CN116522777A - Drilling process drilling rate online prediction method and system based on multi-source information fusion - Google Patents
Drilling process drilling rate online prediction method and system based on multi-source information fusion Download PDFInfo
- Publication number
- CN116522777A CN116522777A CN202310483771.XA CN202310483771A CN116522777A CN 116522777 A CN116522777 A CN 116522777A CN 202310483771 A CN202310483771 A CN 202310483771A CN 116522777 A CN116522777 A CN 116522777A
- Authority
- CN
- China
- Prior art keywords
- drilling
- rate
- prediction model
- speed
- stratum
- 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
- 238000005553 drilling Methods 0.000 title claims abstract description 301
- 238000000034 method Methods 0.000 title claims abstract description 100
- 230000008569 process Effects 0.000 title claims abstract description 44
- 230000004927 fusion Effects 0.000 title claims abstract description 29
- 238000013528 artificial neural network Methods 0.000 claims abstract description 38
- 230000008859 change Effects 0.000 claims abstract description 26
- 238000007781 pre-processing Methods 0.000 claims abstract description 11
- 238000005457 optimization Methods 0.000 claims abstract description 7
- 210000002569 neuron Anatomy 0.000 claims description 40
- 230000000007 visual effect Effects 0.000 claims description 34
- 230000006870 function Effects 0.000 claims description 30
- 238000001914 filtration Methods 0.000 claims description 27
- 238000012549 training Methods 0.000 claims description 25
- 230000009466 transformation Effects 0.000 claims description 20
- 238000005516 engineering process Methods 0.000 claims description 16
- 230000015572 biosynthetic process Effects 0.000 claims description 15
- 238000012216 screening Methods 0.000 claims description 12
- 230000035515 penetration Effects 0.000 claims description 11
- 238000009826 distribution Methods 0.000 claims description 10
- 239000004973 liquid crystal related substance Substances 0.000 claims description 10
- 230000004913 activation Effects 0.000 claims description 5
- 238000004458 analytical method Methods 0.000 claims description 5
- 230000005540 biological transmission Effects 0.000 claims description 5
- 238000011156 evaluation Methods 0.000 claims description 5
- 238000011478 gradient descent method Methods 0.000 claims description 5
- 230000009467 reduction Effects 0.000 claims 1
- 230000009286 beneficial effect Effects 0.000 abstract description 8
- 238000004904 shortening Methods 0.000 abstract description 2
- 230000008901 benefit Effects 0.000 description 3
- 238000011161 development Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 239000011435 rock Substances 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- E—FIXED CONSTRUCTIONS
- E21—EARTH DRILLING; MINING
- E21B—EARTH DRILLING, e.g. DEEP DRILLING; OBTAINING OIL, GAS, WATER, SOLUBLE OR MELTABLE MATERIALS OR A SLURRY OF MINERALS FROM WELLS
- E21B45/00—Measuring the drilling time or rate of penetration
-
- 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
- G06F16/21—Design, administration or maintenance of databases
- G06F16/215—Improving data quality; Data cleansing, e.g. de-duplication, removing invalid entries or correcting typographical errors
-
- 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
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2457—Query processing with adaptation to user needs
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/082—Learning methods modifying the architecture, e.g. adding, deleting or silencing nodes or connections
-
- 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
- G06N3/084—Backpropagation, e.g. using gradient descent
Abstract
The invention provides a drilling process drilling rate online prediction method and a drilling process drilling rate online prediction system based on multi-source information fusion, wherein the method comprises the following steps: the drilling data preprocessing stage is to filter the drilling pressure, the rotating speed and the torque; in the optimizing drilling speed modeling stage, a model structure of a limited Boltzmann machine and a back propagation neural network is improved through a mixed bat algorithm to form a novel mixed bat algorithm optimizing-limited Boltzmann machine-back propagation neural network, and an offline drilling speed prediction model facing a complex stratum environment is established through the algorithm; and in the drilling speed prediction model updating stage, lithology change and time interval are used as updating conditions, and a sliding window strategy is used for updating the drilling speed prediction model on the basis of stratum and drilling multisource information fusion, so that the online prediction of the drilling speed is realized. The beneficial effects of the invention are as follows: the method is beneficial to shortening the drilling period, reducing the operation cost, realizing high-precision prediction of the drilling speed, having effectiveness and laying an important foundation for intelligent optimization control of the drilling process.
Description
Technical Field
The invention relates to the field of geological drilling engineering, in particular to a drilling process drilling rate online prediction method and system based on multi-source information fusion.
Background
The total consumption amount of the resource and the energy in China rises year by year, the contradiction between supply and demand of strategic resource and energy is greatly increased, the external dependency is high, and the guarantee of the resource and energy safety is very important for the national economic development and strategic safety. With the continuous exploitation of medium and shallow resource energy and the continuous exploration of a large amount of deep buried resources, deep geological exploration and development have become necessary. However, the deep ground has complex and severe environment, the lithology of the stratum is rich and various, and the stratum is often subjected to alternating hardness and softness, rock breaking and the like, so that obvious characteristics of multisource variable, low-quality drilling data, strong nonlinearity, severe lithology change and the like exist in the drilling process, the drilling efficiency and the overall benefit are seriously influenced, meanwhile, the drilling rate is a key parameter for determining the drilling efficiency, and the drilling period and the cost are reduced by establishing an accurate drilling rate prediction model. Thus, there is a need to expedite research related to predicting the rate of penetration of advanced deep geological drilling processes.
Existing drill rate prediction studies can be divided into two categories, offline drill rate prediction and online drill rate prediction. The offline drilling speed prediction model is static modeling by utilizing industrial data of nearby well sites or areas, and good prediction effect can be obtained under the condition that the stratum environment is not changed greatly; the online drilling speed prediction model is a dynamic updating model of flow type industrial data in the drilling process, has real-time prediction capability and can acquire higher prediction precision, but the models take time intervals or drilling depths as updating indexes, and stratum lithology information is not fully considered, so that complex stratum environments are still difficult to effectively cope with. Therefore, aiming at the current situation and the difficult problem, a high-precision online prediction model of the drilling speed, which comprehensively considers the fusion of stratum and drilling information, needs to be established, and technical support is provided for realizing intelligent optimal control of the deep geological drilling process.
The foregoing is provided merely for the purpose of facilitating understanding of the technical solutions of the present invention and is not intended to represent an admission that the foregoing is prior art.
Disclosure of Invention
In order to solve the technical problems, the invention provides a drilling process drilling rate online prediction method and a drilling process drilling rate online prediction system based on multi-source information fusion.
In order to achieve the technical purpose, the invention adopts the following technical scheme: a drilling process drilling rate online prediction method based on multi-source information fusion comprises the following steps:
s1: selecting the drilling pressure, the rotating speed and the torque as the input of a drilling speed prediction model, screening and filtering drilling data by using outlier rejection and wavelet filtering analysis, acquiring a stratum drillability identification value by using a lithology identification technology, and judging the lithology change condition of the stratum;
s2: introducing a mixed bat algorithm optimization-a limited Boltzmann machine-a back propagation neural network algorithm, and establishing an offline drilling speed prediction model facing the complex stratum environment through three steps of mixed bat algorithm optimization, limited Boltzmann machine training, back propagation neural network training and back fine tuning;
s3: taking lithology change and time interval as drilling speed prediction model updating conditions, and applying a sliding window strategy to update the drilling speed prediction model in real time on the basis of stratum drillability information and drilling process information;
s4: and inputting the actually obtained drilling weight, rotation speed and torque into the updated drilling speed prediction model to obtain the drilling speed at the next moment.
Further, the specific process of step S1 is as follows:
s11: screening drilling data by using an outlier rejection method based on engineering experience, wherein the drilling data comprises weight on bit, rotating speed, torque and drilling rate, and the measuring range of the drilling data is as follows:
WOB is weight on bit, its unit is KN, RPM is rotational speed, its unit is RPM, torque is Torque, its unit is Nm, ROP is rate of drilling, its unit is cm/min;
s12: and filtering peaks and burrs in drilling data by adopting a wavelet filtering method, wherein a wavelet transformation expression is as follows:
wherein W is f (a, b) is drilling characteristic information after wavelet positive transformation, f (t) is original drilling data, a is a scaling factor, b is a scale factor, t is time, ψ () is a wavelet basis function, and a wavelet inverse transformation expression is:
wherein g (t) is drilling data after wavelet inverse transformation, c ψ Is a wavelet factor;
s13: and marking the drilled core, analyzing the core in a proper environment by using a lithology recognition technology to acquire stratum drillability information, and simultaneously judging whether the stratum lithology changes in real time according to the stratum drillability information at continuous moments.
Further, the step S2 specifically includes the following steps:
s21: introducing trial-and-error and mixed bat algorithm to optimize super parameters of a drilling rate prediction model, wherein the super parameters comprise: batch size, training iteration number, hidden layer neuron number, learning rate and momentum;
s22: extracting key features in drilling data by training a limited boltzmann machine, wherein the limited boltzmann machine comprises a visual layer v and a hidden layer h, and an energy function under a given state (v, h) is defined as follows:
wherein E is θ (v, h) is the energy function of the limited boltzmann machine, θ= { w ij ,c i ,d j Model parameters of the restricted boltzmann machine, p and q are the number of neurons of the visual and hidden layers, c i And d j Bias for the ith visual layer neuron and the jth hidden layer neuron, v i Is the ith visual layer neuron, h j Is the j-th hidden layer neuron, w ij For the weight between the ith visual layer neuron and the jth hidden layer neuron, i=1, 2 …, p, j=1, 2 …, q;
given drilling training set Is the nth k The log likelihood function L (θ) of the restricted boltzmann machine on K is set to:
where r is the r-th drilling data sample, r=1, 2 …, n k ，n k V is the total number of samples r For the r-th visual layer, P (v r ) For visual layer v r Is a marginal probability distribution of (1);
s23: taking the output characteristics of the constrained boltzmann machine as input to a back propagation neural network that includes two parts, forward propagation and back propagation, the forward propagation can be expressed as:
wherein, the liquid crystal display device comprises a liquid crystal display device,for the prediction of the drilling rate of the forward transmission output, f () is the activation function, g r For output layer bias, k represents the kth hidden layer node, k=1, 2 …, n, n is the total number of hidden layer nodes, w kr To conceal the connection weight between the layer and the output layer, x k Is a hidden layer node;
meanwhile, the network structure is continuously adjusted by using back propagation, and the error function of the back propagation neural network can be expressed as:
wherein E is y For the mean square error of the back propagation neural network, n k For the total number of samples,to be the predicted value of drilling speed, y r To obtain the actual value of the drilling speed, the parameters (w) of the back propagation neural network are updated by using a gradient descent method kr ,g r )。
Further, the specific process of step S3 is as follows:
the lithology change and the time interval are simultaneously used as model updating conditions, a sliding window strategy is used for updating a drilling rate prediction model in real time on the basis of stratum and drilling multisource information fusion, and the model updating conditions are expressed as:
where u () is the switching variable of the rate of penetration prediction model update, Δu r State variables, deltau, updated for lithology changes t State variable updated for time interval, r s And r s+1 For formation drillability corresponding to two successive moments, r z Updating threshold for lithology change, t a And t a+1 For two successive time points, t b Is the time interval update threshold, when Deltau r Greater than or equal to r z When the drilling speed prediction model is used, updating according to formation lithology information; when Deltau t Equal to t b When the drilling speed prediction model is used, the drilling speed prediction model is updated by continuously learning the real drilling data under the condition of taking time as a reference;
when one of the above conditions is satisfied, the drilling rate prediction model is updated and predicted in real time.
Further, the method also comprises the step of evaluating the prediction performance of the method by using a model evaluation index, wherein the specific formula is as follows:
wherein RMSE is the root mean square error, NRMSE is the normalized root mean square error, y r Is the actual value of the drilling rate,is the predicted value of drilling speed, r is the r-th drilling data sample, r=1, 2 …, n k ，n k Representing the total number of samples.
An on-line drilling speed prediction system for a drilling process based on multi-source information fusion comprises:
the data preprocessing module is used for selecting the drilling pressure, the rotating speed and the torque as the input of a drilling speed prediction model, screening and filtering drilling data by using outlier rejection and wavelet filtering analysis, acquiring a stratum drillability identification value by using a lithology identification technology, and judging the lithology change condition of the stratum;
the model building module is used for introducing a mixed bat algorithm optimization-limited Boltzmann machine-back propagation neural network algorithm, and building an offline drilling speed prediction model facing the complex stratum environment through three steps of mixed bat algorithm optimization, limited Boltzmann machine training, back propagation neural network training and back fine tuning;
the model updating module is used for taking lithology change and time interval as the updating conditions of the drilling speed prediction model and applying a sliding window strategy to update the drilling speed prediction model in real time on the basis of stratum drillability information and drilling process information;
and the drilling speed prediction module is used for inputting the actually acquired drilling weight, rotation speed and torque into the updated drilling speed prediction model to obtain the drilling speed at the next moment.
Further, in the data preprocessing module, the preprocessing process of drilling data comprises the following steps:
s11: screening drilling data by using an outlier rejection method based on engineering experience, wherein the drilling data comprises weight on bit, rotating speed, torque and drilling rate, and the measuring range of the drilling data is as follows:
WOB is weight on bit, its unit is KN, RPM is rotational speed, its unit is RPM, torque is Torque, its unit is Nm, ROP is rate of drilling, its unit is cm/min;
s12: and filtering peaks and burrs in drilling data by adopting a wavelet filtering method, wherein a wavelet transformation expression is as follows:
wherein W is f (a, b) is drilling characteristic information after wavelet positive transformation, f (t) is original drilling data, a is a scaling factor, b is a scale factor, t is time, ψ () is a wavelet basis function, and a wavelet inverse transformation expression is:
wherein g (t) is drilling data after wavelet inverse transformation, c ψ Is a wavelet factor;
s13: and marking the drilled core, analyzing the core in a proper environment by using a lithology recognition technology to acquire stratum drillability information, and simultaneously judging whether the stratum lithology changes in real time according to the stratum drillability information at continuous moments.
Further, in the model building module, the model building process is as follows:
s21: introducing trial-and-error and mixed bat algorithm to optimize super parameters of a drilling rate prediction model, wherein the super parameters comprise: batch size, training iteration number, hidden layer neuron number, learning rate and momentum;
s22: key features in the drilling data are extracted by training a limited boltzmann machine. The constrained boltzmann machine comprises a visible layer v and a hidden layer h, defining an energy function for a given state (v, h) as:
wherein E is θ (v, h) is the energy function of the limited boltzmann machine, θ= { w ij ,c i ,d j Model parameters of the restricted boltzmann machine, p and q are the number of neurons of the visual and hidden layers, c i And d j Bias for the ith visual layer neuron and the jth hidden layer neuron, v i Is the ith visual layer neuron, h j Is the j-th hidden layer neuron, w ij Hiding for the ith visual layer neuron and the jthWeights between layer neurons, i=1, 2 …, p, j=1, 2 …, q;
given drilling training set Is the nth k The log likelihood function L (θ) of the restricted boltzmann machine on K is set to:
where r is the r-th drilling data sample, r=1, 2 …, n k ，n k V is the total number of samples r For the r-th visual layer, P (v r ) For visual layer v r Is a marginal probability distribution of (1);
s23: taking the output characteristics of the constrained boltzmann machine as input to a back propagation neural network that includes two parts, forward propagation and back propagation, the forward propagation can be expressed as:
wherein, the liquid crystal display device comprises a liquid crystal display device,for the prediction of the drilling rate of the forward transmission output, f () is the activation function, g r For output layer bias, k represents the kth hidden layer node, k=1, 2 …, n, n is the total number of hidden layer nodes, w kr To conceal the connection weight between the layer and the output layer, x k Is a hidden layer node;
meanwhile, the network structure is continuously adjusted by using back propagation, and the error function of the back propagation neural network can be expressed as:
wherein E is y For the mean square error of the back propagation neural network, n k For the total number of samples,to be the predicted value of drilling speed, y r To obtain the actual value of the drilling speed, the parameters (w) of the back propagation neural network are updated by using a gradient descent method jr ,g r )。
Further, in the model updating module, the model updating process is as follows:
the lithology change and the time interval are simultaneously used as model updating conditions, a sliding window strategy is used for updating a drilling rate prediction model in real time on the basis of stratum and drilling multisource information fusion, and the model updating conditions are expressed as:
where u () is the switching variable of the rate of penetration prediction model update, Δu r State variables, deltau, updated for lithology changes t State variable updated for time interval, r s And r s+1 For formation drillability corresponding to two successive moments, r z Updating threshold for lithology change, t a And t a+1 For two successive time points, t b Is the time interval update threshold, when Deltau r Greater than or equal to r z When the drilling speed prediction model is used, updating according to formation lithology information; when Deltau t Equal to t b When the drilling speed prediction model is used, the drilling speed prediction model is updated by continuously learning the real drilling data under the condition of taking time as a reference;
when one of the above conditions is satisfied, the drilling rate prediction model is updated and predicted in real time.
Further, in the drilling rate prediction module, the prediction performance of the method is evaluated by using a model evaluation index, and the specific formula is as follows:
wherein RMSE is the root mean square error, NRMSE is the normalized root mean square error, y r Is the actual value of the drilling rate,is the predicted value of drilling speed, r is the r-th drilling data sample, r=1, 2 …, n k ，n k Representing the total number of samples.
The invention has the beneficial effects based on the technical scheme that:
(1) Screening and filtering drilling data by outlier rejection and wavelet filtering technology, identifying stratum drillability information by lithology identification technology, and providing effective data support for drilling rate modeling and model updating;
(2) An offline drilling rate prediction model is established by adopting a mixed bat algorithm optimization-limited Boltzmann machine-back propagation neural network, and the model has strong learning ability and nonlinear fitting ability and is more suitable for a complex stratum environment;
(3) The lithology change and the time interval are used as model updating conditions, and the drilling rate prediction model is updated in real time through a sliding window strategy on the basis of stratum and drilling multisource information fusion, so that the prediction accuracy of the drilling rate is improved, and the method is beneficial to being applied to actual production.
Drawings
FIG. 1 is a flow chart of a method of an embodiment of the present invention;
FIG. 2 illustrates a model framework for predicting drilling rate;
FIG. 3 Dendonee drilling data distribution;
FIG. 4 filtered Dendonee drilling data distribution;
FIG. 5 formation drillability profile;
fig. 6 shows the result of comparing the actual drilling rate with the predicted drilling rate.
Detailed Description
The achievement of the objects, functional features and advantages of the present invention will be further described with reference to the accompanying drawings, in conjunction with the embodiments.
The embodiment provides a drilling process drilling rate online prediction method based on multi-source information fusion, wherein a flow chart and a frame chart are respectively shown in fig. 1 and fig. 2. The invention divides modeling work into three phases: preprocessing drilling data, optimizing drilling speed modeling and updating a drilling speed prediction model. In the first stage, the abnormal value rejection and wavelet filtering technology is used for filtering the weight on bit, the rotating speed and the torque. In the second stage, an offline drilling rate prediction model is established by adopting a hybrid bat algorithm optimization-limited boltzmann machine-back propagation neural network. And in the third stage, the lithology change and the time interval are used as updating conditions, and a sliding window strategy is used for updating the drilling speed prediction model on the basis of stratum and drilling multisource information fusion so as to realize online prediction of the high-precision drilling speed.
The method comprises the following specific steps:
s1: selecting the drilling pressure, the rotating speed and the torque as the input of a drilling speed prediction model, screening and filtering drilling data by using outlier rejection and wavelet filtering analysis, acquiring a stratum drillability identification value by using a lithology identification technology, and judging the lithology change condition of the stratum;
s11: screening drilling data by using an outlier rejection method based on engineering experience, wherein the drilling data comprises weight on bit, rotating speed, torque and drilling rate, and the measuring range of the drilling data is as follows:
WOB is weight on bit, its unit is KN, RPM is rotational speed, its unit is RPM, torque is Torque, its unit is Nm, ROP is rate of drilling, its unit is cm/min;
s12: and filtering peaks and burrs in drilling data by adopting a wavelet filtering method, wherein a wavelet transformation expression is as follows:
wherein W is f (a, b) is drilling characteristic information after wavelet positive transformation, f (t) is original drilling data, a is a scaling factor, b is a scale factor, t is time, ψ () is a wavelet basis function, and a wavelet inverse transformation expression is:
wherein g (t) is drilling data after wavelet inverse transformation, c ψ Is a wavelet factor;
s13: and marking the drilled core, analyzing the core in a proper environment by using a lithology recognition technology to acquire stratum drillability information, and simultaneously judging whether the stratum lithology changes in real time according to the stratum drillability information at continuous moments.
S2: introducing a mixed bat algorithm optimization-a limited Boltzmann machine-a back propagation neural network algorithm, and establishing an offline drilling speed prediction model facing the complex stratum environment through three steps of mixed bat algorithm optimization, limited Boltzmann machine training, back propagation neural network training and back fine tuning;
s21: introducing trial-and-error and mixed bat algorithm to optimize super parameters of a drilling rate prediction model, wherein the super parameters comprise: batch size, training iteration number, hidden layer neuron number, learning rate and momentum;
s22: extracting key features in drilling data by training a limited boltzmann machine, wherein the limited boltzmann machine comprises a visual layer v and a hidden layer h, and an energy function under a given state (v, h) is defined as follows:
wherein E is θ (v, h) is the energy function of the limited boltzmann machine, θ= { w ij ,c i ,d j Model parameters of the restricted boltzmann machine, p and q are the number of neurons of the visual and hidden layers, c i And d j Bias for the ith visual layer neuron and the jth hidden layer neuron, v i Is the ith visual layer neuron, h j Is the j-th hidden layer neuron, w ij For the weight between the ith visual layer neuron and the jth hidden layer neuron, i=1, 2 …, p, j=1, 2 …, q;
given drilling training set Is the nth k The log likelihood function L (θ) of the restricted boltzmann machine on K is set to:
where r is the r-th drilling data sample, r=1, 2 …, n k ，n k V is the total number of samples r For the r-th visual layer, P (v r ) For visual layer v r Is a marginal probability distribution of (1); meanwhile, a contrast divergence algorithm is adopted to quickly train the limited Boltzmann machine so as to ensure the timeliness of calculation;
s23: taking the output characteristics of the constrained boltzmann machine as input to a back propagation neural network that includes two parts, forward propagation and back propagation, the forward propagation can be expressed as:
wherein, the liquid crystal display device comprises a liquid crystal display device,for the prediction of the drilling rate of the forward transmission output, f () is the activation function, g r For output layer bias, k tableShows the kth hidden layer node, k=1, 2 …, n, n is the total number of hidden layer nodes, w kr To conceal the connection weight between the layer and the output layer, x k Is a hidden layer node;
meanwhile, the network structure is continuously adjusted by using back propagation, and the error function of the back propagation neural network can be expressed as:
wherein E is y For the mean square error of the back propagation neural network, n k For the total number of samples,to be the predicted value of drilling speed, y r To obtain the actual value of the drilling speed, the parameters (w) of the back propagation neural network are updated by using a gradient descent method kr ,g r )。
S3: taking lithology change and time interval as drilling speed prediction model updating conditions, and applying a sliding window strategy to update the drilling speed prediction model in real time on the basis of stratum drillability information and drilling process information;
the lithology change and the time interval are simultaneously used as model updating conditions, a sliding window strategy is used for updating a drilling rate prediction model in real time on the basis of stratum and drilling multisource information fusion, and the model updating conditions are expressed as:
where u () is the switching variable of the rate of penetration prediction model update, Δu r State variables, deltau, updated for lithology changes t State variable updated for time interval, r s And r s+1 For formation drillability corresponding to two successive moments, r z Updating threshold for lithology change, t a And t a+1 For two successive time points, t b Is the time interval update threshold, when Deltau r Greater than or equal toAt r z When the drilling speed prediction model is used, updating according to formation lithology information; when Deltau t Equal to t b When the drilling speed prediction model is used, the drilling speed prediction model is updated by continuously learning the real drilling data under the condition of taking time as a reference;
when one of the above conditions is satisfied, the drilling rate prediction model is updated and predicted in real time.
S4: and inputting the actually obtained drilling weight, rotation speed and torque into the updated drilling speed prediction model to obtain the drilling speed at the next moment.
The method also comprises the step of evaluating the prediction performance of the method by using model evaluation indexes, wherein the specific formula is as follows:
wherein RMSE is the root mean square error, NRMSE is the normalized root mean square error, y r Is the actual value of the drilling rate,is the predicted value of drilling speed, r is the r-th drilling data sample, r=1, 2 …, n k ，n k Is the total number of samples. The smaller the root mean square error and the normalized root mean square error, the better the fitting effect of the predicted drilling rate and the actual drilling rate, and the stronger the prediction capability of the drilling rate prediction model.
The embodiment uses industrial data of drilling sites in northeast Dandong areas of China as an example, and the specific process is as follows:
(1) The actual drilling data (weight on bit, rotation speed, torque and drilling rate) of 122383 group of drilling wells in the eastern Dandong region of China are selected, the data distribution situation is shown in figure 3, the data are displayed in a histogram, and the quality of the drilling data of the drilling wells is very low. The rate of penetration prediction model inputs (weight, speed and torque) and outputs (rate of penetration) are non-gaussian and there are a large number of outliers. For example, some weight and torque measurements exceed 500kN and 60000Nm, respectively, and a few bit rate measurements even exceed 6000cm/min.
Drilling data is screened and filtered by using two preprocessing techniques, as shown in stage 1 in fig. 2, in the first data preprocessing technique, drilling data is screened by using an outlier rejection method in combination with artificial experience, and in the second data preprocessing technique, drilling data is noise-reduced by using a wavelet filtering method. The filtered dandong drilling data distribution is shown in fig. 4, and peaks and burrs in the data are effectively removed. Meanwhile, the formation drillability information is identified by using the lithology identification technology, the formation drillability distribution is shown in fig. 5, and the formation drillability identification value is mainly concentrated between 7 and 10, which indicates that the formation in the current area is relatively hard and is not beneficial to the rapid rock breaking of the drill bit;
(2) An off-line drilling rate prediction model is built using a hybrid batalgorithm optimization-limited boltzmann machine-counter-propagating neural network, as shown in stage 2 in fig. 2, first, the limited boltzmann machine is trained to extract key features of drilling data. Second, the output characteristics of the constrained boltzmann machine are taken as input to the back propagation neural network, and the back propagation neural network is trained and back-tuned. Finally, introducing trial-and-error and mixed bat algorithm to optimize model super parameters, wherein the model optimal super parameters of the method are shown in table 1;
TABLE 1 model optimal superparameter for the method of the present invention
(3) Using lithology change and time interval as model updating conditions, updating a drilling rate prediction model through a sliding window strategy on the basis of stratum and drilling multisource information fusion, as shown in a stage 3 in fig. 2;
(4) To test the effectiveness of the proposed method, it was validated against seven methods of prediction of drilling rate (two off-line and five on-line), the definition of which is shown in table 2. Tables 3 and 6 present the results of comparing the actual rate of penetration with the predicted rate of penetration.
TABLE 2 definition of the method of the invention and seven methods of comparison of the predicted rates of penetration
TABLE 3 comparison of the proposed method with seven methods of prediction of drilling rates
From this, it can be seen that the method disclosed by the invention has excellent performance in capturing the variation trend of the drilling rate, the root mean square error and the normalized root mean square error are 0.0862 and 3.36% respectively, and the prediction accuracy is improved by at least 13% compared with the seven drilling rate prediction methods (M1, M2, M3, M4, M5, M6 and M7). The method can meet the drilling engineering requirement, is beneficial to shortening the drilling period, reducing the operation cost, realizing high-precision prediction of the drilling rate, has effectiveness, and lays an important foundation for intelligent optimization control of the drilling process.
The invention has the beneficial effects that:
(1) Screening and filtering drilling data by outlier rejection and wavelet filtering technology, identifying stratum drillability information by lithology identification technology, and providing effective data support for drilling rate modeling and model updating;
(2) An offline drilling rate prediction model is established by adopting a mixed bat algorithm optimization-limited Boltzmann machine-back propagation neural network, and the model has strong learning ability and nonlinear fitting ability and is more suitable for a complex stratum environment;
(3) The lithology change and the time interval are used as model updating conditions, and the drilling rate prediction model is updated in real time through a sliding window strategy on the basis of stratum and drilling multisource information fusion, so that the prediction accuracy of the drilling rate is improved, and the method is beneficial to being applied to actual production.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or system. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or system that comprises the element.
The foregoing embodiment numbers of the present invention are merely for the purpose of description, and do not represent the advantages or disadvantages of the embodiments. In the unit claims enumerating several means, several of these means may be embodied by one and the same item of hardware. The use of the terms first, second, third, etc. do not denote any order, but rather the terms first, second, third, etc. are used to interpret the terms as labels.
The foregoing description is only of the preferred embodiments of the present invention, and is not intended to limit the scope of the invention, but rather is intended to cover any equivalents of the structures or equivalent processes disclosed herein or in the alternative, which may be employed directly or indirectly in other related arts.
Claims (10)
1. The drilling process drilling rate online prediction method based on multi-source information fusion is characterized by comprising the following steps of:
s1: selecting the drilling pressure, the rotating speed and the torque as the input of a drilling speed prediction model, screening and filtering drilling data by using outlier rejection and wavelet filtering analysis, acquiring a stratum drillability identification value by using a lithology identification technology, and judging the lithology change condition of the stratum;
s2: introducing a mixed bat algorithm optimization-a limited Boltzmann machine-a back propagation neural network algorithm, and establishing an offline drilling speed prediction model facing the complex stratum environment through three steps of mixed bat algorithm optimization, limited Boltzmann machine training, back propagation neural network training and back fine tuning;
s3: taking lithology change and time interval as drilling speed prediction model updating conditions, and applying a sliding window strategy to update the drilling speed prediction model in real time on the basis of stratum drillability information and drilling process information;
s4: and inputting the actually obtained drilling weight, rotation speed and torque into the updated drilling speed prediction model to obtain the drilling speed at the next moment.
2. The method for online prediction of drilling rate of a drilling process based on multi-source information fusion according to claim 1, wherein the step S1 specifically comprises the following steps:
s11: screening drilling data by using an outlier rejection method based on engineering experience, wherein the drilling data comprises weight on bit, rotating speed, torque and drilling rate, and the measuring range of the drilling data is as follows:
WOB is weight on bit, its unit is KN, RPM is rotational speed, its unit is RPM, torque is Torque, its unit is Nm, ROP is rate of drilling, its unit is cm/min;
s12: and filtering peaks and burrs in drilling data by adopting a wavelet filtering method, wherein a wavelet transformation expression is as follows:
wherein W is f (a, b) is drilling characteristic information after wavelet positive transformation, f (t) is original drilling data, and a is extensionThe reduction factor, b is the scale factor, t is the time, ψ is the wavelet basis function, and the wavelet inverse transformation expression is:
wherein g (t) is drilling data after wavelet inverse transformation, c ψ Is a wavelet factor;
s13: and marking the drilled core, analyzing the core in a proper environment by using a lithology recognition technology to acquire stratum drillability information, and simultaneously judging whether the stratum lithology changes in real time according to the stratum drillability information at continuous moments.
3. The method for online prediction of drilling rate of a drilling process based on multi-source information fusion according to claim 1, wherein the step S2 specifically comprises the following steps:
s21: introducing trial-and-error and mixed bat algorithm to optimize super parameters of a drilling rate prediction model, wherein the super parameters comprise: batch size, training iteration number, hidden layer neuron number, learning rate and momentum;
s22: extracting key features in drilling data by training a limited boltzmann machine, wherein the limited boltzmann machine comprises a visual layer v and a hidden layer h, and an energy function under a given state (v, h) is defined as follows:
wherein E is θ (v, h) is the energy function of the limited boltzmann machine, θ= { w ij ，c i ，d j Model parameters of the restricted boltzmann machine, p and q are the number of neurons of the visual and hidden layers, c i And d j Bias for the ith visual layer neuron and the jth hidden layer neuron, v i Is the ith visual layer neuron, h j Is the j-th hidden layer neuron, w ij For the ith visual layer neuron sumThe weight between the j-th hidden layer neurons, i=1, 2 …, p, j=1, 2 …, q;
given drilling training set Is the nth k The log likelihood function L (θ) of the restricted boltzmann machine on K is set to:
where r is the r-th drilling data sample, r=1, 2 …, n k ，n k V is the total number of samples r For the r-th visual layer, P (v r ) For visual layer v r Is a marginal probability distribution of (1);
s23: taking the output characteristics of the constrained boltzmann machine as input to a back propagation neural network that includes two parts, forward propagation and back propagation, the forward propagation can be expressed as:
wherein, the liquid crystal display device comprises a liquid crystal display device,for the prediction of the drilling rate of the forward transmission output, f () is the activation function, g r For output layer bias, k represents the kth hidden layer node, k=1, 2 …, n, n is the total number of hidden layer nodes, w kr To conceal the connection weight between the layer and the output layer, x k Is a hidden layer node;
meanwhile, the network structure is continuously adjusted by using back propagation, and the error function of the back propagation neural network can be expressed as:
wherein E is y For the mean square error of the back propagation neural network, n k For the total number of samples,to be the predicted value of drilling speed, y r To obtain the actual value of the drilling speed, the parameters (w) of the back propagation neural network are updated by using a gradient descent method kr ，g r )。
4. The drilling process drilling rate online prediction method based on multi-source information fusion according to claim 1, wherein the specific process of step S3 is as follows:
the lithology change and the time interval are simultaneously used as model updating conditions, a sliding window strategy is used for updating a drilling rate prediction model in real time on the basis of stratum and drilling multisource information fusion, and the model updating conditions are expressed as:
where u () is the switching variable of the rate of penetration prediction model update, Δu r State variables, deltau, updated for lithology changes t State variable updated for time interval, r s And r s+1 For formation drillability corresponding to two successive moments, r z Updating threshold for lithology change, t a And t a+1 For two successive time points, t b Is the time interval update threshold, when Deltau r Greater than or equal to r z When the drilling speed prediction model is used, updating according to formation lithology information; when Deltau t Equal to t b When the drilling speed prediction model is used, the drilling speed prediction model is updated by continuously learning the real drilling data under the condition of taking time as a reference;
when one of the above conditions is met, the drilling rate prediction model is updated and predicted in real time.
5. The method for online prediction of drilling rate of a drilling process based on multi-source information fusion according to claim 1, wherein the method further comprises the step of evaluating the prediction performance of the method by using model evaluation indexes, wherein the specific formula is as follows:
wherein RMSE is the root mean square error, NRMSE is the normalized root mean square error, y r Is the actual value of the drilling rate,is the predicted value of drilling speed, r is the r-th drilling data sample, r=1, 2 …, n k ，n k Is the total number of samples.
6. The drilling process drilling rate online prediction system based on multi-source information fusion is characterized by comprising:
the data preprocessing module is used for selecting the drilling pressure, the rotating speed and the torque as the input of a drilling speed prediction model, screening and filtering drilling data by using outlier rejection and wavelet filtering analysis, acquiring a stratum drillability identification value by using a lithology identification technology, and judging the lithology change condition of the stratum;
the model building module is used for introducing a mixed bat algorithm optimization-limited Boltzmann machine-back propagation neural network algorithm, and building an offline drilling speed prediction model facing the complex stratum environment through three steps of mixed bat algorithm optimization, limited Boltzmann machine training, back propagation neural network training and back fine tuning;
the model updating module is used for taking lithology change and time interval as the updating conditions of the drilling speed prediction model and applying a sliding window strategy to update the drilling speed prediction model in real time on the basis of stratum drillability information and drilling process information;
and the drilling speed prediction module is used for inputting the actually acquired drilling weight, rotation speed and torque into the updated drilling speed prediction model to obtain the drilling speed at the next moment.
7. The drilling process drilling rate online prediction system based on multi-source information fusion according to claim 6, wherein the data preprocessing module performs preprocessing on drilling data as follows:
s11: screening drilling data by using an outlier rejection method based on engineering experience, wherein the drilling data comprises weight on bit, rotating speed, torque and drilling rate, and the measuring range of the drilling data is as follows:
WOB is weight on bit, its unit is KN, RPM is rotational speed, its unit is RPM, torque is Torque, its unit is Nm, ROP is rate of drilling, its unit is cm/min;
s12: and filtering peaks and burrs in drilling data by adopting a wavelet filtering method, wherein a wavelet transformation expression is as follows:
wherein W is f (a, b) is drilling characteristic information after wavelet positive transformation, f (t) is original drilling data, a is a scaling factor, b is a scale factor, t is time, ψ () is a wavelet basis function, and a wavelet inverse transformation expression is:
wherein g (t) is drilling data after wavelet inverse transformation, c ψ Is a wavelet factor;
s13: and marking the drilled core, analyzing the core in a proper environment by using a lithology recognition technology to acquire stratum drillability information, and simultaneously judging whether the stratum lithology changes in real time according to the stratum drillability information at continuous moments.
8. The drilling process drilling rate online prediction system based on multi-source information fusion according to claim 6, wherein in the modeling module, the modeling process is as follows:
s21: introducing trial-and-error and mixed bat algorithm to optimize super parameters of a drilling rate prediction model, wherein the super parameters comprise: batch size, training iteration number, hidden layer neuron number, learning rate and momentum;
s22: extracting key features in drilling data by training a limited boltzmann machine, wherein the limited boltzmann machine comprises a visual layer v and a hidden layer h, and an energy function under a given state (v, h) is defined as follows:
wherein E is θ (v, h) is the energy function of the limited boltzmann machine, θ= { w ij ，c i ，d j Model parameters of the restricted boltzmann machine, p and q are the number of neurons of the visual and hidden layers, c i And d j Bias for the ith visual layer neuron and the jth hidden layer neuron, v i Is the ith visual layer neuron, h j Is the j-th hidden layer neuron, w ij For the weight between the ith visual layer neuron and the jth hidden layer neuron, i=1, 2 …, p, j=1, 2 …, q;
given drilling training set Is the nth k The log likelihood function L (θ) of the restricted boltzmann machine on K is set to:
where r is the r-th drilling data sample, r=1, 2 …, n k ，n k V is the total number of samples r For the r-th visual layer, P (v r ) For visual layer v r Is a marginal probability distribution of (1);
s23: taking the output characteristics of the constrained boltzmann machine as input to a back propagation neural network that includes two parts, forward propagation and back propagation, the forward propagation can be expressed as:
wherein, the liquid crystal display device comprises a liquid crystal display device,for the prediction of the drilling rate of the forward transmission output, f () is the activation function, g r For output layer bias, k represents the kth hidden layer node, k=1, 2 …, n, n is the total number of hidden layer nodes, w kr To conceal the connection weight between the layer and the output layer, x k Is a hidden layer node;
meanwhile, the network structure is continuously adjusted by using back propagation, and the error function of the back propagation neural network can be expressed as:
wherein E is y For the mean square error of the back propagation neural network, n k For the total number of samples,to be the predicted value of drilling speed, y r To obtain the actual value of the drilling speed, the parameters (w) of the back propagation neural network are updated by using a gradient descent method kr ，g r )。
9. The drilling process drilling rate online prediction system based on multi-source information fusion according to claim 6, wherein in the update model module, the update model process is as follows:
the lithology change and the time interval are simultaneously used as model updating conditions, a sliding window strategy is used for updating a drilling rate prediction model in real time on the basis of stratum and drilling multisource information fusion, and the model updating conditions are expressed as:
where u () is the switching variable of the rate of penetration prediction model update, Δu r State variables, deltau, updated for lithology changes t State variable updated for time interval, r s And r s+1 For formation drillability corresponding to two successive moments, r z Updating threshold for lithology change, t a And t a+1 For two successive time points, t b Is the time interval update threshold, when Deltau r Greater than or equal to r z When the drilling speed prediction model is used, updating according to formation lithology information; when Deltau t Equal to t b When the drilling speed prediction model is used, the drilling speed prediction model is updated by continuously learning the real drilling data under the condition of taking time as a reference;
when one of the above conditions is satisfied, the drilling rate prediction model is updated and predicted in real time.
10. The system for online prediction of drilling rate of a drilling process based on multi-source information fusion according to claim 6, wherein the method further comprises the step of evaluating the prediction performance of the method by using model evaluation indexes, wherein the specific formula is as follows:
wherein RMSE is the root mean square error, NRMSE is the normalized root mean square error, y r Is the actual value of the drilling rate,is the predicted value of drilling speed, r is the r-th drilling data sample, r=1, 2 …, n k ，n k Representing the total number of samples.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310483771.XA CN116522777A (en) | 2023-04-28 | 2023-04-28 | Drilling process drilling rate online prediction method and system based on multi-source information fusion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310483771.XA CN116522777A (en) | 2023-04-28 | 2023-04-28 | Drilling process drilling rate online prediction method and system based on multi-source information fusion |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116522777A true CN116522777A (en) | 2023-08-01 |
Family
ID=87395538
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310483771.XA Pending CN116522777A (en) | 2023-04-28 | 2023-04-28 | Drilling process drilling rate online prediction method and system based on multi-source information fusion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116522777A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117211969A (en) * | 2023-10-17 | 2023-12-12 | 江苏省无锡探矿机械总厂有限公司 | Energy-saving control method and system for hydraulic drilling machine |
-
2023
- 2023-04-28 CN CN202310483771.XA patent/CN116522777A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117211969A (en) * | 2023-10-17 | 2023-12-12 | 江苏省无锡探矿机械总厂有限公司 | Energy-saving control method and system for hydraulic drilling machine |
| CN117211969B (en) * | 2023-10-17 | 2024-03-29 | 江苏省无锡探矿机械总厂有限公司 | Energy-saving control method and system for hydraulic drilling machine |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Negash et al. | Artificial neural network based production forecasting for a hydrocarbon reservoir under water injection | |
| CA2966043C (en) | Method and apparatus for drilling a new well using historic drilling data | |
| Wei et al. | Predicting injection profiles using ANFIS | |
| AU2012370482B2 (en) | Determining optimal parameters for a downhole operation | |
| CN107545577B (en) | Sedimentary facies image segmentation method based on neural network | |
| Nikravesh | Soft computing-based computational intelligent for reservoir characterization | |
| CN113689055B (en) | Oil-gas drilling machinery drilling speed prediction and optimization method based on Bayesian optimization | |
| CN116522777A (en) | Drilling process drilling rate online prediction method and system based on multi-source information fusion | |
| CN106803010A (en) | For the Fuzzy Grey comprehensive evaluation method and device of low permeability reservoir quantitative assessment | |
| CN101266299A (en) | Method for forecasting oil gas utilizing earthquake data object constructional features | |
| Chahar et al. | Data-driven approach for hydrocarbon production forecasting using machine learning techniques | |
| CN110471104A (en) | Poststack seismic reflection mode identification method based on intelligent characteristic study | |
| Shan et al. | Identification of flow units using the joint of WT and LSSVM based on FZI in a heterogeneous carbonate reservoir | |
| CN115481565A (en) | Earth pressure balance shield tunneling parameter prediction method based on LSTM and ant colony algorithm | |
| Chandrasekaran et al. | Drilling efficiency improvement and rate of penetration optimization by machine learning and data analytics | |
| CN114004100B (en) | Oil reservoir assisted history fitting and optimization simulation method | |
| Qodirov et al. | Development of artificial neural network for predicting drill pipe sticking in real-time well drilling process | |
| Bahari et al. | Intelligent drilling rate predictor | |
| CN113494286A (en) | Intelligent dynamic prediction method and system for drilling speed in geological drilling process | |
| CN113323676A (en) | Method for determining cutter head torque of shield tunneling machine by using principal component analysis-length memory model | |
| Xue et al. | Gas well performance prediction using deep learning jointly driven by decline curve analysis model and production data. | |
| CN115434696A (en) | Lithology identification database construction method and lithology identification method | |
| Qin et al. | RCLSTMNet: A Residual-convolutional-LSTM Neural Network for Forecasting Cutterhead Torque in Shield Machine | |
| CN111749675A (en) | Stratum drillability prediction method and system based on cascade model algorithm | |
| Gan et al. | Multi-source information fusion-based dynamic model for online prediction of rate of penetration (ROP) in drilling process |
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 |