Water leakage monitoring method for digital twinning system of coal mine shaft

Abstract

A water leakage monitoring method for a digital twinning system of a coal mine shaft comprises the steps that an RFID antenna and a Twinin-Tag sensor are installed on the wall of the shaft, a monitoring terminal is arranged on the ground, and an RFID reader is arranged at an outlet of the shaft; a shaft monitoring signal is collected in real time through a Twindow-Tag sensor, the shaft monitoring signal is read through an RFID reader, and then shaft monitoring data are obtained by analyzing and processing the shaft monitoring signal; collecting RSSI reading and phase data by using a short time window, and performing state identification by using a decision tree; carrying out coarse segmentation and fine segmentation on the leakage signal, and determining the location of a water leakage point in combination with a change point detection algorithm; and the leakage volume and the leakage rate are estimated. According to the method, the water leakage point can be accurately positioned, the volume and flow velocity information of the leaked water in the shaft can be obtained, so that a reliable basis is provided for subsequent maintenance decision making, and the method is suitable for long-term monitoring operation of the coal mine shaft.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent monitoring, and in particular relates to a water leakage monitoring method for a coal mine shaft digital twin system. Background Art

[0002] As the main access passage between the ground and the mine, the safety of the mine shaft is directly related to the normal production work of the coal mine and the personal safety of the personnel, and is the top priority of the entire underground mining project construction. In the long-term production process, there are certain safety hazards in deep shafts, such as structural damage or water leakage. Among them, shaft leakage is one of the common and dangerous problems. Water seepage will corrode the shaft wall structure, weaken its bearing capacity, and even cause the shaft wall to collapse, seriously threatening the lives of underground workers. Therefore, in order to ensure the safety of coal mine operations and effectively avoid the chain reaction caused by water seepage, it is of great practical significance to conduct water leakage monitoring operations on deep shafts of coal mines.

[0003] Driven by digital twin technology, the development of intelligent mines has entered a new stage. Digital twin technology can map and interact the physical entity of the shaft with the virtual model in real time, thereby realizing dynamic monitoring and early warning of the shaft status. However, the digital twin system of the coal mine shaft requires accurate water leakage monitoring data. How to effectively monitor shaft leakage in a complex underground environment has become a key issue in coal mine shaft safety management and digital transformation.

[0004] Traditional sensor-based monitoring technology is easily disturbed in the complex environment of the shaft, and the stability and durability of the sensor equipment are difficult to guarantee. The RFID-based liquid monitoring method provides a new solution. RFID technology has the advantages of non-contact, long-distance identification, and high reliability, especially in harsh industrial environments. It has shown strong adaptability. RFID tags can not only record the status information of the monitored object, but also realize remote data transmission and intelligent analysis through linkage with external systems. Based on this, RFID technology has gradually become one of the important technologies in mine safety monitoring. In particular, the effective combination of RFID technology and water-absorbing materials can realize early warning and detection of water leakage. After absorbing water, the physical properties of the water-absorbing material change. Then, by embedding RFID tags, remote perception of water changes can be realized. In the existing technology, there is a lack of technology that effectively combines RFID tag technology with shaft leakage detection. For this reason, it is urgent to provide a method for shaft leakage monitoring based on RFID tag technology. Summary of the invention

[0005] In response to the problems existing in the above-mentioned prior art, the present invention provides a water leakage monitoring method for a digital twin system of a coal mine shaft. The method has a low implementation cost and is easy to promote and apply on a large scale. It can accurately locate the leakage point and obtain the volume and flow rate information of the leakage in the shaft, thereby providing a reliable basis for subsequent maintenance decisions. The method is suitable for long-term monitoring operations of coal mine shafts.

[0006] The present invention provides a water leakage monitoring method for a coal mine shaft digital twin system, comprising the following steps:

[0007] Step 1: Arrange monitoring components;

[0008] A number of monitoring points are determined on the wall of the wellbore, and a number of Twin-Tag sensors and a number of RFID antennas are installed at the monitoring points accordingly, and the Twin-Tag sensors and the RFID antennas are connected. At the same time, a monitoring terminal is arranged on the ground, and an RFID reader is arranged at the wellbore exit, and the RFID reader is connected to the number of Twin-Tag sensors through the number of RFID antennas, and the RFID reader is connected to the monitoring terminal; wherein the Twin-Tag sensor includes a water-absorbing sponge, a COTS RFID tag and a plastic package, a pair of COTS RFID tags are attached to the water-absorbing sponge in parallel, and a distance is set between the two; the plastic package covers the outside of the water-absorbing sponge and the pair of COTS RFID tags, and a water-permeable hole is opened thereon;

[0009] Step 2: Obtain wellbore monitoring data;

[0010] Use several Twin-Tag sensors to collect wellbore monitoring signals in real time, read the wellbore monitoring signals through an RFID reader, and then obtain wellbore monitoring data through analysis and processing of the wellbore monitoring signals. The wellbore monitoring data includes the RSSI reading of COTS RFID tag 1, the RSSI reading of COTS RFID tag 2, the phase data of COTS RFID tag 1, and the phase data of COTS RFID tag 2;

[0011] Step 3: Rapid detection of leaks;

[0012] S31: Use a short time window to collect RSSI readings and phase data, and extract signal features that can distinguish different states in each short time window;

[0013] S32: clustering the extracted signal features using t-SNE;

[0014] S33: inputting the clustered signal features into a pre-trained decision tree, using the pre-trained decision tree to identify and mark the state of each short-time window, wherein the state of the short-time window is divided into a normal state, a disturbance state, and a leakage state; when K short-time windows with leakage are continuously identified, triggering an alarm;

[0015] Step 4: Leakage signal extraction;

[0016] S41: For several short time windows marked as leaking state Calculate the window ratio parameter of each time signal according to formula (1): Extract satisfaction And the short time window length m>M continuous m signal windows And m signal windows Perform splicing to form a rough segmentation result of the leakage signal;

[0017]

[0018] S42: Based on the rough segmentation result, perform fine segmentation, and determine the starting point and end point of the leakage event through the change point detection algorithm to achieve accurate positioning of the leakage point; estimate the starting point of the leakage event by calculating the intersection of the local linear fitting of the normal state sample and the local linear fitting of the leakage state sample, and select the starting point of the leakage event from the result set sp = {sp [s] |s∈S}, select a joint estimate to minimize the sum of normalized fitting errors; where s∈S={R1,R2,P1,P2}, R1 is the RSSI reading of COTS RFID tag 1, R2 is the RSSI reading of COTS RFID tag 2, P1 is the phase data of COTS RFID tag 1, and P2 is the phase data of COTS RFID tag 2;

[0019] Step 5: Leakage volume v(S b ) and leakage rate r(S b )

[0020] S51: using an RFID reader to continuously read the wellbore monitoring signal emitted by the Twin-Tag sensor multiple times at different water leakage rates, and obtaining the RSSI reading of COTS RFID tag 1, the RSSI reading of COTS RFID tag 2, the phase data of COTS RFID tag 1, and the phase data of COTS RFID tag 2;

[0021] S52: using the nearest centroid algorithm to integrate the collected wellbore monitoring data to generate high-quality monitoring data;

[0022] S53: Using a search algorithm based on dynamic time warping to track the leakage process; using a dynamic time warping search algorithm to compare the similarity between time series in the monitoring data, in the comparison template sequence S a and the target sequence S b In the process of stretching and squeezing one of the two sequence segments, the best matching function f between the two sequences is found. M , so that the distance metric between two sequence segments is minimized, where and The distance measure between two sequence segments is

[0023] S54: According to formula (2), the matching path based on the dynamic time warping search algorithm is constrained by two parameters α and β to obtain a model-guided dynamic time warping search algorithm; the leakage process is tracked along the restricted search band by using the model-guided dynamic time warping search algorithm, and the unsaturated signal is spatially truncated in advance by combining α and β to avoid unexpected errors; wherein α is the length ratio of the two leakage time series, which is used to guide the search direction, and β is the step size of the restricted search;

[0024]

[0025] S55: Maximizing truncated subsequences by reverse traversal and The similarity between them detects the cutoff point t=(t a ,t b ), and then according to formula (3) the leakage volume v(S b ), according to formula (4), the leakage rate r(S b );

[0026]

[0027] In the formula, v(S a ) is the template sequence S a The leakage volume; L a The template sequence S a The length of the sequence segment; r(S a ) Template sequence S a The leakage rate.

[0028] A7: Based on the leakage progress tracking model, the time scale factor and spatial cutoff ratio are obtained to estimate the leakage volume v(S b ) and leakage rate r(S b ), providing reliable data support for subsequent maintenance decisions.

[0029] As a preferred embodiment, in step 3 S33, the RSSI gap feature f of a pair of COTS RFID tags is obtained according to formula (4), and when the f value changes significantly, it is determined that there is a wellbore leakage;

[0030]

[0031] Where r1 and r2 are the RSSI readings of COTS RFID tag 1 and COTS RFID tag 2, respectively, and t is the interval time for reading the RSSI values ​​of a pair of COTS RFID tags.

[0032] As a preferred embodiment, in step 3 S33, the decision tree uses the maximum information gain function or the minimum Gini index to select the best split feature;

[0033] When the maximum information gain function is used, the information gain function is defined according to formula (6) to measure the contribution of a feature to classification; when the minimum Gini index is used, the Gini index Gini (D) is obtained according to formula (7);

[0034] IG(D,A)=H(D)-H(D|A) (6);

[0035]

[0036] Where H(D) is the entropy of data set D, Where c is the number of categories, p i The proportion of category i in the dataset.

[0037] As a preferred embodiment, in step 1, the RFID reader model is ImpinJ Speedway R420.

[0038] As a preferred embodiment, in step one, when arranging the monitoring components, an alarm terminal is arranged on the ground and connected to the monitoring terminal, and the alarm terminal is used to perform an alarm action.

[0039] Furthermore, in order to ensure detection accuracy, in step 1, a pair of COTS RFID tags are spaced 1 mm apart.

[0040] The present invention uses RFID equipment to transmit passive signals to monitor the humidity data of the wellbore water-absorbing sponge as the input of the digital twin system. The inductive coupling between two adjacent tags is used as the signal feature of the water seepage of the wellbore water-absorbing sponge, which can ensure the synchronization of the physical object and the digital model and improve the accuracy of deep wellbore water leakage monitoring. Specifically, the Twin-Tag sensor includes a pair of COTS RFID tags attached to the water-absorbing sponge, which can make full use of the coupling effect between the pair of COTS RFID tags to effectively reflect the water leakage of the wellbore. Through this coupling effect, not only can the signal change caused by the wellbore water leakage be enhanced, but also the dynamic interference signal caused by multipath propagation and the numerous interferences generated by the mining equipment in the wellbore can be effectively filtered out. By using plastic packaging on the outside of the water-absorbing sponge and a pair of COTS RFID tags, the interference of external environmental factors can be effectively eliminated, and the reliability of the monitoring data can be improved. Therefore, this Twin-Tag sensor structure can ensure the sensitivity of detection, and can quickly capture the signal change when the leakage occurs, thereby significantly improving the detectability and accuracy of the water leakage signal. The change point detection algorithm can be used to more accurately determine the location of the water leakage in the shaft and its leakage characteristics. The model-guided search algorithm based on dynamic time warping is used to track the leakage process, which can accurately estimate the leakage volume and leakage rate, thereby helping to achieve more efficient leakage monitoring and management. The decision tree-based instant leakage state classification method can realize real-time monitoring of the water leakage in the deep shaft of the coal mine, and ensure rapid response when the water leakage occurs, effectively preventing potential safety hazards. For the segmented test data and the pre-learned signal template, the model-guided shaft leakage tracking algorithm is used to ensure that the leakage process can still be tracked along the restricted search when the signal segmentation is incomplete, and then the similarity between the template sequence and the target sequence can be well described, so that the leakage volume and leakage rate can be accurately obtained. In addition, due to channel conflicts, the interrogation frequency of RFID tags will decrease as the number of densely deployed adjacent tags increases. However, the model-guided shaft leakage search method has great flexibility in the face of the reduction of interrogation frequency caused by the increase in the number of tags, so it is capable of multi-point monitoring tasks.

[0041] This method has low implementation cost and is easy to promote and apply on a large scale. It can accurately locate the leakage point and obtain the volume and flow rate information of the leakage in the shaft, thereby providing a reliable basis for subsequent maintenance decisions. It is suitable for long-term monitoring operations in coal mine shafts. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is a schematic diagram of the assembly of the Twin-Tag sensor installed on the wellbore wall;

[0043] Figure 2 is a flow chart of the present invention;

[0044] Figure 3 It is a schematic diagram of the electromagnetic coupling state between a pair of COTS RFID tags in the Twin-Tag sensor of the present invention;

[0045] Figure 4 is a schematic diagram of an equivalent impedance circuit of a pair of COTS RFID tags in a Twin-Tag sensor of the present invention;

[0046] Figure 5 It is a schematic diagram of cluster visualization of the present invention. DETAILED DESCRIPTION

[0047] The present invention will be further described below in conjunction with the embodiments.

[0048] like Figure 1 and Figure 2 As shown, the present invention provides a water leakage monitoring method for a coal mine shaft digital twin system, comprising the following steps:

[0049] Step 1: Arrange monitoring components;

[0050] In order to effectively monitor the water leakage of the shaft wall of a deep coal mine, several monitoring points are determined on the shaft wall, and several Twin-Tag sensors and several RFID antennas are installed correspondingly at the several monitoring points, and the Twin-Tag sensors and the RFID antennas are connected. At the same time, a monitoring terminal is arranged on the ground, and an RFID reader is arranged at the shaft exit, and the RFID reader is connected to the several Twin-Tag sensors through the several RFID antennas, and the RFID reader is connected to the monitoring terminal; wherein the Twin-Tag sensor includes a water-absorbing sponge, a COTS RFID tag and a plastic package, a pair of COTS RFID tags are attached to the water-absorbing sponge in parallel, and a distance is set between the two; the plastic package is coated on the outside of the water-absorbing sponge and the pair of COTS RFID tags, and a water-permeable hole is opened thereon;

[0051] As a preference, the size of the absorbent sponge is 10 cm × 2 cm × 0.4 cm, and it can absorb 20 mL of water;

[0052] To ensure detection accuracy, a pair of COTS RFID tags are spaced 1mm apart. Under this arrangement, a pair of COTS RFID tags will have an RSSI (received signal strength indication) difference of about 5dB. This feature makes the signal recognition process more sensitive and accurate. The plastic packaging is wrapped around the absorbent sponge and the pair of COTS RFID tags, and a water-permeable hole is opened on it. The Twin-Tag sensor is used as a container for collecting leaked fluid.

[0053] Step 2: Obtain wellbore monitoring data;

[0054] Due to the inductive coupling effect between two adjacent tags, even a small amount of wellbore leakage can change the electrolyte between two COTS RFID tags, causing a significant change in the inductive coupling strength, such as Figure 3 As shown in the figure, the change of magnetic flux between a pair of COTSRFID tags due to the change of electrolysis resistance. In order to effectively predict the water leakage in the wellbore under the interference of many mining equipment in the wellbore, the change of coupling strength value between a pair of COTS RFID tags can be used to determine whether a water leakage accident has occurred in the wellbore. The coupling effect between a pair of COTS RFID tags can not only amplify the signal changes caused by water leakage in the wellbore, but also filter out the signal changes caused by other multipath dynamic interference. When mining equipment such as mine hoists or mine workers in the deep wellbore of a coal mine move around the LOS path of the RFID reader tag, the RFID reflection signal collected by the Twin-Tag sensor in the presence of dynamic interference can be observed. Even during the rapid fluctuation of interference, the RSSI gap difference of the two tags remains almost unchanged. There is an interaction between the two tags inside the Twin-Tag sensor. There is a gap between the RSSI measurement values ​​of the two COTS RFID tags before the water leakage accident occurs in the wellbore. When water leakage begins in the deep wellbore, the RSSI of the previous COTS RFID tag rises sharply, and the RSSI gap between the two COTS RFID tags narrows rapidly. It can not only amplify the signal changes caused by wellbore leakage to prevent confusion events, but also filter out signal changes caused by other multipath dynamic interference.

[0055] When the coal mine shaft leaks, the LOS path of the RFID reader tag will not be blocked. Therefore, due to the near-field coupling between the liquid and the tag antenna when the shaft leaks, the backscattered signal will change. To reflect this phenomenon, the equivalent impedance Z(Ψ) is used as Figure 4 As shown in the figure, the wellbore leakage is modeled to further determine the impact of the scattered signal micro-motion on the wellbore leakage monitoring, where Ψ is related to its volume. Then the input impedance Z of the COTS RFID tag is A It can be expressed as: The chip impedance of the COTS RFID tag is denoted as Z C , we can now see that the power reflection coefficient K, which represents the backscattered power ratio, becomes: The impedance matching relationship is expressed as:

[0056] As Ψ increases, the impedance matching relationship is no longer maintained, and the signal received by the RFID reader becomes weaker. However, this before-and-after signal change pattern is too weak to be robustly detected, which inspired the design of the Twin-Tag sensor: considering the mutual coupling during the leakage process, the signal changes in the first stage are "amplified" for early detection. When a leak occurs, the mutual inductance coupling weakens and Δε gradually decreases to zero, but this change does not affect the system monitoring. Therefore, in the first stage, when the wellbore just begins to leak, the RSSI value of COTS RFID tag 1 rises rapidly, and the RSSI difference with COTS RFID tag 2 increases rapidly. This obvious and detectable signal pattern helps to sensitively monitor the wellbore leak in real time.

[0057] Use several Twin-Tag sensors to collect wellbore monitoring signals in real time, read the wellbore monitoring signals through an RFID reader, and then obtain wellbore monitoring data through analysis and processing of the wellbore monitoring signals. The wellbore monitoring data includes the RSSI reading of COTS RFID tag 1, the RSSI reading of COTS RFID tag 2, the phase data of COTS RFID tag 1, and the phase data of COTS RFID tag 2;

[0058] Step 3: Rapid detection of leaks;

[0059] S31: Use a short time window to collect RSSI readings and phase data, and extract signal features that can distinguish different states in each short time window;

[0060] S32: t-SNE is used to cluster the extracted signal features, such as Figure 5 As shown;

[0061] S33: inputting the clustered signal features into a pre-trained decision tree, using the pre-trained decision tree to identify and mark the state of each short-time window, wherein the state of the short-time window is divided into a normal state, a disturbance state, and a leakage state; when K short-time windows with leakage are continuously identified, triggering an alarm;

[0062] The interference state refers to the signal change caused by the mixed event, and the stable gap value is selected to further exclude the wellbore leakage accident. The signal mode of the leakage state is different from other states. It is the superposition of the mutual inductance coupling between tags and the impedance change caused by the leaking liquid.

[0063] Step 4: Leakage signal extraction;

[0064] S41: For several short time windows marked as leaking state Calculate the window ratio parameter of each time signal according to formula (1): Extract satisfaction And the short time window length m>M continuous m signal windows And m signal windows Perform splicing to form a rough segmentation result of the leakage signal;

[0065]

[0066] S42: Based on the rough segmentation result, perform fine segmentation, and determine the starting point and end point of the leakage event through the change point detection algorithm to achieve accurate positioning of the leakage point. After determining the starting point and end point of the leakage event, the location of the leakage point can be effectively determined by combining the position information of several monitoring points; the starting point of the leakage event is estimated by calculating the intersection of the local linear fitting of the normal state sample and the local linear fitting of the leakage state sample, and the result set sp = {sp [s] |s∈S}, select a joint estimate to minimize the sum of normalized fitting errors; where s∈S={R1,R2,P1,P2}, R1 is the RSSI reading of COTS RFID tag 1, R2 is the RSSI reading of COTS RFID tag 2, P1 is the phase data of COTS RFID tag 1, and P2 is the phase data of COTS RFID tag 2;

[0067] Step 5: Leakage volume v(S b ) and leakage rate r(S b )

[0068] S51: using an RFID reader to continuously read the wellbore monitoring signal emitted by the Twin-Tag sensor multiple times at different water leakage rates, and obtaining the RSSI reading of COTS RFID tag 1, the RSSI reading of COTS RFID tag 2, the phase data of COTS RFID tag 1, and the phase data of COTS RFID tag 2;

[0069] S52: using the nearest centroid algorithm to integrate the collected wellbore monitoring data to generate high-quality monitoring data;

[0070] S53: Using a search algorithm based on dynamic time warping to track the leakage process; using a dynamic time warping search algorithm to compare the similarity between time series in the monitoring data, in the comparison template sequence S a and the target sequence S b In the process of stretching and squeezing one of the two sequence segments, the best matching function f between the two sequences is found. M , so that the distance metric between two sequence segments is minimized, where and The distance measure between two sequence segments is

[0071] S54: According to formula (2), the matching path based on the dynamic time warping search algorithm is constrained by two parameters α and β to obtain a model-guided dynamic time warping search algorithm; the leakage process is tracked along the restricted search band by using the model-guided dynamic time warping search algorithm, and the unsaturated signal is spatially truncated in advance by combining α and β to avoid unexpected errors; wherein α is the length ratio of the two leakage time series, which is used to guide the search direction, and β is the step size of the restricted search;

[0072]

[0073] S55: Maximizing truncated subsequences by reverse traversal and The similarity between them detects the cutoff point t=(t a ,t b ), and then according to formula (3) the leakage volume v(S b ), according to formula (4), the leakage rate r(S b );

[0074]

[0075] In the formula, v(S a ) is the template sequence S a The leakage volume can be obtained by pre-measurement; L a The template sequence S a The length of the sequence segment can be obtained by pre-measurement; r(S a ) Template sequence S a The leakage rate can be obtained by pre-measurement.

[0076] A7: Based on the leakage progress tracking model, the time scale factor and spatial cutoff ratio are obtained to estimate the leakage volume v(S b ) and leakage rate r(S b ), providing reliable data support for subsequent maintenance decisions.

[0077] As a preferred embodiment, in step 3 S33, the RSSI gap feature f of a pair of COTS RFID tags is obtained according to formula (4), and when the f value changes significantly, it is determined that there is a wellbore leakage;

[0078]

[0079] Where r1 and r2 are the RSSI readings of COTS RFID tag 1 and COTS RFID tag 2, respectively, and t is the interval time for reading the RSSI values ​​of a pair of COTS RFID tags.

[0080] As a preferred embodiment, in step 3 S33, the decision tree uses the maximum information gain function or the minimum Gini index to select the best split feature;

[0081] When the maximum information gain function is used, the information gain function is defined according to formula (6) to measure the contribution of a feature to classification; when the minimum Gini index is used, the Gini index Gini (D) is obtained according to formula (7);

[0082] IG(D,A)=H(D)-H(D|A) (6);

[0083]

[0084] Where H(D) is the entropy of data set D, Where c is the number of categories, p i The proportion of category i in the dataset.

[0085] As a preferred embodiment, in step 1, the RFID reader model is ImpinJ Speedway R420 with Laird circular polarization antenna. This high-performance reader can not only quickly identify and process signals, but also maintain excellent working stability in complex environments, thereby ensuring real-time monitoring and response capabilities for wellbore water leakage.

[0086] As a preferred embodiment, in step one, when arranging the monitoring components, an alarm terminal is arranged on the ground and connected to the monitoring terminal, and the alarm terminal is used to perform an alarm action.

[0087] The present invention uses RFID equipment to transmit passive signals to monitor the humidity data of the wellbore water-absorbing sponge as the input of the digital twin system. The inductive coupling between two adjacent tags is used as the signal feature of the water seepage of the wellbore water-absorbing sponge, which can ensure the synchronization of the physical object and the digital model and improve the accuracy of deep wellbore water leakage monitoring. Specifically, the Twin-Tag sensor includes a pair of COTS RFID tags attached to the water-absorbing sponge, which can make full use of the coupling effect between the pair of COTS RFID tags to effectively reflect the water leakage of the wellbore. Through this coupling effect, not only can the signal change caused by the wellbore water leakage be enhanced, but also the dynamic interference signal caused by multipath propagation and the numerous interferences generated by the mining equipment in the wellbore can be effectively filtered out. By using plastic packaging on the outside of the water-absorbing sponge and a pair of COTS RFID tags, the interference of external environmental factors can be effectively eliminated, and the reliability of the monitoring data can be improved. Therefore, this Twin-Tag sensor structure can ensure the sensitivity of detection, and can quickly capture the signal change when the leakage occurs, thereby significantly improving the detectability and accuracy of the water leakage signal. The change point detection algorithm can be used to more accurately determine the location of the water leakage in the shaft and its leakage characteristics. The model-guided search algorithm based on dynamic time warping is used to track the leakage process, which can accurately estimate the leakage volume and leakage rate, thereby helping to achieve more efficient leakage monitoring and management. The decision tree-based instant leakage state classification method can realize real-time monitoring of the water leakage in the deep shaft of the coal mine, and ensure rapid response when the water leakage occurs, effectively preventing potential safety hazards. For the segmented test data and the pre-learned signal template, the model-guided shaft leakage tracking algorithm is used to ensure that the leakage process can still be tracked along the restricted search when the signal segmentation is incomplete, and then the similarity between the template sequence and the target sequence can be well described, so that the leakage volume and leakage rate can be accurately obtained. In addition, due to channel conflicts, the interrogation frequency of RFID tags will decrease as the number of densely deployed adjacent tags increases. However, the model-guided shaft leakage search method has great flexibility in the face of the reduction of interrogation frequency caused by the increase in the number of tags, so it is capable of multi-point monitoring tasks.

[0088] This method has low implementation cost and is easy to promote and apply on a large scale. It can accurately locate the leakage point and obtain the volume and flow rate information of the leakage in the shaft, thereby providing a reliable basis for subsequent maintenance decisions. It is suitable for long-term monitoring operations in coal mine shafts.

Claims

1. A water leakage monitoring method for a coal mine shaft digital twin system, characterized in that: The following steps are involved: Step 1: Arrange monitoring components; A number of monitoring points are determined on the wall of the wellbore, and a number of Twin-Tag sensors and a number of RFID antennas are installed at the monitoring points, and the Twin-Tag sensors and the RFID antennas are connected. At the same time, a monitoring terminal is arranged on the ground, and an RFID reader is arranged at the wellbore exit, and the RFID reader is connected to the number of Twin-Tag sensors through the number of RFID antennas, and the RFID reader is connected to the monitoring terminal; wherein the Twin-Tag sensor includes a water-absorbing sponge, a COTS RFID tag and a plastic package, a pair of COTS RFID tags are attached to the water-absorbing sponge in parallel, and a distance is set between the two; the plastic package covers the outside of the water-absorbing sponge and the pair of COTS RFID tags, and a water-permeable hole is opened thereon; Step 2: Obtain wellbore monitoring data; Use several Twin-Tag sensors to collect wellbore monitoring signals in real time, read the wellbore monitoring signals through an RFID reader, and then obtain wellbore monitoring data through analysis and processing of the wellbore monitoring signals. The wellbore monitoring data includes the RSSI reading of COTSRFID tag 1, the RSSI reading of COTSRFID tag 2, the phase data of COTSRFID tag 1, and the phase data of COTSRFID tag 2; Step 3: Rapid detection of leaks; S31: Use a short time window to collect RSSI readings and phase data, and extract signal features that can distinguish different states in each short time window; S32: clustering the extracted signal features using t-SNE; S33: inputting the clustered signal features into a pre-trained decision tree, using the pre-trained decision tree to identify and mark the state of each short-time window, wherein the state of the short-time window is divided into a normal state, a disturbance state, and a leakage state; when K short-time windows with leakage are continuously identified, triggering an alarm; Step 4: Leakage signal extraction; S41: For several short time windows T marked as leakage state j (W) , calculate the ratio parameter of each time signal window according to formula (1) Extract satisfaction And the short time window length m>M continuous m signal windows And m signal windows {T j (W) }m are spliced ​​to form a rough segmentation result of the leakage signal; S42: Based on the coarse segmentation result, perform refined segmentation, and determine the starting point and end point of the leakage event through the change point detection algorithm to achieve accurate positioning of the leakage point; estimate the starting point of the leakage event by calculating the intersection of the local linear fitting of the normal state sample and the local linear fitting of the leakage state sample, and select a joint estimate from the result set sp = {Sp[s]|s∈S} to minimize the sum of normalized fitting errors; where s∈S = {R1, R2, P1, P2}, R1 is the RSSI reading of COTS RFID tag 1, R2 is the RSSI reading of COTS RFID tag 2, P1 is the phase data of COTS RFID tag 1, and P2 is the phase data of COTS RFID tag 2; Step 5: Leakage volume v(S b ) and leakage rate r(S b ) S51: using an RFID reader to continuously read the wellbore monitoring signal emitted by the Twin-Tag sensor multiple times at different water leakage rates, and obtaining the RSSI reading of COTS RFID tag 1, the RSSI reading of COTS RFID tag 2, the phase data of COTS RFID tag 1, and the phase data of COTS RFID tag 2; S52: using the nearest centroid algorithm to integrate the collected wellbore monitoring data to generate high-quality monitoring data; S53: Using a search algorithm based on dynamic time warping to track the leakage process; using a dynamic time warping search algorithm to compare the similarity between time series in the monitoring data, in the comparison template sequence S a and the target sequence S b In the process of stretching and squeezing one of the two sequence segments, the best matching function f between the two sequences is found. M , so that the distance metric between two sequence segments is minimized, where and The distance measure between two sequence segments is S54: According to formula (2), the matching path based on the dynamic time warping search algorithm is constrained by two parameters α and β to obtain a model-guided dynamic time warping search algorithm; the leakage process is tracked along the restricted search band by using the model-guided dynamic time warping search algorithm, and the unsaturated signal is spatially truncated in advance by combining α and β to avoid unexpected errors; wherein α is the length ratio of the two leakage time series, which is used to guide the search direction, and β is the step size of the restricted search; S55: Maximizing truncated subsequences by reverse traversal and The similarity between them detects the cutoff point t=(t a , t b ), and then according to formula (3) the leakage volume v(S b ), according to formula (4), the leakage rate r(S b ); In the formula, v(S a ) is the template sequence S a The leakage volume; L a The template sequence S a The length of the sequence segment; r(S a ) Template sequence S a The leakage rate. A7: Based on the leakage progress tracking model, the time scale factor and spatial cutoff ratio are obtained to estimate the leakage volume v(S b ) and leakage rate r(S b ), providing reliable data support for subsequent maintenance decisions.

2. A water leakage monitoring method for a coal mine shaft digital twin system according to claim 1, characterized in that: In step 3 S33, the RSSI gap feature f of a pair of COTS RFID tags is obtained according to formula (4), and when the f value changes significantly, it is determined that there is a wellbore leakage; Where r1 and r2 are the RSSI readings of COTS RFID tag 1 and COTS RFID tag 2, respectively, and t is the interval time for reading the RSSI values ​​of a pair of COTS RFID tags.

3. A water leakage monitoring method for a coal mine shaft digital twin system according to claim 1, characterized in that: In step 3 S33, the decision tree uses the maximum information gain function or the minimum Gini index to select the best split feature; When the maximum information gain function is used, the information gain function is defined according to formula (6) to measure the contribution of a feature to classification; when the minimum Gini index is used, the Gini index Gini (D) is obtained according to formula (7); IG(D,A)=H(D)-H(D|A) (6); Where H(D) is the entropy of data set D, Where c is the number of categories, p i The proportion of category i in the dataset.

4. A water leakage monitoring method for a coal mine shaft digital twin system according to claim 1, characterized in that: In step 1, the RFID reader model is ImpinJ Speedway R420.

5. The water leakage monitoring method for a coal mine shaft digital twin system according to claim 3 is characterized in that: In step one, when the monitoring components are arranged, an alarm terminal is arranged on the ground and connected to the monitoring terminal, and the alarm terminal is used to perform an alarm action.

6. A water leakage monitoring method for a coal mine shaft digital twin system according to claim 1, characterized in that: In step one, a pair of COTS RFID tags were spaced 1 mm apart.

Images