CN115438823A - Borehole wall instability mechanism analysis and prediction method and system - Google Patents
Borehole wall instability mechanism analysis and prediction method and system Download PDFInfo
- Publication number
- CN115438823A CN115438823A CN202110612424.3A CN202110612424A CN115438823A CN 115438823 A CN115438823 A CN 115438823A CN 202110612424 A CN202110612424 A CN 202110612424A CN 115438823 A CN115438823 A CN 115438823A
- Authority
- CN
- China
- Prior art keywords
- data
- parameters
- borehole wall
- safety risk
- analysis
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/26—Visual data mining; Browsing structured data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/27—Replication, distribution or synchronisation of data between databases or within a distributed database system; Distributed database system architectures therefor
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/951—Indexing; Web crawling techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0635—Risk analysis of enterprise or organisation activities
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/02—Agriculture; Fishing; Mining
Abstract
The invention provides a borehole wall instability mechanism analysis and prediction method and system, and belongs to the field of oil and gas drilling. The method comprises the steps of collecting parameters in a drilling process and constructing a drilling and completion data database; performing dimensionality reduction treatment on data in a well drilling and completion data database to obtain safety risk associated parameters, and establishing a safety risk early warning sample library; forming a borehole wall instability analysis and prediction model by taking the safety risk associated parameter as an input parameter and the borehole diameter expansion rate as an output parameter; and performing borehole wall instability analysis and prediction by using the borehole wall instability analysis and prediction model. The invention forms a borehole wall instability mechanism analysis and prediction method based on data mining by using an advanced big data analysis means, constructing an open distributed database of well drilling and completion data based on a web crawler technology and constructing a borehole diameter expansion rate related data artificial neural network analysis method, deeply reveals a borehole wall instability mechanism and provides guidance for relieving the borehole wall instability problem.
Description
Technical Field
The invention belongs to the field of petroleum and natural gas drilling, and particularly relates to a borehole wall instability mechanism analysis and prediction method and system, which are used for optimizing the performance of borehole wall stability drilling fluid in deep well ultra-deep well formations.
Background
Data mining, which is a nontrivial process that reveals implicit, previously unknown, and potentially valuable information from large amounts of data in a database, is a hot problem for research in the fields of artificial intelligence and databases. The data mining is characterized in that the technology for searching the rule of a large amount of data by analyzing each data mainly comprises three steps of data preprocessing, data mining, evaluation and representation. The data preparation is to select required data from related data sources and integrate the data into a data set for data mining; the rule searching is to find out the rules contained in the data set by a certain method; the law representation is to represent the found laws as much as possible in a way understandable to the user.
In recent years, with the deep ultra-deep layer of the Tarim basin and the Sichuan basin obtaining great oil and gas breakthrough, deep oil and gas reservoirs gradually become the oil and gas resource field with great significance in actual exploration and development. However, deep hydrocarbon reservoirs have special geological conditions and complex ground stress conditions, and the problem of borehole instability of deep strata during drilling is prominent under the action of long time, multiple times, strong tectonic movement, high ground stress and high disturbance stress during drilling. Taking the northward oil-gas field as an example, when the oil-gas field is drilled into the fractured stratum of the Ordovician carbonate rock in the construction process, the block falls seriously, and the drilling time efficiency is seriously influenced. According to incomplete statistics, the 5 exploration evaluation wells drilled in the work area are drilled on the same side for more than 10 times due to serious collapse and block falling of the well wall, the loss time of a single well is up to 242d at most, the accumulated loss time exceeds 669d, and the period of processing the well shaft accounts for 48.2 percent of the total period of drilling at most. The problem of borehole wall instability currently becomes a prominent problem for restricting the efficient well construction of deep oil and gas reservoirs.
The method for analyzing and predicting the borehole wall instability mechanism is mainly developed by relying on a borehole wall stability model. However, due to the complex ground stress conditions of the deep stratum, the development of natural cracks of the stratum and the poor adaptability of the rock fracture criterion, an efficient well wall stabilization model is difficult to establish, the difficulty in analyzing and predicting a well wall stabilization mechanism is high, and the evaluation reliability is insufficient.
Chinese patent publication CN110952978A discloses a drilling leakage fracture width prediction method based on neural network data mining, which provides decision support for leakage stoppage constructors and improves the primary leakage stoppage success rate of the fractured leakage. However, the method needs to collect the drilling and completion related data by manual screening, construct a variable set containing 12 input parameters, manually complete the collection of a large amount of drilling and completion related data, and is time-consuming and labor-consuming, and has poor accuracy; secondly, in the process of drilling loss, the parameters for determining the width of the fracture may not be only 12, but the patent only takes the 12 as input parameters, which may cause the deviation of the model prediction result to be large; thirdly, the method takes the fracture width obtained by imaging logging as a prediction result standard value, the fracture width is only the static fracture width of the stratum and is not the maximum dynamic width of the fracture in the leakage process, and inaccurate selection of the prediction result standard value causes low prediction accuracy of the prediction model.
In more than ten years, with the continuous progress of drilling technology, the number of built-up wells of deep and ultra-deep wells reaches a certain scale, and drilling data is accumulated in a large amount. Especially in the aspects of drilling fluid performance and borehole wall instability, a large amount of field data is accumulated. However, these data have not been converted to useful information and knowledge due to the lack of analytical methods for the data. Therefore, in order to effectively analyze the deep borehole wall instability mechanism and construct a borehole wall stability control technical method, an efficient data mining method is urgently needed to be constructed, a borehole wall instability mechanism analysis and prediction method based on data mining is formed, the borehole wall instability mechanism is deeply disclosed, and guidance is provided for relieving the borehole wall instability problem.
Disclosure of Invention
The invention aims to solve the problems in the prior art, and provides a borehole wall instability mechanism analysis and prediction method and system, which deeply disclose a borehole wall instability mechanism, construct a borehole wall stability prediction method, provide borehole wall stability countermeasures, provide more accurate and rapid analysis and prediction for borehole wall instability, and provide borehole wall stability control countermeasures.
The invention is realized by the following technical scheme:
the invention provides a well wall instability mechanism analysis and prediction method in a first aspect, wherein the method comprises the steps of collecting parameters in a well drilling process and constructing a well drilling and completion data database;
performing dimensionality reduction treatment on data in a well drilling and completion data database to obtain safety risk associated parameters, and establishing a safety risk early warning sample library;
forming a borehole wall instability analysis and prediction model by taking the safety risk associated parameters as input parameters and the borehole diameter expansion rate as output parameters;
and performing borehole wall instability analysis and prediction by using the borehole wall instability analysis and prediction model.
A further development of the invention is that the method comprises:
(1) Collecting various related data of a well wall instability process in a deep well and ultra-deep well drilling process, summarizing data in the data and storing the data in a database, and then capturing key parameters;
(2) Preprocessing the captured key parameters, and storing the preprocessed data into a well drilling and completion data database;
(3) Establishing a safety risk early warning sample library by using the all-factor characteristic parameters in the well drilling and completion data database;
(4) Based on a data mining algorithm, acquiring a borehole wall instability analysis and prediction model by taking safety risk associated parameters as input parameters, taking the borehole diameter expansion rate as output parameters and taking the collected actual value of the actually measured borehole diameter expansion rate of the logging as a standard;
(5) Obtaining safety risk associated parameters from data of oil and gas well drilling to date needing analyzing well wall stability, inputting the safety risk associated parameters into a well wall instability analyzing and predicting model, and outputting predicted well diameter expansion rate by the well wall instability analyzing and predicting model.
The invention is further improved in that the various types of related data in the step (1) comprise: formation data, drilling fluid data and logging related data;
and (2) capturing key parameters by adopting a crawler technology in the step (1).
The invention is further improved in that the full-factor characteristic parameters in the step (3) include 22 parameters, which are respectively: geodetic coordinates, construction type, construction position, formation strike, fault position, formation contact relation, ground stress, rock mechanics parameters, fracture width, porosity, permeability, pump pressure, drilling time, suspended weight, liquid level change, drilling fluid density, discharge capacity, rheological property, particle size distribution, API water loss amount, drilling fluid K + Concentration ofAnd the hole diameter enlargement rate.
A further improvement of the present invention is that the operation of step (3) includes:
collecting the actual value of the actually measured borehole diameter expansion rate obtained by resistivity logging, and carrying out normalization processing on the borehole diameter expansion rate of the stratum with similar horizon, depth and lithology;
and (3) performing dimensionality reduction treatment on 21 parameters except the borehole diameter expansion rate in the all-factor characteristic parameters according to the importance, taking 10 parameters with top 10 importance ranks as safety risk associated parameters, and storing the safety risk associated parameters into a safety risk early warning sample library.
The invention is further improved in that the data mining algorithm in the step (4) adopts: radial basis neural network models, BP neural networks, hopfield networks, ART networks, or Kohonen networks.
In a further development of the invention, the method further comprises:
(6) And finding characteristic parameters related to the drilling fluid from the safety risk related parameters, and adjusting the well wall stable drilling fluid process according to the characteristic parameters and the predicted hole diameter expansion rate.
The invention is further improved in that the operation of the step (6) comprises the following steps:
and obtaining the weight of each parameter in the safety risk associated parameters according to the predicted mapping relation between the hole diameter expansion rate and the safety risk associated parameters, and providing a drilling fluid adjusting scheme according to the characteristic parameters related to the drilling fluid, which are ranked in the front of the weights.
And obtaining a drilling fluid formula with optimized drilling fluid performance by using a drilling fluid performance adjustment expert system according to the drilling fluid adjustment scheme.
In a further development of the invention, the method further comprises:
(7) And (4) storing the safety risk associated parameters obtained in the step (5) into a safety risk early warning sample library, and taking the safety risk associated parameters obtained in the step (5) and the corresponding borehole diameter expansion rate as next training samples.
In a second aspect of the present invention, there is provided a borehole wall instability mechanism analysis and prediction system, the system comprising:
the data acquisition unit is used for collecting various related data of a well wall instability process in the deep well and ultra-deep well drilling process, summarizing and storing data in the data into a database, and then capturing key parameters;
the preprocessing unit is connected with the data acquisition unit and is used for preprocessing the captured key parameters and storing the preprocessed data into a well drilling and completion data database;
the sample library generating unit is connected with the preprocessing unit and used for establishing a safety risk early warning sample library by utilizing the all-factor characteristic parameters in the well drilling and completion data database;
the model generation unit is connected with the sample library generation unit and used for obtaining a borehole wall instability analysis and prediction model based on a data mining algorithm by taking a safety risk associated parameter as an input parameter, taking a borehole diameter expansion rate as an output parameter and taking a collected true value of the borehole diameter expansion rate actually measured by logging as a standard;
a prediction analysis unit: and the safety risk associated parameters are input into the borehole wall instability analysis and prediction model, and the borehole wall instability analysis and prediction model outputs the predicted borehole diameter expansion rate.
Compared with the prior art, the invention has the beneficial effects that: the invention forms a borehole wall instability mechanism analysis and prediction method based on data mining by using an advanced big data analysis means, constructing an open distributed database of well drilling and completion data based on a web crawler technology and constructing a borehole diameter expansion rate related data artificial neural network analysis method, deeply reveals a borehole wall instability mechanism and provides guidance for relieving the borehole wall instability problem.
Drawings
FIG. 1 is a schematic diagram of a Radial Basis Function (RBF) neural network model;
FIG. 2 is a Radial Basis Function (RBF) neural network neuron model;
FIG. 3 is a block diagram of the steps of the method of the present invention;
FIG. 4 is a block diagram of the system of the present invention.
Detailed Description
The invention is described in further detail below with reference to the accompanying drawings:
due to the complex ground stress condition of the deep stratum, the development of natural cracks of the stratum and the poor adaptability of rock fracture criteria, an efficient well wall stability model is difficult to establish, the difficulty in analyzing and predicting a well wall stability mechanism is high, and the evaluation reliability is insufficient. At present, a great deal of field data on the aspects of well drilling and completion characteristics, drilling fluid performance and borehole wall instability are accumulated, but due to the lack of a key parameter acquisition and analysis method for massive data, the actual situation that the data are converted into useful information and knowledge does not exist,
the invention provides a well wall instability analyzing and predicting method based on neural network data mining, which can solve the existing problems, provide more accurate and rapid analysis and prediction for well wall instability and provide well wall stability control countermeasures. The method comprises the following steps: collecting parameters in the drilling process and constructing a drilling and completion data database; cleaning and dimension reduction processing are carried out on data in a well drilling and completion data database to obtain safety risk associated parameters, and an associated parameter sample library is established; forming a borehole wall instability analysis and prediction model by taking the safety risk associated parameter as an input parameter and the borehole diameter expansion rate as an output parameter; and carrying out borehole wall instability analysis and prediction by using the borehole wall instability analysis and prediction model.
The invention specifies 22 parameters which can be obtained in the drilling process, and mainly comprises the following steps: geodetic coordinates, construction type, construction position, stratum strike, fault position, stratum contact relation, ground stress, rock mechanics parameters, fracture width, porosity, permeability, pump pressure, drilling time, hanging weight, liquid level change, drilling fluid density, discharge capacity, rheological property, particle size distribution, API water loss amount and drilling fluid K + Concentration, hole diameter enlargement rate; an open type and distributed well drilling and completion data database constructed based on a web crawler technology accurately captures required parameters; by entering initial association data in a well completion data databaseCleaning, carrying out data dimension reduction processing, determining importance ranking of risk associated parameters, and establishing an associated parameter sample library; and (3) by taking the risk associated parameters as input parameters and the hole diameter expansion rate as output parameters and continuously optimizing neural network prediction model parameters, the model prediction precision is improved, and a borehole wall instability analysis and prediction model is formed.
Specifically, as shown in fig. 3, the method of the present invention comprises the following steps:
(1) Collecting various related data of a well wall instability process in a deep well and ultra-deep well drilling process, summarizing data in the data into a database, searching required key parameters through a crawler technology, and capturing the key parameters to form characteristic parameters required by a prediction model.
The various related data in the deep well and ultra-deep well drilling process comprise: formation data, drilling fluid data, logging related data, and the like.
(2) Preprocessing the captured key parameters: completing missing data, modifying data with wrong format and content, removing and modifying data with wrong logic, removing unnecessary data, performing correlation verification, and finally storing the cleaned data to form a well drilling and completion data database; the pretreatment can be performed by adopting various conventional methods, which are not described herein again.
(3) Forming a safety risk early warning sample library by utilizing the all-factor characteristic parameters of different lithologic stratums in the well drilling and completion data database under different well drilling states;
the full-factor characteristic parameters comprise 22 key parameters which are respectively as follows: geodetic coordinates, construction type, construction position, stratum strike, fault position, stratum contact relation, ground stress, rock mechanics parameters, fracture width, porosity, permeability, pump pressure, drilling time, hanging weight, liquid level change, drilling fluid density, discharge capacity, rheological property, particle size distribution, API water loss amount and drilling fluid K + Concentration, hole diameter enlargement rate.
And collecting the actual value of the actually measured hole diameter expansion rate obtained by resistivity logging, and carrying out normalization processing on the hole diameter expansion rate of the stratum with the similar horizon, depth and lithology, wherein the interval of the normalized hole diameter expansion rate is not more than 0.5m. And (2) performing dimensionality reduction treatment on the associated parameters (21 parameters in 22 key parameters, excluding the borehole diameter expansion rate) according to the importance by adopting a Pearson correlation analysis algorithm, a recursive elimination characteristic algorithm and a random forest algorithm (dimensionality reduction is realized by sequencing the importance of data and only selecting the parameters with the top importance). And according to the importance ranking, determining the index parameters with the importance ranking of 10 above as safety risk associated parameters, and establishing a safety risk early warning sample library, wherein the sample library comprises the safety risk associated parameters, namely the index parameters with the importance ranking of 10 above.
(4) Based on a data mining algorithm, taking safety risk associated parameters in historical drilling data as input parameters, taking the borehole diameter expansion rate as output parameters, taking the actual value of the collected well diameter expansion rate actually measured by logging as a standard, training a neural network by setting parameters such as the number of network layers, iteration times, convergence precision and the like, and generating a prediction model of borehole diameter expansion rate analysis and next-stage borehole diameter expansion rate, namely a borehole wall instability analysis and prediction model by parameter optimization;
the data mining algorithm adopted by the invention includes but is not limited to a neural network algorithm, and the obtained model includes but is not limited to a neural network prediction model, such as an existing Radial Basis Function (RBF) neural network model, a BP neural network, a Hopfield network, an ART network, a Kohonen network and the like.
And (5) obtaining a borehole wall instability analysis and prediction model through the step (4).
The embodiment of the process of supervising, training and optimizing the borehole wall instability analysis and prediction model is as follows:
(1) setting Input Data of a training set, and taking the safety risk associated parameters collected and sorted in the step (3) as Data for neural network training;
(2) setting an Output standard value Output Target of a training set, wherein the standard value is a real borehole diameter expansion rate obtained by logging;
(3) establishing a Radial Basis Function (RBF) neural network model, setting the number of nodes of an input layer, a hidden layer and an output layer, setting a transfer function, and performing unconstrained nonlinear optimization on the neural network model by using a genetic algorithm to finally obtain an optimized borehole wall instability analysis and prediction model.
The above 3 steps are conventional steps for generating a neural network model, and are not described herein again.
(5) The data statistics of the oil and gas well drilling operation to date needing analyzing the well wall stability is input into an analysis database, relevant characteristic parameters (namely obtained safety risk associated parameters) of the interval needing analyzing are extracted according to the same rule, a well wall instability analysis and prediction model is led in to predict the well diameter expansion rate, the well wall instability analysis and the prediction of the well wall stability state of the interval not drilled are deepened, and the well diameter expansion rate of the interval not drilled is obtained by the well wall instability analysis and prediction model.
The method further comprises:
(6) And (3) sorting the data corresponding to the prediction result and the drilling fluid performance characteristic with the front importance: during analysis, due to the change of input and output parameters, the weights of different input parameters may change, at this time, the weight value change of the safety risk associated parameter can be obtained, further, the characteristic parameter related to the drilling fluid in the safety risk associated parameter is obtained, and the well wall stable drilling fluid process is adjusted according to the characteristic parameter and the well diameter expansion rate prediction value:
the hole diameter expansion rate obtained by prediction is used for adjusting the density of the drilling fluid, the viscosity of a funnel, the water loss amount of API (application program interface) and the drilling fluid K + Concentration, etc. of drilling fluid. Specifically, the weights of all parameters in the safety risk associated parameters are obtained according to the neural network mapping relation between the borehole diameter expansion rate and the safety risk associated parameters, and a drilling fluid adjusting scheme is provided according to characteristic parameters related to the drilling fluid and ranked in the front of the weights. How to provide a drilling fluid adjustment scheme according to characteristic parameters related to the drilling fluid is realized by adopting the prior art, and details are not repeated herein.
And then, according to the drilling fluid adjusting scheme, obtaining a specific drilling fluid formula with optimized drilling fluid performance by utilizing the existing drilling fluid performance adjusting expert system. The method provides a faster and effective decision basis for drilling fluid engineers and field construction personnel, improves the operation efficiency of the drilling fluid for maintaining the stability of the well wall, and avoids serious accidents such as well wall instability and even well collapse.
Further, the method further comprises:
(7) And (4) storing the safety risk associated parameters obtained in the step (5) into a safety risk early warning sample library, and taking the safety risk associated parameters obtained in the step (5) and the corresponding borehole diameter expansion rate as training samples for next borehole diameter expansion rate analysis, so as to realize borehole wall instability analysis and continuous increase of the number of predicted samples.
The embodiment of the method of the invention is as follows:
[ EXAMPLES ] A method for producing a semiconductor device
Fig. 1 is a schematic structural diagram of a neural network model in this embodiment. As can be seen from fig. 1, the radial basis function neural network is a three-layer feedforward neural network, and it is the input layer that is responsible for receiving the external information, and for the conversion from the input layer to the hidden layer, the hidden layer is the hidden layer, and at the same time, the nonlinear conversion can be realized in the hidden layer, and the third layer is a function of outputting, and is the final result of inputting. Geological and drilling parameters can be converted into numerical equivalents to be input into the neural network model.
The influence of geological characteristics and drilling fluid performance parameters on borehole wall instability prediction is comprehensively considered, the preprocessed historical geological and drilling fluid parameters are used as input parameters, the borehole diameter expansion rate is used as an output parameter, the real borehole diameter expansion rate obtained through well logging is used as a standard parameter, and a borehole diameter expansion rate neural network prediction model is obtained through supervision training and optimization.
Fig. 2 is a diagram of a Radial Basis (RBF) neural network model architecture. As can be seen from fig. 2, the input of the RBF neural network model is P, which represents various well history parameters, the weight and the threshold are w and b, respectively, which represent predicted well diameter expansion rate feedback values of the RBF model, the number of hidden layer nodes is L, the output of the linear neuron model is a, which represents predicted well diameter expansion rate values. The number of the neurons between the input layer and the output layer of the neural network model is 6 and 1 respectively. In order to improve the calculation accuracy, a K-means self-organizing clustering method is adopted to determine a proper data center for the radial basis function of the hidden layer node, an initial value is given firstly, then the adjustment is carried out slowly, and the optimal number of the hidden layer node is determined through data comparison.
The specific process of optimizing the neural network model supervision training set for predicting the borehole diameter expansion rate is as follows:
the RBF network has K input nodes, the number of hidden layer nodes is the last, the number of output layer nodes is M, the number of learning sample wells is n, and the defined error objective function is as follows:
in the formula: λ is the forgetting factor, δ t ，
y i (n) error, desired output and actual output, respectively, of the output node i. The recursion least square method for training the RBF neural network connection weight is as follows:
and (3) carrying out self-adaptive learning on the network model by adopting a self-adaptive gradient descent method:
s j (n)＝s j (n-1)+ηδ j (n)+α[s j (n-1)-s j (n-2)]
wherein eta is learning rate, eta is more than 0 and less than 0.1, and alpha is more than 0 and less than 0.01.
(4) And (3) importing the geological and drilling fluid data related to the target well into a neural network model, automatically analyzing the hole diameter expansion rate by the model, and analyzing the influence factors of the hole diameter expansion rate and the hole diameter expansion rate state of the lower stage.
In this embodiment, the difference between the target values of the neural network training errors is set to 0.01, and the maximum training times is set to 1000-2000 times.
(5) And (4) sorting the prediction result and the data corresponding to the drilling fluid performance characteristics with the front importance, making a borehole wall stability strategy and adjusting the related performance of the drilling fluid performance in time.
(6) And adding the analysis result and the corresponding characteristic numerical data into a sample library to be used as a training sample of the next analysis project.
(7) The hole diameter expansion rate and the influence factors obtained through prediction are used as a judgment basis for optimizing the performance of the drilling fluid, so that a more accurate and effective decision basis is provided for drilling fluid constructors, the stable working efficiency of the well wall is improved, and the problem of well wall instability is avoided.
The invention also provides a system for analyzing and predicting the borehole wall instability mechanism, which comprises the following embodiments:
[ EXAMPLE II ]
As shown in fig. 4, the system includes:
the data acquisition unit 10 is used for collecting various related data of a well wall instability process in a deep well and ultra-deep well drilling process, summarizing data in the data into a database, and then capturing key parameters;
the preprocessing unit 20 is connected with the data acquisition unit 10 and used for preprocessing the captured key parameters and storing the preprocessed data into a well drilling and completion data database;
the sample library generating unit 30 is connected with the preprocessing unit 20 and used for establishing a safety risk early warning sample library by using the all-factor characteristic parameters in the drilling and completion data database;
the model generating unit 40 is connected with the sample library generating unit 30 and is used for acquiring a borehole wall instability analyzing and predicting model based on a data mining algorithm by taking the safety risk associated parameter as an input parameter, taking the borehole diameter expansion rate as an output parameter and taking the actual value of the collected borehole diameter expansion rate actually measured by logging as a standard;
the prediction analysis unit 50: and the model generating unit 40 is connected to obtain safety risk associated parameters from the data of the oil and gas well drilling to date needing to analyze the well wall stability, input the safety risk associated parameters into the well wall instability analysis and prediction model, and output the predicted well diameter expansion rate by the well wall instability analysis and prediction model.
The method can be used for timely, quickly and effectively analyzing the borehole wall instability mechanism and predicting the borehole wall instability state, provides a quicker and effective decision basis for drilling fluid engineers and field construction personnel, improves the operation efficiency of the drilling fluid for maintaining the borehole wall stability, and avoids serious accidents such as borehole wall instability and even borehole collapse.
Finally, it should be noted that the above-mentioned technical solution is only one embodiment of the present invention, and it will be apparent to those skilled in the art that various modifications and variations can be easily made based on the application method and principle of the present invention disclosed herein, and the method is not limited to the method described in the above-mentioned embodiment of the present invention, so that the above-mentioned embodiment is only preferred and not restrictive.
Claims (10)
1. A borehole wall instability mechanism analysis and prediction method is characterized in that: the method comprises the steps of collecting parameters in a drilling process and constructing a drilling and completion data database;
performing dimensionality reduction treatment on data in a well drilling and completion data database to obtain safety risk associated parameters, and establishing a safety risk early warning sample library;
forming a borehole wall instability analysis and prediction model by taking the safety risk associated parameter as an input parameter and the borehole diameter expansion rate as an output parameter;
and performing borehole wall instability analysis and prediction by using the borehole wall instability analysis and prediction model.
2. The borehole wall instability mechanism analysis and prediction method according to claim 1, characterized in that: the method comprises the following steps:
(1) Collecting various related data of a well wall instability process in a deep well and ultra-deep well drilling process, summarizing and storing data in the data into a database, and then capturing key parameters;
(2) Preprocessing the captured key parameters, and storing the preprocessed data into a well drilling and completion data database;
(3) Establishing a safety risk early warning sample library by using all-factor characteristic parameters in a drilling and completion data database;
(4) Based on a data mining algorithm, acquiring a borehole wall instability analysis and prediction model by taking a safety risk associated parameter as an input parameter, a borehole diameter expansion rate as an output parameter and a collected actual value of the actually measured borehole diameter expansion rate of the logging as a standard;
(5) Obtaining safety risk associated parameters from data of oil and gas well drilling to date needing analyzing well wall stability, inputting the safety risk associated parameters into a well wall instability analyzing and predicting model, and outputting predicted well diameter expansion rate by the well wall instability analyzing and predicting model.
3. The borehole wall instability mechanism analysis and prediction method according to claim 2, characterized in that: the various related data in the step (1) comprise: formation data, drilling fluid data and logging related data;
and (2) capturing key parameters by adopting a crawler technology in the step (1).
4. The borehole wall instability mechanism analysis and prediction method according to claim 3, characterized in that: the full-factor characteristic parameters in the step (3) include 22 parameters, which are respectively: geodetic coordinates, construction type, construction position, formation strike, fault position, formation contact relation, ground stress, rock mechanics parameters, fracture width, porosity, permeability, pump pressure, drilling time, suspended weight, liquid level change, drilling fluid density, discharge capacity, rheological property, particle size distribution, API water loss amount, drilling fluid K + Concentration, hole diameter enlargement rate.
5. The borehole wall instability mechanism analysis and prediction method according to claim 4, characterized in that: the operation of the step (3) comprises:
collecting a true value of an actually measured hole diameter expansion rate obtained by resistivity logging, and carrying out normalization processing on the hole diameter expansion rate of the stratum with the layer position, the depth and the lithology similarity;
and (3) performing dimensionality reduction treatment on 21 parameters except the borehole diameter expansion rate in the all-factor characteristic parameters according to the importance, taking 10 parameters with top 10 importance ranks as safety risk associated parameters, and storing the safety risk associated parameters into a safety risk early warning sample library.
6. The borehole wall instability mechanism analysis and prediction method according to claim 2, characterized in that: the data mining algorithm in the step (4) adopts: radial basis neural network models, BP neural networks, hopfield networks, ART networks, or Kohonen networks.
7. The borehole wall instability mechanism analysis and prediction method according to claim 4, characterized in that: the method further comprises the following steps:
(6) And finding characteristic parameters related to the drilling fluid from the safety risk related parameters, and adjusting the well wall stable drilling fluid process according to the characteristic parameters and the predicted hole diameter expansion rate.
8. The borehole wall instability mechanism analysis and prediction method according to claim 7, characterized in that: the operation of the step (6) comprises the following steps:
and obtaining the weight of each parameter in the safety risk associated parameters according to the mapping relation between the predicted hole diameter expansion rate and the safety risk associated parameters, and providing a drilling fluid adjusting scheme according to the characteristic parameters related to the drilling fluid and in the front of the weight sequence.
And obtaining a drilling fluid formula with optimized drilling fluid performance by using a drilling fluid performance adjustment expert system according to the drilling fluid adjustment scheme.
9. The borehole wall instability mechanism analysis and prediction method according to claim 4, characterized in that: the method further comprises the following steps:
(7) And (4) storing the safety risk associated parameters obtained in the step (5) into a safety risk early warning sample library, and taking the safety risk associated parameters obtained in the step (5) and the corresponding borehole diameter expansion rate as next training samples.
10. A borehole wall instability mechanism analysis and prediction system is characterized in that: the system comprises:
the data acquisition unit is used for collecting various related data of a well wall instability process in the deep well and ultra-deep well drilling process, summarizing data in the data into a database, and then capturing key parameters;
the preprocessing unit is connected with the data acquisition unit and is used for preprocessing the captured key parameters and storing the preprocessed data into a well drilling and completion data database;
the sample library generating unit is connected with the preprocessing unit and used for establishing a safety risk early warning sample library by utilizing the all-factor characteristic parameters in the well drilling and completion data database;
the model generation unit is connected with the sample library generation unit and used for obtaining a borehole wall instability analysis and prediction model based on a data mining algorithm by taking a safety risk associated parameter as an input parameter, taking a borehole diameter expansion rate as an output parameter and taking a collected true value of the borehole diameter expansion rate actually measured by logging as a standard;
a prediction analysis unit: and the safety risk associated parameters are input into the borehole wall instability analysis and prediction model, and the borehole wall instability analysis and prediction model outputs the predicted borehole diameter expansion rate.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110612424.3A CN115438823A (en) | 2021-06-02 | 2021-06-02 | Borehole wall instability mechanism analysis and prediction method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110612424.3A CN115438823A (en) | 2021-06-02 | 2021-06-02 | Borehole wall instability mechanism analysis and prediction method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115438823A true CN115438823A (en) | 2022-12-06 |
Family
ID=84271719
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110612424.3A Pending CN115438823A (en) | 2021-06-02 | 2021-06-02 | Borehole wall instability mechanism analysis and prediction method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115438823A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116822971A (en) * | 2023-08-30 | 2023-09-29 | 长江大学武汉校区 | Well wall risk level prediction method |
-
2021
- 2021-06-02 CN CN202110612424.3A patent/CN115438823A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116822971A (en) * | 2023-08-30 | 2023-09-29 | 长江大学武汉校区 | Well wall risk level prediction method |
| CN116822971B (en) * | 2023-08-30 | 2023-11-14 | 长江大学武汉校区 | Well wall risk level prediction method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108388921B (en) | Overflow loss real-time identification method based on random forest | |
| CN104747185A (en) | Heterogeneous reservoir stratum synthetic classifying evaluation method | |
| CN107122860B (en) | Rock burst danger level prediction method based on grid search and extreme learning machine | |
| CN112529341B (en) | Drilling well leakage probability prediction method based on naive Bayesian algorithm | |
| CN103617147A (en) | Method for identifying mine water-inrush source | |
| CN105760673A (en) | Fluvial facies reservoir earthquake sensitive parameter template analysis method | |
| CN111985610A (en) | System and method for predicting pumping efficiency of oil pumping well based on time sequence data | |
| CN115906675B (en) | Well position and injection and production parameter joint optimization method based on time sequence multi-target prediction model | |
| Aljubran et al. | Deep learning and time-series analysis for the early detection of lost circulation incidents during drilling operations | |
| CN113283180A (en) | K-means and SVR combination-based tight reservoir horizontal well fracturing productivity prediction method and application | |
| CN114117881A (en) | Sand production risk prediction method and system | |
| CN112085242B (en) | Low-resistance layer accurate prediction method and device based on big data deep learning method | |
| CN116384554A (en) | Method and device for predicting mechanical drilling speed, electronic equipment and computer storage medium | |
| CN114004100B (en) | Oil reservoir assisted history fitting and optimization simulation method | |
| Bajolvand et al. | Optimization of controllable drilling parameters using a novel geomechanics-based workflow | |
| CN115438823A (en) | Borehole wall instability mechanism analysis and prediction method and system | |
| CN115907236B (en) | Underground complex situation prediction method based on improved decision tree | |
| CN115438090A (en) | Borehole wall instability type analysis and prediction method and system based on lost block image characteristics | |
| CN116562428A (en) | Fracturing construction parameter optimization method based on machine learning | |
| CN116910638A (en) | Dominant seepage channel intelligent identification method based on big data | |
| CN115809411A (en) | Improved decision tree lithology identification method based on logging data | |
| CN115964667A (en) | River-lake lithofacies well logging identification method based on deep learning and resampling | |
| CN113537706A (en) | Oil field production increasing measure optimization method based on intelligent integration | |
| CN113052374A (en) | Data-driven intelligent prediction method for casing loss depth of oil well | |
| RU2300632C1 (en) | Horizontal well output estimation method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |