A streaming big data analysis and dynamic preprocessing method for deep geological drilling process

Abstract

The present invention discloses a method for analyzing and dynamically preprocessing streaming big data of deep geological drilling processes, which mainly solves the problem that traditional methods are difficult to effectively remove various types of drilling process data noise such as spikes and burrs under the conditions of streaming big data during drilling. The data preprocessing work is carried out in three steps. First, considering the streaming big data of the drilling process, the limiting filter is used to remove outliers in the process data; then, the sliding window strategy is used to realize the iterative calculation of dynamic streaming big data; finally, the Savitzky Golay filter is used in each window to smooth the process data to further improve the data quality. The method proposed in the present invention overcomes the defects of the existing data analysis and preprocessing methods, such as slow operation speed and inability to run online, and lays a good foundation for further analysis and utilization of data to improve the safety and efficiency of deep geological drilling processes.

Description

Technical Field

[0001] The present invention relates to a method for stream-type big data analysis and dynamic preprocessing of a deep geological drilling process, and belongs to the field of intelligent control of complex geological drilling processes. Background Art

[0002] Resource and energy security is a crucial component of national security and a key to sustainable national economic development. Ensuring resource and energy security should be based on domestic resources. With the increasing depletion of shallow mineral resources, the development of deep mineralization theory, and the discovery of a large number of deep mineral resources both domestically and internationally, deep geological exploration and development have become inevitable. However, deep geological drilling processes present complex stratigraphic conditions and low data value density. Traditional methods struggle to effectively remove spikes, burrs, and other types of drilling process noise from streaming big data in-drilling, creating significant challenges for further analysis and utilization of this data to improve the safety and efficiency of deep geological drilling processes. Therefore, in-depth analysis and preprocessing of streaming big data from deep geological drilling processes is of great scientific and engineering significance for improving data quality and achieving safe and efficient deep drilling. Summary of the Invention

[0003] The purpose of this invention is to propose a method for streaming big data analysis and dynamic preprocessing of deep geological drilling processes. This method establishes a streaming big data analysis and dynamic preprocessing framework for the drilling process by integrating the ideas of static data preprocessing and sliding windows. It can effectively improve the quality of streaming data in deep geological drilling processes and lay an important foundation for further analysis and utilization of data to improve the safety and efficiency of deep geological drilling processes.

[0004] In order to achieve the above technical objectives, the technical solution adopted by the present invention is: a method for streaming big data analysis and dynamic preprocessing of deep geological drilling process, comprising the following steps:

[0005] S1: Considering the raw data collected during the actual drilling process, a limiting filter is applied to remove outliers in the raw process data in combination with the distribution characteristics of the process data and the manual operation experience of the driller / driver;

[0006] S2: Use a sliding window strategy to put the original data after removing outliers into the data window, and implement iterative calculation of dynamic streaming big data through window sliding;

[0007] S3: Using Savitzky-Golay filtering in each window to smooth the input process data to further improve data quality. After smoothing, the core parameters are obtained. The core parameters include the clipping filter threshold, the sliding window width and sliding distance, and the polynomial order and window width of the Savitzky-Golay filter.

[0008] S4: Based on the values of the core parameters, the optimal method for streaming big data analysis and dynamic preprocessing of the deep geological drilling process shown in steps S1-S4 is obtained, and then applied to actual drilling projects.

[0009] Furthermore, step S1 specifically includes the following process:

[0010] 1) Considering the original data collected during the actual drilling process, the kernel density estimation method is used to extract the distribution characteristics of these original data and obtain the probability density estimation of the original data;

[0011] 2) Based on the probability density estimation of the original data, the limiting filter is used in combination with the distribution characteristics of the process data and the manual operation experience of the driller / captain to remove outliers in the original data.

[0012] Furthermore, the raw data includes: weight on bit, hook load, drilling speed, rotation speed and torque.

[0013] Furthermore, step S2 specifically includes the following process:

[0014] 1) Using a sliding window strategy, the data after removing outliers is placed into a data window, and limited process data is analyzed and processed in the window to speed up the processing and meet the real-time computing requirements;

[0015] 2) After the data in the window is processed, some of the drilled data in the current window is discarded by sliding the window, and some new data is added at the same time to form a new window, thereby realizing iterative calculation of dynamic streaming big data.

[0016] The beneficial effects of the present invention based on its technical solution are:

[0017] (1) The present invention provides a method for analyzing and dynamically preprocessing streaming big data of deep geological drilling process. First, the streaming big data of drilling process is considered. The outliers in the process data are removed by using limiting filtering combined with the distribution characteristics of process data and the manual operation experience of the driller / captain, and the drilling data is preliminarily processed.

[0018] (2) The present invention provides a method for analyzing and dynamically preprocessing streaming big data in deep geological drilling processes. A sliding window strategy is introduced to dynamically process the streaming drilling big data after removing outliers, thereby accelerating the processing speed to meet real-time computing requirements. Savitzky-Golay filtering is then used in each window to improve the smoothness of the drilling data and reduce data noise.

[0019] (3) The present invention provides a method for analyzing and dynamically preprocessing deep geological drilling process streaming big data. First, the method collects drilling process data from actual drilling projects, conducts simulation experiments and analysis, determines the core parameters of the relevant methods, and then conducts engineering application practice based on this data, which is closely integrated with the actual drilling projects. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 Streaming big data analysis and preprocessing framework for deep geological drilling process;

[0021] Figure 2 Probability density estimation of raw data from actual drilling projects;

[0022] Figure 3 Probability density estimation of processed data in actual drilling engineering;

[0023] Figure 4 Some engineering application results. DETAILED DESCRIPTION

[0024] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments:

[0025] Example:

[0026] A streaming big data analysis and dynamic preprocessing method for deep geological drilling process. First, construct Figure 1 The illustrated framework for analyzing and preprocessing streaming big data during deep geological drilling processes consists of a deep geological drilling process streaming big data analysis and dynamic preprocessing layer and a basic automation layer. The basic automation layer includes a rotary control system, a feed control system, and a circulation control system. Each system can acquire data from monitoring elements, and the actuators execute corresponding instructions. In the deep geological drilling process streaming big data analysis and dynamic preprocessing layer, the data preprocessing process is divided into three stages. The first stage considers the drilling process streaming big data and uses a limiting filter combined with the distribution characteristics of the process data and the manual operation experience of the driller / captain. The second stage uses a sliding window strategy to place the collected process data into a data window, analyze and process the limited process data within the window, and implement iterative calculation of dynamic streaming big data through window sliding. The third stage uses a Savitzky-Golay filter to smooth the process data in each window. The Savitzky-Golay filter can improve the smoothness of the drilling data, reduce the interference of data noise, and further improve data quality. Then, the method of the present invention was used to carry out simulation experiments based on the original data of the drilling process. The core parameter values of the method of the present invention were obtained by combining the distribution characteristics of the process data, the manual operation experience of the driller / captain, and the trial and error method. Then, the method of the present invention was applied to engineering projects. Some of the engineering application results are shown in Figure 2. Figure 4 The specific steps are as follows:

[0027] (1) Based on the drilling process data collected from an actual project (this implementation takes the 3000-meter scientific drilling project in Dandong, Liaoning as an example), a simulation experiment is carried out to determine the core parameters (limiting filter threshold, sliding window width, sliding distance, polynomial order, etc.) of the relevant methods (limiting filter, sliding window and Savitzky Golay filter) in the present invention. The main steps of the simulation experiment are as follows:

[0028] 1) The original data of drilling process including weight on bit, hook load, drilling speed, rotation speed and torque collected in actual engineering are analyzed to obtain the data statistics table shown in Table 1. The distribution of the above data is calculated by using the kernel density estimation method, and the following is obtained: Figure 2 The probability density estimation of some of the original data shown shows that the data is non-Gaussian distributed, the data distribution is discrete, and the corresponding data probability density estimation value is low;

[0029] Table 1 Statistics of some original data of Liaoning Dandong 3000m scientific drilling project

[0030] Parameter name Maximum Minimum average value Standard deviation Drilling pressure (kg) 9993 0 734.9 985.0 Hook load (kg) 20160 0 5777.7 4697.1 Drilling speed (cm / min) 999.4 0 4.0 38.7 Speed (rpm) 1023 0 118.7 127.6 Torque (Nm) 9987 0 514.3 582.1

[0031] 2) Then, based on the distribution characteristics of process data and the manual operation experience of the driller / captain, the outliers in the original drilling data are removed by using the limiting filter. Figure 2 The probability density estimate of the raw WOB data shown in the figure shows that the WOB data is primarily distributed between 0 and 5000 kg. However, due to factors such as measurement error, a few data points differ significantly from the rest of the data and can be considered outliers. To more accurately reflect the data distribution characteristics and reduce data noise, a clipping filter method can be used, combining the data distribution characteristics with the manual operation experience of the driller / captain. The upper threshold of the clipping filter is set to approximately 5000 kg and the lower threshold is set to 0 kg to remove outliers in the WOB data.

[0032] 3) After removing outliers, a sliding window strategy is used to place the above data into a data window, accelerating processing to meet real-time computing needs. This strategy also enables iterative updates of dynamic streaming big data through window sliding. Two core parameters in the sliding window strategy influence the effectiveness of data iteration: the sliding window width and the sliding distance. A longer sliding window width increases the data diversity within the window, but also increases data processing time, and vice versa. A smaller sliding distance increases the amount of overlapping data between windows, and vice versa. These parameters can be determined using a trial-and-error approach based on data distribution characteristics.

[0033] 4) Use Savitzky Golay filtering in the window to smooth the limited process data and further improve the data quality. There are two core parameters in Savitzky Golay filtering that affect the filtering effect, namely the order of the polynomial and the width of the sliding window. The larger the order of the polynomial, the higher the fitting accuracy, but the lower the generalization performance, and vice versa; the smaller the sliding window width, the smaller the difference between the data before and after filtering, and vice versa. The above core parameters can be determined by trial and error based on the data distribution characteristics. For example, observe the data smoothing effect under different core parameter combinations. If there are many spikes and glitches in the data, the polynomial order should be reduced and the sliding window width should be increased, and vice versa.

[0034] 5) Determine the core parameters of the method according to the above steps, and use the method of the present invention to Figure 2 The original data in the 3000-meter scientific drilling project in Dandong, Liaoning Province were processed to obtain the probability density statistics of the processed data. Figure 3 It is clear that after analysis and preprocessing by the method of the present invention, the distribution of process data is more concentrated, and the spikes and burrs in the data are removed. The probability density estimate of the processed data is much larger than that of the original data and approximately conforms to the Gaussian distribution, which improves the quality of deep geological drilling process data and lays an important foundation for the engineering application of the method of the present invention.

[0035] (2) After the values of the relevant core parameters in the present invention are determined, the method of the present invention is applied to an engineering project, and the method of the present invention is compared with the method of limiting filtering combined with a sliding window, and the following is obtained: Figure 4 The results of some engineering applications are shown in FIG. It can be seen from the engineering application results that the method of the present invention further removes noise from deep geological drilling process data and improves the quality of process data.

[0036] The beneficial effects of the present invention are: it can ensure the noise filtering effect while taking into account the real-time computing speed, remove the peaks and burrs in the streaming big data of the drilling process, improve the data quality of the deep geological drilling process, and lay an important foundation for further analysis and utilization of data to improve the safety and efficiency of the deep geological drilling process.

[0037] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for streaming big data analysis and dynamic preprocessing in deep geological drilling process, characterized in that: The following steps are involved: S1: Considering the raw data collected during the actual drilling process, the outliers in the raw data are removed by applying a limiting filter combined with the distribution characteristics of the process data and the manual operation experience of the driller / captain; The following processes are included: 1) Considering the original data collected during the actual drilling process, the kernel density estimation method is used to extract the distribution characteristics of these original data and obtain the probability density estimation of the original data; 2) Based on the probability density estimation of the original data, the limit filter is used in combination with the distribution characteristics of the process data and the manual operation experience of the driller / captain to remove outliers in the original data; S2: Use the sliding window strategy to put the original data after removing outliers into the data window, and realize the iterative calculation of dynamic streaming big data through the form of window sliding; The following processes are included: 1) Using a sliding window strategy, the data after removing outliers is placed into a data window, and limited process data is analyzed and processed in the window to speed up the processing and meet the real-time computing requirements; 2) After the data in the window is processed, some of the drilled-in data in the current window is discarded by sliding the window, while some new data is added to form a new window, thereby achieving iterative calculation of dynamic streaming big data; S3: Using Savitzky-Golay filtering in each window to smooth the input process data to further improve data quality. After smoothing, the core parameters are obtained. The core parameters include the clipping filter threshold, the sliding window width and sliding distance, and the polynomial order and window width of the Savitzky-Golay filter. S4: Based on the values of the core parameters, the optimal method for streaming big data analysis and dynamic preprocessing of the deep geological drilling process shown in steps S1-S4 is obtained, and then applied to actual drilling projects.

2. A method for streaming big data analysis and dynamic preprocessing of deep geological drilling process according to claim 1, characterized in that: The raw data include: weight on bit, hook load, drilling speed, rotation speed and torque.

Images