System for determining rock properties from 3-dimensional data cloud with machine learning

Abstract

The invention relates to a system for determining rock properties from a 3-dimensional data cloud with machine learning.

Description

[0001] DESCRIPTION SYSTEM FOR DETERMINING ROCK PROPERTIES FROM 3-DIMENSIONAL DATA CLOUD WITH MACHINE LEARNING

[0002] Technical Field of the Invention

[0003] The invention relates to a system for determining rock properties from a 3-dimensional data cloud with machine learning.

[0004] State of the Art

[0005] Today, the physical properties of rocks, such as density and porosity, are widely used in mining operations, the construction industry, and geological research. These parameters have become routine in rock mechanics experiments and are performed to determine the engineering properties of almost all rock materials. Determination of the actual porosity of the rock is laborious, expensive and requires skilled personnel. Carrying out these experiments involves a series of sample preparation procedures This, in turn leads to demanding and time-consuming tasks such as cutting and correcting cores or preparing cube samples even before the experiment begins.

[0006] A review of the literature indicates that the apparent porosity parameter is more practical and commonly used Apparent density and porosity parameters are determined by caliper method for shaped samples and Archimedes method for unshaped samples according to the standards. Experiments on shaped samples are laborious and time-consuming as they require the preparation of core or cube samples. Furthermore, it is almost impossible to prepare high quality core samples from weathered and weak rock samples. For this reason, the Archimedes method is more preferred as it does not require sample preparation for researchers. However, the Archimedes method has disadvantages such as underreading the sample weight on the scale and loss of material worn away and broken off in water, arising from the researcher's individual errors. In addition, it is not possible to take a reliable measurement as extremely weak samples will disintegrate in the water.

[0007] In the density experiments conducted today, the rock sample is saturated with water for periods of 2, 4, 6, 8 and 24 hours (for saturated density). At each period, the rock sample is weighed and its weight is recorded. If the sample is saturated with water after 24 hours, the measured weight should be constant. The same process is performed on fresh rock samples with natural moisture content by conducting weight measurements at fixed intervals at a constant temperature, and the measurement continues until the sample's weight stabilizes. Here, researchers determine parameters such as dry and saturated density, porosity, natural moisture content, and time-dependent weight loss in studies lasting a minimum of 48 hours. In the light of the experiments, it takes a minimum of 48 hours to determine the physical parameters of the rock material. In addition, since all of the experiments are carried out in a laboratory environment, it takes longer time for the results of the experiments to come out. This results in slower process control, reducing the cost-time performance.

[0008] The possibility of obtaining the most accurate density and physical property parameters of the rock fragment with on-site measurement will lead to improved accuracy and reduced cost. However, there is currently no experiment or analysis device capable of performing on-site measurements. Delays in measurements from rock type to physical properties will increase the cost and lead to engineering errors. Because the natural moisture content of the material will have changed and will lead to individual errors by the researcher in an artificial environment.

[0009] The long duration of experimental studies, loss of samples due to researcher errors, and consequently inaccurate measurement results lead to increased costs in mining applications such as reinforcement, transportation, and explosive charging. Almost all methods used to determine the physical properties of rock fragments involve water. Whether the rock is saturated with water is determined through periodic weight measurements. This process not only prolongs the experiment but also causes the disintegration of weak rock materials under the effect of water, resulting in sample loss.

[0010] It is of great importance to develop a system developed for fast, accurate, and contactless determination of the physical properties (such as density, volume, and porosity) of the samples, which, compared to traditional methods, analyzes the characteristics of the samples instantly with multiple measurements per second using three-dimensional scanning technology and machine learning in a water-free environment, thus preventing problems such as researcher errors, sample loss, long drying and manual measurement processes, and additionally, through its portable and WiFi-connected device, allows samples to be analyzed on-site, reducing unnecessary transportation and labor costs, while optimizing processes without compromising the accuracy and speed of the results.

[0011] Summary and Objects of the Invention

[0012] The invention relates to a system for determining rock properties from a 3-dimensional data cloud with machine learning.

[0013] An object of the invention is to enable accurate and contactless determination of the physical properties (such as density, volume, porosity) of rock samples. This allows to minimize researcher errors, sample loss, and manual measurement processes.

[0014] Another object of the invention is that it allows to significantly reduce the measurement time compared to traditional methods (up to 3-5 hours). By making multiple measurements per second, it enables results to be obtained quickly.

[0015] Another object of the invention is to enable the volume of samples to be calculated and modeled in a virtual environment using three-dimensional scanner technology. It analyzes the surface properties of the rock such as color and roughness with image processing technology.

[0016] Another object of the invention is to enable automatic classification of porosity and other physical properties of different rock types using machine learning. It increases the accuracy of the results by generating validation coefficients with data from previous experiments.

[0017] Another object of the invention is to enable on-site analysis of samples thanks to the portable nature of the device. This enables quick decision-making in the field, allowing unnecessary samples to be eliminated and reducing costs.

[0018] Another object of the invention is to reduce the cost of experiments by achieving fast results without requiring water. Another object of the invention is to enable the device to develop its own literature by transferring the experimental data obtained via WiFi connection to the data cloud.

[0019] Another object of the invention is to provide solutions that can be used not only in mining but also in other fields where material density and physical properties are important.

[0020] Detailed Description of the Invention

[0021] The invention relates to a system for determining rock properties from a 3-dimensional data cloud with machine learning.

[0022] The heater surrounds the sample container with resistors to heat the rock sample up to 500 degrees Celsius. Here, heat control is performed in 3 dimensions, ensuring that each surface of the sample is heated equally.

[0023] A heat sensor is a device used to ensure that the sample reaches the desired temperature quickly and that this temperature is maintained precisely. This sensor continuously measures the current temperature of the sample and optimizes the heating process required to reach the set target temperature. By responding instantly to temperature changes, the heat sensor prevents the sample from overheating or underheating, thus increasing the accuracy and repeatability of the experimental conditions. It also offers an efficient heating mechanism, saving energy during this process.

[0024] Weight Sensor; This sensor, with a sensitivity of 0.01 , is designed to measure the weight of the sample with high accuracy. The system takes 10 measurements per second, calculates the statistical average of these measurements, and produces a single weight value for each second. This process increases the reliability of the measurement results and minimizes random errors. The weight sensor quickly detects sudden changes in the sample's weight, providing accurate and stable measurements. This makes it particularly suitable for use in fields where precision is critical, such as laboratory work, industrial weighing systems, or process control applications. In the invention, the rock material is dried and its initial dry weight is precisely measured. The rock material is immersed in water and the water saturation process is initiated. During the process, 10 weight measurements are taken per second and the average of these measurements is used to determine an average weight measurement per second by the processor.

[0025] 3-Dimensional Scanner: it is used for precise assessment of the surface and internal structure of the rock sample. The scanner scans the outer surface of the sample with high accuracy, generating coordinate data in the form of a point cloud. This coordinate data is decomposed into polygons representing the outer surface of the sample and then into tetrahedrons representing its internal volume. In this way, the data obtained from the scanner allows the total volume of the sample to be calculated with high accuracy. The volume of visible voids and pores is calculated by the microprocessor to determine the porosity properties of the rock sample.

[0026] The microprocessor receives and processes data from the temperature sensor, weight sensor, and 3-dimensional scanner. The microprocessor is in communication with the heater. Microprocessor calculates volume from 3-dimensional scanner data. After modeling the sample, the microprocessor classifies the sample for rocks with different porosity from surface porosity properties. The microprocessor analyzes the color and roughness surface properties of the rock from the 3-dimensional scanner data. The microprocessor follows the weight sensor and calculates an average weight measurement per second using the average of the measurements taken within a specified time period. The microprocessor determines a validation coefficient based on the type of rock material, using the maximum and minimum water saturation weighttime curves obtained from experimental results in the literature stored in the cloud data system. The microprocessor verifies the water saturation point based on rock type using machine learning with a determined confidence interval parameter, The microprocessor accurately determines the water saturation point through machine learning and validation coefficient and reports the results. The microprocessor, with the data it receives from the temperature sensor, transfers the numerical data to the researcher graphically over the interface when the rock completely loses its natural moisture. The microprocessor continuously improves itself by storing the results of each experiment in the cloud database via wireless connection. In this way, the device will develop its own literature and provide approximate porosity information according to the rock type, allowing subsequent experiments to be performed in a shorter time. The interface works on an electronic device to present the data calculated by the microprocessor to the user.

[0027] The operation of the pre-operated system was reduced to 3-5 hours depending on the type and porosity of the rock sample. The system performs volume calculation by modeling the rock sample with a three-dimensional scanner in a waterless environment, enabling more accurate measurements to be completed in less time. In addition to the physical parameters of the rock samples, the obtained three- dimensional model and image processing technology provide information regarding the color and roughness of the rock. Thanks to the rapid execution of all measurements, significant cost-performance benefits will be achieved. Sensors used in three- dimensional scanners, volume calculation techniques used in three-dimensional modeling, image processing software, and machine learning based on rock type will provide data for determining experiment durations.

[0028] Measuring the physical properties of rock on-site and identifying specific rock types is important for engineers and researchers. The portability of the device provides significant gains for the operation in situations that constantly change or are likely to change depending on the formation, such as reinforcement, explosive charging, and transportation. The prototype to be produced is designed to be modular so that the analyses can provide earlier and more accurate results. This will enable on-site experiments independent of testing and analysis institutions, allowing for quick decision-making as a result of this experiment and savings in costs and labor. The device will perform a preliminary analysis of samples, allowing researchers and the industry to eliminate unnecessary samples in the field. This will reduce costs by eliminating the need to bring all kinds of samples from the field to the laboratory.

Claims

CLAIMS1. A system for determining rock properties from a 3-dimensional data cloud with machine learning, characterized in that it comprises:- At least one heater that surrounds the sample container in which the rock sample is placed with resistors to heat the rock sample,- at least one temperature sensor that continuously measures the current temperature of the sample so that it reaches the set target temperature,- at least one weight sensor that measures the weight of the sample,- at least one 3-dimensional scanner that scans the outer surface of the sample, generates coordinate data in the form of a point cloud, and decomposes this coordinate data into polygons representing the outer surface of the sample and then into tetrahedrons representing its internal volume,- at least one microprocessor that receives and processes data from temperature sensor, weight sensor, and 3-dimensional scanner; calculates volume using 3- dimensional scanner data; classifies the sample for rocks with different porosity from surface porosity properties after the sample is modeled; analyzes the color and roughness surface properties of the rock from 3-dimensional scanner data; follows the weight sensor and calculates an average weight measurement per second using the average of the measurements taken within a specified time period, determines a validation coefficient based on the type of rock material, using the maximum and minimum water saturation weight-time curves obtained from experimental results in the literature stored in the cloud database, verifies the water saturation point based on rock type using machine learning with a determined confidence interval parameter, and determines the water saturation point through machine learning and the validation coefficient, and reports the results,- at least one interface that works on the electronic device, presents the data calculated by the microprocessor to the user, and displays this data graphically.

2. The system for determining rock properties from a 3-dimensional data cloud with machine learning according to claim 1 , characterized in that it comprises a microprocessor that detects the moment when the rock completely loses its natural moisture with the data received from the temperature sensor.

3. The system for determining rock properties from a 3-dimensional data cloud with machine learning according to claim 1 , characterized in that it comprises a microprocessor that stores the results of each experiment in a cloud database via wireless connection.

4. The system for determining rock properties from a 3-dimensional data cloud with machine learning according to claim 1 , characterized in that it comprises at least one heater for heating the rock sample up to 500 degrees Celsius.

5. The system for determining rock properties from a 3-dimensional data cloud with machine learning according to claim 1 , characterized in that it comprises a processor that takes 10 weight measurements per second from a weight sensor and calculates an average weight measurement per second using the average of these measurements.

6. The system for determining rock properties from a 3-dimensional data cloud with machine learning according to claim 1 , characterized in that it comprises a weight sensor with an accuracy of 0.01 .