Safety early warning methods and related devices for harmful flow patterns in oil and gas gathering and transmission riser systems
By combining LSTM and EMD technologies, an adaptive early warning model was established, which solved the problem that existing harmful flow pattern early warning methods require a large amount of training data. This enabled real-time early warning without initial training, ensuring the safe production of oil and gas fields.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- XI AN JIAOTONG UNIV
- Filing Date
- 2023-09-21
- Publication Date
- 2026-07-03
AI Technical Summary
Existing methods for early warning of harmful flow patterns require extensive initial training, and it is difficult to determine the parameter correspondence between laboratory models and field extrapolation through theoretical analysis, leading to difficulties in field application.
By employing an LSTM neural network model combined with EMD decomposition technology, an adaptive early warning model is established through preprocessing and feature extraction of the differential pressure signal of the riser system, enabling real-time prediction and alarm of harmful flow patterns.
It can provide early warning of harmful flow patterns without requiring a large amount of initial training data, reduce the workload of on-site commissioning, ensure normal production in oil and gas fields, and provide redundant time for risk assessment.
Smart Images
Figure CN117332278B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of oil and gas field safety production assurance technology, and relates to a method and related device for early warning of harmful flow patterns in oil and gas gathering and transmission riser systems. Background Technology
[0002] Offshore oil and gas gathering and transportation systems are the main facilities for collecting, processing, and storing extracted crude oil, water, and natural gas, before transporting them by ship via single-point mooring or through subsea pipelines. Compared to separating oil and gas before gathering and transporting them through separate oil and water pipelines, gas-liquid mixed transportation can greatly simplify engineering construction, reduce workload, shorten construction period, and lower project investment. This advantage is even more pronounced in more challenging and deep-water oilfields. However, due to the influence of seabed topography, subsea pipelines often have uneven structures. In the early and later stages of oilfield development, when gas and liquid flow rates are relatively low, the stratified flow of gas and liquid within the pipe can form long liquid plugs. If a downward-sloping pipeline exists, the liquid phase accumulates at the bottom of the riser, forming liquid plugs that can be one or even several times the length of the riser's height, resulting in severe slugging flow. Severe slugging flow not only causes significant fluctuations in pipe pressure and gas-liquid flow rates but also causes liquid to be ejected from the riser at high speed in a short period. This gas-liquid ejection process leads to sudden changes in flow rates at the wellhead or riser outlet, resulting in drastic changes in pipeline flow parameters. Significant pressure fluctuations cause pipeline vibrations, which can exacerbate pipe wall erosion and corrosion. On-site measures are typically implemented to control unstable flow patterns, but these measures are usually in standby mode. If the formation of these harmful flow patterns can be predicted in advance, allowing flow control facilities to respond automatically, adverse operating conditions during production can be effectively avoided, ensuring the safety of subsea pipeline flow.
[0003] Existing early warning technologies for hazardous flow patterns require extensive initial training before application. Training the identification model necessitates obtaining raw data from experiments within a specific gas-liquid flow range. However, due to the complexity of flow and the diversity of pipeline routes, it is difficult to determine the parameter correspondences when extrapolating laboratory models to the field through theoretical analysis. Therefore, an adaptive safety early warning method for hazardous flow patterns in oil and gas gathering and transmission riser systems is needed. To overcome this difficulty, there is an urgent need to develop an adaptive safety early warning method that uses a smaller training database or even eliminates the need for initial experiments to obtain a training set. Summary of the Invention
[0004] The purpose of this invention is to solve the technical problems in the existing technology of early warning technology for harmful flow patterns, which requires a lot of initial training, the training of the identification model requires obtaining raw data from experiments within a certain range of gas and liquid flow rates, and it is difficult to determine the parameter correspondence when extrapolating the laboratory model to the field through theoretical analysis. The invention provides a safety early warning method and related device for harmful flow patterns in oil and gas gathering and transmission riser systems.
[0005] To achieve the above objectives, the present invention employs the following technical solution:
[0006] In a first aspect, the present invention provides a method for safety early warning of harmful flow patterns in an oil and gas gathering and transmission riser system, comprising the following steps:
[0007] S1: The collected horizontal loop differential pressure signals and riser differential pressure signals in the gathering and transmission-riser system are preprocessed to obtain the historical sequence of the average horizontal loop differential pressure, the historical sequence of the average riser differential pressure, and the historical sequence of the riser differential pressure amplitude.
[0008] S2: The riser pressure difference mean sequence is selected and the first prediction model is established based on LSTM to predict the riser pressure difference mean prediction sequence;
[0009] S3: Based on the riser differential pressure average prediction sequence, riser differential pressure historical sequence, and riser differential pressure real-time sequence, extract the amplitude characteristic values of the three for comparison as the first discrimination condition; at the same time, combine the horizontal loop differential pressure average historical sequence threshold as the second discrimination condition, and jointly determine the flow state in the gathering and transport-riser system.
[0010] S4: If the flow state is transient, preprocess the transient data and combine it with the historical sequence of riser differential pressure amplitude obtained in S1 to form a new riser differential pressure amplitude sequence, and then establish a second prediction model based on LSTM.
[0011] S5: Based on the second prediction model, a new riser pressure differential amplitude prediction sequence is obtained, and it is determined whether a harmful flow pattern will appear. If a harmful flow pattern appears, an alarm mechanism is triggered.
[0012] Secondly, the present invention provides a hazardous flow pattern safety early warning system for oil and gas gathering and transmission riser systems, comprising:
[0013] The preprocessing module is used to preprocess the collected horizontal loop differential pressure signal and riser differential pressure signal in the collection and transmission-riser system to obtain the historical sequence of the average horizontal loop differential pressure, the historical sequence of the average riser differential pressure, and the historical sequence of the riser differential pressure amplitude.
[0014] The first prediction model construction module is used to select the riser pressure difference mean sequence, build the first prediction model based on LSTM, and predict the riser pressure difference mean prediction sequence.
[0015] The flow state determination module is used to extract the amplitude feature values of the riser differential pressure average prediction sequence, riser differential pressure historical sequence, and riser differential pressure real-time sequence as the first discrimination condition; at the same time, it combines the horizontal loop differential pressure average historical sequence threshold as the second discrimination condition, and together determine the flow state in the gathering and transport-riser system.
[0016] The second prediction model construction module is used to preprocess the transient data if the flow state is transient, and combine it with the historical sequence of riser differential pressure amplitude obtained in S1 to form a new riser differential pressure amplitude sequence, and then establish the second prediction model based on LSTM.
[0017] The harmful flow pattern early warning module is used to obtain a new riser pressure differential amplitude prediction sequence based on the second prediction model, and to determine whether a harmful flow pattern will appear. If a harmful flow pattern appears, an alarm mechanism is triggered.
[0018] Thirdly, the present invention provides a computer device including a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the steps of the method described above.
[0019] Fourthly, the present invention provides a computer-readable storage medium storing a computer program, characterized in that the computer program, when executed by a processor, implements the steps of the method described above.
[0020] Compared with the prior art, the present invention has the following beneficial effects:
[0021] This invention discloses a method and related device for early warning of harmful flow patterns in oil and gas gathering and transportation riser systems. It utilizes a sliding window approach to calculate the mean sequence, and obtains predicted values for future times under steady-state conditions using an EMD-LSTM combined prediction model. Adaptive features and thresholds are extracted from steady-state data and future data under different operating conditions. Based on the "steady-state" assumption, the predicted values are compared with measured values to automatically determine transient flows within the system. The method predicts the changing trend of riser differential pressure amplitude during transient processes, continuously judging whether harmful flow patterns occur, thereby triggering an early warning mechanism. This adaptive safety early warning method shields the differences between different experimental systems. The early warning model is trained online in real-time, eliminating the need for pre-training through experiments. This significantly reduces the workload of on-site debugging and can be transferred to the field without affecting normal oil and gas field production. Furthermore, the adaptive early warning process includes two stages: transient discrimination and harmful flow pattern safety early warning. The transient process provides a certain amount of redundancy time for secondary judgment of potential risks. Moreover, this method can directly use the original signal or use the components obtained after EMD decomposition of the original signal. The former is faster, while the latter is more accurate; the choice can be made flexibly based on the actual situation on-site. Attached Figure Description
[0022] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.
[0023] Figure 1 This is a flowchart of the method of the present invention;
[0024] Figure 2 This is a schematic diagram of the system of the present invention;
[0025] Figure 3 This is a schematic diagram of the signal acquisition location for this invention;
[0026] Figure 4 This is a flowchart of the LSTM prediction modeling process of the present invention;
[0027] Figure 5 This is an overall flowchart of the harmful flow pattern safety early warning method of the present invention;
[0028] Figure 6 This is the result of the first prediction model in this embodiment of the invention;
[0029] Figure 7 This is the result of the second prediction model in the embodiment of the present invention;
[0030] Figure 8 This is a schematic diagram of the computer device structure of the present invention. Detailed Implementation
[0031] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.
[0032] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.
[0033] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.
[0034] In the description of the embodiments of the present invention, it should be noted that if terms such as "upper," "lower," "horizontal," or "inner" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of the invention is in use, they are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. Furthermore, terms such as "first" and "second" are only used to distinguish descriptions and should not be construed as indicating or implying relative importance.
[0035] Furthermore, the use of the term "horizontal" does not imply that the component must be absolutely horizontal, but rather that it can be slightly tilted. For example, "horizontal" simply means that its direction is more horizontal than "vertical," and does not mean that the structure must be completely horizontal, but can be slightly tilted.
[0036] In the description of the embodiments of the present invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in the present invention according to the specific circumstances.
[0037] The present invention will now be described in further detail with reference to the accompanying drawings:
[0038] See Figure 1 and Figure 5 This invention discloses a method for safety early warning of harmful flow patterns in oil and gas gathering and transportation riser systems, comprising the following steps:
[0039] S1: The collected horizontal loop differential pressure signals and riser differential pressure signals in the gathering and transmission-riser system are preprocessed to obtain the historical sequence of the average horizontal loop differential pressure, the historical sequence of the average riser differential pressure, and the historical sequence of the riser differential pressure amplitude.
[0040] S2: The riser pressure difference mean sequence is selected and the first prediction model is established based on LSTM to predict the riser pressure difference mean prediction sequence;
[0041] S3: Based on the riser differential pressure average prediction sequence, riser differential pressure historical sequence, and riser differential pressure real-time sequence, extract the amplitude characteristic values of the three for comparison as the first discrimination condition; at the same time, combine the horizontal loop differential pressure average historical sequence threshold as the second discrimination condition, and jointly determine the flow state in the gathering and transport-riser system.
[0042] S4: If the flow state is transient, preprocess the transient data and combine it with the historical sequence of riser differential pressure amplitude obtained in S1 to form a new riser differential pressure amplitude sequence, and then establish a second prediction model based on LSTM.
[0043] S5: Based on the second prediction model, a new riser pressure differential amplitude prediction sequence is obtained, and it is determined whether a harmful flow pattern will appear. If a harmful flow pattern appears, an alarm mechanism is triggered.
[0044] In one feasible embodiment of the present invention, see [link to relevant documentation]. Figure 3 S1 includes:
[0045] S101) Collect the horizontal loop differential pressure signal DP1 and the riser differential pressure signal DP2 in the collection and transmission-riseer system, and perform noise reduction processing on the collected signals based on the Savitzky-Golay filtering method of local polynomial least squares fitting.
[0046] S102) For the denoised horizontal loop differential pressure signal DP1 and riser differential pressure signal DP2, select a sampling interval of Δt, a mean window sampling point count of n, a window sliding step count of N, and truncate the signal forward for a time length of (N+n)Δt to obtain the historical sequence M of the mean horizontal loop differential pressure. DP1 Historical sequence of average riser pressure differential M DP2 Historical sequence of riser pressure differential amplitude A DP2 The specific process is as follows:
[0047]
[0048]
[0049] A DP2,i =max(DP2) ,i-n DP2 ,i-n+1 DP2 ,i-n+2 ...,DP2 ,i-1 )-min(DP2 ,i-n DP2 ,i-n+1 DP2 ,i-n+2 ...,DP2 ,i-1 )
[0050] i = n+1, n+2, ..., n+N
[0051] Among them, M DP1,i M is the mean of DP1 in the i-th mean window; DP2,i Let A be the DP2 mean of the i-th mean window; DP2,i Let be the DP2 amplitude of the i-th mean window.
[0052] In one feasible embodiment of the present invention, see [link to relevant documentation]. Figure 4 S2 includes:
[0053] S201) The historical sequence of the average riser pressure differential M DP2 The data standardization process is as follows:
[0054] M his,DP2,max =max(M DP2,n+1 M DP2,n+2 ,...,M DP2,n+N )
[0055] M his,DP2,min =min(M) DP2,n+1 M DP2,n+2 ,...,M DP2,n+N )
[0056]
[0057] Among them, M his,DP2,max M is the historical series of the average riser pressure differential. DP2 The maximum value of M; his,DP2,min M is the historical series of the average riser pressure differential. DP2 The minimum value of M; DP2,nor This is the historical data of the average riser pressure differential after standardization;
[0058] S202) The standardized average historical data of riser pressure difference is used as the input and the predicted quantity. The first prediction model is established based on LSTM, and the prediction step of the first prediction model is set to s1.
[0059] S203) Starting from window i = n + N + 1, the mean riser pressure difference prediction sequence M for future times is obtained using the first prediction model. pred,DP2 .
[0060] In one feasible embodiment of the present invention, S3 includes:
[0061] S301) Based on the historical sequence of average riser pressure differential M DP2 The amplitude characteristic value A of the historical sequence of the average riser pressure difference was calculated. his,DP2,M The specific process is as follows:
[0062] A DP2,M,i =max(M DP2,i-m M DP2,i-m+1 M DP2,i-m+2 ...,M DP2,i-1 )-min(M DP2,i-m M DP2,i-m+1 M DP2,i-m+2 ...,M DP2,i-1 )
[0063] i = n+m+1, n+m+2, ..., n+N
[0064] A his,DP2,M =max(A DP2,M,n+N-4 A DP2,M,n+N-3 A DP2,M,n+N-2 A DP2,M,n+N-1 A DP2,M,n+N )
[0065] Among them, A DP2,M,i M is the amplitude window of the i-th value. DP2 The range of the sequence; m is the number of mean values in the amplitude sliding window;
[0066] S302) Predict sequence M based on the average riser pressure difference. pred,DP2 Solve for the amplitude characteristic value A of the predicted mean value of riser pressure difference. pred,DP2,M,i :
[0067] A pred,DP2,M,i
[0068] =max(M pred,DP2,i-m M pred,DP2,i-m+1 M pred,DP2,i-m+2 ...,M pred,DP2,i-1 )-min(M pred,DP2,i-m M pred,DP2,i-m+1 M pred,DP2,i-m+2 ...,M pred,DP2,i-1 )
[0069] i = n+m+1, n+m+2, ..., n+N
[0070] S303) When the last time corresponding to the i-th (i≥n+N+1) window arrives, solve for the amplitude characteristic value A of the measured mean value of the riser pressure difference. real,DP2,M,i :
[0071] A real,DP2,M,i =max(M real,DP2,i-m M real,DP2,i-m+1 M real,DP2,i-m+2 ...,M real,DP2,i-1 )-min(M real,DP2,i-m M real,DP2,i-m+1 M real,DP2,i-m+2 ...,M real,DP2,i-1 )
[0072] S304) Solve for the maximum value M of the historical series of the horizontal loop pressure difference mean. his,DP1,max With minimum value M his,DP1,min :
[0073] M his,DP1,max =max(M DP1,n+1 M DP1,n+2 ,...,MDP1,n+N )
[0074] M his,DP1,min =min(M) DP1,n+1 M DP1,n+2 ,...,M DP1,n+N )
[0075] S305) Determine whether the flow state in the system is transient based on the following conditions;
[0076]
[0077] |A real,DP2,M,i -A pred,DP2,M,i |>|A his,DP2,M -A pred,DP2,M,i | (2)
[0078] (a) If equations (1) and (2) are not satisfied at the same time, the flow state is determined to be steady state, and the process ends when the last time corresponding to window i = n + N + s1 is reached.
[0079] (b) If both equation (1) and equation (2) are satisfied, the flow state is determined to be transient flow, and an early warning is triggered.
[0080] In a feasible embodiment of the present invention, S5 includes: setting the prediction step number of the second prediction model to s2, and predicting a new riser pressure difference amplitude prediction sequence A through the second prediction model. pred,DP2 Determine whether a harmful flow pattern exists in the system based on the following conditions;
[0081] A pred,DP2,i ′>0.2ρgh, i=n+N+1,n+N+2,...,n+N+s2 (3)
[0082] Where ρ is the fluid density inside the riser; h is the fluid height difference inside the riser; and g is the acceleration due to gravity.
[0083] (a) If equation (3) is not satisfied, it is determined that the system will not have a harmful flow pattern, and the process will end when the last time corresponding to window i = n + N + s2 is reached.
[0084] (b) If equation (3) is satisfied, it is determined that a harmful flow pattern will appear in the system, and an alarm mechanism will be triggered.
[0085] In a feasible embodiment of the present invention, the establishment of the first prediction model and the second prediction model further includes: performing empirical mode decomposition (EMD) on the standardized riser pressure difference mean data, then establishing the first prediction model or the second prediction model for each layer component and the residual component respectively, and then reconstructing the prediction results of each layer component and the residual component model through the inverse process of EMD.
[0086] In a feasible embodiment of the present invention, in S101, the sampling interval is Δt = 1 second, and the minimum setting value of the mean window sampling point number window sliding step N is 300; in S104, the mean number m of the amplitude sliding window is 40; in S3, the prediction step s1 of the first prediction model is set to 40; the value of k in equation (1) is set to 5; and in S5, the prediction step s2 of the second prediction model is set to 40.
[0087] like Figure 2 As shown, this embodiment of the invention discloses a hazardous flow pattern safety early warning system for oil and gas gathering and transportation riser systems, comprising:
[0088] The preprocessing module is used to preprocess the collected horizontal loop differential pressure signal and riser differential pressure signal in the collection and transmission-riser system to obtain the historical sequence of the average horizontal loop differential pressure, the historical sequence of the average riser differential pressure, and the historical sequence of the riser differential pressure amplitude.
[0089] The first prediction model construction module is used to select the riser pressure difference mean sequence, build the first prediction model based on LSTM, and predict the riser pressure difference mean prediction sequence.
[0090] The flow state determination module is used to extract the amplitude feature values of the riser differential pressure average prediction sequence, riser differential pressure historical sequence, and riser differential pressure real-time sequence as the first discrimination condition; at the same time, it combines the horizontal loop differential pressure average historical sequence threshold as the second discrimination condition, and together determine the flow state in the gathering and transport-riser system.
[0091] The second prediction model construction module is used to preprocess the transient data if the flow state is transient, and combine it with the historical sequence of riser differential pressure amplitude obtained in S1 to form a new riser differential pressure amplitude sequence, and then establish the second prediction model based on LSTM.
[0092] The harmful flow pattern early warning module is used to obtain a new riser pressure differential amplitude prediction sequence based on the second prediction model, and to determine whether a harmful flow pattern will appear. If a harmful flow pattern appears, an alarm mechanism is triggered.
[0093] See Figure 8 This invention discloses a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. The device is characterized in that the processor executes the computer program to implement the steps of the method for safety early warning of harmful flow patterns in the oil and gas gathering and transmission riser system.
[0094] This invention discloses a computer-readable storage medium storing a computer program, characterized in that the computer program, when executed by a processor, implements the steps of the hazardous flow pattern safety early warning method for the oil and gas gathering and transmission riser system.
[0095] See Figure 6 and Figure 7 Another feasible specific embodiment of the present invention discloses a method for safety early warning of harmful flow patterns in oil and gas gathering and transportation riser systems, comprising the following steps:
[0096] Select at least two differential pressure signals along the pipeline route of the gathering and transmission riser system, see [reference]. Figure 1 The horizontal loop differential pressure signal DP1 and the riser differential pressure signal DP2 were selected. Before training with historical data sequences, the warning signals needed to be preprocessed. The DP1 and DP2 signal data were filtered in real time during acquisition. The mean sequence M of signals DP1 and DP2 was obtained using a window sliding translation method. DP1,i and M DP2.i and amplitude sequence A DP1,i and A DP2.i .
[0097] The input to the first prediction model and the predicted M DP2.i standardization:
[0098]
[0099] After preprocessing, the early warning signal has two prediction technology routes.
[0100] Technical Route 1: First, the data M to be predicted... DP2.i The empirical mode (EMD) is decomposed into several intrinsic eigenfunctions (IMFs) and one residual component. Next, each component is trained using an LSTM neural network, and the optimal parameters are selected to obtain the predicted value for each component. Finally, the predicted values of each component are superimposed and reconstructed to obtain the final predicted value M. pred,DP2 .
[0101] Technical Route 2: Directly predict the data M DP2.i The predicted value M is obtained by training an LSTM neural network and selecting the optimal parameters. pred,DP2 .
[0102] For example: tintval=1s,twindow=40s,ttraining=300s,tpred=40s.
[0103] 1) The process of solving for the mean and range sequences
[0104] From t=41s to t=340s
[0105]
[0106] A x,i =max(x i-40 ,x i-39 ,…,x i-2 ,x i-1 )-min(x i-40 ,x i-39 ,…,x i-2 ,x i-1 ), (i=41,42,…,340; x=DP5)
[0107] 2) The process of solving the predicted values
[0108] From t=41s to t=340s,
[0109] t = 340s, using M x,41 t to M x,340 (x = DP5), used as the training set for input LSTM;
[0110] First step prediction:
[0111] Use M x,41 To M x,340 The predicted value M is obtained through the LSTM model. pred,M,341 ;
[0112] Second step prediction:
[0113] Use M x,42 To M x,340 and M pred,M,341 The predicted value M is obtained through the LSTM model. pred,M,342 ;
[0114] Third step prediction:
[0115] Use M x,43 To M x,340 M pred,M,341 and M pred,M,342 The predicted value M is obtained through the LSTM model. pred,M,343 ;
[0116] ...
[0117] Fortieth step prediction:
[0118] Use M x,80 To M x,340 M pred,M,341 and M pred,M,379 The predicted value M is obtained through the LSTM model. pred,M,380 ;
[0119] This embodiment predicts a total of 40 steps.
[0120] The adaptive safety early warning process for hazardous flow patterns includes two judgments and two warnings. The first judgment is the flow state within the system. Upper and lower thresholds are set for the horizontal loop differential pressure signal DP1. When the measured value deviates from the threshold, a transient condition is detected within the system. The riser differential pressure signal DP2 is determined by the historical value range A. his,DP2,M Predicted range A pred,DP2,M The range of the measured value A DP2,M,i The comparison determines whether a transient situation will occur within the system. If a transient situation is detected, a first warning is issued; no control measures are needed for this first warning. The second warning is for the presence of a harmful flow pattern within the system. Using the range of historical data and transient process data as the training set for an LSTM model, the trend of the range curve fluctuation at future times is predicted. When the range value exceeds the system's set warning value, a harmful flow pattern is detected, issuing a second warning and requiring necessary safety control measures. The prediction of the range curve fluctuation trend is compared with... Figure 4 The prediction process is the same.
[0121] The distinction between stable and harmful flow patterns is as follows:
[0122] Stable flow pattern: The overall pressure difference fluctuation amplitude of the riser is less than 20% of the static pressure difference within the riser;
[0123] Harmful flow pattern: The overall pressure difference fluctuation amplitude of the riser shall not be less than 20% of the static pressure difference within the riser.
[0124] 1) The process of solving the threshold of the horizontal loop differential pressure signal DP1
[0125] For example: from t=41s to t=340s
[0126] M his,DP1,max =max(M DP1,41 M DP1,42 ,…,M DP1,339 M DP1,340 )
[0127] M his,DP1,min =min(M) DP1,41 M DP1,42 ,…,M DP1,339 M DP1,340 )
[0128]
[0129] 2) The process of solving the historical range of riser differential pressure signal DP2
[0130] For example: from t=41s to t=340s, i=340
[0131] A his,DP2,i =max(M DP2,i-40 M DP2,i-39 ,…,M DP2,i-2 M DP2,i-1 )-min(M DP2,i-40 M DP2,i-39 ,…,M DP2,i-2 M DP2,i-1 )
[0132] A his,DP2,M =max(A DP2,M,i-4 A DP2,M,i-3 A DP2,M,i-2 A DP2,M,i-1 A DP2,M,i )
[0133] 3) The process of solving the range of the predicted value of riser differential pressure signal DP2
[0134] For example: from t=41s to t=340s, i=341,342,…380
[0135] A pred,DP2,M,i =max(M pred,DP2,i-40 M pred,DP2,i-39 ,…,M pred,DP2,i-2 M pred,DP2,i-1 )-min(M pred,DP2,i-40 M pred,DP2,i-39 ,…,M pred,DP2,i-2 M pred,DP2,i-1 )
[0136] 4) The process of solving the range of the measured value of the riser pressure differential signal DP2
[0137] For example: from t=41s to t=340s, i=341,342,…380
[0138] A DP2,M,i =max(M DP2,i-40 M DP2,i-39 ,…,M DP2,i-2 M DP2,i-1 )-min(M DP2,i-40 M DP2,i-39 ,…,M DP2,i-2 M DP2,i-1 )
[0139] See Figure 6 Data before 787 seconds is historical data, and data after 787 seconds is future data. The upper and lower thresholds of the DP1 signal mean sequence, calculated from the historical data, are 30.35 kPa and 32.88 kPa, respectively. In the future, the mean M... DP1,788 To M DP1,796All values are within the range of [30.35 kPa, 32.88 kPa], and the average values M for five consecutive periods from 793 s to 797 s are all within this range. DP1,793 To M DP1,797 Outside the threshold shown in the figure, the horizontal loop differential pressure signal DP1 is determined at 797s. An EMD-LSTM model is trained to predict the future values of the DP5 signal, ranging from 787s to 826s. At time 797s, A... DP2,M,797 For 0.88 kPa, A pred,DP2,M,797 The Pa is 0.64 kPa, A his,DP2,M The pressure is 0.67 kPa, and the condition is satisfied for 5 consecutive seconds |A DP2,M,i -A pred,DP2,M |>|A his,DP2,M -A pred,DP2,M The riser differential pressure signal DP2 is determined at 801s. The transient determination time of the system based on the results of the dual signals is 801s.
[0140] See Figure 7 The EMD-LSTM combined prediction model is trained every 40 seconds, with the prediction time for future extreme values also set at 40 seconds. The training length is set to 300 seconds. The training set for the first model is traced from t=801s (confirmed as a transient moment within the system) to t=301s, and the training set for the second model is traced from t=841s (including the model's first prediction) to t=341s. In this embodiment, the riser height h is 16.3m, and the liquid density ρ = 1000kg / m³. 3 According to A pred,DP2,i >0.2ρgh, the warning value is 32.14 kPa; the predicted value A at t=848s. pred,DP2,848 The pressure was 32.88 kPa, exceeding the warning threshold, and a warning was issued.
[0141] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0142] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0143] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0144] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0145] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.
Claims
1. A safety early warning method for harmful flow patterns in oil and gas gathering and transmission-riseer systems, characterized in that, Includes the following steps: S1: The collected horizontal loop differential pressure signals and riser differential pressure signals in the gathering and transmission-riser system are preprocessed to obtain the historical sequence of the average horizontal loop differential pressure, the historical sequence of the average riser differential pressure, and the historical sequence of the riser differential pressure amplitude. S2: The historical series of riser pressure difference mean is selected and the first prediction model is established based on LSTM to predict the riser pressure difference mean prediction series; S3: Based on the riser differential pressure average prediction sequence, riser differential pressure average historical sequence, and riser differential pressure average real-time sequence, extract the amplitude characteristic values of the three for comparison as the first discrimination condition; at the same time, combine the horizontal loop differential pressure average historical sequence threshold as the second discrimination condition, and jointly determine the flow state in the gathering and transport-riser system; Specifically, it includes: S301) Based on the historical sequence of average riser pressure differential M DP2 The amplitude characteristic value A of the historical sequence of the average riser pressure difference was calculated. his,DP2,M The specific process is as follows: Among them, A DP2,M,i M is the amplitude window of the i-th value. DP2 The range of the sequence; m is the number of mean values in the amplitude sliding window; n is the number of sampling points in the mean window; N is the number of sliding steps in the window; S302) Predict sequence M based on the average riser pressure difference pred,DP2 Solve for the amplitude characteristic value A of the predicted mean value of riser pressure difference. pred,DP2,M,i : S303) When the last time corresponding to the i-th (i≥n+N+1) window is reached, solve for the amplitude characteristic value A of the measured mean value of the riser pressure difference. real,DP2,M,i : S304) Solve for the maximum value M of the historical series of the horizontal loop pressure difference mean. his,DP1,max With minimum value M his,DP1,min : S305) Determine whether the flow state in the system is transient based on the following conditions; , k consecutive windows (1) （2） (a) If equations (1) and (2) are not satisfied at the same time, the flow state is determined to be steady state flow, and the process ends when the last time corresponding to window i = n+N+s1 is reached. (b) If both equation (1) and equation (2) are satisfied, the flow state is determined to be transient flow, and an early warning is triggered. S4: If the flow state is transient, preprocess the transient data and combine it with the historical sequence of riser differential pressure amplitude obtained in S1 to form a new riser differential pressure amplitude sequence, and then establish a second prediction model based on LSTM. S5: Based on the second prediction model, a new riser pressure differential amplitude prediction sequence is obtained, and it is determined whether a harmful flow pattern will appear. If a harmful flow pattern appears, an alarm mechanism is triggered.
2. The method for safety early warning of harmful flow patterns in oil and gas gathering and transmission-riseer systems according to claim 1, characterized in that, S1 includes: S101) Collect the horizontal loop differential pressure signal DP1 and the riser differential pressure signal DP2 in the collection and transmission-riseer system, and perform noise reduction processing on the collected signals based on the Savitzky-Golay filtering method of local polynomial least squares fitting. S102) For the denoised horizontal loop differential pressure signal DP1 and riser differential pressure signal DP2, select a sampling interval of Δt, a mean window sampling point count of n, a window sliding step count of N, and truncate the signal forward for a time length of (N+n)Δt to obtain the historical sequence M of the mean horizontal loop differential pressure. DP1 Historical sequence of average riser pressure differential M DP2 Historical sequence of riser pressure differential amplitude A DP2 The specific process is as follows: Among them, M DP1,i M is the mean of DP1 in the i-th mean window; DP2,i Let A be the DP2 mean of the i-th mean window; DP2,i Let be the DP2 amplitude of the i-th mean window.
3. The method for safety early warning of harmful flow patterns in oil and gas gathering and transmission-riseer systems according to claim 1, characterized in that, S2 includes: S201) The historical sequence of the average riser pressure differential M DP2 The data standardization process is as follows: Among them, M his,DP2,max M is the historical series of the average riser pressure differential. DP2 The maximum value of M; his,DP2,min M is the historical series of the average riser pressure differential. DP2 The minimum value of M; DP2,nor This is the historical data of the average riser pressure differential after standardization; S202) The standardized average historical data of riser pressure difference is used as the input and the predicted quantity. The first prediction model is established based on LSTM, and the prediction step of the first prediction model is set to s1. (S203) Starting from window i=n+N+1, the first prediction model is used to predict the mean riser pressure difference prediction sequence M for future times. pred,DP2 .
4. The method for safety early warning of harmful flow patterns in oil and gas gathering and transmission-riseer systems according to claim 1, characterized in that, S5 includes: setting the prediction step number of the second prediction model to s2, and predicting a new riser pressure difference amplitude prediction sequence A through the second prediction model. pred,DP2 ′ Determine whether a harmful flow pattern exists within the system based on the following conditions; （3） in, The fluid density inside the riser; The height difference of the fluid inside the riser; It is the acceleration due to gravity; (a) If equation (3) is not satisfied, it is determined that the system will not have a harmful flow pattern, and the process will end when the last time corresponding to window i = n+N+s2 is reached; (b) If equation (3) is satisfied, it is determined that a harmful flow pattern will appear in the system, and an alarm mechanism will be triggered.
5. The method for safety early warning of harmful flow patterns in oil and gas gathering and transmission-riseer systems according to claim 2, characterized in that, The establishment of the first prediction model and the second prediction model also includes: performing Empirical Mode Decomposition (EMD) on the standardized riser pressure difference mean data, then establishing the first prediction model or the second prediction model for each layer component and the residual component respectively, and then reconstructing the prediction results of each layer component and the residual component model through the inverse process of EMD.
6. The method for safety early warning of harmful flow patterns in oil and gas gathering and transmission-riseer systems according to claim 5, characterized in that, In S102, the sampling interval is Δt = 1 second, and the minimum setting value of the number of sampling points and the number of sliding steps N of the mean window is 300; in S301, the number of mean values m of the amplitude sliding window is 40; in S2, the prediction step s1 of the first prediction model is set to 40; the value of k in equation (1) is set to 5; in S5, the prediction step s2 of the second prediction model is set to 40.
7. A hazardous flow pattern safety early warning system for an oil and gas gathering and transmission riser system, characterized in that, include: The preprocessing module is used to preprocess the collected horizontal loop differential pressure signal and riser differential pressure signal in the collection and transmission-riser system to obtain the historical sequence of the average horizontal loop differential pressure, the historical sequence of the average riser differential pressure, and the historical sequence of the riser differential pressure amplitude. The first prediction model construction module is used to select the historical sequence of the average riser pressure difference to build the first prediction model based on LSTM and predict the average riser pressure difference prediction sequence. The flow state determination module is used to extract the amplitude feature values of the riser differential pressure average prediction sequence, the riser differential pressure average historical sequence, and the riser differential pressure average real-time sequence and compare them as the first discrimination condition; at the same time, it combines the horizontal loop differential pressure average historical sequence threshold as the second discrimination condition, and jointly determine the flow state in the gathering and transmission-riser system. Specifically, it includes: S301) Based on the historical sequence of average riser pressure differential M DP2 The amplitude characteristic value A of the historical sequence of the average riser pressure difference was calculated. his,DP2,M The specific process is as follows: Among them, A DP2,M,i M is the amplitude window of the i-th value. DP2 The range of the sequence; m is the number of mean values in the amplitude sliding window; n is the number of sampling points in the mean window; N is the number of sliding steps in the window; S302) Predict sequence M based on the average riser pressure difference pred,DP2 Solve for the amplitude characteristic value A of the predicted mean value of riser pressure difference. pred,DP2,M,i : S303) When the last time corresponding to the i-th (i≥n+N+1) window is reached, solve for the amplitude characteristic value A of the measured mean value of the riser pressure difference. real,DP2,M,i : S304) Solve for the maximum value M of the historical series of the horizontal loop pressure difference mean. his,DP1,max With minimum value M his,DP1,min : S305) Determine whether the flow state in the system is transient based on the following conditions; , k consecutive windows (1) （2） (a) If equations (1) and (2) are not satisfied at the same time, the flow state is determined to be steady state flow, and the process ends when the last time corresponding to window i = n+N+s1 is reached. (b) If both equation (1) and equation (2) are satisfied, the flow state is determined to be transient flow, and an early warning is triggered. The second prediction model construction module is used to preprocess the transient data if the flow state is transient, and combine it with the historical sequence of riser differential pressure amplitude obtained in S1 to form a new riser differential pressure amplitude sequence, and then establish the second prediction model based on LSTM. The harmful flow pattern early warning module is used to obtain a new riser pressure differential amplitude prediction sequence based on the second prediction model, and to determine whether a harmful flow pattern will appear. If a harmful flow pattern appears, an alarm mechanism is triggered.
8. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1-6.
9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1-6.