Real-time Monitoring Method of In-situ Stress Based on Multi-source Data Fusion

Abstract

This application relates to the technical field of data processing, specifically to a real-time in-situ stress monitoring method based on multi-source data fusion. The method includes: collecting three-dimensional in-situ stress data at the in-situ stress monitoring location to construct an in-situ stress ratio depth sequence, a horizontal in-situ stress difference depth sequence, a maximum horizontal in-situ stress depth sequence, and a minimum horizontal in-situ stress depth sequence; calculating stress ratio outliers based on the differences between the in-situ stress ratio depth sequences; obtaining maximum horizontal stress outliers and minimum horizontal stress outliers respectively according to the distribution relationships between the horizontal in-situ stress difference depth sequence and the maximum horizontal in-situ stress depth sequence, and the minimum horizontal in-situ stress depth sequence; obtaining the data fusion weights at each acquisition moment of each sensor based on the stress ratio outliers, the maximum horizontal stress outliers, and the minimum horizontal stress outliers, and completing the real-time monitoring of in-situ stress. Thereby improving the reliability and confidence of in-situ stress monitoring data.

Description

Technical Field

[0001] This application relates to the technical field of data processing, and specifically relates to a real-time ground stress monitoring method based on multi-source data fusion. Background Art

[0002] Ground stress refers to the forces existing within the rocks of the Earth's crust, which are caused by factors such as the Earth's tectonic activities, gravitational forces, and surface processes. By monitoring ground stress in real time, it can help researchers understand the geodynamic processes, prevent natural disasters, ensure the safety of engineering construction, and the development of resources such as oil fields and coal.

[0003] Accurate ground stress monitoring data is crucial for decision-making processes such as geological disaster prevention and resource development. However, the monitoring data of ground stress may be affected by various factors, such as systematic deviations in monitoring equipment, environmental noise, installation deviations of equipment, etc., which can lead to deviations in the collected ground stress data. And the data fusion method based on the weighted average method is a commonly used multi-source data fusion technology. This method performs weighted averaging on multiple data sources of ground stress and reasonably allocates weights, which can reduce the impact of errors or outliers in a single data source on the overall ground stress monitoring results and reduce the possibility of some data sources with important information being ignored or having a reduced contribution. Summary of the Invention

[0004] To solve the above technical problems, this application provides a real-time ground stress monitoring method based on multi-source data fusion to solve existing problems.

[0005] The real-time ground stress monitoring method based on multi-source data fusion of this application adopts the following technical solutions:

[0006] An embodiment of this application provides a real-time ground stress monitoring method based on multi-source data fusion, and this method includes the following steps:

[0007] Collect three-dimensional ground stress data at the ground stress monitoring locations at each moment, and obtain the vertical ground stress sequence, maximum horizontal ground stress sequence, minimum horizontal ground stress sequence, and formation depth value of each sensor on each grating line;

[0008] Based on the vertical ground stress sequence, maximum horizontal ground stress sequence, minimum horizontal ground stress sequence, and formation depth value of each sensor, construct the ground stress ratio-depth sequence, horizontal ground stress difference-depth sequence, and maximum horizontal ground stress-depth sequence of each grating line at different acquisition moments;

[0009] According to the differences between the ground stress ratio-depth sequences of each grating line at each acquisition moment and the data points corresponding to each acquisition moment of each sensor within the ground stress ratio-depth sequence, determine the stress ratio outliers at each acquisition moment in the ground stress ratio sequence of each sensor;

[0010] Determine the maximum horizontal stress anomaly values at each acquisition moment in the maximum horizontal stress sequence of each sensor according to the distribution relationship between the horizontal stress difference depth sequence and the maximum horizontal stress depth sequence of each grating line at each acquisition moment;

[0011] Adopt the construction method of the maximum horizontal stress depth sequence to obtain the minimum horizontal stress depth sequence. For the data points in the horizontal stress difference depth sequence and the minimum horizontal stress depth sequence, use the same calculation method as the maximum horizontal stress anomaly value to obtain the minimum horizontal stress anomaly values at each acquisition moment in the minimum horizontal stress sequence;

[0012] Determine the data fusion weights at each acquisition moment of each sensor based on the stress ratio anomaly values, maximum horizontal stress anomaly values, and minimum horizontal stress anomaly values at each acquisition moment of each sensor;

[0013] Use the data fusion weights at each acquisition moment of each sensor to complete the real-time monitoring of the ground stress.

[0014] Preferably, the construction method of the ground stress ratio depth sequence is as follows:

[0015] Calculate the mean value between the data points at each acquisition moment in the maximum horizontal stress sequence of each sensor and the data points at the corresponding acquisition moment in the minimum horizontal stress sequence, denoted as the first mean value. The first mean values of all acquisition moments form the average horizontal stress sequence of each sensor;

[0016] Calculate the ratio between the data points at each acquisition moment in the average horizontal stress sequence of each sensor and the data points at the corresponding acquisition moment in the vertical stress sequence. The ratios of all acquisition moments form the ground stress ratio sequence of each sensor;

[0017] Arrange the values of the data points at the same acquisition moment in the ground stress ratio sequences of all sensors on each grating line in ascending order according to the formation depth values of the sensors to form the ground stress ratio depth sequence of each grating line at the same acquisition moment.

[0018] Preferably, the calculation method of the stress ratio anomaly values at each acquisition moment in the ground stress ratio sequence of each sensor is as follows:

[0019] Calculate the Pearson correlation coefficient between the values of the data points in the ground stress ratio depth sequence of each grating line at each acquisition moment and the sequence composed of the subscript values of the data in the ground stress ratio depth sequence, as the ratio depth correlation coefficient of the ground stress ratio depth sequence of each grating line at each acquisition moment;

[0020] Calculate the ratio sequence outliers of the in-situ stress ratio depth sequence for each grating line at each acquisition time based on the differences and ratio depth correlation coefficients between the in-situ stress ratio depth sequences of each grating line at each acquisition time;

[0021] Calculate the stress ratio characteristic values at each acquisition time in the in-situ stress ratio sequence for each sensor based on the formation depth values of each sensor and the values of the data points at each acquisition time in the in-situ stress ratio sequence;

[0022] Take the product of the ratio sequence outliers and the stress ratio characteristic values at the corresponding time as the stress ratio outliers at each acquisition time in the in-situ stress ratio sequence for each sensor.

[0023] Preferably, the calculation method for the ratio sequence outliers of the in-situ stress ratio depth sequence for each grating line at each acquisition time is as follows:

[0024] Calculate the mean value of the DTW distances between the in-situ stress ratio depth sequence of each grating line at each acquisition time and the in-situ stress ratio depth sequences of the remaining grating lines at the corresponding acquisition time, denoted as the second mean value;

[0025] Take the product of the calculation result of the exponential function with the natural constant as the base and the ratio depth correlation coefficient as the exponent and the second mean value as the ratio sequence outliers of the in-situ stress ratio depth sequence for each grating line at each acquisition time.

[0026] Preferably, the calculation formula for the stress ratio characteristic values at each acquisition time in the in-situ stress ratio sequence for each sensor is:

[0027]

[0028] In the formula, s ′ m ′ ,n,t represents the stress ratio characteristic value at the t-th acquisition time in the in-situ stress ratio sequence of the n-th sensor on the m-th grating line; H m,n represents the value of the formation depth of the n-th sensor on the m-th grating line; b m,n,t represents the value of the data point at the t-th acquisition time in the in-situ stress ratio sequence of the n-th sensor on the m-th grating line; β is a tuning constant; max() represents the function that outputs the maximum value in the parentheses.

[0029] Preferably, the construction method for the horizontal in-situ stress difference depth sequence and the maximum horizontal in-situ stress depth sequence is as follows:

[0030] The horizontal in-situ stress difference sequence of each sensor on each grating line is obtained according to the difference between the data points at each acquisition time in the maximum horizontal in-situ stress sequence and the data points at the corresponding acquisition time in the minimum horizontal in-situ stress sequence;

[0031] The values of the data points at the same acquisition time in the horizontal in-situ stress difference sequences of all sensors on each grating line are arranged in ascending order according to the formation depth values of the sensors to form the horizontal in-situ stress difference depth sequence of each grating line at the same acquisition time;

[0032] The values of the data points at the same corresponding acquisition time in the maximum horizontal in-situ stress sequences of all sensors on each grating line are arranged in ascending order according to the formation depth values of the sensors to form the maximum horizontal in-situ stress depth sequence of each grating line at the same acquisition time.

[0033] Preferably, the method for determining the maximum horizontal in-situ stress anomaly value at each acquisition time in the maximum horizontal in-situ stress sequence of each sensor is as follows:

[0034] According to the horizontal in-situ stress difference depth sequence and the maximum horizontal in-situ stress depth sequence of each grating line at each acquisition time, the difference sequence anomaly value of the horizontal in-situ stress difference rate depth sequence at each acquisition time is obtained, and the maximum horizontal in-situ stress depth correlation coefficient of the maximum horizontal in-situ stress depth sequence is obtained;

[0035] Based on the data points in the maximum horizontal in-situ stress depth sequence, linear fitting is performed, and the Euclidean distance from each data point to the straight line is calculated to obtain the point-line distance of each data point in the maximum horizontal in-situ stress depth sequence.

[0036] The calculation formula for the maximum horizontal in-situ stress anomaly value at each acquisition time in the maximum horizontal in-situ stress sequence of each sensor is as follows:

[0037]

[0038] In the formula, represents the stress sequence anomaly value of the maximum horizontal in-situ stress depth sequence of the mth grating line at the tth acquisition time; w ′ m,t represents the difference sequence anomaly value of the horizontal in-situ stress difference depth sequence of the mth grating line at the tth acquisition time; represents the maximum horizontal in-situ stress depth correlation coefficient of the maximum horizontal in-situ stress depth sequence;

[0039] Calculate the mean of the differences in the point-line distances between the data point of the nth sensor in the maximum horizontal in-situ stress depth sequence at the tth acquisition moment of the mth grating line and the remaining data points, denoted as the third mean. Denote the product of the third mean and the point-line distance of the data point of the nth sensor in the maximum horizontal in-situ stress depth sequence at the tth acquisition moment of the mth grating line as is the maximum horizontal stress anomaly value of the data point at the tth acquisition moment in the maximum horizontal in-situ stress sequence of the nth sensor of the mth grating line.

[0040] Preferably, the method for obtaining the difference sequence anomaly value of the horizontal in-situ stress difference rate depth sequence of each grating line at each acquisition moment and obtaining the maximum horizontal in-situ stress depth correlation coefficient of the maximum horizontal in-situ stress depth sequence includes:

[0041] Perform a first-order difference on the horizontal in-situ stress difference depth sequence to obtain the horizontal in-situ stress difference rate depth sequence of each grating line at each acquisition moment;

[0042] Calculate the Pearson correlation coefficient between the horizontal in-situ stress difference rate depth sequence of each grating line at each acquisition moment and the sequence composed of the subscript values of each data in the horizontal in-situ stress difference rate depth sequence, as the difference rate depth correlation coefficient of the horizontal in-situ stress difference rate depth sequence of each grating line at each acquisition moment. Denote the opposite number of the difference rate depth correlation coefficient as the difference sequence anomaly value;

[0043] Based on the maximum horizontal in-situ stress depth sequence, use the same calculation method of the difference rate depth correlation coefficient to obtain the maximum horizontal in-situ stress depth correlation coefficient of the maximum horizontal in-situ stress depth sequence of each grating line at each acquisition moment.

[0044] Preferably, the calculation method of the data fusion weight of each acquisition moment of each sensor is:

[0045] Calculate the product of the stress ratio anomaly value at each acquisition moment in the in-situ stress ratio sequence of each sensor, the maximum horizontal stress anomaly value at each acquisition moment in the maximum horizontal in-situ stress sequence, and the minimum horizontal stress anomaly value at each acquisition moment in the minimum horizontal in-situ stress sequence;

[0046] Take the normalized result of the reciprocal of the product as the weight distribution coefficient of each acquisition moment of each sensor;

[0047] Take the proportion of the weight distribution coefficient of each acquisition moment of each sensor in the weight distribution coefficients of all sensors at each acquisition moment as the data fusion weight of each acquisition moment of each sensor.

[0048] Preferably, the method for real-time monitoring of in-situ stress based on the data fusion weights at each acquisition time of each sensor is as follows:

[0049] Calculate the product of the vertical in-situ stress in the original three-dimensional in-situ stress data of each sensor at each acquisition time and the data fusion weight at each acquisition time, and take the mean of the products of all sensors as the vertical in-situ stress monitoring data at each acquisition time at the in-situ stress monitoring location;

[0050] For the maximum horizontal in-situ stress and the minimum horizontal in-situ stress, use the same acquisition method for the vertical in-situ stress monitoring data to obtain the maximum horizontal in-situ stress monitoring data and the minimum horizontal in-situ stress monitoring data at each acquisition time at the in-situ stress monitoring location;

[0051] Based on the vertical in-situ stress monitoring data, the maximum horizontal in-situ stress monitoring data, and the minimum horizontal in-situ stress monitoring data, monitor the in-situ stress in combination with a preset threshold.

[0052] This application has at least the following beneficial effects:

[0053] This application constructs stress ratio outliers by analyzing the distribution of data points in the stress ratio sequence and the stress ratio depth sequence constructed from each in-situ stress sequence. Its beneficial effect is to improve the recognition accuracy when there are data deviation anomalies in the vertical in-situ stress in the original three-dimensional in-situ stress data; by analyzing the distribution of data points in the horizontal stress difference depth sequence and the maximum horizontal stress depth sequence to construct the difference sequence outliers and stress sequence outliers to construct the maximum horizontal stress outliers, its beneficial effect is to improve the discrimination between the maximum horizontal in-situ stress and the minimum horizontal in-situ stress when there are data deviation anomalies in the original three-dimensional in-situ stress data; by combining the stress ratio outliers, the maximum horizontal stress outliers, and the minimum horizontal stress outliers to construct the sampling data anomaly feature values to construct the data fusion weights, its beneficial effect is to improve the discrimination of the data deviation degrees when different sensors collect the original three-dimensional in-situ stress data, and adaptively obtain the weight values of each sensor in the original three-dimensional in-situ stress data at each acquisition time for data fusion, reduce the influence of the original three-dimensional in-situ stress data with large deviations on the overall in-situ stress monitoring results, and improve the reliability and confidence of the in-situ stress monitoring data. Description of the Drawings

[0054] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0055] Figure 1 It is a flowchart of the steps of the in-situ stress real-time monitoring method based on multi-source data fusion provided by this application;

[0056] Figure 2 It is a distribution diagram of fiber Bragg gratings and sensors at the in-situ stress monitoring locations provided by this application;

[0057] Figure 3 It is a flowchart of the steps of the maximum horizontal stress anomaly values at each acquisition moment in the maximum horizontal in-situ stress sequence provided by this application. Specific embodiments

[0058] In order to further elaborate on the technical means and effects adopted by this application to achieve the predetermined invention purpose, the following combines the accompanying drawings and preferred embodiments to elaborate in detail on the specific embodiments, structures, features and their effects of the in-situ stress real-time monitoring method based on multi-source data fusion proposed according to this application. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0059] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which this application belongs.

[0060] The following specifically describes the specific solution of the in-situ stress real-time monitoring method based on multi-source data fusion provided by this application with reference to the accompanying drawings.

[0061] An in-situ stress real-time monitoring method based on multi-source data fusion provided by an embodiment of this application. Specifically, the following in-situ stress real-time monitoring method based on multi-source data fusion is provided. Please refer to Figure 1 , The method includes the following steps:

[0062] Step 1, collect the three-dimensional in-situ stress data of the in-situ stress monitoring locations at each moment, and obtain the vertical in-situ stress sequence, maximum horizontal in-situ stress sequence, minimum horizontal in-situ stress sequence and formation depth value of each sensor on each grating line.

[0063] Uniformly embed M grating lines vertically at the in-situ stress monitoring location, and uniformly install M high-precision three-dimensional in-situ stress sensors based on fiber Bragg gratings on each grating line, so that there are N sensors at the same formation depth of the in-situ stress monitoring location to collect the original three-dimensional in-situ stress data at each position at different depths of the in-situ stress monitoring location, which are respectively the vertical in-situ stress, the maximum horizontal in-situ stress, and the minimum horizontal in-situ stress. The empirical values of M and N are 16 and 20 respectively, and the data sampling time interval and sampling time of each sensor are taken as the empirical values of 20 seconds and 10 minutes respectively. This embodiment only provides a scheme for the distribution method of fiber Bragg gratings and sensors and the data sampling time, and other implementers can set it by themselves. For the convenience of subsequent processing, the set composed of all sensors on the m-th grating line is denoted as the sensor set G m , obtain the set G m the values of the formation depths where each sensor in is located, and denote the n-th sensor in the set G m as the sensor g m,n .

[0064] Furthermore, in an embodiment of the present application, the distribution of fiber Bragg gratings and sensors in the in-situ stress monitoring location is as Figure 2 shown.

[0065] Use the "max-min" normalization method to perform dimensionless processing on the original three-dimensional in-situ stress data of each sensor and the values of the formation depths of all sensors collected respectively. The "max-min" normalization method is a well-known existing technology and will not be elaborated here. Obtain the dimensionless processed in-situ stress sequences and formation depths of each sensor. Taking the sensor g m,n as an example, the value of its formation depth is H m,n , and the in-situ stress sequences include the vertical in-situ stress sequence the maximum horizontal in-situ stress sequence and the minimum horizontal in-situ stress sequence where the values of the data points in each sequence obtained are arranged in ascending order of the data point collection time.

[0066] So far, the vertical in-situ stress sequence, the maximum horizontal in-situ stress sequence, the minimum horizontal in-situ stress sequence, and the formation depth values of each sensor on each grating line are obtained.

[0067] Step 2, construct the in-situ stress ratio-depth sequence, the horizontal in-situ stress difference-depth sequence, and the maximum horizontal in-situ stress-depth sequence of each grating line at each collection time.

[0068] Generally, at the same acquisition time, the ratio of the average horizontal geostress to the vertical geostress will decrease with the increase of the formation depth and approach 1. The average horizontal geostress is the average of the maximum horizontal geostress and the minimum horizontal geostress collected by the sensor at each acquisition time. In addition, since sensors at the same formation depth on different grating lines generally have similar geological conditions, such as rock type and formation structure, and are subject to similar external influences, such as surface load and climate change, the ratio of the average horizontal geostress to the vertical geostress of sensors in different sensor sets at the same acquisition time has similar data changes with the change of the formation depth of the sensor.

[0069] Based on the above analysis, the sensor g m,n For example, calculate the sequence The average value between the data points at the same position in the , and the average horizontal ground stress series Calculate the sequence separately Each data point in the sequence The ratio of the data points at the same position in is recorded as the ground stress ratio, and the ground stress ratio sequence B is obtained. m,n .

[0070] Further, the calculation sequence The difference between the data points at the same position in the horizontal in-situ stress difference is recorded as the horizontal in-situ stress difference, and the horizontal in-situ stress difference sequence ΔA is obtained. m,n . Take the set G m For example, get the set G m The value of the tth data point in the horizontal stress difference sequence of each sensor in , all the obtained values ​​are arranged in ascending order according to the value of the formation depth of the sensor, and the set G is obtained. m The depth sequence of horizontal stress difference at the tth acquisition time GA m,t .

[0071] The set G m The values ​​of the t-th data point in the maximum horizontal ground stress sequence of all sensors in are arranged in ascending order according to the values ​​of the formation depth of the sensors, and the set G is obtained. m The maximum horizontal stress depth sequence at the tth acquisition time

[0072] At this point, the in-situ stress ratio depth sequence, the horizontal in-situ stress difference depth sequence and the maximum horizontal in-situ stress depth sequence of each grating line at each acquisition moment are obtained.

[0073] Step 3, determining the stress ratio abnormal value at each acquisition moment in the geostress ratio sequence of each sensor according to the difference between the geostress ratio depth sequences of each grating line at each acquisition moment and the data points of each sensor at each acquisition moment in the geostress ratio depth sequence.

[0074] Taking the set G m as an example, obtain the value of the t-th data point in the in-situ stress ratio sequence of each sensor in the set G m Arrange all the obtained values in ascending order according to the formation depth value of the sensor to obtain the set G m In the in-situ stress ratio depth sequence GB m,t at the t-th acquisition moment, calculate the Pearson correlation coefficient between the sequence of subscript values of each data point in the sequence GB m,t and the value of the data point, and obtain the ratio depth correlation coefficient p m,t of the sequence GB m,t where the subscript values of each data point are numbered sequentially from 1 in ascending order according to the formation depth value. The calculation of the Pearson correlation coefficient is a well-known technology, and the specific process will not be elaborated here.

[0075] Construct the stress ratio outlier to characterize the possibility of abnormality of each data point in the in-situ stress ratio sequence of each sensor. Calculate the stress ratio outlier S m,n of the t-th data point B m,n,t in the sequence B m,n,t :

[0076]

[0077] S m,n,t = s ′ m,t * s ′ m ′ ,n,t

[0078] In the formula, s ′ m,t represents the ratio sequence outlier of the sequence GB m,t ; DGB m,t represents the mean value of the DTW distance between the set G m and the in-situ stress ratio depth sequence of the remaining sensor sets at the t-th acquisition moment; p m,t represents the ratio depth correlation coefficient of the sequence GB m,t ; β represents the tuning parameter constant. To prevent the denominator from being 0, β takes an empirical value of 0.01; s ′ m ′ ,n,t represents the stress ratio eigenvalue at the t-th acquisition moment in the in-situ stress ratio sequence of the sensor g m,n ; H m,n represents the formation depth value of the sensor g m,n ; b m,n,t represents the sequence Bm,n the t-th data point B in m,n,t value; max() represents a function that outputs the maximum value within the parentheses.

[0079] It should be understood that the embodiments of the present application only provide a method for measuring the difference between two sensor set sequences, namely the DTW algorithm. As other implementation manners, on the premise of achieving the purpose of measuring the difference between two sensor set sequences, existing technology measurement methods can be used to measure the difference between two sensor set sequences, and this embodiment does not make special restrictions on this.

[0080] set G m The less similar the data changes are between it and the in-situ stress ratio depth sequence of the remaining sensor sets at the t-th acquisition moment, the greater the value of DGB m,t ; the weaker the negative correlation between the in-situ stress ratio of the data points in the sequence GB m,t and the formation depth, the greater the value of p m,t ; then the greater the value of s ′ m,t , the greater the possibility that the data points in the sequence GB m,t are abnormal;

[0081] While the formation depth of the sensor g m,n is getting larger, the value of the data point B m,n,t is getting closer to 1, the value of is getting smaller, or while the formation depth of the sensor g m,n is getting smaller, the value of the data point B m,n,t is getting less close to 1, the value of is getting smaller, then the possibility that the data point B m,n,t is abnormal is getting smaller, and the value of s ′ m ′ ,n,t is getting smaller; then the possibility that the data point B m,n,t is abnormal is getting larger, the value of S m,n,t is getting larger, and the possibility that the vertical in-situ stress or the maximum horizontal in-situ stress and the minimum horizontal in-situ stress collected by the sensor g m,n at the t-th acquisition moment have data deviation abnormalities is getting larger.

[0082] Thus far, the stress ratio anomaly values at each acquisition moment in the in-situ stress ratio sequence of each sensor are obtained.

[0083] Step 4: Calculate the maximum horizontal stress anomaly value at each acquisition time in the maximum horizontal stress sequence of each sensor and the minimum horizontal stress anomaly value at each acquisition time in the minimum horizontal stress sequence respectively according to the distribution relationship between the horizontal stress difference depth sequence of each grating line at each acquisition time and the maximum horizontal stress depth sequence and the minimum horizontal stress depth sequence.

[0084] Furthermore, the maximum horizontal stress and the minimum horizontal stress will show a linear growth relationship with the formation depth, and the difference in the stress values between the two will increase with the increase in depth.

[0085] Based on the above analysis, perform a first-order difference processing on the sequence GA m,t to obtain the horizontal stress difference change rate depth sequence ΔGA m,t , and use the same method as calculating the correlation coefficient p m,t to obtain the difference change rate depth correlation coefficient p m,t of the sequence ΔGA ′ m,t . Denote the opposite number of the correlation coefficient p ′ m,t as the difference sequence anomaly value w m,t of the sequence ΔGA ′ m,t . The larger the value of w ′ m,t , the weaker the positive correlation between the values of the data points in the sequence ΔGA m,t and the formation depth, and the greater the possibility that the data points in the sequence ΔGA m,t are abnormal. Then, the greater the possibility that the maximum horizontal stress and the minimum horizontal stress collected by the sensors in the set G m are abnormal at the t-th acquisition time.

[0086] Taking the set G m as an example, use the same method as calculating the correlation coefficient p m,t to obtain the maximum horizontal stress depth correlation coefficient of the sequence

[0087] Secondly, perform a linear fitting on the data points in the sequence . Taking the serial number and the value of the data points in the sequence as the abscissa and ordinate respectively, obtain the fitting straight line . Calculate the Euclidean distance from each data point in the sequence to the straight line to obtain the point-line distance of each data point in the sequence . Linear fitting is a well-known technology, and the specific process will not be elaborated here.

[0088] Construct the maximum horizontal stress anomaly value to characterize the possibility of anomalies occurring at each data point in the maximum horizontal in-situ stress sequence of each sensor. Calculate the sequence at the t-th acquisition moment data point of the maximum horizontal stress anomaly value

[0089]

[0090] In the formula, represents the stress sequence anomaly value at the t-th acquisition moment in the sequence ; w ′ m,t represents the difference sequence anomaly value of the sequence ΔGA m,t ; represents the maximum horizontal in-situ stress depth correlation coefficient of the sequence ; represents the product of the mean of the difference in the point-line distance between the data point m,n corresponding to the sensor g in the sequence and the remaining data points in the sequence and the point-line distance of the data point ;

[0091] Among them, the greater the possibility of anomalies in the maximum horizontal in-situ stress and minimum horizontal in-situ stress collected by the sensors in the set G m at the t-th acquisition moment, the greater the value of w ′ m,t ; the less the relationship between the value of the data point in the sequence and the formation depth shows a linear growth relationship, the greater the value of ; then the greater the possibility of anomalies in the data points in the sequence , the greater the point-line distance of the data point, and the greater the difference in the point-line distance between it and the remaining data points in , the greater the value of , the greater the possibility of anomalies in the data point m,n ; the greater the possibility of abnormal data deviation in the maximum horizontal in-situ stress collected by the sensor g at the t-th acquisition moment, the greater the value of

[0092] In an embodiment of the present application, the step flow chart of the method for obtaining the maximum horizontal stress anomaly value at each acquisition moment in the maximum horizontal in-situ stress sequence of each sensor is as Figure 3 shown.

[0093] ​Construct the minimum horizontal in-situ stress depth sequence based on the same method as above and replace the sequence with the sequence Construct the data point at the t-th acquisition moment in the sequence of the minimum horizontal stress anomaly value

[0094] Thus, the maximum horizontal stress anomaly values at each acquisition moment in the maximum horizontal in-situ stress sequence of each sensor and the minimum horizontal stress anomaly values at each acquisition moment in the minimum horizontal in-situ stress sequence are obtained.

[0095] Step 5: Calculate the data fusion weight at each acquisition moment of each sensor based on the stress ratio anomaly value, maximum horizontal stress anomaly value, and minimum horizontal stress anomaly value at each acquisition moment of each sensor.

[0096] Furthermore, taking the sensor g m,n as an example, multiply the ratio sequence anomaly value S m,n,t , the maximum horizontal stress anomaly value , and the minimum horizontal stress anomaly value to obtain the sampling data anomaly feature value f m,n of the sensor g m,n,t at the t-th acquisition moment. Construct the data fusion weight, which is used to characterize the weight value of each sensor when performing data fusion on the original three-dimensional in-situ stress data at each acquisition moment. Calculate the data fusion weight γ m,n of the sensor g m,n,t at the t-th acquisition moment as follows:

[0097]

[0098] where α m,n,t represents the weight distribution coefficient of the sensor g m,n at the t-th acquisition moment; f m,n,t represents the sampling data anomaly feature value of the sensor g m,n at the t-th acquisition moment; β represents the tuning parameter constant. To prevent the denominator from being zero, β takes an empirical value of 0.01; norm() represents a function for normalizing the value inside the parentheses.

[0099] The larger the value of the eigenvalue F m,n,t , the greater the possibility that the original three-dimensional in-situ stress data collected by the sensor g m,n at the t-th acquisition moment has a large deviation. Therefore, a smaller weight should be assigned when using the three-dimensional in-situ stress data for data fusion. The smaller the value of α m,n,t , the smaller the influence of the data deviation existing in the three-dimensional in-situ stress data on the overall monitoring result of the in-situ stress. Then, for the sensor g​m,n The data fusion weight γ at the t-th acquisition moment m,n,t The smaller the value is.

[0100] Thus, the data fusion weights at each acquisition moment of each sensor are obtained.

[0101] Step 6: Use the data fusion weights at each acquisition moment of each sensor to complete the real-time monitoring of in-situ stress.

[0102] Taking the vertical in-situ stress in the original three-dimensional in-situ stress data acquired at the t-th acquisition moment as an example, calculate the product of the vertical in-situ stress in the original three-dimensional in-situ stress data acquired by each sensor at the t-th acquisition moment and the data fusion weight of the sensor at the t-th acquisition moment respectively. Take the mean value of all the obtained products as the vertical in-situ stress monitoring value at the monitoring location of in-situ stress at the t-th acquisition moment. Obtain the vertical in-situ stress monitoring data at each acquisition moment of the in-situ stress monitoring location in the same way, and upload it to the in-situ stress monitoring system.

[0103] Based on the same method as above, replace the vertical in-situ stress with the maximum horizontal in-situ stress and the minimum horizontal in-situ stress respectively, and obtain the maximum horizontal in-situ stress monitoring data and the minimum horizontal in-situ stress monitoring data at each acquisition moment of the in-situ stress monitoring location respectively, and upload them to the in-situ stress monitoring system.

[0104] Collectively refer to the obtained vertical in-situ stress monitoring data, maximum horizontal in-situ stress monitoring data, and minimum horizontal in-situ stress monitoring data as three-dimensional in-situ stress monitoring data. If the value of any one of the three-dimensional in-situ stress monitoring data exceeds the monitoring threshold, an alarm is issued to remind the staff. Among them, the monitoring threshold is set by the staff according to the geological conditions of the in-situ stress monitoring location.

[0105] It should be noted that: the above sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. And the above description of specific embodiments of this specification is made. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0106] Each embodiment in this specification is described in a progressive manner. The same or similar parts between each embodiment can be referred to each other, and the key point of each embodiment is to illustrate the differences from other embodiments.

[0107] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them; any modification of the technical solutions recorded in the foregoing embodiments, or any equivalent replacement of some of the technical features, does not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application, and shall all be included within the protection scope of the present application.

Claims

1. A real-time in-situ stress monitoring method based on multi-source data fusion, comprising: Collecting three-dimensional in-situ stress data at each moment to obtain the vertical, maximum horizontal, minimum horizontal in-situ stress sequences and formation depth values of each sensor on each grating line; Calculating the mean value between the data points at each acquisition moment in the maximum horizontal in-situ stress sequence and the corresponding data points in the minimum horizontal in-situ stress sequence of each sensor, each forming an average horizontal in-situ stress sequence; Calculating the ratio of the data at the corresponding acquisition moment between the average horizontal in-situ stress sequence and the vertical in-situ stress sequence to form an in-situ stress ratio sequence; Arranging the values of the data points at the same acquisition moment in the in-situ stress ratio sequences of all sensors on each grating line in ascending order according to the formation depth values of the sensors to form the in-situ stress ratio depth sequence of each grating line at the same acquisition moment; Constructing the horizontal in-situ stress difference and maximum horizontal in-situ stress depth sequences of each grating line at each acquisition moment based on the vertical, maximum horizontal, minimum horizontal in-situ stress sequences and formation depth values of each sensor; Calculating the Pearson correlation coefficient between the values of each data point in the in-situ stress ratio depth sequence of each grating line at each acquisition moment and the sequence composed of the subscript values of each data in the in-situ stress ratio depth sequence as the ratio depth correlation coefficient of the in-situ stress ratio depth sequence; calculating the ratio sequence outliers of the in-situ stress ratio depth sequence of each grating line at each acquisition moment according to the differences between the in-situ stress ratio depth sequences and the ratio depth correlation coefficient; calculating the stress ratio characteristic values of each acquisition moment in each in-situ stress ratio sequence according to the formation depth values and the values of the data points at each acquisition moment in the in-situ stress ratio sequence; Taking the product of the ratio sequence outliers and the stress ratio characteristic values at the corresponding moments as the stress ratio outliers of each acquisition moment in the in-situ stress ratio sequence of each sensor; Determining the maximum horizontal stress outliers at each acquisition moment according to the distribution relationship between the horizontal in-situ stress difference depth sequence and the maximum horizontal in-situ stress depth sequence; Obtaining the minimum horizontal in-situ stress depth sequence by using the construction method of the maximum horizontal in-situ stress depth sequence, and obtaining the minimum horizontal stress outliers at each acquisition moment by using the same calculation method as the maximum horizontal stress outliers; Determining the data fusion weights of each acquisition moment of each sensor based on the stress ratio outliers, maximum horizontal stress outliers, and minimum horizontal stress outliers of each acquisition moment of each sensor; Completing the real-time monitoring of in-situ stress by using the data fusion weights of each acquisition moment of each sensor.

2. The real-time in-situ stress monitoring method based on multi-source data fusion according to claim 1, wherein The calculation method of the ratio sequence outliers of the in-situ stress ratio depth sequence of each grating line at each acquisition moment is as follows: Calculating the mean value of the DTW distances between the in-situ stress ratio depth sequence of each grating line at each acquisition moment and the in-situ stress ratio depth sequences of the remaining grating lines at the corresponding acquisition moment, denoted as the second mean value; Taking the product of the calculation result of the exponential function with the natural constant as the base and the ratio depth correlation coefficient as the exponent and the second mean value as the ratio sequence outliers of the in-situ stress ratio depth sequence of each grating line at each acquisition moment.

3. The real-time in-situ stress monitoring method based on multi-source data fusion according to claim 1, characterized in that, The calculation formula for the stress ratio characteristic values of each acquisition moment in the in-situ stress ratio sequence of each sensor is: In the formula, represents the eigenvalue of the stress ratio at the th acquisition moment in the sequence of the stress ratio of the nth sensor on the mth grating line; represents the value of the formation depth of the nth sensor on the mth grating line; represents the value of the data point at the th acquisition moment in the sequence of the stress ratio of the nth sensor on the mth grating line; is the parameter adjustment constant; represents the function that outputs the maximum value in the parentheses.

4. The real-time in-situ stress monitoring method based on multi-source data fusion according to claim 1, characterized in that The construction method of the horizontal in-situ stress difference depth sequence and the maximum horizontal in-situ stress depth sequence is as follows: Obtain the horizontal in-situ stress difference sequence of each sensor on each grating line according to the difference between the data points at each acquisition time in the maximum horizontal in-situ stress sequence and the corresponding data points at the same acquisition time in the minimum horizontal in-situ stress sequence; Arrange the values of the data points at the same acquisition time in the horizontal in-situ stress difference sequences of all sensors on each grating line in ascending order according to the formation depth values of the sensors to form the horizontal in-situ stress difference depth sequence of each grating line at the same acquisition time; Arrange the values of the data points at the same corresponding acquisition time in the maximum horizontal in-situ stress sequences of all sensors on each grating line in ascending order according to the formation depth values of the sensors to form the maximum horizontal in-situ stress depth sequence of each grating line at the same acquisition time.

5. The real-time in-situ stress monitoring method based on multi-source data fusion according to claim 4, characterized in that, The determination method of the maximum horizontal stress anomaly value at each acquisition time in the maximum horizontal in-situ stress sequence of each sensor is as follows: Obtain the difference sequence anomaly value of the horizontal in-situ stress difference rate depth sequence at each acquisition time according to the horizontal in-situ stress difference depth sequence and the maximum horizontal in-situ stress depth sequence of each grating line at each acquisition time, and obtain the maximum horizontal in-situ stress depth correlation coefficient of the maximum horizontal in-situ stress depth sequence; Based on the data points in the maximum horizontal in-situ stress depth sequence, perform linear fitting, calculate the Euclidean distance from each data point to the straight line, and obtain the point-line distance of each data point in the maximum horizontal in-situ stress depth sequence; The maximum horizontal stress anomaly value at each acquisition time in the maximum horizontal in-situ stress sequence of each sensor, the calculation formula is: In the formula, , represents the stress sequence outlier of the maximum horizontal in-situ stress depth sequence of the m-th grating line at the -th acquisition moment; represents the difference sequence outlier of the horizontal in-situ stress difference depth sequence of the m-th grating line at the -th acquisition moment; represents the maximum horizontal in-situ stress depth correlation coefficient of the maximum horizontal in-situ stress depth sequence. Calculate the mean value of the difference in the line segment distances between the data point of the nth sensor and the remaining data points in the maximum horizontal in-situ stress depth sequence at the th acquisition moment of the mth grating line, which is denoted as the third mean value. Multiply the third mean value by the line segment distance between the data point of the nth sensor and the th acquisition moment of the mth grating line in the maximum horizontal in-situ stress depth sequence, and denote the product as ; is the maximum horizontal stress anomaly value of the data point at the th acquisition moment in the maximum horizontal in-situ stress sequence of the nth sensor of the mth grating line.

6. The real-time in-situ stress monitoring method based on multi-source data fusion according to claim 5, wherein The obtaining of the difference sequence anomaly value of the horizontal in-situ stress difference rate depth sequence of each grating line at each acquisition time and the obtaining of the maximum horizontal in-situ stress depth correlation coefficient of the maximum horizontal in-situ stress depth sequence include: Perform a first-order difference on the horizontal in-situ stress difference depth sequence to obtain the horizontal in-situ stress difference rate depth sequence of each grating line at each acquisition time; Calculate the Pearson correlation coefficient between the horizontal in-situ stress difference rate depth sequence of each grating line at each acquisition time and the sequence composed of the subscript values of each data in the horizontal in-situ stress difference rate depth sequence as the difference rate depth correlation coefficient of the horizontal in-situ stress difference rate depth sequence of each grating line at each acquisition time, and record the opposite number of the difference rate depth correlation coefficient as the difference sequence anomaly value; Based on the maximum horizontal in-situ stress depth sequence, use the same calculation method as the difference rate depth correlation coefficient to obtain the maximum horizontal in-situ stress depth correlation coefficient of the maximum horizontal in-situ stress depth sequence of each grating line at each acquisition time.

7. The real-time in-situ stress monitoring method based on multi-source data fusion according to claim 1, characterized in that The calculation method of the data fusion weight at each acquisition time of each sensor is as follows: Calculate the product of the stress ratio anomaly value at each acquisition time in the in-situ stress ratio sequence of each sensor, the maximum horizontal stress anomaly value at each acquisition time in the maximum horizontal in-situ stress sequence, and the minimum horizontal stress anomaly value at each acquisition time in the minimum horizontal in-situ stress sequence; Take the normalized result of the reciprocal of the product as the weight distribution coefficient of each sensor at each acquisition time. Take the proportion of the weight distribution coefficient of each sensor at each acquisition moment in the weight distribution coefficients of all sensors at each acquisition moment as the data fusion weight of each sensor at each acquisition moment.

8. The real-time in-situ stress monitoring method based on multi-source data fusion according to claim 1, characterized in that, The method for completing real-time monitoring of in-situ stress based on the data fusion weight of each sensor at each acquisition moment is as follows: Calculate the product of the vertical in-situ stress in the original three-dimensional in-situ stress data of each sensor at each acquisition moment and the data fusion weight at each acquisition moment, and take the mean value of the products of all sensors as the vertical in-situ stress monitoring data at each acquisition moment at the in-situ stress monitoring location; For the maximum horizontal in-situ stress and the minimum horizontal in-situ stress, use the same acquisition method for the vertical in-situ stress monitoring data to obtain the maximum horizontal in-situ stress monitoring data and the minimum horizontal in-situ stress monitoring data at each acquisition moment at the in-situ stress monitoring location; Based on the vertical in-situ stress monitoring data, the maximum horizontal in-situ stress monitoring data, and the minimum horizontal in-situ stress monitoring data, monitor the in-situ stress in combination with a preset threshold.

Images