Remote control method for underground equipment

Abstract

The invention relates to the technical field of remote control, and discloses a remote control method for underground equipment, which comprises the following steps: constructing an edge-cloud collaborative decision architecture, reading real-time data by an edge node to obtain equipment operation and chamber environment data, and comparing the data with a preset equipment safety threshold; acquiring working condition data, calling historical operation data of cloud equipment, judging whether to adjust safety thresholds such as a power upper limit of a charger, and if so, adjusting; triggering an alarm and adjusting parameters when the real-time data exceeds a threshold value, calling maintenance historical data and combining the maintenance historical data with the periodic environment data of the chamber to predict the residual life of the equipment, and scheduling standby equipment for replacement when the equipment reaches the life; calling a fault early warning model, inputting real-time data to obtain an early warning result, generating a state report in combination with last maintenance data, and pushing the state report to ground operation and maintenance personnel through networking; before an instruction is issued, an edge node lightweight digital twin engine is started to simulate an execution effect, and the instruction is issued after parameters are corrected. According to the invention, the safety and efficiency of remote control of underground equipment can be improved.

Description

Technical Field

[0001] This application relates to the technical field of remote control, and in particular to a method for remote control of downhole equipment. Background Technology

[0002] Currently, the communication and data processing systems in the field of remote control of underground equipment in coal mines have significant deficiencies. Traditional underground communication often uses a single network solution, which is limited by explosion-proof standards. Some solutions have limited coverage and insufficient transmission stability in edge areas, while others rely on centralized cloud data processing, resulting in excessively high latency for emergency control commands.

[0003] Meanwhile, downhole data acquisition often processes equipment operation data and chamber environment data in isolation. The preset equipment safety thresholds are mostly fixed values ​​and cannot be dynamically adjusted according to real-time operating conditions such as gas concentration, humidity, and load, which can easily lead to safety control adaptation deviations. Furthermore, the information on the status of downhole equipment and the ground monitoring platform is not synchronized in a timely manner, making it difficult for maintenance personnel to obtain comprehensive control information in a timely manner, which restricts the safety and efficiency of remote control.

[0004] As can be seen from the above, how to improve the safety and efficiency of remote control of downhole equipment still needs to be addressed. Summary of the Invention

[0005] To improve the safety and efficiency of remote control of downhole equipment, this application provides a method for remote control of downhole equipment.

[0006] Firstly, this application provides a method for remote control of downhole equipment, employing the following technical solution: A method for remote control of downhole equipment, comprising: Distributed sensors and edge computing nodes are deployed in underground equipment and chambers to construct an edge-cloud collaborative decision-making architecture. Data is transmitted based on a hybrid power line carrier and 5G explosion-proof communication link. The edge nodes read the real-time underground data collected by the distributed sensors to obtain the underground equipment operation data and chamber environment data. The real-time underground data is synchronously uploaded to the cloud platform for storage and the edge node for local storage. The real-time underground data is then compared with the pre-set equipment safety thresholds. The system acquires current operating condition data for the underground equipment, including chamber gas concentration, ambient humidity, equipment load, and tunnel communication strength. It then retrieves historical operating data stored on a cloud platform and determines whether the equipment safety thresholds need adjustment based on the current operating condition data and the historical operating data. If so, the system adjusts the equipment safety thresholds, which include the charger output power limit, the battery swapping robot operating speed limit, and the ventilation system airflow limit. If the real-time data exceeds the equipment's safety threshold, a corresponding alarm mechanism is triggered, and the downhole equipment's operating parameters are adjusted accordingly. Historical equipment maintenance data stored on the cloud platform is retrieved, and the remaining expected service life of the downhole equipment is predicted based on this historical data and the real-time data. During the prediction of the remaining expected service life, the corresponding chamber is determined based on the collected data, and the periodic environmental change data corresponding to that chamber is retrieved. The final remaining expected service life is determined based on this periodic environmental change data and the historical equipment maintenance data. When the downhole equipment reaches its remaining expected service life, a backup device is scheduled to replace the target device through an edge-cloud collaborative architecture. Retrieve a pre-trained equipment failure early warning model, input the real-time downhole data into the equipment failure early warning model, and obtain the corresponding equipment early warning result; generate a corresponding equipment status report based on the equipment early warning result and the previously collected equipment maintenance history data; connect the downhole equipment management system to the ground monitoring platform, and push the equipment status report to the ground operation and maintenance personnel through a remote communication module; Before issuing equipment control commands, a lightweight digital twin engine deployed on the edge node is activated to simulate the command execution effect by combining the real-time data and the equipment status report; the equipment operating parameters are corrected according to the simulation results, and the equipment control commands are issued to the downhole equipment after the correction is completed.

[0007] Optionally, in the process of transmitting data based on the explosion-proof communication link of the hybrid power line carrier and 5G network, the method further includes: In the hybrid network, explosion-proof relay nodes are deployed. The edge nodes detect the communication strength of each area in the roadway in real time. When the communication strength of a certain area is detected to be lower than the preset transmission threshold, the dynamic switching mechanism of the explosion-proof relay nodes is automatically triggered to switch the data transmission link to the backup explosion-proof relay node in the coverage area.

[0008] Optionally, in the process of determining whether to adjust the equipment safety threshold based on the working condition data and the equipment's historical operating data, the method further includes: The working condition data, including chamber gas concentration, ambient humidity, equipment load, and roadway communication intensity, are assigned weights using a dynamic weighting algorithm. The weight of chamber gas concentration increases as its value increases, while the weights of ambient humidity and equipment load increase as their values ​​exceed the normal range. Based on the equipment fault records under the corresponding weight parameters in the historical operating data of the equipment, the current threshold adjustment demand is calculated. When the demand is higher than the preset judgment value, the equipment safety threshold adjustment is performed.

[0009] Optionally, in the process of predicting the remaining expected service life of downhole equipment based on the equipment maintenance history data and the real-time data, the method further includes: A lightweight TinyML federated learning model is deployed on the edge node. The lightweight TinyML federated learning model processes the device's running data and maintains historical data locally on the edge node, restricting the uploading of raw data to the cloud platform and only synchronizing the calculation results of the lightweight TinyML federated learning model to the cloud. At the same time, the prediction parameters of the lightweight TinyML federated learning model are corrected by combining the fluctuation patterns of chamber temperature and humidity in the periodic environmental change data.

[0010] Optionally, in the process of retrieving a pre-trained equipment failure early warning model and applying it to downhole equipment condition monitoring, the method further includes: At preset intervals, newly collected downhole real-time data and newly added equipment maintenance historical data are input into the equipment fault early warning model to iteratively train and update the equipment fault early warning model; wherein, the preset interval is set according to the downhole equipment operating intensity; During the update process, parameter modules with prediction accuracy higher than the preset accuracy threshold in the equipment fault early warning model are retained, while parameter modules with accuracy lower than the preset accuracy threshold are replaced.

[0011] Optionally, the method further includes: When the real-time data exceeds the equipment safety threshold and the parameter exceeding the threshold is the gas concentration in the chamber, in addition to triggering the alarm mechanism and adjusting the equipment operating parameters, the power input of the high-power charger in the chamber is automatically cut off, the backup ventilation system of the chamber is started, and a high gas emergency alarm message is sent to the ground monitoring platform through the remote communication module. At the same time, the access rights of personnel in the chamber are locked until the gas concentration drops to a safe range.

[0012] Optionally, the method further includes: Before issuing equipment control commands to the downhole equipment and when the controlled object is a mobile battery swapping robot, the lightweight digital twin engine also combines the tunnel layout data of the current chamber of the robot and the target scheduling chamber, as well as the real-time passage status data, to generate the optimal path for the robot to move across chambers. Meanwhile, the edge node establishes a pre-communication link with the edge node of the target scheduling chamber, and automatically switches the control signal receiving node when the robot enters the target chamber area.

[0013] Secondly, this application provides a remote control system for downhole equipment, which adopts the following technical solution: A remote control system for downhole equipment, comprising: The architecture construction and data acquisition comparison module deploys distributed sensors and edge computing nodes in the underground equipment and chambers to build an edge-cloud collaborative decision-making architecture. Data is transmitted based on a hybrid power line carrier and 5G explosion-proof communication link. The edge nodes read the real-time underground data collected by the distributed sensors to obtain the underground equipment operation data and chamber environment data. The real-time underground data is synchronously uploaded to the cloud platform for storage and the edge node for local storage. Then, the real-time underground data is compared with the pre-set equipment safety thresholds. The equipment safety threshold adjustment module acquires current working condition data of the underground equipment, including chamber gas concentration, ambient humidity, equipment load, and roadway communication intensity; retrieves historical equipment operation data stored on the cloud platform, and determines whether the equipment safety threshold needs to be adjusted based on the working condition data and the historical equipment operation data; if so, the equipment safety threshold is adjusted, including the upper limit of charger output power, the upper limit of battery swapping robot operating speed, and the lower limit of ventilation system airflow. The over-threshold response and equipment lifespan management module triggers a corresponding alarm mechanism if the real-time data exceeds the equipment safety threshold, and coordinates with the control system to adjust the operating parameters of the downhole equipment. It retrieves historical equipment maintenance data stored on the cloud platform and predicts the remaining expected lifespan of the downhole equipment based on this historical data and the real-time data. During the prediction of the remaining expected lifespan, it determines the corresponding chamber for the downhole equipment based on the collected data, retrieves the periodic environmental change data corresponding to that chamber, and determines the final remaining expected lifespan based on this periodic environmental change data and the historical equipment maintenance data. When the downhole equipment reaches its remaining expected lifespan, it schedules backup equipment to replace the target equipment through an edge-cloud collaborative architecture. The early warning and status report push module retrieves a pre-trained equipment fault early warning model, inputs the real-time downhole data into the equipment fault early warning model, and obtains the corresponding equipment early warning result; generates a corresponding equipment status report based on the equipment early warning result and the previously collected equipment maintenance history data; connects the downhole equipment management system to the ground monitoring platform, and pushes the equipment status report to the ground operation and maintenance personnel through the remote communication module. The control command pre-verification and issuance module enables a lightweight digital twin engine deployed on edge nodes before issuing equipment control commands. It combines the real-time data and the equipment status report to simulate the command execution effect. Based on the simulation results, it corrects the equipment operating parameters and issues the equipment control commands to the downhole equipment after the correction is completed.

[0014] Thirdly, this application provides a remote control system for downhole equipment, which adopts the following technical solution: A remote control system for downhole equipment includes a processor, wherein the processor runs a program for the remote control method for downhole equipment described in any one of the above-mentioned methods.

[0015] Fourthly, this application provides a storage medium, which adopts the following technical solution: A storage medium storing a program for the remote control method of downhole equipment as described in any one of the above.

[0016] In summary, this application includes at least one of the following beneficial technical effects: By constructing an edge-cloud collaborative architecture and a hybrid network of power line carrier + 5G, coupled with dynamic switching of explosion-proof relay nodes, the problems of high latency and unstable coverage in traditional communication are solved. At the same time, equipment operation and chamber environment data are collected synchronously, and safety thresholds are dynamically adjusted based on working conditions and historical data to avoid fixed threshold adaptation deviations. This reduces the risk of delayed emergency control response and ensures that equipment operating parameters are in line with real-time working conditions. It improves the safety of remote control from the perspective of communication and parameter management, and reduces ineffective scheduling losses to improve efficiency. In terms of enhancing safety and efficiency, measures such as triggering alarms when thresholds are exceeded, adjusting equipment parameters, and power-off ventilation in high-gas scenarios are implemented to strengthen emergency safety protection. By combining edge-side TinyML federated learning with periodic environmental data, the lifespan of equipment is accurately predicted, and backup equipment is promptly dispatched to reduce downtime. Before issuing instructions, parameters are pre-verified and corrected using digital twins to reduce execution risks. At the same time, status reports are pushed to ensure timely intervention by maintenance personnel. Path optimization and pre-communication switching for cross-chamber robot scheduling further improve operational efficiency, achieving a comprehensive improvement in both safety and efficiency. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating a method for remote control of downhole equipment according to an exemplary embodiment.

[0018] Figure 2 This is a structural block diagram of a remote control system for downhole equipment, illustrated according to an exemplary embodiment. Detailed Implementation

[0019] The embodiments of this application are described in detail below, and examples of the embodiments are shown in the accompanying drawings.

[0020] In the description of this specification, the references to "certain embodiments," "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples" refer to specific features, structures, materials, or characteristics described in connection with the described embodiment or example, which are included in at least one embodiment or example of this application. 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.

[0021] This application discloses a method for remote control of downhole equipment, referring to... Figure 1 ,include: The S100 deploys distributed sensors and edge computing nodes in underground equipment and chambers to build an edge-cloud collaborative decision-making architecture. Data is transmitted based on a hybrid power line carrier and 5G explosion-proof communication link. The edge nodes read real-time underground data collected by distributed sensors to obtain underground equipment operation data and chamber environment data. The real-time underground data is synchronously uploaded to the cloud platform for storage and the edge nodes for local storage. The real-time underground data is then compared with the pre-set equipment safety thresholds.

[0022] The specific execution process of S100 includes: Step 1, Hardware Deployment and Architecture Construction: First, two types of core hardware deployments were completed at key underground locations. One was to deploy distributed sensors on underground equipment such as chargers, battery swapping robots, and ventilation equipment, as well as at the entrances and exits of the chambers, in areas with dense equipment, and in areas prone to gas accumulation, to ensure comprehensive capture of equipment operating status and environmental changes. The other was to deploy edge computing nodes at key signal coverage nodes such as the middle of the chamber and the transfer points in the tunnels, and at the same time build an edge-cloud collaborative decision-making architecture. The edge nodes are responsible for local real-time data processing and emergency command response, while the cloud platform is responsible for historical data storage and global scheduling logic operations. The two form a collaborative mode of "local fast response + global optimal scheduling" through data interaction.

[0023] Step 2, Communication Link Construction and Data Transmission: A hybrid network explosion-proof communication link is constructed based on power line carrier and 5G technologies. Power line carrier technology is used to achieve long-distance, dust-resistant data transmission (adapting to complex underground wiring environments) by relying on existing underground power supply lines. Combined with 5G technology, it meets the requirements for high bandwidth, low latency, and real-time data transmission. At the same time, the entire link adopts an explosion-proof design to meet underground safety standards. During data transmission, the edge node acts as an "intermediate hub," receiving raw data collected by distributed sensors and synchronously transmitting data to the cloud platform through this hybrid link. Step 3, Data Reading, Classification, and Storage: Edge nodes actively read real-time downhole data collected by distributed sensors and classify it into two core data categories: one is "downhole equipment operation data" (such as charger output power, battery swapping robot running speed, and ventilation system air volume), and the other is "chamber environment data" (such as chamber gas concentration, ambient temperature, and humidity). The two types of data are then synchronously stored in two locations: a cloud platform (for long-term retention, subsequent historical data retrieval, and global analysis) and an edge node local storage module (for data backup and fast local access during network outages).

[0024] Step 4: Preliminary comparison of real-time data with preset thresholds: Edge nodes call the locally stored "equipment safety thresholds" (pre-set according to the safety operation standards of underground equipment and the safety standards of the chamber environment, such as the upper limit of charger output power and the safe value of gas concentration, etc.) and compare the classified real-time data (equipment operation data and environmental data) with the corresponding thresholds one by one. For example, the real-time gas concentration is compared with the gas safety threshold, and the real-time charger power is compared with the power upper limit. This provides an initial judgment basis for the current equipment operation and environmental status to determine whether they are within the safe range, and provides an initial judgment basis for subsequent S200 threshold adjustment and S300 over-threshold response.

[0025] By deploying hardware and building an architecture, a complete data link is constructed, consisting of "sensing (sensors) - processing (edge ​​nodes) - storage (edge ​​+ cloud) - transmission (hybrid communication)," which solves the problems of incomplete data acquisition, unstable transmission, and scattered storage in traditional remote control, providing data and hardware support for all subsequent control steps. At the same time, its hybrid networking explosion-proof communication link takes into account both anti-interference and low latency, local processing and storage at edge nodes ensure fast data retrieval, and explosion-proof design and classified storage respectively ensure communication security and data integrity, effectively guaranteeing data timeliness and security.

[0026] S200: Obtain current working condition data of the underground equipment, including chamber gas concentration, ambient humidity, equipment load, and roadway communication strength; retrieve historical operating data of the equipment stored on the cloud platform; determine whether the equipment safety threshold needs to be adjusted based on the working condition data and the historical operating data; if so, adjust the equipment safety threshold, which includes the upper limit of charger output power, the upper limit of battery swapping robot operating speed, and the lower limit of ventilation system airflow.

[0027] The specific execution process of S200 includes: Step 1, Accurately Obtain Working Condition Data: Based on the distributed sensors and edge computing nodes deployed by S100, data on the current working conditions of underground equipment are collected in a directional manner. Among them, the gas concentration and ambient humidity data of the chamber are directly retrieved from the chamber environment sensors, the equipment load data (such as the real-time load of the charger and the load of the battery swapping robot) are obtained from the sensors on the equipment itself, and the tunnel communication strength data is generated by the edge nodes detecting the signal strength of the hybrid network link. After all the data is initially verified by the edge nodes (abnormal fluctuation data is removed), a complete "current working condition dataset" is formed to ensure that the data can truly reflect the real-time environment and load status of the equipment.

[0028] Step 2, retrieve historical operating data of the equipment: The edge node initiates a retrieval request to the cloud platform that has stored data in S100 via a hybrid communication link of power line carrier and 5G to obtain the historical operating data of the target downhole equipment. This data includes equipment operating parameters from the past 3 months to 1 year (such as charger power under different loads and robot operating speed under different humidity), historical threshold adjustment records, equipment failure and normal operation data under similar working conditions, and the data is classified and labeled according to "working condition type" (such as high gas working condition and high load working condition) to facilitate matching with the current working condition data.

[0029] Step 3, Determine the threshold adjustment requirement: Establish a "current operating condition - historical data" comparison model: First, align the current operating condition data with the historical data of the same type of operating condition (e.g., the current gas concentration of 1.2% corresponds to the historical gas concentration of 1.0% to 1.5%). Analyze whether the fixed threshold under the historical operating conditions has ever resulted in "thresholds that are too loose leading to risks (e.g., excessive power causing temperature rise)" or "thresholds that are too strict affecting efficiency (e.g., low air volume causing humidity accumulation)". Then, combine the special characteristics of the current operating condition (e.g., the current ambient humidity is 10% higher than the historical operating conditions with the same gas concentration) to comprehensively judge whether the current preset equipment safety threshold is appropriate, and form a judgment result of "needs adjustment" or "no need for adjustment".

[0030] Step 4, Precisely adjust the equipment safety threshold: If the judgment result is "adjustment required", then differentiated adjustments will be made to the safety thresholds for different types of equipment: the upper limit of charger output power will be adjusted according to the current equipment load and ambient humidity (e.g., when the load exceeds the rated value by 80% and the humidity exceeds 85%, the upper limit of power will be reduced by 15% to 20%), the upper limit of battery swapping robot running speed will be adjusted according to the tunnel communication strength and gas concentration (e.g., when the communication strength is lower than the S100 preset transmission threshold and the gas concentration exceeds 0.8%, the upper limit of speed will be reduced by 0.2m / s to 0.3m / s), and the lower limit of ventilation system air volume will be adjusted according to the chamber gas concentration and ambient humidity (e.g., when the gas concentration exceeds 1.0%, the lower limit of air volume will be increased by 200m³ / h to 300m³ / h); the adjusted thresholds will be synchronously stored on the edge node local and cloud platforms, replacing the original preset thresholds.

[0031] By acquiring current working condition data such as the gas concentration and ambient humidity of the underground equipment, and retrieving historical operating data of the equipment stored in the cloud, the safety thresholds of the equipment, such as the upper limit of the charger output power, are determined and adjusted. This not only solves the defect of fixed safety thresholds in traditional remote control, but also achieves dynamic adaptation between the threshold and the working conditions (avoiding efficiency reduction due to overly strict thresholds or risk due to overly lenient thresholds). Furthermore, relying on historical operating data provides practical support for threshold adjustment, thereby improving the scientific nature of threshold setting.

[0032] S300: If real-time data exceeds the equipment safety threshold, the corresponding alarm mechanism is triggered, and the downhole equipment is controlled to adjust its operating parameters. The system retrieves historical equipment maintenance data stored on the cloud platform and predicts the remaining expected service life of the downhole equipment based on the historical data and real-time data. During the prediction of the remaining expected service life, the downhole equipment is identified based on the collected data, and the corresponding periodic environmental change data for that chamber is retrieved. The final remaining expected service life is determined based on the periodic environmental change data and the historical equipment maintenance data. When the downhole equipment reaches its remaining expected service life, a backup device is scheduled to replace the target device through an edge-cloud collaborative architecture.

[0033] The specific execution process of S300 includes: Step 1, Threshold Exceedance Detection and Immediate Response Trigger: Using the adjusted equipment safety thresholds of S200 as the judgment standard, and comparing them with the real-time downhole data (equipment operation data + chamber environment data) collected and synchronized by S100: if a certain type of data exceeds the corresponding threshold (such as the chamber gas concentration exceeding the safe value, the charger output power exceeding the upper limit, or the battery swapping robot speed exceeding the upper limit), a dual immediate response is immediately triggered. First, an alarm mechanism is activated, issuing a warning through the downhole audible and visual alarm, and simultaneously pushing alarm information containing "the type of parameter exceeding the threshold, the magnitude of the exceedance, and the location" to the ground monitoring platform; second, the downhole equipment is linked to adjust its operating parameters, such as reducing the charger power when the gas concentration exceeds the limit, lowering the operating speed of the battery swapping robot when the speed exceeds the limit, and increasing the air volume of the ventilation system when the air volume is insufficient, to quickly curb the expansion of risks.

[0034] Step 2, Retrieval of historical equipment maintenance data and preliminary lifespan prediction: While triggering the immediate response, the edge node sends a data retrieval request to the cloud platform to obtain complete maintenance history data of the device currently exceeding the threshold, including the device's past maintenance time, maintenance content (such as component replacement, parameter calibration), number of failures and causes, and post-repair operational stability records. Subsequently, this maintenance history data is correlated with the real-time data collected by the S100 (such as current device load, ambient humidity, and operating temperature). Based on the "matching degree between historical failures and real-time operating conditions", the remaining expected service life of the device is initially predicted (for example, if a charger is prone to failure under high load in the past, and is currently under high load and has been repaired more than 3 times, the remaining service life is initially predicted to be short).

[0035] Step 3, Retrieval of chamber periodic environmental data and correction of lifespan prediction: Based on the unique identifier of the downhole equipment (such as a battery swapping robot or a set of chargers) collected from the data, determine its assigned chamber (such as the No. 1 charging and swapping chamber or the No. 3 auxiliary chamber). Retrieve the periodic environmental change data of the assigned chamber from the cloud platform, including the monthly average humidity for the past 3 months / half a year, the quarterly gas concentration fluctuation range, and the environmental temperature change patterns in different seasons. Combine this periodic environmental data with the equipment maintenance history data to revise the preliminary lifespan prediction results. For example, if the humidity in the chamber to which a certain piece of equipment belongs is consistently high during the rainy season, and historical data shows that a high humidity environment will accelerate the aging of equipment components, then the remaining expected lifespan needs to be further shortened based on the preliminary prediction, ultimately obtaining a "final remaining expected lifespan" that is more in line with the actual working conditions.

[0036] Step 4, Triggering and Execution of Backup Equipment Scheduling: Edge nodes monitor the remaining expected lifespan of downhole equipment in real time. When the remaining expected lifespan of a certain piece of equipment drops to a preset critical value (e.g., only 7 days / 10 days left) or has reached its final remaining expected lifespan, the edge node sends a backup equipment scheduling request to the cloud platform. The cloud platform, through the edge-cloud collaborative architecture, queries the location, status, and compatibility of currently idle backup equipment (e.g., backup chargers, backup battery swapping robots) and issues scheduling instructions to the edge nodes where the corresponding backup equipment is located. After receiving the instructions, the target edge node controls the backup equipment to move to the target equipment location and stops the target equipment from running, completing the replacement of the backup equipment with the target equipment and ensuring that downhole charging and battery swapping operations are not interrupted.

[0037] On the one hand, by triggering an alarm mechanism when real-time data exceeds the threshold and adjusting the operating parameters of downhole equipment in conjunction with it, the risk spread can be quickly contained, solving the problem of traditional remote control only alarming but not intervening, and enabling real-time management of safety risks such as gas exceeding limits and equipment overload.

[0038] On the other hand, by combining historical equipment maintenance data stored in the cloud with periodic environmental change data of the corresponding chamber, the accuracy of predicting the remaining expected service life of downhole equipment can be improved, avoiding equipment waste caused by misjudgment or sudden failures caused by omission. At the same time, relying on the edge-cloud collaborative architecture, backup equipment can be scheduled to replace the equipment when it reaches its remaining expected service life, preventing work stoppages caused by equipment failure and replacement, and ensuring the continuity of critical downhole operations.

[0039] The S400 retrieves a pre-trained equipment fault early warning model, inputs real-time downhole data into the model, and obtains the corresponding equipment early warning results. Based on the equipment early warning results and the previously collected historical equipment maintenance data, it generates a corresponding equipment status report. The downhole equipment management system is connected to the ground monitoring platform, and the equipment status report is pushed to ground maintenance personnel through a remote communication module.

[0040] The specific execution process of S400 includes: Step 1, Retrieval of the pre-trained fault warning model: The fault warning model is pre-trained based on historical fault data of underground equipment (such as charger overload faults, battery swapping robot jamming faults, insufficient ventilation system air pressure faults, etc.), normal operation data of equipment, and corresponding chamber environment data. It is lightweight and deployed on edge computing nodes (adapting to the computing power requirements of underground edge terminals). During execution, the edge nodes directly call the fault warning model stored locally, without relying on cloud transmission, avoiding the delay in calling the fault warning model due to network fluctuations.

[0041] Step 2, Real-time downhole data input and early warning result acquisition: The edge nodes input the collected and classified "real-time downhole data" (including equipment operation data such as charger output voltage, battery swapping robot motor temperature, and chamber environment data such as gas concentration and humidity) into the fault early warning model. The fault early warning model analyzes the correlation between the data and fault characteristics through preset algorithms (such as feature matching and risk level determination), and outputs the corresponding equipment early warning results, including the early warning risk level (such as low, medium, and high), the equipment components that may be faulty (such as charger heat dissipation module and robot wheels), and the risk cause speculation (such as "charger temperature is too high, suspected cooling fan failure").

[0042] Step 3, Generation of Equipment Status Report: Edge nodes retrieve the "last collected equipment maintenance history data" (including the last maintenance time, maintenance items, replaced parts models, fault repair records, etc.) stored on the cloud platform, and integrate it with the above-mentioned early warning results to generate an equipment status report. The report must clearly present three core contents: current early warning information (risk level, faulty parts, and causes), last maintenance details (e.g., "last maintenance was 30 days ago, charger cooling fan was replaced"), and correlation analysis between the two (e.g., "current charger temperature warning, 30 days since the last cooling fan replacement, fan operation status needs to be checked"), to ensure that the report information is complete and has operational guidance significance.

[0043] Step 4, System Network Connection and Device Status Report Push: First, establish a stable connection between the downhole equipment management system and the ground monitoring platform (relying on the power line carrier and 5G hybrid network explosion-proof communication link built by S100 to ensure that data transmission complies with downhole explosion-proof standards); then, through the "remote communication module" (an explosion-proof module adapted to the complex downhole environment) deployed at the edge node, push the generated equipment status report to the display terminal of the ground monitoring platform and the mobile terminals of ground maintenance personnel (such as explosion-proof mobile phones and maintenance tablets) to achieve real-time information synchronization.

[0044] By retrieving and analyzing real-time downhole data from the fault early warning model, hidden dangers can be identified before obvious equipment failures occur, reducing the probability of sudden downtime. At the same time, by combining the early warning results with historical data from the last equipment maintenance, a status report is generated, providing accurate basis for operation and maintenance decisions and avoiding the waste of time and costs from blind repairs. Furthermore, by connecting the downhole equipment management system with the ground monitoring platform, the report can be pushed to ground operation and maintenance personnel, achieving real-time synchronization of ground and well information. This allows operation and maintenance personnel to grasp the equipment status without going down into the well, prepare tools and spare parts in advance, and shorten fault response time.

[0045] Before issuing equipment control commands, the S500 activates a lightweight digital twin engine deployed on edge nodes, combining real-time data and equipment status reports to simulate the execution effect of commands; based on the simulation results, it corrects the equipment operating parameters, and after the correction is completed, it issues equipment control commands to the downhole equipment.

[0046] The specific execution process of S500 includes: Step 1, enable the lightweight digital twin engine at the edge: First, the lightweight digital twin engine deployed on the downhole edge computing node is started (this engine is adapted to the computing power and storage resources of the downhole edge node and does not rely on cloud computing power). At the same time, two types of core data are called: one is the real-time downhole data collected by the distributed sensors in S100 and synchronized to the edge node (including equipment operation data such as the current power of the charger and environmental data such as the current gas concentration in the chamber); the other is the equipment status report generated by S400 (including equipment warning results such as "the motor temperature of the battery swapping robot is too high" and historical maintenance records such as "the cooling fan was replaced during the last maintenance"), to provide data support for subsequent simulations.

[0047] Step 2, simulate the effect of executing device control commands: Determine the type of equipment control command to be issued (e.g., "increase the charger output power from 80% to 90%" or "adjust the battery swapping robot's running speed to 0.5m / s"), and input the command parameters into the lightweight digital twin engine. Based on the real-time data and status reports, the engine constructs a virtual mapping scenario of the underground equipment and the chamber environment, simulating the changes in the equipment's operating status after the command is executed (e.g., the temperature change trend of the charger after the power is increased, the energy consumption and passage safety of the robot after the speed is adjusted), as well as the potential impact on the surrounding environment (e.g., whether the power increase will lead to excessive local electromagnetic field strength, and whether the change in robot speed will affect the work efficiency).

[0048] Step 3: Adjust the equipment operating parameters based on the simulation results: Compare the simulation results with the adjusted equipment safety thresholds of S200 (e.g., whether the simulated charger temperature is close to the upper temperature limit, whether the robot energy consumption exceeds the load threshold): If the simulation results meet the safety thresholds and operational requirements (e.g., the temperature stabilizes within a safe range after power increase, while also improving charging efficiency), then retain the original command parameters; if the simulation results exceed the safety thresholds (e.g., the temperature will exceed the threshold after power increase) or do not meet operational requirements (e.g., the robot speed is too fast, causing path planning conflicts), then correct the command parameters (e.g., reduce the charger power increase to 85%, adjust the robot speed to 0.4 m / s), and re-simulate and verify until the parameters meet both safety and efficiency requirements.

[0049] Step 4: Issue the revised equipment control instructions: After the parameters are corrected, the final equipment control command is sent from the edge node to the corresponding downhole equipment (such as chargers and battery swapping robots) through the power line carrier and 5G hybrid network explosion-proof communication link built by S100. At the same time, the command content and correction record are uploaded to the cloud platform (for subsequent historical data retrieval and command execution effect tracking), completing the closed-loop control of "simulation-correction-sending".

[0050] By enabling a lightweight digital twin engine on edge nodes before issuing equipment control commands, and combining real-time data and equipment status reports to simulate command execution effects and correct parameters, this approach avoids the problem of inappropriate parameters caused by the traditional direct issuance of commands without pre-verification, reducing equipment failures or environmental safety hazards. It also ensures that the corrected control parameters fit the current chamber environment and equipment status, avoiding the drawbacks of "one-size-fits-all" commands. At the same time, relying on edge node deployment ensures the real-time performance and reliability of simulation and command issuance, achieving safe, accurate, and efficient control of downhole equipment.

[0051] The following case illustrates this point: In the No. 1 charging and battery swapping chamber of a coal mine, three 100kW high-power chargers and two mobile battery swapping robots are deployed to supply energy to 10 unmanned new energy wide-body vehicles underground. The solution proposed in this application achieves a dual improvement in safety and efficiency. During daily operation, the distributed sensors deployed on the S100 system collect data such as methane concentration and charger load in real time. This data is synchronized to edge nodes and the cloud via a hybrid power line carrier and 5G communication link. The edge nodes quickly perform a preliminary comparison of the data with preset thresholds locally, laying the foundation for subsequent management and control. When the humidity in the chamber rises to 82% (normal range 40% to 70%) and the charger load reaches 90% of the rated value, the S200 uses a dynamic weighting algorithm to increase the humidity and load weights to 28% and 37% respectively. It retrieves historical data from the cloud and finds that similar operating conditions have caused charger overheating failures due to overly loose thresholds. It then automatically lowers the charger power limit from 100kW to 82kW, which avoids the risk of equipment overload and ensures the basic charging needs of wide-body vehicles, solving the problem of traditional fixed thresholds that either limit power and affect efficiency or pose hidden dangers at full load.

[0052] During the afternoon shift one day, the gas sensor data in Chamber 1 suddenly jumped to 0.85% (the safety threshold of 0.5%). The S300 immediately triggered an emergency response: the edge node cut off the charger power supply within 1 second, activated the backup ventilation system, and simultaneously pushed an alarm message containing the exceedance level and handling status to the ground monitoring platform, locking the chamber access control to prevent personnel from entering. Simultaneously, historical maintenance data of the charger showed that it had been running continuously for 180 days and had experienced three instances of abnormal heat dissipation. Combined with the cyclical environmental data of Chamber 1's average humidity of 78% during the rainy season over the past three months, the lightweight TinyML federated learning model revised the predicted remaining lifespan, reducing it from 30 days to 12 days. The cloud platform, through an edge-cloud collaborative architecture, scheduled the backup charger in Chamber 2. After the S500 digital twin engine simulated the path, the mobile battery swapping robot guided it to complete the replacement within 30 minutes, without interrupting the wide-body vehicle battery swapping operation. This shortened the replacement time by 45 minutes compared to traditional manual replacement, avoiding transportation efficiency losses due to downtime.

[0053] When the battery-swapping robot in Chamber 1 needs to be dispatched to Chamber 3 to recharge a faulty wide-body vehicle, the fault warning model deployed by the S400 system first analyzes the robot's real-time data and finds that its motor temperature has increased by 8°C from the baseline. Combined with the previous maintenance record (bearing replacement 15 days ago), a "medium-risk warning" report is generated and pushed to the ground maintenance terminal. After the maintenance personnel remotely instruct the robot to execute the dispatch task, the S500 digital twin engine, based on the tunnel layout and real-time traffic data, plans the optimal path: "Chamber 1 → Main Tunnel → Transfer Node → Chamber 3". Simultaneously, a pre-communication link is established between the edge nodes of Chambers 1 and 3. When the robot reaches the transfer node, the control signal automatically switches to the edge node of Chamber 3, with no communication interruption throughout the process. The path is shortened by 5 minutes compared to traditional manual planning, and the battery-swapping operation is successfully completed upon arrival. Throughout the process, the fault warning proactively avoids the risk of robot downtime, and path optimization and communication switching ensure dispatch efficiency, achieving "early detection of potential problems and zero dispatch errors".

[0054] In this embodiment of the application, during the data transmission process based on the explosion-proof communication link of the hybrid power line carrier and 5G network, the method further includes: Step 1, Deployment of Explosion-Proof Relay Nodes: In the underground hybrid network link, explosion-proof relay nodes (which must meet underground explosion-proof safety standards) are evenly deployed in areas prone to signal attenuation, such as roadway ends, bends, and densely packed equipment areas. Each relay node has a preset coverage area and backup link identifier to ensure that the coverage areas of adjacent relay nodes overlap (avoiding signal blind spots). At the same time, all relay nodes are connected to edge computing nodes to form a two-layer transmission network of "main link + backup relay link".

[0055] Step 2, Real-time detection of communication strength at edge nodes: Edge computing nodes use built-in signal detection modules to sample the hybrid network communication strength in various areas of the roadway in real time (sampling frequency is synchronized with data transmission frequency). They focus on monitoring the data transmission link strength between distributed sensors, underground equipment and edge nodes, record the real-time signal value of each transmission link, and continuously compare it with the preset transmission threshold (set according to the stability requirements of underground data transmission to ensure data transmission without packet loss and with low latency).

[0056] Step 3, trigger dynamic switching mechanism judgment: When the edge node detects that the communication strength of a certain area is lower than the preset transmission threshold (such as the signal attenuation of the direct link between the end sensor of the alley and the edge node, and the strength drops below the threshold), the dynamic switching judgment logic of the explosion-proof relay node is immediately triggered: first, identify the backup explosion-proof relay nodes within the coverage area of ​​the area, confirm the current communication load and operating status of the backup nodes (excluding faulty or overloaded backup nodes), and filter out the available target backup relay nodes.

[0057] Step 4, Data Transmission Link Switching Execution: The edge node sends a link switching command to the target backup explosion-proof relay node, and simultaneously sends a link adjustment signal to the distributed sensors and downhole equipment in the area, switching the original direct transmission link to an indirect transmission link of "sensor / equipment → target backup relay node → edge node". During the switching process, the continuity of data transmission is checked synchronously (to avoid data loss during the switching interval). After the new link transmission is stable (signal strength rises back above the threshold and there is no packet loss in data transmission), the dynamic switching is completed, and the stable data transmission of the hybrid network continues to be maintained. In this embodiment of the application, the method further includes, in the process of determining whether the equipment safety threshold needs to be adjusted based on working condition data and historical equipment operating data: Step 1, Dynamic Weight Algorithm Initialization and Parameter Setting: First, preset the dynamic weight algorithm model on the edge node or cloud platform, and clarify the basic weight and adjustment rules of each parameter in the working condition data: Set the initial basic weights of the chamber gas concentration, ambient humidity, equipment load, and tunnel communication intensity (e.g., 30%, 20%, 30%, 20% respectively), and define the "normal range" of each parameter (e.g., the normal range of gas concentration is 0% to 0.5%, the normal range of ambient humidity is 40% to 70%, the normal range of equipment load is 50% to 80% of the rated load, and the normal range of tunnel communication intensity is ≥-85dBm), and preset the weight increment rules (e.g., set the increment gradient based on the magnitude or value of the parameter exceeding the normal range).

[0058] Step 2: Calculate the actual weight of each parameter based on real-time working condition data: The edge node reads the current working condition data of the downhole equipment obtained from S200 and substitutes it into the dynamic weight algorithm model: If the gas concentration value in the chamber increases (e.g., from 0.3% to 0.6%), its weight is increased according to preset rules (e.g., for every 0.1% exceeding the normal upper limit, the weight increases by 5%); if the ambient humidity exceeds the normal range (e.g., from 65% to 80%), the weight is increased by the magnitude of the exceedance (e.g., for every 10% exceeding the normal upper limit, the weight increases by 8%); if the equipment load exceeds the normal range (e.g., from 70% of the rated load to 90%), the weight is increased by the magnitude of the exceedance (e.g., for every 10% exceeding the normal upper limit, the weight increases by 7%); if the tunnel communication strength is lower than the normal range (e.g., from -80dBm to -95dBm), the weight is increased by the magnitude of the decrease (e.g., for every 5dBm below the normal lower limit, the weight increases by 6%). Finally, the real-time actual weight of each parameter is obtained (the sum of the weights of each parameter is 100%).

[0059] Step 3: Retrieve corresponding fault records from historical equipment operation data: Edge nodes retrieve historical equipment operation data stored on the cloud platform via a hybrid communication link, filter out historical records with "parameter weight distribution" similar to the current working condition data (e.g., if the current gas concentration weight is 40% and the load weight is 35%, then filter out historical records where the sum of the weights of these two parameters is ≥70%), extract equipment fault information from these historical records, and statistically analyze the fault type (e.g., overload fault, gas-related fault), fault frequency, and threshold deviation at the time of the fault (e.g., the charger power exceeds the threshold by 15% at the time of the fault).

[0060] Step 4: Combine fault records to calculate the threshold adjustment demand and construct a demand calculation model. Use "actual weight of each parameter × corresponding historical fault frequency × fault threshold deviation rate" as the individual demand contribution value, and sum them to obtain the current total demand for threshold adjustment. For example, if the actual weight of gas concentration is 40%, the corresponding historical fault frequency is 8 times, and the fault threshold deviation rate is 20%, then the contribution value of this item is 40% × 8 × 20% = 0.64. Similarly, calculate the contribution values ​​of environmental humidity, equipment load, and roadway communication intensity, and sum them to obtain the total demand (e.g., total demand = 0.64 + 0.28 + 0.42 + 0.16 = 1.5).

[0061] Step 5: Based on the demand level, determine whether to perform threshold adjustment by calling the preset "threshold adjustment demand level judgment value" (set according to the downhole equipment safety operation requirements and historical adjustment effects, such as a judgment value of 1.2). Compare the calculated current total demand level with the judgment value: if the total demand level is higher than the judgment value (e.g., 1.5 > 1.2), then trigger the equipment safety threshold adjustment command and enter the "adjust equipment safety threshold" execution stage in S200; if the total demand level is lower than or equal to the judgment value, then maintain the original equipment safety threshold unchanged.

[0062] By pre-setting a dynamic weighting algorithm in the S200 threshold adjustment judgment and allocating the actual weight of each parameter based on real-time working condition data, and combining the equipment's historical fault records to calculate the threshold adjustment demand, the accuracy and scientific nature of the equipment safety threshold adjustment judgment are achieved, avoiding empirical judgment errors and ensuring that the threshold adjustment is adapted to real-time working conditions and is necessary.

[0063] In this embodiment of the application, the method for predicting the remaining expected service life of downhole equipment based on historical and real-time equipment maintenance data further includes: Step 1, Deployment of the lightweight TinyML federated learning model: Deploy a lightweight TinyML federated learning model adapted to the computing power of the edge nodes in each underground chamber (the lightweight TinyML federated learning model is pre-processed and compressed to reduce the occupation of computing power and storage resources). The lightweight TinyML federated learning model is pre-set with "equipment operation data parsing module", "historical data matching module", and "lifespan prediction calculation module", and establishes association with the local storage module and data transmission module of the edge node to ensure that it can read local data and only output the calculation results.

[0064] Step 2: The lightweight TinyML federated learning model reads data and performs local processing: The edge node retrieves two types of core data from the local storage of the lightweight TinyML federated learning model: one is the real-time operating data of the downhole equipment collected and stored by S100 (such as the cumulative working time of the charger and the motor loss value of the battery swapping robot), and the other is the equipment maintenance history data synchronized from the cloud to the local machine by S300 (such as past maintenance time, replacement parts type, and post-maintenance performance recovery rate). The local processing module of the lightweight TinyML federated learning model cleans and extracts features from the two types of data (such as extracting key features such as "maintenance interval duration" and "real-time loss rate"). The original data is not uploaded to the cloud platform throughout the process, and only the processed feature data is retained for model calculation. Step 3: The lightweight TinyML federated learning model performs a preliminary calculation of the remaining expected service life. The "life prediction calculation module" of the lightweight TinyML federated learning model uses the locally processed feature data to call a preset life prediction algorithm (such as a fitting algorithm based on the equipment wear rate and maintenance cycle) to perform a preliminary calculation of the remaining expected service life of the downhole equipment (such as a preliminary prediction of the charger's remaining service life of 800 hours), and temporarily stores the preliminary calculation result in the result cache area of ​​the edge node. Step 4: Retrieve periodic environmental data and analyze fluctuation patterns: Edge nodes retrieve periodic environmental change data (such as the daily temperature fluctuation range and monthly average humidity trend of the chamber in the past 3 months) from the cloud platform via a hybrid communication link. The "Environmental Data Adaptation Module" of the lightweight TinyML federated learning model analyzes the data patterns: identify key environmental factors such as the cumulative duration of temperature above 30℃ and the frequency of humidity above 75%, and determine the degree of impact of these factors on equipment aging (e.g., for every 100 hours of temperature exceeding 30℃, the equipment lifespan decreases by 5%; for every 50 hours of humidity exceeding 75%, the lifespan decreases by 3%).

[0065] Step 5: Correct the prediction parameters of the lightweight TinyML federated learning model and output the final results: The lightweight TinyML federated learning model adjusts the core parameters in the "lifespan prediction calculation module" according to the degree of influence of environmental factors on lifespan (e.g., the "basic loss coefficient" is corrected from 0.0012 to 0.0015 to match the accelerated aging characteristics under high humidity environment); the remaining expected lifespan is recalculated based on the corrected parameters (e.g., the remaining lifespan of the charger is adjusted from 800 hours to 720 hours after correction); finally, the final calculation results (not the original data) are synchronized to the cloud platform through the communication link, and stored locally on the edge node for use in the subsequent "scheduling backup equipment" stage of S300. By deploying a lightweight TinyML federated learning model at edge nodes, local processing of equipment operation data and maintenance history data is achieved (restricting the uploading of raw data to the cloud). At the same time, the prediction parameters of the lightweight TinyML federated learning model are corrected by combining the periodic fluctuation patterns of temperature and humidity in the well chamber, thereby improving the prediction accuracy of the remaining expected service life of downhole equipment and ensuring the security of data transmission.

[0066] In this embodiment of the application, the method further includes, during the process of retrieving a pre-trained equipment fault early warning model and applying it to downhole equipment condition monitoring: Step 1, Initialize the preset iteration cycle and accuracy threshold: In the model management module of the cloud platform or edge node, first set the "model iteration preset cycle" according to the operating intensity of the downhole equipment: For equipment with high operating intensity (such as chargers and battery swapping robots that work for more than 16 hours a day), set the cycle to 24-48 hours; for equipment with low operating intensity (such as ventilation auxiliary equipment that works for less than 8 hours a day), set the cycle to 72-120 hours. At the same time, preset the "parameter module accuracy threshold" (such as 90%) as the standard for subsequent judgment on whether to retain the module (that is, if the module prediction accuracy is ≥90%, it is retained; if it is <90%, it is replaced).

[0067] Step 2, Iteration Cycle Triggering and Data Preparation: The timing module on the edge node or in the cloud monitors the iteration cycle in real time. When the preset cycle is reached, the model iteration process is automatically triggered: First, "new real-time downhole data" (new equipment operation data and chamber environment data added by distributed sensors within the cycle, such as temperature fluctuations and current changes of the charger in the past 24 hours) are collected, and "new equipment maintenance history data" (equipment maintenance records, component replacement information, and fault handling results generated within the cycle, such as gear replacement records and post-maintenance operating parameters of a robot) are retrieved simultaneously. The two types of data are preprocessed (outliers are cleaned, data formats are standardized, and fault correlation features are extracted, such as the feature combination of "sudden temperature rise + unstable current") to form the model iteration training dataset.

[0068] Step 3, Input new data to perform iterative training of the model: Input the preprocessed iterative training dataset into the equipment fault early warning model and start incremental iterative training: The equipment fault early warning model fine-tunes existing parameters (such as fault feature weights and fault type judgment thresholds) based on the new data, rather than retraining the entire dataset. For example, if the new data shows that "when the humidity in the chamber is >80%, the frequency of charger insulation faults increases", the equipment fault early warning model will automatically increase the weight of "humidity feature" in charger fault judgment. At the same time, the real-time prediction accuracy of each "parameter module" (such as charger fault identification module, battery swapping robot motion fault module, and ventilation equipment wind pressure fault module) is recorded during the training process.

[0069] Step 4, Parameter Module Accuracy Evaluation and Screening: After iterative training, the model management module independently evaluates the prediction accuracy of each parameter module. By comparing the module's prediction results for "known fault cases in new data" with the actual fault situations, the accuracy of each module is calculated (e.g., the charger fault module correctly predicts 92 out of 100 known fault cases, with an accuracy rate of 92%; the battery swapping robot motion fault module correctly predicts 85 out of 100, with an accuracy rate of 85%). The accuracy of each module is compared with a preset accuracy threshold (90%), and modules with an accuracy rate ≥90% are screened (e.g., the charger fault module is retained), while modules with an accuracy rate <90% are marked (e.g., the battery swapping robot motion fault module is marked).

[0070] Step 5, Low-accuracy module replacement and model activation: Replace the marked low-accuracy parameter modules. If a pre-trained backup module exists (such as a backup optimization module for motion faults in the battery swapping robot), replace it directly. If no backup module exists, retrain the module based on new data and historical high-accuracy data until the accuracy is ≥90%. Combine the "retained module + replacement / new training module" into an updated equipment fault warning model, and synchronize it to the edge node (for local real-time monitoring) and the cloud platform (for global model management) to ensure that subsequent S400 equipment status monitoring is performed based on the updated model, while backing up the old model for rollback in case of anomalies.

[0071] By setting the iteration cycle according to the operating intensity of downhole equipment, the equipment fault early warning model is incrementally trained using newly collected real-time data and newly added maintenance historical data. High-accuracy parameter modules are retained and low-accuracy modules are replaced to continuously maintain the model's early warning accuracy and ensure the accuracy and reliability of downhole equipment status monitoring in S400.

[0072] In this embodiment of the application, the method further includes: Step 1, Threshold Parameter Judgment and Emergency Trigger: The edge node continuously receives real-time data on the gas concentration in the chamber collected by the distributed sensors and compares it with the "safe threshold for gas concentration in the chamber" (e.g., 0.5%) adjusted by S200. When the gas concentration value is detected to exceed the threshold (e.g., rises to 0.6% or above), the "gas exceeding the threshold special emergency procedure" is immediately triggered, and the regular exceeding threshold response logic is suspended, prioritizing the execution of emergency operations for high-risk scenarios.

[0073] Step 2, core safety measures are executed simultaneously: After the emergency process is initiated, the edge node simultaneously issues two key commands through the underground equipment control bus: First, it sends a "power cut-off command" to all high-power chargers (such as equipment with a rated power ≥10kW) in the chamber, triggering the explosion-proof power-off module built into the equipment to cut off the main power input and avoid the risk of gas explosion caused by electric sparks generated during the operation of high-power equipment; Second, it sends a "start command" to the backup ventilation system of the chamber (a redundant device independent of the main ventilation system) to start the backup fan, increase the air circulation in the chamber, accelerate the dilution and diffusion of gas, and at the same time monitor the operating status of the ventilation system (such as fan speed and air pressure) in real time through sensors to ensure that the equipment starts normally.

[0074] Step 3, Emergency Alarm and Personnel Access Control: The edge node sends a "high gas emergency alarm message" to the ground monitoring platform through the hybrid network explosion-proof communication link. The message includes the chamber number exceeding the threshold, the real-time gas concentration value, and the current status of emergency measures (such as "charger is powered off, backup ventilation is activated"). At the same time, it triggers an audible and visual alarm on the ground platform to ensure that maintenance personnel are informed immediately. Simultaneously, it sends an "access lock command" to the personnel access control system (such as explosion-proof access control, electronic fence controller) at the chamber entrance to close the access control and prevent personnel from entering. At the same time, it pushes a "chamber entry restriction reminder" to the underground worker positioning system to prevent personnel from accidentally entering high-risk areas.

[0075] Step 4, Status Monitoring and Emergency Deactivation: Edge nodes continuously monitor changes in methane concentration in the chamber and the effectiveness of emergency measures: methane concentration data is collected in real time through sensors and updated to the ground platform every 10 seconds; when the methane concentration drops below the safe threshold (e.g., stable at 0.4% or below) and remains so for more than 5 minutes, a "methane concentration restored to safe status" notification is first sent to the ground monitoring platform. After confirmation by maintenance personnel, emergency measures are gradually deactivated. A "permission unlock command" is first sent to the personnel access control system to restore normal access to the chamber; if the main ventilation system is normal, the backup ventilation system can be shut down; finally, based on the instructions of maintenance personnel, it is determined whether to restart the high-power charger (if it is confirmed that there are no other risks on site, the equipment power is gradually restored).

[0076] When the gas concentration in the chamber exceeds the safety threshold, in addition to routine alarms and parameter adjustments, the system automatically cuts off the power to the high-power charger, activates backup ventilation, sends an emergency alarm to the ground, and locks the access of personnel in the chamber. This effectively prevents the risk of gas explosion, ensures the safety of personnel underground, and achieves rapid and accurate risk management in high-risk scenarios.

[0077] In this embodiment of the application, the method further includes: Step 1, Control Object Determination and Data Retrieval: After receiving the "Mobile Battery Swapping Robot Control Command Request", the edge node first determines that the control object is a mobile battery swapping robot through the device identifier and that cross-chamber scheduling (such as scheduling from chamber A to chamber B) is required. Then, it retrieves two types of core data from the cloud platform through the hybrid communication link: one is "the tunnel layout data of the current chamber (chamber A) and the target scheduling chamber (chamber B) of the robot" (including pre-stored geographical information such as tunnel direction, passage width, fixed obstacle position, and turning node coordinates); the other is "real-time tunnel traffic status data" (collected by distributed position sensors and video monitoring equipment, including the current position of other mobile devices in the tunnel, temporary obstacles (such as piles of maintenance tools), and information on congested sections).

[0078] Step 2: Lightweight Digital Twin Engine Generates Optimal Path: The edge node activates the lightweight digital twin engine, importing the retrieved lane layout data and real-time traffic status data into the engine's "path planning module." The engine first constructs a 3D digital model of the lane from chamber A to chamber B, marking fixed obstacles and real-time obstacle areas. Then, based on the principle of "shortest path + obstacle avoidance + low congestion," it calculates multiple candidate paths (e.g., Path 1: Chamber A → Main Lane 1 → Turning Node C → Main Lane 2 → Chamber B; Path 2: Chamber A → Auxiliary Lane 1 → Turning Node D → Main Lane 2 → Chamber B). By comparing the estimated travel time, obstacle avoidance difficulty, and frequency of intersections with other equipment for each candidate path, the optimal path is selected (e.g., Path 1 has an estimated travel time of 12 minutes and no congestion, so it is determined as the optimal path). The path parameters (including node coordinates, suggested driving speed, and turning prompts) are then embedded into the control command plan.

[0079] Step 3, edge nodes establish pre-communication links: While the digital twin engine generates the optimal path, the edge nodes of the current chamber (chamber A) send a "pre-communication link establishment request" to the edge nodes of the target scheduling chamber (chamber B) through hybrid networking. The request information includes the device ID of the mobile battery swapping robot, the estimated time of entry into chamber B, and the required communication bandwidth. After receiving the request, the edge node of chamber B verifies its own communication resources (such as current idle bandwidth and signal coverage). After confirming availability, it sends a "link confirmation response" back to the edge node of chamber A. Both parties establish a temporary pre-communication link based on the explosion-proof communication protocol and simultaneously test the link stability (such as transmission delay and data packet loss rate) to ensure that the link meets the real-time transmission requirements of the robot control signals.

[0080] Step 4, Automatic Switching of Robot Movement and Control Signal Receiving Nodes: The edge node sends control commands containing the optimal path to the mobile battery swapping robot, and the robot starts moving across chambers according to the path; at the same time, the robot's built-in position sensor collects its own coordinates in real time and feeds back the position information to the edge node of chamber A every 5 seconds; when the edge node of chamber A detects that the robot's coordinates have entered the "signal coverage threshold area of ​​the edge node of chamber B" (such as within 10 meters of the entrance of chamber B), it automatically triggers the "control signal receiving node switching command": on the one hand, it notifies the edge node of chamber A to stop sending control signals, and on the other hand, it activates the established pre-communication link, instructing the robot to switch the control signal receiving end from the edge node of chamber A to the edge node of chamber B.

[0081] Step 5, Post-Switch Status Confirmation and Continuous Command Execution: After receiving the signal feedback from the robot after the switch, the edge node of chamber B sends a "Switch Successful Confirmation" message to the edge node of chamber A, and at the same time continuously sends control signals to the robot (such as adjusting the driving speed and indicating the upcoming turning node); the lightweight digital twin engine synchronously monitors the robot's driving status in the tunnel of chamber B. If a new temporary obstacle is encountered, the path is corrected in real time and the control commands are updated through the edge node of chamber B until the robot safely arrives at the target scheduling chamber, completing the cross-chamber scheduling. In the scenario of cross-chamber scheduling where the controlled object is a mobile battery swapping robot, the optimal movement path is generated by combining the tunnel layout and real-time traffic data through a lightweight digital twin engine. At the same time, a pre-communication link is established with the edge node of the target chamber, and the control signal receiving node is automatically switched when the robot enters the target area. This effectively ensures the safe and efficient movement of the robot across chambers and the continuous and stable control signal, avoiding path congestion or communication interruption from affecting the scheduling.

[0082] This application discloses a remote control system for downhole equipment, referring to... Figure 2 ,include: The architecture construction and data acquisition comparison module 001 deploys distributed sensors and edge computing nodes in the underground equipment and chambers to build an edge-cloud collaborative decision-making architecture. Data is transmitted based on a hybrid power line carrier and 5G explosion-proof communication link. The edge nodes read the real-time underground data collected by the distributed sensors to obtain the underground equipment operation data and chamber environment data. The real-time underground data is synchronously uploaded to the cloud platform for storage and the edge node for local storage. Then, the real-time underground data is compared with the pre-set equipment safety thresholds. The equipment safety threshold adjustment module 002 acquires the current working condition data of the underground equipment, including the gas concentration in the chamber, ambient humidity, equipment load, and roadway communication strength; it retrieves the historical operating data of the equipment stored on the cloud platform, and determines whether the equipment safety threshold needs to be adjusted based on the working condition data and the historical operating data; if so, it adjusts the equipment safety threshold, which includes the upper limit of the charger output power, the upper limit of the battery swapping robot operating speed, and the lower limit of the ventilation system air volume. The over-threshold response and equipment lifespan management module 003 triggers a corresponding alarm mechanism if real-time data exceeds the equipment safety threshold, and coordinates with the control system to adjust the operating parameters of the downhole equipment. It retrieves historical equipment maintenance data stored on the cloud platform and predicts the remaining expected lifespan of the downhole equipment based on this historical data and real-time data. During the prediction of the remaining expected lifespan, it determines the corresponding chamber based on the collected data, retrieves the corresponding periodic environmental change data for that chamber, and determines the final remaining expected lifespan based on this periodic environmental change data and the historical equipment maintenance data. When the downhole equipment reaches its remaining expected lifespan, it uses an edge-cloud collaborative architecture to schedule backup equipment to replace the target equipment. The early warning and status report push module 004 retrieves a pre-trained equipment fault early warning model, inputs real-time downhole data into the equipment fault early warning model, and obtains the corresponding equipment early warning results; it generates a corresponding equipment status report based on the equipment early warning results and the previously collected equipment maintenance history data; it connects the downhole equipment management system to the ground monitoring platform and pushes the equipment status report to the ground operation and maintenance personnel through the remote communication module. The control command pre-verification and issuance module 005 enables a lightweight digital twin engine deployed on the edge node before issuing equipment control commands. It combines real-time data and equipment status reports to simulate the command execution effect. Based on the simulation results, it corrects the equipment operating parameters and then issues the equipment control commands to the downhole equipment.

[0083] This application also discloses a method for remote control of downhole equipment, including a processor, wherein the processor runs a program for the method for remote control of downhole equipment as described in any one of the above embodiments.

[0084] This application also discloses a storage medium storing a program for the remote control method of downhole equipment described in any one of the above embodiments.

[0085] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A method for remote control of downhole equipment, characterized in that, include: Distributed sensors and edge computing nodes are deployed in underground equipment and chambers to construct an edge-cloud collaborative decision-making architecture. Data is transmitted based on a hybrid power line carrier and 5G explosion-proof communication link. The edge nodes read the real-time underground data collected by the distributed sensors to obtain the underground equipment operation data and chamber environment data. The real-time underground data is synchronously uploaded to the cloud platform for storage and the edge node for local storage. The real-time underground data is then compared with the pre-set equipment safety thresholds. The system acquires current operating condition data for the underground equipment, including chamber gas concentration, ambient humidity, equipment load, and tunnel communication strength. It then retrieves historical operating data stored on a cloud platform and determines whether the equipment safety thresholds need adjustment based on the current operating condition data and the historical operating data. If so, the system adjusts the equipment safety thresholds, which include the charger output power limit, the battery swapping robot operating speed limit, and the ventilation system airflow limit. If the real-time data exceeds the equipment's safety threshold, a corresponding alarm mechanism is triggered, and the downhole equipment's operating parameters are adjusted accordingly. Historical equipment maintenance data stored on the cloud platform is retrieved, and the remaining expected service life of the downhole equipment is predicted based on this historical data and the real-time data. During the prediction of the remaining expected service life, the corresponding chamber is determined based on the collected data, and the periodic environmental change data corresponding to that chamber is retrieved. The final remaining expected service life is determined based on this periodic environmental change data and the historical equipment maintenance data. When the downhole equipment reaches its remaining expected service life, a backup device is scheduled to replace the target device through an edge-cloud collaborative architecture. Retrieve a pre-trained equipment failure early warning model, input the real-time downhole data into the equipment failure early warning model, and obtain the corresponding equipment early warning results; Based on the equipment early warning results and the previously collected equipment maintenance history data, a corresponding equipment status report is generated; The downhole equipment management system is connected to the ground monitoring platform, and the equipment status report is pushed to the ground maintenance personnel through the remote communication module; Before issuing equipment control commands, a lightweight digital twin engine deployed on the edge node is activated to simulate the command execution effect by combining the real-time data and the equipment status report; the equipment operating parameters are corrected according to the simulation results, and the equipment control commands are issued to the downhole equipment after the correction is completed.

2. The method for remote control of downhole equipment according to claim 1, characterized in that, In the process of transmitting data through an explosion-proof communication link based on a hybrid power line carrier and 5G network, the method also includes: In a hybrid network, explosion-proof relay nodes are deployed. The edge nodes detect the communication strength of each area in the roadway in real time. When the communication strength of a certain area is detected to be lower than a preset transmission threshold, the dynamic switching mechanism of the explosion-proof relay nodes is automatically triggered to switch the data transmission link to the backup explosion-proof relay node in the coverage area.

3. The method for remote control of downhole equipment according to claim 2, characterized in that, In the process of determining whether to adjust the equipment safety threshold based on the operating condition data and the equipment's historical operating data, the method further includes: The working condition data, including chamber gas concentration, ambient humidity, equipment load, and roadway communication intensity, are assigned weights using a dynamic weighting algorithm. The weight of chamber gas concentration increases as its value increases, while the weights of ambient humidity and equipment load increase as their values ​​exceed the normal range. Based on the equipment fault records under the corresponding weight parameters in the historical operating data of the equipment, the current threshold adjustment demand is calculated. When the demand is higher than the preset judgment value, the equipment safety threshold adjustment is performed.

4. The method for remote control of downhole equipment according to claim 3, characterized in that, In the process of predicting the remaining expected service life of downhole equipment based on the equipment maintenance history data and the real-time data, the method further includes: A lightweight TinyML federated learning model is deployed on the edge node. The lightweight TinyML federated learning model processes the device's running data and maintains historical data locally on the edge node, restricting the uploading of raw data to the cloud platform and only synchronizing the calculation results of the lightweight TinyML federated learning model to the cloud. At the same time, the prediction parameters of the lightweight TinyML federated learning model are corrected by combining the fluctuation patterns of chamber temperature and humidity in the periodic environmental change data.

5. The method for remote control of downhole equipment according to claim 4, characterized in that, The method also includes the following steps in retrieving and applying pre-trained equipment failure early warning models to downhole equipment condition monitoring: At preset intervals, newly collected downhole real-time data and newly added equipment maintenance historical data are input into the equipment fault early warning model to iteratively train and update the equipment fault early warning model; wherein, the preset interval is set according to the downhole equipment operating intensity; During the update process, parameter modules with prediction accuracy higher than the preset accuracy threshold in the equipment fault early warning model are retained, while parameter modules with accuracy lower than the preset accuracy threshold are replaced.

6. The remote control method for downhole equipment according to claim 5, characterized in that, The method also includes: When the real-time data exceeds the equipment safety threshold and the parameter exceeding the threshold is the gas concentration in the chamber, in addition to triggering the alarm mechanism and adjusting the equipment operating parameters, the power input of the high-power charger in the chamber is automatically cut off, the backup ventilation system of the chamber is started, and a high gas emergency alarm message is sent to the ground monitoring platform through the remote communication module. At the same time, the access rights of personnel in the chamber are locked until the gas concentration drops to a safe range.

7. The method for remote control of downhole equipment according to claim 6, characterized in that, The method also includes: Before issuing equipment control commands to the downhole equipment and when the controlled object is a mobile battery swapping robot, the lightweight digital twin engine also combines the tunnel layout data of the current chamber of the robot and the target scheduling chamber, as well as the real-time passage status data, to generate the optimal path for the robot to move across chambers. Meanwhile, the edge node establishes a pre-communication link with the edge node of the target scheduling chamber, and automatically switches the control signal receiving node when the robot enters the target chamber area.

8. A remote control system for downhole equipment, characterized in that, include: The architecture construction and data acquisition comparison module deploys distributed sensors and edge computing nodes in the underground equipment and chambers to build an edge-cloud collaborative decision-making architecture. Data is transmitted based on a hybrid power line carrier and 5G explosion-proof communication link. The edge nodes read the real-time underground data collected by the distributed sensors to obtain the underground equipment operation data and chamber environment data. The real-time underground data is synchronously uploaded to the cloud platform for storage and the edge node for local storage. Then, the real-time underground data is compared with the pre-set equipment safety thresholds. The equipment safety threshold adjustment module acquires current working condition data of the underground equipment, including chamber gas concentration, ambient humidity, equipment load, and roadway communication intensity; retrieves historical equipment operation data stored on the cloud platform, and determines whether the equipment safety threshold needs to be adjusted based on the working condition data and the historical equipment operation data; if so, the equipment safety threshold is adjusted, including the upper limit of charger output power, the upper limit of battery swapping robot operating speed, and the lower limit of ventilation system airflow. The over-threshold response and equipment lifespan management module triggers a corresponding alarm mechanism if the real-time data exceeds the equipment safety threshold, and coordinates with the control system to adjust the operating parameters of the downhole equipment. It retrieves historical equipment maintenance data stored on the cloud platform and predicts the remaining expected lifespan of the downhole equipment based on this historical data and the real-time data. During the prediction of the remaining expected lifespan, it determines the corresponding chamber for the downhole equipment based on the collected data, retrieves the periodic environmental change data corresponding to that chamber, and determines the final remaining expected lifespan based on this periodic environmental change data and the historical equipment maintenance data. When the downhole equipment reaches its remaining expected lifespan, it schedules backup equipment to replace the target equipment through an edge-cloud collaborative architecture. The early warning and status report push module retrieves a pre-trained equipment fault early warning model, inputs the real-time downhole data into the equipment fault early warning model, and obtains the corresponding equipment early warning results. Based on the equipment early warning results and the previously collected equipment maintenance history data, a corresponding equipment status report is generated; The downhole equipment management system is connected to the ground monitoring platform, and the equipment status report is pushed to the ground maintenance personnel through the remote communication module; The control command pre-verification and issuance module enables a lightweight digital twin engine deployed on edge nodes before issuing equipment control commands. It combines the real-time data and the equipment status report to simulate the command execution effect. Based on the simulation results, it corrects the equipment operating parameters and issues the equipment control commands to the downhole equipment after the correction is completed.

9. A remote control system for downhole equipment, characterized in that, Includes a processor, wherein the processor runs a program for the remote control method for downhole equipment as described in any one of claims 1-7.

10. A storage medium, characterized in that, The program stores the remote control method for downhole equipment as described in any one of claims 1-7.

Images