Mining mine pressure intelligent analyzing and monitoring method based on machine learning

Abstract

The invention relates to the technical field of data processing, in particular to an intelligent mine pressure analyzing and monitoring method based on machine learning, and the method comprises the steps: collecting the pressure data of any hydraulic support, and obtaining a pressure data sequence; obtaining a plurality of pressure subsequences according to maximum value data and minimum value data in the pressure data sequence; target pressure subsequences are screened, the pressure deviation degree of each target pressure subsequence is obtained, and a pressure deviation degree sequence is formed; performing STL decomposition on the pressure deviation degree sequence to obtain a corresponding trend term sequence, and obtaining a pressure anomaly index according to the trend term sequence; the method comprises the following steps: acquiring pressure anomaly indexes of all hydraulic supports, performing mine pressure anomaly monitoring and early warning according to the pressure anomaly indexes of all the hydraulic supports, taking pressure data between maximum value data and minimum value data as periodic pressure data, and performing mine pressure anomaly monitoring according to the pressure data change amplitude difference of multiple pieces of periodic pressure data. And the abnormity monitoring accuracy is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a method for intelligent analysis and monitoring of mine pressure based on machine learning. Background Art

[0002] Mine pressure refers to the rock pressure generated by ore body destruction, rock deformation and geological pressure during the mining process. It reflects the stress state of the rock layer or surrounding rock in the mine and directly affects the stability and safety of the mine. Hydraulic supports are usually used as supporting structures to offset the pressure of rock production. However, due to the limited bearing capacity of hydraulic supports, abnormal mine pressure may lead to accidents such as mine collapse. Therefore, it is necessary to monitor and analyze the mine pressure in the mine in real time to avoid abnormal mine pressure, provide timely and accurate warnings for the mining process, ensure the safety of miners, reduce the occurrence of dangerous accidents, and improve the safety management level of mines.

[0003] In the prior art, for coal mine pressure monitoring, the pressure cycle data monitored by the hydraulic support during the coal mining cycle is usually obtained, and the pressure cycle data is analyzed using static methods such as fixed thresholds to achieve monitoring and identification of abnormal mine pressure. However, due to the complex and changeable geological conditions underground in coal mines and the different advancement speeds during the coal mining cycle, the pressure data change characteristics monitored by the hydraulic support during different coal mining cycles may be different, resulting in the fixed threshold method being difficult to accurately identify the abnormal trend of mine pressure, resulting in untimely abnormal warnings.

[0004] Therefore, how to improve the accuracy of abnormal mine pressure warning using pressure data monitored by hydraulic supports has become an urgent problem to be solved. Summary of the invention

[0005] In view of this, an embodiment of the present invention provides a method for intelligent analysis and monitoring of mine pressure based on machine learning to solve the problem of how to improve the accuracy of abnormal mine pressure warning using pressure data monitored by hydraulic supports.

[0006] An embodiment of the present invention provides a method for intelligent analysis and monitoring of mine pressure based on machine learning, the method comprising the following steps: In the process of coal mining, the pressure data of any hydraulic support is collected to obtain a pressure data sequence composed of the pressure data at the current sampling moment and each previous historical sampling moment; Acquire the maximum value data and the minimum value data in the pressure data sequence, combine the pressure data between the adjacent maximum value data and the minimum value data into a pressure subsequence, obtain multiple pressure subsequences, and acquire the data change characteristic value of each pressure subsequence; Filter the target pressure subsequence according to the position of each pressure subsequence, obtain the pressure deviation degree of each target pressure subsequence according to the difference in data change characteristic value and sequence time length between each target pressure subsequence and the previous pressure subsequence, and form a pressure deviation degree sequence; Perform STL decomposition on the pressure deviation degree sequence to obtain a corresponding trend item sequence, obtain a first-order difference sequence of the trend item sequence, perform sliding traversal on the first-order difference sequence using a preset sliding window and sliding step size to obtain multiple target sliding windows, and obtain the pressure anomaly index of any hydraulic support at the current sampling time according to the mean of the differential values ​​in each of the target sliding windows; The pressure anomaly indexes of all hydraulic supports at the current sampling time are obtained, and the mine pressure anomaly monitoring and early warning are performed according to the pressure anomaly indexes of all hydraulic supports at the current sampling time.

[0007] Preferably, obtaining the data change characteristic value of each pressure subsequence includes: For any pressure subsequence, first-order difference processing is performed on the pressure subsequence to obtain a difference sequence, and the average value and standard deviation of the difference sequence are obtained. Using an exponential function with a natural constant as the base, the normalized value of the standard deviation and the sum of the normalized values ​​of the standard deviation are mapped to obtain the data change characteristic value of the pressure subsequence.

[0008] Preferably, screening the target pressure subsequence according to the position of each of the pressure subsequences comprises: According to the position of each of the pressure subsequences, the next n pressure subsequences are marked as target pressure subsequences.

[0009] Preferably, obtaining the pressure deviation degree of each target pressure subsequence according to the data change characteristic value difference and sequence time difference between each target pressure subsequence and the previous pressure subsequence includes: For any target pressure subsequence, obtain any pressure subsequence before the any target pressure subsequence as a reference subsequence, obtain the ratio of the data change characteristic value between the any target pressure subsequence and the reference subsequence, recorded as a first ratio, obtain the ratio between the pressure mean of the any target pressure subsequence and the pressure mean of the reference subsequence, recorded as a second ratio, and obtain the relative deviation degree of the any target pressure subsequence compared with the reference subsequence according to the product of the first ratio and the second ratio; Obtaining the absolute value of the difference between the sequence duration of any target pressure subsequence and the sequence duration of the reference subsequence, and using the inverse of the absolute value of the difference as an independent variable of a preset exponential function to obtain a weight factor of the reference subsequence, wherein the sequence duration refers to the duration of the time period corresponding to the pressure subsequence; The relative deviation degree of any target pressure subsequence compared with each reference subsequence and the weight factor of each reference subsequence are obtained, and all relative deviation degrees are weighted summed and averaged to obtain the pressure deviation degree of any target pressure subsequence.

[0010] Preferably, obtaining the pressure anomaly index of any hydraulic support at the current sampling moment according to the mean of the difference values ​​in each target sliding window includes: For any target sliding window, obtain the mean of the differential values ​​in the any target sliding window, obtain the pressure deviation degree associated with each differential value in the any target sliding window in the pressure deviation degree sequence, and obtain the average pressure deviation degree; The mean of the differential values ​​is used as an independent variable of an exponential function with a natural constant as a base to obtain a corresponding exponential function value, and a local pressure anomaly characteristic value of any target sliding window is obtained according to the product of the exponential function value and the average pressure deviation degree; According to the local pressure anomaly characteristic value of each target sliding window, the pressure anomaly index of any hydraulic support at the current sampling moment is obtained.

[0011] Preferably, obtaining the pressure anomaly index of any hydraulic support at the current sampling moment according to the local pressure anomaly characteristic value of each target sliding window includes: The smallest local pressure anomaly characteristic value is selected from the local pressure anomaly characteristic values ​​of all target sliding windows to obtain a reference value, the target sliding window to which the current sampling moment belongs is obtained, recorded as the window to be analyzed, the ratio of the average difference value in the window to be analyzed to the last difference value is calculated to obtain the difference proportion at the current sampling moment, the ratio of the reference value to the local pressure anomaly characteristic value of the window to be analyzed is calculated, recorded as the fourth ratio, the product of the difference proportion and the fourth ratio is normalized to obtain the pressure anomaly index of any hydraulic support at the current sampling moment.

[0012] Preferably, the monitoring and early warning of abnormal mine pressure according to the abnormal pressure indicators of all hydraulic supports at the current sampling time includes: If the pressure abnormality index of any hydraulic support at the current sampling time is greater than the preset abnormality index threshold, then any hydraulic support is determined to be a suspected abnormal hydraulic support, and the number of suspected abnormal hydraulic supports is counted. If the number is 0, then it is determined that the mine pressure at the current sampling time is normal; If the number is 1, the hydraulic supports in the preset neighborhood range of the suspected abnormal hydraulic supports are obtained, recorded as neighborhood hydraulic supports, and the difference between the pressure abnormality index of each neighborhood hydraulic support at the current sampling time and the pressure abnormality index at the previous sampling time is obtained. If there is at least one positive difference, an abnormal warning is issued for the mine pressure at the suspected abnormal hydraulic support; If the number is greater than 1, the neighboring hydraulic supports of each suspected abnormal hydraulic support are obtained respectively. If there is at least one positive number in the difference between the pressure abnormality index at the current sampling moment and the pressure abnormality index at the previous sampling moment for each neighboring hydraulic support of any suspected abnormal hydraulic support, an abnormal warning is issued for the mine pressure at the suspected abnormal hydraulic support.

[0013] Compared with the prior art, the embodiments of the present invention have the following beneficial effects: The present invention uses the pressure data monitored by any hydraulic support as the analysis object, collects the pressure data sequence at the current moment, and forms a pressure subsequence with the pressure data between the maximum value data and the minimum value data in the pressure data sequence, which is used to characterize a periodic sequence in which the pressure changes from the minimum to the maximum, and obtains the data change characteristic value of each pressure subsequence. The larger the data change characteristic value, the greater the pressure change fluctuation in the corresponding period. Then, according to the periodic pressure change difference between different pressure subsequences, the overall relative deviation degree (that is, the pressure deviation degree) of each pressure subsequence compared with the historical pressure data is obtained, and the pressure deviation degree sequence is decomposed into trend items to obtain the trend item sequence. According to the data change characteristics in the trend item sequence, the pressure anomaly index of any hydraulic support at the current sampling moment is obtained, which is used to judge the object of mine pressure anomaly monitoring. The larger the pressure anomaly index, the more likely there is mine pressure anomaly. The difference in the pressure data change amplitude in multiple cycles is used to monitor the mine pressure anomaly, which improves the accuracy of anomaly monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0015] Figure 1This is a method flow chart of a method for intelligent analysis and monitoring of mine pressure based on machine learning provided in Example 1 of the present invention. DETAILED DESCRIPTION

[0016] Embodiments of the present disclosure are described in detail below, and examples of the embodiments are shown in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present disclosure, but should not be understood as limiting the present disclosure.

[0017] It should be noted that the terms "first", "second", etc. in the specification of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. Instead, they are merely examples of devices and methods consistent with some aspects of the present disclosure.

[0018] In order to illustrate the technical solution of the present invention, specific embodiments are provided below for illustration.

[0019] See also Figure 1 , is a method flow chart of a method for intelligent analysis and monitoring of mine pressure based on machine learning provided in Embodiment 1 of the present invention, such as Figure 1 As shown, the method may include: Step S101, in the process of coal mining, pressure data of any hydraulic support is collected to obtain a pressure data sequence composed of pressure data at the current sampling moment and pressure data at each previous historical sampling moment.

[0020] During the mining process of coal mines, the mine pressure in the mine cave may change due to the loosening of rocks, etc. Since the supporting force of the hydraulic support is limited, it is necessary to timely and accurately monitor and warn of the abnormal mine pressure during the mining process of coal mines to ensure the safety of the mine caves during the mining process of coal mines. Therefore, in the embodiment of the present invention, the pressure data collected by the pressure sensor in the hydraulic support during coal mining is used as the analysis object of the abnormal mine pressure to reflect the mine pressure situation in the mine cave in real time.

[0021] During the coal mining process, multiple hydraulic supports are set up. In the embodiment of the present invention, one of the hydraulic supports is taken as an example to analyze the pressure data collected under the hydraulic support. The pressure data analysis of other hydraulic supports is the same. In the coal mining process, the current sampling time t is used to represent the real-time collection, and the pressure data of the hydraulic support at the current time and the pressure data of each historical sampling time before the current sampling time are obtained to form a pressure data sequence at the current sampling time, which is used to analyze whether the pressure borne by the hydraulic support has undergone a drastic change due to abnormal mine pressure.

[0022] At this point, the pressure data sequence at the current sampling moment is obtained.

[0023] Step S102, obtaining the maximum value data and the minimum value data in the pressure data sequence, combining the pressure data between adjacent maximum value data and minimum value data into a pressure subsequence, obtaining multiple pressure subsequences, and obtaining the data change characteristic value of each of the pressure subsequences.

[0024] Taking into account the complex and changeable geological conditions underground in the coal mine during coal mining, and the influence of factors such as the different advancement speeds of each coal mining process, resulting in certain differences in the changes in the pressure data of the hydraulic support collected during each coal mining process. Therefore, the pressure data sequence at the current sampling moment is divided into periods, and one period represents the pressure data changes within a period of time. Specifically, the derivative method is used to obtain the maximum data and minimum data in the pressure data sequence, and all the pressure data between an adjacent maximum data and a minimum data are combined into a pressure subsequence, so as to obtain multiple pressure subsequences in the pressure data, and one pressure subsequence represents a period in which the pressure changes from the minimum value to the maximum value.

[0025] During the coal mining process, the rate of change of the pressure data of the hydraulic support reflects the response speed and stability of the support system during the coal mining process. A fast rate of change of the pressure data may indicate poor geological conditions or unstable mining operations, resulting in dramatic changes in the force of the hydraulic support, increasing the risk of the hydraulic support's support timeliness; while a slow rate of change of the pressure data indicates that the support system works relatively smoothly, which helps to maintain the stability and safety of the excavation surface. However, the coal mining process in different time periods has different times, and the pressure size and changes felt by the hydraulic support during the time may also be different, resulting in different pressure data change characteristics in each cycle of the hydraulic support monitoring. Therefore, in order to more intuitively reflect the pressure data change characteristics in each cycle, the data change characteristic values ​​of each pressure subsequence are obtained separately, and the specific acquisition method is as follows: For any pressure subsequence, first-order difference processing is performed on the pressure subsequence to obtain a difference sequence, and the average value and standard deviation of the difference sequence are obtained. Using an exponential function with a natural constant as the base, the normalized value of the standard deviation and the sum of the normalized values ​​of the standard deviation are mapped to obtain the data change characteristic value of the pressure subsequence.

[0026] In one implementation, taking the i-th pressure subsequence as an example, the calculation expression of the data change characteristic value of the i-th pressure subsequence is:

[0027] in, represents the data change characteristic value of the i-th pressure subsequence, represents an exponential function with a natural constant as base, represents the normalization function, represents the average value of the difference sequence of the i-th pressure subsequence, Represents the standard deviation of the difference sequence of the i-th pressure subsequence.

[0028] It should be noted that the first-order difference processing and the acquisition of the average value and the standard deviation are all existing technologies and will not be described in detail here. Characterizes the overall pressure change direction and magnitude of the i-th pressure subsequence, The larger the value of , the greater the pressure increase rate of the i-th pressure subsequence, and the greater the probability of abnormal pressure change on the hydraulic support in the corresponding time period; Characterizes the stability of the change of the i-th pressure subsequence, The larger the value of is, the worse the pressure stability in the corresponding period is, and the greater the probability of abnormal pressure change on the hydraulic support is. The sum is used as the independent variable of the exponential function with the natural constant as the base, so as to avoid the data change characteristic value of the pressure subsequence analyzed in the pressure data sequence monitored by the hydraulic support being 0, which affects the subsequent calculations.

[0029] At this point, the data change characteristic value of each pressure subsequence can be obtained.

[0030] Step S103, filtering the target pressure subsequence according to the position of each pressure subsequence, obtaining the pressure deviation degree of each target pressure subsequence according to the difference in data change characteristic values ​​and sequence duration between each target pressure subsequence and the previous pressure subsequence, and forming a pressure deviation degree sequence.

[0031] The greater the difference between the data change characteristic values ​​of each pressure subsequence, the greater the difference in pressure change on the hydraulic support, and the more likely there is danger in the coal mining process at the current sampling moment. Considering that the probability of danger at the beginning of coal mining is very small and can be ignored, the first few pressure subsequences obtained at the current sampling moment can be assumed to be free of abnormalities, and no subsequent pressure deviation analysis is required. Therefore, in an embodiment of the present invention, the last n pressure subsequences are marked as target pressure subsequences according to the position of each of the pressure subsequences. Preferably, n is 10. In another embodiment, each pressure subsequence after the first 5 pressure subsequences can be marked as a target pressure subsequence. It is worth noting that the target pressure subsequence refers to the pressure subsequence with the highest probability of pressure abnormality.

[0032] After obtaining the target pressure subsequence, the overall relative deviation degree of each target pressure subsequence compared with the historical pressure data (that is, the pressure subsequence before each target pressure subsequence) is obtained according to the data change characteristic value difference and sequence time difference between each target pressure subsequence and the historical pressure data (that is, the pressure subsequence before each target pressure subsequence). That is, the pressure deviation degree is obtained by: For any target pressure subsequence, obtain any pressure subsequence before the any target pressure subsequence as a reference subsequence, obtain the ratio of the data change characteristic value between the any target pressure subsequence and the reference subsequence, recorded as a first ratio, obtain the ratio between the pressure mean of the any target pressure subsequence and the pressure mean of the reference subsequence, recorded as a second ratio, and obtain the relative deviation degree of the any target pressure subsequence compared with the reference subsequence according to the product of the first ratio and the second ratio; Obtaining the absolute value of the difference between the sequence duration of any target pressure subsequence and the sequence duration of the reference subsequence, and using the inverse of the absolute value of the difference as an independent variable of a preset exponential function to obtain a weight factor of the reference subsequence, wherein the sequence duration refers to the duration of the time period corresponding to the pressure subsequence; The relative deviation degree of any target pressure subsequence compared with each reference subsequence and the weight factor of each reference subsequence are obtained, and all relative deviation degrees are weighted summed and averaged to obtain the pressure deviation degree of any target pressure subsequence.

[0033] In one embodiment, the Take the target pressure subsequence as an example, and obtain the first All pressure subsequences before the target pressure subsequence are taken as reference subsequences. Take the reference subsequence as an example and get the The mean pressure value of the target pressure subsequence and the duration of the corresponding time period (that is, the sequence duration) are obtained in the same way. The pressure mean of the reference subsequence and the duration of the corresponding time period (that is, the sequence duration), and then according to the The reference subsequence and The data change characteristic value difference, sequence length difference and pressure mean difference between the target pressure subsequences are obtained. The pressure deviation degree of a target pressure subsequence compared with all reference subsequences is calculated as follows:

[0034] in, Indicates The pressure deviation degree of the target pressure subsequence compared with all reference subsequences, L represents the The number of reference subsequences for each target pressure subsequence, Indicates The data change characteristic value of the target pressure subsequence, Indicates The mean pressure of the target pressure subsequences, Indicates The sequence duration of the target pressure subsequence, Indicates The data variation characteristic value of the reference subsequence, Indicates The pressure mean of the reference subsequences, Indicates The sequence duration of the reference subsequences, represents a natural constant, and || represents the absolute value symbol.

[0035] It should be noted that Indicates The target pressure subsequence and the The first ratio of the data change characteristic values ​​of the reference subsequences. The larger the first ratio, the greater the The greater the change of the target pressure subsequence compared with the sth reference subsequence, Indicates The second ratio of the pressure mean of the sth target pressure subsequence to the sth reference subsequence. The larger the second ratio, the greater the pressure mean of the sth reference subsequence. The target pressure subsequence is compared with the The greater the load on the reference subsequence; Indicates The target pressure subsequence and the The smaller the absolute value of the difference between the sequence lengths of the two reference subsequences, the closer the time lengths of the two pressure subsequences are, and the greater the degree of mutual reference between the pressure data between the two. As the The weight factor when calculating between the target pressure subsequence and the sth reference subsequence, Indicates The relative deviation between the target pressure subsequence and the sth reference subsequence is then used to obtain the mean value obtained by weighted summation of the relative deviations of all reference subsequences to represent the sth target pressure subsequence. The overall relative deviation degree of a target pressure subsequence compared to the overall historical pressure data is also called the pressure deviation degree.

[0036] Similarly, the pressure deviation degree of each target pressure subsequence compared with all reference subsequences is obtained to form a pressure deviation degree sequence, which is used to analyze the pressure variation law between target pressure subsequences, that is, the pressure difference variation law between each cycle.

[0037] Step S104, perform STL decomposition on the pressure deviation degree sequence to obtain a corresponding trend item sequence, obtain a first-order difference sequence of the trend item sequence, use a preset sliding window and sliding step to perform sliding traversal on the first-order difference sequence to obtain multiple target sliding windows, and obtain the pressure anomaly index of any hydraulic support at the current sampling moment according to the mean of the differential values ​​in each of the target sliding windows.

[0038] When the pressure load of the rock formation is stable, the corresponding pressure deviation degree sequence is close to a stable horizontal straight line in overall change; however, when the actual coal mining work is carried out, the rock formation may be damaged, and the corresponding load of the hydraulic support may have a more obvious change trend with the continuous advancement of each coal mining period. Therefore, in order to understand the overall load change law of the hydraulic support in the monitoring time series, the pressure deviation degree sequence is further decomposed by STL to obtain the corresponding trend item sequence, where STL decomposition belongs to the existing technology and will not be repeated here.

[0039] Furthermore, the trend item sequence is subjected to first-order difference processing to obtain the corresponding first-order difference sequence, which is used to facilitate the subsequent analysis of characteristics such as the rate of change of the pressure deviation degree. Therefore, in an embodiment of the present invention, a sliding window with a sliding step size of 1 and a window length of 5 is set, and the sliding window is used to slide on the first-order difference sequence. The sliding window after each sliding is used as the target sliding window to obtain multiple target sliding windows. The first-order difference sequence is divided to analyze the local rate of change of the pressure deviation degree sequence, and then the mean of the differential values ​​corresponding to each target sliding window is calculated according to all the differential values ​​contained in each target sliding window, and then the pressure anomaly index of the hydraulic support at the current sampling time is obtained according to the mean of the differential values ​​in each target sliding window. The specific process is as follows: For any target sliding window, obtain the mean of the differential values ​​in the any target sliding window, obtain the pressure deviation degree associated with each differential value in the any target sliding window in the pressure deviation degree sequence, and obtain the average pressure deviation degree; The mean of the differential values ​​is used as an independent variable of an exponential function with a natural constant as a base to obtain a corresponding exponential function value, and a local pressure anomaly characteristic value of any target sliding window is obtained according to the product of the exponential function value and the average pressure deviation degree; According to the local pressure anomaly characteristic value of each target sliding window, the pressure anomaly index of any hydraulic support at the current sampling moment is obtained.

[0040] In one embodiment, the Take the target sliding window as an example, and get the The mean of the difference values ​​contained in the target sliding window is denoted as , in the pressure deviation degree sequence, obtain the The pressure deviation degree associated with each differential value in the target sliding window is obtained to obtain the average pressure deviation degree, which is recorded as , then according to and Get the The calculation expression of the local pressure anomaly characteristic value of a target sliding window is:

[0041] in, Indicates The local pressure anomaly characteristic value of the target sliding window, e represents the natural constant.

[0042] It should be noted that The smaller the value of The smaller the change of the pressure deviation degree subsequence corresponding to the target sliding window is, the more stable the load data of the hydraulic support will be during the cyclic coal mining process, and the smaller the probability of abnormal mine pressure will be. The smaller the value is, the closer the overall change of the load data of the hydraulic support in similar coal mining processes is. The greater the probability that the pressure data in the time period corresponding to the target sliding window is normal data.

[0043] Similarly, the local pressure anomaly characteristic value of each target sliding window is obtained, and then the pressure anomaly index of the hydraulic support at the current sampling time is obtained according to the local pressure anomaly characteristic value of each target sliding window. Specifically: The smallest local pressure anomaly characteristic value is selected from the local pressure anomaly characteristic values ​​of all target sliding windows to obtain a reference value, the target sliding window to which the current sampling moment belongs is obtained, recorded as the window to be analyzed, the ratio of the mean difference value in the window to be analyzed to the last difference value is calculated to obtain the difference proportion at the current sampling moment, the ratio of the reference value to the local pressure anomaly characteristic value of the window to be analyzed is calculated, recorded as the fourth ratio, the product of the difference proportion and the fourth ratio is normalized to obtain the pressure anomaly index of any hydraulic support at the current sampling moment.

[0044] In one embodiment, the minimum local pressure anomaly characteristic value is used as the reference value of the hydraulic support at the current sampling moment, and the target sliding window involved in the current sampling moment is recorded as the window to be analyzed, that is, the last target sliding window, which is used to detect whether the pressure of the hydraulic support at the current sampling moment is abnormal. In this embodiment of the present invention, the last differential value in the window to be analyzed is recorded as the pressure change difference value characterizing the current sampling moment. Therefore, according to the difference between the last differential value in the window to be analyzed and all differential values ​​and the difference between the local pressure anomaly characteristic value and the reference value, the pressure anomaly index of the hydraulic support at the current sampling moment is obtained, and the calculation expression of the pressure anomaly index is:

[0045] in, Indicates the abnormal pressure index of the hydraulic support at the current sampling time. represents the normalization function, It represents the local pressure anomaly characteristic value of the window to be analyzed, that is, the local pressure anomaly characteristic value of the target sliding window involved in the current sampling moment. represents the reference value, which is also the minimum characteristic value of local pressure anomaly. Represents the last difference value in the window to be analyzed, Represents the mean of the difference values ​​in the window to be analyzed.

[0046] It should be noted that The larger the value is, the more different the local pressure anomaly characteristic value of the window to be analyzed is from the reference value. The more abnormal the pressure monitored by the hydraulic support at the current sampling time is, the greater the pressure anomaly index of the corresponding hydraulic support at the current sampling time is. It is used to characterize the ratio of the pressure difference at the current sampling moment to the mean pressure difference of the entire window. The larger the ratio, the greater the possibility that the pressure data monitored at the current sampling moment is abnormal data, and the greater the pressure anomaly index of the corresponding hydraulic support at the current sampling moment.

[0047] At this point, the pressure anomaly index of a hydraulic support in the coal mining process at the current sampling time is obtained.

[0048] Step S105, obtaining the pressure anomaly indexes of all hydraulic supports at the current sampling time, and performing mine pressure anomaly monitoring and early warning according to the pressure anomaly indexes of all hydraulic supports at the current sampling time.

[0049] According to the method of step S101 to step S104, the pressure abnormality index of each hydraulic support at the current sampling time is respectively set, and the abnormal index threshold is set to 0.35. If the pressure abnormality index of any hydraulic support at the current sampling time is greater than the preset abnormal index threshold, then any hydraulic support is determined to be a suspected abnormal hydraulic support.

[0050] Further, the number of suspected abnormal hydraulic supports is counted. If the number is 0, it is determined that the mine pressure at the current sampling time is normal and no abnormal warning is required; on the contrary, if the number is not 0, it is necessary to determine whether the suspected abnormal hydraulic support is caused by abnormal mine pressure or by problems with the hydraulic support itself. The specific judgment method is: If the number is 1, the hydraulic supports in the preset neighborhood range of the suspected abnormal hydraulic supports are obtained, recorded as neighborhood hydraulic supports, and the difference between the pressure abnormality index of each neighborhood hydraulic support at the current sampling time and the pressure abnormality index at the previous sampling time is obtained. If there is at least one positive difference, an abnormal warning is issued for the mine pressure at the suspected abnormal hydraulic support; If the number is greater than 1, the neighboring hydraulic supports of each suspected abnormal hydraulic support are obtained respectively. If there is at least one positive number in the difference between the pressure abnormality index at the current sampling moment and the pressure abnormality index at the previous sampling moment for each neighboring hydraulic support of any suspected abnormal hydraulic support, an abnormal warning is issued for the mine pressure at the suspected abnormal hydraulic support.

[0051] In one embodiment, the preset neighborhood range is set to include the range of 6 hydraulic supports closest to the suspected abnormal hydraulic support. When there is one suspected abnormal hydraulic support, and the pressure abnormality index of at least one of its neighboring hydraulic supports at the current sampling time is greater than the pressure abnormality index at the previous sampling time, it means that the pressure at the suspected abnormal hydraulic support may be abnormal, and an abnormal mine pressure warning is required; otherwise, it means that the pressure abnormality of the suspected abnormal hydraulic support is due to its own monitoring error or performance problem, and the hydraulic support needs to be maintained and replaced in time. When there are multiple suspected abnormal hydraulic supports, if the pressure abnormality index of at least one of the neighboring hydraulic supports of any suspected abnormal hydraulic support at the current sampling time is greater than the pressure abnormality index at the previous sampling time, it means that the pressure at the suspected abnormal hydraulic support may be abnormal, and an abnormal mine pressure warning is required; otherwise, it means that there are quality problems between the hydraulic supports.

[0052] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.

Claims

1. A mine pressure intelligent analysis and monitoring method based on machine learning, characterized in that: The method comprises: In the process of coal mining, the pressure data of any hydraulic support is collected to obtain a pressure data sequence composed of the pressure data at the current sampling moment and each previous historical sampling moment; Acquire the maximum value data and the minimum value data in the pressure data sequence, combine the pressure data between the adjacent maximum value data and the minimum value data into a pressure subsequence, obtain multiple pressure subsequences, and acquire the data change characteristic value of each pressure subsequence; Filter the target pressure subsequence according to the position of each pressure subsequence, obtain the pressure deviation degree of each target pressure subsequence according to the difference in data change characteristic value and sequence time length between each target pressure subsequence and the previous pressure subsequence, and form a pressure deviation degree sequence; Perform STL decomposition on the pressure deviation degree sequence to obtain a corresponding trend item sequence, obtain a first-order difference sequence of the trend item sequence, perform sliding traversal on the first-order difference sequence using a preset sliding window and sliding step size to obtain multiple target sliding windows, and obtain the pressure anomaly index of any hydraulic support at the current sampling time according to the mean of the differential values ​​in each of the target sliding windows; The pressure anomaly indexes of all hydraulic supports at the current sampling time are obtained, and the mine pressure anomaly monitoring and early warning are performed according to the pressure anomaly indexes of all hydraulic supports at the current sampling time.

2. The method for intelligent analysis and monitoring of mine pressure based on machine learning according to claim 1 is characterized in that: The obtaining of the data change characteristic value of each pressure subsequence includes: For any pressure subsequence, first-order difference processing is performed on the pressure subsequence to obtain a difference sequence, and the average value and standard deviation of the difference sequence are obtained. Using an exponential function with a natural constant as the base, the normalized value of the standard deviation and the sum of the normalized values ​​of the standard deviation are mapped to obtain the data change characteristic value of the pressure subsequence.

3. The method for intelligent analysis and monitoring of mine pressure based on machine learning according to claim 1 is characterized in that: The step of screening the target pressure subsequence according to the position of each of the pressure subsequences comprises: According to the position of each of the pressure subsequences, the next n pressure subsequences are marked as target pressure subsequences.

4. The method for intelligent analysis and monitoring of mine pressure based on machine learning according to claim 1 is characterized in that: Obtaining the pressure deviation degree of each target pressure subsequence according to the data change characteristic value difference and sequence time difference between each target pressure subsequence and the previous pressure subsequence includes: For any target pressure subsequence, obtain any pressure subsequence before the any target pressure subsequence as a reference subsequence, obtain the ratio of the data change characteristic value between the any target pressure subsequence and the reference subsequence, recorded as a first ratio, obtain the ratio between the pressure mean of the any target pressure subsequence and the pressure mean of the reference subsequence, recorded as a second ratio, and obtain the relative deviation degree of the any target pressure subsequence compared with the reference subsequence according to the product of the first ratio and the second ratio; Obtaining the absolute value of the difference between the sequence duration of any target pressure subsequence and the sequence duration of the reference subsequence, and using the inverse of the absolute value of the difference as an independent variable of a preset exponential function to obtain a weight factor of the reference subsequence, wherein the sequence duration refers to the duration of the time period corresponding to the pressure subsequence; The relative deviation degree of any target pressure subsequence compared with each reference subsequence and the weight factor of each reference subsequence are obtained, and all relative deviation degrees are weighted summed and averaged to obtain the pressure deviation degree of any target pressure subsequence.

5. The method for intelligent analysis and monitoring of mine pressure based on machine learning according to claim 1 is characterized in that: The method of obtaining the pressure anomaly index of any hydraulic support at the current sampling time according to the mean of the difference values ​​in each target sliding window includes: For any target sliding window, obtain the mean of the differential values ​​in the any target sliding window, obtain the pressure deviation degree associated with each differential value in the any target sliding window in the pressure deviation degree sequence, and obtain the average pressure deviation degree; The mean of the differential values ​​is used as an independent variable of an exponential function with a natural constant as a base to obtain a corresponding exponential function value, and a local pressure anomaly characteristic value of any target sliding window is obtained according to the product of the exponential function value and the average pressure deviation degree; According to the local pressure anomaly characteristic value of each target sliding window, the pressure anomaly index of any hydraulic support at the current sampling moment is obtained.

6. The method for intelligent analysis and monitoring of mine pressure based on machine learning according to claim 5 is characterized in that: The obtaining of the pressure anomaly index of any hydraulic support at the current sampling time according to the local pressure anomaly characteristic value of each target sliding window includes: The smallest local pressure anomaly characteristic value is selected from the local pressure anomaly characteristic values ​​of all target sliding windows to obtain a reference value, the target sliding window to which the current sampling moment belongs is obtained, recorded as the window to be analyzed, the ratio of the average difference value in the window to be analyzed to the last difference value is calculated to obtain the difference proportion at the current sampling moment, the ratio of the reference value to the local pressure anomaly characteristic value of the window to be analyzed is calculated, recorded as the fourth ratio, the product of the difference proportion and the fourth ratio is normalized to obtain the pressure anomaly index of any hydraulic support at the current sampling moment.

7. The method for intelligent analysis and monitoring of mine pressure based on machine learning according to claim 1 is characterized in that: The abnormal mine pressure monitoring and early warning according to the abnormal pressure indicators of all hydraulic supports at the current sampling time includes: If the pressure abnormality index of any hydraulic support at the current sampling time is greater than the preset abnormality index threshold, then any hydraulic support is determined to be a suspected abnormal hydraulic support, and the number of suspected abnormal hydraulic supports is counted. If the number is 0, then it is determined that the mine pressure at the current sampling time is normal; If the number is 1, the hydraulic supports in the preset neighborhood range of the suspected abnormal hydraulic supports are obtained, recorded as neighborhood hydraulic supports, and the difference between the pressure abnormality index of each neighborhood hydraulic support at the current sampling time and the pressure abnormality index at the previous sampling time is obtained. If there is at least one positive difference, an abnormal warning is issued for the mine pressure at the suspected abnormal hydraulic support; If the number is greater than 1, the neighboring hydraulic supports of each suspected abnormal hydraulic support are obtained respectively. If there is at least one positive number in the difference between the pressure abnormality index at the current sampling moment and the pressure abnormality index at the previous sampling moment for each neighboring hydraulic support of any suspected abnormal hydraulic support, an abnormal warning is issued for the mine pressure at the suspected abnormal hydraulic support.

Images