033] In the following description, the terms "first\second" involved are merely used to distinguish similar objects and do not represent a specific ordering of the objects. It is understandable that "first\second" can be interchanged with a specific order or sequence where permitted, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein.

[0034] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application pertains. The terms used herein are for the purpose of describing the embodiments of this application only and are not intended to limit this application.

[0035] The processing flow of the fluid identification method for a marine riser provided in an embodiment of the present application is described. Figure 1 , Figure 1 This is a schematic diagram of the processing flow of the fluid identification method for the marine riser provided in the embodiment of the present application. Figure 1 , will combine Figure 1 Steps S101-S105 are shown for explanation.

[0036] Step S101: obtaining liquid holdup data of at least one target location in a marine riser.

[0037] In some embodiments, target locations may include key locations such as the riser inlet, the middle of the descending section, the arch bend, the sag point, and the middle of the hanging section. Conductivity probes installed at these target locations can collect liquid holdup data, reflecting the fluid properties at different locations within the marine riser. This holdup data may include liquid holdup values ​​at different locations and times within the marine riser. This holdup data can be used to analyze the flow state and characteristics of gas-liquid two-phase flow.

[0038] In some embodiments, step S101 may include a conductive probe based on the target position to collect the liquid holdup of the fluid in the marine riser to obtain liquid holdup data. Specifically, a two-phase flow transmission experimental platform inside the marine riser is first established. The experimental platform includes an air compressor, a submersible pump, a liquid flow meter, a gas flow meter, a tee, a valve, a water storage tank, a circulation pipeline, and a conductivity probe. Water and air enter their respective pipelines through the submersible pump and the air compressor respectively, and are merged into a horizontal pipeline through the tee. After a period of fusion, they enter the deep-sea flexible riser. After a period of time, the fluid flow in the riser stabilizes, and the liquid holdup data collection begins. Conductivity probes are used and installed at the riser inlet, the middle of the descending section, the arch bend point, the vertical bend point and the middle of the hanging section. The conductivity probe can collect the flow data of the fluid in the pipeline in real time, including the dynamic changes of the flow under different flow shapes. By changing the gas-liquid phase conversion velocity, the flow type in the pipeline is transformed. The gas phase conversion velocity range is 0.0131~0.0278m / s, and the liquid phase conversion velocity range is 0.0233~0.0311m / s. After the flow type transformation, the liquid holdup signal is recorded for 300s when bubbly flow, annular flow, slug flow, severe slug flow type I, type II, and type III appear, respectively. The sampling frequency is 1024HZ to collect the liquid holdup data at each target position.

[0039] Step S102 : pre-processing the liquid holdup data to obtain a first sample data set; the first sample data set includes the liquid holdup data after frame processing.

[0040] Step S103: Based on the first sample data set, a preset long short-term memory network model is trained to obtain a first prediction model; the first prediction model is used to predict the liquid holdup time history curve of the fluid in the marine riser.

[0041] In some embodiments, the first sample dataset may include preprocessed liquid holdup data, including frame-processed liquid holdup data. The first sample dataset may be used to train a long short-term memory (LSTM) network model to predict the time-dependent liquid holdup curve of the fluid in the marine riser. The LSTM network model may include an input layer, a hidden layer, and an output layer. The LSTM network model may be used to capture temporal dependencies in time series data and predict the time-dependent liquid holdup curve of the fluid in the marine riser. The first prediction model may include a trained LSTM network model.

[0042] Step S104, obtaining a second prediction model; the second prediction model is obtained by training a preset convolutional neural network model using a second sample data set corresponding to the liquid holdup time history curve; the second sample data set includes liquid holdup time history curves of known manifold categories.

[0043] In some embodiments, the second sample dataset may include liquid holdup history curve data with known manifold types. The second sample dataset may be used to train a convolutional neural network model to identify the manifold type of the fluid in the marine riser. The convolutional neural network model may include a convolution layer, a pooling layer, a flattening layer, and a fully connected layer. The convolutional neural network model may be used to extract local and global features from the liquid holdup history curve to identify the manifold type of the fluid in the marine riser. The second prediction model may include a trained convolutional neural network model.

[0044] Step S105 : Identify the fluid to be tested in the marine riser to be tested based on the first prediction model and the second prediction model to obtain a fluid identification result.

[0045] In some embodiments, the fluid identification results may include the flow shape classification of the fluid within the marine riser based on the first and second prediction models. The fluid identification results can be used to analyze the fluid state within the marine riser and formulate appropriate decision recommendations based on actual process requirements and safety standards.

[0046] In some embodiments, step S105 may include: based on the first prediction model, performing flow prediction on the fluid to be measured in the marine riser to be measured to obtain a liquid holdup time history curve of the fluid to be measured; based on the second prediction model, performing manifold identification on the liquid holdup time history curve of the fluid to be measured to obtain a manifold category of the fluid to be measured; and determining a fluid identification result based on the liquid holdup time history curve and the manifold category.

[0047] As an example, the output results of the first prediction model and the second prediction model are integrated. The fluid identification result is determined based on the predicted liquid holdup time history curve and the identified flow type. The fluid identification result can be combined with the actual process requirements and safety standards of offshore oil production to generate corresponding decision suggestions. For example, when the fluid identification result is that the fluid is identified as a slug flow and the predicted flow fluctuation is large, the corresponding decision suggestion generated is to recommend adjusting the operating parameters of the production equipment, such as reducing the oil production speed, to reduce the impact of the slug flow on the equipment. When the fluid identification result is that the fluid is identified as an annular flow and the flow is stable in an appropriate range, the corresponding decision suggestion generated is to maintain the current production operation state.

[0048] The method of the embodiment of the present application effectively captures the complex characteristics of gas-liquid two-phase flow by acquiring liquid holdup data from at least one target location within a marine riser and preprocessing this data to construct a first sample dataset. A first prediction model, obtained by training the preprocessed data using a long short-term memory network (LSTM), can accurately predict the time history of liquid holdup, thereby providing highly accurate results for flow prediction. Simultaneously, a second prediction model, obtained by training the time history of liquid holdup for known flow pattern categories using a convolutional neural network (CNN), can efficiently identify the flow pattern of the fluid and enable rapid classification of complex flow states. This method combines the advantages of LSTM and CNN, not only improving the accuracy of flow prediction but also enhancing the real-time and reliability of flow pattern identification. Furthermore, by monitoring changes in flow and flow pattern in real time, it can provide early warning of potential safety risks, accurately and in real time predict flow changes within the pipeline, and identify flow pattern states.

[0049] In some embodiments, the processing flow of the fluid identification method of the marine riser is as follows: Figure 2 ,like Figure 2 As shown, preprocessing the liquid holdup data in step S102 to obtain a first sample data set may include:

[0050] Step S201 : Processing the liquid holdup data using a sliding window algorithm, and normalizing the processed data to obtain time series data.

[0051] In some embodiments, the data obtained by data processing is normalized, and the obtained time series data can be expressed by the following formula.

[0052]

[0053] Where x is the data obtained by data processing, is the minimum value of the data obtained by data processing, is the maximum value of the data obtained by data processing, The time series data are obtained after normalization.

[0054] Step S202: performing frame processing on the time series data to obtain frame-processed data.

[0055] As an example, the data after framing processing can be expressed by the following formula.

[0056]

[0057] in, Represents a four-dimensional array containing data of four target locations, each of which is a time series segment of length T. (where i=1,2,3,4) represents a time series data sequence of a target location. (where t=1,2,…,T, i=1,2,3,4) represents the liquid holdup value of target position i at time point t.

[0058] Step S203: Perform variational mode decomposition and channel attention mechanism processing on the frame-processed data to obtain feature data.

[0059] Step S204: construct a first sample data set based on the feature data.

[0060] In this embodiment, variational modal decomposition is used to decompose the frame-processed data into multiple modalities, and the channel attention mechanism is used to assign different weights to the decomposed data of the same modality.

[0061] As an example, the collected liquid holdup data is cleaned to remove noise and outliers. A sliding window algorithm is used to analyze n consecutive data points in the liquid holdup data. If the deviation between a data point and the window mean exceeds a set threshold of three standard deviations, the data point is identified as an outlier and corrected, replacing the outlier with interpolated values ​​from adjacent data points. The min-max normalization formula is used to map the data to the interval [0, 1] to generate time series data. The time series data is then framed, dividing the continuous time series data into segments of fixed length to generate the framed data. Variational mode decomposition is used to decompose the framed data into multiple modes to extract distinct features. A channel-wise attention mechanism is then used to assign different weights to data within the same mode, generating key feature data. Finally, a first sample dataset is constructed based on the feature data. Different modes can characterize the trend, periodicity, and randomness of the liquid holdup data. Different weights represent the varying importance of different modal components to the holdup data.

[0062] In some embodiments, the processing flow of the fluid identification method of the marine riser is as follows: Figure 3 , like Figure 3 As shown, in step S103, based on the first sample data set, a preset long short-term memory network model is trained to obtain a first prediction model, which may specifically include:

[0063] Step S301 : dividing a first sample data set into a first training data set and a first validation data set based on a preset first ratio.

[0064] In this embodiment, the preset first ratio may include a preset dataset division ratio, for example, 70% for training and 30% for validation. The preset first ratio may be used to divide the first sample dataset into a first training dataset and a first validation dataset. The first sample dataset may include preprocessed liquid holdup data, including frame-processed liquid holdup data.

[0065] Step S302: input the first training data set into a preset long short-term memory network model, train the preset long short-term memory network model, and obtain a trained long short-term memory network model.

[0066] In this embodiment, the network structure of the long short-term memory network model includes an input layer, a hidden layer, and an output layer. The input layer converts the preprocessed one-dimensional traffic data into a three-dimensional tensor suitable for LSTM input, with the dimensions (number of samples, time step, number of features), where the time step is the frame length of the framed data and the number of features is 1. The hidden layer is composed of multiple stacked LSTM units. The LSTM unit calculation process is as follows:

[0067] Input Gate :

[0068] Forget Gate :

[0069] Output Gate :

[0070] Candidate memory units :

[0071] memory unit :

[0072] Hidden State :

[0073] in, is the sigmoid activation function, tanh is the hyperbolic tangent activation function, ⊙ is the element-by-element multiplication, w is the weight matrix, is the weight matrix in the input gate that weights the input at the current moment, is the weight matrix in the input gate that weights the hidden state at the previous moment, is the weight matrix for weighting the current moment input in the forget gate, is the weight matrix for weighting the hidden state of the previous moment in the forget gate, is the weight matrix in the output gate that weights the input at the current moment, is the weight matrix in the output gate that weights the hidden state at the previous moment, is the weight matrix for weighting the input at the current moment in the candidate memory unit, is the weight matrix that weights the hidden state of the previous moment in the candidate memory unit, b is the bias vector, is the bias vector in the input gate corresponding to the current input, is the bias vector in the input gate corresponding to the hidden state at the previous moment, is the bias vector corresponding to the current input in the forget gate, is the bias vector in the forget gate corresponding to the hidden state at the previous moment, is the bias vector corresponding to the current input in the output gate, is the bias vector in the output gate corresponding to the hidden state at the previous moment, is the bias vector corresponding to the current input in the candidate memory unit, is the bias vector corresponding to the hidden state of the previous moment in the candidate memory unit, x t Input for the current moment, h t-1 is the hidden state at the last moment, C t-1 The memory unit at the previous moment. Multiple LSTM units are connected sequentially, with the hidden state of the previous unit serving as the input for the next unit. The output layer maps the hidden state of the last LSTM unit to the predicted traffic value through a fully connected layer, using a linear activation function.

[0074] In this embodiment, the training data in the first training data set is input into the long short-term memory network model, and the loss function (mean square error) is calculated by the back propagation algorithm (based on the chain rule). , N is the sample size, is the predicted value, is the true value) about the gradient of the model parameters and updates the model parameters to achieve the prediction of traffic in the future.

[0075] Step S303: Determine a first prediction result of the trained long short-term memory network model based on the first verification data set.

[0076] In this embodiment, the first validation dataset is input into the trained model to obtain a first prediction result, i.e., predicted liquid holdup data. The predicted liquid holdup data is denormalized, and each mode is reconstructed and summed according to its weights to obtain the first prediction result, i.e., the predicted liquid holdup time history curve.

[0077] Step S304: In response to the first prediction result satisfying a preset prediction condition, the trained long short-term memory network model is determined as the first prediction model.

[0078] In some embodiments, step S304 may include: determining the coefficient of determination, root mean square error, and mean absolute error corresponding to the first prediction result; performing a goodness of fit evaluation on the first prediction result based on the coefficient of determination, root mean square error, and mean absolute error to obtain an evaluation result; in response to the evaluation result meeting a preset prediction condition, determining the trained long short-term memory network model as the first prediction model.

[0079] In this embodiment, the first prediction result may include a predicted liquid holdup time history curve. The coefficient of determination (R²) can be used to evaluate the goodness of fit between the predicted result and the actual value. The closer the value is to 1, the better the model fit effect. The root mean square error (RMSE) can be used to measure the accuracy of the model prediction. The smaller the value, the closer the predicted result is to the actual value. The mean absolute error (MAE) can be used to measure the average error of the model prediction. The smaller the value, the more accurate the prediction result. The goodness of fit evaluation may include a process of comprehensively evaluating the model prediction results using indicators such as the coefficient of determination, root mean square error, and mean absolute error. The goodness of fit evaluation may be used to determine the degree of fit of the model to the data and the prediction performance. The preset prediction conditions may include model performance indicator thresholds set according to actual application requirements. The first prediction model can be used to predict the liquid holdup time history curve of the fluid in the marine riser.

[0080] As an example, the goodness of fit of the first prediction result is evaluated based on the coefficient of determination, the root mean square error, and the mean absolute error, and the evaluation result can be expressed by the following formula.

[0081]

[0082] Where m represents the number of samples, Indicates the actual value, Represents the predicted value.

[0083] In some embodiments, the processing flow of the fluid identification method of the marine riser is as follows: Figure 4 , like Figure 4 As shown, in step S104, the preset convolutional neural network model is trained using the second sample data set corresponding to the liquid holdup time history curve, which may specifically include:

[0084] Step S401 : dividing the two sample data sets into a second training data set and a second validation data set based on a preset second ratio.

[0085] Step S402: input the second training data set into the preset convolutional neural network model, train the preset convolutional neural network model, and obtain a trained convolutional neural network model.

[0086] In this embodiment, the convolutional neural network architecture is as follows:

[0087] The first layer is a convolutional layer. The input shape is (2048, 1), 32 convolution kernels of size 16 are used, the stride is 1, the padding is Same, batch normalization is not used, the activation function is Leaky ReLU, and the output shape is (2048, 32).

[0088] Second layer: Convolutional layer. Input shape (2048, 32), 32 convolution kernels of size 16, stride 1, same padding, batch normalization, activation function Leaky ReLU, output (2048, 32).

[0089] Layer 3: Pooling layer. Input: (2048, 32), pooling kernel size: 4, stride: 4, valid padding, no batch normalization or activation function, output: (512, 32).

[0090] Layer 4: Convolutional layer. Input: (512, 32), 64 kernels of size 16, stride 1, same padding, no batch normalization, leaky ReLU activation, output: (512, 64).

[0091] Layer 5: Convolutional layer. Input: (512, 64), 64 kernels of size 16, stride 1, same padding, batch normalization, leaky ReLU activation, output: (512, 64).

[0092] Layer 6: Pooling layer. Input: (512, 64), pooling kernel size: 4, stride: 4, valid padding, no batch normalization or activation function, output: (128, 64).

[0093] Layer 7: Convolutional layer. Input: (128, 64), 128 kernels of size 16, stride 1, same padding, no batch normalization, leaky ReLU activation, output: (128, 128).

[0094] Layer 8: Convolutional layer. Input: (128, 128), 128 kernels of size 16, stride 1, same padding, batch normalization, leaky ReLU activation, output: (128, 128).

[0095] Layer 9: Pooling layer. Input: (128, 128), pooling kernel size: 4, stride: 4, valid padding, no batch normalization or activation function, output: (32, 128).

[0096] Layer 10: Flatten layer. Flattens the input (32, 128) into a one-dimensional vector with an output shape of (4096).

[0097] Layer 11: Fully connected layer. Input (4096), output (512), no batch normalization, activation function LeakyReLU.

[0098] Layer 12: Fully connected layer. Input (512), output (128), no batch normalization, activation function LeakyReLU.

[0099] Layer 13: Softmax layer. Input (128), output (5), used for multi-classification tasks.

[0100] Step S403: Determine the classification accuracy of the trained convolutional neural network model based on the second validation data set.

[0101] Step S404: In response to the classification accuracy meeting the preset classification condition, the trained convolutional neural network model is determined as the second prediction model.

[0102] In this embodiment, the preset second ratio may include a predetermined dataset division ratio, such as 80% for training and 20% for validation. The preset second ratio may be used to divide the second sample dataset into a second training dataset and a second validation dataset. The second sample dataset may include a dataset containing holdup history curves for known flowform categories. The classification accuracy may include the proportion of classification results obtained by the trained convolutional neural network model on the second validation dataset that are consistent with the actual category. The preset classification conditions may include a classification accuracy threshold set based on actual application requirements, such as requiring a classification accuracy greater than 90%. The second prediction model may be used to identify the flowform category of the fluid within the marine riser.

[0103] refer to Figure 5, the application scenario of the fluid identification method of the marine riser provided in the embodiment of the present application Figure 1 , applied to the two-phase flow circulation pipeline experiment of deep-sea gently corrugated riser.

[0104] The deep-sea gentle corrugated riser may include: a water tank, a water pump, an air pump, valves, a gas flow meter, a liquid flow meter, a tee, a conductivity probe, and a circulation pipeline.

[0105] During the experiment, the water pump and air pump were first turned on, pumping water from the water tank and air from the air pump into their respective pipelines. The water and air flowed through a liquid flowmeter and a gas flowmeter, respectively, and flow data was measured and recorded. The gas and liquid then merged at the tee and entered the horizontal section of the pipeline, where they fully developed to form a stable gas-liquid two-phase flow. The gas-liquid two-phase flow then entered the gently wavy riser. Conductivity probes (No. 1-5) installed at different locations on the riser (touchdown point 1, mid-descendant section 2, arch section 3, vertical bend section 4, and mid-overhang section 5) measured the liquid holdup in real time and recorded a time-dependent liquid holdup curve. By varying the gas-liquid phase conversion velocity, the flow pattern within the pipeline was altered. Liquid holdup signals were recorded for 300 seconds at a sampling frequency of 1024 Hz for each of the following flow patterns: bubbly flow, annular flow, slugging flow, severe slugging flow type I, type II, and type III. This ensured the acquisition of good data samples for subsequent analysis and model training.

[0106] refer to Figure 6 , the application scenario of the fluid identification method of the marine riser provided in the embodiment of the present application Figure 2 , the network structure applied to the long short-term memory network model.

[0107] The structure of the Long Short-Term Memory (LSTM) network consists of multiple LSTM units. Each unit contains a memory unit ( ), a hidden state ( h k ), as well as input gate, forget gate and output gate, used to control the flow of information. Data input sequence x t Enter each unit in turn, and the hidden state of the previous unit is used as the input of the next unit. x t-1 、 x t 、 x k+1 Represent the input data at different times, LSTMCell t−1 、 LSTMCell t 、 LSTMCell t+1 : represent different LSTM units respectively, C t-1 、C t 、 C t+1 Represent different memory units, h k-1 、h k 、h k+1 Represent the hidden states at different times, 、 、 They represent the predicted values ​​at different times.

[0108] refer to Figure 7 , the application scenario of the fluid identification method of the marine riser provided in the embodiment of the present application Figure 3 ,,applied to the intra-unit computation process of long short-term memory network model.

[0109] The calculation process within the unit of the LSTM model is as follows:

[0110] Input Gate :

[0111] Forget Gate :

[0112] Output Gate :

[0113] Candidate memory units :

[0114] memory unit :

[0115] Hidden State :

[0116] in, is the sigmoid activation function, tanh is the hyperbolic tangent activation function, ⊙ is the element-by-element multiplication, w is the weight matrix, b is the bias vector, x t Input for the current moment, h t-1 is the hidden state at the last moment, C t-1 It is the memory unit of the previous moment.

[0117] refer to Figure 8 , the application scenario of the fluid identification method of the marine riser provided in the embodiment of the present application Figure 4 , applied to the long short-term memory network model to predict liquid holdup data.

[0118] Get the original data;

[0119] Data cleaning and normalization can include: preprocessing the raw data, including removing noise and outliers, and performing normalization operations.

[0120] Variational modal decomposition may include: performing variational modal decomposition on preprocessed data to extract multiple modal features of different measurement points.

[0121] Channel attention can include: applying a channel attention mechanism to the decomposed modal data to enhance key features.

[0122] Inputting the LSTM network model may include: inputting the modal data processed by the channel attention into a long short-term memory (LSTM) network for training, and processing the modal data of each measurement point separately.

[0123] The modal reconstruction summation may include: performing modal reconstruction summation on each modal prediction result output by the LSTM network to obtain reconstructed data.

[0124] Denormalization may include: performing denormalization processing on the reconstructed data to restore the dimension of the original data to obtain predicted liquid holdup data.

[0125] Next, the exemplary structure of the software modules included in the fluid identification device 90 for the marine riser provided in the embodiment of the present application will be described. In some embodiments, for example, Figure 9 As shown, the fluid identification device 90 of the marine riser may include:

[0126] A first acquisition module 901 is configured to acquire liquid holdup data of at least one target location in a marine riser;

[0127] A preprocessing module 902 is configured to preprocess the liquid holdup data to obtain a first sample data set; the first sample data set includes the liquid holdup data after frame processing;

[0128] A training module 903 is configured to train a preset long short-term memory network model based on the first sample data set to obtain a first prediction model; the first prediction model is used to predict a liquid holdup time history curve of the fluid in the marine riser;

[0129] A second acquisition module 904 is configured to acquire a second prediction model; the second prediction model is obtained by training a preset convolutional neural network model using a second sample data set corresponding to the liquid holdup time history curve; the second sample data set includes liquid holdup time history curves of known manifold categories;

[0130] The identification module 905 is configured to identify the fluid to be measured in the marine riser to be measured based on the first prediction model and the second prediction model to obtain a fluid identification result.

[0131] In some embodiments, the target location includes at least one of the following: a riser inlet, a middle of a descending section, a dome point, a vertical bend point, and a middle of a hanging section; the first acquisition module 901 can be used to collect the liquid holdup of the fluid in the marine riser based on a conductive probe at the target location to obtain the liquid holdup data.

[0132] In some embodiments, the preprocessing module 902 can be used to: process the liquid holdup data through a sliding window algorithm, and normalize the data obtained from the data processing to obtain time series data; frame the time series data to obtain frame-processed data; perform variational mode decomposition and channel attention mechanism processing on the frame-processed data to obtain feature data; the variational mode decomposition is used to decompose the frame-processed data into multiple modes; the channel attention mechanism processing is used to assign different weights to the data of the same mode obtained by decomposition; and construct the first sample data set based on the feature data.

[0133] In some embodiments, the training module 903 can be used to: divide the first sample data set into a first training data set and a first verification data set based on a preset first ratio; input the first training data set into a preset long short-term memory network model, train the preset long short-term memory network model, and obtain a trained long short-term memory network model; determine a first prediction result of the trained long short-term memory network model based on the first verification data set; in response to the first prediction result satisfying a preset prediction condition, determine the trained long short-term memory network model as the first prediction model.

[0134] In some embodiments, the training module 903 can be used to: determine the coefficient of determination, root mean square error and mean absolute error corresponding to the first prediction result; perform a goodness of fit evaluation on the first prediction result based on the coefficient of determination, root mean square error and mean absolute error to obtain an evaluation result; in response to the evaluation result meeting the preset prediction conditions, determine the trained long short-term memory network model as the first prediction model.

[0135] In some embodiments, the fluid identification device 90 of the marine riser may further include a second training module, which may be used to: divide the two sample data sets into a second training data set and a second verification data set based on a preset second ratio; input the second training data set into a preset convolutional neural network model, train the preset convolutional neural network model to obtain a trained convolutional neural network model; determine the classification accuracy of the trained convolutional neural network model based on the second verification data set; and determine the trained convolutional neural network model as the second prediction model in response to the classification accuracy satisfying a preset classification condition.

[0136] In some embodiments, the identification module 905 can be used to: predict the flow rate of the fluid to be measured in the marine riser to be measured based on the first prediction model to obtain a liquid holdup time history curve of the fluid to be measured; perform manifold identification on the liquid holdup time history curve of the fluid to be measured based on the second prediction model to obtain a manifold category of the fluid to be measured; and determine the fluid identification result based on the liquid holdup time history curve and the manifold category.

[0137] It should be noted that the description of the device in the embodiment of the present application is similar to the description of the method embodiment above, and has similar beneficial effects as the method embodiment, so it will not be repeated here. Figures 1 to 8 The present invention shall be understood by reference to the description of any of the accompanying drawings.

[0138] According to an embodiment of the present application, the present application also provides an electronic device and a non-transitory computer-readable storage medium.

[0139] Figure 10 A schematic block diagram of an example electronic device 800 that can be used to implement an embodiment of the present application is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present application described and / or claimed herein.

[0140] like Figure 10As shown, electronic device 800 includes a computing unit 801, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 802 or a computer program loaded from a storage unit 808 into a random access memory (RAM) 803. RAM 803 may also store various programs and data required for the operation of electronic device 800. Computing unit 801, ROM 802, and RAM 803 are interconnected via a bus 804. An input / output (I / O) interface 805 is also connected to bus 804.

[0141] Multiple components in the electronic device 800 are connected to the I / O interface 805, including an input unit 806, such as a keyboard, a mouse, etc.; an output unit 807, such as various types of displays, speakers, etc.; a storage unit 808, such as a magnetic disk, an optical disk, etc.; and a communication unit 809, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 809 allows the electronic device 800 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0142] The computing unit 801 can be any general-purpose and / or specialized processing component with processing and computing capabilities. Some examples of the computing unit 801 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 801 performs the various methods and processes described above, such as the method for identifying fluids in a marine riser. For example, in some embodiments, the method for identifying fluids in a marine riser can be implemented as a computer software program tangibly embodied in a machine-readable medium, such as the storage unit 808. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 800 via the ROM 802 and / or the communication unit 809. When the computer program is loaded into the RAM 803 and executed by the computing unit 801, one or more steps of the method for identifying fluids in a marine riser described above can be performed. Alternatively, in other embodiments, the computing unit 801 may be configured to execute the fluid identification method for the marine riser in any other appropriate manner (eg, by means of firmware).

[0143] Various embodiments of the systems and techniques described above can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chip systems (SOCs), programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0144] The program code for implementing the methods of the present application can be written in any combination of one or more programming languages. Such program code can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the program code is executed by the processor or controller, the functions / operations specified in the flow charts and / or block diagrams are implemented. The program code can be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0145] In the context of this application, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or apparatus. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of machine-readable storage media may include an electrical connection based on one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), optical fibers, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0146] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0147] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), and the Internet.

[0148] A computer system may include a client and a server. The client and server are generally remote from each other and typically interact through a communication network. The client-server relationship arises through computer programs running on the respective computers and having a client-server relationship with each other. The server may be a cloud server, a server in a distributed system, or a server integrated with a blockchain.

[0149] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this application can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this application can be achieved. This is not a limitation herein.

[0150] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. A method for identifying fluid in a marine riser, characterized in that: The method comprises: Acquiring liquid holdup data at at least one target location within the marine riser; Preprocessing the liquid holdup data to obtain a first sample data set; the first sample data set includes the liquid holdup data after frame processing; Based on the first sample data set, a preset long short-term memory network model is trained to obtain a first prediction model; the first prediction model is used to predict a liquid holdup time history curve of the fluid in the marine riser; Obtaining a second prediction model; the second prediction model is obtained by training a preset convolutional neural network model using a second sample data set corresponding to the liquid holdup time history curve; the second sample data set includes liquid holdup time history curves of known manifold categories; Identifying the fluid to be measured in the marine riser to be measured based on the first prediction model and the second prediction model to obtain a fluid identification result; The identifying the fluid to be measured in the marine riser to be measured based on the first prediction model and the second prediction model to obtain a fluid identification result includes: predicting the flow rate of the fluid to be measured in the marine riser to be measured based on the first prediction model to obtain a liquid holdup time history curve of the fluid to be measured; performing manifold identification on the liquid holdup time history curve of the fluid to be measured based on the second prediction model to obtain a manifold category of the fluid to be measured; and determining the fluid identification result based on the liquid holdup time history curve and the manifold category.

2. The method according to claim 1, characterized in that The obtaining of liquid holdup data of at least one target location in the marine riser comprises: The target position includes at least one of the following: the riser entrance, the middle of the descending section, the arch bend point, the vertical bend point and the middle of the hanging section; The liquid holdup of the fluid in the marine riser is collected based on the conductive probe at the target position to obtain the liquid holdup data.

3. The method according to claim 1, characterized in that The preprocessing of the liquid holdup data to obtain a first sample data set includes: Processing the liquid holdup data using a sliding window algorithm, and normalizing the data obtained from the data processing to obtain time series data; Performing frame processing on the time series data to obtain frame-processed data; The data after the frame processing is subjected to variational mode decomposition and channel attention mechanism processing to obtain feature data; the variational mode decomposition is used to decompose the data after the frame processing into multiple modes; the channel attention mechanism processing is used to assign different weights to the data of the same mode obtained by decomposition; The first sample data set is constructed based on the feature data.

4. The method according to claim 1, wherein The step of training a preset long short-term memory network model based on the first sample data set to obtain a first prediction model includes: Based on a preset first ratio, dividing the first sample data set into a first training data set and a first validation data set; Inputting the first training data set into a preset long short-term memory network model, training the preset long short-term memory network model, and obtaining a trained long short-term memory network model; Determining a first prediction result of the trained long short-term memory network model based on the first validation data set; In response to the first prediction result satisfying a preset prediction condition, the trained long short-term memory network model is determined as the first prediction model.

5. The method according to claim 4, characterized in that In response to the first prediction result satisfying a preset prediction condition, determining the trained long short-term memory network model as the first prediction model includes: Determining the coefficient of determination, root mean square error, and mean absolute error corresponding to the first prediction result; Performing a goodness of fit evaluation on the first prediction result based on the coefficient of determination, the root mean square error, and the mean absolute error to obtain an evaluation result; In response to the evaluation result satisfying a preset prediction condition, the trained long short-term memory network model is determined as the first prediction model.

6. The method according to claim 1, characterized in that The training of a preset convolutional neural network model using a second sample data set corresponding to the liquid holdup time history curve includes: Based on a preset second ratio, dividing the two sample data sets into a second training data set and a second validation data set; Inputting the second training data set into a preset convolutional neural network model, training the preset convolutional neural network model to obtain a trained convolutional neural network model; Determining the classification accuracy of the trained convolutional neural network model based on the second validation data set; In response to the classification accuracy meeting a preset classification condition, the trained convolutional neural network model is determined as the second prediction model.

7. A fluid identification device for a marine riser, characterized in that: include: A first acquisition module is configured to acquire liquid holdup data of at least one target location in the marine riser; a preprocessing module, configured to preprocess the liquid holdup data to obtain a first sample data set; the first sample data set includes the liquid holdup data after frame processing; A training module, configured to train a preset long short-term memory network model based on the first sample data set to obtain a first prediction model; the first prediction model is used to predict a liquid holdup time history curve of the fluid in the marine riser; a second acquisition module for acquiring a second prediction model; the second prediction model is obtained by training a preset convolutional neural network model using a second sample data set corresponding to the liquid holdup time history curve; the second sample data set includes liquid holdup time history curves of known manifold categories; an identification module, configured to identify the fluid to be tested in the marine riser to be tested based on the first prediction model and the second prediction model, and obtain a fluid identification result; The identifying the fluid to be measured in the marine riser to be measured based on the first prediction model and the second prediction model to obtain a fluid identification result includes: predicting the flow rate of the fluid to be measured in the marine riser to be measured based on the first prediction model to obtain a liquid holdup time history curve of the fluid to be measured; performing manifold identification on the liquid holdup time history curve of the fluid to be measured based on the second prediction model to obtain a manifold category of the fluid to be measured; and determining the fluid identification result based on the liquid holdup time history curve and the manifold category.

8. An electronic device, characterized in that: include: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 6.

9. A non-transitory computer-readable storage medium storing computer instructions, characterized in that: The computer instructions are used to enable a computer to execute the method according to any one of claims 1 to 6.

Images