Underground pipe gallery equipment remote control method based on Internet of Things

Abstract

The invention relates to the technical field of the Internet of Things and intelligent control, and discloses an underground pipe gallery equipment remote control method based on the Internet of Things, which comprises the following steps: collecting environment and equipment state data through a heterogeneous sensor network, extracting a dynamic linkage logic rule based on an edge node, constructing a virtual decision node to generate a control candidate instruction set, and a distributed consensus control instruction is formed through a double-domain game model of the energy consumption sensitivity and the risk level. According to the method, the real-time adaptive adjustment of the control threshold and the dynamic verification of the logic consistency between the devices can be realized, so that the reconstruction of the response priority and the conflict avoidance can be completed under the condition that centralized intervention is not needed. The method has the remarkable beneficial effects that the control logic is driven by data to automatically evolve, risk response and energy efficiency optimization are considered in the instruction generation process, the problems of rule stiffness, response lag and resource allocation splitting in an existing system are effectively solved, and a stable, efficient and easy-to-deploy solution is provided for equipment collaboration in a complex scene.

Description

Technical Field

[0001] The present invention relates to a remote control method for underground pipe gallery equipment based on the Internet of Things, and belongs to the technical field of the Internet of Things and intelligent control. Background Art

[0002] As a crucial component of smart city infrastructure, underground utility corridors and their remote equipment control systems play a key role in ensuring urban energy delivery, communication channels, and environmental safety. These corridors typically house critical infrastructure such as electrical cables, communication lines, and gas pipelines. These diverse devices operate under a wide range of dynamic conditions, influenced by factors such as temperature and humidity, gas concentrations, and power fluctuations. To ensure operational continuity and safety, the industry has reached a consensus on the need for stable, efficient, and intelligent remote control methods.

[0003] The current mainstream solution typically adopts a chain structure of "sensor acquisition - central server processing - control command issuance." Sensors are deployed at key locations in the tunnel to collect environmental parameters such as temperature, humidity, and harmful gas concentrations, as well as the operating status of equipment such as fans and pumps. These are then uploaded to the cloud platform, where the central server conducts unified analysis and generates a control strategy. However, this approach has exposed a number of problems in practice, including:

[0004] 1. Response lag: The central server's processing is subject to network transmission delays and computing queue bottlenecks, which causes the system to respond untimely to emergencies (such as gas leaks, equipment overloads, etc.), easily causing accidents to escalate.

[0005] 2. Rigid rules: Control strategies are mostly based on fixed threshold settings or static rule base reasoning. In actual operation, they are difficult to adapt to equipment aging, environmental fluctuations or unknown scenarios, resulting in reduced control accuracy.

[0006] 3. Device conflict: Multi-device control often lacks a coordination mechanism. For example, when multiple high-energy-consuming devices respond to the same command at the same time, it will cause power surges or task scheduling conflicts, and lack the ability to optimize global resources.

[0007] To overcome the above problems, some systems in the industry have attempted to introduce edge computing architectures, reducing response latency by moving some decision-making logic down to local nodes. However, such methods generally still rely on preset control strategies or manually adjusted parameters, lacking a data-driven rule evolution mechanism, making it difficult to achieve true system adaptability. In addition, existing solutions generally fail to effectively integrate the two key variables of equipment energy consumption and risk level. They only prioritize equipment in high-risk scenarios, ignoring the refined control of energy efficiency management in daily operations, which can easily lead to resource waste and control redundancy during long-term operation.

[0008] Therefore, how to construct dynamic linkage rules driven by multi-source data and realize distributed consensus control based on edge nodes while taking into account energy consumption sensitivity and risk response priority has become the technical problem to be solved by the present invention. Summary of the Invention

[0009] The present invention provides a remote control method for underground pipe gallery equipment based on the Internet of Things, the main purpose of which is to solve the problems of response delay, control conflict and rule incompatibility.

[0010] To achieve the above-mentioned object, the present invention provides a remote control method for underground pipe gallery equipment based on the Internet of Things, which is characterized by comprising the following steps:

[0011] Through the heterogeneous sensor network deployed in the underground tunnel, real-time collection of environmental parameters including E={e1,e2,...,e m} and equipment operating status S={s1,s2,...,s n}Multi-dimensional data stream;

[0012] Based on the multi-dimensional data stream, the edge computing node is used to automatically extract the dynamic linkage logic rules between the environment and the device operation. The linkage logic rules are used to adaptively adjust the control threshold of the device according to the real-time environmental changes, where the control threshold T of device i is i Adjustment amount ΔT i Calculated based on the following formula:

[0013] ΔT i =f(E,S)

[0014] Where f is the correlation function learned from historical data, which is used to characterize the impact of environmental parameters E and equipment operating status S on the control threshold of equipment i;

[0015] In each edge computing node, the corresponding physical device is mapped to a virtual decision node, and each virtual decision node generates a candidate instruction set including at least one control instruction based on the received local real-time data stream and the dynamic linkage logic rule;

[0016] The virtual decision nodes collaborate with each other through a logical consistency verification protocol and select the optimal control instructions from their respective candidate instruction sets based on the energy entropy weight-risk dual-domain game model to form a distributed consensus decision for each device in the tunnel. The energy entropy weight-risk dual-domain game model uses the real-time energy consumption sensitivity of the equipment and the real-time risk level of the tunnel as dynamic game parameters, and achieves a balance between risk response priority and energy consumption optimal constraints through an asymmetric Pareto frontier screening algorithm.

[0017] According to the results of the distributed consensus decision, the control priority of each device is dynamically adjusted, and the execution sequence of the control instructions is automatically reconstructed based on the prediction of the event propagation chain. The compatibility of the instruction combination is verified through logical deduction to avoid device action conflicts.

[0018] Preferably, it also includes a self-evolution mechanism of logic rules, specifically: introducing a mapping library of historical events and operation effects, and automatically correcting the weights of the dynamic linkage logic rules through implicit feedback on the operating status of the equipment, so that the control strategy can adaptively evolve as the operating status of the corridor changes.

[0019] Preferably, in the energy entropy weight-risk dual-domain game model, the real-time energy consumption sensitivity of the equipment is obtained through the following steps: based on the historical operation data and real-time energy consumption curve of the equipment, an energy entropy weight coefficient matrix is constructed to quantify the energy consumption sensitivity of different equipment under different environmental conditions.

[0020] Preferably, it also includes a self-feedback calibration mechanism of energy entropy weight, specifically: establishing an implicit feedback loop between device energy consumption and task completion, when the actual energy consumption C of the device is actual Deviation from the energy consumption C predicted based on energy entropy weight predicted When the preset threshold value θ is exceeded, the online correction of the energy entropy weight coefficient is automatically triggered, wherein the predicted energy consumption C predicted Calculated based on the following formula:

[0021] C predicted =g(E,S,W entropy )

[0022] Among them, g is the energy consumption prediction model based on historical data, W entropy The correction process is based on the real-time correlation analysis between the equipment operating status and environmental parameters.

[0023] Preferably, in the step of forming a distributed consensus decision for each device in the tunnel, an asymmetric game model is constructed between the decision nodes, and the energy consumption, response speed and failure probability of the equipment are used as game parameters. The optimal control instruction combination is obtained by solving the Nash equilibrium, thereby achieving global optimization of resource allocation for multiple devices.

[0024] Preferably, it also includes a dynamic priority adjustment algorithm, specifically: adjusting the priority of the control instruction in real time according to the health index of the equipment, wherein the equipment health index is calculated based on the vibration spectrum entropy value and current harmonic distortion rate of the equipment, and high-risk equipment with a lower health index obtains preemptive control authority.

[0025] Preferably, after the step of collecting multi-dimensional data streams including environmental parameters and equipment operating status in real time through a heterogeneous sensor network deployed in the underground tunnel, the extracted environment-equipment dynamic linkage logic rules are further embedded with energy entropy weight coefficients, so that the data stream correlation analysis is extended to the energy consumption dimension.

[0026] Preferably, the virtual decision nodes collaborate through a logic consistency verification protocol, and in the step of selecting the optimal control instruction from their respective candidate instruction sets based on the energy entropy weight-risk dual-domain game model, the candidate instruction set of each virtual decision node adds energy consumption constraints, and realizes the logical closed loop of instruction screening through localized dual-domain game, and the collaborative protocol between nodes adds energy consumption consistency verification to avoid global energy consumption imbalance caused by multi-node decision-making.

[0027] Preferably, in the step of dynamically adjusting the control priority of each device according to the result of the distributed consensus decision, an energy consumption sensitivity threshold is introduced. When the energy consumption entropy weight of the device exceeds the critical value, the elastic priority is automatically obtained. The elastic priority allows the device to temporarily break through the risk level limit within a preset range; the heterogeneous sensor network supports mainstream industrial communication protocols, including Modbus and OPC UA.

[0028] Compared with the problems described in the background technology, the beneficial effects of the present invention are:

[0029] 1. By introducing a multi-source, real-time data perception mechanism at the edge, the dynamic relationship between the environment and device status can be efficiently captured and logically modeled locally, avoiding the response lag and lack of adaptability associated with fixed threshold settings in traditional control strategies. As a continuously evolving intermediary mechanism, linkage logic rules are triggered by the data itself and drive threshold adjustments in real time, rather than relying on manual push from a central node. This significantly improves remote control's situational responsiveness and control strategy flexibility in complex operating conditions.

[0030] 2. By mapping event propagation chains to control instruction sequences, control behaviors can be dynamically sequenced and restructured based on risk levels. A logical deduction mechanism is embedded within the control instruction generation process, eliminating the need for isolated judgments about the execution timing of different devices. Instead, compatibility checks at the global semantic level automatically avoid potential conflicts between devices, ensuring the continuity and coordination of control system execution in emergency situations.

[0031] 3. By solidifying the causal relationship between historical operating data and control effects in the form of feedback in the process of adjusting rule weights, the system has acquired the ability to adaptively adjust to environmental changes. Rule evolution no longer relies on explicit modeling by engineers, but instead completes dynamic iteration through an implicit data learning mechanism, fundamentally alleviating the problems of high maintenance costs and poor adaptability of the rule base, enabling the control system to have the technical characteristics of sustainable optimization in the time dimension, and extracting the energy consumption performance of equipment under different operating conditions into structured parameters with context-aware capabilities, so that energy consumption no longer exists as a static reference quantity, but becomes a real-time driving factor in the game model. By introducing the intermediary dimension of energy entropy weight, the system can adjust the basis for energy consumption decisions according to environmental fluctuations, achieving deep technical coupling between energy efficiency regulation and task scheduling. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 This is the asymmetric Pareto optimization control flow chart driven by energy entropy weight and risk level of the present invention.

[0033] Figure 2 Schematic diagram of the game decision-making mechanism based on energy entropy weight and risk level parameters of the present invention.

[0034] Figure 3 This is a flow chart of the energy consumption and risk collaborative optimization control in the dual-domain game state space of the present invention.

[0035] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0036] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0037] The present application provides an Internet of Things-based remote control method for underground pipe gallery equipment, which includes the following steps:

[0038] Through the heterogeneous sensor network deployed in the underground tunnel, real-time collection of environmental parameters including E={e1,e2,...,e m} and equipment operating status S={s1,s2,...,s n}Multi-dimensional data stream;

[0039] Based on the multi-dimensional data stream, the edge computing node is used to automatically extract the dynamic linkage logic rules between the environment and the device operation. The linkage logic rules are used to adaptively adjust the control threshold of the device according to the real-time environmental changes, where the control threshold T of device i is i Adjustment amount ΔT i Calculated based on the following formula:

[0040] ΔT i =f(E,S)

[0041] Where f is the correlation function learned from historical data, which is used to characterize the impact of environmental parameters E and equipment operating status S on the control threshold of equipment i;

[0042] In each edge computing node, the corresponding physical device is mapped to a virtual decision node, and each virtual decision node generates a candidate instruction set including at least one control instruction based on the received local real-time data stream and the dynamic linkage logic rule;

[0043] The virtual decision nodes collaborate with each other through a logical consistency verification protocol and select the optimal control instructions from their respective candidate instruction sets based on the energy entropy weight-risk dual-domain game model to form a distributed consensus decision for each device in the tunnel. The energy entropy weight-risk dual-domain game model uses the real-time energy consumption sensitivity of the equipment and the real-time risk level of the tunnel as dynamic game parameters, and achieves a balance between risk response priority and energy consumption optimal constraints through an asymmetric Pareto frontier screening algorithm.

[0044] According to the results of the distributed consensus decision, the control priority of each device is dynamically adjusted, and the execution sequence of the control instructions is automatically reconstructed based on the prediction of the event propagation chain. The compatibility of the instruction combination is verified through logical deduction to avoid device action conflicts.

[0045] Preferably, it also includes a self-evolution mechanism of logic rules, specifically: introducing a mapping library of historical events and operation effects, and automatically correcting the weights of the dynamic linkage logic rules through implicit feedback on the operating status of the equipment, so that the control strategy can adaptively evolve as the operating status of the corridor changes.

[0046] Preferably, in the energy entropy weight-risk dual-domain game model, the real-time energy consumption sensitivity of the equipment is obtained through the following steps: based on the historical operation data and real-time energy consumption curve of the equipment, an energy entropy weight coefficient matrix is constructed to quantify the energy consumption sensitivity of different equipment under different environmental conditions.

[0047] Preferably, it also includes a self-feedback calibration mechanism of energy entropy weight, specifically: establishing an implicit feedback loop between device energy consumption and task completion, when the actual energy consumption C of the device is actual Deviation from the energy consumption C predicted based on energy entropy weight predicted When the preset threshold value θ is exceeded, the online correction of the energy entropy weight coefficient is automatically triggered, wherein the predicted energy consumption C predicted Calculated based on the following formula:

[0048] C predicted =g(E,S,W entropy )

[0049] Among them, g is the energy consumption prediction model based on historical data, W entropy The correction process is based on the real-time correlation analysis between the equipment operating status and environmental parameters.

[0050] Preferably, in the step of forming a distributed consensus decision for each device in the tunnel, an asymmetric game model is constructed between the decision nodes, and the energy consumption, response speed and failure probability of the equipment are used as game parameters. The optimal control instruction combination is obtained by solving the Nash equilibrium, thereby achieving global optimization of resource allocation for multiple devices.

[0051] Preferably, it also includes a dynamic priority adjustment algorithm, specifically: adjusting the priority of the control instruction in real time according to the health index of the equipment, wherein the equipment health index is calculated based on the vibration spectrum entropy value and current harmonic distortion rate of the equipment, and high-risk equipment with a lower health index obtains preemptive control authority.

[0052] Preferably, after the step of collecting multi-dimensional data streams including environmental parameters and equipment operating status in real time through a heterogeneous sensor network deployed in the underground tunnel, the extracted environment-equipment dynamic linkage logic rules are further embedded with energy entropy weight coefficients, so that the data stream correlation analysis is extended to the energy consumption dimension.

[0053] Preferably, the virtual decision nodes collaborate through a logic consistency verification protocol, and in the step of selecting the optimal control instruction from their respective candidate instruction sets based on the energy entropy weight-risk dual-domain game model, the candidate instruction set of each virtual decision node adds energy consumption constraints, and realizes the logical closed loop of instruction screening through localized dual-domain game, and the collaborative protocol between nodes adds energy consumption consistency verification to avoid global energy consumption imbalance caused by multi-node decision-making.

[0054] Preferably, in the step of dynamically adjusting the control priority of each device according to the result of the distributed consensus decision, an energy consumption sensitivity threshold is introduced. When the energy consumption entropy weight of the device exceeds the critical value, the elastic priority is automatically obtained. The elastic priority allows the device to temporarily break through the risk level limit within a preset range; the heterogeneous sensor network supports mainstream industrial communication protocols, including Modbus and OPC UA.

[0055] Example 1: In an underground integrated pipe gallery scenario, a heterogeneous sensor network is deployed to continuously collect an environmental parameter set E = {e1, e2, ..., e mEnvironmental parameters include temperature (unit: °C), humidity (unit: %RH), hydrogen sulfide concentration (unit: ppm), oxygen concentration (unit: %), and noise level (unit: dB). Each parameter is derived from standardized industrial sensors. For example, temperature and humidity sensors use SHT31 series modules, while gas sensors use electrochemical modules (such as the MQ series). These sensors are connected to edge nodes via the Modbus RTU protocol.

[0056] At the same time, the equipment operation state set S={s1,s2,...,s n Data is collected from various control devices, such as fans, pumps, and smoke exhaust systems. Status parameters include operating current (unit: A), vibration amplitude (unit: mm / s), power factor (dimensionless), and device start / stop status (Boolean). These parameters are collected and reported synchronously by the power monitoring unit (such as the DTSU666 series) in the device control module.

[0057] The control threshold is adjusted by ΔT i =f(E,S) is the core calculation logic, where f is a nonlinear regression model built using historical data. This model runs locally on the edge node based on the support vector regression (SVR) algorithm. The model training data comes from 30 consecutive days of equipment operation data and matching records of corresponding environmental parameters. Variable T i Indicates the current control threshold of the i-th device, such as the upper limit of the fan speed, the starting frequency of the pump, etc. The unit is consistent with the specific device parameters. ΔT i Indicates the amount of adjustment that needs to be made to the threshold in the current state. The value range is adaptively set within ±20% based on the safety factor.

[0058] Virtual decision nodes are built within edge computing devices. Each node generates a set of candidate control instructions based on the real-time status data of the physical devices it manages and the results of the aforementioned model calculations. Each instruction consists of the instruction type (start, stop, slowdown, etc.), the target device identifier, and the execution condition. Decision nodes exchange candidate instruction sets using a lightweight consistency check protocol that uses a consistent hashing mechanism to ensure logical mutual exclusion between instructions.

[0059] During the instruction screening phase, the system introduces a dual-domain game model to balance energy consumption and risk level, and constructs a game matrix based on the following input parameters:

[0060] Wentropy: Energy entropy weight coefficient, which indicates the sensitivity of each device to overall energy consumption fluctuations. Its value is calculated by the ratio of the standard deviation to the mean of the energy consumption per unit time of the device under different environmental conditions.

[0061] Risklevel: Risk level, constructed by weighting the degree of environmental parameter exceeding the limit, using a fuzzy logic reasoning model to process multiple risk factors;

[0062] Cactual: actual energy consumption, derived from meter data;

[0063] Cpredicted: Predicted energy consumption, calculated using the function Cpredicted = g(E, S, Wentropy), where g is a multivariate regression model built based on historical data, implemented using the Gradient Boosted Tree (GBDT) algorithm. This model is built by regressing the device's historical power curve.

[0064] When the deviation between Cactual and Cpredicted exceeds the dynamically set offset threshold θ (usually in the range of 5% to 10%, and the specific value is set based on the device response time and grid capacity), the system automatically triggers the adjustment of the energy entropy weight coefficient. The adjustment mechanism uses sliding average calibration and moderately suppresses fluctuations based on the real-time risk level to ensure the stability of the overall strategy.

[0065] Before entering the execution queue, all candidate instructions must pass through a logic deduction module to verify whether the execution order is subject to timing conflicts or resource contention. Based on an event propagation chain model, this deduction logic maps possible environmental change events into a time-series trigger chain. State transition diagrams are used to determine the execution compatibility of instructions across multiple devices, thus avoiding power load anomalies caused by the simultaneous activation of multiple high-power devices.

[0066] This implementation also establishes a device health index mechanism, based on the entropy of the device's vibration spectrum and the current harmonic distortion rate. Data is collected in a 10-second cycle, using a fast Fourier transform to analyze the vibration signal in the frequency domain and extract spectral entropy. Current harmonics are filtered and then the distortion rate is calculated. Devices with an index value below a set safety threshold automatically have their control command execution priority increased. This mechanism enables the system to implement protective resource allocation for key devices in a dynamic environment.

[0067] Through the above implementation plan, the present invention realizes the autonomous control logic evolution on the edge side without relying on the intervention of the central server, and enables the control instructions to form a structured balance between energy efficiency and risk avoidance. It is particularly suitable for urban underground pipeline corridor scenarios with high-density equipment distribution and complex environmental fluctuation characteristics.

[0068] Example 2: To further illustrate the specific structure and operation mechanism of the remote control method for underground pipe gallery equipment based on the Internet of Things of the present invention, Figure 1 To the attached Figure 3 , gives the complete process of generating distributed consensus instructions under the coordination of different modules, as detailed below:

[0069] Figure 1 The control flow chart of the asymmetric Pareto optimization driven by energy entropy weight and risk level is shown. In this structure, the energy entropy weight calculation module first analyzes the historical energy consumption curve of the equipment, and then constructs an entropy weight coefficient matrix to measure the energy consumption sensitivity of different equipment under various working conditions. The entropy weight coefficient matrix is passed as one of the inputs to the asymmetric Pareto optimizer. At the same time, the system dynamically evaluates the risk level through real-time environmental parameter mapping, and the risk level assessment module outputs the risk level parameter, both of which serve as input variables of the optimizer. After the candidate instruction set is generated, the asymmetric Pareto optimizer combines the above-mentioned multi-objective optimization input to execute the Nash equilibrium solution process, and finally generates a distributed consensus instruction that meets the dual constraints of energy consumption optimization and risk response.

[0070] Figure 2 The diagram shows a game-based decision-making mechanism based on energy entropy weights and risk level parameters, where the core unit of control decision-making is the game decision point. In this structure, on the one hand, the entropy weight coefficient input is constructed through energy entropy weight calculation, historical energy consumption analysis, and prediction model correction. This parameter reflects the device's sensitive response to energy consumption changes. On the other hand, the risk parameter input is formed through risk level assessment, failure probability monitoring, and emergency response strategy. It is used to describe the multidimensional risk level that the device may face during operation. After receiving the entropy weight coefficient input and risk parameter input, the game decision point first performs Pareto optimization, comprehensively considering the state weights of each device in the energy consumption and risk dimensions. It then solves the optimal control solution using a Nash equilibrium solution algorithm, and finally issues control instructions to the device layer through the instruction output module.

[0071] Figure 3 This paper describes a control flow chart for the collaborative optimization of energy consumption and risk within a dual-domain game state space. The system evolves and outputs control strategies within the dual-domain game state space, starting with the initial game state. Two types of input guide the selection of game paths. Environmental parameter input drives the system into the energy optimization domain, where entropy weight coefficients are calculated to characterize the sensitivity of device energy consumption characteristics. Risk event triggering guides the system into the risk control domain, where risk level assessments are performed to quantify the potential operational risks faced by the device or system. The game input variables generated by these two subdomains are fed into the Pareto optimization module and then into the Nash equilibrium solution step to ensure a trade-off between energy efficiency and risk avoidance during the optimization process. The final consensus instruction output ensures collaborative execution and conflict avoidance among multiple devices based on global semantic consistency.

[0072] Example 3: This example adds the following steps and mechanisms based on the original heterogeneous data collection and edge modeling process: For example, in the threshold adjustment function ΔT i=f(E,S), in order to avoid the redundant interference of variables E and S in high-dimensional space, this embodiment introduces a priori dimensionality reduction mechanism based on principal component analysis (PCA) in the modeling stage, retaining the factors affecting ΔT i The first k components with the most significant changes are selected, where k is selected based on the cumulative contribution rate reaching more than 90%. This processing path clarifies the input dimension of the function f, making the variable have an impact on ΔT i The impact logic has a stable interpretation boundary. In this process, each environmental parameter e i With the state parameter s i All of these maintain unit consistency and timestamp alignment to ensure the rigor and consistency of the input data source, and are all extended implementation methods known to ordinary technicians in this field.

[0073] In the energy consumption prediction process, to improve the generalization ability of the model, this embodiment adopts a three-stage modeling path to clarify the implementation logic of the function g, such as the preprocessing stage: normalizing the environmental parameters E and the equipment status S to ensure that variables of different dimensions enter the same model system; the model training stage: using the gradient boosting tree (GBDT) regression algorithm to fit the historical energy consumption;

[0074] W entropy Input: It is no longer used as a single static weight in the fitting, but as a sample grouping factor to construct a conditional model subset to improve the robustness of the prediction under different operating conditions. entropy The value of is generated by the following path:

[0075]

[0076] Here, σ represents the standard deviation of the device's energy consumption over a continuous 24-hour period, and μ represents its mean. This definition is refreshed in real time before each decision update and aligned with the environmental context to ensure it accurately reflects the actual impact of energy consumption fluctuations on the control model.

[0077] In order to avoid the problem of inconsistent energy consumption constraints among multiple nodes in the distributed consensus generation, a "node energy consumption verification factor" mechanism is added in this embodiment. Specifically, after each virtual decision node generates a candidate control instruction set, it calculates the total energy consumption C of the device under the instruction combination. t , which is determined by the C corresponding to the target device of the candidate instruction. predicted Aggregation is obtained. When multiple nodes engage in a game of logical consistency, only instruction combinations whose total energy consumption difference does not exceed a threshold ∈ are retained. The recommendation ∈ is set to ±5% of the node's average energy consumption. This mechanism effectively curbs global imbalances caused by local optimal solutions and achieves energy consistency across coordinated control nodes by setting a unified energy consumption tolerance band.

[0078] Before a control instruction enters the execution phase, the system introduces an instruction "compatibility state graph," which consists of resource competition relationships among all candidate instructions. Each instruction establishes a bidirectional dependency chain with resources as nodes. If two instructions have mutually exclusive key resources (such as the same power distribution unit or the same pipe section equipment control channel), the corresponding edges in the graph are marked with conflict markers. When the system runs instruction sequence reasoning, it performs topological sorting based on the directed acyclic structure of the graph to ensure that high-priority tasks can be executed first without causing resource deadlocks, thereby enhancing the timing robustness of the control execution sequence.

[0079] Considering that the energy consumption curve of the equipment may fluctuate in the short term due to environmental changes in actual operation, the system has an actual energy consumption of C actual and predicted energy consumption C predicted A dynamic window recalculation mechanism is introduced into the deviation correction mechanism. W is triggered only when the system detects that the deviation of two consecutive cycles is greater than the set threshold θ (for example, 8%), rather than a single deviation event. entropy This processing method effectively filters short-term sudden disturbances and avoids unstable convergence of the game model due to overly sensitive adjustments. It is an extended implementation method known to ordinary technicians in this field.

[0080] Example 4: In this example, to further improve the optimization process and collaborative working mechanism of device control, an improved method based on a dynamic model adjustment and feedback mechanism is proposed. By introducing a multi-level data calibration and model update mechanism, the adaptive capability of the control strategy is optimized, and the efficiency and accuracy of multi-device collaborative work are improved. Specific steps include: Multi-dimensional data calibration and feedback mechanism: During implementation, a real-time calibration mechanism based on device and environmental data is first introduced to ensure that each control node can dynamically adjust its model input based on the latest device status and environmental parameters. During the control threshold adjustment process, the environmental parameters and device status are reduced in dimension using the principal component analysis (PCA) method to avoid redundant interference caused by high-dimensional data and optimize the input dimension of the model. This process helps reduce computational complexity and improve the accuracy and real-time performance of control command generation; Dynamic adjustment of the association between the model and historical data: An adaptive adjustment method based on a dynamic regression model is used to perform real-time analysis of historical data and device operating status. Specifically, by analyzing the correlation between device operating data and environmental parameters over the past 30 days, the threshold change is adjusted to ensure the precise response of the control strategy under different operating conditions. The threshold change (ΔTi) is updated in real time based on historical data and current status changes, thus ensuring the stability and predictability of equipment control.

[0081] Game Model Optimization and Energy Efficiency Management: We further optimized the computational process of the dual-domain game model and introduced a multi-objective optimization algorithm to comprehensively consider the trade-off between device energy efficiency and risk level. During the game decision-making phase, we set energy consumption optimization constraints and risk response priorities, and conducted multiple iterations of optimization to ensure that each device's control instructions strike the optimal balance between energy efficiency and safety. In particular, by adding an energy entropy weight coefficient, the device's sensitivity to energy efficiency and risk response capabilities are dynamically adjusted, ensuring that the overall system energy efficiency and safety levels are simultaneously improved.

[0082] Device coordination and conflict avoidance: After control instructions are generated, a new device coordination and conflict avoidance mechanism has been added. By introducing an event propagation chain model, potential device conflicts are analyzed to ensure that resource contention or power load conflicts do not occur when multiple devices respond simultaneously. A timing deduction module globally verifies the execution order of control instructions after they are generated, preventing conflicts when multiple high-energy-consuming devices are activated simultaneously or when controlling the same pipe section.

[0083] Adaptive feedback mechanism and real-time correction: During energy consumption monitoring and adjustment, a dynamic deviation threshold is established by combining the device's real-time energy consumption data with its predicted energy consumption. When the deviation between actual and predicted energy consumption exceeds the set threshold, the energy entropy weight coefficient is automatically corrected, adjusting the model weights to reduce the impact of short-term fluctuations. This mechanism effectively eliminates instability in the device response process and improves system robustness. These are all extended implementation methods known to those skilled in the art.

[0084] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0085] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A remote control method for underground pipe gallery equipment based on the Internet of Things, characterized in that: The following steps are involved: Through the heterogeneous sensor network deployed in the underground tunnel, real-time collection of environmental parameters including E={e1,e2,...,e m } and equipment operating status S={s1,s2,...,s n }Multi-dimensional data stream; Based on the multi-dimensional data stream, the edge computing node is used to automatically extract the dynamic linkage logic rules between the environment and the device operation. The linkage logic rules are used to adaptively adjust the control threshold of the device according to the real-time environmental changes, where the control threshold T of device i is i Adjustment amount ΔT i Calculated based on the following formula: ΔT i =f(E,S), Where f is the correlation function learned from historical data, which is used to characterize the impact of environmental parameters E and equipment operating status S on the control threshold of equipment i; In each edge computing node, the corresponding physical device is mapped to a virtual decision node, and each virtual decision node generates a candidate instruction set including at least one control instruction based on the received local real-time data stream and the dynamic linkage logic rule; The virtual decision nodes collaborate with each other through a logical consistency verification protocol and select the optimal control instructions from their respective candidate instruction sets based on the energy entropy weight-risk dual-domain game model to form a distributed consensus decision for each device in the tunnel. The energy entropy weight-risk dual-domain game model uses the real-time energy consumption sensitivity of the equipment and the real-time risk level of the tunnel as dynamic game parameters, and achieves a balance between risk response priority and energy consumption optimal constraints through an asymmetric Pareto frontier screening algorithm. According to the results of the distributed consensus decision, the control priority of each device is dynamically adjusted, and the execution sequence of the control instructions is automatically reconstructed based on the prediction of the event propagation chain. The compatibility of the instruction combination is verified through logical deduction to avoid device action conflicts.

2. The remote control method for underground pipe gallery equipment based on the Internet of Things according to claim 1 is characterized in that: It also includes a self-evolution mechanism for logical rules, specifically: introducing a mapping library of historical events and operation effects, and automatically correcting the weights of the dynamic linkage logical rules through implicit feedback on the operating status of the equipment, so that the control strategy can adaptively evolve as the operating status of the corridor changes.

3. The remote control method for underground pipe gallery equipment based on the Internet of Things according to claim 1 is characterized in that: In the energy entropy weight-risk dual-domain game model, the real-time energy consumption sensitivity of the equipment is obtained through the following steps: based on the historical operation data and real-time energy consumption curve of the equipment, an energy entropy weight coefficient matrix is constructed to quantify the energy consumption sensitivity of different equipment under different environmental conditions.

4. The remote control method for underground pipe gallery equipment based on the Internet of Things according to claim 3 is characterized in that: It also includes a self-feedback calibration mechanism for energy entropy weight, specifically: establishing an implicit feedback loop between device energy consumption and task completion, when the actual energy consumption of the device C atctual Deviation from the energy consumption C predicted based on energy entropy weight predicted When the preset threshold value θ is exceeded, the online correction of the energy entropy weight coefficient is automatically triggered, wherein the predicted energy consumption C predicted Calculated based on the following formula: C predicted =g(E,S,W entropy ), Among them, g is the energy consumption prediction model based on historical data, W entropy is the energy entropy weight coefficient.

5. The remote control method for underground pipe gallery equipment based on the Internet of Things according to claim 1 is characterized in that: In the step of forming a distributed consensus decision for each device in the tunnel, an asymmetric game model is constructed between the decision nodes, and the energy consumption, response speed and failure probability of the equipment are used as game parameters. The optimal control instruction combination is obtained by solving the Nash equilibrium.

6. The remote control method for underground pipe gallery equipment based on the Internet of Things according to claim 1 is characterized in that: It also includes a dynamic priority adjustment algorithm, specifically: adjusting the priority of control instructions in real time according to the health index of the equipment, wherein the equipment health index is calculated based on the vibration spectrum entropy value and current harmonic distortion rate of the equipment, and high-risk equipment with a lower health index obtains preemptive control authority.

7. The remote control method for underground pipe gallery equipment based on the Internet of Things according to claim 1 is characterized in that: After the step of collecting multi-dimensional data streams including environmental parameters and equipment operating status in real time through a heterogeneous sensor network deployed in the underground pipeline corridor, the extracted environment-equipment dynamic linkage logic rules are further embedded with energy entropy weight coefficients.

8. The remote control method for underground pipe gallery equipment based on the Internet of Things according to claim 1 is characterized in that: The virtual decision nodes collaborate through a logic consistency verification protocol, and in the step of selecting the optimal control instruction from their respective candidate instruction sets based on the energy entropy weight-risk dual-domain game model, the candidate instruction set of each virtual decision node adds energy consumption constraints, and a logical closed loop of instruction screening is realized through localized dual-domain game, and the collaborative protocol between nodes adds energy consumption consistency verification.

9. The remote control method for underground pipe gallery equipment based on the Internet of Things according to claim 1 is characterized in that: In the step of dynamically adjusting the control priority of each device based on the result of the distributed consensus decision, an energy consumption sensitivity threshold is introduced. When the energy consumption entropy weight of a device exceeds a critical value, the device automatically obtains an elastic priority. The elastic priority allows the device to temporarily break through the risk level limit within a preset range. Heterogeneous sensor networks support mainstream industrial communication protocols, including Modbus and OPC UA.

Images