CN-122021981-A - Working condition prediction and evaluation method and system for deep sea drilling fluid lifting system

CN122021981ACN 122021981 ACN122021981 ACN 122021981ACN-122021981-A

Abstract

The application discloses a working condition prediction evaluation method and a working condition prediction evaluation system of a deep sea drilling fluid lifting system, and relates to the technical field of marine engineering equipment and working condition prediction evaluation; the method comprises the steps of building a machine learning agent model, training, predicting working conditions in real time, optimizing parameters, outputting a working condition predicting result, a frequency recommending result and/or a capacity boundary analyzing result in real time, wherein the working condition predicting result comprises a predicted flow and a predicted lift corresponding to various power supply frequencies and pump configuration combinations, the frequency recommending result comprises an optimal power supply frequency which meets a preset working target set by a user and meets system safety constraints, and the capacity boundary analyzing result comprises the lifting capacity of a deep sea drilling fluid lifting system under the limit condition. The application can rapidly and accurately predict the operation working condition points (flow and lift) of the double-pump serial/parallel configuration under different power supply frequencies based on the drilling fluid density which changes in real time, and can intelligently recommend the optimal power supply frequency.

Inventors

QIN RULEI
XIE WENWEI
HE GUOLEI
YU YANJIANG
CHEN HAOWEN
YIN GUOLE
XU BENCHONG

Assignees

中国地质科学院勘探技术研究所

Dates

Publication Date: 20260512
Application Date: 20250904

Claims (10)

1. The working condition prediction and evaluation method for the deep sea drilling fluid lifting system is characterized by comprising the following steps of: Acquiring historical working condition data of a deep sea drilling fluid lifting system, and generating a working condition point sample database based on a physical model; constructing a machine learning agent model, and training the machine learning agent model by utilizing the working point samples in the working point sample database to obtain a trained machine learning agent model, wherein the machine learning agent model is a model which takes drilling fluid density, power supply frequency and pump configuration mode as input and takes predicted flow and predicted lift under corresponding working conditions as output for predicting the working conditions of a deep sea drilling fluid lifting system; The method comprises the steps of collecting current drilling fluid density in real time, carrying out real-time working condition prediction and parameter optimization by utilizing a trained machine learning agent model according to the current drilling fluid density, and outputting real-time working condition prediction results, frequency recommendation results and/or capability boundary analysis results, wherein the working condition prediction results comprise prediction flow and prediction lift corresponding to various power supply frequencies and pump configuration combinations, the frequency recommendation results comprise optimal power supply frequencies which meet preset operation targets set by users and meet system safety constraints, and the capability boundary analysis results comprise lifting capability of a deep sea drilling fluid lifting system under limit conditions.
2. The method for predicting and evaluating the working condition of the deep sea drilling fluid lifting system according to claim 1, wherein the method for predicting and evaluating the working condition of the deep sea drilling fluid lifting system is characterized by obtaining historical working condition data of the deep sea drilling fluid lifting system and generating a working condition point sample database based on a physical model, and specifically comprises the following steps: Acquiring historical working condition data of a deep sea drilling fluid lifting system; Based on historical working condition data of the deep sea drilling fluid lifting system, constructing a hydraulic characteristic model of a pipeline system, calculating the along-distance friction pressure loss by adopting a Darcy-Weissebach formula, calculating a friction coefficient by adopting a Haaland formula, and obtaining a pipeline total lift demand curve by combining an effective static pressure head caused by the density difference of the sea water and the drilling fluid in the pipeline; based on historical working condition data of the deep sea drilling fluid lifting system, constructing a double-pump lifting unit performance model, acquiring equivalent performance curves of a single pump under different power supply frequencies, and establishing equivalent performance curves under a double-pump serial and parallel configuration according to the rules of superposition of the same flow under the same flow and superposition of the same flow under parallel connection during serial connection; And solving the intersection point of the pipeline total lift demand curve and the equivalent performance curve under the serial and parallel configuration of the double pumps by adopting a numerical method aiming at each group of parameter combinations in a preset parameter space to obtain balanced flow and lift under the corresponding working condition and generate a working condition point sample database, wherein the parameter combinations comprise drilling fluid density, power supply frequency and pump configuration modes.
3. The method for predicting and evaluating the working condition of a deep sea drilling fluid lifting system according to claim 2, wherein the expression of the pipeline total lift demand curve is: H system (Q)＝H f +H static (ρ); Wherein H system represents a pipeline total head demand curve, H f represents along-distance friction pressure consumption, H static represents an effective static pressure head caused by the density difference of sea water and drilling fluid in a pipe, Q is flow, and ρ is drilling fluid density.
4. The method for predicting and evaluating the working condition of the deep sea drilling fluid lifting system according to claim 2, wherein the expression of the equivalent performance curve of the dual pump series and parallel configuration is: tandem configuration, namely, the lift superposition is expressed as: H series (Q)＝H pump1 (Q)+H pump2 (Q); Wherein H series (Q) is the pump lift under the double-pump serial configuration, and H pump1 (Q) and H pump2 (Q) are the respective lifts of the first pump and the second pump under the double-pump serial configuration respectively; parallel configuration, namely, under the same lift, flow superposition is expressed as: Q parallel (H)＝Q pump1 (H)+Q pump2 (H); Wherein, Q parallel (H) is the flow rate in the dual pump parallel configuration, and Q pump1 (H) and Q pump2 (H) are the flow rates of the first pump and the second pump in the dual pump parallel configuration, respectively.
5. The method for predicting and evaluating the working condition of a deep sea drilling fluid lifting system according to claim 2, wherein the intersection point of the pipeline total lift demand curve and the equivalent performance curve of the double pump in series and parallel configuration is calculated by adopting the following steps: H pump_effective (Q,f freq ,Config)＝H system (Q,ρ); Wherein, H pump_effective (Q,f freq , config) is the lift/flow supply under the cooperation of multiple pumps, H system represents the total lift demand curve of the pipeline, Q is the flow, ρ is the drilling fluid density, config represents the pump configuration mode, and f freq is the power supply frequency.
6. The method for predicting and evaluating the working condition of the deep sea drilling fluid lifting system according to claim 1, wherein a machine learning agent model is constructed, and the machine learning agent model is trained by using working condition point samples in the working condition point sample database, so as to obtain a trained machine learning agent model, and the method specifically comprises the following steps: Dividing the working condition point sample database into a training set and a testing set; Respectively carrying out standardization processing on continuous numerical characteristics in the training set and the testing set, and carrying out independent heat coding processing on classified characteristics in the training set and the testing set to obtain a preprocessed training set and a preprocessed testing set, wherein the continuous numerical characteristics comprise drilling fluid density and power supply frequency, and the classified characteristics comprise a pump configuration mode; Constructing a machine learning agent model, taking drilling fluid density, power supply frequency and pump configuration mode in the preprocessed training set as input, and taking predicted flow and predicted lift under corresponding working conditions as output, and training the machine learning agent model to obtain a trained model; and performing performance evaluation on the trained model by using the preprocessed test set, and taking the model with the optimal performance as a trained machine learning agent model.
7. The method for predicting and evaluating the working condition of the deep sea drilling fluid lifting system according to claim 1, wherein the method is characterized by collecting the current drilling fluid density in real time, and according to the current drilling fluid density, performing real-time working condition prediction and parameter optimization by using the trained machine learning agent model, and outputting a real-time working condition prediction result, and specifically comprises the following steps: Collecting the current drilling fluid density in real time; Based on the current drilling fluid density, traversing two pump configuration modes of all preset power supply frequencies and double pump serial configuration and double pump parallel configuration, respectively inputting the current drilling fluid density, each power supply frequency and each pump configuration mode into the trained machine learning agent model to obtain corresponding predicted flow and predicted lift under various power supply frequencies and pump configuration combinations, and taking the predicted flow and the predicted lift as the working condition prediction results.
8. The method for predicting and evaluating the working condition of a deep sea drilling fluid lifting system according to claim 7, wherein after the steps of collecting the current drilling fluid density in real time, predicting the working condition in real time and optimizing parameters by using the trained machine learning agent model according to the current drilling fluid density, and outputting the real-time working condition prediction result, the method for predicting and evaluating the working condition of the deep sea drilling fluid lifting system further comprises: acquiring a preset operation target set by a user, wherein the preset operation target comprises a target flow and/or a target lift; Screening power supply frequency and pump configuration combinations meeting system safety constraints in the working condition prediction result according to the preset working target, wherein the system safety constraints comprise the highest rotating speed of the pump, the lowest rotating speed of the pump and the maximum power of the pump; Taking the power supply frequency meeting the system safety constraint and the power supply frequency corresponding to the pump configuration combination as the optimal power supply frequency to obtain a frequency recommendation result; and recommending the optimal power supply frequency in the frequency recommendation result to a user.
9. The method for predicting and evaluating the working condition of a deep sea drilling fluid lifting system according to claim 7, wherein after the steps of collecting the current drilling fluid density in real time, predicting the working condition in real time and optimizing parameters by using the trained machine learning agent model according to the current drilling fluid density, and outputting the real-time working condition prediction result, the method for predicting and evaluating the working condition of the deep sea drilling fluid lifting system further comprises: Fixing the pump configuration mode and the power supply frequency, gradually adjusting the drilling fluid density, calling the trained machine learning agent model, and respectively inputting the fixed pump configuration mode, the fixed power supply frequency and the drilling fluid densities into the trained machine learning agent model to obtain the predicted flow and the predicted lift under corresponding working conditions; and determining the maximum drilling fluid density which can be processed by the deep sea drilling fluid lifting system under corresponding configuration according to the predicted flow, the predicted lift, the drilling fluid density, the pump configuration mode, the power supply frequency and the pump physical limit parameters.
10. A deep sea drilling fluid lifting system working condition prediction evaluation system, characterized in that the deep sea drilling fluid lifting system working condition prediction evaluation system applies the deep sea drilling fluid lifting system working condition prediction evaluation method according to any one of claims 1-9, and the deep sea drilling fluid lifting system working condition prediction evaluation system comprises: The data generation module is used for acquiring historical working condition data of the deep sea drilling fluid lifting system and generating a working condition point sample database based on a physical model; The model training module is used for constructing a machine learning agent model, training the machine learning agent model by utilizing the working point samples in the working point sample database to obtain a trained machine learning agent model, wherein the machine learning agent model is a model which takes drilling fluid density, power supply frequency and pump configuration mode as input and takes predicted flow and predicted lift under corresponding working conditions as output and is used for predicting the working conditions of a deep sea drilling fluid lifting system; the real-time prediction and optimization module is used for collecting current drilling fluid density in real time, performing real-time working condition prediction and parameter optimization by utilizing the trained machine learning agent model according to the current drilling fluid density, and outputting a real-time working condition prediction result, a frequency recommendation result and/or a capacity boundary analysis result, wherein the working condition prediction result comprises a predicted flow and a predicted lift corresponding to various power supply frequencies and pump configuration combinations, the frequency recommendation result comprises an optimal power supply frequency which accords with a preset working target set by a user and meets system safety constraint, and the capacity boundary analysis result comprises the lifting capacity of the deep sea drilling fluid lifting system under a limit condition.

Description

Working condition prediction and evaluation method and system for deep sea drilling fluid lifting system Technical Field The application relates to the technical field of marine engineering equipment and working condition prediction and evaluation, in particular to a working condition prediction and evaluation method and system of a deep sea drilling fluid lifting system. Background In deep sea oil and gas exploration and development, a drilling fluid lifting system is core equipment for maintaining stable wellbore pressure and realizing safe and efficient return of drilling cuttings. In order to cope with complex and changeable working conditions in a deep water environment, the drilling fluid lifting system has become a key component of advanced drilling processes such as the drilling fluid lifting system and the like because the drilling fluid lifting system can provide higher lift, larger flow and redundant backup of the system. The double-pump lifting unit can be operated in a serial connection mode or a parallel connection mode so as to adapt to different operation requirements. Currently, the determination of the operating point (i.e., actual flow and head) of a dual pump lift unit on an engineering site is largely dependent on conventional "pump-line curve" mapping methods. According to the method, an equivalent performance curve of the pump at a specific rotating speed and a resistance curve of a pipeline system are manually or semi-manually drawn on a coordinate graph, and the intersection point of the two curves is a stable working condition point of the system. However, the prior art suffers from the following significant drawbacks: (1) The response is lag, and the real-time regulation and control are not possible. In the deep sea drilling process, rheological parameters such as density, viscosity and the like of the drilling fluid can be dynamically changed. After each change, the characteristic curve of the pipeline is changed, the intersection point is required to be drawn again, the calculation process is complex and time-consuming, and the requirements of on-site 'instant prediction and real-time regulation' on the working condition cannot be met. (2) The calculated amount is large, and the coverage working condition is limited. When two configurations of double pump series/parallel connection and variable frequency adjustment (usually, the power supply frequency is continuously or discretely adjustable in a certain range) are considered, the number of working condition combinations to be analyzed is huge. The traditional graphic method is difficult to quickly and comprehensively evaluate all potential operation schemes, and limits the operation optimization potential of the system. (3) The accuracy is not enough due to the dependence of manpower. The graph method relies on manual graph reading and interpretation, subjective errors are easy to introduce, complex nonlinear relations are difficult to process, prediction accuracy is low, and reliability of operation decision is affected. Therefore, a technical solution is needed that can rapidly, accurately and intelligently predict the working states of a drilling fluid lifting system under different drilling fluid densities, different operation configurations and different power supply frequencies, and can provide optimal decision support. Disclosure of Invention The application aims to provide a working condition prediction evaluation method and system for a deep sea drilling fluid lifting system, which can rapidly and accurately predict operation working condition points (flow and lift) of double pumps in series/parallel configuration under different power supply frequencies based on drilling fluid density changing in real time and can intelligently recommend optimal power supply frequency. In order to achieve the above object, the present application provides the following. In a first aspect, the application provides a working condition prediction and evaluation method for a deep sea drilling fluid lifting system, which specifically comprises the following steps of. And acquiring historical working condition data of the deep sea drilling fluid lifting system, and generating a working condition point sample database based on a physical model. The method comprises the steps of constructing a machine learning agent model, training the machine learning agent model by using working point samples in a working point sample database to obtain a trained machine learning agent model, wherein the machine learning agent model is a model which takes drilling fluid density, power supply frequency and pump configuration mode as input and takes predicted flow and predicted lift under corresponding working conditions as output and is used for predicting the working conditions of a deep sea drilling fluid lifting system. The method comprises the steps of collecting current drilling fluid density in real time, carrying out real-time working condition predic