Pigging period monitoring method and system based on pipeline effusion and hydrate

Abstract

The invention discloses a pigging period monitoring method and system based on pipeline effusion and hydrate, and the method comprises the steps: collecting data, and carrying out the preprocessing of the data, and obtaining a data set; performing multi-dimensional analysis on the data set, and extracting data features; constructing a pigging period prediction model based on the data features; acquiring historical pigging data to train the pigging period prediction model; and constructing an expert scoring model to obtain an expert score, verifying and optimizing the trained pigging prediction model based on the expert score, and performing pigging real-time monitoring based on the optimized pigging period prediction model. The pipeline cleaning data is monitored in real time through the pipeline cleaning period prediction model to obtain task feedback data, the pipeline cleaning prediction model is verified and optimized in combination with expert scores, pipeline cleaning management is carried out based on the optimized pipeline cleaning period prediction model, monitoring and safety evaluation on the health state of a pipeline are more facilitated, and the pipeline cleaning efficiency is improved. And the accurate prediction period of the pipe cleaning operation is given.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of pipeline application, in particular to a pigging cycle monitoring method and system based on pipeline liquid accumulation and hydrate. BACKGROUND

[0002] During gas transmission, due to the existence of pipeline elevation difference, pipeline undulation, temperature change and other conditions, liquid accumulation is easy to form in the pipeline, and the existence of liquid accumulation will have adverse effects on pipeline transmission. In order to solve the problem of pipeline liquid accumulation, regular pipeline pigging operation is the first choice. However, if the pigging operation is set too frequently, not only the cost of the gas pipeline in the running process will increase, but also certain operation risks may be caused; and if the pigging cycle is set too long, the pigging effect will be greatly weakened.

[0003] At present, pipeline pigging operation of each oil and gas mine is faced with complex working conditions such as "upstream and downstream linkage", the flow and pressure of the pipeline change frequently, and the pigging operation of a single pipeline also needs to fully consider the overall pipeline pigging work arrangement and the pigging execution plan of the upstream and downstream, so the existing indicators cannot support the accuracy of the pigging prediction for the overall pipeline under complex working conditions. The model needs to continuously enrich the relevant prediction parameters to adapt to the field working condition data, and needs to use the "expert model + data model" method to repeatedly iterate and optimize, dynamically simulate and calculate the characteristic parameters of each pipeline, predict the running trend data, set different training sets and parameters for simulation calculation and trend analysis, optimize the calculation methods of elevation difference, hydrate and the like, and further improve the accuracy of the model. The accurate management of "one line one model" is realized.

[0004] However, with the continuous iteration and development of the model, each pipeline has its own specific model, and from model development to management and monitoring, a large amount of manpower and time is needed, and various compliance and safety risks exist. How to efficiently manage and monitor the overall model system has become a challenge. SUMMARY

[0005] The technical problem to be solved by the present application is that the pipeline pigging process is complicated due to liquid accumulation and hydrate in the pipeline, and the existing technology cannot accurately control the pigging process. The purpose is to provide a pigging cycle monitoring method and system based on pipeline liquid accumulation and hydrate, to obtain task feedback data by constructing a pigging cycle prediction model and performing real-time monitoring on the pigging data by the pigging cycle prediction model; to perform data backtracking according to the task feedback data, to verify and optimize the pigging prediction model in combination with expert scoring, to perform pigging management based on the optimized pigging cycle prediction model, and to be more conducive to monitoring and safety evaluation of the health status of the pipeline, and to give an accurate prediction cycle of pigging operation.

[0006] The present application is realized by the following technical solutions:

[0007] The first aspect of the application provides a pigging cycle monitoring method based on pipeline liquid accumulation and hydrate, characterized in that it comprises the following specific steps:

[0008] Collecting data, pre-processing the data to obtain a data set;

[0009] Performing multi-dimensional analysis on the data set to extract data features;

[0010] Building a pigging cycle prediction model based on the data features;

[0011] Training the pigging cycle prediction model by obtaining historical pigging data;

[0012] Building an expert scoring model to obtain expert scores, verifying and optimizing the trained pigging prediction model based on the expert scores, and performing real-time monitoring of pigging based on the optimized pigging cycle prediction model.

[0013] The application builds a pigging cycle prediction model to obtain task feedback data from real-time monitoring of pigging data by the pigging cycle prediction model, performs data backtracking according to the task feedback data, verifies and optimizes the pigging prediction model in combination with expert scores, and performs pigging management based on the optimized pigging cycle prediction model, which is more conducive to monitoring and safety evaluation of the health status of the pipeline and giving an accurate prediction cycle of pigging operations.

[0014] Further, the data preprocessing comprises:

[0015] Cleaning the data to correct errors, missing or repeated parts in the data;

[0016] Performing institutionalized processing on the cleaned data to obtain a formatted data set.

[0017] Further, the data set comprises pipeline data, dynamic data and pigging operation data;

[0018] The pipeline data comprises static data such as pipeline type, pipeline length, pipe diameter, pipeline elevation and surveying and mapping data;

[0019] The dynamic data comprises pipeline operation real-time data such as starting point pressure, ending point pressure, instantaneous flow, measurement temperature and derived data such as pipeline transportation efficiency and friction coefficient;

[0020] The pigging operation data collects pigging operation records of previous years, including launch time, receive time and interference amount data.

[0021] Further, the data feature extraction specifically comprises:

[0022] Obtaining static parameters, dynamic parameters and simulation software simulation data of hydrate and liquid accumulation through the pipeline;

[0023] According to the previous pigging operation, the pigging characteristic data is extracted, and the data column affecting the pigging characteristics is selected as the characteristic column.

[0024] Further, the simulation software simulation data of hydrate and liquid accumulation includes:

[0025] The minimum pressure drop model, droplet model and liquid film model are constructed to obtain the conditions of liquid accumulation, and the mechanism model is constructed to simulate the gas-liquid two-phase flow data in the pipeline, so as to obtain the simulation software simulation data of hydrate and liquid accumulation.

[0026] Further, the combination of data backtracking for verifying and optimizing the pigging prediction model specifically includes:

[0027] The pigging prediction result is verified, and after the pigging operation prediction message is pushed, it is judged whether the prediction result is effective or not.

[0028] Based on the effective prediction result, the pigging prediction model is iteratively optimized, and the invalid pigging suggestion is eliminated.

[0029] The second aspect of the application provides a system applied to a pipeline liquid accumulation and hydrate based pigging cycle monitoring method, including a data layer, a platform layer and a terminal layer;

[0030] The data layer is used for pipeline production data acquisition service;

[0031] The platform layer is used for processing the basic data required by the pipeline pigging cycle prediction model, including data acquisition, data analysis, model management, pigging cycle prediction and verification optimization data; the platform layer is also used for pipeline data query, model query and pipeline production data dynamic monitoring and analysis;

[0032] The terminal layer is used for completing data monitoring, task feedback, data processing and data analysis through the production network.

[0033] Further, it also includes a pipeline management module, a model management module, a pipeline monitoring module and a task management module;

[0034] The pipeline management module is used for providing and maintaining the data required by the pigging cycle prediction process;

[0035] The model management module is used for realizing the visual management of the whole life cycle of pigging prediction model training to verification;

[0036] The pipeline monitoring module is used for integrating the pipeline map, and monitoring the pipeline efficiency, key parameters and pigging early warning in real time according to the pipeline map;

[0037] The task management module is used for repeatedly iterating and optimizing the prediction model according to the pigging cycle prediction and real-time monitoring data.

[0038] Further, in the pipeline management module, the data required by the pigging cycle prediction process includes:

[0039] Pipeline data, map configuration data, elevation data and dynamic data;

[0040] The pipeline data includes pipeline ID, management unit ID, pipeline name, thickness, outer diameter, material, length, transmission medium, starting point station and ending point station;

[0041] The map configuration data includes pipeline name, longitude and latitude;

[0042] The elevation data includes pipeline elevation;

[0043] The dynamic data includes temperature, flow, transmission difference and pressure.

[0044] Further, the model management module includes a model creation module, a model training module, a model verification module and a model release module;

[0045] The model creation module is used for creating a new pigging cycle prediction model;

[0046] The model training module is used for setting time column, target column and other training parameters according to the selected data set, and training the model;

[0047] The model verification module is used for verifying the accuracy of the prediction of the model;

[0048] The model release module is used for pigging cycle prediction of the verified model.

[0049] Compared with the prior art, the present application has the following advantages and beneficial effects:

[0050] By constructing the pigging cycle prediction model, the task feedback data obtained by the real-time monitoring of the pigging data by the pigging cycle prediction model is obtained; the data is traced back according to the task feedback data, the pigging prediction model is verified and optimized in combination with expert scoring, the pigging management is carried out based on the optimized pigging cycle prediction model, and it is more beneficial to realize the monitoring and safety evaluation of the health state of the pipeline, and the prediction cycle of the accurate pigging operation is given. BRIEF DESCRIPTION OF DRAWINGS

[0051] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments. It should be understood that the following drawings only show some of the embodiments of the present application, and therefore should not be considered as a limitation to the scope. Other related drawings can also be obtained by those skilled in the art without any creative effort, based on these drawings. In the drawings:

[0052] Figure 1 The flow chart in the embodiment of the present application;

[0053] Figure 2 The frame chart of the system in the embodiment of the present application. DETAILED DESCRIPTION

[0054] In order to make the purpose, technical solutions and advantages of the present application more clear and understandable, the following will further describe the present application in detail with embodiments and drawings. The exemplary embodiments of the present application and their descriptions are only used to explain the present application, and should not be considered as a limitation to the present application.

[0055] As a possible embodiment, as shown in Figure 1 The embodiment provides a pigging cycle monitoring method based on pipeline liquid accumulation and hydrate, which comprises the following specific steps: collecting data, pre-processing the data to obtain a data set; performing multi-dimensional analysis on the data set to extract data features; constructing a pigging cycle prediction model based on the data features; obtaining historical pigging data to train the pigging cycle prediction model; constructing an expert scoring model to obtain expert scores, verifying and optimizing the trained pigging prediction model based on the expert scores, and performing real-time monitoring of pigging based on the optimized pigging cycle prediction model. The task feedback data obtained by the pigging cycle prediction model for real-time monitoring of pigging data, in combination with the expert scores, is used to verify and optimize the pigging prediction model, and the pigging management is performed based on the optimized pigging cycle prediction model, which is more conducive to monitoring and safety evaluation of the health status of the pipeline, and giving an accurate prediction cycle of pigging operation.

[0056] In a possible embodiment, the pigging cycle monitoring method based on pipeline liquid accumulation and hydrate provided by the embodiment can be applied to the field of natural gas gathering and transportation.

[0057] In a possible embodiment, the implementation path of the pigging cycle monitoring method based on pipeline liquid accumulation and hydrate provided by the embodiment comprises:

[0058] Data integration -> data preprocessing -> establishment of prediction model -> real-time monitoring -> pigging suggestion -> pigging prediction result verification -> model iteration optimization.

[0059] In a possible implementation, the data preprocessing includes: cleaning the data, correcting errors, missing or repeated parts in the data, performing data quality assessment and data cleaning operation to ensure the accuracy of the data. The cleaned data is processed in an institutionalized manner to obtain a formatted data set to facilitate subsequent analysis, which can include formatting dates and times, standardizing numerical values, converting data types, etc.

[0060] In a possible implementation, the data set includes pipeline data, dynamic data and pigging operation data; the pipeline data includes static data such as pipeline type, pipeline length, pipe diameter, pipeline elevation and surveying and mapping data, which are mainly obtained from the regional lake master data interface, and the system can be expanded for properties not covered by the master data; the dynamic data includes pipeline operation real-time data such as starting point pressure, ending point pressure, instantaneous flow, metering temperature and derived data such as pipeline transportation efficiency and friction coefficient; the pigging operation data collects pigging operation records of previous years including data such as ball launch time, ball collection time and excess amount.

[0061] In a possible implementation, since the dependent variable characteristics of pigging cannot be directly reflected, static parameters such as the diameter, length and elevation difference of the pipeline through which the pig passes, as well as real-time data such as flow, pressure difference and temperature difference, are determined, combined with pipeline transportation efficiency, simulation data of hydrate and liquid accumulation by simulation software, and pigging characteristic data is extracted according to previous pigging operation conditions. Specifically, static parameters, dynamic parameters and simulation data of hydrate and liquid accumulation by simulation software are obtained; pigging characteristic data is extracted according to previous pigging operation conditions, and data columns that affect pigging characteristics are selected as characteristic columns.

[0062] In a possible implementation, the simulation data of hydrate and liquid accumulation by simulation software includes:

[0063] A minimum pressure drop model, a liquid drop model and a liquid film model are constructed to obtain the conditions for the formation of liquid accumulation, a mechanism model is constructed to simulate the gas-liquid two-phase flow data in the pipeline, and simulation data of hydrate and liquid accumulation by simulation software is obtained. The pipeline health degree is calculated by weighting the three predicted values after predicting the amount of hydrate and liquid accumulation and the pipeline transportation efficiency, so that the pigging plan can be made.

[0064] In a possible implementation, the expert scoring model includes pipeline transportation efficiency, pressure difference change trend, upstream and downstream linkage pigging experience and pipeline health index evaluation.

[0065] In a possible implementation, the pigging prediction model is verified and optimized in combination with data backtracking, which specifically includes:

[0066] The pigging prediction result is verified, and after the prediction message is pushed in each pigging operation, it is determined whether the prediction result is valid.

[0067] Based on the effective prediction results, the pigging prediction model is iteratively optimized, and invalid pigging suggestions are eliminated.

[0068] When the model is iteratively optimized, the pigging model optimizes the pigging suggestions fed back to the operating personnel, and the data before and after the effective pigging times are added to the model for optimization every day, and invalid pigging suggestions are eliminated. The model is verified by historical effective pigging every month to improve the accuracy of the model.

[0069] In a possible implementation, after the pigging prediction model is verified and optimized, the real-time operating condition data (temperature, inlet pressure, outlet pressure, instantaneous flow) of the gathering pipeline are monitored through the pSpace database, and the pipeline operating data in the past 2 hours are extracted according to the data structure of the training set and input into the model to obtain the pipeline health degree of the pipeline during this period. According to the set threshold, it is judged whether to pig or not, and the corresponding management personnel are pushed, and the subsequent feedback to the suggestion is tracked.

[0070] The pigging prediction result verification is that the pigging operating personnel manually judges whether the prediction result of each pigging operation is effective after the prediction message is pushed, and the system is fed back.

[0071] As a possible implementation, as shown in Figure 2 The embodiment provides a pigging cycle monitoring system based on pipeline liquid accumulation and hydrate, which comprises a data layer, a platform layer and a terminal layer.

[0072] The data layer is used for pipeline production data collection service;

[0073] The platform layer is used for processing of basic data required by the pipeline pigging cycle prediction model, including data collection, data analysis, model management, pigging cycle prediction and verification optimization data; the platform layer is also used for pipeline data query, model query and dynamic monitoring and analysis of pipeline production data;

[0074] The terminal layer is used for completing data monitoring, task feedback, data processing and data analysis through a production network.

[0075] The embodiment builds a system covering the whole life cycle of pipeline pigging operation from the perspective of pipeline hydrate and liquid accumulation prediction, which comprises a data layer, a platform layer and a terminal layer, realizes effective prediction of pipeline operation cycle and visualized unified management of the prediction model, and evaluates the health index of the pipeline.

[0076] In a possible implementation, the system further comprises a pipeline management module, a model management module, a pipeline monitoring module and a task management module.

[0077] In a possible implementation, the pipeline management module is configured to provide and maintain data required by the pigging cycle prediction process; the required data is synchronized from the master data platform to ensure data consistency, and can also be customized with mine-specific fields.

[0078] In the pipeline management module, the data required by the pigging cycle prediction process includes:

[0079] Pipeline data, map configuration data, elevation data, and dynamic data;

[0080] The pipeline data includes pipeline ID, management unit ID, pipeline name, thickness, outer diameter, material, length, transmission medium, starting point station, and ending point station.

[0081] The map configuration data includes pipeline name, longitude, and latitude.

[0082] The elevation data includes pipeline elevation.

[0083] The dynamic data includes temperature, flow, transmission difference, and pressure.

[0084] In a possible implementation, the model management module includes a model creation module, a model training module, a model verification module, and a model release module.

[0085] The model creation module is configured to create a new pigging cycle prediction model.

[0086] The model training module is configured to set time column, target column, and other training parameters according to the selected data set, and train the model.

[0087] The model verification module is configured to verify the accuracy of the model prediction.

[0088] The model release module is configured to perform pigging cycle prediction on the verified model.

[0089] The model management module is configured to realize visual management of the whole life cycle of the pigging prediction model from training to verification.

[0090] In a possible implementation, the pipeline monitoring module is configured to integrate a pipeline map, and perform real-time monitoring on pipeline transmission efficiency, key parameters, and pigging early warning reminders according to the pipeline map.

[0091] When real-time monitoring is performed, how to perform pigging monitoring needs to be analyzed by big data analysis technology, specifically including any one or a combination of at least two of an artificial intelligence algorithm engine, deep learning, a weighted average method, a D-S argument theory or a rough set theory, typical but non-limiting combinations including a combination of an artificial intelligence algorithm engine, deep learning, a weighted average method and a D-S argument theory, a combination of an artificial intelligence algorithm engine, deep learning and a weighted average method, a combination of deep learning, a weighted average method, a D-S argument theory and a rough set theory, or a combination of an artificial intelligence algorithm engine, deep learning, a weighted average method, a D-S argument theory and a rough set theory.

[0092] In one possible implementation, task management mainly realizes closed-loop management of pigging operations and models. After the model is applied, according to pigging cycle prediction and real-time monitoring data, it is helpful to more scientifically formulate a pigging operation plan, and through repeated iteration optimization of the prediction model, the accuracy and rationality of the model are improved.

[0093] The above specific embodiments further specifically describe the purposes, technical solutions and beneficial effects of the present application. It should be understood that the above description is only a specific embodiment of the present application and is not used to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the protection scope of the present application.

Claims

1. A method for monitoring pipeline cleaning cycles based on accumulated liquid and hydrates, characterized in that, The specific steps include the following: Collect data, preprocess the data, and obtain a dataset; Perform multidimensional analysis on the dataset to extract data features; A cleaning cycle prediction model is constructed based on data characteristics; Historical cleaning data was used to train a cleaning cycle prediction model; An expert scoring model was constructed to obtain expert scores. The trained cleaning cycle prediction model was then validated and optimized based on the expert scores. Real-time cleaning monitoring was then conducted based on the optimized cleaning cycle prediction model.

2. The method for monitoring pipeline cleaning cycles based on pipeline effluent and hydrates according to claim 1, characterized in that, The data preprocessing includes: Clean the data and correct any errors, missing or duplicate parts. The cleaned data is then structured to obtain a formatted dataset.

3. The method for monitoring pipeline cleaning cycles based on pipeline effluent and hydrates according to claim 1 or 2, characterized in that, The dataset includes pipeline data, dynamic data, and pipeline cleaning operation data; The pipeline data includes static data such as pipeline type, pipeline length, pipe diameter, pipeline elevation, and survey data; The dynamic data includes real-time pipeline operation data such as starting pressure, ending pressure, instantaneous flow rate, metering temperature, and derived data such as pipeline efficiency and friction coefficient. The data collection for the cleaning operation includes historical cleaning operation records, such as ball launch time, ball return time, and interference data.

4. The method for monitoring pipeline cleaning cycles based on pipeline liquid and hydrate as described in claim 1, characterized in that, The specific steps for extracting data features include: Obtain simulation data of hydrates and accumulated liquid through static and dynamic parameters of pipelines and simulation software; Based on the data from previous cleaning operations, cleaning characteristic data were extracted, and data columns that affect cleaning characteristics were selected as feature columns.

5. The method for monitoring pipeline cleaning cycles based on pipeline effluent and hydrates according to claim 4, characterized in that, The steps for obtaining simulation data on hydrates and accumulated liquids by the simulation software include: A minimum pressure drop model, a droplet model, and a liquid film model were constructed to obtain the conditions for liquid accumulation. A mechanistic model was then constructed to simulate the gas-liquid two-phase flow data in the pipeline, and simulation data of hydrates and liquid accumulation were obtained from the simulation software.

6. The method for monitoring pipeline cleaning cycles based on pipeline effluent and hydrates according to claim 1, characterized in that, The verification and optimization of the pigging prediction model by combining data backtracking specifically includes: Verification of the cleaning and drainage prediction results: After each cleaning and drainage operation prediction message is pushed, it is determined whether the prediction result is valid. The cleaning prediction model is iteratively optimized based on effective prediction results, and invalid cleaning suggestions are eliminated.

7. A system for monitoring the pipeline cleaning cycle based on pipeline effluent and hydrate as described in any one of claims 1-6, characterized in that, It includes the data layer, platform layer, and terminal layer; The data layer is used for pipeline production data acquisition services; The platform layer is used for processing the basic data required by the pipeline cleaning cycle prediction model, including data acquisition, data analysis, model management, cleaning cycle prediction, and verification and optimization data; the platform layer is also used for pipeline data query, model query, and dynamic monitoring and analysis of pipeline production data; The terminal layer is used to complete data monitoring, task feedback, data processing, and data analysis through the production network.

8. The pipeline cleaning cycle monitoring system based on pipeline liquid and hydrate as described in claim 7, characterized in that, It also includes a pipeline management module, a model management module, a pipeline monitoring module, and a task management module; The pipeline management module is used to provide and maintain the data required for the pipeline cleaning cycle prediction process; The model management module is used to realize visualized management of the entire lifecycle of the pigging prediction model, from training to validation. The pipeline monitoring module is used to integrate pipeline maps and monitor pipeline transportation efficiency, key parameters, and cleaning early warning reminders in real time based on the pipeline maps. The task management module is used to iteratively optimize the prediction model based on the cleaning cycle prediction and real-time monitoring data.

9. The pipeline cleaning cycle monitoring system based on pipeline liquid and hydrate as described in claim 8, characterized in that, The data required for the pipeline cleaning cycle prediction process in the pipeline management module includes: Pipeline data, map configuration data, elevation data, and dynamic data; The pipeline data includes pipeline ID, management unit ID, pipeline name, thickness, outer diameter, material, length, transmission medium, starting station, and ending station; The map configuration data includes pipeline name, longitude, and latitude; The elevation data includes pipeline elevations; The dynamic data includes temperature, flow rate, differential pressure, and pressure.

10. The pipeline cleaning cycle monitoring system based on pipeline liquid and hydrate as described in claim 8, characterized in that, The model management module includes a model creation module, a model training module, a model validation module, and a model publishing module; The model creation module is used to create a new cleaning cycle prediction model; The model training module is used to train the model by setting the time column, target column and other training parameters according to the selected dataset; The model validation module is used to validate the accuracy of the model's predictions; The model publishing module is used to predict the cleaning cycle of the validated model.

Images