Intelligent prediction method, device, equipment, medium and program product for cementing quality

Abstract

The present disclosure relates to the technical field of oil drilling, and particularly relates to a cementing quality intelligent prediction method, device, equipment, medium and program product. The method comprises the following steps: obtaining correlation coefficients of each logging data based on the logging data of a target oil and gas well through a Spearman correlation coefficient; selecting logging data with an absolute value of the correlation coefficient greater than a preset threshold as input parameters for predicting the cementing quality; dividing the input parameters into continuous parameters and non-continuous parameters based on the data continuity of the input parameters; inputting the continuous parameters and the non-continuous parameters into a mixed neural network model respectively to obtain the cementing quality of the target oil and gas well. By adopting the mixed neural network model, the continuous parameters and the non-continuous parameters are input into different neural network models respectively, the cementing quality prediction with a high accuracy can be realized, the operation efficiency and the prediction accuracy are high, and the cementing quality can be quantitatively evaluated.

Description

Technical Field

[0001] This disclosure relates to the field of oil drilling technology, and in particular to a method, apparatus, equipment, medium, and program product for intelligent prediction of cementing quality. Background Technology

[0002] Cementing is a crucial part of drilling engineering. The quality of cementing has a significant impact on the lifespan, productivity, and overall benefits of exploration and development of oil and gas wells. Moreover, with the continuous deepening of oil exploration and development, higher requirements are being placed on the scientific nature and precision of cementing design, thus leading to increasingly higher demands on cementing quality.

[0003] Currently, commonly used cementing quality prediction methods are mainly divided into traditional cementing quality calculation methods and big data-driven methods. However, the numerical simulation results of traditional cementing quality calculation methods cannot reflect the real situation and are mostly qualitative analyses, unable to achieve quantitative calculations. While big data-driven methods have produced good results in cementing quality prediction, they often have poor transfer effects and are difficult to achieve accurate predictions in different wells and blocks. Summary of the Invention

[0004] This disclosure provides a method, apparatus, equipment, medium, and program product for intelligent prediction of cementing quality, so as to achieve quantitative calculation and improve migration effect.

[0005] Firstly, this disclosure provides an intelligent prediction method for cementing quality, including:

[0006] Based on the logging data of the target oil and gas well, the correlation between each logging data and cementing quality is calculated using the Spearman correlation coefficient to obtain the correlation coefficient of each logging data. The logging data is the data obtained by measuring and recording the target oil and gas well during the drilling process.

[0007] Logging data with an absolute value of correlation coefficient greater than a preset threshold are selected as input parameters for predicting cementing quality.

[0008] Based on the data continuity of the input parameters, the input parameters are divided into continuous parameters and discontinuous parameters. The continuous parameters are time-series data that depend on time series, and the discontinuous parameters are static data that do not depend on time series.

[0009] The continuous parameters and the discontinuous parameters are respectively input into a pre-trained hybrid neural network model for prediction to obtain the cementing quality of the target oil and gas well.

[0010] In some embodiments, based on logging data from the target oil and gas well, the correlation between each logging data point and cementing quality is calculated using the Spearman correlation coefficient, including:

[0011] Data preprocessing is performed on logging data from the target oil and gas wells;

[0012] The data preprocessing includes: deleting missing values ​​in the logging data; performing data interpolation on the well inclination and azimuth data in the logging data; performing one-hot encoding on discontinuous data in the logging data; and performing normalization on the logging data.

[0013] Based on the preprocessed logging data, the correlation between each logging data point and cementing quality is calculated using the Spearman correlation coefficient.

[0014] In some embodiments, the logging data is normalized, including:

[0015] Based on the characteristic data in the logging data, the average value and standard deviation of the characteristic data are obtained;

[0016] Calculate the difference between the feature data and the average value of the feature data;

[0017] Based on the ratio of the difference to the standard deviation, the normalized feature data is calculated.

[0018] In some embodiments, after obtaining the cementing quality of the target oil and gas well, the method further includes:

[0019] Calculate the SHAP value of each input parameter, wherein the SHAP value is used to characterize the degree of influence of each input parameter on the cementing quality of the target oil and gas well;

[0020] Based on the largest SHAP value, the input parameters with the greatest impact are obtained for quantitative and qualitative analysis of cementing quality.

[0021] In some embodiments, the construction and training process of the hybrid neural network model includes:

[0022] An initial hybrid neural network model is constructed based on a feedforward neural network model and a gated recurrent unit model. The initial hybrid neural network model includes a gated recurrent unit input layer, a feedforward neural network input layer, a feature splicing layer, a dropout layer, and three feedforward neural network layers.

[0023] Data preprocessing and correlation analysis were performed on the logging data of the sample oil and gas wells to obtain sample input parameters for model input;

[0024] The initial hybrid neural network model is trained based on the sample input parameters until it converges, thereby obtaining the hybrid neural network model and storing its model weights.

[0025] In some embodiments, the continuous parameters and the discontinuous parameters are respectively input into a pre-trained hybrid neural network model for prediction to obtain the cementing quality of the target oil and gas well, including:

[0026] Read the stored model weights to obtain the pre-trained hybrid neural network model;

[0027] The continuous parameters are input into the gated recurrent unit input layer of the hybrid neural network model, and the gated recurrent unit model performs data processing; and the discontinuous parameters are input into the feedforward neural network input layer of the hybrid neural network model, and the feedforward neural network model performs data processing.

[0028] The cementing quality of the target oil and gas well is obtained through the prediction of the hybrid neural network model.

[0029] Secondly, this disclosure provides a smart cementing quality prediction device, comprising:

[0030] The data processing module is used to calculate the correlation between each logging data and cementing quality based on the logging data of the target oil and gas well using the Spearman correlation coefficient, and obtain the correlation coefficient of each logging data. The logging data is the data obtained by measuring and recording the target oil and gas well during the drilling process.

[0031] The input parameter selection module is used to select logging data with an absolute value of correlation coefficient greater than a preset threshold as input parameters for predicting cementing quality.

[0032] The parameter classification module is used to classify the input parameters into continuous parameters and discontinuous parameters based on the data continuity of the input parameters. The continuous parameters are time series data that depend on time series, and the discontinuous parameters are static data that do not depend on time series.

[0033] The intelligent prediction module is used to input the continuous parameters and the discontinuous parameters into a pre-trained hybrid neural network model to obtain the predicted cementing quality of the target oil and gas well.

[0034] Thirdly, this disclosure provides a computer device including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method described in the foregoing aspects.

[0035] Fourthly, this disclosure provides a computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the steps of the method described in the above aspects.

[0036] Fifthly, this disclosure provides a computer program product, including a computer program / instructions that, when executed by a processor, implement the steps of the methods described in the foregoing aspects.

[0037] This disclosure provides a method, device, equipment, medium, and program product for intelligent prediction of cementing quality. By dividing the input parameters into continuous and discontinuous parameters and employing a hybrid neural network model, the continuous and discontinuous parameters are input into different neural network models respectively. This can achieve high-accuracy prediction of cementing quality. Compared with traditional cementing quality calculation methods, it has higher operating efficiency and prediction accuracy, and can quantitatively evaluate cementing quality.

[0038] The hybrid neural network model constructed by combining a feedforward neural network model and a gated recurrent unit model exhibits good transferability and can achieve relatively accurate predictions across different wells and blocks.

[0039] By calculating the SHAP value and analyzing the input parameters with the highest impact, qualitative and quantitative analysis of the factors affecting cementing quality (i.e., input parameters) can be achieved. This allows for advance prediction of cementing quality before cementing operations and analysis of the input parameters that currently have a significant impact on cementing quality, in order to guide cementing operations.

[0040] It should be understood that both the foregoing general description and the following detailed description are exemplary and intended to provide further illustration of the claimed technology. Attached Figure Description

[0041] The above and other objects, features, and advantages of this disclosure will become more apparent from the more detailed description of the embodiments thereof in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of this disclosure and form part of the specification. They are used together with the embodiments of this disclosure to explain the disclosure and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same components or steps.

[0042] The present disclosure will be described in more detail below based on embodiments and with reference to the accompanying drawings:

[0043] Figure 1 A flowchart of the intelligent prediction method for cementing quality provided in the embodiments of this disclosure.

[0044] Figure 2 This is a structural diagram illustrating a hybrid neural network model applied according to an embodiment of the present disclosure.

[0045] Figure 3 This is a flowchart illustrating yet another application of the intelligent prediction method for cementing quality according to an embodiment of this disclosure.

[0046] Figure 4 This is a schematic diagram of the structure of a cementing quality intelligent device according to an embodiment of the present disclosure.

[0047] Figure 5 This is a schematic diagram illustrating the structure of a computer device according to an embodiment of the present disclosure.

[0048] Figure 6 This is a schematic diagram illustrating a computer program product according to an embodiment of the present disclosure.

[0049] In the accompanying drawings, the same parts are referred to by the same reference numerals, and the drawings are not drawn to scale. Detailed Implementation

[0050] To enable those skilled in the art to better understand the technical solutions of this disclosure, and to fully understand and implement the process of how this disclosure applies technical means to solve technical problems and achieve corresponding technical effects, the technical solutions in the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, not all embodiments. The embodiments of this disclosure and the various features within them can be combined with each other without conflict, and the resulting technical solutions are all within the protection scope of this disclosure. All other embodiments obtained by those skilled in the art based on the embodiments of this disclosure without creative effort should fall within the protection scope of this disclosure.

[0051] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this disclosure are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this disclosure described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0052] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0053] Example 1

[0054] Figure 1 A flowchart illustrating an intelligent prediction method for cementing quality provided in this embodiment of the disclosure. Figure 1 As shown, a smart prediction method for cementing quality includes S101-S104:

[0055] S101: Based on the logging data of the target oil and gas well, the correlation between each logging data and cementing quality is calculated using the Spearman correlation coefficient to obtain the correlation coefficient of each logging data.

[0056] Among them, logging data is the data obtained by measuring and recording the target oil and gas well during the drilling process, such as well depth, azimuth, well diameter, well inclination angle, sonic transit time, natural gamma, resistivity, spontaneous potential, centralizer position, etc.

[0057] Specifically, S101 includes:

[0058] (i) Perform data preprocessing on logging data of the target oil and gas wells;

[0059] Data preprocessing specifically includes:

[0060] (1) Delete missing values ​​in well logging data.

[0061] The collected logging data usually contains missing values ​​such as incompleteness, inconsistency, inaccuracy, and missing data. In this embodiment, the method for handling missing values ​​is deletion. That is, when a missing value is found, the row containing the current data is directly deleted to prevent it from having an adverse effect on the hybrid neural network model.

[0062] (2) Perform data interpolation on the well inclination and azimuth data in the logging data.

[0063] Most logging data uses a step size of 1m, but the step size for well inclination azimuth data is often 30m. Therefore, to ensure data alignment, well inclination azimuth data needs to be interpolated to obtain a more complete or continuous dataset. Optionally, this embodiment uses cubic spline interpolation, which approximates the data using piecewise continuous low-order polynomials to obtain a smooth curve. The spline interpolation curve is smooth and continuous, and can better approximate the actual curve.

[0064] (3) Perform unique thermal encoding on discontinuous data in well logging data.

[0065] For some discontinuous data (i.e. discontinuous parameters) in logging data, such as drilling fluid type and whether there is a casing centralizer, one-hot encoding is performed to meet the input requirements of the hybrid neural network model.

[0066] (4) Standardize and normalize the logging data.

[0067] Normalization is primarily used to handle differences in the dimensionality of data between quality features. For example, a sample may have one feature dimension ranging from 0 to 1, while another feature dimension ranges from 100 to 10000. It's difficult to compare their relative changes. To address this issue, this embodiment performs data normalization, specifically including:

[0068] Based on the characteristic data in the logging data, the average value and standard deviation of the characteristic data are obtained;

[0069] Calculate the difference between the feature data and the average value of the feature data;

[0070] Based on the ratio of the difference to the standard deviation, the normalized feature data is calculated.

[0071] That is, the normalization formula is:

[0072]

[0073] Where x is the normalized and standardized feature data, μ is the mean of the feature data, σ is the standard deviation of the feature data, and X is the feature data.

[0074] (ii) Conduct correlation analysis on the logging data after data preprocessing.

[0075] Correlation analysis refers to the analysis of two or more correlated variables (i.e., influencing parameters) to measure the degree of correlation between the two variables. Correlation analysis is necessary before modeling to identify the parameters that significantly impact the cementing quality of oil and gas wells, preventing excessive irrelevant variables from being input into the hybrid neural network model and affecting its running speed and accuracy.

[0076] Specifically, correlation analysis includes:

[0077] Based on the preprocessed logging data, a dataset of the target oil and gas wells is obtained;

[0078] The correlation analysis of the logging data included in the dataset was performed using Spearman correlation coefficient. The correlation between each logging data and cementing quality was calculated, and the correlation coefficient of each logging data was obtained.

[0079] S102: Select logging data with an absolute value of correlation coefficient greater than a preset threshold as input parameters for predicting cementing quality.

[0080] This embodiment uses the Spearman correlation coefficient to calculate the correlation coefficients of various influencing parameters. When the correlation coefficient is less than 0, it is called negative correlation; when it is greater than 0, it is called positive correlation; and when it is equal to 0, it is called zero correlation. Optionally, this embodiment uses a preset threshold of 0.3 as an example to filter logging data with an absolute value of correlation coefficient greater than 0.3 as input parameters for the hybrid neural network model. For example, after correlation analysis, this embodiment selects well depth, azimuth, well diameter, well inclination angle, sonic transit time, natural gamma, resistivity, spontaneous potential, casing centering, average return velocity of grouting, average return velocity of displacement slurry, cement slurry contact time, and centralizer position as input parameters for the hybrid neural network model. The output of the hybrid neural network model is the cementing quality.

[0081] S103: Based on the data continuity of the input parameters, the input parameters are divided into continuous parameters and discontinuous parameters.

[0082] Among them, continuous parameters are time-series data that depend on time, while discontinuous parameters are static data that do not depend on time. For example, parameters related to well depth will change with the depth of the oil and gas well, that is, they have depth correlation. These parameters are continuous parameters. On the other hand, parameters such as the position of the centralizer will not change significantly over time. They are static and do not depend on time, that is, discontinuous parameters.

[0083] S104: Input the continuous and discontinuous parameters into the pre-trained hybrid neural network model for prediction to obtain the cementing quality of the target oil and gas well.

[0084] The hybrid neural network model is built upon a feedforward neural network (BP) model and a gated recurrent unit (GRU) model. For example... Figure 2 The diagram shows the structure of a hybrid neural network model, which includes two input layers (i.e., a feedforward neural network input layer and a gated recurrent unit input layer), a feature concatenation layer, a dropout layer, and three feedforward neural network layers. In the diagram, (2×10) indicates that the input dimension is 2 and the number of neurons is 10.

[0085] S104 specifically includes:

[0086] Read the stored model weights to obtain the pre-trained hybrid neural network model;

[0087] Continuous parameters are input into the gated recurrent unit input layer of the hybrid neural network model, and the data is processed by the gated recurrent unit model; and discontinuous parameters are input into the feedforward neural network input layer of the hybrid neural network model, and the data is processed by the feedforward neural network model.

[0088] The cementing quality of the target oil and gas well is obtained through prediction using a hybrid neural network model.

[0089] Continuous and discontinuous parameters have different characteristics. For example, directly inputting continuous parameters into a backpropagation (BP) network will result in the loss of the parameter's sequence information, thus requiring different neural network models. Specifically, discontinuous parameters are input into the feedforward neural network's input layer, while continuous parameters are input into the gated recurrent unit (GRU) input layer. By combining BP and GRU, the hybrid neural network model can simultaneously utilize static and time-series information, thereby obtaining more accurate prediction results.

[0090] Optionally, after S104, this embodiment further includes:

[0091] Calculate the SHAP value of each input parameter, and based on the largest SHAP value, obtain the input parameter with the greatest impact for quantitative and qualitative analysis of cementing quality.

[0092] The SHAP value is used to characterize the degree of influence of each input parameter on the output result. This embodiment uses the SHAP value (SHapley Additive exPlanations) to achieve interpretability of cementing quality influencing factors, mainly reflected in the interpretation of machine learning model prediction results. The SHAP value is a game theory-inspired method that aims to decompose the model's prediction results into the specific contribution of each input feature, thereby providing global and local interpretability of the model's predictions.

[0093] The core idea of ​​SHAP is to distribute the contribution of feature values ​​to different features, calculate the Shapley value for each feature, and multiply it by the feature value to obtain the contribution of that feature to the prediction result. The calculation formula is as follows:

[0094] SHAP=∑(f(x)-E[f(x)])(v(s∪{i})-v(S))

[0095] Where f(x) represents the output of the model, E[f(x)] represents the expected value of the output, v(S) represents the SHAP value of the feature subset S, and v(S∪{i}) represents the SHAP value of the new subset after adding feature i.

[0096] Example 2

[0097] Based on the above embodiments, combined with Figure 2 The structure diagram shown illustrates the following steps in the construction and training process of the hybrid neural network model:

[0098] Step 1: Construct an initial hybrid neural network model based on the feedforward neural network model (BP) and the gated recurrent unit model (GRU).

[0099] A single-layer feedforward neural network model contains only one output layer. The values ​​of the nodes in the output layer (output values) are directly obtained by multiplying the input values ​​by the weight values. Taking one element as an example, the transformation relationship from input to output is as follows:

[0100]

[0101] Where x = [x0, x1, x2, x3, ..., x n ] T It is the input feature vector, W ji It is x i To S j The connection weights, and the output y j (j=1,2,3,…m) represents the classification results based on different characteristics. S j It is the input (weighted sum) of the neuron, θ j It is a threshold (bias), used to adjust the activation of neurons, f(S) j ) represents the neuron's output after activation by the activation function.

[0102] A multilayer feedforward neural network has one input layer, one or more hidden layers, and one output layer. The input-output transformation relationship in a multilayer perceptron network is as follows:

[0103]

[0104] At this point, each layer is equivalent to a single-layer feedforward neural network. For example, for the q-th layer, it forms a q-th layer. n-1 A 3D hyperplane. It performs linear classification of the input patterns at this layer, but due to the combination of multiple layers, it can ultimately achieve more complex classification of the input patterns.

[0105] The Gated Recurrent Unit (GRU) model consists of update gates and reset gates. The reset gate rt controls how much information from the previous state is written into the current candidate set ht-1, and the update gate zt controls the extent to which information from the previous state is incorporated into the current state. The calculation formula is as follows:

[0106] z t =σ(W z h t-1 +U z x t +b z )

[0107] r t =σ(W r h t-1 +U r x t +b r )

[0108]

[0109] Where σ is the sigmoid activation function, W z W r W h U z U r U h b z b r b h These are the learning parameters for the neural network model. This represents the candidate hidden state of the current neuron. GRU has two significant features: the reset gate helps capture short-term dependencies in the sequence; the update gate helps capture long-term dependencies in the sequence.

[0110] The design principle of the hybrid neural network model in this embodiment is as follows: the Dropout layer is placed between the BP layers, rather than before or after them. In the feature extraction stage (BP 64-dimensional), this layer mainly learns non-linear features, and Dropout should not be performed too early, otherwise information loss will be significant. After Dropout in the low-dimensional BP layer (32→3), the model continues to reduce dimensionality before finally performing classification or regression. No further Dropout is added here because the network is already shallow enough, and the impact of regularization is minimal. Dropout is placed in the middle layer: here, Dropout has the greatest effect because deep neurons are prone to overfitting. This layer ensures that the model does not overly rely on certain features, improving generalization ability.

[0111] A three-layer backpropagation (BP) neural network was designed. A 64-neuron BP layer was used to fuse features extracted by GRU and static BP. A 32-neuron BP layer performed further feature extraction to reduce dimensionality. The final output layer consisted of a 3-neuron BP layer. Since this hybrid neural network model performs 3-class classification, the output layer must have 3 neurons.

[0112] Dropout layers are used to prevent overfitting and avoid the model fitting the training data too closely. A common principle is that Dropout layers are usually added after deeper backpropagation (BP) networks because deeper layers are prone to overfitting. Generally, only 1-2 Dropout layers are added; adding more will result in the loss of useful information.

[0113] Optionally, this embodiment uses two layers for both the gated recurrent unit (GRU) input layer and the feedforward neural network input layer. Taking the GRU input layer as an example, the first GRU input layer in this embodiment directly processes the raw time series data (2×10-dimensional input). This layer captures local temporal correlations. The second GRU input layer extracts high-level temporal features, and the output of the first GRU input layer is passed to the second GRU input layer, forming higher-dimensional temporal features (32-dimensional). Since a single GRU input layer may not be sufficient to learn complex temporal dependencies, it may lose long-term patterns. Increasing the depth helps improve the model's expressive power, enabling it to better learn complex time series features. Each additional layer significantly increases computational cost. Empirically, two input layers are sufficient, and many time series tasks also use two GRU or LSTM input layers. Experimental results show that two layers generally perform better than one layer, but are more stable than three layers.

[0114] Step 2: Perform data preprocessing and correlation analysis on the logging data of the sample oil and gas wells to obtain sample input parameters for model input.

[0115] For details on how to obtain the sample input parameters, please refer to step S101 of Example 1.

[0116] Step 3: Train the initial hybrid neural network model based on the sample input parameters until the initial hybrid neural network model converges, obtain the hybrid neural network model, and store the model weights of the hybrid neural network model.

[0117] Specifically, model training includes the following steps S1-S5:

[0118] S1. Divide the sample input parameters into a training set and a validation set.

[0119] All sample input parameters and data are shuffled and divided into training and validation sets in a 7:3 ratio, which are used for training and testing the hybrid neural network model, respectively.

[0120] S2. Train the constructed initial hybrid neural network model based on the training set.

[0121] The initial hybrid neural network model was trained using the following parameters: optimizer Adam, loss function crossentropy, initial learning rate 0.0001, batch size 128, and number of iterations 200.

[0122] This embodiment uses cross-entropy (loss function) to calculate the classification results of the model. Cross-entropy measures the difference between two different probability distributions of the same random variable, representing the difference between the true probability distribution and the predicted probability distribution. The smaller the cross-entropy value, the better the model's prediction performance. In the program, it is used in conjunction with the softmax function to process the output results so that the sum of the predicted values ​​for multiple categories is 1. Then, the loss is calculated using cross-entropy. The formula for calculating cross-entropy is as follows:

[0123]

[0124] Where M is a variable representing the number of categories; the parameter y ic This indicates whether the class belongs to the same category as sample i; 1 indicates the same category, and 0 indicates different category. Parameter p ic x represents the predicted probability that sample i belongs to class c, acc represents the classification accuracy of model f on dataset D, m is the total number of samples, and Loss represents the average loss of the model. i Let y represent the data of the i-th sample. i This represents the true label of the i-th sample. N is the total number of samples, and L... i It is the loss value of the i-th sample.

[0125] S3. Based on the training results and validation set of the initial hybrid neural network model, perform model testing to confirm that the initial hybrid neural network model does not exhibit overfitting or underfitting. Specifically,

[0126] The model was tested based on the training results and validation set of the initial hybrid neural network model, and the confusion matrix was obtained as follows:

[0127]

[0128] Among them, T P Indicates a true positive result, F N Indicates a false negative, F P Indicates a false positive, T N It indicates a true negative.

[0129] Based on the confusion matrix, the classification accuracy (also known as classification precision), recall, and precision of the hybrid neural network model are calculated.

[0130] Classification accuracy A cc The calculation formula is:

[0131]

[0132] The formula for calculating recall R is:

[0133]

[0134] The formula for calculating precision P is:

[0135]

[0136] By examining the classification accuracy, recall, and precision, one can determine whether the initial hybrid neural network model is overfitting or underfitting.

[0137] S4. Obtain the test accuracy of the model test.

[0138] S5. If the test accuracy is greater than or equal to the accuracy threshold, confirm that the initial hybrid neural network model has converged, obtain the hybrid neural network model, save the current model weights, and save the file for future reading. The accuracy threshold can be, for example, 90%.

[0139] If the test accuracy is less than the accuracy threshold, adjust the parameters of the initial hybrid neural network model and repeat steps S2-S5 until the test accuracy is greater than or equal to the accuracy threshold.

[0140] Example 3

[0141] Based on the above embodiments, this embodiment provides an application example.

[0142] like Figure 3 The diagram shown is another flowchart of a smart prediction method for cementing quality provided in this embodiment, including the following steps S201-S205:

[0143] S201. Well logging data collection and preprocessing.

[0144] The logging data used in the hybrid neural network model of this embodiment includes: well depth, azimuth, well diameter, inclination angle, sonic transit time, natural gamma, resistivity, spontaneous potential, casing centering, average grout return velocity, average displacement return velocity, cement slurry contact time, and centralizer position. Since the smallest unit of data input into the hybrid neural network model is 1m, if the logging data is not aligned, data preprocessing is required according to the following (1)-(4):

[0145] (1) Delete missing values ​​in well logging data.

[0146] The collected logging data usually contains missing values ​​such as incompleteness, inconsistency, inaccuracy, and missing data. In this embodiment, the method for handling missing values ​​is deletion. That is, when a missing value is found, the row containing the current data is directly deleted to prevent it from having an adverse effect on the hybrid neural network model.

[0147] The logging data is shown in Table 1. For example, if the well diameter at position 3002m and the grouting return velocity at position 3001m are missing, the entire row of data at positions 3002m and 3001m should be deleted to ensure the effectiveness of the model.

[0148] Table 1 Examples of logging data

[0149] well deep Azimuth Well diameter Sound wave time difference Gamma Casing centering Grouting return speed Plasma return rate 3000 15° 20 45 22 45 33 66 3001 15° 21 44 35 44 Null 64 3002 15° Null 46 61 46 35 65 3003 15° 21 43 44 43 32 62 … … … … … … … …

[0150] (2) Perform data interpolation on the well inclination and azimuth data in the logging data.

[0151] Most logging data uses a step size of 1 meter, but the well inclination and azimuth data are mostly 30 meters. Therefore, well depth and inclination data cannot be directly spliced ​​together. To ensure data alignment, data interpolation processing is required for the well inclination and azimuth data to obtain a more complete or continuous dataset. Optionally, this embodiment uses cubic spline interpolation to interpolate the well inclination data into a dataset with one data point per meter. By using piecewise continuous low-order polynomials to approximate the data, a smooth curve is obtained. The spline interpolation curve is smooth and continuous, and can better approximate the actual curve.

[0152] (3) Perform unique thermal encoding on discontinuous data in well logging data.

[0153] For discontinuous data in well logging data, such as centralizer locations, one-hot encoding is used. Alphabetical data is encoded using one-hot encoding to meet the input requirements of hybrid neural network models. For example, at a corresponding depth location, such as 3000m, if there is no centralizer, the data is 0; if there is a centralizer, the data is 1.

[0154] (4) Standardize and normalize the logging data.

[0155] Normalization is primarily used to handle differences in the dimensionality of data between quality features. For example, a sample may have one feature dimension ranging from 0 to 1, while another feature dimension ranges from 100 to 10000. It's difficult to compare their relative changes. To address this issue, this embodiment performs data normalization. The normalization formula is:

[0156]

[0157] Where x is the normalized and standardized feature data, μ is the mean of the feature data, σ is the standard deviation of the feature data, and X is the feature data.

[0158] The normalization formula is used to normalize all logging data except for the position of the centralizer.

[0159] This embodiment establishes a high-quality logging dataset through (1)-(4), which can provide a data foundation for subsequent models.

[0160] S202, Training of Hybrid Neural Network Model for Cementing Quality of Oil and Gas Wells.

[0161] First, correlation analysis is performed on the logging dataset obtained through (1)-(4) based on the Spearman correlation coefficient. When the correlation coefficient is less than 0, it is called negative correlation; when it is greater than 0, it is called positive correlation; and when it is equal to 0, it is called zero correlation. In this embodiment, a preset threshold of 0.3 is used as an example to filter the influence parameters with an absolute value of correlation coefficient greater than 0.3 as input parameters of the hybrid neural network model.

[0162] Based on correlation analysis, this embodiment selects well depth, azimuth, well diameter, well inclination angle, sonic transit time, natural gamma, resistivity, spontaneous potential, casing centering, average return velocity of grouting, average return velocity of grout replacement, cement slurry contact time, and centralizer position as input parameters for the hybrid neural network model. The output of the hybrid neural network model is the cementing quality.

[0163] The process of building and training a hybrid neural network model includes:

[0164] Dataset partitioning: All sample input parameters and data are shuffled and divided into training and validation sets in a 7:3 ratio, which are used for training and testing the hybrid neural network model, respectively.

[0165] Model Setup: Based on the PyTorch deep learning architecture, using a feedforward neural network (BP) model and a gated recurrent unit (GRU) model, following... Figure 2 The diagram shown establishes a hybrid neural network model. The hybrid neural network model includes two input layers, a feature concatenation layer, a dropout layer, and three feedforward neural network layers. In the diagram, (2×10) indicates that the input dimension is 2 and the number of neurons is 10.

[0166] Model Training: The hybrid neural network model was trained using the following parameters: optimizer Adam, loss function crossentropy, initial learning rate 0.0001, batch size 128, and number of iterations 200.

[0167] S203, Evaluation and Parameter Optimization of Hybrid Neural Network Model for Cementing Quality of Oil and Gas Wells.

[0168] In practice, the evaluation and parameter tuning of the hybrid neural network model for cementing quality in oil and gas wells includes the following steps:

[0169] Based on the training results of the hybrid neural network model and the validation set, the model is tested to confirm that it does not exhibit overfitting or underfitting. Specifically,

[0170] The model was tested based on the training results and validation set of the hybrid neural network model, and the confusion matrix was obtained as follows:

[0171]

[0172] Among them, T P Indicates a true positive result, F N Indicates a false negative, F P Indicates a false positive, T N It indicates a true negative.

[0173] Based on the confusion matrix, the classification accuracy (also known as classification precision), recall, and precision of the hybrid neural network model are calculated.

[0174] Classification accuracy A cc The calculation formula is:

[0175]

[0176] The formula for calculating recall R is:

[0177]

[0178] The formula for calculating precision P is:

[0179]

[0180] By examining classification accuracy, recall, and precision, one can determine whether the hybrid neural network model is overfitting or underfitting.

[0181] Obtain the test accuracy of the model. If the test accuracy is greater than or equal to the accuracy threshold, confirm that the hybrid neural network model has converged, save the current model weights, and save the file for future reading. The accuracy threshold can be, for example, 90%.

[0182] If the test accuracy is less than the accuracy threshold, the parameters of the hybrid neural network model are adjusted and training is repeated until the test accuracy is greater than or equal to the accuracy threshold.

[0183] S204. Intelligent prediction of cementing quality in oil and gas wells.

[0184] Referring to the steps in S201 of this embodiment, a dataset of oil and gas wells to be predicted is established.

[0185] Read the model weights saved in S203 and retrieve the pre-trained hybrid neural network model;

[0186] Using a trained hybrid neural network model, input parameters are given to the oil and gas well to be predicted, and the hybrid neural network model outputs the cementing quality per meter of the oil and gas well to be predicted.

[0187] S205. Analysis and calculation of factors affecting cementing quality of oil and gas wells.

[0188] Calculate the SHAP value of each input parameter, and based on the largest SHAP value, obtain the input parameter with the greatest impact for quantitative and qualitative analysis of cementing quality.

[0189] The SHAP value is used to characterize the degree of influence of each input parameter on the output result. This embodiment uses the SHAP value (SHapley Additive exPlanations) to achieve interpretability of cementing quality influencing factors, mainly reflected in the interpretation of machine learning model prediction results. The SHAP value is a game theory-inspired method that aims to decompose the model's prediction results into the specific contribution of each input feature, thereby providing global and local interpretability of the model's predictions.

[0190] The core idea of ​​SHAP is to distribute the contribution of feature values ​​to different features, calculate the Shapley value for each feature, and multiply it by the feature value to obtain the contribution of that feature to the prediction result. The calculation formula is as follows:

[0191] SHAP=∑(f(x)-E[f(x)])(v(s∪{i})-v(S))

[0192] Where f(x) represents the output of the model, E[f(x)] represents the expected value of the output, v(S) represents the SHAP value of the feature subset S, and v(S∪{i}) represents the SHAP value of the new subset after adding feature i.

[0193] This embodiment uses the PyTorch deep learning framework to implement a hybrid neural network model for cementing quality in oil and gas wells. Different neural network models are input into the data characteristics of well logging, as well as continuous and discontinuous data, to ultimately output the cementing quality. The hybrid neural network model in this embodiment was trained using actual data from the oilfield, achieving a stable loss value of 0.12 and a cementing quality prediction accuracy of up to 93%, which meets engineering requirements.

[0194] Example 4

[0195] Based on the above embodiments, this embodiment provides a smart cementing quality prediction device, such as... Figure 4 As shown, the intelligent prediction device for cementing quality includes:

[0196] The data processing module 401 is used to calculate the correlation between each logging data and cementing quality based on the logging data of the target oil and gas well using the Spearman correlation coefficient, and to obtain the correlation coefficient of each logging data. The logging data is the data obtained by measuring and recording the target oil and gas well during the drilling process.

[0197] The input parameter selection module 402 is used to select logging data with an absolute value of correlation coefficient greater than a preset threshold as input parameters for predicting cementing quality.

[0198] The parameter classification module 403 is used to classify the input parameters into continuous parameters and discontinuous parameters based on the data continuity of the input parameters. The continuous parameters are time series data that depend on time series, and the discontinuous parameters are static data that do not depend on time series.

[0199] The intelligent prediction module 404 is used to input the continuous parameters and the discontinuous parameters into a pre-trained hybrid neural network model to obtain the predicted cementing quality of the target oil and gas well.

[0200] In one or more embodiments, the data processing module 401 is used to: perform data preprocessing on the logging data of the target oil and gas well;

[0201] The data preprocessing includes: deleting missing values ​​in the logging data; performing data interpolation on the well inclination and azimuth data in the logging data; performing one-hot encoding on discontinuous data in the logging data; and performing normalization on the logging data.

[0202] Based on the preprocessed logging data, the correlation between each logging data point and cementing quality is calculated using the Spearman correlation coefficient.

[0203] In one or more embodiments, the data processing module 401 is further configured to: obtain the average value and standard deviation of the feature data based on the feature data in the logging data;

[0204] Calculate the difference between the feature data and the average value of the feature data;

[0205] Based on the ratio of the difference to the standard deviation, the normalized feature data is calculated.

[0206] The cementing quality intelligent prediction device is also used to: after obtaining the cementing quality of the target oil and gas well, calculate the SHAP value of each input parameter, wherein the SHAP value is used to characterize the degree of influence of each input parameter on the cementing quality of the target oil and gas well;

[0207] Based on the largest SHAP value, the input parameters with the greatest impact are obtained for quantitative and qualitative analysis of cementing quality.

[0208] In one or more embodiments, the intelligent prediction module 404 is used to: read the stored model weights to obtain a pre-trained hybrid neural network model;

[0209] The continuous parameters are input into the gated recurrent unit input layer of the hybrid neural network model, and the gated recurrent unit model performs data processing; and the discontinuous parameters are input into the feedforward neural network input layer of the hybrid neural network model, and the feedforward neural network model performs data processing.

[0210] The cementing quality of the target oil and gas well is obtained through the prediction of the hybrid neural network model.

[0211] The intelligent cementing quality prediction device is also used in the construction and training process of the hybrid neural network model, which includes:

[0212] A hybrid neural network model is constructed based on a feedforward neural network model and a gated recurrent unit model; the hybrid neural network model includes two input layers, a feature concatenation layer, a dropout layer, and three feedforward neural network layers.

[0213] Data preprocessing and correlation analysis were performed on the logging data of the sample oil and gas wells to obtain sample input parameters for model input;

[0214] The hybrid neural network model is trained based on the sample input parameters until the hybrid neural network model converges, and the model weights of the current hybrid neural network model are stored.

[0215] The cementing quality intelligent prediction device and the cementing quality intelligent prediction method provided in this disclosure are based on the same inventive concept and have the same beneficial effects as the methods they adopt, operate or implement.

[0216] Example 5

[0217] Based on the above embodiments, this disclosure also provides a computer device to execute the above-described intelligent prediction method for cementing quality. Please refer to... Figure 5 It illustrates a schematic diagram of a computer device provided by some embodiments of this disclosure. For example... Figure 5As shown, the computer device 50 includes: a processor 500, a memory 501, a bus 502, and a communication interface 503. The processor 500, the communication interface 503, and the memory 501 are connected via the bus 502. The memory 501 stores a computer program that can run on the processor 500. When the processor 500 runs the computer program, it executes the cementing quality intelligent prediction method provided in any of the foregoing embodiments of this disclosure.

[0218] The memory 501 may include high-speed random access memory (RAM) or non-volatile memory, such as at least one disk storage device. Communication between this device network element and at least one other network element is achieved through at least one communication interface 503 (which can be wired or wireless), such as the Internet, wide area network, local area network, metropolitan area network, etc.

[0219] Bus 502 can be an ISA bus, PCI bus, or EISA bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. The memory 501 is used to store programs. After receiving an execution instruction, the processor 500 executes the program. The intelligent prediction method for cementing quality disclosed in any of the foregoing embodiments of this disclosure can be applied to the processor 500, or implemented by the processor 500.

[0220] The processor 500 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of the processor 500 or by instructions in software form. The processor 500 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), an off-the-shelf programmable gate array (FPTA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this disclosure. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this disclosure can be directly embodied in the execution of a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software modules may reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The storage medium is located in memory 501. The processor 500 reads the information in memory 501 and, in conjunction with its hardware, completes the steps of the above method.

[0221] The computer equipment provided in this disclosure and the intelligent prediction method for cementing quality provided in this disclosure are based on the same inventive concept and have the same beneficial effects as the methods they employ, operate, or implement.

[0222] This disclosure also provides a computer-readable storage medium corresponding to the intelligent prediction method for cementing quality provided in the foregoing embodiments. The computer-readable storage medium is an optical disc, on which a computer program (i.e., a computer program product) is stored. When the computer program is run by a processor, it executes the intelligent prediction method for cementing quality provided in any of the foregoing embodiments.

[0223] It should be noted that examples of the computer-readable storage medium may also include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other optical and magnetic storage media, which will not be elaborated here.

[0224] The computer-readable storage medium provided in the above embodiments of this disclosure and the intelligent prediction method for cementing quality provided in the embodiments of this disclosure are based on the same inventive concept and have the same beneficial effects as the methods adopted, run or implemented by the application programs stored therein.

[0225] This disclosure also provides a computer program product; please refer to [reference needed]. Figure 6 The computer program product 600 carries program code, namely computer program 601. The instructions included in the computer program 601 can be used to execute the steps of the intelligent prediction method for cementing quality described in the above method embodiments. For details, please refer to the above method embodiments, which will not be repeated here.

[0226] The aforementioned computer program product can be implemented through hardware, software, or a combination thereof. In one optional embodiment, the computer program product is specifically embodied in a computer storage medium; in another optional embodiment, the computer program product is specifically embodied in a software product, such as a software development kit (SDK), etc.

[0227] Computer-readable storage media can be implemented by any type of volatile or non-volatile storage device or a combination thereof. Computer-readable storage media may include, but are not limited to, random access memory (RAM), read-only memory (ROM), flash memory, EPROM memory, EEPROM memory, registers, and computer storage media (e.g., hard disks, floppy disks, solid-state drives, removable disks, CD-ROMs, DVD-ROMs, Blu-ray discs, etc.).

[0228] Computer-readable storage media may also store at least one computer-executable program / instruction, such as computer-readable instructions. Computer-readable storage media include, but are not limited to, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and / or cache memory. Computer-readable storage media may include, for example, read-only memory (ROM), hard disk, flash memory, etc. For example, a non-transitory computer-readable storage medium may be connected to a computing device such as a computer, and then, when the computing device executes the computer-readable instructions stored on the computer-readable storage medium, the various methods described above can be performed.

[0229] In addition, the computer device may include (but is not limited to) a data bus, an input / output (I / O) bus, a display, and input / output devices (e.g., keyboard, mouse, speakers, etc.).

[0230] The processor can communicate with external devices via the I / O bus through wired or wireless networks.

[0231] In one embodiment, the at least one computer-executable instruction may also be compiled into or comprise a software product / computer program product, wherein one or more computer-executable instructions are executed by a processor to perform the steps of the various functions and / or methods in the embodiments described herein.

[0232] The basic principles of this disclosure have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in this disclosure are merely examples and not limitations, and should not be considered as essential features of each embodiment of this disclosure. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not limit the scope of this disclosure to the necessity of employing the aforementioned specific details for implementation.

[0233] The block diagrams of devices, apparatuses, devices, and systems disclosed herein are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, devices, and systems can be connected, arranged, and configured in any manner. Words such as “comprising,” “including,” “having,” etc., are open-ended terms meaning “including but not limited to,” and are used interchangeably with them. The terms “or” and “and” as used herein refer to the terms “and / or,” and are used interchangeably with them unless the context clearly indicates otherwise. The term “such as” as used herein refers to the phrase “such as but not limited to,” and is used interchangeably with it.

[0234] Additionally, as used herein, the “or” used in a list of items beginning with “at least one” indicates a separate list, such that a list of, for example, “at least one of A, B, or C” means A or B or C, or AB or AC or BC, or ABC (i.e., A and B and C). Furthermore, the word “exemplary” does not imply that the described example is preferred or better than other examples.

[0235] It should also be noted that in the systems and methods of this disclosure, the components or steps can be decomposed and / or recombined. These decompositions and / or recombinations should be considered as equivalent solutions to this disclosure.

[0236] Various changes, substitutions, and modifications can be made to the technology described herein without departing from the teachings defined by the appended claims. Furthermore, the scope of the claims of this disclosure is not limited to the specific aspects of the processes, machines, manufactures, events, means, methods, and actions described above. Currently existing or later-developed processes, machines, manufactures, events, means, methods, or actions that perform substantially the same function or achieve substantially the same result as the corresponding aspects described herein can be utilized. Therefore, the appended claims include such processes, machines, manufactures, events, means, methods, or actions within their scope.

[0237] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use this disclosure. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of this disclosure. Therefore, this disclosure is not intended to be limited to the aspects shown herein, but rather to be carried out within the widest scope consistent with the principles and novel features disclosed herein.

[0238] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of this disclosure to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations therein.

Claims

1. A method for intelligent prediction of cementing quality, characterized in that, include: Based on the logging data of the target oil and gas well, the correlation between each logging data and cementing quality is calculated using the Spearman correlation coefficient to obtain the correlation coefficient of each logging data. The logging data is the data obtained by measuring and recording the target oil and gas well during the drilling process. Logging data with an absolute value greater than a preset threshold are selected as input parameters for predicting cementing quality. Based on the data continuity of the input parameters, the input parameters are divided into continuous parameters and discontinuous parameters. The continuous parameters are time-series data that depend on time series, and the discontinuous parameters are static data that do not depend on time series. The continuous parameters and the discontinuous parameters are respectively input into a pre-trained hybrid neural network model for prediction to obtain the cementing quality of the target oil and gas well.

2. The method according to claim 1, characterized in that, Based on logging data from the target oil and gas wells, the correlation between the logging data and cementing quality is calculated using the Spearman correlation coefficient, including: Data preprocessing is performed on logging data from the target oil and gas wells; The data preprocessing includes: deleting missing values ​​in the logging data; performing data interpolation on the well inclination and azimuth data in the logging data; performing one-hot encoding on discontinuous data in the logging data; and performing normalization on the logging data. Based on the preprocessed logging data, the correlation between each logging data point and cementing quality is calculated using the Spearman correlation coefficient.

3. The method according to claim 2, characterized in that, The logging data is normalized and standardized, including: Based on the characteristic data in the logging data, the average value and standard deviation of the characteristic data are obtained; Calculate the difference between the feature data and the average value of the feature data; Based on the ratio of the difference to the standard deviation, the normalized feature data is calculated.

4. The method according to claim 1, characterized in that, After obtaining the cementing quality of the target oil and gas well, the process also includes: Calculate the SHAP value of each input parameter, wherein the SHAP value is used to characterize the degree of influence of each input parameter on the cementing quality of the target oil and gas well; Based on the largest SHAP value, the input parameters with the greatest impact are obtained for quantitative and qualitative analysis of cementing quality.

5. The method according to claim 1, characterized in that, The construction and training process of the hybrid neural network model includes: An initial hybrid neural network model is constructed based on a feedforward neural network model and a gated recurrent unit model. The initial hybrid neural network model includes a gated recurrent unit input layer, a feedforward neural network input layer, a feature splicing layer, a dropout layer, and three feedforward neural network layers. Data preprocessing and correlation analysis were performed on the logging data of the sample oil and gas wells to obtain sample input parameters for model input; The initial hybrid neural network model is trained based on the sample input parameters until it converges, thereby obtaining the hybrid neural network model and storing its model weights.

6. The method according to claim 1, characterized in that, The continuous parameters and the discontinuous parameters are respectively input into a pre-trained hybrid neural network model for prediction to obtain the cementing quality of the target oil and gas well, including: Read the stored model weights to obtain the pre-trained hybrid neural network model; The continuous parameters are input into the gated recurrent unit input layer of the hybrid neural network model, and the gated recurrent unit model performs data processing; and the discontinuous parameters are input into the feedforward neural network input layer of the hybrid neural network model, and the feedforward neural network model performs data processing. The cementing quality of the target oil and gas well is obtained through the prediction of the hybrid neural network model.

7. A smart cementing quality prediction device, characterized in that, include: The data processing module is used to calculate the correlation between each logging data and cementing quality based on the logging data of the target oil and gas well using the Spearman correlation coefficient, and obtain the correlation coefficient of each logging data. The logging data is the data obtained by measuring and recording the target oil and gas well during the drilling process. The input parameter selection module is used to select logging data with an absolute value of correlation coefficient greater than a preset threshold as input parameters for predicting cementing quality. The parameter classification module is used to classify the input parameters into continuous parameters and discontinuous parameters based on the data continuity of the input parameters. The continuous parameters are time series data that depend on time series, and the discontinuous parameters are static data that do not depend on time series. The intelligent prediction module is used to input the continuous parameters and the discontinuous parameters into a pre-trained hybrid neural network model to obtain the predicted cementing quality of the target oil and gas well.

8. A computer device, comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the steps of the method according to any one of claims 1 to 6.

10. A computer program product comprising a computer program / instructions, characterized in that, When executed by a processor, the computer program implements the steps of the method according to any one of claims 1 to 6.

Images