Construction method of fracturing difficulty prediction model, prediction method and related device

Abstract

The invention relates to the technical field of petroleum engineering fracturing prediction, in particular to a construction method of a fracturing difficulty prediction model, a prediction method and a related device, and the construction method comprises the following steps: constructing a training sample and a test sample; selecting a kernel function, and combining a maximum likelihood estimation method to complete Gaussian process regression training by using the training sample; iteratively correcting the fracturing difficulty level prediction model by using the test sample, and outputting the fracturing difficulty level prediction model meeting the precision requirement; and selecting different kernel functions, comparing prediction errors of the fracturing difficulty level prediction models corresponding to the different kernel functions by using the test samples, and outputting the fracturing difficulty level prediction model with the minimum error. According to the method, fracturing construction basic data are combined, the fracturing difficulty degree of the non-fractured section is predicted, early warning is carried out on the stage section with high construction difficulty and high construction risk in advance, constructors are assisted to make preparations in advance, the construction risk is reduced, and the fracturing construction efficiency is improved.

Description

Technical Field

[0001] The invention relates to the technical field of petroleum engineering fracturing prediction, and is a construction method of a fracturing difficulty prediction model, a prediction method and a related device. Background Art

[0002] At present, unconventional and ultra-low permeability resources account for more than 85% of the newly discovered reserves of oil and natural gas in my country. As the old oil fields in the east generally enter the middle and late stages of production, the development of unconventional resources as an important strategic successor resource is imperative. For unconventional resources, due to their ultra-low porosity (porosity <10%) and ultra-low permeability (permeability <0.1mD) resource characteristics, hydraulic fracturing technology cannot be avoided in the process of large-scale development. Fracturing has been listed as one of the three core technologies of petroleum engineering along with geophysical prospecting (seismic method) and drilling.

[0003] During the fracturing construction of unconventional reservoirs, downhole accidents such as pressure channeling, casing change or casing damage often occur, which makes the fracturing construction unable to proceed normally, seriously affecting the construction progress and the effect of reservoir production enhancement. At present, during the fracturing construction process, there is no relevant technology that can reflect the key parameters of the fracturing difficulty through the fracturing section, evaluate the fracturing difficulty of the fracturing section, predict the fracturing difficulty of the subsequent construction sections, and give early warning for the sections with high construction difficulty and high construction risk. Summary of the invention

[0004] The present invention provides a method for constructing a fracturing difficulty prediction model, a prediction method and related devices, which overcome the shortcomings of the above-mentioned prior art and can effectively solve the problem in the prior art that the fracturing difficulty of an unfractured section cannot be predicted based on the basic fracturing construction data of the fracturing section.

[0005] One of the technical solutions of the present invention is achieved by the following measures: a method for constructing a fracturing difficulty prediction model, comprising:

[0006] Construct training samples and test samples, wherein the training samples include the fracturing difficulty grading results and basic fracturing construction data of the first n-stage fracturing sections, based on the basic fracturing construction data of the first n-stage fracturing sections, the first n-stage fracturing sections are graded for fracturing difficulty using the normalized index weighted average and threshold value, and the fracturing difficulty grading results of the first n-stage fracturing sections are obtained, and the test samples include the fracturing difficulty grading results and basic fracturing construction data of the n+1-stage fracturing sections and the n+2-stage fracturing sections, and the basic fracturing construction data include the maximum displacement and maximum construction pressure in the section, the pressure before the slug and the displacement before the slug in the section, the time to reach the maximum displacement in the pre-fluid stage in the section, and the maximum pressure and maximum displacement in the pre-fluid stage in the section;

[0007] The kernel function is selected, and the Gaussian process regression training is completed using the training samples in combination with the maximum likelihood estimation method to obtain a fracturing difficulty prediction model;

[0008] Iteratively modify the fracturing difficulty prediction model using test samples, and output a fracturing difficulty prediction model that meets accuracy requirements;

[0009] Select different kernel functions, repeat the above steps, use the test samples to compare the prediction errors of the fracturing difficulty prediction models corresponding to different kernel functions, and output the fracturing difficulty prediction model with the smallest error.

[0010] The following are further optimizations and / or improvements to the above technical solutions:

[0011] The above-mentioned method of iteratively correcting the fracturing difficulty prediction model using the test samples and outputting a fracturing difficulty prediction model that meets the accuracy requirements includes:

[0012] Input the basic data of fracturing construction of the n+1th fracturing stage into the fracturing difficulty prediction model, and output the corresponding predicted weighted average value;

[0013] The threshold value is used to determine the classification prediction result of the fracturing difficulty, and compared with the classification result of the n+1th level fracturing stage in the test sample to determine whether the classification prediction result is correct;

[0014] In response to incorrectness, the threshold value is adjusted, and the n+1th level fracturing stage is included in the training sample, the first n+1 level fracturing stages are reclassified for fracturing difficulty, the training sample and the test sample are updated, and the training sample is used to repeat the training to obtain a fracturing difficulty prediction model;

[0015] If the response is correct, the n+1th level fracturing stage is included in the training sample, the training sample and the test sample are updated, and the training sample is used to repeat the training to obtain a fracturing difficulty prediction model;

[0016] The retrained fracturing difficulty prediction model is used to predict the fracturing difficulty of the n+2-th fracturing stage, and the confidence level corresponding to the fracturing difficulty prediction result is determined;

[0017] Determine whether the confidence is greater than the confidence setting value, and if so, output the corresponding fracturing difficulty prediction model, and if not, repeat the above steps until the confidence is greater than the confidence setting value, and output the corresponding fracturing difficulty prediction model.

[0018] The above training samples are constructed, including:

[0019] Obtaining basic fracturing construction data of the first n-stage fracturing stages, wherein the basic fracturing construction data includes the maximum displacement and maximum construction pressure in the stage, the pressure before the slug and the displacement before the slug in the stage, the time when the maximum displacement is reached in the pre-fluid stage in the stage, and the maximum pressure and maximum displacement in the pre-fluid stage in the stage;

[0020] Based on the basic data of fracturing construction, the displacement-pressure index of the first n-stage fracturing stages, the slug front pressure-displacement ratio, and the ratio of the maximum pressure to the maximum displacement in the pre-fluid stage are obtained;

[0021] The four indicators of displacement-pressure index of the first n-stage fracturing stages, the slug front pressure displacement ratio, the ratio of the maximum pressure to the maximum displacement in the pre-fluid stage, and the time to reach the maximum displacement in the pre-fluid stage are normalized;

[0022]

[0023] Among them, x is the data that needs to be normalized; x min is the smallest data in the data set; x max It is the largest data in the dataset;

[0024] Get the weighted average of the four normalized indicators of the first n fracturing stages

[0025]

[0026] in, is the normalized displacement-pressure index; is the normalized slug front pressure ratio; It is the normalized ratio of the maximum pressure to the maximum displacement in the pre-fluid stage; is the time for the normalized pre-fluid stage to reach the maximum displacement;

[0027] Set the threshold value to Compared with the threshold value, the fracturing difficulty classification results of the first n fracturing stages;

[0028] The training samples are constructed by combining the fracturing difficulty classification results of the first n fracturing stages and the basic data of fracturing construction.

[0029] Based on the basic data of fracturing construction, the displacement-pressure index of the first n-stage fracturing stages, the slug front pressure-displacement ratio and the ratio of the maximum pressure to the maximum displacement in the pre-fluid stage are obtained, including:

[0030] The displacement-pressure index of the first n-stage fracturing stage is calculated by the following formula:

[0031]

[0032] Where F is the displacement-pressure index; Q i,max (i=1,2,...,n) is the maximum displacement in the first n-stage fracturing stage; P i,max (i=1,2,...,n) is the maximum operation pressure in the first n-stage fracturing stage;

[0033] The slug front pressure ratio of the first n-stage fracturing stage is calculated by the following formula:

[0034]

[0035] Among them, R i,slug is the slug front pressure ratio; P i,slug (i=1,2,...,n) is the pre-slug pressure in the first n-stage fracturing stage; Q i,slug (i=1,2,...,n) is the slug front displacement in the first n-stage fracturing stage;

[0036] The ratio of the maximum pressure and the maximum displacement of the pre-fluid stage in the first n-stage fracturing stage is calculated by the following formula:

[0037]

[0038] Among them, R i,pad It is the ratio of the maximum pressure to the maximum displacement in the pre-liquid stage; P i,pad (i=1,2,...,n) is the maximum pressure of the pre-fluid stage in the first n-stage fracturing stage; Q i,pad (i=1, 2, ..., n) is the maximum displacement of the pre-fluid stage in the first n fracturing stages.

[0039] The second technical solution of the present invention is achieved by the following measures: a method for predicting the difficulty of fracturing, comprising:

[0040] Obtaining basic fracturing construction data of the predicted fracturing section, wherein the basic fracturing construction data includes the maximum displacement and maximum construction pressure in the section, the pressure before the slug and the displacement before the slug in the section, the time when the maximum displacement is reached in the pre-fluid stage in the section, and the maximum pressure and maximum displacement in the pre-fluid stage in the section;

[0041] Input the basic data of fracturing construction of the predicted fracturing section into the fracturing difficulty prediction model and output the corresponding weighted average value. The fracturing difficulty prediction model is constructed according to the construction method of the fracturing difficulty prediction model;

[0042] The weighted average The result is compared with the threshold value corresponding to the fracturing difficulty prediction model to determine the fracturing difficulty classification result of the fracturing section to be predicted.

[0043] The third technical solution of the present invention is achieved by the following measures: a device for constructing a fracturing difficulty prediction model, comprising:

[0044] A sample construction unit, constructing training samples and test samples, wherein the training samples include the fracturing difficulty grading results and basic fracturing construction data of the first n-stage fracturing sections, based on the basic fracturing construction data of the first n-stage fracturing sections, the first n-stage fracturing sections are graded for fracturing difficulty using the normalized index weighted average and threshold value, and the fracturing difficulty grading results of the first n-stage fracturing sections are obtained, and the test samples include the fracturing difficulty grading results and basic fracturing construction data of the n+1-stage fracturing sections and the n+2-stage fracturing sections, and the basic fracturing construction data include the maximum displacement and maximum construction pressure in the section, the pressure before the slug and the displacement before the slug in the section, the time to reach the maximum displacement in the pre-fluid stage in the section, and the maximum pressure and maximum displacement in the pre-fluid stage in the section;

[0045] The model training unit selects a kernel function and uses the training samples in combination with the maximum likelihood estimation method to complete Gaussian process regression training and obtain a fracturing difficulty prediction model;

[0046] A model testing unit, which uses the test samples to iteratively correct the fracturing difficulty prediction model and outputs a fracturing difficulty prediction model that meets the accuracy requirements;

[0047] The model output unit selects different kernel functions, repeats the above steps, uses the test samples to compare the prediction errors of the fracturing difficulty prediction models corresponding to different kernel functions, and outputs the fracturing difficulty prediction model with the smallest error.

[0048] The following are further optimizations and / or improvements to the above technical solutions:

[0049] The above-mentioned sample construction unit includes a training sample construction module and a test sample construction module, wherein the training sample construction module includes:

[0050] The sample data acquisition module acquires the basic data of the fracturing construction of the first n-stage fracturing stages, wherein the basic data of the fracturing construction includes the maximum displacement and the maximum construction pressure in the stage, the pressure before the slug and the displacement before the slug in the stage, the time when the maximum displacement is reached in the pre-fluid stage in the stage, and the maximum pressure and the maximum displacement in the pre-fluid stage in the stage;

[0051] The index acquisition module obtains the displacement-pressure index of the first n-stage fracturing stages, the slug front pressure-displacement ratio, and the ratio of the maximum pressure to the maximum displacement in the pre-fluid stage based on the basic data of the fracturing construction;

[0052] The first processing module normalizes the four indicators of the displacement-pressure index of the first n-stage fracturing stages, the slug front pressure displacement ratio, the ratio of the maximum pressure to the maximum displacement in the pre-fluid stage, and the time to reach the maximum displacement in the pre-fluid stage;

[0053]

[0054] Among them, x is the data that needs to be normalized; x min is the smallest data in the data set; x max It is the largest data in the dataset;

[0055] The second processing module obtains the weighted average of the four normalized indicators of the first n-stage fracturing stages

[0056]

[0057] in, is the normalized displacement-pressure index; is the normalized slug front pressure ratio; It is the normalized ratio of the maximum pressure to the maximum displacement in the pre-fluid stage; is the time for the normalized pre-fluid stage to reach the maximum displacement;

[0058] Comparison module, set the threshold value, the first n fracturing stages Compared with the threshold value, the fracturing difficulty classification results of the first n fracturing stages;

[0059] The sample formation module combines the fracturing difficulty classification results of the first n fracturing stages and the basic data of fracturing construction to construct training samples.

[0060] The fourth technical solution of the present invention is achieved by the following measures: a device for predicting the difficulty of fracturing, comprising:

[0061] A data acquisition unit is used to acquire basic fracturing construction data of the predicted fracturing section, wherein the basic fracturing construction data includes the maximum displacement and maximum construction pressure in the section, the pressure before the slug and the displacement before the slug in the section, the time when the maximum displacement is reached in the pre-fluid stage in the section, and the maximum pressure and maximum displacement in the pre-fluid stage in the section;

[0062] The prediction unit inputs the basic data of the fracturing construction of the fracturing section to be predicted into the fracturing difficulty prediction model and outputs the corresponding weighted average value. The fracturing difficulty prediction model is constructed according to the construction method of the fracturing difficulty prediction model;

[0063] Grading unit, the weighted average The result is compared with the threshold value corresponding to the fracturing difficulty prediction model to determine the fracturing difficulty classification result of the fracturing section to be predicted.

[0064] The beneficial effects of the present invention include:

[0065] (1) The present invention combines the basic data of fracturing construction (maximum displacement and maximum construction pressure within a segment, pressure before the slug and displacement before the slug within a segment, time to reach maximum displacement during the pre-fluid stage within a segment, maximum pressure and maximum displacement during the pre-fluid stage within a segment), and fully considers the influencing factors of various aspects of on-site construction to achieve the prediction of the difficulty of fracturing the unfractured segment, effectively reflecting the difficulty of fracturing construction, so as to give early warning to the stages with high construction difficulty and high construction risk, assist construction personnel to prepare in advance, reduce possible construction risks, improve fracturing construction efficiency, ensure fracturing construction effect, and enhance reservoir production increase and transformation effect.

[0066] (2) The Gaussian process regression model of the present invention does not need to define a fitting function with a definite form, but can effectively capture the nonlinear abstract mapping relationship between the dependent variable and the independent variable, and fully utilize the sample data for fitting and prediction. It is very suitable for low-dimensional and small sample regression prediction problems, making the model training fast and effective. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Attached Figure 1 The present invention is a flow chart of a model building method.

[0068] Attached Figure 2 The present invention is a flow chart of another model building method.

[0069] Attached Figure 3 This is a flow chart of the method for constructing training samples in the present invention.

[0070] Attached Figure 4 The figure is a flow chart of the method for predicting the difficulty of fracturing of the present invention.

[0071] Attached Figure 5 This is a schematic diagram of the prediction results of the difficulty of fracturing at each stage of the DaHW1506 well in an embodiment of the present invention.

[0072] Attached Figure 6 This is a schematic diagram of the prediction results of the difficulty of fracturing at each stage of the DaHW1507 well in an embodiment of the present invention.

[0073] Attached Figure 7 This is a schematic diagram of the prediction results of the difficulty of fracturing at each stage of the DaHW1508 well in an embodiment of the present invention.

[0074] Attached Figure 8 This is a schematic diagram of the prediction results of the difficulty of fracturing at each stage of the DaHW1509 well in an embodiment of the present invention.

[0075] Attached Fig. 9 It is a schematic diagram of the structure of the model building device of the present invention.

[0076] Attached Fig.10It is a structural diagram of the training sample construction module in the present invention.

[0077] Attached Fig.11 It is a schematic diagram of the structure of the device for predicting the difficulty of fracturing according to the present invention. DETAILED DESCRIPTION

[0078] The present invention is not limited by the following embodiments, and specific implementation methods can be determined based on the technical solution of the present invention and actual conditions.

[0079] The present invention will be further described below in conjunction with embodiments and drawings:

[0080] Embodiment 1: As attached Figure 1 As shown, the embodiment of the present invention discloses a method for constructing a fracturing difficulty prediction model, comprising:

[0081] Step S101, constructing training samples and test samples, wherein the training samples include the fracturing difficulty grading results and basic fracturing construction data of the first n-stage fracturing sections, based on the basic fracturing construction data of the first n-stage fracturing sections, the first n-stage fracturing sections are graded for fracturing difficulty using the normalized index weighted average and the threshold value, and the fracturing difficulty grading results of the first n-stage fracturing sections are obtained, and the test samples include the fracturing difficulty grading results and basic fracturing construction data of the n+1-stage fracturing section and the n+2-stage fracturing section, and the basic fracturing construction data include the maximum displacement and maximum construction pressure in the section, the pressure before the slug and the displacement before the slug in the section, the time to reach the maximum displacement in the pre-fluid stage in the section, and the maximum pressure and maximum displacement in the pre-fluid stage in the section;

[0082] Step S102, selecting a kernel function, and using training samples in combination with a maximum likelihood estimation method to complete Gaussian process regression training to obtain a fracturing difficulty prediction model;

[0083] Step S103, iteratively correcting the fracturing difficulty prediction model using the test samples, and outputting a fracturing difficulty prediction model that meets the accuracy requirements;

[0084] Step S104, select different kernel functions, repeat the above steps, use test samples to compare the prediction errors of the fracturing difficulty prediction models corresponding to different kernel functions, and output the fracturing difficulty prediction model with the smallest error.

[0085] In the above embodiment, in step S102, a kernel function is selected, and Gaussian process regression training is performed using training samples in combination with the maximum likelihood estimation method to obtain a fracturing difficulty prediction model, which specifically includes:

[0086] (1) Assume an implicit function f(x);

[0087] (2) Select the kernel function and the initial value of the kernel function iteration, and calculate the covariance matrix. Specifically, in this embodiment, the basic data of the fracturing operation in the training sample is regarded as a sample point taken from a multidimensional Gaussian distribution. Let x i ∈R d represents the input matrix, and the independent variables in the input matrix include the maximum displacement and maximum operation pressure of the first n-stage fracturing stage, the pressure and displacement before the slug in the stage, the time to reach the maximum displacement in the pre-fluid stage in the stage, and the maximum pressure and maximum displacement in the pre-fluid stage in the stage; let y i ∈R represents the output, which is the weighted average of the normalized indicators This constructs the Gaussian distribution relationship of the implicit function between y and x as shown below;

[0088] f(x)~GP(m(x),K)

[0089] Where GP(m(x),K) is the Gaussian distribution; m(x) is the mean function; and K is the covariance matrix.

[0090] If there is a new input x * , then the Gaussian process regression and x * The corresponding target data expression is as follows:

[0091]

[0092] Where K is the covariance matrix of the Gaussian process regression model.

[0093] (3) Calculate the predicted value according to the Gaussian distribution, use the maximum likelihood estimation to iteratively optimize the parameters of the kernel function, obtain the kernel function parameters that satisfy the maximum likelihood estimation, and calculate the optimized covariance matrix.

[0094] (4) Based on the optimized covariance matrix, the optimized Gaussian distribution relationship is obtained, thereby training a fracturing difficulty prediction model.

[0095] In the above embodiment, in step S103, the fracturing difficulty prediction model is iteratively corrected using the test sample, that is, the test sample is brought into the trained fracturing difficulty prediction model to correct the threshold value used for fracturing difficulty classification.

[0096] In the above embodiment, in step S104, different kernel functions are selected, and steps S102 to S103 are repeated to obtain fracturing difficulty prediction models corresponding to different kernel functions, and the fracturing difficulty prediction models corresponding to all kernel functions are analyzed for errors using test samples, and the fracturing difficulty prediction model with the smallest error is used as the fracturing difficulty prediction model for the final fracturing difficulty prediction. It should also be noted that the number of kernel functions selected here can be set as needed, and may include exponential kernel functions, square exponential kernel functions, quadratic rational kernel functions, etc. The error analysis here is any one of mean square error, root mean square error, mean absolute error, and mean relative error absolute value.

[0097] The present invention combines the basic data of fracturing construction (maximum displacement and maximum construction pressure within a segment, pressure before the slug within a segment and displacement before the slug, time to reach maximum displacement in the pre-fluid stage within a segment, maximum pressure and maximum displacement in the pre-fluid stage within a segment), and on the basis of fully considering the influencing factors of various aspects of on-site construction, realizes the prediction of the difficulty of fracturing in the unfractured segment, effectively reflects the difficulty of fracturing construction, and can make early warnings for stages with high construction difficulty and high construction risk, assist construction personnel to prepare in advance, reduce possible construction risks, improve fracturing construction efficiency, ensure fracturing construction effects, and enhance reservoir production enhancement transformation effects. Further, the present invention constructs a model using Gaussian process regression, which does not require the definition of a fitting function with a certain form, can effectively capture the nonlinear abstract mapping relationship between the dependent variable and the independent variable, and fully utilizes sample data for fitting prediction, which is very suitable for low-dimensional and small sample regression prediction problems, making model training fast and effective.

[0098] Embodiment 2: As attached Figure 2 As shown, the embodiment of the present invention discloses a method for constructing a fracturing difficulty prediction model, comprising:

[0099] Step S201, constructing training samples and test samples, wherein the training samples include the fracturing difficulty grading results and basic fracturing construction data of the first n-level fracturing stages, based on the basic fracturing construction data of the first n-level fracturing stages, the first n-level fracturing stages are graded for fracturing difficulty using the normalized index weighted average and the threshold value, and the fracturing difficulty grading results of the first n-level fracturing stages are obtained, and the test samples include the fracturing difficulty grading results and basic fracturing construction data of the n+1-level fracturing stage and the n+2-level fracturing stage, and the basic fracturing construction data include the maximum displacement and maximum construction pressure in the stage, the pressure before the slug and the displacement before the slug in the stage, the time to reach the maximum displacement in the pre-fluid stage in the stage, and the maximum pressure and maximum displacement in the pre-fluid stage in the stage;

[0100] Step S202, selecting a kernel function, and using the training samples in combination with the maximum likelihood estimation method to complete Gaussian process regression training to obtain a fracturing difficulty prediction model;

[0101] Step S203, inputting the basic data of the fracturing construction of the n+1th level fracturing stage into the fracturing difficulty prediction model, and outputting the corresponding predicted weighted average value;

[0102] Step S204, using the threshold value to determine the classification prediction result of the fracturing difficulty, and comparing it with the classification result of the n+1th level fracturing stage in the test sample to determine whether the classification prediction result is correct;

[0103] Step S205, in response to incorrectness, adjusting the threshold value, and including the n+1th level fracturing stage in the training sample, re-classifying the fracturing difficulty of the first n+1th level fracturing stage, updating the training sample and the test sample, and then repeating the training with the training sample to obtain a fracturing difficulty prediction model;

[0104] Step S206, in response to a correct result, the n+1th level fracturing stage is included in the training sample, the training sample and the test sample are updated, and the training sample is used to repeat the training to obtain a fracturing difficulty prediction model;

[0105] Step S207, using the retrained fracturing difficulty prediction model to predict the fracturing difficulty of the n+2-th level fracturing stage, and determining the confidence level corresponding to the fracturing difficulty prediction result;

[0106] Step S208, determine whether the confidence is greater than the confidence setting value, and if so, output the corresponding fracturing difficulty prediction model, and if not, repeat the above steps until the confidence is greater than the confidence setting value, and output the corresponding fracturing difficulty prediction model.

[0107] Step S209, select different kernel functions, repeat the above steps, use test samples to compare the prediction errors of the fracturing difficulty prediction models corresponding to different kernel functions, and output the fracturing difficulty prediction model with the smallest error.

[0108] In the above embodiment, steps S203 to S208 use the test samples to iteratively correct the fracturing difficulty prediction model, and output a fracturing difficulty prediction model that meets the accuracy requirements. Specifically:

[0109] In step S203 to step S204, the basic fracturing construction data of the n+1th level fracturing stage is input into the fracturing difficulty prediction model, and the corresponding predicted weighted average value is output. The threshold value is used to determine the fracturing difficulty classification prediction result, and it is compared with the classification result of the n+1th level fracturing stage in the test sample to determine whether the classification prediction result is correct.

[0110] In step S205, if the classification prediction result is incorrect, the threshold value is corrected and adjusted, and then the n+1-th level fracturing stage is included in the training sample. At this time, since the threshold value is corrected, it is necessary to re-classify the fracturing difficulty of the first n+1-th level fracturing stage based on the basic data of fracturing construction of the first n+1-th level fracturing stage, and use the normalized index weighted average value and the corrected threshold value to classify the fracturing difficulty of the first n+1-th level fracturing stage, and then update the training sample and the test sample. The updated training sample includes the fracturing difficulty classification result and the basic data of fracturing construction of the first n+1-th level fracturing stage, and the updated test sample includes the fracturing difficulty classification result and the basic data of fracturing construction of the n+2-th level fracturing stage. After the training sample is updated, step S202 is repeated using the updated training sample to train and obtain the corresponding fracturing difficulty prediction model.

[0111] In step S206, if the classification prediction result is correct, the n+1th level fracturing stage is included in the training sample, and the first n+1th level fracturing stage is re-classified for fracturing difficulty based on the basic fracturing construction data of the first n+1th level fracturing stage using the normalized index weighted average and the threshold value, and then the training sample and the test sample are updated. The updated training sample includes the fracturing difficulty classification result and the basic fracturing construction data of the first n+1th level fracturing stage, and the updated test sample includes the fracturing difficulty classification result and the basic fracturing construction data of the n+2th level fracturing stage. After the training sample is updated, step S202 is repeated using the updated training sample to train and obtain the corresponding fracturing difficulty prediction model.

[0112] In step S207 and step S208, the fracturing difficulty prediction model retrained in step S205 and step S206 is used to predict the fracturing difficulty of the n+2th level fracturing stage, and the confidence corresponding to the fracturing difficulty prediction result is determined, and it is judged whether the confidence is greater than the confidence setting value (the confidence setting value is set according to the actual situation and can be set to 90%). If the response is yes, the corresponding fracturing difficulty prediction model is output. If the response is no, steps S201 to S208 are repeated until the confidence is greater than the confidence setting value.

[0113] In the above embodiment, steps S201 to S208 are a process of constructing a fracturing difficulty prediction model for a kernel function. In step S209, a different kernel function is selected, and steps S201 to S208 are repeated to obtain a corresponding fracturing difficulty prediction model.

[0114] Embodiment 3: As attached Figure 3 As shown, the embodiment of the present invention discloses a method for constructing a fracturing difficulty prediction model, which is a further optimization of the above embodiment, wherein constructing a training sample includes:

[0115] Step S301, obtaining basic fracturing construction data of the first n-stage fracturing stages, wherein the basic fracturing construction data includes the maximum displacement and maximum construction pressure in the stage, the pressure before the slug and the displacement before the slug in the stage, the time when the maximum displacement is reached in the pre-fluid stage in the stage, and the maximum pressure and maximum displacement in the pre-fluid stage in the stage;

[0116] Step S302, based on the basic data of fracturing construction, obtaining the displacement-pressure index of the first n-stage fracturing stages, the slug front pressure displacement ratio, and the ratio of the maximum pressure to the maximum displacement in the pre-fluid stage;

[0117] Step S303, normalizing the displacement-pressure index of the first n-stage fracturing stages, the slug front pressure displacement ratio, the ratio of the maximum pressure to the maximum displacement in the pre-fluid stage, and the time to reach the maximum displacement in the pre-fluid stage;

[0118]

[0119] Among them, x is the data that needs to be normalized; x min is the smallest data in the data set; x max It is the largest data in the dataset;

[0120] Step S304, obtaining the weighted average of the four normalized indicators of the first n-stage fracturing stages

[0121]

[0122] in, is the normalized displacement-pressure index; is the normalized slug front pressure ratio; It is the normalized ratio of the maximum pressure to the maximum displacement in the pre-fluid stage; is the time for the normalized pre-fluid stage to reach the maximum displacement;

[0123] Step S305, setting a threshold value, Compared with the threshold value, the fracturing difficulty classification results of the first n fracturing stages;

[0124] Step S306, constructing a training sample by combining the fracturing difficulty classification results of the first n fracturing stages and the basic data of fracturing construction.

[0125] In the above embodiment, in step S302, based on the basic data of fracturing construction, the displacement-pressure index of the first n-stage fracturing stages, the slug front pressure displacement ratio, and the ratio of the maximum pressure to the maximum displacement in the pre-fluid stage are obtained, including:

[0126] (1) The displacement-pressure index of the first n-stage fracturing stage is calculated by the following formula:

[0127]

[0128] Among them, F is the displacement-pressure index; Q i,max (i=1,2,...,n) is the maximum displacement in the first n-stage fracturing stage; P i,max (i=1,2,...,n) is the maximum operation pressure in the first n stages of fracturing;

[0129] (2) The slug front pressure ratio of the first n-stage fracturing stages is calculated by the following formula:

[0130]

[0131] Among them, R i,slug is the slug front pressure ratio; P i,slug (i=1,2,...,n) is the pre-slug pressure in the first n-stage fracturing stage; Q i,slug (i=1,2,...,n) is the slug front displacement in the first n-stage fracturing stage;

[0132] (3) The ratio of the maximum pressure and the maximum displacement of the pre-fluid stage of the first n-stage fracturing stage is calculated by the following formula:

[0133]

[0134] Among them, R i,pad It is the ratio of the maximum pressure to the maximum displacement in the pre-liquid stage; P i,pad (i=1,2,...,n) is the maximum pressure of the pre-fluid stage in the first n-stage fracturing stage; Q i,pad (i=1, 2, ..., n) is the maximum displacement of the pre-fluid stage in the first n fracturing stages.

[0135] In the above embodiment, the threshold value is and The fractured sections can be divided into three categories: Easy to fracture; higher than The hard fracturing is difficult; the medium fracturing is between the two. When ω<1.24, it is easy to fracture; when ω>1.62, it is difficult to fracture; and the value between the two is medium.

[0136] Embodiment 4: As attached Figure 4 As shown, the embodiment of the present invention discloses a method for predicting the difficulty of fracturing, comprising:

[0137] Step S401, obtaining basic fracturing construction data of the predicted fracturing section, wherein the basic fracturing construction data includes the maximum displacement and maximum construction pressure in the section, the pressure before the slug and the displacement before the slug in the section, the time when the maximum displacement is reached in the pre-fluid stage in the section, and the maximum pressure and maximum displacement in the pre-fluid stage in the section;

[0138] Step S402: Input the basic data of the fracturing construction of the fracturing section to be predicted into the fracturing difficulty prediction model, and output the corresponding weighted average value. The fracturing difficulty prediction model is constructed according to the method described in the above embodiment;

[0139] Step S403: weighted average The result is compared with the threshold value corresponding to the fracturing difficulty prediction model to determine the fracturing difficulty classification result of the fracturing section to be predicted.

[0140] Example 5, introduces specific data and applies the method of the above example to construct a fracturing difficulty prediction model, as follows:

[0141] (1) Data collection of DaHW1506 well

[0142] According to the on-site fracturing construction records, the maximum displacement and maximum construction pressure of the first n fracturing stages, the pressure before the slug and the displacement before the slug, the time to reach the maximum displacement during the pre-fluid stage within the stage, and the maximum pressure and maximum displacement during the pre-fluid stage within the stage were collected, as shown in Table 1.

[0143] (2) Calculation of the fracturing difficulty index of DaHW1506 well

[0144] The displacement-pressure index of the first n fractured stages, the slug front pressure displacement ratio, the ratio of the maximum pressure to the maximum displacement in the pre-fluid stage, and the time to reach the maximum displacement in the pre-fluid stage are determined, as shown in Table 2.

[0145] (3) Data collection of DaHW1507 well

[0146] According to the on-site fracturing construction records, the maximum displacement and maximum construction pressure of the first n fracturing stages, the pressure before the slug and the displacement before the slug, the time to reach the maximum displacement during the pre-fluid stage within the stage, and the maximum pressure and maximum displacement during the pre-fluid stage within the stage were collected, as shown in Table 3.

[0147] (4) Calculation of the fracturing difficulty index of DaHW1507 well

[0148] The displacement-pressure index of the first n fractured stages, the slug front pressure displacement ratio, the ratio of the maximum pressure to the maximum displacement in the pre-fluid stage, and the time to reach the maximum displacement in the pre-fluid stage are determined, as shown in Table 4.

[0149] (5) Data collection of DaHW1508 well

[0150] According to the on-site fracturing construction records, the maximum displacement and maximum construction pressure of the first n fracturing stages, the pressure before the slug and the displacement before the slug, the time to reach the maximum displacement during the pre-fluid stage within the stage, and the maximum pressure and maximum displacement during the pre-fluid stage within the stage were collected, as shown in Table 5.

[0151] (6) Calculation of the fracturing difficulty index of DaHW1508 well

[0152] Determine the displacement-pressure index of the first n fractured stages, the slug front pressure displacement ratio, the ratio of the maximum pressure to the maximum displacement in the pre-fluid stage, and the time to reach the maximum displacement in the pre-fluid stage, as shown in Table 6.

[0153] (7) Data collection of DaHW1509 well

[0154] According to the on-site fracturing construction records, the maximum displacement and maximum construction pressure of the first n fracturing stages, the pressure before the slug and the displacement before the slug, the time to reach the maximum displacement during the pre-fluid stage within the stage, and the maximum pressure and maximum displacement during the pre-fluid stage within the stage were collected, as shown in Table 7.

[0155] (8) Calculation of the fracturing difficulty index of DaHW1509 well

[0156] Determine the displacement-pressure index of the first n fractured stages, the slug front pressure displacement ratio, the ratio of the maximum pressure to the maximum displacement in the pre-fluid stage, and the time to reach the maximum displacement in the pre-fluid stage, as shown in Table 8.

[0157] (9) Based on the classification results of the first n-stage fracturing stages of the DaHW1506 well, a Gaussian process regression model for predicting the difficulty of fracturing is trained, and the parameters of the Gaussian process regression model are optimized using the maximum likelihood estimation method, and the fracturing difficulty of the n+1th fracturing stage is predicted in real time on site. The prediction results of the Gaussian process regression model for the n+1th fracturing stage are accepted or modified, and the Gaussian process regression model for the n+2th fracturing stage is trained again based on the classification results of the first n+1 fracturing stages, until a Gaussian process regression model for predicting the difficulty of fracturing in the unfractured stage with a confidence level greater than 90% is obtained. The prediction results are shown in the attached figure. Figure 5 The above process was used to train and iteratively optimize the Gaussian process regression model for wells DaHW1507 to DaHW1509, and the prediction results are shown in the attached figure. Figures 6 to 8 shown.

[0158] Embodiment 6: As attached Fig. 9 As shown, an embodiment of the present invention discloses a device for constructing a fracturing difficulty prediction model, comprising:

[0159] A sample construction unit, constructing training samples and test samples, wherein the training samples include the fracturing difficulty grading results and basic fracturing construction data of the first n-stage fracturing sections, based on the basic fracturing construction data of the first n-stage fracturing sections, the first n-stage fracturing sections are graded for fracturing difficulty using the normalized index weighted average and threshold value, and the fracturing difficulty grading results of the first n-stage fracturing sections are obtained, and the test samples include the fracturing difficulty grading results and basic fracturing construction data of the n+1-stage fracturing sections and the n+2-stage fracturing sections, and the basic fracturing construction data include the maximum displacement and maximum construction pressure in the section, the pressure before the slug and the displacement before the slug in the section, the time to reach the maximum displacement in the pre-fluid stage in the section, and the maximum pressure and maximum displacement in the pre-fluid stage in the section;

[0160] The model training unit selects a kernel function and uses the training samples in combination with the maximum likelihood estimation method to complete Gaussian process regression training and obtain a fracturing difficulty prediction model;

[0161] A model testing unit, which uses the test samples to iteratively correct the fracturing difficulty prediction model and outputs a fracturing difficulty prediction model that meets the accuracy requirements;

[0162] The model output unit selects different kernel functions, repeats the above steps, uses the test samples to compare the prediction errors of the fracturing difficulty prediction models corresponding to different kernel functions, and outputs the fracturing difficulty prediction model with the smallest error.

[0163] The sample construction unit is constructed, including a training sample construction module and a test sample construction module, wherein as shown in the attached Fig.10 As shown, the training sample building module includes:

[0164] The sample data acquisition module acquires the basic data of the fracturing construction of the first n-stage fracturing stages, wherein the basic data of the fracturing construction includes the maximum displacement and the maximum construction pressure in the stage, the pressure before the slug and the displacement before the slug in the stage, the time when the maximum displacement is reached in the pre-fluid stage in the stage, and the maximum pressure and the maximum displacement in the pre-fluid stage in the stage;

[0165] The index acquisition module obtains the displacement-pressure index of the first n-stage fracturing stages, the slug front pressure-displacement ratio, and the ratio of the maximum pressure to the maximum displacement in the pre-fluid stage based on the basic data of the fracturing construction;

[0166] The first processing module normalizes the four indicators of the displacement-pressure index of the first n-stage fracturing stages, the slug front pressure displacement ratio, the ratio of the maximum pressure to the maximum displacement in the pre-fluid stage, and the time to reach the maximum displacement in the pre-fluid stage;

[0167]

[0168] Among them, x is the data that needs to be normalized; x minis the smallest data in the data set; x max It is the largest data in the dataset;

[0169] The second processing module obtains the weighted average of the four normalized indicators of the first n-stage fracturing stages

[0170]

[0171] in, is the normalized displacement-pressure index; is the normalized slug front pressure ratio; It is the normalized ratio of the maximum pressure to the maximum displacement in the pre-fluid stage; is the time for the normalized pre-fluid stage to reach the maximum displacement;

[0172] Comparison module, set the threshold value, the first n fracturing stages Compared with the threshold value, the fracturing difficulty classification results of the first n fracturing stages;

[0173] The sample formation module combines the fracturing difficulty classification results of the first n fracturing stages and the basic data of fracturing construction to construct training samples.

[0174] Embodiment 7: As attached Fig.11 As shown, the embodiment of the present invention discloses a device for predicting the difficulty of fracturing, comprising:

[0175] A data acquisition unit is used to acquire basic fracturing construction data of the predicted fracturing section, wherein the basic fracturing construction data includes the maximum displacement and maximum construction pressure in the section, the pressure before the slug and the displacement before the slug in the section, the time when the maximum displacement is reached in the pre-fluid stage in the section, and the maximum pressure and maximum displacement in the pre-fluid stage in the section;

[0176] The prediction unit inputs the basic data of the fracturing construction of the fracturing section to be predicted into the fracturing difficulty prediction model and outputs the corresponding weighted average value. The fracturing difficulty prediction model is constructed according to the method described in the above embodiment;

[0177] Grading unit, the weighted average The result is compared with the threshold value corresponding to the fracturing difficulty prediction model to determine the fracturing difficulty classification result of the fracturing section to be predicted.

[0178] Embodiment 8: The embodiment of the present invention discloses a storage medium, on which is stored a computer program that can be read by a computer, and the computer program is configured to execute a method for predicting the difficulty of fracturing when it is run.

[0179] The above storage medium may include, but is not limited to: a USB flash drive, a read-only memory, a mobile hard disk, a magnetic disk or an optical disk, and other media that can store computer programs.

[0180] Embodiment 9: The embodiment of the present invention discloses an electronic device, including a processor and a memory, wherein the memory stores a computer program, and the computer program is loaded and executed by the processor to implement a method for predicting the difficulty of fracturing.

[0181] The processor may be a central processing unit (CPU), a general purpose processor, a digital signal processor (DSP), an ASIC, an FPGA or other programmable logic device, a transistor logic device, a hardware component or any combination thereof. It may implement or execute various exemplary logic blocks, modules and circuits described in conjunction with the disclosure of this application. It may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, and the like. The memory may include, but is not limited to, various media that can store computer programs, such as a USB flash drive, a read-only memory, a mobile hard disk, a magnetic disk or an optical disk.

[0182] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of complete hardware embodiments, complete software embodiments, or embodiments in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code. The scheme in the embodiments of the present application can be implemented in various computer languages, for example, object-oriented programming language Java and literal scripting language JavaScript, etc.

[0183] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0184] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0185] The above technical features constitute the best embodiment of the present invention, which has strong adaptability and best implementation effect. Non-essential technical features can be added or reduced according to actual needs to meet the requirements of different situations.

[0186] Table 1 Construction data of DaHW1506 well

[0187]

[0188] Table 2 Summary of key indicators of DaHW1506 well

[0189]

[0190] Table 3 Construction data of DaHW1507 well

[0191]

[0192] Table 4 Summary of key indicators of DaHW1507 well

[0193]

[0194]

[0195] Table 5 Construction data of DaWH1508 well

[0196]

[0197]

[0198] Table 6 Summary of key indicators of DaHW well

[0199]

[0200]

[0201] Table 7 Construction data of DaHW1509 well

[0202]

[0203] Table 8 Summary of key indicators of DaHW1509 well

[0204]

[0205]

Claims

1. A method for constructing a fracturing difficulty prediction model, characterized in that: include: Construct training samples and test samples, wherein the training samples include the fracturing difficulty grading results and basic fracturing construction data of the first n-stage fracturing sections, based on the basic fracturing construction data of the first n-stage fracturing sections, the first n-stage fracturing sections are graded for fracturing difficulty using the normalized index weighted average and threshold value, and the fracturing difficulty grading results of the first n-stage fracturing sections are obtained, and the test samples include the fracturing difficulty grading results and basic fracturing construction data of the n+1-stage fracturing sections and the n+2-stage fracturing sections, and the basic fracturing construction data include the maximum displacement and maximum construction pressure in the section, the pressure before the slug and the displacement before the slug in the section, the time to reach the maximum displacement in the pre-fluid stage in the section, and the maximum pressure and maximum displacement in the pre-fluid stage in the section; The kernel function is selected, and the Gaussian process regression training is completed using the training samples in combination with the maximum likelihood estimation method to obtain a prediction model for the difficulty of fracturing. Iteratively modify the fracturing difficulty prediction model using test samples, and output a fracturing difficulty prediction model that meets accuracy requirements; Select different kernel functions, repeat the above steps, use the test samples to compare the prediction errors of the fracturing difficulty prediction models corresponding to different kernel functions, and output the fracturing difficulty prediction model with the smallest error.

2. The method for constructing a fracturing difficulty prediction model according to claim 1, characterized in that: The method of iteratively correcting the fracturing difficulty prediction model using the test samples and outputting the fracturing difficulty prediction model that meets the accuracy requirements includes: Input the basic data of fracturing construction of the n+1th fracturing stage into the fracturing difficulty prediction model, and output the corresponding predicted weighted average value; The threshold value is used to determine the classification prediction result of the fracturing difficulty, and compared with the classification result of the n+1th level fracturing stage in the test sample to determine whether the classification prediction result is correct; In response to incorrectness, the threshold value is adjusted, and the n+1th level fracturing stage is included in the training sample, the first n+1 level fracturing stages are reclassified for fracturing difficulty, the training sample and the test sample are updated, and the training sample is used to repeat the training to obtain a fracturing difficulty prediction model; If the response is correct, the n+1th level fracturing stage is included in the training sample, the training sample and the test sample are updated, and the training sample is used to repeat the training to obtain a fracturing difficulty prediction model; The retrained fracturing difficulty prediction model is used to predict the fracturing difficulty of the n+2-th fracturing stage, and the confidence level corresponding to the fracturing difficulty prediction result is determined; Determine whether the confidence is greater than the confidence setting value, and if so, output the corresponding fracturing difficulty prediction model, and if not, repeat the above steps until the confidence is greater than the confidence setting value, and output the corresponding fracturing difficulty prediction model.

3. The method for constructing a fracturing difficulty prediction model according to claim 1 or 2, characterized in that: The constructing of training samples comprises: Obtaining basic fracturing construction data of the first n-stage fracturing stages, wherein the basic fracturing construction data includes the maximum displacement and maximum construction pressure in the stage, the pressure before the slug and the displacement before the slug in the stage, the time when the maximum displacement is reached in the pre-fluid stage in the stage, and the maximum pressure and maximum displacement in the pre-fluid stage in the stage; Based on the basic data of fracturing construction, the displacement-pressure index of the first n-stage fracturing stages, the slug front pressure-displacement ratio, and the ratio of the maximum pressure to the maximum displacement in the pre-fluid stage are obtained; The four indicators of displacement-pressure index of the first n-stage fracturing stages, the slug front pressure displacement ratio, the ratio of the maximum pressure to the maximum displacement in the pre-fluid stage, and the time to reach the maximum displacement in the pre-fluid stage are normalized; Among them, x is the data that needs to be normalized; x min is the smallest data in the data set; x max It is the largest data in the dataset; Get the weighted average of the four normalized indicators of the first n fracturing stages in, is the normalized displacement-pressure index; is the normalized slug front pressure ratio; It is the normalized ratio of the maximum pressure to the maximum displacement in the pre-fluid stage; is the time for the normalized pre-fluid stage to reach the maximum displacement; Set the threshold value to Compared with the threshold value, the fracturing difficulty classification results of the first n fracturing stages; The training samples are constructed by combining the fracturing difficulty classification results of the first n fracturing stages and the basic data of fracturing construction.

4. The method for constructing a fracturing difficulty prediction model according to claim 3, characterized in that: The method of obtaining the displacement-pressure index of the first n-stage fracturing stages, the slug front pressure-displacement ratio, and the ratio of the maximum pressure to the maximum displacement in the pre-fluid stage based on the basic fracturing construction data includes: The displacement-pressure index of the first n-stage fracturing stage is calculated by the following formula: Where F is the displacement-pressure index; Q i,max (i=1,2,...,n) is the maximum displacement in the first n-stage fracturing stage; P i,max (i=1,2,...,n) is the maximum operation pressure in the first n-stage fracturing stage; The slug front pressure ratio of the first n-stage fracturing stage is calculated by the following formula: Among them, R i,slug is the slug front pressure ratio; P i,slug (i=1,2,...,n) is the pre-slug pressure in the first n-stage fracturing stage; Q i,slug (i=1,2,...,n) is the slug front displacement in the first n-stage fracturing stage; The ratio of the maximum pressure and the maximum displacement of the pre-fluid stage in the first n-stage fracturing stage is calculated by the following formula: Among them, R i,pad It is the ratio of the maximum pressure to the maximum displacement in the pre-liquid stage; P i,pad (i=1,2,...,n) is the maximum pressure of the pre-fluid stage in the first n-stage fracturing stage; Q i,pad (i=1, 2, ..., n) is the maximum displacement of the pre-fluid stage in the first n fracturing stages.

5. A method for predicting the difficulty of fracturing, characterized in that: include: Obtaining basic fracturing construction data of the predicted fracturing section, wherein the basic fracturing construction data includes the maximum displacement and maximum construction pressure in the section, the pressure before the slug and the displacement before the slug in the section, the time when the maximum displacement is reached in the pre-fluid stage in the section, and the maximum pressure and maximum displacement in the pre-fluid stage in the section; Input the basic data of fracturing construction of the predicted fracturing section into the fracturing difficulty prediction model and output the corresponding weighted average value. The fracturing difficulty prediction model is constructed according to the method according to any one of claims 1 to 4; The weighted average The result is compared with the threshold value corresponding to the fracturing difficulty prediction model to determine the fracturing difficulty classification result of the fracturing section to be predicted.

6. A device for constructing a fracturing difficulty prediction model using the method according to any one of claims 1 to 4, characterized in that: include: A sample construction unit, constructing training samples and test samples, wherein the training samples include the fracturing difficulty grading results and basic fracturing construction data of the first n-stage fracturing sections, based on the basic fracturing construction data of the first n-stage fracturing sections, the first n-stage fracturing sections are graded for fracturing difficulty using the normalized index weighted average and threshold value, and the fracturing difficulty grading results of the first n-stage fracturing sections are obtained, and the test samples include the fracturing difficulty grading results and basic fracturing construction data of the n+1-stage fracturing sections and the n+2-stage fracturing sections, and the basic fracturing construction data include the maximum displacement and maximum construction pressure in the section, the pressure before the slug and the displacement before the slug in the section, the time to reach the maximum displacement in the pre-fluid stage in the section, and the maximum pressure and maximum displacement in the pre-fluid stage in the section; The model training unit selects a kernel function and uses the training samples in combination with the maximum likelihood estimation method to complete Gaussian process regression training and obtain a fracturing difficulty prediction model; A model testing unit, which uses the test samples to iteratively correct the fracturing difficulty prediction model and outputs a fracturing difficulty prediction model that meets the accuracy requirements; The model output unit selects different kernel functions, repeats the above steps, uses the test samples to compare the prediction errors of the fracturing difficulty prediction models corresponding to different kernel functions, and outputs the fracturing difficulty prediction model with the smallest error.

7. The device for constructing a fracturing difficulty prediction model according to claim 6, characterized in that: The sample construction unit includes a training sample construction module and a test sample construction module, wherein the training sample construction module includes: The sample data acquisition module acquires the basic data of the fracturing construction of the first n-stage fracturing stages, wherein the basic data of the fracturing construction includes the maximum displacement and the maximum construction pressure in the stage, the pressure before the slug and the displacement before the slug in the stage, the time when the maximum displacement is reached in the pre-fluid stage in the stage, and the maximum pressure and the maximum displacement in the pre-fluid stage in the stage; The index acquisition module obtains the displacement-pressure index of the first n-stage fracturing stages, the slug front pressure-displacement ratio, and the ratio of the maximum pressure to the maximum displacement in the pre-fluid stage based on the basic data of the fracturing construction; The first processing module normalizes the four indicators of the displacement-pressure index of the first n-stage fracturing stages, the slug front pressure displacement ratio, the ratio of the maximum pressure to the maximum displacement in the pre-fluid stage, and the time to reach the maximum displacement in the pre-fluid stage; Among them, x is the data that needs to be normalized; x min is the smallest data in the data set; x max It is the largest data in the dataset; The second processing module obtains the weighted average of the four normalized indicators of the first n-stage fracturing stages in, is the normalized displacement-pressure index; is the normalized slug front pressure ratio; It is the normalized ratio of the maximum pressure to the maximum displacement in the pre-fluid stage; is the time for the normalized pre-fluid stage to reach the maximum displacement; Comparison module, set the threshold value, the first n fracturing stages Compared with the threshold value, the fracturing difficulty classification results of the first n fracturing stages; The sample formation module combines the fracturing difficulty classification results of the first n fracturing stages and the basic data of fracturing construction to construct training samples.

8. A device for predicting the difficulty of fracturing using the method of claim 5, characterized in that: include: A data acquisition unit is used to acquire basic fracturing construction data of the predicted fracturing section, wherein the basic fracturing construction data includes the maximum displacement and maximum construction pressure in the section, the pressure before the slug and the displacement before the slug in the section, the time when the maximum displacement is reached in the pre-fluid stage in the section, and the maximum pressure and maximum displacement in the pre-fluid stage in the section; The prediction unit inputs the basic data of the fracturing construction of the fracturing section to be predicted into the fracturing difficulty prediction model and outputs the corresponding weighted average value. The fracturing difficulty prediction model is constructed according to the method according to any one of claims 1 to 5; Grading unit, the weighted average The result is compared with the threshold value corresponding to the fracturing difficulty prediction model to determine the fracturing difficulty classification result of the fracturing section to be predicted.

9. A storage medium, characterized in that: The storage medium stores a computer program that can be read by a computer, and the computer program is configured to execute the method for predicting the difficulty of fracturing according to any one of claim 6 when running.

10. An electronic device, characterized in that: It comprises a processor and a memory, wherein a computer program is stored in the memory, and the computer program is loaded and executed by the processor to implement the method for predicting the difficulty of fracturing as claimed in any one of claim 6.

Images