Mine hoist system twin monitoring method based on cloud edge collaboration

Abstract

The application discloses a mine hoisting system twin monitoring method based on cloud edge cooperation, comprising the following steps: constructing a mine hoisting system twin monitoring model, wherein the twin monitoring model comprises a fault diagnosis model, an advanced prediction model, a steel wire rope surface damage identification model and a three-dimensional simulation model; determining the deployment position of an edge server by using an improved k-means++ method; associating a terminal device with the edge server, distributing tasks on the terminal device to the edge server and a cloud server, and cooperatively processing the tasks by combining the edge server and the cloud server. The application constructs a mine hoisting system twin monitoring model under a cloud edge cooperation computing framework, which is beneficial to realizing full-state health monitoring of the mine hoisting system. The application distributes computing tasks by deploying an edge computing task distribution strategy, unloads the computing tasks to suitable edge servers, reduces the time delay of the twin monitoring model calculation, and guarantees the real-time performance of the mine hoisting system twin monitoring and early warning.

Description

Technical Field

[0001] This invention relates to the field of mine monitoring, and in particular to a twin monitoring method for mine hoisting systems based on cloud-edge collaboration. Background Technology

[0002] Mine hoisting systems comprise numerous sub-components distributed over a wide area, thus requiring a large number of diverse sensors to comprehensively monitor their operational status. However, directly transmitting massive amounts of data to the cloud for processing leads to delays and slow computation speeds, making real-time feedback to the control center impossible. Therefore, monitoring the operational status of mine hoisting systems and promptly addressing faults is of significant practical importance for the safe and efficient production of mines.

[0003] Digital twins refer to the construction of a virtual representation of the actual condition of mechanical equipment. Based on virtual-real mapping and evolution methods, they reflect the spatiotemporal evolution of key performance parameters of the equipment. Technically, digital twin technology has the potential to monitor the real-time operating status of equipment in mine hoisting systems. However, traditional cloud computing methods transmit sensor data to a central cloud server for centralized processing. This single cloud computing approach can overload the underground communication network. Furthermore, the wide spatial distribution and long data transmission distances of mine hoisting systems cause data transmission delays, resulting in untimely processing of equipment operating data and affecting the real-time data interaction between the digital twin virtual entity and the physical entity of the mine hoisting system. Summary of the Invention

[0004] Purpose of the invention: To address the above problems, the purpose of this invention is to provide a twin monitoring method for mine hoisting systems based on cloud-edge collaboration.

[0005] Technical solution: The present invention provides a twin monitoring method for mine hoisting systems based on cloud-edge collaboration, comprising:

[0006] A twin monitoring model for a mine hoisting system is constructed, which includes a fault diagnosis model, an advanced prediction model, a wire rope surface damage identification model, and a three-dimensional simulation model.

[0007] The deployment locations of edge servers are determined using an improved k-means++ method;

[0008] By associating terminal devices with edge servers, tasks on the terminal devices are distributed to edge servers and cloud servers, and tasks are collaboratively processed by the edge and cloud.

[0009] Furthermore, the construction of a twin monitoring model for the mine hoisting system includes:

[0010] Sensor signals from various sub-components in the mine hoisting system are collected to construct a database D1 representing system faults, a database D2 representing system performance parameters, a database D3 representing wire rope fault information, and a database D4 representing system operating conditions. The data in each database are preprocessed and then divided into training and testing sets according to a certain ratio.

[0011] A fault diagnosis model M1 is constructed. Vibration signal data from the test set in database D1 are input into the trained fault diagnosis model M1, and the health status of the equipment is evaluated based on the output results. This represents a sub-model for head wheel fault diagnosis driven by vibration signals. This represents a sub-model for intelligent drum fault diagnosis driven by vibration signals. A sub-model for fault diagnosis of motors driven by vibration signals;

[0012] A forward prediction model M2 is constructed. Real-time operational data from the devices in the D2 test set is input into the trained forward prediction model M2. The output is compared with a threshold to evaluate the device's operational status over a future period. A predictive sub-model representing the motor current signal. A sub-prediction model representing the wire rope tension signal. A predictive sub-model representing the hydraulic oil pressure signal of an intelligent brake;

[0013] A wire rope surface damage recognition model M3 is constructed. The wire rope image data in the D3 test set of the database is input into the trained wire rope surface damage recognition model M3 to obtain the wire rope damage results.

[0014] A three-dimensional simulation model M4 is constructed. Real-time temperature and stress data from database D4 are input into the trained three-dimensional simulation model M4. The obtained temperature and stress field data are compared with thresholds to evaluate the equipment's operating status. A three-dimensional finite element simulation system for mine hoisting systems. For the temperature field model, This is a stress field model.

[0015] Furthermore, determining the deployment location of edge servers using the improved k-means++ method includes:

[0016] In the mine hoisting system, each sensor is connected to a terminal device. The terminal device processes the data acquired by the sensors and transmits it to the edge server. The expression for calculating the importance of each terminal device is as follows:

[0017] WT i =α1Q i +α2D i +α3P i

[0018] In the formula, Q i D represents the workload of terminal device i. i P represents the amount of data loaded by the terminal device i. i The priority of the load task of terminal device i is indicated by α1, α2, and α3, which are respectively the weight coefficients of the load task, data volume, and priority of terminal device i.

[0019] The location of the edge server is determined based on the importance of each terminal device, and the distance between the edge server and the terminal device is calculated as follows:

[0020]

[0021] In the formula, This represents the actual transmission distance between terminal device i and the corresponding cluster center point.

[0022] Distance d i Instead of the Euclidean distance in the k-means++ algorithm, the improved k-means++ algorithm is used to find the 5 cluster center points of all terminal devices on the ground, and the 5 cluster center points are used as the deployment locations of edge servers.

[0023] Furthermore, the construction of a twin monitoring model for the mine hoisting system includes:

[0024] The fault diagnosis model, the advanced prediction model, and the wire rope surface damage identification model are deployed on edge servers and cloud servers.

[0025] Deploy the 3D simulation model on a cloud server;

[0026] Communication connection between edge servers and cloud servers.

[0027] Furthermore, associating terminal devices with edge servers includes:

[0028] Associate all terminal devices located underground with a single edge server deployed in the mine;

[0029] Associating terminal devices located above ground with edge servers located above ground involves the following steps:

[0030] Based on the signal priority, data volume, task load, and distance to the edge server of the terminal device, the matching value between terminal device i and edge server k is calculated. The calculation expression is as follows:

[0031]

[0032] In the formula, d(i,k) represents the actual distance between the terminal device and the edge server;

[0033] Select the largest match value between all terminal devices and all edge servers. If edge server k does not reach its maximum load limit, then the association between terminal device i and edge server k is determined; if edge server k reaches its maximum load limit, then the edge server with the largest matching value among the other edge servers within the maximum latency range that have not exceeded the maximum load limit is selected and associated with terminal device i; if all edge servers exceed their maximum load limits, then the edge server with the lowest load is selected and associated with terminal device i, until all terminal devices are associated with edge servers.

[0034] Furthermore, the expression for calculating the latency of computation task i on edge server j is as follows:

[0035] T ij =TR ij +TC ij +TW ij

[0036] In the formula, TR ij Indicates communication delay, TC ij Indicates the computation delay, TW ij This indicates a delay.

[0037] Furthermore, associating the terminal device with the edge server includes:

[0038] Under the constraints of interaction latency between physical equipment in a mine hoisting system and its twin monitoring model, interaction latency between different physical devices, and edge server resource limitations, a deployment model for the edge server is established. This deployment model is described as an integer programming problem, and the optimization expression is:

[0039] Min:

[0040]

[0041] (2)D i ij ,1≤j≤5

[0042] (3)T i,max >T ij ,i=1,2,3…,n,j=1,2,3,4,5

[0043] In the formula, n represents the number of terminal devices, m represents the number of computing tasks, and x​ij This represents the association between computation task i and edge server j. If computation task i is executed on edge server j, then x... ij =1, otherwise x ij =0; Constraint (1) indicates that each edge server can only process one computing task at a time, and each computing task can only be executed on one edge server; Constraint (2) indicates that the data storage space D required when computing task i requests allocation. i It must be less than the remaining storage capacity S of the edge server ij Constraint (3) indicates that the processing delay of each computation task is less than the maximum tolerance time T of the task. i,max .

[0044] Furthermore, distributing tasks from terminal devices to edge servers and cloud servers includes:

[0045] When computing task i is submitted to its associated edge server k, edge server k determines whether all edge servers at the edge can handle computing task i. If they cannot, computing task i is offloaded to the cloud; if they can, the latency T of all edge servers processing the computing task is calculated. ij Edge server k will handle each latency T ij The comparison is performed, and the edge server with the lowest latency is selected for task offloading.

[0046] Furthermore, after computing task i is offloaded to the cloud, the cloud server uses the twin monitoring model to complete the task's computation and returns the computation results to the edge server.

[0047] Furthermore, after the edge server processes computing task i, it sends the computing results to the cloud server;

[0048] The cloud server compares the calculation results with pre-set thresholds to assess the operating status of each piece of equipment in the mine hoisting system and issues an early warning if any abnormality occurs.

[0049] Beneficial effects: Compared with the prior art, the significant advantages of this invention are:

[0050] 1. Under the cloud-edge collaborative computing framework, this invention deploys a vibration signal-driven fault diagnosis model, an advanced prediction model, a wire rope surface defect identification model, and a three-dimensional simulation model to achieve full-state health monitoring of the mine hoisting system.

[0051] 2. This invention uses an edge-side computing task allocation strategy to offload computing tasks to appropriate edge servers, reducing the latency of twin monitoring model computation and ensuring the real-time performance of the system's twin monitoring and early warning. Attached Figure Description

[0052] Figure 1 A graph showing the cloud-edge collaborative computing relationship of the twin monitoring model;

[0053] Figure 2 This is a flowchart illustrating the connection between terminal devices and edge servers. Detailed Implementation

[0054] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments.

[0055] The cloud-edge collaborative twin monitoring method for mine hoisting systems described in this embodiment includes:

[0056] A twin monitoring model for a mine hoisting system is constructed, which includes a fault diagnosis model, an advanced prediction model, a wire rope surface damage identification model, and a three-dimensional simulation model.

[0057] The deployment locations of edge servers are determined using an improved k-means++ method;

[0058] By associating terminal devices with edge servers, tasks on the terminal devices are distributed to edge servers and cloud servers, and tasks are collaboratively processed by the edge and cloud.

[0059] The cloud-edge collaborative twin monitoring method for mine hoisting systems described in this embodiment designs twin monitoring methods for each component in the mine hoisting system. It combines real-time data processing at the edge with the powerful computing capabilities of the cloud, and uses a cloud-edge collaborative computing model to handle computationally intensive and latency-sensitive tasks generated by the twin monitoring of the mine hoisting system, thereby enabling real-time monitoring and early warning of the equipment operating status of the mine hoisting system.

[0060] Furthermore, such as Figure 1 The diagram shows the cloud-edge collaborative computing relationship of the twin monitoring model. The twin monitoring model for the mine hoisting system includes:

[0061] Sensor signals from various sub-components in the mine hoisting system are collected to construct a database D1 representing system faults, a database D2 representing system performance parameters, a database D3 representing wire rope fault information, and a database D4 representing system operating conditions. The data in each database are preprocessed and then divided into training and testing sets according to a certain ratio.

[0062] A fault diagnosis model M1 is constructed. Vibration signal data from the test set in database D1 are input into the trained fault diagnosis model M1, and the health status of the equipment is evaluated based on the output results. This represents a sub-model for head wheel fault diagnosis driven by vibration signals. This represents a sub-model for intelligent drum fault diagnosis driven by vibration signals. A sub-model for fault diagnosis of motors driven by vibration signals;

[0063] A forward prediction model M2 is constructed. Real-time operational data from the devices in the D2 test set is input into the trained forward prediction model M2. The output is compared with a threshold to evaluate the device's operational status over a future period. A predictive sub-model representing the motor current signal. A sub-prediction model representing the wire rope tension signal. A predictive sub-model representing the hydraulic oil pressure signal of an intelligent brake;

[0064] A wire rope surface damage recognition model M3 is constructed. The wire rope image data in the D3 test set of the database is input into the trained wire rope surface damage recognition model M3 to obtain the wire rope damage results.

[0065] A three-dimensional simulation model M4 is constructed. Real-time temperature and stress data from database D4 are input into the trained three-dimensional simulation model M4. The obtained temperature and stress field data are compared with thresholds to evaluate the equipment's operating status. A three-dimensional finite element simulation system for mine hoisting systems. For the temperature field model, This is a stress field model.

[0066] Data signals from various sub-components of the mine hoisting system are collected using sensors and constructed into databases D1, D2, D3, and D4. Database D1, representing system faults, includes vibration signal data from the sheave, drum, and motor. Database D2, representing system performance parameters, includes motor current signal data, wire rope tension signal data, and intelligent brake hydraulic oil pressure signal data. Database D3, representing wire rope fault information, includes wire rope image data. Database D4, representing system operating conditions, includes sheave speed, hoisting container speed and stopping position, drum speed and stress, and motor temperature data. After preprocessing the data in each database, such as denoising and data imputation, they are proportionally divided into training and testing sets. The twin monitoring model is trained using data from the training set.

[0067] Fault diagnosis model M1 includes a vibration signal-driven sheave fault diagnosis sub-model. Vibration signal driven intelligent drum fault diagnosis sub-model Sub-model for motor fault diagnosis driven by vibration signal Each fault diagnosis sub-model All models are composed of the same deep neural network, which includes ResNet and BiLSTM. ResNet is a residual module that uses CNN to extract local features, and BiLSTM is a bidirectional long short-term memory network that extracts temporal features. The two types of features are fused and input into the classification layer FC to obtain the diagnostic results. The mathematical model of each fault diagnosis sub-model is as follows:

[0068] Y1=FC(ResNet(X1)+BiLSTM(X1))

[0069] In the formula, X1 represents the preprocessed vibration signal, and Y1 represents the output of the deep neural network. The deep neural network is used to obtain classification results, which are then mapped one-to-one with fault types to determine the equipment's health status during operation. Fault types include sheave wheel sway, abnormal noise, and bearing failure; drum shell cracking, main shaft bending and breakage, and bearing failure; motor rotor deformation and gear damage. After diagnosing a fault, an alert is sent to the cloud via the edge server. Based on the actual situation, the system selects whether to suspend operation and perform equipment repairs.

[0070] The advanced prediction model M2 includes a prediction sub-model for the motor current signal. Sub-prediction model of wire rope tension signal Predictive sub-model of hydraulic oil pressure signal of intelligent brake Each sub-model is an LSTM model. Based on the D2 database, the LSTM model is trained using existing runtime data. Then, real-time runtime data is input into the trained LSTM model to achieve advanced prediction of key parameters. The mathematical model for each sub-model is as follows:

[0071] Y2 = LSTM(X2)

[0072] In the formula, X2 represents the real-time operating data in database D2, serving as input data; Y2 represents the prediction result of the operating data over a future period, which is 10 minutes in this embodiment, but can be set according to specific circumstances. An LSTM model is used to construct an advanced prediction model for motor current signals, wire rope tension signals, and intelligent brake hydraulic oil pressure signals to obtain prediction data. Then, based on whether the prediction data exceeds a threshold, the equipment operating status over a future period is assessed. Finally, the current operating status of the equipment is adjusted to ensure that the mine hoisting system remains at a normal level throughout the entire operation.

[0073] The wire rope surface damage identification model M3 is a YOLOv5 network. It is trained on the D3 database, and then the captured wire rope images are input into the trained model M3 to obtain the wire rope damage results. The mathematical model of the wire rope surface damage identification model M3 is as follows:

[0074] Y3 = YOLOv5(X3)

[0075] In the formula, X3 is the input image data, and Y3 is the output damage recognition result.

[0076] The 3D simulation model M4 includes a 3D finite element simulation system for mine hoisting systems. Temperature field model and stress field model Based on the database D4 characterizing the system's operating conditions, three-dimensional finite element simulations are performed on key components such as motors, intelligent sheaves, and intelligent drums to obtain simulated temperature and stress field information. Temperature and stress field datasets are then constructed, with input data defined as temperature and stress data at the nodes where sensors are deployed within the key components, and output data as the temperature and stress fields of the entire component. A deep learning convolutional neural network is trained using the constructed dataset D4. The trained convolutional neural network is then used to construct temperature and stress field models. Real-time temperature and stress data from the database D4 are input into these models to obtain reconstructed temperature and stress field data. Temperature and stress thresholds are set for each component of the mine hoisting system during operation, and alarms are issued when abnormal equipment conditions occur.

[0077] In mine hoisting systems, intelligent hoisting containers are mainly deployed underground, with a wide spatial distribution and sparse computational tasks. A single edge server can handle all computational tasks for the intelligent hoisting containers within the mine. Motors, brakes, intelligent drums, and intelligent sheaves are located above ground, in a more concentrated position. Their computational tasks include fault diagnosis of motors and bearings using vibration signals, predictive analysis of motor current, wire rope tension, and intelligent brake hydraulic pressure using LSTM networks, and identification of wire rope surface damage using wire rope images. This requires deploying five high-performance edge servers around the terminal equipment, and pairing each terminal equipment with one edge server at the edge, utilizing these edge servers to process the computational tasks.

[0078] Furthermore, determining the deployment location of edge servers using the improved k-means++ method includes:

[0079] In the mine hoisting system, each sensor is connected to a terminal device. The terminal device processes the data acquired by the sensors and transmits it to the edge server. The expression for calculating the importance of each terminal device is as follows:

[0080] WT i =α1Q i +α2D i +α3P i

[0081] In the formula, Q i D represents the workload of terminal device i. i P represents the amount of data loaded by the terminal device i. i The priority of the load task of terminal device i is indicated by α1, α2, and α3, which are respectively the weight coefficients of the load task, data volume, and priority of terminal device i.

[0082] The location of the edge server is determined based on the importance of each terminal device, and the distance between the edge server and the terminal device is calculated as follows:

[0083]

[0084] In the formula, This represents the actual transmission distance between terminal device i and the corresponding cluster center point.

[0085] Distance d i Instead of the Euclidean distance in the k-means++ algorithm, the improved k-means++ algorithm is used to find the 5 cluster center points of all terminal devices on the ground, and the 5 cluster center points are used as the deployment locations of edge servers.

[0086] Furthermore, the construction of a twin monitoring model for the mine hoisting system includes:

[0087] The fault diagnosis model, the advanced prediction model, and the wire rope surface damage identification model are deployed on edge servers and cloud servers.

[0088] Deploy the 3D simulation model on a cloud server;

[0089] Communication connection between edge servers and cloud servers.

[0090] Furthermore, associating terminal devices with edge servers includes:

[0091] Associate all terminal devices located underground with a single edge server deployed in the mine;

[0092] like Figure 2 As shown, the process of associating terminal devices located above ground with edge servers located above ground includes:

[0093] Based on the signal priority, data volume, task load, and distance to the edge server of the terminal device, the matching value between terminal device i and edge server k is calculated. The calculation expression is as follows:

[0094]

[0095] In the formula, d(i,k) represents the actual distance between the terminal device and the edge server;

[0096] Select the largest match value between all terminal devices and all edge servers. If edge server k has not reached its maximum load limit, then determine the maximum matching value. The association relationship between the corresponding terminal device i and the edge server k is determined. If the edge server reaches its maximum load limit, the edge server with the largest matching value among the other edge servers within the maximum latency range that have not exceeded the maximum load limit is selected to associate with the terminal device i. If all edge servers exceed the maximum load limit, the edge server with the lowest load is selected to associate with the terminal device i. The matched terminal devices and their matching values ​​are removed, and the matching of the remaining terminal devices continues until all terminal devices are associated with the edge servers.

[0097] The expression for calculating the latency of computation task i on edge server j is:

[0098] T ij =TC ij +TR ij +TW ij

[0099] In the formula, TC ij TR indicates communication delay. ij Indicates the computation delay, TW ij This indicates a delay.

[0100] Communication Delay TC ij This refers to the communication latency between the terminal device and the edge server, primarily consisting of uplink communication for transmitting data from the terminal device to the edge server. Specifically, this includes vibration signals from the motor, brake, intelligent drum, and intelligent sheave; motor current signals; wire rope tension signals and image signals; and hydraulic signals from the brake tank. Therefore, the uplink communication latency of computation task i on edge server j is:

[0101]

[0102] In the formula, N ij D represents the number of hops required to transmit computation task i to the edge server. iB represents the amount of data for computation task i, and B represents the network bandwidth between devices.

[0103] Calculate the delay TR ij This represents the time delay for the edge server to process the transmitted real-time data, and its calculation expression is:

[0104]

[0105] In the formula, Q i C represents the amount of work required to compute task i. j This represents the computing power of edge server j.

[0106] When computation task i is offloaded to edge server j, it cannot be processed in a timely manner due to the backlog of previous tasks, resulting in a delay of TW. ij This refers to the time required to process all existing tasks on edge server j, calculated as follows:

[0107]

[0108] In the formula, Q j This represents the backlog of tasks in edge server j.

[0109] Furthermore, associating the terminal device with the edge server includes:

[0110] Under the constraints of interaction latency between physical equipment in a mine hoisting system and its twin monitoring model, interaction latency between different physical devices, and edge server resource limitations, a deployment model for the edge server is established. This deployment model is described as an integer programming problem, and the optimization expression is:

[0111] Min:

[0112]

[0113] (2)D i ij ,1≤j≤5

[0114] (3)T i,max >T ij ,i=1,2,3…,n,j=1,2,3,4,5

[0115] In the formula, n represents the number of terminal devices, m represents the number of computing tasks, and x ij This represents the association between computation task i and edge server j. If computation task i is executed on edge server j, then x... ij =1, otherwise x ij ​=0; Constraint (1) indicates that each edge server can only process one computing task at a time, and each computing task can only be executed on one edge server; Constraint (2) indicates that the data storage space D required when computing task i requests allocation. i It must be less than the remaining storage capacity S of the edge server ij Constraint (3) indicates that the processing delay of each computation task is less than the maximum tolerance time T of the task. i,max .

[0116] Furthermore, distributing tasks from terminal devices to edge servers and cloud servers includes:

[0117] When computing task i is submitted to its associated edge server k, edge server k determines whether all edge servers at the edge can handle computing task i. If they cannot, computing task i is offloaded to the cloud; if they can, the latency T of all edge servers processing the computing task is calculated. ij Edge server k will handle each latency T ij The comparison is performed, and the edge server with the lowest latency is selected for task offloading.

[0118] Once computation task i is offloaded to the cloud, the cloud server uses the twin monitoring model to complete the computation of the task and returns the computation results to the edge server.

[0119] After the edge server processes computing task i, it sends the computing result to the cloud server.

[0120] The cloud server compares the calculation results with pre-set thresholds for temperature, stress, speed, etc. of each component during operation to evaluate the operating status of each piece of equipment in the mine hoisting system and issue an early warning if any abnormality occurs.

Claims

1. A twin monitoring method for mine hoisting systems based on cloud-edge collaboration, characterized in that, include: A twin monitoring model for a mine hoisting system is constructed, which includes a fault diagnosis model, an advanced prediction model, a wire rope surface damage identification model, and a three-dimensional simulation model. The construction of a twin monitoring model for mine hoisting systems includes: Collect sensor signals from various sub-components of the mine hoisting system to construct a database characterizing system faults. Databases characterizing system performance parameters A database characterizing wire rope fault information and a database characterizing system operating conditions The data from each database were preprocessed and then divided into training and testing sets according to a set ratio. Build a fault diagnosis model , database Test-collected vibration signal data is input into the trained fault diagnosis model. The system evaluates the health status of equipment based on the output results; the fault diagnosis model is one such model. , This represents a sub-model for head wheel fault diagnosis driven by vibration signals. This represents a sub-model for intelligent drum fault diagnosis driven by vibration signals. A sub-model for fault diagnosis of motors driven by vibration signals; Building advanced prediction models , database Real-time operational data from the devices in the test set are input into the trained advanced prediction model. In this process, the output results are compared with thresholds to assess the equipment's operating status over a future period; among them, the advanced prediction model... , A predictive sub-model representing the motor current signal. A sub-prediction model representing the wire rope tension signal. A predictive sub-model representing the hydraulic oil pressure signal of an intelligent brake; Constructing a surface damage identification model for steel wire ropes , database The test set of wire rope image data is input into the trained wire rope surface damage recognition model. In the process, the damage results of the wire rope were obtained; Constructing a 3D simulation model , database Real-time temperature and stress data are input into the trained 3D simulation model. In this process, the obtained temperature field data and stress field data are compared with thresholds to evaluate the operating status of the equipment; the three-dimensional simulation model... , A three-dimensional finite element simulation system for mine hoisting systems. For the temperature field model, For stress field model; The deployment locations of edge servers are determined using an improved k-means++ method; By associating terminal devices with edge servers, tasks on the terminal devices are distributed to edge servers and cloud servers, and tasks are collaboratively processed by the edge and cloud.

2. The twin monitoring method for mine hoisting systems based on cloud-edge collaboration according to claim 1, characterized in that, Determining the deployment location of edge servers using the improved k-means++ method includes: In the mine hoisting system, each sensor is connected to a terminal device. The terminal device processes the data acquired by the sensors and transmits it to the edge server. The expression for calculating the importance of each terminal device is as follows: ， In the formula, Indicates terminal device The workload of the load. Indicates terminal device The amount of data in the workload task. Indicates terminal device Priority of load tasks , , These are respectively represented as terminal devices. Weighting coefficients for workload, data volume, and priority; The location of the edge server is determined based on the importance of each terminal device, and the distance between the edge server and the terminal device is calculated as follows: ， In the formula, Indicates terminal device The actual transmission distance between the particle and the corresponding cluster center point; Distance Instead of the Euclidean distance in the k-means++ algorithm, the improved k-means++ algorithm is used to find the 5 cluster center points of all terminal devices on the ground, and the 5 cluster center points are used as the deployment locations of edge servers.

3. The twin monitoring method for mine hoisting systems based on cloud-edge collaboration according to claim 2, characterized in that, After constructing a twin monitoring model of the mine hoisting system, the following are included: The fault diagnosis model, the advanced prediction model, and the wire rope surface damage identification model are deployed on edge servers and cloud servers. Deploy the 3D simulation model on a cloud server; Communication connection between edge servers and cloud servers.

4. The twin monitoring method for mine hoisting systems based on cloud-edge collaboration according to claim 3, characterized in that, Associating terminal devices with edge servers includes: Associate all terminal devices located underground with a single edge server deployed in the mine; Associating terminal devices located above ground with edge servers located above ground involves the following steps: Based on the signal priority, data volume, task load, and distance to the edge server of the terminal device, the matching value between terminal device i and edge server k is calculated. The calculation expression is as follows: ， In the formula, This indicates the actual distance between the terminal device and the edge server; Select the largest match value between all terminal devices and all edge servers. If the edge server If the maximum load limit has not been reached, then the terminal device is identified. With edge servers The relationship between them; if the edge server reaches its maximum load limit, then select the edge server with the largest matching value among the other edge servers within the maximum latency range that have not exceeded their maximum load limit and pair it with the terminal device. Establish a connection; if all edge servers exceed their maximum load limit, select the edge server with the lowest load and the terminal device. Establish connections until all terminal devices are associated with the edge server.

5. The twin monitoring method for mine hoisting systems based on cloud-edge collaboration according to claim 4, characterized in that, Computational tasks The expression for calculating latency on edge server j is: ， In the formula, Indicates communication delay. Indicates the calculation delay. This indicates a delay.

6. The twin monitoring method for mine hoisting systems based on cloud-edge collaboration according to claim 5, characterized in that, Associating the terminal device with the edge server includes: Under the constraints of interaction latency between physical equipment in a mine hoisting system and its twin monitoring model, interaction latency between different physical devices, and edge server resource limitations, a deployment model for the edge server is established. This deployment model is described as an integer programming problem, and the optimization expression is: ， ， ， ， In the formula, Indicates the number of terminal devices. Indicates the number of computational tasks. Represents computational task With edge servers The relationship between them, if the computation task On the edge server If executed above, then ,otherwise Constraint (1) indicates that each edge server can only process one computing task at a time, and each computing task can only be executed on one edge server; Constraint (2) indicates that the computing task Data storage space required when requesting allocation It must be less than the remaining storage capacity of the edge server. Constraint (3) indicates that the processing delay of each computation task is less than the maximum tolerable time of the task. .

7. The twin monitoring method for mine hoisting systems based on cloud-edge collaboration according to claim 6, characterized in that, Distributing tasks from terminal devices to edge servers and cloud servers includes: When computing tasks Submitted to the associated edge server At that time, edge server Determine if all edge servers at the edge can handle the computation task. If it cannot be processed, then the computation task will be performed. Offload the computation to the cloud; if possible, calculate the latency of all edge servers processing the task. Edge server Each delay The comparison is performed, and the edge server with the lowest latency is selected for task offloading.

8. The twin monitoring method for mine hoisting systems based on cloud-edge collaboration according to claim 7, characterized in that, When computing tasks After being unloaded to the cloud, the cloud server uses the twin monitoring model to complete the task calculations and returns the calculation results to the edge server.

9. The twin monitoring method for mine hoisting systems based on cloud-edge collaboration according to claim 8, characterized in that, When the edge server processes computing tasks Then, the calculation results are sent to the cloud server; The cloud server compares the calculation results with pre-set thresholds to assess the operating status of each piece of equipment in the mine hoisting system and issues an early warning if any abnormality occurs.

Images