Drilling rock stratum exploration method and system based on computer vision

Abstract

The invention discloses a drilling rock stratum exploration method and system based on computer vision, and the method comprises the steps: building algorithm input dimensions according to the porosity, permeability, lithology and stratum sequence drilling geological data of a drilling rock stratum under different stratum pressures; establishing a drilling rock stratum porosity, permeability, lithology and stratum sequence identification algorithm according to the algorithm input dimension, performing exploration risk assessment according to the porosity, permeability, lithology and stratum sequence abnormal data of the drilling rock stratum in the rock core samples with different stratum temperatures, and generating an exploration risk assessment coefficient; and finally, mining risk prevention measures are set according to the exploration risk assessment coefficient, and human resources and material resources are scheduled based on the mining risk prevention measures. According to the method, automatic classification and anomaly detection can be carried out on the drilling geological data, potential exploration risks are found in time, the risk assessment process is optimized, and human intervention is reduced.

Description

Technical Field

[0001] The present invention relates to the field of drilling rock formation exploration, and in particular to a drilling rock formation exploration method and system based on computer vision. Background Art

[0002] With the continuous growth of global energy demand, drilling operations have become an important part of oil and gas extraction. However, geological exploration work during the drilling process often faces complex challenges, including rock formation identification, porosity and permeability measurement, and potential mining risk assessment. Traditional drilling exploration methods usually rely on manual operation and limited sensor data, and have problems such as low data acquisition accuracy and poor real-time performance, and it is difficult to fully cover the dynamic changes of different strata. In addition, when processing deep drilling geological data, existing technologies often fail to detect abnormal situations in a timely manner, resulting in inaccurate prediction of mining risks, affecting production efficiency and safety. Therefore, a drilling rock formation exploration method based on computer vision came into being. It uses advanced computer vision technology, combined with a multimodal sensing platform and algorithm optimization, to improve data acquisition and risk assessment capabilities during drilling, and further optimize resource scheduling and mining safety.

[0003] Current drilling rock exploration technology mainly relies on traditional physical detection tools and data analysis methods, such as seismic exploration and drilling fluid analysis. Although it can provide geological information to a certain extent, it still has some obvious shortcomings. First, traditional technologies are usually slow in acquiring geological data and rely on a lot of manual intervention, resulting in poor real-time performance and difficulty in meeting the needs of modern drilling operations for rapid response. Secondly, most existing exploration methods are limited to single-dimensional data acquisition, lack of comprehensive analysis of complex formation characteristics, and cannot fully reflect the diversity of geological changes. For example, traditional core analysis can only be done through manual inspection, which often ignores dynamic changes during the drilling process and easily misses potential risk points. In addition, existing risk assessment systems mostly rely on historical data for prediction, which makes it difficult to cope with the complex effects of dynamic factors such as formation pressure on rock stability, resulting in inaccurate assessment results. Summary of the invention

[0004] The present invention overcomes the deficiencies of the prior art and provides a drilling rock formation exploration method and system based on computer vision.

[0005] A first aspect of the present invention provides a drilling rock formation exploration method based on computer vision, comprising the following steps: Build a multimodal computer vision sensing platform, optimize the multimodal computer vision sensing platform, obtain the optimized multimodal computer vision sensing platform, and use the multimodal computer vision sensing platform to control the RockSense+Seequent Leapfrog system to obtain drilling geological data at different locations and depths of drilling rock formations; The porosity, permeability, lithology, and stratigraphic sequence drilling geological data of the drilling rock formation under different formation pressures are obtained through the rock formation edge computing gateway, and the algorithm input dimension is built according to the porosity, permeability, lithology, and stratigraphic sequence drilling geological data of the drilling rock formation under different formation pressures, and the drilling rock formation porosity, permeability, lithology, and stratigraphic sequence identification algorithm is built according to the algorithm input dimension; Through the drilling rock porosity, permeability, lithology, and stratigraphic sequence identification algorithm, the drilling geological data of different regional locations and different depths of the drilling rock formation are identified, and the abnormal data of porosity, permeability, lithology, and stratigraphic sequence of the drilling rock formation in the core samples with different formation temperatures are obtained. Based on the abnormal data of porosity, permeability, lithology, and stratigraphic sequence of the drilling rock formation in the core samples with different formation temperatures, the exploration risk assessment is performed to generate the exploration risk assessment coefficient; Set mining risk prevention measures according to the exploration risk assessment coefficient, and dispatch human and material resources based on the mining risk prevention measures.

[0006] Furthermore, in this method, a multimodal computer vision sensing platform is constructed, and the optimized multimodal computer vision sensing platform is obtained by optimizing the multimodal computer vision sensing platform, which specifically includes: Build a multimodal computer vision sensing platform, and test the multimodal computer vision sensing platform to obtain the fluctuation range of the rock formation hardness threshold monitored by the RockSense+Seequent Leapfrog system at different formation pressures, and build a prediction model for the fluctuation of the rock formation hardness threshold monitored by the RockSense+Seequent Leapfrog system through random forest; Inputting the fluctuation range of the threshold value of rock formation hardness monitored by the RockSense+Seequent Leapfrog system for different formation pressures of the multimodal computer vision sensing platform into the prediction model of the threshold value fluctuation of rock formation hardness monitored by the RockSense+Seequent Leapfrog system for training, and obtaining the trained prediction model of the threshold value fluctuation of rock formation hardness monitored by the RockSense+Seequent Leapfrog system; Obtain the fluctuation range of the threshold value of rock hardness monitored by the RockSense+SeequentLeapfrog system of the multimodal computer vision sensing platform within a preset time and input it into the trained prediction model of the threshold value fluctuation of rock hardness monitored by the RockSense+SeequentLeapfrog system to make predictions, and obtain the fluctuation of the threshold value of rock hardness monitored by the RockSense+SeequentLeapfrog system of the multimodal computer vision sensing platform of the unit depth crack density and morphology; The real-time RockSense+Seequent Leapfrog system monitoring fluctuation range of the multimodal computer vision sensing platform is obtained. When the real-time RockSense+Seequent Leapfrog system monitoring fluctuation range is greater than the threshold fluctuation of the rock formation hardness monitoring by the RockSense+Seequent Leapfrog system of the multimodal computer vision sensing platform of unit depth crack density and morphology, the real-time RockSense+Seequent Leapfrog system monitoring fluctuation range is adjusted according to the threshold fluctuation of the rock formation hardness monitoring by the RockSense+Seequent Leapfrog system of the multimodal computer vision sensing platform of unit depth crack density and morphology to obtain the optimized multimodal computer vision sensing platform.

[0007] Furthermore, in this method, the porosity, permeability, lithology, and stratigraphic sequence drilling geological data of the drilling rock formation under different formation pressures are obtained through the rock formation edge computing gateway, and the algorithm input dimension is built according to the porosity, permeability, lithology, and stratigraphic sequence drilling geological data of the drilling rock formation under different formation pressures, and the drilling rock formation porosity, permeability, lithology, and stratigraphic sequence identification algorithm is built according to the algorithm input dimension, specifically including: The porosity, permeability, lithology, and stratigraphic sequence drilling geological data of the drilling rock formation under different formation pressures are obtained through the rock formation edge computing gateway, and the porosity, permeability, lithology, and stratigraphic sequence drilling geological data of the drilling rock formation under different formation pressures are classified into porosity, permeability, lithology, and stratigraphic sequence types to obtain the drilling geological data of each porosity, permeability, lithology, and stratigraphic sequence type. According to the drilling geological data of each porosity, permeability, lithology, and stratigraphic sequence type, the algorithm input dimensions of each porosity, permeability, lithology, and stratigraphic sequence type are established; The drilling geological data in the algorithm input dimensions of each porosity, permeability, lithology, and stratigraphic sequence type are input into the abnormal One-Class SVM model to obtain the abnormal porosity, permeability, lithology, and stratigraphic sequence data in different groups of drilling geological data, and determine whether the abnormal porosity, permeability, lithology, and stratigraphic sequence data in the drilling geological data exceed the preset interval, and introduce Autoencoder; When the porosity, permeability, lithology, and stratigraphic sequence abnormal data existing in the drilling geological data exceeds a preset interval, the porosity, permeability, lithology, and stratigraphic sequence abnormal data corresponding to the porosity, permeability, lithology, and stratigraphic sequence type label information of the unit depth drilling geological data are obtained as the relevant porosity, permeability, lithology, and stratigraphic sequence abnormal data; when the porosity, permeability, lithology, and stratigraphic sequence abnormal data existing in the drilling geological data exceeds a preset value, the porosity, permeability, lithology, and stratigraphic sequence abnormal data corresponding to the porosity, permeability, lithology, and stratigraphic sequence type label information of the unit depth drilling geological data are used as the relevant porosity, permeability, lithology, and stratigraphic sequence abnormal data; The relevant porosity, permeability, lithology, and stratigraphic sequence abnormal data are input into Autoencoder, so that the encoding layer is concentrated on the relevant porosity, permeability, lithology, and stratigraphic sequence abnormal data, and the encoding layer neuron dimension is obtained. Based on Threshold-based Methods, a drilling rock formation porosity, permeability, lithology, and stratigraphic sequence identification algorithm is built, and the encoding layer neuron dimension in the algorithm input dimension is input into the drilling rock formation porosity, permeability, lithology, and stratigraphic sequence identification algorithm for safety peak judgment.

[0008] Furthermore, in this method, the drilling geological data of different regional locations and different depths of the drilling rock formation are identified by using the drilling rock formation porosity, permeability, lithology, and stratigraphic sequence identification algorithm to obtain the abnormal data of the porosity, permeability, lithology, and stratigraphic sequence of the drilling rock formation in the core samples with different formation temperatures, and the exploration risk assessment is performed based on the abnormal data of the porosity, permeability, lithology, and stratigraphic sequence of the drilling rock formation in the core samples with different formation temperatures, and the exploration risk assessment coefficient is generated, which specifically includes: Input the drilling geological data of different regional locations and different depths of the drilling rock formation into the drilling rock formation porosity, permeability, lithology, and stratigraphic sequence identification algorithm for identification, obtain the porosity, permeability, lithology, and stratigraphic sequence abnormal data of the drilling rock formation in the core samples of different formation temperatures, and preset different mining red lines for oil and gas reserves exploration; According to different mining red lines for oil and gas exploration, the porosity, permeability, lithology, and stratigraphic sequence abnormal data of the drilling rock formations in the core samples with different formation temperatures are divided into different levels of oil and gas exploration reserves, and the porosity, permeability, lithology, and stratigraphic sequence abnormal data of the drilling rock formations in the core samples with different formation temperatures are obtained to explore different levels of oil and gas reserves, and the mineral distribution data at the preset depth where the sedimentation phenomenon occurs are obtained; An exploration risk assessment is generated based on the data of mineral distribution at preset depths of different magnitudes of oil and gas reserves and sedimentation phenomena in the core samples of different formation temperatures, and the data of porosity, permeability, lithology, and stratigraphic sequence abnormalities of the drilled rock formations in the core samples of different formation temperatures, and an exploration risk assessment coefficient is generated.

[0009] Furthermore, in this method, mining risk prevention measures are set according to the exploration risk assessment coefficient, specifically including: Obtain mineral distribution data at a preset depth where sedimentation occurs according to the exploration risk assessment coefficient, build search tags based on the mineral distribution data at the preset depth where sedimentation occurs, perform searches based on the search tags, and obtain fluid samples from the drilling strata; The position distribution of the oil and gas layer and the water layer is obtained according to the fluid sample of the drilling rock formation, and a material content determination device for the position distribution data of the oil and gas layer and the water layer is built, and the position distribution of the oil and gas layer and the water layer is input into the material content determination device for the position distribution data of the oil and gas layer and the water layer to perform material content determination, and obtain different carbon concentration content determination results from high to low concentrations; Based on the measurement results of different carbon concentrations from high to low, mining risk prevention measures for preset depths are set, and the mining risk prevention measures are documented in a report.

[0010] Furthermore, in this method, human and material resources are dispatched based on mining risk prevention measures, specifically including: Obtain the stress and strain depth information of the crustal movement per unit depth, as well as the porosity, permeability, lithology, and mineral distribution data of the preset depth of the stratigraphic sequence where underground rivers exist; calculate the influence weights of the rock stability of each crustal movement location and the mineral distribution data of the preset depth where sedimentation occurs under different faults based on the stress and strain depth information of the crustal movement per unit depth and the mineral distribution data of the preset depth where sedimentation occurs; A multimodal parameter comprehensive evaluation model is introduced, and the coefficient of the rock stability influence weight is set according to the multimodal parameter comprehensive evaluation model. The rock stability influence weight under different faults is initialized according to the location of each crustal movement and the mineral distribution data at the preset depth of the sedimentation phenomenon, and the parameter category of the crustal movement multimodal parameter comprehensive evaluation model is obtained to obtain the mining cost of the crustal movement per unit area. Determine whether the influence weight of the rock formation stability under different faults of the mineral distribution data at the location where the crust movement is located and the preset depth where the sedimentation occurs is greater than the preset influence weight of the rock formation stability. When the influence weight of the rock formation stability under different faults of the mineral distribution data at the location where the crust movement is located and the preset depth where the sedimentation occurs is less than the preset influence weight of the rock formation stability, output the mining cost of the crust movement per unit area; When the weight of the impact of the mineral distribution data at the location where the crustal movement is located and the preset depth of the sedimentation phenomenon on the rock stability under different faults is greater than the preset rock stability impact weight, the mining cost per unit area of ​​the crustal movement will be re-evaluated until the weight of the impact of the mineral distribution data at the location where the crustal movement is located and the preset depth of the sedimentation phenomenon on the rock stability under different faults is less than the preset rock stability impact weight, and human and material resources are dispatched according to the mining risk prevention measures.

[0011] A second aspect of the present invention provides a drilling rock formation exploration system based on computer vision, the system comprising: Multimodal computer vision sensor data integration module, which builds a multimodal computer vision sensor platform and optimizes the multimodal computer vision sensor platform to obtain the optimized multimodal computer vision sensor platform. The multimodal computer vision sensor platform is used to control the RockSense+Seequent Leapfrog system to obtain drilling geological data at different locations and depths of drilling rock formations. The rock formation edge computing gateway data acquisition module, which obtains the porosity, permeability, lithology, and stratigraphic sequence drilling geological data of the drilling rock formation under different formation pressures through the rock formation edge computing gateway, and builds the algorithm input dimension according to the porosity, permeability, lithology, and stratigraphic sequence drilling geological data of the drilling rock formation under different formation pressures, and builds the drilling rock formation porosity, permeability, lithology, and stratigraphic sequence identification algorithm according to the algorithm input dimension; An exploration risk assessment coefficient generation module, which uses a drilling rock porosity, permeability, lithology, and stratigraphic sequence identification algorithm to identify drilling geological data at different locations and depths of the drilling rock formation, obtains abnormal data on the porosity, permeability, lithology, and stratigraphic sequence of the drilling rock formation in core samples at different formation temperatures, and performs exploration risk assessment based on the abnormal data on the porosity, permeability, lithology, and stratigraphic sequence of the drilling rock formation in core samples at different formation temperatures, and generates an exploration risk assessment coefficient; Different resource scheduling modules, which set mining risk prevention measures according to the exploration risk assessment coefficient, and schedule human and material resources based on the mining risk prevention measures.

[0012] Beneficial effects: The present invention builds a multimodal computer vision sensing platform, optimizes the multimodal computer vision sensing platform, obtains the optimized multimodal computer vision sensing platform, controls the RockSense+Seequent Leapfrog system through the multimodal computer vision sensing platform to obtain drilling geological data of different areas and depths of drilling rock formations, and then obtains the porosity, permeability, lithology, and formation sequence drilling geological data of the drilling rock formation under different formation pressures through the rock formation edge computing gateway, and builds an algorithm input dimension according to the porosity, permeability, lithology, and formation sequence drilling geological data of the drilling rock formation under different formation pressures, and builds a drilling rock formation porosity, permeability, lithology, and formation sequence identification algorithm according to the algorithm input dimension, so as to obtain the porosity, permeability, lithology, and formation sequence of the drilling rock formation through the drilling rock formation edge computing gateway. The lithology and stratigraphic sequence identification algorithm identifies the drilling geological data of different regional locations and different depths of the drilling rock formation, obtains the porosity, permeability, lithology, and stratigraphic sequence abnormal data of the drilling rock formation in the core samples of different formation temperatures, and performs exploration risk assessment based on the porosity, permeability, lithology, and stratigraphic sequence abnormal data of the drilling rock formation in the core samples of different formation temperatures, generates an exploration risk assessment coefficient, and finally sets the mining risk prevention measures according to the exploration risk assessment coefficient, and performs human resources and material resources scheduling based on the mining risk prevention measures. The present invention can greatly improve the accuracy and efficiency of data acquisition during drilling by integrating advanced visual sensing technology and intelligent algorithms. By building a multimodal computer vision sensing platform and combining the RockSense and Seequent Leapfrog systems, the rock formation characteristics can be monitored in real time at different formation locations and depths, and key geological data such as porosity, permeability, and lithology can be automatically identified. This technology can acquire and process large-scale image and sensor data in real time, greatly improving the real-time and comprehensiveness of data acquisition. In addition, algorithm models (such as One-Class SVM and Autoencoder) can automatically classify and detect anomalies in drilling geological data, timely discover potential exploration risks, optimize the risk assessment process, and reduce human intervention. In general, the present invention provides a more efficient and accurate technical means for drilling operations, solving many deficiencies of existing technologies in real-time data processing, risk assessment, and resource scheduling. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 A flow chart of the method of the present invention is shown; Figure 2 The system module composition diagram of the present invention is shown. DETAILED DESCRIPTION

[0014] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application is further described in detail below in conjunction with the drawings and specific embodiments.

[0015] like Figure 1 As shown, the first aspect of the present invention provides a drilling rock formation exploration method based on computer vision, comprising the following steps: Step A1, building a multimodal computer vision sensing platform, optimizing the multimodal computer vision sensing platform, obtaining an optimized multimodal computer vision sensing platform, and controlling the RockSense+Seequent Leapfrog system through the multimodal computer vision sensing platform to obtain drilling geological data of different areas and different depths of drilling rock formations; Step A2: Obtain the drilling geological data of porosity, permeability, lithology, and stratigraphic sequence of the drilling rock formation under different formation pressures through the rock formation edge computing gateway, and build an algorithm input dimension according to the drilling geological data of porosity, permeability, lithology, and stratigraphic sequence of the drilling rock formation under different formation pressures, and build a drilling rock formation porosity, permeability, lithology, and stratigraphic sequence identification algorithm according to the algorithm input dimension; Step A3, identifying the drilling geological data of different areas and depths of the drilling rock formations by using the drilling rock formation porosity, permeability, lithology, and stratigraphic sequence identification algorithm, obtaining the abnormal data of the porosity, permeability, lithology, and stratigraphic sequence of the drilling rock formations in the core samples with different formation temperatures, and performing exploration risk assessment based on the abnormal data of the porosity, permeability, lithology, and stratigraphic sequence of the drilling rock formations in the core samples with different formation temperatures, and generating an exploration risk assessment coefficient; Step A4: Set mining risk prevention measures according to the exploration risk assessment coefficient, and dispatch human and material resources based on the mining risk prevention measures.

[0016] It should be noted that the present invention processes the target porosity, permeability, lithology, and stratigraphic sequence abnormal data in the algorithm input dimension through Autoencoder, so that the encoding layer is concentrated on the target porosity, permeability, lithology, and stratigraphic sequence abnormal data in the algorithm input dimension, which can suppress the interference of multi-scale characteristics on the drilling rock formation porosity, permeability, lithology, and stratigraphic sequence identification algorithm, and can improve the identification accuracy of drilling rock formation porosity, permeability, lithology, and stratigraphic sequence identification.

[0017] Furthermore, a multimodal computer vision sensing platform is constructed, and the optimized multimodal computer vision sensing platform is obtained by optimizing the multimodal computer vision sensing platform, specifically including: Build a multimodal computer vision sensing platform, and test the multimodal computer vision sensing platform to obtain the fluctuation range of the rock formation hardness threshold monitored by the RockSense+Seequent Leapfrog system at different formation pressures, and build a prediction model for the fluctuation of the rock formation hardness threshold monitored by the RockSense+Seequent Leapfrog system through random forest; Inputting the fluctuation range of the threshold value of rock formation hardness monitored by the RockSense+Seequent Leapfrog system for different formation pressures of the multimodal computer vision sensing platform into the prediction model of the threshold value fluctuation of rock formation hardness monitored by the RockSense+Seequent Leapfrog system for training, and obtaining the trained prediction model of the threshold value fluctuation of rock formation hardness monitored by the RockSense+Seequent Leapfrog system; Obtain the fluctuation range of the threshold value of rock hardness monitored by the RockSense+SeequentLeapfrog system of the multimodal computer vision sensing platform within a preset time and input it into the trained prediction model of the threshold value fluctuation of rock hardness monitored by the RockSense+SeequentLeapfrog system to make predictions, and obtain the fluctuation of the threshold value of rock hardness monitored by the RockSense+SeequentLeapfrog system of the multimodal computer vision sensing platform of the unit depth crack density and morphology; The real-time RockSense+Seequent Leapfrog system monitoring fluctuation range of the multimodal computer vision sensing platform is obtained. When the real-time RockSense+Seequent Leapfrog system monitoring fluctuation range is greater than the threshold fluctuation of the rock formation hardness monitoring by the RockSense+Seequent Leapfrog system of the multimodal computer vision sensing platform of unit depth crack density and morphology, the real-time RockSense+Seequent Leapfrog system monitoring fluctuation range is adjusted according to the threshold fluctuation of the rock formation hardness monitoring by the RockSense+Seequent Leapfrog system of the multimodal computer vision sensing platform of unit depth crack density and morphology to obtain the optimized multimodal computer vision sensing platform.

[0018] It should be noted that RockSense is a technology launched by Schlumberger, which is mainly used to measure and analyze the mechanical properties of rocks and other related data in real time during the drilling process. By monitoring and analyzing various parameters (such as torque, speed, drilling pressure, etc.) during the drilling process of the drill bit, combined with relevant algorithms and models, the characteristics of the rock encountered, such as the hardness and strength of the rock, can be inferred. Seequent Leapfrog is a powerful geological modeling and analysis software developed by Seequent. It can integrate data from multiple sources, including geological, geophysical, geochemical and other data, and build a three-dimensional geological model through advanced algorithms and visualization technology. It has a wide range of applications in geological exploration, resource assessment, underground engineering design and other fields. Users can use this software to analyze and display geological structures in detail, predict the distribution of geological bodies and changes in properties, etc.

[0019] When RockSense is combined with Seequent Leapfrog to form the RockSense+Seequent Leapfrog system, it means combining the rock property data (from RockSense) acquired in real time during the drilling process with the powerful geological modeling and analysis capabilities of SeequentLeapfrog. In this method, this system can more accurately monitor the fluctuation range of the rock hardness threshold under different formation pressures. Specifically, RockSense provides real-time rock mechanical property data, while Seequent Leapfrog integrates, models and analyzes this data and other related geological data to help researchers and engineers better understand the characteristics and changing laws of underground rock formations, thereby providing a more reliable basis for subsequent predictions and decisions. For example, in the fields of oil and gas extraction, underground mineral resource development, etc., this combined system can more accurately evaluate the formation conditions, optimize the mining plan, and improve the efficiency and safety of resource extraction. Through this method, the real-time RockSense+Seequent Leapfrog system monitoring fluctuation range can be adjusted according to the fluctuation of the threshold value of rock hardness monitored by the RockSense+Seequent Leapfrog system of the multimodal computer vision sensing platform of unit depth crack density and morphology, so that the monitoring fluctuation range of the RockSense+SeequentLeapfrog system controlled by the human-machine integrated control network meets the predetermined requirements and ensures the stability of the multimodal computer vision sensing platform.

[0020] Furthermore, the drilling geological data of porosity, permeability, lithology, and stratigraphic sequence of the drilling rock formation under different formation pressures are obtained through the rock formation edge computing gateway, and the algorithm input dimension is built according to the drilling geological data of porosity, permeability, lithology, and stratigraphic sequence of the drilling rock formation under different formation pressures. According to the algorithm input dimension, the drilling rock formation porosity, permeability, lithology, and stratigraphic sequence identification algorithm is built, which specifically includes: The porosity, permeability, lithology, and stratigraphic sequence drilling geological data of the drilling rock formation under different formation pressures are obtained through the rock formation edge computing gateway, and the porosity, permeability, lithology, and stratigraphic sequence drilling geological data of the drilling rock formation under different formation pressures are classified into porosity, permeability, lithology, and stratigraphic sequence types to obtain the drilling geological data of each porosity, permeability, lithology, and stratigraphic sequence type. According to the drilling geological data of each porosity, permeability, lithology, and stratigraphic sequence type, the algorithm input dimensions of each porosity, permeability, lithology, and stratigraphic sequence type are established; The drilling geological data in the algorithm input dimensions of each porosity, permeability, lithology, and stratigraphic sequence type are input into the abnormal One-Class SVM model to obtain the abnormal porosity, permeability, lithology, and stratigraphic sequence data in different groups of drilling geological data, and determine whether the abnormal porosity, permeability, lithology, and stratigraphic sequence data in the drilling geological data exceed the preset interval, and introduce Autoencoder; When the porosity, permeability, lithology, and stratigraphic sequence abnormal data existing in the drilling geological data exceeds a preset interval, the porosity, permeability, lithology, and stratigraphic sequence abnormal data corresponding to the porosity, permeability, lithology, and stratigraphic sequence type label information of the unit depth drilling geological data are obtained as the relevant porosity, permeability, lithology, and stratigraphic sequence abnormal data; when the porosity, permeability, lithology, and stratigraphic sequence abnormal data existing in the drilling geological data exceeds a preset value, the porosity, permeability, lithology, and stratigraphic sequence abnormal data corresponding to the porosity, permeability, lithology, and stratigraphic sequence type label information of the unit depth drilling geological data are used as the relevant porosity, permeability, lithology, and stratigraphic sequence abnormal data; The relevant porosity, permeability, lithology, and stratigraphic sequence abnormal data are input into Autoencoder, so that the encoding layer is concentrated on the relevant porosity, permeability, lithology, and stratigraphic sequence abnormal data, and the encoding layer neuron dimension is obtained. Based on Threshold-based Methods, a drilling rock formation porosity, permeability, lithology, and stratigraphic sequence identification algorithm is built, and the encoding layer neuron dimension in the algorithm input dimension is input into the drilling rock formation porosity, permeability, lithology, and stratigraphic sequence identification algorithm for safety peak judgment.

[0021] It should be noted that the porosity, permeability, lithology, and stratigraphic sequence drilling geological data include crack porosity, permeability, lithology, and stratigraphic sequence drilling geological data, corrosion porosity, permeability, lithology, and stratigraphic sequence drilling geological data, and burned porosity, permeability, lithology, and stratigraphic sequence drilling geological data. Since there may be at least two porosity, permeability, lithology, and stratigraphic sequence types in the algorithm input dimension of each porosity, permeability, lithology, and stratigraphic sequence type, such as crack porosity, permeability, lithology, and stratigraphic sequence, corrosion porosity, permeability, lithology, and stratigraphic sequence, during training, multiple porosity, permeability, lithology, and stratigraphic sequence types of algorithm input dimensions should be input into the Threshold-based Methods is used for training. When there is at least two types of label information of porosity, permeability, lithology and stratigraphic sequence in a drilling geological data during training, it will be disturbed. By fusing Autoencoder with this method, the encoding layer can be concentrated on the relevant abnormal data of porosity, permeability, lithology and stratigraphic sequence, and the multi-scale porosity, permeability, lithology and stratigraphic sequence abnormal data can be suppressed from interfering with the drilling rock formation porosity, permeability, lithology and stratigraphic sequence identification algorithm, thereby improving the prediction accuracy of the model.

[0022] Furthermore, in this method, the drilling geological data of different regional locations and different depths of the drilling rock formation are identified by using the drilling rock formation porosity, permeability, lithology, and stratigraphic sequence identification algorithm to obtain the abnormal data of the porosity, permeability, lithology, and stratigraphic sequence of the drilling rock formation in the core samples with different formation temperatures, and the exploration risk assessment is performed based on the abnormal data of the porosity, permeability, lithology, and stratigraphic sequence of the drilling rock formation in the core samples with different formation temperatures, and the exploration risk assessment coefficient is generated, which specifically includes: Input the drilling geological data of different regional locations and different depths of the drilling rock formation into the drilling rock formation porosity, permeability, lithology, and stratigraphic sequence identification algorithm for identification, obtain the porosity, permeability, lithology, and stratigraphic sequence abnormal data of the drilling rock formation in the core samples of different formation temperatures, and preset different mining red lines for oil and gas reserves exploration; According to different mining red lines for oil and gas exploration, the porosity, permeability, lithology, and stratigraphic sequence abnormal data of the drilling rock formations in the core samples with different formation temperatures are divided into different levels of oil and gas exploration reserves, and the porosity, permeability, lithology, and stratigraphic sequence abnormal data of the drilling rock formations in the core samples with different formation temperatures are obtained to explore different levels of oil and gas reserves, and the mineral distribution data at the preset depth where the sedimentation phenomenon occurs are obtained; An exploration risk assessment is generated based on the data of mineral distribution at preset depths of different magnitudes of oil and gas reserves and sedimentation phenomena in the core samples of different formation temperatures, and the data of porosity, permeability, lithology, and stratigraphic sequence abnormalities of the drilled rock formations in the core samples of different formation temperatures, and an exploration risk assessment coefficient is generated.

[0023] It should be noted that the standards for different levels of exploration oil and gas reserves can be set according to porosity, permeability, lithology, type of stratigraphic sequence, porosity, permeability, lithology, and size of stratigraphic sequence. Among them, different levels of exploration oil and gas reserves include different levels of low exploration oil and gas reserves, different levels of medium exploration oil and gas reserves, different levels of high exploration oil and gas reserves, etc.

[0024] Furthermore, mining risk prevention measures are set according to the exploration risk assessment coefficient, including: Obtain mineral distribution data at a preset depth where sedimentation occurs according to the exploration risk assessment coefficient, build search tags based on the mineral distribution data at the preset depth where sedimentation occurs, perform searches based on the search tags, and obtain fluid samples from the drilling strata; The position distribution of the oil and gas layer and the water layer is obtained according to the fluid sample of the drilling rock formation, and a material content determination device for the position distribution data of the oil and gas layer and the water layer is built, and the position distribution of the oil and gas layer and the water layer is input into the material content determination device for the position distribution data of the oil and gas layer and the water layer to perform material content determination, and obtain different carbon concentration content determination results from high to low concentrations; Based on the measurement results of different carbon concentrations from high to low, mining risk prevention measures for preset depths are set, and the mining risk prevention measures are documented in a report.

[0025] It should be noted that, since the crustal movement may be limited, this method can be used to formulate more reasonable mining risk prevention measures.

[0026] Furthermore, human and material resources are dispatched based on the mining risk prevention measures, including: Obtain the stress and strain depth information of the crustal movement per unit depth, as well as the porosity, permeability, lithology, and mineral distribution data of the preset depth of the stratigraphic sequence where underground rivers exist; calculate the influence weights of the rock stability of each crustal movement location and the mineral distribution data of the preset depth where sedimentation occurs under different faults based on the stress and strain depth information of the crustal movement per unit depth and the mineral distribution data of the preset depth where sedimentation occurs; A multimodal parameter comprehensive evaluation model is introduced, and the coefficient of the rock stability influence weight is set according to the multimodal parameter comprehensive evaluation model. The rock stability influence weight under different faults is initialized according to the location of each crustal movement and the mineral distribution data at the preset depth of the sedimentation phenomenon, and the parameter category of the crustal movement multimodal parameter comprehensive evaluation model is obtained to obtain the mining cost of the crustal movement per unit area. Determine whether the influence weight of the rock formation stability under different faults of the mineral distribution data at the location where the crust movement is located and the preset depth where the sedimentation occurs is greater than the preset influence weight of the rock formation stability. When the influence weight of the rock formation stability under different faults of the mineral distribution data at the location where the crust movement is located and the preset depth where the sedimentation occurs is less than the preset influence weight of the rock formation stability, output the mining cost of the crust movement per unit area; When the weight of the impact of the mineral distribution data at the location where the crustal movement is located and the preset depth of the sedimentation phenomenon on the rock stability under different faults is greater than the preset rock stability impact weight, the mining cost per unit area of ​​the crustal movement will be re-evaluated until the weight of the impact of the mineral distribution data at the location where the crustal movement is located and the preset depth of the sedimentation phenomenon on the rock stability under different faults is less than the preset rock stability impact weight, and human and material resources are dispatched according to the mining risk prevention measures.

[0027] It should be noted that the introduction of a multi-modal parameter comprehensive evaluation model refers to the use of an evaluation model that can comprehensively consider multiple different types of parameters. These parameters can cover different modal information, such as geological data, mechanical data, chemical data, etc., in order to comprehensively and accurately evaluate various characteristics related to the rock formation. The implementation process is as follows: First, the coefficients of the weights affecting rock formation stability are set through the multimodal parameter comprehensive evaluation model. Since different factors have different influences on rock formation stability, the model will set corresponding weight coefficients for each relevant factor based on professional knowledge and experience. For example, factors such as rock hardness, structural integrity, and groundwater conditions may have different influences on rock formation stability. By setting weight coefficients, the extent of these factors can be quantified.

[0028] Next, the mineral distribution data of the locations of crustal movement and the preset depths of sedimentation phenomena are used to initialize the parameter categories of the crustal movement multimodal parameter comprehensive evaluation model with the weights of the rock formation stability under different faults. This step is to initialize the parameter categories of the model by combining specific geological data, such as the location information of crustal movement, the distribution of minerals at a specific depth, etc., with the weights of the influence of different faults on rock formation stability. In other words, according to these actual geological conditions and influencing factors, the initial parameter configuration is set for the model so that it can more accurately reflect the actual geological conditions.

[0029] Finally, the mining cost of crustal movement per unit area is obtained. After completing the setting and initialization of the above model parameters, the model is used to analyze and evaluate the crustal movement within a unit area, taking into account the influence of various factors such as rock stability on the mining process, and then calculating the mining cost of crustal movement per unit area. This cost calculation result can provide an important reference for subsequent resource mining decisions, helping decision makers to rationally plan mining plans, reduce costs, and improve mining efficiency.

[0030] By introducing a multimodal parameter comprehensive evaluation model, we can integrate a variety of geological data, comprehensively evaluate factors such as rock stability, and ultimately derive the mining cost of crustal movement per unit area, providing a scientific basis for geological resource development.

[0031] It should be noted that, through this method, the drilling rock formation to be repaired at a preset depth can be configured, so that each maintenance area can respond quickly, and the scheduling of human and material resources for crustal movement is more reasonable.

[0032] In addition, the RockSense+Seequent Leapfrog system is controlled by the multimodal computer vision sensing platform to obtain drilling geological data of different areas and depths of the drilling rock formation, and the following steps may also be included: System construction and equipment calibration: According to the actual situation of the drilling site, various sensors of the multimodal computer vision sensing platform, such as cameras and lidar, are reasonably installed and arranged to ensure that they can cover the key areas of drilling operations and have a stable connection with the RockSense+Seequent Leapfrog system. The installed sensors are calibrated to ensure the accuracy and reliability of their measurement data. This includes adjusting the camera's focal length, viewing angle, white balance and other parameters, as well as calibrating the distance measurement accuracy of the lidar.

[0033] Set data acquisition parameters: According to the drilling objectives and geological characteristics, set appropriate data acquisition parameters in the RockSense + SeequentLeapfrog system, such as sampling interval, measurement range, data accuracy, etc. Determine the data acquisition strategy for different depths and regional locations. For example, according to the change of drilling depth, adjust the scanning frequency and resolution of the sensor to ensure that sufficiently detailed geological data can be obtained at different depths.

[0034] Real-time data acquisition: During the drilling operation, a multimodal computer vision sensing platform is used to collect relevant data of the drilling rock formation in real time. The camera can capture the image information of the borehole wall to identify the structure, color, texture and other characteristics of the rock formation; the lidar can measure the distance and shape of the borehole wall and obtain the three-dimensional geometric information of the rock formation. The collected multimodal data is transmitted to the RockSense+Seequent Leapfrog system in real time for storage and preliminary processing. The system will integrate and preprocess the data to remove noise and abnormal data and improve data quality.

[0035] Data labeling and feature extraction: Manual or automatic labeling of collected images and other data to mark different rock formation areas, geological structure features, etc. Using computer vision and machine learning algorithms, key geological features such as rock formation boundaries, thickness, dip, and orientation are extracted from the labeled data. These features will serve as an important basis for subsequent data analysis and modeling.

[0036] Depth and location association: Combine the drilling depth record and sensor location information to accurately associate the collected geological data with the specific depth and regional location. By establishing a depth-location coordinate system, the data collected at different time points are mapped to the corresponding depth and location to form a continuous geological data profile.

[0037] Data analysis and processing: In the RockSense+Seequent Leapfrog system, various geological data analysis algorithms and models are used to conduct in-depth analysis of the collected data. For example, the geological structure and evolution history of the drilling area can be inferred through methods such as stratigraphic comparison, lithology identification, and structural analysis. The analysis results are visualized to generate geological profiles, three-dimensional geological models, etc., so that geological engineers and related personnel can intuitively understand the drilling geological conditions.

[0038] Data verification and update: Compare and verify the analyzed geological data with existing geological information, drilling records, etc. to check the accuracy and reliability of the data. During the drilling operation, as new data is continuously collected, the existing geological models and data are updated and corrected in a timely manner to ensure that they can accurately reflect the current drilling geological conditions.

[0039] Data storage and sharing: The collected raw data, processed data, and analysis results are properly stored to establish a geological data database. As needed, the relevant data is shared with other departments or partners to provide support for subsequent geological research, resource assessment, drilling engineering design, etc.

[0040] In addition, the method may also include the following steps: obtaining different regional positions of each sensor parameter set under each lithology, characteristic information of drilling geological data at different depths, and the lithology of the area where the unit depth RockSense+Seequent Leapfrog system is located through a rock formation edge computing gateway, and obtaining different regional positions of each sensor parameter and characteristic information of drilling geological data at different depths of the RockSense+Seequent Leapfrog system according to the different regional positions of each sensor parameter set under each lithology, characteristic information of drilling geological data at different depths, and the lithology of the area where the unit depth RockSense+SeequentLeapfrog system is located; obtaining the adjustable range of the sensor parameter of the camera of the unit depth RockSense+SeequentLeapfrog system, judging whether there is at least one sensor parameter among the sensor parameters that makes the characteristic information of drilling geological data at different regional positions and different depths of the RockSense+Seequent Leapfrog system greater than the preset characteristic threshold of drilling geological data at different regional positions and different depths; if there is at least one sensor parameter among the sensor parameters that makes the RockSense+Seequent Leapfrog system When the characteristic information of drilling geological data at different regional locations and different depths of the Leapfrog system is greater than the preset data thresholds of drilling geological data at different regional locations and different depths, a sensor parameter will be randomly output to make the characteristic information of drilling geological data at different regional locations and different depths of the RockSense+Seequent Leapfrog system greater than the preset data thresholds of drilling geological data at different regional locations and different depths and used as the sensor parameter of the RockSense+Seequent Leapfrog system; if there is no sensor parameter among the sensor parameters that makes the characteristic information of drilling geological data at different regional locations and different depths of the RockSense+Seequent Leapfrog system greater than the preset data thresholds of drilling geological data at different regional locations and different depths, the data collection period and data collection point of the RockSense+Seequent Leapfrog system will be adjusted until at least one sensor parameter among the sensor parameters makes the characteristic information of drilling geological data at different regional locations and different depths of the RockSense+Seequent Leapfrog system greater than the preset data thresholds of drilling geological data at different regional locations and different depths.

[0041] It should be noted that due to the influence of the environment, no matter how the sensing parameters are adjusted, there is no parameter that makes the characteristic information of drilling geological data at different regional locations and different depths of the RockSense+Seequent Leapfrog system greater than the preset characteristic threshold of drilling geological data at different regional locations and different depths, making it impossible to obtain drilling geological data of the predetermined standard. This method can further improve the rationality of data acquisition.

[0042] like Figure 2 As shown, the second aspect of the present invention provides a drilling rock formation exploration system based on computer vision, the system comprising: Multimodal computer vision sensor data integration module, which builds a multimodal computer vision sensor platform and optimizes the multimodal computer vision sensor platform to obtain the optimized multimodal computer vision sensor platform. The multimodal computer vision sensor platform is used to control the RockSense+Seequent Leapfrog system to obtain drilling geological data at different locations and depths of drilling rock formations. The rock formation edge computing gateway data acquisition module, which obtains the porosity, permeability, lithology, and stratigraphic sequence drilling geological data of the drilling rock formation under different formation pressures through the rock formation edge computing gateway, and builds the algorithm input dimension according to the porosity, permeability, lithology, and stratigraphic sequence drilling geological data of the drilling rock formation under different formation pressures, and builds the drilling rock formation porosity, permeability, lithology, and stratigraphic sequence identification algorithm according to the algorithm input dimension; An exploration risk assessment coefficient generation module, which uses a drilling rock porosity, permeability, lithology, and stratigraphic sequence identification algorithm to identify drilling geological data at different locations and depths of the drilling rock formation, obtains abnormal data on the porosity, permeability, lithology, and stratigraphic sequence of the drilling rock formation in core samples at different formation temperatures, and performs exploration risk assessment based on the abnormal data on the porosity, permeability, lithology, and stratigraphic sequence of the drilling rock formation in core samples at different formation temperatures, and generates an exploration risk assessment coefficient; Different resource scheduling modules, which set mining risk prevention measures according to the exploration risk assessment coefficient, and schedule human and material resources based on the mining risk prevention measures.

[0043] Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various equivalent changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A drilling rock formation exploration method based on computer vision, characterized in that: The following steps are involved: Building a multimodal computer vision sensing platform, optimizing the multimodal computer vision sensing platform to obtain an optimized multimodal computer vision sensing platform, and controlling the RockSense+Seequent Leapfrog system to obtain drilling geological data at different locations and depths of drilling rock formations based on the multimodal computer vision sensing platform; The porosity, permeability, lithology, and stratigraphic sequence drilling geological data of the drilling rock formation under different formation pressures are obtained through the rock formation edge computing gateway, and the algorithm input dimension is established based on the porosity, permeability, lithology, and stratigraphic sequence drilling geological data of the drilling rock formation under different formation pressures, and the drilling rock formation porosity, permeability, lithology, and stratigraphic sequence identification algorithm is established based on the algorithm input dimension; According to the drilling rock formation porosity, permeability, lithology, and stratigraphic sequence identification algorithm, the drilling geological data of different regional locations and different depths of the drilling rock formation are identified, and the porosity, permeability, lithology, and stratigraphic sequence abnormal data of the drilling rock formation in the core samples with different formation temperatures are obtained, and the exploration risk assessment is performed based on the porosity, permeability, lithology, and stratigraphic sequence abnormal data of the drilling rock formation in the core samples with different formation temperatures, and an exploration risk assessment coefficient is generated; Exploitation risk prevention measures are set according to the exploration risk assessment coefficient, and human and material resources are dispatched based on the exploration risk prevention measures.

2. The computer vision-based drilling rock exploration method according to claim 1, characterized in that: The porosity, permeability, lithology, and stratigraphic sequence drilling geological data of the drilling rock formation under different formation pressures are obtained through the rock formation edge computing gateway, and the algorithm input dimension is built based on the porosity, permeability, lithology, and stratigraphic sequence drilling geological data of the drilling rock formation under different formation pressures. The porosity, permeability, lithology, and stratigraphic sequence identification algorithm of the drilling rock formation is built based on the algorithm input dimension, specifically including: The porosity, permeability, lithology, and stratigraphic sequence drilling geological data of the drilling rock formation under different formation pressures are obtained through the rock formation edge computing gateway, and the porosity, permeability, lithology, and stratigraphic sequence drilling geological data of the drilling rock formation under different formation pressures are classified into porosity, permeability, lithology, and stratigraphic sequence types to obtain the drilling geological data of each porosity, permeability, lithology, and stratigraphic sequence type, and the algorithm input dimensions of each porosity, permeability, lithology, and stratigraphic sequence type are established based on the drilling geological data of each porosity, permeability, lithology, and stratigraphic sequence type; Input the drilling geological data in the algorithm input dimensions of each porosity, permeability, lithology, and stratigraphic sequence type into the abnormal One-Class SVM model, obtain the abnormal porosity, permeability, lithology, and stratigraphic sequence data in different groups of drilling geological data, and determine whether the abnormal porosity, permeability, lithology, and stratigraphic sequence data in the drilling geological data exceeds a preset interval, and introduce Autoencoder; If the porosity, permeability, lithology, and stratigraphic sequence abnormal data in the drilling geological data exceeds a preset interval, the porosity, permeability, lithology, and stratigraphic sequence abnormal data corresponding to the porosity, permeability, lithology, and stratigraphic sequence type label information of the unit depth drilling geological data are obtained as the relevant porosity, permeability, lithology, and stratigraphic sequence abnormal data; if the porosity, permeability, lithology, and stratigraphic sequence abnormal data in the drilling geological data exceeds a preset value, the porosity, permeability, lithology, and stratigraphic sequence abnormal data corresponding to the porosity, permeability, lithology, and stratigraphic sequence type label information of the unit depth drilling geological data are used as the relevant porosity, permeability, lithology, and stratigraphic sequence abnormal data; The relevant porosity, permeability, lithology, and stratigraphic sequence abnormal data are input into the Autoencoder, so that the encoding layer is concentrated on the relevant porosity, permeability, lithology, and stratigraphic sequence abnormal data, the encoding layer neuron dimension is obtained, and a drilling rock formation porosity, permeability, lithology, and stratigraphic sequence identification algorithm is built based on Threshold-based Methods. The encoding layer neuron dimension is input into the drilling rock formation porosity, permeability, lithology, and stratigraphic sequence identification algorithm for safety peak judgment.

3. The computer vision-based drilling rock exploration method according to claim 1, characterized in that: Building a multimodal computer vision sensing platform, and optimizing the multimodal computer vision sensing platform to obtain an optimized multimodal computer vision sensing platform, specifically including: Build a multimodal computer vision sensing platform, and test the multimodal computer vision sensing platform to obtain the fluctuation range of the threshold value of rock hardness monitored by the RockSense+Seequent Leapfrog system at different formation pressures of the multimodal computer vision sensing platform, and build a prediction model for the fluctuation of the threshold value of rock hardness monitored by the RockSense+Seequent Leapfrog system through random forest; Inputting the fluctuation range of the threshold value of rock formation hardness monitored by the RockSense+Seequent Leapfrog system for different formation pressures of the multimodal computer vision sensing platform into the prediction model for the fluctuation of the threshold value of rock formation hardness monitored by the RockSense+Seequent Leapfrog system for training, and obtaining the trained prediction model for the fluctuation of the threshold value of rock formation hardness monitored by the RockSense+Seequent Leapfrog system; Obtaining the fluctuation range of the threshold value of rock hardness monitored by the RockSense+Seequent Leapfrog system of the multimodal computer vision sensing platform within a preset time and inputting it into the trained prediction model of the threshold value fluctuation of rock hardness monitored by the RockSense+Seequent Leapfrog system for prediction, and obtaining the fluctuation of the threshold value of rock hardness monitored by the RockSense+Seequent Leapfrog system of the multimodal computer vision sensing platform of the unit depth crack density and morphology; A real-time RockSense+Seequent Leapfrog system monitoring fluctuation range of the multimodal computer vision sensing platform is obtained. If the real-time RockSense+Seequent Leapfrog system monitoring fluctuation range is greater than a threshold fluctuation of rock formation hardness monitored by the RockSense+Seequent Leapfrog system of the multimodal computer vision sensing platform of unit depth crack density and morphology, the real-time RockSense+Seequent Leapfrog system monitoring fluctuation range is adjusted according to the threshold fluctuation of rock formation hardness monitored by the RockSense+Seequent Leapfrog system of the multimodal computer vision sensing platform of unit depth crack density and morphology to obtain an optimized multimodal computer vision sensing platform.

4. The computer vision-based drilling rock exploration method according to claim 1, characterized in that: According to the drilling rock formation porosity, permeability, lithology, and stratigraphic sequence identification algorithm, the drilling geological data of different regional locations and different depths of the drilling rock formation are identified, and the porosity, permeability, lithology, and stratigraphic sequence abnormal data of the drilling rock formation in the core samples with different formation temperatures are obtained. Based on the porosity, permeability, lithology, and stratigraphic sequence abnormal data of the drilling rock formation in the core samples with different formation temperatures, exploration risk assessment is performed to generate an exploration risk assessment coefficient, which specifically includes: Input the drilling geological data of different regional locations and different depths of the drilling rock formation into the porosity, permeability, lithology, and stratigraphic sequence identification algorithm for identification, obtain the porosity, permeability, lithology, and stratigraphic sequence abnormal data of the drilling rock formation in the core samples with different formation temperatures, and preset different mining red lines for oil and gas reserves exploration; According to the different red lines for oil and gas exploration, the porosity, permeability, lithology, and stratigraphic sequence abnormal data of the drilling rock formations in the core samples with different formation temperatures are divided into different levels for oil and gas exploration, and the porosity, permeability, lithology, and stratigraphic sequence abnormal data of the drilling rock formations in the core samples with different formation temperatures are obtained to explore different levels of oil and gas reserves, and the mineral distribution data at the preset depth where the sedimentation phenomenon occurs are obtained; An exploration risk assessment is generated based on the porosity, permeability, lithology, and stratigraphic sequence abnormal data of the drilling rock formations in the core samples with different formation temperatures, the mineral distribution data of different levels of oil and gas reserves, the preset depths of sedimentation phenomena, and the porosity, permeability, lithology, and stratigraphic sequence abnormal data of the drilling rock formations in the core samples with different formation temperatures, and an exploration risk assessment coefficient is generated.

5. The method for drilling rock formation exploration based on computer vision according to claim 1, characterized in that: The mining risk prevention measures are set according to the exploration risk assessment coefficient, including: Obtaining mineral distribution data at a preset depth where sedimentation occurs according to the exploration risk assessment coefficient, building a search tag based on the mineral distribution data at the preset depth where sedimentation occurs, and performing a search according to the search tag to obtain a fluid sample of the drilling stratum; Obtaining the position distribution of the oil and gas layer and the water layer according to the fluid sample of the drilling rock formation, and building a material content determination device for the position distribution data of the oil and gas layer and the water layer, inputting the position distribution of the oil and gas layer and the water layer into the material content determination device for the position distribution data of the oil and gas layer and the water layer to perform material content determination, and obtaining different carbon concentration content determination results from high to low concentrations; Based on the determination results of different carbon concentrations from high to low, mining risk prevention measures for a preset depth are set, and the mining risk prevention measures are documented and reported.

6. The method for drilling rock formation exploration based on computer vision according to claim 1, characterized in that: Based on the mining risk prevention measures, human and material resources are dispatched, including: Obtain the stress and strain depth information of the crustal movement per unit depth, as well as the porosity, permeability, lithology, and mineral distribution data of the preset depth of the stratigraphic sequence where the underground river exists; calculate the influence weight of the rock stability of each crustal movement location and the mineral distribution data of the preset depth where the sedimentation phenomenon occurs under different faults based on the stress and strain depth information of the crustal movement per unit depth and the mineral distribution data of the preset depth where the sedimentation phenomenon occurs; A multimodal parameter comprehensive evaluation model is introduced, and the coefficient of the rock stability influence weight is set according to the multimodal parameter comprehensive evaluation model. The rock stability influence weight under different faults is initialized according to the mineral distribution data of the location of each crustal movement and the preset depth of the sedimentation phenomenon, and the parameter category of the crustal movement multimodal parameter comprehensive evaluation model is obtained to obtain the mining cost of the crustal movement per unit area; Determine whether the influence weight of the rock formation stability under different faults of the mineral distribution data at the location where the crust movement is located and the preset depth where the sedimentation phenomenon occurs is greater than the preset influence weight of the rock formation stability; if the influence weight of the rock formation stability under different faults of the mineral distribution data at the location where the crust movement is located and the preset depth where the sedimentation phenomenon occurs is less than the preset influence weight of the rock formation stability, output the mining cost of the crust movement per unit area; If the weight of the impact of the mineral distribution data at the location where the crust movement is located and the preset depth of the sedimentation phenomenon on the rock stability under different faults is greater than the preset rock stability impact weight, the mining cost per unit area of ​​the crust movement will be re-evaluated until the weight of the impact of the mineral distribution data at the location where the crust movement is located and the preset depth of the sedimentation phenomenon on the rock stability under different faults is less than the preset rock stability impact weight, and human and material resources are dispatched according to the mining risk prevention measures.

7. A drilling rock formation exploration system based on computer vision, implemented according to the drilling rock formation exploration method based on computer vision according to any one of claims 1 to 6, characterized in that: include: Multimodal computer vision sensor data integration module, which builds a multimodal computer vision sensor platform and optimizes the multimodal computer vision sensor platform to obtain the optimized multimodal computer vision sensor platform. The multimodal computer vision sensor platform is used to control the RockSense+Seequent Leapfrog system to obtain drilling geological data at different locations and depths of drilling rock formations. The rock formation edge computing gateway data acquisition module, which obtains the porosity, permeability, lithology, and stratigraphic sequence drilling geological data of the drilling rock formation under different formation pressures through the rock formation edge computing gateway, and builds the algorithm input dimension according to the porosity, permeability, lithology, and stratigraphic sequence drilling geological data of the drilling rock formation under different formation pressures, and builds the drilling rock formation porosity, permeability, lithology, and stratigraphic sequence identification algorithm according to the algorithm input dimension; An exploration risk assessment coefficient generation module, which uses a drilling rock porosity, permeability, lithology, and stratigraphic sequence identification algorithm to identify drilling geological data at different locations and depths of the drilling rock formation, obtains abnormal data on the porosity, permeability, lithology, and stratigraphic sequence of the drilling rock formation in core samples at different formation temperatures, and performs exploration risk assessment based on the abnormal data on the porosity, permeability, lithology, and stratigraphic sequence of the drilling rock formation in core samples at different formation temperatures, and generates an exploration risk assessment coefficient; Different resource scheduling modules, which set mining risk prevention measures according to the exploration risk assessment coefficient, and schedule human and material resources based on the mining risk prevention measures.

Images