Open pit coal mine slope monitoring method based on Beidou GNSS

Abstract

The invention discloses a Beidou GNSS-based open pit coal mine slope monitoring method. The method comprises the steps of determining a target data length value of an open pit coal mine slope; acquiring an initial data set corresponding to the target data length value in the Beidou GNSS of the open pit coal mine slope; determining a data feature set corresponding to the initial data set, and determining the data feature set as a target data set; training the initial monitoring model based on the target data set to obtain a target monitoring model; obtaining current monitoring data of the open pit coal mine slope, and determining a current data characteristic value corresponding to the current monitoring data; and inputting the current data characteristic value into the target monitoring model to obtain a current state classification result of the open pit coal mine slope. According to the method, the target monitoring model learns the association relationship in the open-pit coal mine slope monitoring data, so that the open-pit coal mine slope monitoring data is automatically judged and classified based on the target monitoring model, and the accuracy of the monitoring result of the open-pit coal mine slope is improved.

Description

Technical Field

[0001] This invention relates to the field of coal mine monitoring technology, and in particular to a monitoring method and system for open-pit coal mine slopes based on BeiDou GNSS. Background Technology

[0002] Currently, slope landslides are characterized by their suddenness, complexity, and unpredictability, making them one of the major safety accidents in open-pit mines. Slope landslides can be caused by a variety of factors, including geological factors (such as dip angle, fault structure, and stratigraphic structure), slope engineering factors (such as height and slope angle), the physical and mechanical characteristics of the soil and rock, and other factors (such as groundwater, blasting shocks, and earthquakes), all of which play a dominant role in landslide accidents. Furthermore, the slopes are constantly changing during open-pit mining, making their stress state and deformation patterns even more complex. Therefore, how to monitor open-pit slope safety accidents for timely warnings and reduce such accidents is an urgent problem to be solved.

[0003] In existing technologies, open-pit mines can be monitored using ground-based slope radar, GNSS monitoring, deep displacement monitoring, and InSAR monitoring to prevent slope disasters. However, these existing technologies lack in-depth analysis and application of open-pit coal mine slope monitoring data, which reduces the accuracy of the monitoring results. Summary of the Invention

[0004] The present invention aims to at least partially solve one of the technical problems in the related art.

[0005] To address this, this invention proposes a monitoring method and system for open-pit coal mine slopes based on BeiDou GNSS. The method identifies the data feature set corresponding to the initial dataset containing the target data length value in BeiDou GNSS for the open-pit coal mine slope as the target dataset. The initial monitoring model is then trained based on this target dataset to obtain the target monitoring model. This model learns the correlations within the open-pit coal mine slope monitoring data, enabling automatic identification and classification of the monitoring data and improving the accuracy of the monitoring results.

[0006] To achieve the above objectives, this invention proposes a method for monitoring open-pit coal mine slopes based on BeiDou GNSS, the method comprising:

[0007] Determine the target data length value for open-pit coal mine slopes;

[0008] Obtain the initial dataset corresponding to the target data length value in BeiDou GNSS of the open-pit coal mine slope;

[0009] Determine the data feature set corresponding to the initial dataset, and define the data feature set as the target dataset;

[0010] The initial monitoring model is trained based on the target dataset to obtain the target monitoring model;

[0011] Obtain the current monitoring data of the open-pit coal mine slope and determine the current data feature value corresponding to the current monitoring data;

[0012] The current data feature values ​​are input into the target monitoring model to obtain the current state classification result of the open-pit coal mine slope.

[0013] The monitoring method for open-pit coal mine slopes based on BeiDou GNSS in this invention may also have the following additional technical features:

[0014] In one embodiment of the present invention, determining the target data length value of an open-pit coal mine slope includes:

[0015] Obtain historical monitoring data of the open-pit coal mine slope;

[0016] Based on the historical monitoring data, the target anomaly probability of the open-pit coal mine slope is determined;

[0017] Determine the anomaly probability corresponding to the length of each candidate data;

[0018] The candidate data length corresponding to the anomaly probability with the smallest difference from the target anomaly probability is determined as the target data length value of the open-pit coal mine slope.

[0019] In one embodiment of the present invention, determining the data feature set corresponding to the initial dataset and determining the data feature set as the target dataset includes:

[0020] Based on the initial dataset, at least one data feature value corresponding to different times and directions is determined by calculation;

[0021] At least one data feature value corresponding to different directions at different times is determined as the data feature set corresponding to the initial dataset, and the data feature set is determined as the target dataset.

[0022] In one embodiment of the present invention, the at least one data feature value includes: maximum value, minimum value, mean, mode, median, skewness, kurtosis, starting value, ending value, data volume within the error range of the maximum value, and data volume within the error range of the mode.

[0023] In one embodiment of the present invention, training the initial monitoring model based on the target dataset to obtain the target monitoring model includes:

[0024] Determine the target state classification result corresponding to each time step in the target dataset;

[0025] The target dataset is labeled based on the target state classification results at each time point, and the labeled target dataset is determined as the training dataset.

[0026] The data feature values ​​from the training dataset are input into the initial monitoring model to obtain the corresponding predicted state classification results;

[0027] The predicted state classification result and the corresponding target state classification result are input into the loss function to obtain the corresponding loss value;

[0028] The network parameters in the initial monitoring model are updated using the loss value until the initial monitoring model converges, or the target monitoring model is obtained when the number of iterations of the network parameters reaches a preset number.

[0029] In one embodiment of the present invention, determining the target state classification result corresponding to each time step in the target dataset includes:

[0030] Determine whether at least one data feature value in the target dataset at different directions at the target time meets the stability discrimination condition;

[0031] If at least one data feature value in different directions at the target time meets the stability discrimination condition, then the target state classification result at the target time is determined to be stable; otherwise, the target state classification result at the target time is determined to be risky.

[0032] In one embodiment of the present invention, determining whether at least one data feature value in the target dataset at different directions at a target time meets the stability discrimination condition includes:

[0033] Determine whether at least one data feature value in the target dataset at different directions at the target time meets the first discrimination condition, the second discrimination condition, or the third discrimination condition;

[0034] If at least one data feature value in different directions at the target time meets the first discrimination condition, the second discrimination condition, or the third discrimination condition, then it is determined that at least one data feature value in different directions at the target time meets the stable discrimination condition; otherwise, it is determined that at least one data feature value in different directions at the target time does not meet the stable discrimination condition.

[0035] In one embodiment of the present invention, the first discrimination condition includes:

[0036] X max -X min≤S

[0037] |Z-μ|<0.2μ

[0038] |γ1|<1

[0039] |β²⁻³| < 0.5

[0040] Among them, X max X is the maximum value. min Z is the minimum value, Z is the median value, μ is the mean value, γ1 is the skewness, β2 is the kurtosis, and S is the error threshold for the open-pit coal mine slope stability monitoring data.

[0041] In one embodiment of the present invention, the second discrimination condition includes:

[0042] X max -X min >S

[0043] |X max -X z |

[0044] |γ1|>1

[0045] |β²⁻³|>0.5

[0046] |μ-Z|≤0.1(X max -X min )

[0047] TS / N > 2.7%

[0048] Among them, X z Z is the termination value, Z is the median value, TS is the amount of data within the maximum error range, and N is the amount of monitored data.

[0049] To achieve the above objectives, another aspect of the present invention proposes a monitoring system for open-pit coal mine slopes based on BeiDou GNSS, the system comprising:

[0050] The first determining module is used to determine the target data length value of the open-pit coal mine slope;

[0051] The first acquisition module is used to acquire the initial dataset corresponding to the target data length value in the BeiDou GNSS of the open-pit coal mine slope;

[0052] The second determining module is used to determine the data feature set corresponding to the initial dataset and determine the data feature set as the target dataset.

[0053] The training module is used to train the initial monitoring model based on the target dataset to obtain the target monitoring model;

[0054] ​The second acquisition module is used to acquire the current monitoring data of the open-pit coal mine slope and determine the current data feature value corresponding to the current monitoring data;

[0055] The monitoring module is used to input the current data feature values ​​into the target monitoring model to obtain the current state classification result of the open-pit coal mine slope.

[0056] The monitoring method and system for open-pit coal mine slopes based on BeiDou GNSS of this invention can determine the data feature set corresponding to the initial dataset corresponding to the target data length value in BeiDou GNSS of open-pit coal mine slopes as the target dataset, and train the initial monitoring model based on the target dataset to obtain the target monitoring model. This allows the target monitoring model to learn the correlation relationships in the monitoring data of open-pit coal mine slopes, thereby enabling automatic discrimination and classification of the monitoring data of open-pit coal mine slopes based on the target monitoring model, and improving the accuracy of the monitoring results of open-pit coal mine slopes.

[0057] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0058] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0059] Figure 1 This is a flowchart illustrating a method for monitoring open-pit coal mine slopes based on BeiDou GNSS according to an embodiment of the present invention.

[0060] Figure 2 This is a schematic diagram of a stable type according to an embodiment of the present invention;

[0061] Figure 3 This is a schematic diagram of a monitoring system for open-pit coal mine slopes based on BeiDou GNSS, according to an embodiment of the present invention. Detailed Implementation

[0062] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0063] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0064] The following describes, with reference to the accompanying drawings, a monitoring method and system for open-pit coal mine slopes based on BeiDou GNSS, according to an embodiment of the present invention.

[0065] Figure 1 This is a flowchart illustrating the monitoring method for open-pit coal mine slopes based on BeiDou GNSS according to an embodiment of the present invention.

[0066] like Figure 1 As shown, the method may include the following steps:

[0067] Step 101: Determine the target data length value of the open-pit coal mine slope.

[0068] In one embodiment of the present invention, the monitoring data of BeiDou GNSS can be identified as the monitoring data of open-pit coal mine slope.

[0069] In one embodiment of the present invention, the feedback cycle of open-pit coal mine monitoring data can be 1 hour, and 24 sets of valid data can be acquired every day. Based on this, the data length used for subsequent analysis needs to be determined comprehensively according to the deformation state of the slope and the analysis requirements, so as to determine the optimal data length of the monitoring data, establish a suitable data feature set, make the data length closer to the real data, and thus make the subsequent analysis results more accurate.

[0070] In one embodiment of the present invention, the aforementioned monitoring data can be three columns of data: dis_x, dis_y, and dis_z. Dis_x represents the horizontal displacement in the X direction, with north as positive; dis_y represents the horizontal displacement in the Y direction, with east as positive; and dis_z represents the vertical displacement in the Z direction, with upward as positive. Furthermore, in one embodiment of the present invention, the total planar displacement dis can be determined from the dis_x and dis_y data, such as... Therefore, in one embodiment of the present invention, the directions corresponding to the above-mentioned monitoring data may include the planar direction, the X-direction horizontal displacement, the Y-direction horizontal displacement, and the Z-direction vertical displacement. It should be noted that, in one embodiment of the present invention, the positive and negative values ​​of the referenced data only represent the displacement direction.

[0071] In one embodiment of the present invention, the method for determining the target data length value of an open-pit coal mine slope may include the following steps:

[0072] Step 1011: Obtain historical monitoring data of open-pit coal mine slopes;

[0073] Step 1012: Determine the target anomaly probability of the open-pit coal mine slope based on historical monitoring data;

[0074] Step 1013: Determine the anomaly probability corresponding to the length of each candidate data;

[0075] Step 1014: The candidate data length corresponding to the anomaly probability with the smallest difference from the target anomaly probability is determined as the target data length value for the open-pit coal mine slope.

[0076] In one embodiment of the present invention, after obtaining historical monitoring data of the open-pit coal mine slope, the average detection value H and standard deviation σ in different directions can be calculated based on the historical monitoring data in different directions, and (H-3σ, H+3σ) is determined as the normal fluctuation range of the monitoring data in different directions. Furthermore, in another embodiment of the present invention, monitoring data outside the normal fluctuation range of the monitoring data in different directions are determined as abnormal data in those directions. The number of abnormal data in a certain direction is divided by the number of monitoring data in that direction to determine the abnormal probability in that direction, and the average of the abnormal probabilities in all directions is determined as the target abnormal probability of the open-pit coal mine slope.

[0077] Furthermore, in one embodiment of the present invention, the maximum value of the candidate data length set and the slope monitoring data can be determined empirically. For example, the candidate data length set is {3, 10, 15, 30, 90}, in days.

[0078] Furthermore, in one embodiment of the present invention, when the slope monitoring data exceeds the maximum value three times consecutively in a certain direction, a warning will be issued; when it exceeds the maximum value four times consecutively, a risk warning will be issued; and when it exceeds the maximum value five times consecutively, an automatic early warning will be issued, which can be identified as abnormal data in that direction. Also, in one embodiment of the present invention, the average probability P of abnormal monitoring data exceeding the maximum value five times consecutively in different directions can be determined as the anomaly probability corresponding to each candidate data length. Table 1 is a statistical table of candidate data length and anomaly discrimination probability proposed in an embodiment of the present invention.

[0079] Table 1

[0080]

[0081] As shown in Table 1, when the candidate data length is 90 days, the corresponding anomaly probability is 0.23%, which is the smallest absolute value of the difference between the candidate data length and the target anomaly probability. Based on this, the candidate data length of 90 days is determined as the target data length value for open-pit coal mine slopes, which can meet the requirements for slope disaster early warning.

[0082] Step 102: Obtain the initial dataset corresponding to the target data length value in BeiDou GNSS for the open-pit coal mine slope.

[0083] In one embodiment of the present invention, after determining the target data length value of the open-pit coal mine slope through the above steps, the initial dataset corresponding to the target data length value in the BeiDou GNSS of the open-pit coal mine slope can be obtained. For example, the monitoring data of the open-pit coal mine slope in the BeiDou GNSS of the previous 90 days can be determined as the initial dataset.

[0084] Step 103: Determine the data feature set corresponding to the initial dataset, and define the data feature set as the target dataset.

[0085] In one embodiment of the present invention, after obtaining the initial dataset through the above steps, the data feature set corresponding to the initial dataset can be determined, and the data feature set can be determined as the target dataset, so that the corresponding slope deformation state can be optimized based on the target dataset.

[0086] In one embodiment of the present invention, the method for determining the data feature set corresponding to the initial dataset and determining the data feature set as the target dataset may include: based on the initial dataset, calculating and determining at least one data feature value corresponding to different directions at different times, determining the at least one data feature value corresponding to different directions at different times as the data feature set corresponding to the initial dataset, and determining the data feature set as the target dataset.

[0087] Furthermore, in one embodiment of the present invention, the aforementioned at least one data feature value may include: maximum value, minimum value, mean, mode, median, skewness, kurtosis, starting value, ending value, data volume within the error range of the maximum value, and data volume within the error range of the mode.

[0088] Furthermore, in one embodiment of the present invention, the aforementioned maximum value X max : The maximum value of the displacement data after cleaning within the preset data length; minimum value X min : Minimum value of displacement data after cleaning within the preset data length; Mean μ: Mean value. N is the initial data size; Mode M: the most frequent value within the preset data length; Median Z: the median value of the monitored data within the preset data length; Skewness γ1: measures the asymmetry of the distribution, γ1=[Σ(xi-μ)^3 / N] / σ^3, where σ 2 σ is the variance, the average of the squared differences between the monitored data and the mean. 2 =Σ(xi-μ) 2 / N; Kurtosis β2: measures the sharpness of the distribution, β2=[Σ(xi-μ)^4 / N] / σ^4; Initial value X q: The first data point within the preset data length; Termination value X z : The final data within the preset data length; the data volume TS within the maximum error range: the deformation amount within the preset data length up to the maximum value [X]. max -S,X max The amount of data within the mode error range TM: the amount of data within the preset time constant where the deformation is between the mode [M-0.5S, M+0.5S].

[0089] In one embodiment of the present invention, the preset data length can be set as needed, such as 5 days; or it can be a target data length value.

[0090] Step 104: Train the initial monitoring model based on the target dataset to obtain the target monitoring model.

[0091] In one embodiment of the present invention, after determining the target dataset through the above steps, the initial monitoring model can be trained based on the target dataset to obtain the target monitoring model.

[0092] Furthermore, in one embodiment of the present invention, the method for training an initial monitoring model based on a target dataset to obtain a target monitoring model may include the following steps:

[0093] Step 1041: Determine the target state classification result corresponding to each time step in the target dataset;

[0094] Step 1042: Label the target dataset based on the target state classification results at each time step, and determine the labeled target dataset as the training dataset;

[0095] Step 1043: Input the data feature values ​​from the training dataset into the initial monitoring model to obtain the corresponding predicted state classification results;

[0096] Step 1044: Input the predicted state classification result and the corresponding target state classification result into the loss function to obtain the corresponding loss value;

[0097] Step 1045: Update the network parameters in the initial monitoring model using the loss value until the initial monitoring model converges, or until the number of iterations of the network parameters reaches a preset number, to obtain the target monitoring model.

[0098] In one embodiment of the present invention, the method for determining the target state classification result at each time step in the target dataset may include the following steps:

[0099] Step a: Determine whether at least one data feature value in the target dataset at different directions at the target time meets the stability discrimination condition;

[0100] Step b: If at least one data feature value in different directions at the target time meets the stability discrimination condition, then the target state classification result at the target time is determined to be stable; otherwise, the target state classification result at the target time is determined to be risky.

[0101] Furthermore, in one embodiment of the present invention, the method for determining whether at least one data feature value in different directions at a target time in the target dataset meets the stability discrimination condition may include: determining whether at least one data feature value in different directions at a target time in the target dataset meets the first discrimination condition, the second discrimination condition, or the third discrimination condition; if it is determined that at least one data feature value in different directions at a target time meets the first discrimination condition, the second discrimination condition, or the third discrimination condition, then it is determined that at least one data feature value in different directions at a target time meets the stability discrimination condition; otherwise, it is determined that at least one data feature value in different directions at a target time does not meet the stability discrimination condition.

[0102] In one embodiment of the present invention, the first discrimination condition may include:

[0103] X max -X min ≤S

[0104] |Z-μ|<0.2μ

[0105] |γ1|<1

[0106] |β²⁻³| < 0.5

[0107] Among them, X max X is the maximum value. min Z is the minimum value, Z is the median value, μ is the mean value, γ1 is the skewness, β2 is the kurtosis, and S is the error threshold for open-pit coal mine slope stability monitoring data.

[0108] Furthermore, in one embodiment of the present invention, the second discrimination condition may include:

[0109] X max -X min >S

[0110] |X max -X z |

[0111] |γ1|>1

[0112] |β²⁻³|>0.5

[0113] |μ-Z|≤0.1(X max -X min )

[0114] ​TS / N > 2.7%

[0115] Among them, X z Z is the termination value, Z is the median value, TS is the amount of data within the maximum error range, and N is the amount of monitored data.

[0116] Furthermore, in one embodiment of the present invention, the third discrimination condition may include:

[0117] X max -X min >S

[0118] |X max -X z |

[0119] |γ1|>1

[0120] |β²⁻³|>0.5

[0121] 0.1(X max -X min )≤|μ-Z|≤0.2(X max -X min )

[0122] TS / N+TM / N>50%

[0123] Among them, X z Z is the termination value, Z is the median, and TM is the amount of data within the mode error range.

[0124] In one embodiment of the present invention, the state of a stable slope can be further subdivided into three states: stable, deformation tending towards stability, and step-stable (e.g.) Figure 2 (As shown in the diagram). The left diagram shows a slope in a stable state where the cumulative displacement remains relatively constant over a certain period. The middle diagram shows a slope that initially deformed, then the deformation leveled out, and the cumulative displacement almost stopped changing. This slope deformed first and then stabilized, with no risk of instability. This is a step-stable slope. The right diagram shows a slope that stabilized after multiple deformations. This is a deformation-to-stabilization slope. All three slope states described above do not pose a landslide risk and can be considered stable.

[0125] In one embodiment of the present invention, if at least one data feature value in a certain direction at the target time meets the first discrimination condition, the state classification result of that direction is determined to be stable; if at least one data feature value in a certain direction at the target time meets the second discrimination condition, the state classification result of that direction is determined to be deformation-to-stable; if at least one data feature value in a certain direction at the target time meets the third discrimination condition, the state classification result of that direction is determined to be step-stable.

[0126] ​Furthermore, in one embodiment of the present invention, if the state classification results in different directions at the target time are all stable, or the deformation tends to be stable, or the step is stable, it indicates that there is no risk of landslide in different directions at the target time, and the target state classification result at the target time is determined to be stable; otherwise, the target state classification result at the target time is determined to be risky.

[0127] Furthermore, in one embodiment of the present invention, after obtaining the target state classification results corresponding to each time step through the above steps, the target dataset can be labeled based on the target state classification results at each time step, and the labeled target dataset can be determined as the training dataset.

[0128] In one embodiment of the present invention, the initial monitoring model described above may be a convolutional network.

[0129] Furthermore, in one embodiment of the present invention, the method for training the initial monitoring model based on the training dataset is the same as that in the prior art, and will not be described in detail here.

[0130] Step 105: Obtain the current monitoring data of the open-pit coal mine slope and determine the current data feature value corresponding to the current monitoring data.

[0131] In one embodiment of the present invention, after obtaining the current monitoring data of the open-pit coal mine slope, the current data feature values ​​corresponding to different directions within the preset data length can be obtained by the method described above.

[0132] Step 106: Input the current data feature values ​​into the target monitoring model to obtain the current state classification results of the open-pit coal mine slope.

[0133] In one embodiment of the present invention, current data feature values ​​from different directions are input into the target monitoring model to obtain the current state classification result of the open-pit coal mine slope, so that it can be determined whether to issue an alarm based on the current state classification result. Specifically, in one embodiment of the present invention, if the current state classification result of the open-pit coal mine slope obtained by inputting current data feature values ​​from different directions into the target monitoring model is stable, then a stable state classification result without alarm can be output; if the current state classification result of the open-pit coal mine slope obtained by inputting current data feature values ​​from different directions into the target monitoring model is risky, then a risky state classification result can be output and an alarm can be issued, so that maintenance personnel can issue timely warnings, thereby effectively preventing the occurrence of open-pit mine slope disasters.

[0134] The monitoring method for open-pit coal mine slopes based on BeiDou GNSS in this invention can determine the data feature set corresponding to the initial dataset of the target data length value in BeiDou GNSS of the open-pit coal mine slope as the target dataset, and train the initial monitoring model based on the target dataset to obtain the target monitoring model. This allows the target monitoring model to learn the correlation in the monitoring data of the open-pit coal mine slope, thereby enabling automatic discrimination and classification of the monitoring data of the open-pit coal mine slope based on the target monitoring model, and improving the accuracy of the monitoring results of the open-pit coal mine slope.

[0135] Figure 3 This is a schematic diagram of the structure of a monitoring system for open-pit coal mine slopes based on BeiDou GNSS, according to an embodiment of the present invention.

[0136] like Figure 3 As shown, the system may include:

[0137] The first determining module 301 is used to determine the target data length value of the open-pit coal mine slope;

[0138] The first acquisition module 302 is used to acquire the initial dataset corresponding to the target data length value in the Beidou GNSS of the open-pit coal mine slope, which is the target state classification result.

[0139] The second determining module 303 is used to determine the data feature set corresponding to the initial dataset of the target state classification result, and to determine the data feature set of the target state classification result as the target dataset.

[0140] Training module 304 is used to train the initial monitoring model based on the target dataset to obtain the target monitoring model;

[0141] The second acquisition module 305 is used to acquire the current monitoring data of the open-pit coal mine slope of the target state classification result, and determine the current data feature value corresponding to the current monitoring data of the target state classification result;

[0142] The monitoring module 306 is used to input the current data feature value of the target state classification result into the target state classification result target monitoring model to obtain the current state classification result of the open-pit coal mine slope.

[0143] In one embodiment of the present invention, the first determining module 301 is specifically used for:

[0144] Obtain historical monitoring data of open-pit coal mine slopes based on target state classification results;

[0145] Based on historical monitoring data of target state classification results, the target anomaly probability of open-pit coal mine slopes is determined.

[0146] Determine the anomaly probability corresponding to the length of each candidate data;

[0147] The candidate data length corresponding to the anomaly probability with the smallest difference between the target anomaly probability and the target state classification result is determined as the target data length value of the open-pit coal mine slope in the target state classification result.

[0148] In one embodiment of the present invention, the second determining module 303 is specifically used for:

[0149] Based on the initial dataset, at least one data feature value corresponding to different directions at different times is determined by calculation;

[0150] At least one data feature value corresponding to different directions at different times is determined as the data feature set corresponding to the initial dataset, and the data feature set is determined as the target dataset.

[0151] In one embodiment of the present invention, the target state classification result includes at least one data feature value, including: maximum value, minimum value, mean, mode, median, skewness, kurtosis, starting value, ending value, data volume within the error range of the maximum value, and data volume within the error range of the mode.

[0152] In one embodiment of the present invention, the training module 304 is specifically used for:

[0153] Determine the target state classification result at each time step in the target dataset;

[0154] The target dataset is labeled based on the target state classification results at each time step, and the labeled target dataset is determined as the training dataset.

[0155] The data feature values ​​from the training dataset are input into the initial monitoring model to obtain the corresponding predicted state classification results;

[0156] Input the predicted state classification result and the corresponding target state classification result into the loss function to obtain the corresponding loss value;

[0157] The network parameters in the initial monitoring model are updated using the loss value until the initial monitoring model converges, or until the number of iterations of the network parameters reaches a preset number, thus obtaining the target monitoring model.

[0158] In one embodiment of the present invention, the training module 304 is further configured to:

[0159] Determine whether at least one data feature value in the target dataset at different times in different directions meets the stability discrimination criteria;

[0160] If at least one data feature value in different directions at the target time meets the stability discrimination condition, then the target state classification result at the target time is determined to be stable; otherwise, the target state classification result at the target time is determined to be risky.

[0161] In one embodiment of the present invention, the training module 304 is further configured to:

[0162] Determine whether at least one data feature value in the target dataset at different directions at the target time meets the first, second, or third discrimination condition;

[0163] If at least one data feature value in different directions at the target time meets the first, second, or third discrimination condition, then at least one data feature value in different directions at the target time meets the stability discrimination condition; otherwise, at least one data feature value in different directions at the target time does not meet the stability discrimination condition.

[0164] In one embodiment of the present invention, the first discrimination condition includes:

[0165] X max -X min ≤S

[0166] |Z-μ|<0.3μ

[0167] |γ1|<1

[0168] |β3-3|<0.5

[0169] Among them, X max X is the maximum value. min Z is the minimum value, Z is the median value, μ is the mean value, γ1 is the skewness, β3 is the kurtosis, and S is the error threshold for open-pit coal mine slope stability monitoring data.

[0170] In one embodiment of the present invention, the second discrimination condition includes:

[0171] X max -X min >S

[0172] |X max -X z |

[0173] |γ1|>1

[0174] |β3-3|>0.5

[0175] |μ-Z|≤0.1(X max -X min )

[0176] TS / N > 3.7%​

[0177] Among them, X z Z is the termination value, Z is the median value, TS is the amount of data within the maximum error range, and N is the amount of monitored data.

[0178] The BeiDou GNSS-based open-pit coal mine slope monitoring system of this invention can determine the data feature set corresponding to the initial dataset of the target data length value in the BeiDou GNSS of the open-pit coal mine slope as the target dataset, and train the initial monitoring model based on the target dataset to obtain the target monitoring model. This allows the target monitoring model to learn the correlation in the open-pit coal mine slope monitoring data, thereby enabling automatic discrimination and classification of the open-pit coal mine slope monitoring data based on the target monitoring model, improving the accuracy of the monitoring results of the open-pit coal mine slope.

[0179] In this specification, the use of terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refers to a specific feature, structure, material, or characteristic described in connection with that embodiment or example, which is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0180] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

Claims

1. A monitoring method for open-pit coal mine slopes based on BeiDou GNSS, characterized in that, The method includes: Determine the target data length value for open-pit coal mine slopes; Obtain the initial dataset corresponding to the target data length value in BeiDou GNSS of the open-pit coal mine slope; Determine the data feature set corresponding to the initial dataset, and define the data feature set as the target dataset; The initial monitoring model is trained based on the target dataset to obtain the target monitoring model; Obtain the current monitoring data of the open-pit coal mine slope and determine the current data feature value corresponding to the current monitoring data; The current data feature values ​​are input into the target monitoring model to obtain the current state classification result of the open-pit coal mine slope.

2. The method according to claim 1, characterized in that, The determination of the target data length value of the open-pit coal mine slope includes: Obtain historical monitoring data of the open-pit coal mine slope; Based on the historical monitoring data, the target anomaly probability of the open-pit coal mine slope is determined; Determine the anomaly probability corresponding to the length of each candidate data; The candidate data length corresponding to the anomaly probability with the smallest difference from the target anomaly probability is determined as the target data length value of the open-pit coal mine slope.

3. The method according to claim 1, characterized in that, The step of determining the data feature set corresponding to the initial dataset and determining the data feature set as the target dataset includes: Based on the initial dataset, at least one data feature value corresponding to different directions at different times is determined by calculation; At least one data feature value corresponding to different directions at different times is determined as the data feature set corresponding to the initial dataset, and the data feature set is determined as the target dataset.

4. The method according to claim 3, characterized in that, The at least one data feature value includes: maximum value, minimum value, mean, mode, median, skewness, kurtosis, starting value, ending value, data volume within the error range of the maximum value, and data volume within the error range of the mode.

5. The method according to claim 1, characterized in that, The step of training the initial monitoring model based on the target dataset to obtain the target monitoring model includes: Determine the target state classification result corresponding to each time step in the target dataset; The target dataset is labeled based on the target state classification results at each time point, and the labeled target dataset is determined as the training dataset. The data feature values ​​from the training dataset are input into the initial monitoring model to obtain the corresponding predicted state classification results; The predicted state classification result and the corresponding target state classification result are input into the loss function to obtain the corresponding loss value; The network parameters in the initial monitoring model are updated using the loss value until the initial monitoring model converges, or the target monitoring model is obtained when the number of iterations of the network parameters reaches a preset number.

6. The method according to claim 5, characterized in that, Determining the target state classification result corresponding to each time step in the target dataset includes: Determine whether at least one data feature value in the target dataset at different directions at the target time meets the stability discrimination condition; If at least one data feature value in different directions at the target time meets the stability discrimination condition, then the target state classification result at the target time is determined to be stable; otherwise, the target state classification result at the target time is determined to be risky.

7. The method according to claim 6, characterized in that, Determining whether at least one data feature value in the target dataset at different directions at the target time meets the stability discrimination condition includes: Determine whether at least one data feature value in the target dataset at different directions at the target time meets the first discrimination condition, the second discrimination condition, or the third discrimination condition; If at least one data feature value in different directions at the target time meets the first discrimination condition, the second discrimination condition, or the third discrimination condition, then it is determined that at least one data feature value in different directions at the target time meets the stable discrimination condition; otherwise, it is determined that at least one data feature value in different directions at the target time does not meet the stable discrimination condition.

8. The method according to claim 7, characterized in that, The first discrimination condition includes: X max -X min ≤S |Z-μ|<0.2μ |γ1|<1 |β2-3|<0.5 Among them, X max X is the maximum value. min Z is the minimum value, Z is the median value, μ is the mean value, γ1 is the skewness, β2 is the kurtosis, and S is the error threshold for the open-pit coal mine slope stability monitoring data.

9. The method according to claim 7, characterized in that, The second discrimination condition includes: X max -X min >S |X max -X z |<S |γ1|>1 |β2-3|＞0.5 |μ-Z|≤0.1(X max -X min ) TS / N > 2.7% Among them, X z Z is the termination value, Z is the median value, TS is the amount of data within the maximum error range, and N is the amount of monitored data.

10. A monitoring system for open-pit coal mine slopes based on BeiDou GNSS, characterized in that, The system includes: The first determining module is used to determine the target data length value of the open-pit coal mine slope; The first acquisition module is used to acquire the initial dataset corresponding to the target data length value in the BeiDou GNSS of the open-pit coal mine slope; The second determining module is used to determine the data feature set corresponding to the initial dataset and determine the data feature set as the target dataset. The training module is used to train the initial monitoring model based on the target dataset to obtain the target monitoring model; The second acquisition module is used to acquire the current monitoring data of the open-pit coal mine slope and determine the current data feature value corresponding to the current monitoring data; The monitoring module is used to input the current data feature values ​​into the target monitoring model to obtain the current state classification result of the open-pit coal mine slope.

Images