A method, device and apparatus for determining the stop pressure in a pipeline

Abstract

The embodiments of this specification provide a method, device and equipment for determining the stop pressure in a pipeline, including: obtaining index data of a target pipeline, the index data changing with the pipeline stop time; determining the pressure in the target pipeline at different stop times according to the index data and a pipeline stop pressure prediction model; wherein the pipeline stop pressure prediction model is obtained by training a preset autoencoder network model based on stop information, the neuron structure of the preset autoencoder network model considers time series characteristics during the updating process, and the parameter information of the preset autoencoder network model is determined based on a particle swarm optimization algorithm. The embodiments of this specification can improve the accuracy of prediction of the stop pressure of a refined oil pipeline.

Description

Technical Field

[0001] The present application relates to the technical field of pipeline transmission, and in particular to a method, a device and an apparatus for determining a stop-transmission pressure in a pipeline. Background Art

[0002] The shutdown operation of refined oil pipelines often occurs during the commissioning and operation of refined oil pipelines. The pressure in the pipeline changes dramatically during the shutdown and restart process. During the pipeline shutdown period, on-site personnel cannot predict the pressure changes in the pipeline in advance. When the pressure in the pipeline drops, it is often mistaken for pipeline leakage or drilling and oil theft accidents, which requires personnel to patrol and investigate, thus reducing work efficiency. Therefore, it is becoming increasingly important to study the pressure changes during the shutdown of refined oil pipelines.

[0003] At present, the research on the change of the stop pressure of refined oil pipeline mainly relies on empirical formulas and numerical simulation of a single pipeline. However, since the relevant data of the actual refined oil pipeline is constantly changing, the accuracy of the prediction of the stop pressure of refined oil pipeline can be easily reduced if the research on the change of the stop pressure relies only on empirical formulas.

[0004] Therefore, the industry is in urgent need of a technical solution that can solve the above technical problems. Summary of the invention

[0005] The embodiments of this specification provide a method, device and equipment for determining the stop-flow pressure in a pipeline, which can improve the accuracy of predicting the stop-flow pressure of a refined oil pipeline.

[0006] The present specification provides a method, device and equipment for determining the stop pressure in a pipeline, which is implemented in the following manner.

[0007] A method for determining the stop pressure in a pipeline comprises: obtaining index data of a target pipeline, wherein the index data varies with the pipeline stop time; determining the pressure in the target pipeline at different stop times according to the index data and a pipeline stop pressure prediction model; wherein the pipeline stop pressure prediction model is obtained by training a preset autoencoder network model based on stop information, the neuron structure of the preset autoencoder network model considers time series characteristics during the updating process, and the parameter information of the preset autoencoder network model is determined based on a particle swarm optimization algorithm.

[0008] A training method for a pipeline stop pressure prediction model comprises: constructing a preset autoencoder network model; the neuron structure of the preset autoencoder network model takes time series characteristics into consideration during the updating process, and parameter information of the preset autoencoder network model is determined based on a particle swarm optimization algorithm; obtaining pipeline stop information; the stop information comprises stop pressure and index data corresponding to the stop pressure, and the index data varies with the pipeline stop time; inputting the index data into the preset autoencoder network model to obtain an output result; calculating the error between the output result and the stop pressure; and correcting the preset autoencoder network model based on the error to obtain a pipeline stop pressure prediction model.

[0009] A device for determining the stop pressure in a pipeline, comprising: an index data acquisition module, used to acquire index data of a target pipeline, wherein the index data changes with the pipeline stop time; a pipeline pressure determination module, used to determine the pressure in the target pipeline at different stop times according to the index data and a pipeline stop pressure prediction model; wherein the pipeline stop pressure prediction model is obtained by training a preset autoencoder network model based on stop information, the neuron structure of the preset autoencoder network model considers time series characteristics during the updating process, and the parameter information of the preset autoencoder network model is determined based on a particle swarm optimization algorithm.

[0010] A training device for a pipeline stop pressure prediction model comprises: a construction module for constructing a preset autoencoder network model; the neuron structure of the preset autoencoder network model considers time series characteristics during the updating process, and the parameter information of the preset autoencoder network model is determined based on a particle swarm optimization algorithm; an acquisition module for acquiring pipeline stop information; the stop information includes stop pressure and index data corresponding to the stop pressure, and the index data changes with the pipeline stop time; the acquisition module is used to input the index data into the preset autoencoder network model to obtain an output result; a calculation module is used to calculate the error between the output result and the stop pressure; and a correction module is used to correct the preset autoencoder network model based on the error to obtain a pipeline stop pressure prediction model.

[0011] A device for determining the stop-transmission pressure in a pipeline comprises a processor and a memory for storing instructions executable by the processor. When the processor executes the instructions, the steps of any method embodiment in the embodiments of this specification are implemented.

[0012] A computer-readable storage medium stores computer instructions, which, when executed, implement the steps of any method embodiment in the embodiments of this specification.

[0013] The present specification provides a method, device and equipment for determining the stop pressure in a pipeline. In some embodiments, the pipeline stop pressure prediction model is obtained by obtaining the stop information of the pipeline, training the preset autoencoder network model according to the stop pressure included in the stop information and the index data corresponding to the stop pressure. Furthermore, the stop pressure of pipelines with different stop durations is predicted according to the obtained pipeline stop pressure prediction model. Since the index data is data that changes with the pipeline stop time determined based on the mechanism model, the neuron structure of the preset autoencoder network model considers the time series characteristics during the updating process, and the parameter information of the preset autoencoder network model is determined based on the particle swarm optimization algorithm, so that the obtained pipeline stop pressure prediction model has high accuracy and strong versatility for predicting the stop pressure of the refined oil pipeline. The implementation scheme provided in the present specification can improve the accuracy of the prediction of the stop pressure of the refined oil pipeline. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The drawings described herein are used to provide a further understanding of this specification, constitute a part of this specification, and do not constitute a limitation of this specification. In the drawings:

[0015] Figure 1 A flow chart of a method for determining the stop pressure in a pipeline provided in an embodiment of this specification;

[0016] Figure 2 A schematic diagram of the basic unit structure of the LSTM provided in the embodiments of this specification;

[0017] Figure 3 A schematic diagram showing the change of pipeline pressure of pipeline A with the stoppage time provided in the embodiment of this specification;

[0018] Figure 4 A schematic diagram of prediction errors of different prediction models provided in the embodiments of this specification in pipeline A;

[0019] Figure 5 A schematic diagram showing the change of pipeline pressure of pipeline B with the stoppage time provided in the embodiment of this specification;

[0020] Figure 6 A schematic diagram of prediction errors of different prediction models provided in the embodiments of this specification in pipeline B;

[0021] Figure 7 A schematic diagram showing the change of pipeline pressure of pipeline C with the stoppage time provided in the embodiment of this specification;

[0022] Figure 8 A schematic diagram of prediction errors of different prediction models provided in the embodiments of this specification in pipeline C;

[0023] Fig. 9A schematic diagram of the module structure of a device for determining the stop pressure in a pipeline provided in an embodiment of this specification;

[0024] Fig.10 A schematic diagram of the structure of a training device for a pipeline shut-down pressure prediction model provided in an embodiment of this specification;

[0025] Fig.11 A hardware structure block diagram of a server for determining the stop pressure in a pipeline provided in an embodiment of this specification. DETAILED DESCRIPTION

[0026] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments in this specification, not all of the embodiments. Based on one or more embodiments in this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the embodiments of this specification.

[0027] The following is an example of a specific application scenario to illustrate the implementation plan of this specification. Specifically, Figure 1 A flow chart of a method for determining the stop pressure in a pipeline provided in an embodiment of this specification. Although this specification provides method operation steps or device structures as shown in the following embodiments or drawings, the method or device may include more or fewer operation steps or module units after partial combination based on routine or no creative labor.

[0028] An implementation scheme provided in this specification can be applied to a client, a server, etc. The client can include a terminal device, such as a smart phone, a tablet computer, etc. The server can include a single computer device, or a server cluster composed of multiple servers, or a server structure of a distributed system, etc.

[0029] It should be noted that the following embodiment description does not limit the technical solutions in other application scenarios that can be extended based on this specification. Figure 1 As shown, in an embodiment of a method for determining a stop-transmission pressure in a pipeline provided in this specification, the method may include the following steps.

[0030] S0: Obtaining index data of the target pipeline, wherein the index data changes with the pipeline stoppage time.

[0031] In the embodiments of this specification, the target pipeline can be understood as a pipeline for which the stop pressure needs to be predicted. The index data that changes with the pipeline stop time may at least include the stop duration, the ambient temperature, and the oil product temperature after the stop. The index data can be used to predict the stop pressure of the pipeline. It should be noted that during the operation of the hot oil pipeline, it is inevitable that the pipeline will stop due to various reasons. The reasons for the stop can be divided into accidental stop and planned stop, but whether it is an accidental stop or a planned stop, the pipeline stop time cannot be longer than the safe stop time. In the embodiments of this specification, the stop can be understood as the stop of the transportation of refined oil, and the pipeline stop can be understood as the pipeline stopping the transportation of refined oil. Correspondingly, the stop duration can be understood as the time when the pipeline stops transporting refined oil, and the stop information can be understood as some information of the pipeline during the period when the pipeline stops transporting refined oil.

[0032] In some embodiments, the index data may be pre-stored in a database, memory, etc., so that when the index data needs to be obtained, it can be obtained from the database, memory, etc. in real time, thereby improving information processing efficiency. Of course, the index data may also be stored in other devices, systems, etc., such as a SCADA (Supervisory Control And Data Acquisition) system of a finished oil pipeline, which is not limited in this specification.

[0033] S2: Determine the pressure in the target pipeline at different outage durations according to the indicator data and the pipeline outage pressure prediction model; wherein the pipeline outage pressure prediction model is obtained by training a preset autoencoder network model based on outage information, the neuron structure of the preset autoencoder network model considers time series characteristics during the updating process, and the parameter information of the preset autoencoder network model is determined based on a particle swarm optimization algorithm.

[0034] In the embodiment of the present specification, after the index data of the target pipeline is obtained, the index data can be input into the pipeline stop pressure prediction model to determine the stop pressure of the target pipeline at different stop durations.

[0035] In some embodiments, before inputting indicator data into a pipeline stop pressure prediction model to determine the stop pressure of a target pipeline at different stop times, a preset autoencoder network model can be pre-trained based on the stop information to obtain a pipeline stop pressure prediction model. In this way, when it is necessary to predict the pipeline stop pressure subsequently, the indicator data of the pipeline to be predicted can be directly input into the model to obtain a prediction result, thereby improving the prediction efficiency.

[0036] In the embodiments of this specification, the training of the pipeline shut-down pressure prediction model may include the following steps:

[0037] S20: constructing a preset autoencoder network model; the neuron structure of the preset autoencoder network model considers time series characteristics during the updating process, and the parameter information of the preset autoencoder network model is determined based on a particle swarm optimization algorithm;

[0038] S21: Acquire pipeline stop information; the stop information includes stop pressure and index data corresponding to the stop pressure, and the index data changes with pipeline stop time;

[0039] S22: inputting the indicator data into a preset autoencoding network model to obtain an output result;

[0040] S23: Calculating the error between the output result and the stop pressure;

[0041] S24: Correcting the preset autoencoder network model based on the error to obtain a pipeline shut-down pressure prediction model.

[0042] In some embodiments, the preset autoencoder network model may be a long short-term memory neural network model. The parameter information of the preset autoencoder network model may be determined based on a particle swarm optimization algorithm. The parameter information may include an input weight matrix, a recursive weight matrix, a hidden output weight matrix, a bias vector, and the like. Long-Short Term Memory (LSTM) is evolved based on a recurrent neural network (RNN). RNN forms a self-loop structure to pass the information extracted at each moment to the next moment, and the time series characteristics can be taken into account. Since RNN has a vanishing gradient, it cannot handle long dependency problems. In LSTM, the outputs of the three gates are calculated using a sigmoid function, and the output is between 0 and 1. If the output is close to 0, it may not be that the information of the previous moment has a significant impact on the moment, and there is no need to return to adjust the weights through gradient descent. If the output is close to 1, the gradient can be smoothly propagated in the network, which largely avoids the situation where the gradient drops to zero and the gradient disappears. Therefore, LSTM can solve the problem of gradient disappearance to a certain extent through gate control. The sigmoid function can be understood as an activation function that can map variables to between (0,1).

[0043] like Figure 2 As shown, Figure 2 The basic unit structure diagram of LSTM provided in the embodiment of this specification. At time t, the unit state c at time t-1 is input to the LSTM unit. t-1 , hidden state h t-1 And the new information X t (T t ,Tet ,t), for the convenience of subsequent description, x t represents (t,Te t ,T t ). When the entire unit is in an activated state, the three main components (forget gate, input gate, and output gate) will calculate the input information according to their respective operation properties, and the activation function of each gate will limit the output value of the gate to between 0 and 1. Specifically, the information calculation and transmission of LSTM can be divided into the following steps:

[0044] (1) LSTM first determines what information of the cell state at the previous moment needs to be retained. This step is completed in the forget gate. The sigmoid function in the forget gate checks the h at the previous moment. t-1 And the new input information x t , and use this to output a vector f between 0 and 1 t , vector f t The size of can indicate how much information is retained at the previous moment. 0 means that all information at the previous moment is discarded, and 1 means that all information at the previous moment is retained. The function of sigmoid is as shown in formula (1), t The calculation of is shown in formula (2).

[0045]

[0046] f t =σ(W xf ·x t +W hf ·h t-1 +b f ) (2)

[0047] (2) After determining the information that needs to be retained at the previous moment, we can further determine what new information to add to the unit state. Specifically, we first analyze the input h t-1 and x t To calculate the input gate i t , in order to decide which information needs to be updated, and then calculate the candidate information through the tanh function The candidate information will decide whether to update to the new unit state based on the information that needs to be updated calculated in the previous step. The tanh function is as shown in formula (3), i t The calculation of is as shown in formula (4), The calculation of is shown in formula (5).

[0048]

[0049] i t =σ(W xi ·x t +Whi ·h t-1 +b i ) (4)

[0050]

[0051] (3) After obtaining the information that needs to be updated and the candidate new information, the information at the previous moment can be updated, that is, the unit state c output at this moment can be determined. t Specifically, the old cell state c can be retained by the forget gate t-1 The partial information of the second step is used to calculate the candidate information obtained for updating The unit status information at this moment is updated, as shown in formula (6).

[0052]

[0053] (4) After completing the update of the unit state, the final output content can be determined. Specifically, the new information x t and the hidden state h at the previous moment t-1 Pass it to the output gate, use the sigmoid function and tanh function to determine the output of the hidden state at this moment, and output the final prediction result P of the pressure at time t t . The specific details are shown in formulas (7), (8) and (9).

[0054] o t =σ(W xo ·x t +W ho ·h t-1 +b o ) (7)

[0055] h t =o t *tanh(c t ) (8)

[0056] P t =σ(W hy ·h t +b y ) (9)

[0057] Among them, σ(x) is the sigmoid function, f t is the value of the forget gate at time t, x t is the input information at time t, h t-1 is the state of the hidden layer at time t-1 (referred to as “hidden state”), i t is the value of the input gate at time t, is the candidate information at time t, c t is the cell state at time t, ct-1 is the unit state at time t-1, o t is the value of the output gate at time t, h t is the state of the hidden layer at time t, W xf , W xi , W xo and W xc are the input weight matrices, W hf , W hi , W ho and W hc are the recursive weight matrices, W hy is the hidden output weight matrix, b f , b i , b o , b c and b y are the corresponding deviation vectors, p t is the predicted pressure value at time t.

[0058] Of course, the above is only an exemplary description, and the preset autoencoder network model is not limited to the above example. Technical personnel in the relevant field may make other changes under the inspiration of the technical essence of this application, but as long as the functions and effects achieved are the same or similar to those of this application, they should be covered within the scope of protection of this application.

[0059] In some embodiments, the stop information may include the stop pressure and index data corresponding to the stop pressure. The index data varies with the pipeline stop time. The stop information may be understood as some information during the pipeline stop period.

[0060] In some implementation scenarios, before obtaining the pipeline stop information, the influencing factors of the pipeline stop pressure can be determined according to the preset relationship between the oil product temperature and time after the pipeline is stopped, and the preset relationship between the pipeline pressure and the oil product temperature. Then, the influencing factors that change with the pipeline stop time are screened from the influencing factors of the pipeline stop pressure, and the data corresponding to the influencing factors that change with the pipeline stop time are used as indicator data. Among them, the preset relationship between the oil product temperature and time after the pipeline is stopped, and the preset relationship between the pipeline pressure and the oil product temperature can be understood as a mechanism model. It should be noted that the finished oil pipeline is usually transported in a closed manner at normal temperature, and the finished oil has the physical properties of expansion and compression. Due to the expansion of the oil product, there is a certain relationship between the temperature and volume of the oil product, and the change in volume is related to the pipeline pressure. Therefore, it can be considered that the difference between the temperature of the pipeline oil and the pipeline environment is the main factor causing the change in pipeline pressure. In some implementation scenarios, in the initial stage of the stop, the temperature of the oil product in the pipe is significantly higher than the ambient temperature, and the heat exchange between the pipeline and the soil is strong. As time increases, the difference between the oil product temperature and the ambient temperature is small, that is, the oil product temperature will change with time after the pipeline is stopped. Since changes in oil temperature will cause changes in oil volume, resulting in changes in pressure within the pipe, the pipeline pressure will also change with the oil temperature.

[0061] In some implementation scenarios, the preset relationship between the oil product temperature and time after the pipeline stops transporting oil, and the preset relationship between the pipeline pressure and the oil product temperature can be expressed as:

[0062]

[0063]

[0064]

[0065] Where K1 is the total heat transfer coefficient, D, D1, and D2 are the average diameter of the pipe, the inner diameter of the pipe, and the outer diameter of the pipe, respectively. y 、c g are the specific heat capacities of oil and steel, ρ y , g are the densities of oil and steel, Te t is the ambient temperature, T t is the oil temperature after the oil is shut down for t hours, T0 is the oil temperature when the oil is shut down, t is the shut down time, α is the volume expansion coefficient of the pipeline, δ is the wall thickness of the pipeline, E is the Young's elastic modulus of the pipe, β is the volume expansion coefficient of the oil, K2 is the elastic coefficient of the oil, T i is the oil temperature at the i-th moment, T i-1 is the oil temperature at the i-1th moment, ΔP is the pressure difference between the i-th moment and the i-1th moment, and A represents the coefficient.

[0066] In some implementation scenarios, factors affecting the pipeline stop pressure determined based on the preset relationship between the oil temperature and time after pipeline stoppage, and the preset relationship between the pipeline pressure and the oil temperature may include the average diameter of the pipeline, the inner diameter of the pipeline, the outer diameter of the pipeline, the volume expansion coefficient of the pipeline, the wall thickness of the pipeline, the Young's coefficient of the pipe, the specific heat capacity of steel, the density of steel, the specific heat capacity of oil, the density of oil, the volume expansion coefficient of the oil, the elastic coefficient of the oil, the oil temperature, the ambient temperature, the stop pressure, the stop duration, the oil temperature after the stoppage, etc.

[0067] In some implementation scenarios, after determining the influencing factors of pipeline stop pressure, for the convenience of analysis, these influencing factors can be divided into two categories. The first category does not change with the stop time, such as soil properties such as total heat transfer coefficient, basic pipeline data, initial oil temperature T0, etc. The second category changes with the stop time, such as stop duration, ambient temperature, oil product temperature, etc. In some implementation scenarios, the values ​​of the first category of influencing factors are generally obtained by estimation, and their values ​​are different for different pipelines.

[0068] In some implementation scenarios, after the influencing factors of pipeline outage pressure are classified, the influencing factors that change with the pipeline outage time can be screened out, and then the data corresponding to the influencing factors that change with the pipeline outage time are used as indicator data.

[0069] In some implementation scenarios, after screening out the influencing factors that change with the pipeline outage time, the data corresponding to the influencing factors that change with the pipeline outage time can be stored in a database, memory, etc. In this way, when the outage information needs to be obtained, it can be obtained from the database, memory, etc. in real time, thereby improving information processing efficiency.

[0070] In some implementation scenarios, after screening out the factors that affect the pipeline outage time, the ambient temperature, oil product temperature, and outage pressure corresponding to different outage durations can be obtained, and a sample database of outage pipeline pressure changes can be established to train the pipeline outage pressure prediction model based on the sample database. The ambient temperature can be obtained from weather websites, such as https: / / www.wunderground.com / .

[0071] In the embodiments of the present specification, after the pipeline outage information is obtained, the outage information can be used to train a preset autoencoder network model to obtain a pipeline outage pressure prediction model.

[0072] In some embodiments, before inputting the indicator data into the preset autoencoder network model, the stop-transmission pressure and the indicator data can be normalized to obtain the normalized stop-transmission pressure and the normalized indicator data, and then the normalized indicator data is input into the preset autoencoder network model to obtain the output result, and then the error between the output result and the normalized stop-transmission pressure is calculated, and the preset autoencoder network model is corrected based on the error to obtain the pipeline stop-transmission pressure prediction model.

[0073] In some implementation scenarios, due to the stop pressure P t And the stop time t, ambient temperature Te t , oil temperature T t The value range of is different, so the data can be normalized before model training. In some implementation scenarios, the minimum-maximum method can be used for normalization to constrain the values ​​of each dimension to the range of 0-1.

[0074] Among them, the minimum-maximum normalization formula can be expressed as:

[0075]

[0076] Among them, x norm is the normalized value of each variable, x orig is the original value of each variable, x max and x min are the maximum and minimum values ​​of each variable respectively.

[0077] In the embodiments of this specification, by normalizing the data before model training, data in different ranges are processed into a consistent range, which can accelerate the convergence of the training model while maintaining data relevance and improve the speed of model training.

[0078] In some embodiments, after obtaining the pipeline stoppage information, the indicator data can be input into a preset autoencoder network model to obtain an output result, and then the error between the output result and the stoppage pressure is calculated. Based on the error, the preset autoencoder network model is corrected to obtain a pipeline stoppage pressure prediction model.

[0079] In some embodiments, after obtaining the pipeline shut-down pressure prediction model, the effectiveness of the obtained pipeline shut-down pressure prediction model can be evaluated by preset performance indicators. For example, in some implementation scenarios, mean absolute error (MAE), root mean square error (RMSE) and mean absolute percentage error (MAPE) can be used as performance indicators of the prediction model. Specifically, MAE, RMSE and MAPE can be calculated by the following formula:

[0080]

[0081]

[0082]

[0083] Among them, n is the number of samples, p t is the predicted pressure value at time t, r i is the true value of the pressure of the i-th sample at the t-th moment, that is, the stop pressure.

[0084] In the embodiments of this specification, after the pipeline shut-down pressure prediction model is obtained, the shut-down pressure of pipelines with different shut-down durations can be predicted using the model.

[0085] In some embodiments, after determining the stop pressure of pipelines with different stop times using a pipeline stop pressure prediction model, the predicted value can be compared with the actual detection value. When the predicted value differs greatly from the actual detection value, there may be an abnormality in the pipeline. At this time, relevant management personnel can be notified to conduct emergency inspections.

[0086] In some embodiments, a pipeline outage pressure prediction model can be used to predict and monitor the pipeline pressure during the outage of the refined oil pipeline in real time, thereby reducing personnel inspections and improving work efficiency.

[0087] Of course, the above is only an exemplary description, and the embodiments of this specification are not limited to the above examples. Technical personnel in the relevant field may make other changes based on the technical essence of this application. However, as long as the functions and effects achieved are the same or similar to those of this application, they should be included in the scope of protection of this application.

[0088] The above method is described below in conjunction with a specific embodiment. However, it should be noted that the specific embodiment is only for better illustrating the present application and does not constitute an improper limitation on the present application.

[0089] In this embodiment, in order to verify the accuracy and universality of the obtained pipeline outage pressure prediction model, three pipelines with different outage durations were selected for analysis and verification. The specific outage information is shown in Table 1, where the outage durations of pipelines A, B, and C increase in sequence, which can be used to represent the short, medium, and long outage periods, respectively. In order to illustrate the superiority of considering time series characteristics, several traditional machine learning algorithms such as artificial neural network (ANN), support vector machine (SVR), decision tree (DT), and random forest (RF) can be selected for prediction at the same time. Among them, ANN includes an input layer, a hidden layer, and an output layer. Information flows through each layer in sequence, but will not be transmitted between network neurons in the same layer, which results in the output information being only related to the size of the input value and cannot be associated with the input order of the input value. SVR can be used for classification and regression. The basic idea is to find a "hyperplane" with the maximum interval so that the distance from the existing data sample to the "hyperplane" is the largest. The basic idea of ​​DT is to divide the feature space into different units, and determine the corresponding output according to the division, which is similar to the process of building a tree. RF is an integrated learning algorithm that is composed of multiple decision trees and comprehensively considers all decision trees to determine the final prediction result.

[0090] In this embodiment, in order to maintain the original time series characteristics of the pipeline stop pressure, the sample data required by the pipeline stop pressure prediction model obtained in this application is not disrupted, but in order to improve the accuracy of the predictions of the ANN, SVR, DT and RF prediction models, the sample data can be disrupted. Among them, the sample data can be pre-stored in the sample database. The implementation environment of different prediction models is the open source scikit-learn and Tensorflow libraries. Scikit-learn is a free software machine learning library for the Python programming language, and TensorFlow is a symbolic mathematical system based on data flow programming, which can be applied to the programming implementation of various machine learning algorithms.

[0091] Table 1 The corresponding shutdown information of the three pipelines

[0092] pipeline Pipeline A Pipeline B Pipeline C Stop transmission start time 2019-5-5 21:00:00 2019-6-10 23:00:00 2019-3-30 16:00:00 Stop transmission end time 2019-5-7 12:00:00 2019-6-15 23:00:00 2019-4-11 15:00:00 Stoppage time (h) 39 120 287

[0093] In this embodiment, the pressure change of pipeline A at the terminal from May 5 to May 7, 2019 is taken as an example. The suspension time is 39 hours. The schematic diagram of the change of pipeline pressure with suspension time is as follows: Figure 3 As shown. Figure 3 It can be seen that the pipeline pressure has been on a downward trend, which is most likely due to the fact that the oil temperature has been higher than the ambient temperature, and the oil and soil have exchanged heat, causing the pressure in the pipe to drop. At this point, the sample data can be input into different prediction models to obtain the following: Figure 4The prediction error diagram shown in FIG. 1 is a schematic diagram of the prediction error, where LSTM represents the pipeline shut-down pressure prediction model obtained in this application, and the three performance indicators corresponding to each prediction model are RMSE, MAE, and MAPE from left to right. Figure 4 It can be seen that SVR has the largest prediction error, followed by ANN, and LSTM has the best prediction effect, with MAE and RMSE close to 0. In order to intuitively show the prediction effect of LSTM, the predicted value and the true value are compared in Figure 3 It can be seen that the two are almost overlapping, indicating the reliability of the prediction results.

[0094] In this embodiment, taking the case where pipeline B stops transporting gas from June 10 to June 15, 2019, the duration of the stoppage is 120 hours. The schematic diagram of the change of pipeline pressure with the stoppage time is as follows: Figure 5 As shown. Figure 5 It can be seen that the pipeline pressure first decreases and then becomes almost stable, which may be caused by the fact that the oil temperature gradually decreases and is not much different from the ambient temperature. At this time, the sample data can be input into different prediction models to obtain the following Figure 6 The prediction error diagram shown in FIG. 1 is a schematic diagram of the prediction error, where LSTM represents the pipeline shut-down pressure prediction model obtained in this application, and the three performance indicators corresponding to each prediction model are RMSE, MAE, and MAPE from left to right. Figure 6 It can be seen that although LSTM is not much different from other prediction models in terms of MAE and RMSE, LSTM is obviously the smallest in terms of MAPE, and its advantage is more prominent. In order to intuitively show the prediction effect of LSTM, the predicted value and the true value are compared in Figure 5 is displayed in the

[0095] In this embodiment, taking the case where pipeline C is shut down from March 30 to April 11, 2019 as an example, the shut down time is 287 hours. The schematic diagram of the change of pipeline pressure with the shut down time is as follows: Figure 7 As shown. Figure 7 It can be seen that the pipeline pressure first drops, then stabilizes and then rises. This may be because the oil temperature first drops and then the pressure tends to stabilize when it is close to the ambient temperature. However, due to seasonal changes (from March to April), the ambient temperature rises, causing heat exchange from the ambient temperature to the oil temperature, which in turn causes the pipeline pressure to rise. At this point, the sample data can be input into different prediction models to obtain the following: Figure 8 The prediction error diagram shown in FIG. 1 is a schematic diagram of the prediction error, where LSTM represents the pipeline shut-down pressure prediction model obtained in this application, and the three performance indicators corresponding to each prediction model are RMSE, MAE, and MAPE from left to right. Figure 8 It can be seen that LSTM still has the smallest performance among the three performance indicators and performs best. Similarly, in order to intuitively demonstrate the prediction effect of LSTM, the predicted value and the true value are compared in Figure 7 is displayed in the

[0096] In the embodiments of this specification, by comparing the above three pipelines with different outage durations, it can be seen that the pipeline outage pressure prediction model obtained in this application has outstanding prediction effects on different pipelines and different outage periods, with higher accuracy and strong versatility. Especially when the outage duration is short and the sample data is small, since it takes into account the time series characteristics, it has good prediction effects and strong robustness.

[0097] It should be noted that the above implementation is only schematically illustrated by taking three pipelines with different outage durations as examples. The embodiments of this specification can also predict the outage pressure of pipelines with other outage durations, and this application does not limit this.

[0098] From the above description, it can be seen that the embodiments of the present application can achieve the following technical effects: the pipeline stop information can be obtained, and the preset autoencoder network model can be trained according to the stop pressure included in the stop information and the index data corresponding to the stop pressure to obtain the pipeline stop pressure prediction model. Furthermore, the stop pressure of pipelines with different stop times can be determined according to the obtained pipeline stop pressure prediction model. Since the index data is data that changes with the pipeline stop time based on the mechanism model, the neuron structure of the preset autoencoder network model considers the time series characteristics during the update process, and the parameter information of the preset autoencoder network model is determined based on the particle swarm optimization algorithm, so that the obtained pipeline stop pressure prediction model has high accuracy and strong versatility for predicting the stop pressure of the refined oil pipeline.

[0099] In this specification, each embodiment of the above method is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. For relevant parts, please refer to the partial description of the method embodiment.

[0100] Based on the above-mentioned method for determining the stop pressure in a pipeline, one or more embodiments of this specification also provide a device for determining the stop pressure in a pipeline. The device may include a system (including a distributed system), software (application), module, component, server, client, etc. using the method described in the embodiment of this specification and a device combined with necessary implementation hardware. Based on the same innovative concept, the device in one or more embodiments provided in the embodiment of this specification is as described in the following embodiments. Since the implementation scheme of the device to solve the problem is similar to the method, the implementation of the specific device in the embodiment of this specification can refer to the implementation of the aforementioned method, and the repetitions will not be repeated. As used below, the term "unit" or "module" can be a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceived.

[0101] Specifically, Fig. 9 A schematic diagram of a module structure of a device for determining a stop pressure in a pipeline provided in an embodiment of this specification is shown in FIG. Fig. 9 As shown, a device for determining the stop-transmission pressure in a pipeline provided in this specification may include: an indicator data acquisition module 120 and a pipeline pressure determination module 122.

[0102] The index data acquisition module 120 can be used to acquire the index data of the target pipeline, and the index data changes with the pipeline stop time;

[0103] The pipeline pressure determination module 122 can be used to determine the pressure in the target pipeline at different outage durations based on the indicator data and the pipeline outage pressure prediction model; wherein the pipeline outage pressure prediction model is obtained by training a preset autoencoder network model based on the outage information, the neuron structure of the preset autoencoder network model considers the time series characteristics during the updating process, and the parameter information of the preset autoencoder network model is determined based on the particle swarm optimization algorithm.

[0104] Fig.10 A schematic diagram of a training device structure for a pipeline shut-down pressure prediction model provided in an embodiment of this specification is shown in FIG. Fig.10 As shown, a training device for a pipeline shut-down pressure prediction model provided in this specification may include: a construction module 210 , an acquisition module 212 , an acquisition module 214 , a calculation module 216 , and a correction module 218 .

[0105] The construction module 210 can be used to construct a preset autoencoder network model; the neuron structure of the preset autoencoder network model considers the time series characteristics during the updating process, and the parameter information of the preset autoencoder network model is determined based on the particle swarm optimization algorithm;

[0106] The acquisition module 212 may be used to acquire pipeline stop information; the stop information includes a stop pressure and index data corresponding to the stop pressure, and the index data changes with the pipeline stop time;

[0107] The obtaining module 214 may be used to input the indicator data into a preset autoencoding network model to obtain an output result;

[0108] A calculation module 216, which can be used to calculate the error between the output result and the stop pressure;

[0109] The correction module 218 can be used to correct the preset autoencoder network model based on the error to obtain a pipeline shut-down pressure prediction model.

[0110] It should be noted that the above-mentioned device may also include other implementation modes according to the description of the method embodiment. The specific implementation mode may refer to the description of the relevant method embodiment, which will not be described one by one here.

[0111] This specification also provides an embodiment of a device for determining the stop pressure in a pipeline, including a processor and a memory for storing processor executable instructions, and the processor can implement any of the above method embodiments when executing the instructions. For example, when the instructions are executed by the processor, the implementation includes the following steps: obtaining index data of the target pipeline, the index data changes with the pipeline stop time; determining the pressure in the target pipeline at different stop durations according to the index data and the pipeline stop pressure prediction model; wherein the pipeline stop pressure prediction model is obtained by training a preset autoencoder network model based on the stop information, the neuron structure of the preset autoencoder network model considers the time series characteristics during the update process, and the parameter information of the preset autoencoder network model is determined based on the particle swarm optimization algorithm.

[0112] It should be noted that the above-mentioned device may also include other implementations according to the description of the method or device embodiment. The specific implementation methods can refer to the description of the relevant method embodiments, which will not be described one by one here.

[0113] The method embodiments provided in this specification can be executed in a mobile terminal, a computer terminal, a server or a similar computing device. Taking running on a server as an example, Fig.11 The hardware structure block diagram of a server for determining the stop pressure of pipeline provided in the embodiment of this specification, the server may be the device for determining the stop pressure of pipeline or the system for determining the stop pressure of pipeline in the above embodiment. Fig.11 As shown, the server 10 may include one or more (only one is shown in the figure) processors 100 (the processor 100 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory 200 for storing data, and a transmission module 300 for communication functions. A person skilled in the art can understand that Fig.11 The structure shown is only for illustration and does not limit the structure of the above electronic device. Fig.11 More or fewer components shown in the figure may also include other processing hardware, such as a database or multi-level cache, GPU, or other hardware with Fig.11 Different configurations are shown.

[0114] The memory 200 can be used to store software programs and modules of application software, such as the program instructions / modules corresponding to the method for determining the stop pressure in the pipeline in the embodiment of this specification. The processor 100 executes various functional applications and data processing by running the software programs and modules stored in the memory 200. The memory 200 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 200 may further include a memory remotely arranged relative to the processor 100, and these remote memories may be connected to the computer terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0115] The transmission module 300 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of a computer terminal. In one example, the transmission module 300 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission module 300 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0116] The above is a description of a specific embodiment of the present specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or sequential order shown to achieve the desired results.

[0117] Each embodiment in this application is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for hardware + program embodiments, since they are basically similar to method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. It should be understood by those skilled in the art that one or more embodiments of this specification can be provided as methods, devices or equipment. Therefore, one or more embodiments of this specification may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware.

[0118] The above is only an example of one or more embodiments of this specification and is not intended to limit one or more embodiments of this specification. For those skilled in the art, one or more embodiments of this specification may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of this application should be included in the scope of the claims.

Claims

1. A method for determining the stop pressure in a pipeline. It is characterized in that include: Acquire indicator data of a target pipeline, wherein the indicator data changes with pipeline shutdown time; According to the indicator data and the pipeline outage pressure prediction model, the pressure in the target pipeline at different outage durations is determined; wherein the pipeline outage pressure prediction model is obtained by training a preset autoencoder network model based on outage information, the neuron structure of the preset autoencoder network model considers time series characteristics during the updating process, and the parameter information of the preset autoencoder network model is determined based on a particle swarm optimization algorithm, and the parameter information includes: an input weight matrix, a recursive weight matrix, a hidden output weight matrix, and a bias vector; The preset autoencoder network model is a long short-term memory neural network model, including: a forget gate, an input gate, and an output gate; Accordingly, the neuron structure of the preset autoencoder network model is updated in the following manner: At time t, the cell state c at time t-1 is transmitted to the neuron structure of the preset autoencoder network model. t-1 , hidden state h t-1 And the new information x t ; Use the sigmoid function of the forget gate to view the hidden state h at time t-1 t-1 and the new information x t , and output a vector f between 0 and 1 t As the value of the forget gate at time t, it is used to determine the cell state c at time t-1 t-1 Information that needs to be retained; By analyzing the hidden state h at time t-1 t-1 and the new information x t , calculate the input gate i t As the value of the input gate at time t, the information that needs to be updated is determined based on this, and then the candidate information is calculated through the tanh function Use the forget gate to retain the cell state c at time t-1 according to the following formula t-1 Partial information, candidate information For the cell state c at time t t To update: The cell state c at time t t and the hidden state h at time t-1 t-1 Pass it to the output gate, and use the sigmoid function and tanh function, combined with parameter information, to determine the hidden state h at time t t , and output the predicted result P of the pressure at time t t .

2. A training method for a pipeline shut-down pressure prediction model, It is characterized in that include: Build a preset autoencoder network model; The neuron structure of the preset autoencoder network model considers time series characteristics during the updating process, and the parameter information of the preset autoencoder network model is determined based on a particle swarm optimization algorithm, and the parameter information includes: an input weight matrix, a recursive weight matrix, a hidden output weight matrix, and a bias vector; Acquiring pipeline stop information; the stop information includes a stop pressure and index data corresponding to the stop pressure, wherein the index data changes with the pipeline stop time; Inputting the indicator data into a preset autoencoding network model to obtain an output result; Calculating the error between the output result and the stop pressure; Correcting the preset autoencoder network model based on the error to obtain a pipeline shut-down pressure prediction model; The preset autoencoder network model is a long short-term memory neural network model, including: a forget gate, an input gate, and an output gate; Accordingly, the neuron structure of the preset autoencoder network model is updated in the following manner: At time t, the cell state c at time t-1 is transmitted to the neuron structure of the preset autoencoder network model. t-1 , hidden state h t-1 And the new information x t ; Use the sigmoid function of the forget gate to view the hidden state h at time t-1 t-1 and the new information x t , and output a vector f between 0 and 1 t As the value of the forget gate at time t, it is used to determine the cell state c at time t-1 t-1 Information that needs to be retained; By analyzing the hidden state h at time t-1 t-1 and the new information x t , calculate the input gate i t As the value of the input gate at time t, the information that needs to be updated is determined based on this, and then the candidate information is calculated through the tanh function Use the forget gate to retain the cell state c at time t-1 according to the following formula t-1 Partial information, candidate information For the cell state c at time t t To update: The cell state c at time t t and the hidden state h at time t-1 t-1 Pass it to the output gate, and use the sigmoid function and tanh function, combined with parameter information, to determine the hidden state h at time t t , and output the predicted result P of the pressure at time t t .

3. The method according to claim 2, It is characterized in that Before obtaining the pipeline stop information, the method includes: Determine the factors affecting the pipeline shut-down pressure based on the preset relationship between the oil product temperature and time after the pipeline shut-down, and the preset relationship between the pipeline pressure and the oil product temperature; Selecting the influencing factors that change with the pipeline shutdown time from the influencing factors of the pipeline shutdown pressure; The data corresponding to the influencing factors that change with the pipeline outage time are used as indicator data.

4. The method according to claim 3, It is characterized in that The factors affecting the change of pipeline stoppage time at least include the stoppage time, the ambient temperature, and the oil product temperature after the stoppage.

5. The method according to claim 3, It is characterized in that The preset relationship between the oil product temperature and time after the pipeline stops transporting oil, and the preset relationship between the pipeline pressure and the oil product temperature are: Among them, K 1 is the total heat transfer coefficient, D, D 1 , D 2 are the average diameter of the pipeline, the inner diameter of the pipeline and the outer diameter of the pipeline, respectively, y 、c g are the specific heat capacities of oil and steel, ρ y , g are the densities of oil and steel, Te t is the ambient temperature, T t T is the oil temperature after stopping the oil flow for t hours. 0 is the oil temperature at the beginning of the oil stoppage, t is the stoppage duration, α is the volume expansion coefficient of the pipeline, δ is the pipeline wall thickness, E is the Young's elastic modulus of the pipe, β is the volume expansion coefficient of the oil, K 2 is the elastic coefficient of the oil, T i is the oil temperature at the i-th moment, T i-1 is the oil temperature at the i-1th moment, and ΔP is the pressure difference between the i-th moment and the i-1th moment.

6. The method according to claim 2, It is characterized in that Before inputting the indicator data into the preset autoencoding network model, the method includes: Normalizing the stop-transmission pressure and the index data to obtain a normalized stop-transmission pressure and normalized index data; Inputting the normalized index data into a preset autoencoder network model to obtain an output result; The error between the output result and the normalized stop-transmission pressure is calculated.

7. A device for determining the stop pressure in a pipeline, It is characterized in that include: An index data acquisition module, used to acquire index data of a target pipeline, wherein the index data changes with the pipeline stoppage time; A pipeline pressure determination module is used to determine the pressure in the target pipeline at different stoppage durations according to the indicator data and the pipeline stoppage pressure prediction model; wherein the pipeline stoppage pressure prediction model is obtained by training a preset autoencoder network model based on the stoppage information, the neuron structure of the preset autoencoder network model considers the time series characteristics during the updating process, and the parameter information of the preset autoencoder network model is determined based on the particle swarm optimization algorithm, and the parameter information includes: input weight matrix, recursive weight matrix, hidden output weight matrix, and bias vector; The preset autoencoder network model is a long short-term memory neural network model, including: a forget gate, an input gate, and an output gate; Accordingly, the device is also used to update the neuron structure of the preset autoencoder network model in the following manner: At time t, the cell state c at time t-1 is transmitted to the neuron structure of the preset autoencoder network model. t-1 , hidden state h t-1 And the new information x t ; Use the sigmoid function of the forget gate to view the hidden state h at time t-1 t-1 and the new information x t , and output a vector f between 0 and 1 t As the value of the forget gate at time t, it is used to determine the cell state c at time t-1 t-1 Information that needs to be retained; By analyzing the hidden state h at time t-1 t-1 and the new information x t , calculate the input gate i t As the value of the input gate at time t, the information that needs to be updated is determined based on this, and then the candidate information is calculated through the tanh function Use the forget gate to retain the cell state c at time t-1 according to the following formula t-1 Partial information, candidate information For the cell state c at time t t To update: The cell state c at time t t and the hidden state h at time t-1 t-1 Pass it to the output gate, and use the sigmoid function and tanh function, combined with parameter information, to determine the hidden state h at time t t , and output the predicted result P of the pressure at time t t .

8. A training device for a pipeline shut-down pressure prediction model, It is characterized in that include: A construction module for building a preset autoencoder network model; The neuron structure of the preset autoencoder network model considers time series characteristics during the updating process, and the parameter information of the preset autoencoder network model is determined based on a particle swarm optimization algorithm, and the parameter information includes: an input weight matrix, a recursive weight matrix, a hidden output weight matrix, and a bias vector; An acquisition module, used to acquire pipeline stop information; the stop information includes a stop pressure and index data corresponding to the stop pressure, and the index data changes with the pipeline stop time; An acquisition module, used for inputting the indicator data into a preset autoencoding network model to obtain an output result; A calculation module, used for calculating the error between the output result and the stop pressure; A correction module, used to correct the preset autoencoder network model based on the error to obtain a pipeline shut-down pressure prediction model; The preset autoencoder network model is a long short-term memory neural network model, including: a forget gate, an input gate, and an output gate; Accordingly, the construction module updates the neuron structure of the preset autoencoder network model in the following manner: At time t, the cell state c at time t-1 is transmitted to the neuron structure of the preset autoencoder network model. t-1 , hidden state h t-1 And the new information x t ; Use the sigmoid function of the forget gate to view the hidden state h at time t-1 t-1 and the new information x t , and output a vector f between 0 and 1 t As the value of the forget gate at time t, it is used to determine the cell state c at time t-1 t-1 Information that needs to be retained; By analyzing the hidden state h at time t-1 t-1 and the new information x t , calculate the input gate i t As the value of the input gate at time t, the information that needs to be updated is determined based on this, and then the candidate information is calculated through the tanh function Use the forget gate to retain the cell state c at time t-1 according to the following formula t-1 Partial information, candidate information For the cell state c at time t t To update: The cell state c at time t t and the hidden state h at time t-1 t-1 Pass it to the output gate, and use the sigmoid function and tanh function, combined with parameter information, to determine the hidden state h at time t t , and output the predicted result P of the pressure at time t t .

9. A device for determining the stop pressure in a pipeline. It is characterized in that The method comprises a processor and a memory for storing processor-executable instructions, wherein the processor implements the steps of the method according to any one of claims 1 to 6 when executing the instructions.

10. A computer-readable storage medium, It is characterized in that Computer instructions are stored thereon, and when the instructions are executed, the steps of the method according to any one of claims 1 to 6 are implemented.

Images