High-position directional long drill hole three-dimensional space parameter prediction method

Abstract

The invention relates to a high-position directional long drill hole three-dimensional space parameter prediction method, and belongs to the field of coal mine drilling. Aiming at the problems of low prediction precision, single data, poor model adaptability and the like existing in a traditional high-level borehole design method, the invention provides an intelligent prediction model based on improved gray correlation analysis GRA and a Bat algorithm optimized back propagation neural network BA-BP. According to the method, gray correlation analysis is optimized through a dynamic weight adjustment mechanism, influence factors are screened, the weight and bias parameters of the BP neural network are optimized in combination with a BA algorithm, and the model prediction precision and the iteration efficiency are improved. The model can comprehensively consider seven key parameters such as the mining height, the mining face advancing speed and the rock stratum compressive strength, accurate prediction of the vertical distance and the horizontal distance of the drill hole is achieved, the gas extraction efficiency and the mine safety are remarkably improved, and high engineering application value is achieved.

Description

Technical Field

[0001] The present invention belongs to the field of coal mine drilling, and relates to a method for predicting three-dimensional space parameters of high-position directional long boreholes. Background Art

[0002] With the continuous development of coal mining technology, the problem of goaf gas has become increasingly prominent, seriously threatening the safe production of coal mines. As an effective gas drainage method, the layout horizon of the trajectory of high-position directional long boreholes directly affects the gas drainage efficiency. Therefore, how to accurately predict the three-dimensional space parameters of high-position directional long boreholes has become an urgent problem to be solved in coal mine safe production.

[0003] Traditional borehole design methods mainly rely on theoretical calculations, simulation analysis and empirical judgments. Although these methods can guide borehole design to a certain extent, they have the following limitations:

[0004] Mainly based on idealized models, they cannot fully consider the complexity of actual mine conditions, such as geological structures, coal seam occurrence states, etc., resulting in deviations between prediction results and actual situations.

[0005] They consume a large amount of time and computing resources, and the simulation results are greatly affected by the setting of model parameters, making it difficult to ensure the accuracy of prediction results.

[0006] Relying on the experience and professional knowledge of designers, they are easily affected by subjective factors and lack objectivity and scientificity.

[0007] In recent years, with the rapid development of artificial intelligence technology, technologies such as neural networks and machine learning have been widely used in the field of coal mine safe production. These technologies can effectively process complex data information and establish non-linear mapping relationships, providing new ideas for predicting the three-dimensional space parameters of high-position directional long boreholes.

[0008] However, there are still some problems in the existing prediction methods based on neural networks:

[0009] The setting of parameters of the neural network model has a great influence on the prediction results, and it is necessary to conduct repeated experiments and adjustments according to specific problems, lacking theoretical guidance.

[0010] Some neural network models perform well on training samples, but have poor generalization ability on test samples and are difficult to be popularized and applied.

[0011] Therefore, the present invention aims to provide a BA-BP neural network prediction model based on improved grey relational degree analysis, which can effectively solve the limitations of existing methods and realize the intelligent and accurate prediction of three-dimensional space parameters of high-position directional long boreholes. Summary of the Invention

[0012] In view of this, the purpose of the present invention is to provide a method for predicting the three-dimensional space parameters of high-level directional long boreholes. In the prior art, there are problems such as the inability to fully consider the complex and variable actual working conditions of the mine, low prediction accuracy, and complex and time-consuming operations. Therefore, in order to solve the above problems, the present invention provides a method for predicting the three-dimensional space parameters of high-level directional long boreholes.

[0013] To achieve the above object, the present invention provides the following technical solutions:

[0014] A method for predicting the three-dimensional space parameters of high-level directional long boreholes, comprising the following steps:

[0015] Step 1: Collect data on multiple factors affecting the three-dimensional space parameters of high-level directional long boreholes, including burial depth, coal seam dip angle, mining height, coal mining advance speed, rock stratum structure, uniaxial compressive strength of overlying rock, and goaf diagonal length;

[0016] Step 2: Use the improved grey relational analysis method to quantitatively evaluate the correlation between each influencing factor and the borehole effect, and screen out the influencing factors;

[0017] Step 3: Construct a BA-BP neural network model, including an input layer, a hidden layer, and an output layer, where the number of nodes in the input layer corresponds to the number of influencing factors, and the number of nodes in the output layer is 2, corresponding to the predicted borehole trajectory parameters respectively;

[0018] Step 4: Use the particle swarm optimization algorithm to optimize the weight and bias parameters of the BP neural network;

[0019] Step 5: Use the trained BA-BP neural network model to predict the three-dimensional space parameters of high-level directional long boreholes.

[0020] Further, the improved grey relational analysis method includes the following steps:

[0021] Step 2-1: Dynamically adjust the weight coefficients of each influencing factor according to the data change trend and amplitude;

[0022] Step 2-2: Calculate the correlation degree between each influencing factor and the borehole effect, and screen out the influencing factors according to the correlation degree.

[0023] Further, the particle swarm optimization algorithm includes the following steps:

[0024] Step 4-1: Initialize the particle swarm, and each particle represents a set of weight and bias parameters of the BP neural network;

[0025] Step 4-2: Calculate the fitness value of each particle, that is, the error between the prediction result and the actual data;

[0026] Step 4-3: Update the velocity and position of each particle according to the fitness value;

[0027] Step 4-4: Repeat Step 4-2 and 4-3 until the termination condition is met.

[0028] Furthermore, the three-dimensional space parameters of the high-level directional long borehole include the vertical distance from the coal seam roof and the horizontal distance.

[0029] A prediction system for the three-dimensional space parameters of a high-level directional long borehole, the system comprising:

[0030] A data acquisition module for acquiring data on multiple factors affecting the three-dimensional space parameters of the high-level directional long borehole;

[0031] A data processing module for preprocessing and feature extraction of the acquired data;

[0032] A model training module for training a BA-BP neural network model;

[0033] A model prediction module for predicting the three-dimensional space parameters of the high-level directional long borehole;

[0034] A result display module for displaying the prediction results.

[0035] The beneficial effects of the present invention are as follows:

[0036] (1) The improved grey relational analysis method can effectively screen out the main influencing factors, and the BA-BP neural network model can establish a more accurate prediction model, improving the accuracy of the prediction results.

[0037] (2) Compared with the traditional theoretical calculation, simulation analysis and empirical judgment methods, the present invention can quickly predict the borehole trajectory parameters and significantly improve the design efficiency.

[0038] (3) By precisely controlling the borehole trajectory, the gas drainage efficiency is improved, the risk of gas accumulation and overrun is reduced, and the safe production of coal mines is ensured.

[0039] (4) The optimized borehole trajectory can better match the gas occurrence state, improve the gas drainage efficiency, and promote the efficient utilization of resources.

[0040] (5) The present invention applies artificial intelligence technology to the field of coal mine drilling, promotes the intelligent development of coal mines, and provides technical support for the construction of smart mines.

[0041] Other advantages, objects, and features of the present invention will be set forth in part in the following description, and in part will be obvious to those skilled in the art upon examination of the following, or may be learned by practice of the present invention. The objects and other advantages of the present invention may be realized and obtained by the following description of the specification. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be described in detail preferably with reference to the accompanying drawings, where:

[0043] Figure 1 For optimizing the flowchart;

[0044] Figure 2 For the high-level borehole three-dimensional space parameter BA-BP neural network model structure. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0045] The following specific examples illustrate the embodiments of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments. The details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the drawings provided in the following embodiments only illustrate the basic concept of the present invention schematically. Without conflict, the following embodiments and the features in the embodiments can be combined with each other.

[0046] Among them, the drawings are only for illustrative purposes, showing only schematic diagrams, not physical diagrams, and should not be construed as limiting the present invention; in order to better illustrate the embodiments of the present invention, some components in the drawings will be omitted, enlarged, or reduced, which do not represent the dimensions of the actual product; for those skilled in the art, it is understandable that some well-known structures and their descriptions in the drawings may be omitted.

[0047] In the drawings of the embodiments of the present invention, the same or similar reference numerals correspond to the same or similar components; in the description of the present invention, it should be understood that if there are terms such as "upper", "lower", "left", "right", "front", "rear", etc. indicating the orientation or positional relationship, they are based on the orientation or positional relationship shown in the drawings. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation. Therefore, the terms describing the positional relationship in the drawings are only for illustrative purposes and should not be construed as limiting the present invention. For those of ordinary skill in the art, the specific meanings of the above terms can be understood according to specific circumstances.

[0048] Please refer to Figure 1 andFigure 2 , Improve the Grey Relational Analysis (GRA): Grey Relational Analysis is based on grey system theory and evaluates the correlation degree between sequences by calculating the similarity between them. The key lies in determining the reference sequence and the comparison sequences, and analyzing the internal relationship between data by calculating the correlation degree between them. Aiming at the limitations of fixed weight allocation and insufficient data dynamics, a dynamic weight adjustment mechanism based on the data change trend and amplitude is proposed. By analyzing the historical and current states of the data, the weight coefficients are dynamically adjusted to more accurately reflect the actual influence of each factor. An improved correlation degree calculation formula is introduced, which incorporates the dynamic weight coefficient W i , and the calculation formula is:

[0049]

[0050] δ′(X 0 ,X i )=W i ·δ(X 0 ,X i )

[0051] where W i is dynamically determined according to the data change trend and amplitude.

[0052] Steps for comprehensive evaluation using grey relational analysis:

[0053] (1) Determine the evaluation index system according to the evaluation purpose and collect evaluation data. Suppose n data sequences form the following matrix:

[0054]

[0055] where m is the number of indicators,

[0056] X i ′=(X i ′(1),X i ′(2),···X i ′(m)) T , i=1,2,···,n

[0057] (2) Determine the reference data series

[0058] The reference data series should be an ideal comparison standard. It can be composed of the optimal values (or the worst values) of each indicator to form the reference data series, or other reference values can be selected according to the evaluation purpose.

[0059] X 0 ′=(x′ 0 (1), x′ 0 (2),…, x′ 0 (m))

[0060] (3) Nondimensionalize the index data

[0061] The dimensionless data sequence forms the following matrix:

[0062]

[0063] (4) Calculate the absolute difference between the corresponding elements of the index sequence (comparison sequence) of each evaluated object and the reference sequence one by one, that is, x 0 (k) - x i (k), i = 1, …, n; k = 1, …, m, where n is the number of evaluated objects.

[0064] (5) Determine min i＝1 n min k＝1 m|x 0 (k) - x i (k)| and max i＝1 n max k＝1 m|x 0 (k) - x i (k)|.

[0065] (6) Calculate the correlation coefficient.

[0066] Calculate the correlation coefficients between the corresponding elements of each comparison sequence and the reference sequence respectively.

[0067]

[0068] In the formula, ρ is the resolution coefficient, which takes values in (0, 1). If ρ is smaller, the difference between the correlation coefficients is greater and the discrimination ability is stronger. ρ takes 0.5.

[0069]

[0070] (7) Calculate the correlation order

[0071] Calculate the mean value of the correlation coefficients between the corresponding elements of each index of each evaluated object (comparison sequence) and the reference sequence respectively to reflect the correlation between each evaluated object and the reference sequence, and call it the correlation order, denoted as:

[0072]

[0073] (8) If the roles played by each index in the comprehensive evaluation are different, the weighted average value of the correlation coefficients can be calculated, that is

[0074]

[0075] In the formula: W K is the weight of each index.

[0076] (9) Obtain the comprehensive evaluation result according to the association order of each observation object.

[0077] Sample selection and data processing: Construct a BA-BP neural network to analyze the influencing factors of the three-dimensional space parameters of high-position directional long boreholes, and select the actual trajectory parameter data of high-position directional long boreholes constructed in a specific mining area for comprehensive evaluation. Refer to the measured data of M typical mines in China as the sample set for BA-BP neural network training and testing. Use Matlab software for simulation training. From the collected data, the data of the first M - N mines are used as the learning samples for network training, while the data of the remaining N mines are used as the test samples for model performance verification. During the BA-BP neural network training process, the sample set selected for the three-dimensional space parameters of high-position directional long boreholes includes burial depth (f 1 ), coal seam dip angle (f 2 ), mining height (f 3 ), coal mining advance speed (f 4 ), rock stratum structure (f 5 ), uniaxial compressive strength of overlying rock (f 6 ), and inclined length of goaf (f 7 ) and other 7 factors. The input vector P(7×M - N) represents the factor source affecting the three-dimensional space parameters of high-position directional long boreholes, while the target output T(2×M - N) corresponds to the actual vertical distance (vertical distance from the coal seam roof) and horizontal distance (horizontal projection distance from the goaf side of the return airway) of each sample. There are many influencing parameter data with large differences, and it is necessary to normalize each data to minimize the impact on the prediction result. Use min-max normalization to scale the range of data features to between the minimum and maximum values ([0.1]) to achieve normalization.

[0078]

[0079] Where: X norm is the processed data; X is the sample data; X max is the maximum value in the sample data; X min is the minimum value in the sample data.

[0080] Parameter determination: The BA-BP neural network consists of three layers: an input layer, a hidden layer, and an output layer, forming a three-layer backpropagation neural network. The input layer has 7 nodes, the hidden layer has M - N nodes, and the output layer has 2 nodes. The sigmoid activation function is used to achieve the non-linear mapping between the hidden layer and the output layer. The measured data sample sets of M mines are used as the training population, and the measured data of N mines are used as the prediction samples. The value range of each parameter in the algorithm is limited between [-1, 1]. The population size is set to ≥60 to ensure the diversity of the algorithm. The initial pulse loudness is ≤0.3, the pulse emission rate is ≤0.6, both the pulse loudness attenuation coefficient and the pulse emission rate increase coefficient are ≤0.8, and the frequency range of echolocation is between [0, 3] to simulate the unique navigation and hunting behaviors of bats. The upper limit of the number of algorithm iterations is set to ≤6000 times, and the prediction error target is ≤0.001.

[0081] Prediction model: First, the grey relational analysis is used to quantitatively evaluate the correlation between each influencing factor and the borehole effect, and the influencing factors are screened out. The particle swarm optimization BA algorithm is used to optimize the weight and bias parameters of the BP neural network to improve the prediction accuracy and iteration rate.

[0082] The following takes a certain coal mine as an example to illustrate the implementation process of the present invention:

[0083] 1. Data collection

[0084] Collect data such as the mining depth, coal seam dip angle, mining height, coal mining advance speed, rock stratum structure, uniaxial compressive strength of overlying rock, and goaf inclined length of this coal mine.

[0085] Collect the actual trajectory parameter data of the long high-level directional boreholes that have been constructed in this coal mine, including the vertical distance and horizontal distance from the coal seam roof.

[0086] 2. Sample selection

[0087] Select the actual trajectory parameter data of 30 long high-level directional boreholes that have been constructed in this coal mine as the training samples.

[0088] Select 10 prediction samples of long high-level directional boreholes that have not been constructed in this coal mine.

[0089] 3. Data processing

[0090] Normalize the sample data to scale the range of data features to between [0, 1].

[0091] 4. Model construction

[0092] Construct a BA-BP neural network model with 7 inputs and 2 outputs.

[0093] Set the population size to 100, the upper limit of the number of iterations to 6000, and the prediction error target to 0.001.

[0094] 5. Model Training

[0095] Use the improved GRA to quantitatively evaluate the correlation between each influencing factor and the drilling effect, and screen out the influencing factors.

[0096] Adopt the BA algorithm to optimize the weight and bias parameters of the BP neural network.

[0097] 6. Model Prediction

[0098] Use the trained model to predict 10 prediction samples to obtain the predicted drilling trajectory parameters.

[0099] 7. Result Analysis

[0100] Compare and analyze the prediction results with the actual drilling data to evaluate the prediction accuracy of the model.

[0101] The results show that the prediction accuracy of the model is high, the average error is controlled within 5%, and the degree of coincidence with the actual drilling data is high.

[0102] 8. Application Effect

[0103] For the high-position directional long boreholes designed by the method of the present invention, the gas extraction efficiency is increased by 30%, and the risk of gas accidents is significantly reduced.

[0104] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the purpose and scope of the present technical solution, and they should all be covered within the scope of the claims of the present invention.

Claims

1. A method for predicting three-dimensional spatial parameters of high-position directional long drilling, characterized in that: The following steps are involved: Step 1: Collect data on multiple factors that affect the three-dimensional spatial parameters of high-position directional long drilling, including burial depth, coal seam inclination, mining height, coal mining advancement speed, rock layer structure, uniaxial compressive strength of overburden rock, and oblique length of goaf; Step 2: Use the improved grayscale correlation analysis method to quantitatively evaluate the correlation between each influencing factor and the drilling effect, and screen out the influencing factors; Step 3: Construct a BA-BP neural network model, including an input layer, a hidden layer, and an output layer, where the number of nodes in the input layer corresponds to the number of influencing factors, and the number of nodes in the output layer is 2, corresponding to the predicted drilling trajectory parameters; Step 4: Use particle swarm optimization algorithm to optimize the weight and bias parameters of BP neural network; Step 5: Use the trained BA-BP neural network model to predict the three-dimensional spatial parameters of high-position directional long drilling holes.

2. The method for predicting three-dimensional spatial parameters of high-position directional long drilling according to claim 1 is characterized in that: The improved grayscale correlation analysis method comprises the following steps: Step 2-1: Dynamically adjust the weight coefficient of each influencing factor according to the trend and magnitude of data changes; Step 2-2: Calculate the correlation between each influencing factor and the drilling effect, and select the influencing factors based on the correlation.

3. The method for predicting three-dimensional spatial parameters of high-position directional long drilling according to claim 1 is characterized in that: The particle swarm optimization algorithm comprises the following steps: Step 4-1: Initialize the particle swarm, where each particle represents a set of weights and bias parameters of the BP neural network; Step 4-2: Calculate the fitness value of each particle, that is, the error between the predicted result and the actual data; Step 4-3: Update the speed and position of each particle according to the fitness value; Step 4-4: Repeat steps 4-2 and 4-3 until the termination condition is met.

4. The method for predicting three-dimensional spatial parameters of high-position directional long drilling according to claim 1 is characterized in that: The three-dimensional spatial parameters of the high-position directional long drilling hole include the vertical distance and the horizontal distance from the coal seam roof.

5. A high-position directional long drilling three-dimensional spatial parameter prediction system, characterized by: The system includes: Data acquisition module, used to collect data on multiple factors that affect the three-dimensional spatial parameters of high-position directional long drilling; A data processing module is used to preprocess and extract features from the collected data; Model training module, used to train BA-BP neural network model; Model prediction module, used to predict the three-dimensional spatial parameters of high-position directional long drilling; The result display module is used to display the prediction results.

Images