Tunnel PLC device wireless ad hoc network construction method supporting APP linkage

Abstract

The application discloses a tunnel PLC equipment wireless ad hoc network construction method supporting APP linkage and belongs to the field of tunnel communication, and comprises the following steps: S1, constructing a signal comprehensive attenuation model based on tunnel geometric parameters and obtaining a three-dimensional mixed channel matrix; S2, obtaining a virtual model by hierarchically expressing multi-dimensional data of a tunnel physical topology, a communication state and device operation by using a digital twin model; S3, designing a dual-link switching module supporting PLC communication and WiFi communication switching; and S4, designing multi-protocol conversion and management authority of APP collaborative control. The tunnel PLC equipment wireless ad hoc network construction method supporting APP linkage is adopted, technologies such as laser scanning modeling, digital twin, reinforcement learning and multi-protocol conversion are fused, the communication reliability, real-time performance and anti-interference problems of PLC equipment ad hoc network in a complex tunnel environment are solved, and the whole-process optimization from environment perception to intelligent maintenance is realized.

Description

Technical Field

[0001] This invention relates to the field of tunnel communication technology, and in particular to a method for constructing a wireless self-organizing network for tunnel PLC devices that supports APP linkage. Background Technology

[0002] With the accelerating pace of intelligent and unmanned tunnel development, programmable logic controller (PLC) equipment within tunnels requires the construction of high-density, highly reliable communication networks to support core needs such as real-time monitoring, remote control, and intelligent maintenance. However, the unique curved geometry of tunnels (e.g., curve radii R can be as small as 20 meters), multi-branch topology (branch density D can reach 3-5 per 100 meters), and strong electromagnetic interference environment (e.g., pulse noise spectrum covering 2-100MHz generated by motor start-up and shutdown) pose severe challenges to the environmental adaptability of communication systems. Traditional single communication solutions (such as pure power line communication PLC or wireless communication) and static control modes are no longer sufficient to meet the stringent requirements of tunnel scenarios for communication reliability (success rate >95%), real-time performance (control delay <200ms), and anti-interference capabilities. A systematic solution integrating environmental perception and intelligent algorithms is urgently needed.

[0003] Specifically, existing power line communication (PLC) is based on transmission line theory and optimizes the coupling circuit through impedance matching formulas, but it does not consider the surface reflection effect of tunnel curves. Moreover, the traditional attenuation model is only applicable to straight road scenarios. When there are curves, the signal is additionally attenuated due to surface reflection, and the measured communication success rate drops sharply from 85% on straight roads to below 60% on curves.

[0004] Traditional wireless ad hoc networks use backpressure algorithms to calculate queue differences and optimize routing, but do not introduce curvature attenuation factors. Multipath fading caused by tunnel bends leads to wireless signal outage rates exceeding 30%, and fixed frequency bands (such as 2.4GHz) are susceptible to co-channel interference caused by reflections from metal structures, resulting in spectrum utilization rates consistently below 40%.

[0005] In the field of ad hoc network optimization, existing reinforcement learning algorithms (such as SARSA) are designed only for single links and have two limitations: (1) Static topology assumption: relying on a preset routing table, when a new branch is added to the tunnel or a device fails, the topology map cannot be updated in real time, and the route reconstruction delay exceeds 5 minutes, which leads to increased data congestion. (2) Fixed spectrum allocation: without the integration of spectrum awareness mechanism, the action only includes route selection and does not involve frequency band switching. Co-channel interference leads to a transmission bit error rate as high as 15%. Summary of the Invention

[0006] The purpose of this invention is to provide a method for constructing a wireless self-organizing network for tunnel PLC devices that supports APP linkage, thereby solving the aforementioned technical problems.

[0007] To achieve the above objectives, this invention provides a method for constructing a wireless self-organizing network for tunnel PLC devices that supports APP linkage, comprising the following steps:

[0008] S1. Obtain three-dimensional point cloud data of the tunnel through laser scanning, extract the tunnel geometric parameters, construct a signal comprehensive attenuation model based on the tunnel geometric parameters, and obtain a three-dimensional hybrid channel matrix based on the signal comprehensive attenuation model;

[0009] S2. A virtual model is obtained by using a digital twin model to represent multi-dimensional data of tunnel physical topology, communication status, and equipment operation in layers, and the virtual model is synchronized with the physical topology in real time through a dynamic update mechanism.

[0010] S3. Based on a virtual model, design a dual-link switching module that supports switching between PLC communication and Wi-Fi communication;

[0011] S4. Based on the virtual model and dual-link switching module, design multi-protocol conversion and management permissions for APP collaborative control to realize cross-protocol communication and hierarchical control between PLC and APP.

[0012] Therefore, the present invention employs the above-mentioned method for constructing a wireless self-organizing network for tunnel PLC devices that supports APP linkage, and has the following beneficial effects:

[0013] 1. Complex Environment Modeling and Improved Communication Reliability: By extracting geometric parameters such as tunnel curvature radius and branch density through laser scanning, a comprehensive attenuation model is constructed, including curve reflection attenuation and branch load attenuation. This overcomes the limitation of traditional models that are only applicable to straight roads. In actual tests, the communication success rate on curves increased from 60% to 89%, and on straight roads, it increased to 97%. At the same time, based on the virtual model, physical topology changes are mapped in real time, and the route reconstruction delay is less than 1 second. Compared with traditional static algorithms, the efficiency is improved by more than 90%, ensuring uninterrupted communication during topology changes.

[0014] 2. Reinforcement learning-driven dynamic routing: By introducing a curvature penalty factor and a spectrum allocation function, the reward function integrates multiple objectives such as latency, packet loss rate, and energy efficiency, resulting in a 40% reduction in end-to-end latency and an increase in spectrum utilization from 40% to 75%.

[0015] 3. Dual-link adaptive switching: The PLC / Wi-Fi link is dynamically selected based on channel quality indicators. In areas with strong interference, the PLC anti-interference link is used first, reducing the bit error rate from 15% to 5%.

[0016] 4. Dynamic multi-protocol conversion: The protocol conversion engine based on finite state machine supports real-time mapping between Modbus RTU and MQTT, and the transmission delay of emergency commands is compressed to less than 50ms to meet the real-time requirements of remote control.

[0017] 5. LSTM Fault Prediction: By deploying edge computing units at critical nodes, the probability of device failure can be predicted based on the input vector, providing early warning up to 2 hours in advance and reducing maintenance costs by 30%.

[0018] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0019] Figure 1 This is a flowchart of the method for constructing a wireless self-organizing network for tunnel PLC devices that supports APP linkage according to the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are merely illustrative of the embodiments of the present invention and are not intended to limit the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of this application. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout.

[0021] It should be noted that the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion, such as a process, method, system, product, or server that includes a series of steps or units, not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such process, method, product, or device.

[0022] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0023] like Figure 1 As shown, the method for constructing a wireless self-organizing network for tunnel PLC devices that support APP linkage includes the following steps:

[0024] S1. Obtain three-dimensional point cloud data of the tunnel through laser scanning, extract the tunnel geometric parameters, construct a signal comprehensive attenuation model based on the tunnel geometric parameters, and obtain a three-dimensional hybrid channel matrix based on the signal comprehensive attenuation model;

[0025] Step S1 specifically includes the following steps:

[0026] S11. Extract the following tunnel geometric parameters: tunnel radius of curvature R, branch length L. b Branch density D i,j The straight-line distance l between node i and node j i,j The height H and width W of the tunnel;

[0027] S12, Based on tunnel geometric parameters M G ={R,L b D i,j ,l i,j Construct a comprehensive signal attenuation model for H and W:

[0028] δ i,j =δ B (i,j)+10log 10 (N(f)+1) (1);

[0029] In the formula, δ i,j δ represents the combined signal attenuation between node i and node j; B (i,j) represents the branch load attenuation between node i and node j; N(f) represents the background noise power spectral density at frequency f;

[0030] in,

[0031]

[0032] δ R (i,j)=δ0(i,j)·α(R) N (3);

[0033] In the formula, δ R (i,j) represents the curvature surface reflection attenuation between node i and node j; δ0(i,j) represents the straight-line foundation attenuation, and Both k1 and k2 represent frequency-dependent attenuation coefficients. When using a PLC communication link, k1 = 0.2 and k2 = 0.01; when using a Wi-Fi communication link, k1 = 0.1 and k2 = 0.02. cw Let represent the carrier frequency; α(R) represent the curvature attenuation factor, and N represents the number of times the signal is reflected within the curve, and

[0034] S13. Construct a three-dimensional hybrid channel matrix H = [h] based on the integrated attenuation model. i,j ] n×n Where n×n represents the number of rows and columns of the matrix, h i,j This represents the proportion of signal transmission from node i to node j, and P t and P r These represent the transmit power and receive power, respectively.

[0035] S2. A virtual model is obtained by using a digital twin model to represent multi-dimensional data of tunnel physical topology, communication status, and equipment operation in layers, and a dynamic update mechanism is used to achieve real-time synchronization between the virtual model and the physical topology.

[0036] Step S2 specifically includes the following steps:

[0037] S21. Construct the following virtual model G. D (t):

[0038] G D (t)=(V D ,ε D ,S D (t),C D (t),A D (t)) (4);

[0039] In the formula, V D Let V represent the set of nodes, and V D ={v i |i=1,2,…,n},v i The attribute v represents node i. i =(x i ,y i ,z i ,R i ,T i ), x i ,y i ,z i R represents the three-dimensional coordinates of node i; i R represents the radius of curvature of the curve at the location of node i, and when node i is on a straight section, R i =∞,T i T represents the communication type of node i. i =1,2,3 represent PLC communication, Wi-Fi communication, and relay communication, respectively;

[0040] ε D Denotes the set of communication links, and ε D ={e i,j |i,h∈V D}, e i,j This represents the link attribute from node i to node j, e i,j =(l i,j ,L b D i,j );

[0041] S D (t) represents the device state, and S D(t)=(H(t),Z(t),L(t),E(t),P(t)), where H(t) represents the real-time three-dimensional hybrid channel matrix; Z(t), L(t) and E(t) represent the data queue length vector, load vector and remaining energy vector of the real-time node, respectively; and P(t) represents the permission vector.

[0042] C D (t) represents the communication state, and C D (t)=(H(t),Γ(t),F(t)), where Γ(t) represents the signal-to-noise ratio matrix, Γ(t)=[γ i,j (t)] n×n , γ i,j (t) represents the signal-to-noise ratio from node i to node j at time t, P r,i,j (t) represents the received power, N i,j F(t) represents the noise power, F(t) represents the spectrum allocation matrix, and F(t) = [f alloc (i,j,t)] n×n f alloc (i,j,t) represents the spectrum allocation function from node i to node j at time t, f alloc (i,j,t)∈F={2MHz,5MHz,…,100MHz}, where F represents the set of available frequency bands, and F={2MHz,5MHz,8MHz,…,100MHz}, where the frequency bands increase in MHz increments;

[0043] A D (t) represents the action decision, and A D (t)=(X(t),π(t)), where X(t) represents the routing matrix, X(t)=[x i,j (t)] n×n x i,j (t) represents an element of the routing matrix, and x i,j (t)∈{0,1},∑ j x i,j (t)≤1, when x i,j When π(t) = 1, it indicates that node i selects node j as the next-hop transmission node, π(t) represents the policy function, and S D (t)×C D (t)→A D (t);

[0044] In step S21, the node set V is filtered. D The key nodes are identified, and edge computing units are deployed at these key nodes. A pre-trained LSTM neural network model is used to predict faults. The input vector X of the LSTM neural network model is set as follows: t =[δ i,j(t),T delay (t),P drop (t),E res (t),R j (t),L i The prediction formula is as follows: (t)]

[0045]

[0046] In the formula, X represents the probability of a failure occurring in the next moment. t ,X t-1 ,…,X t-n Let represent the input vectors at times t, t-1, ..., tn, respectively.

[0047] S22. Design the update trigger conditions and update mechanism, where the trigger conditions are as follows:

[0048] (‖ΔG D (t)‖ F >∈)∨(t mod T sync =0) (5);

[0049] In the formula, ΔG D (t) represents the parameter change of the virtual model, and ΔG D (t)=(ΔV D (t),Δε D (t),ΔS D (t),ΔC D (t)), ΔV D (t), Δε D (t), ΔS D (t) and ΔC D (t) represents the changes in node, link, device status, and communication status, respectively; ‖·‖ F Represents the Frobenius norm; ∈ indicates a threshold value; mod indicates the modulo operation; T sync Indicates the timing synchronization period;

[0050] The update mechanism is as follows:

[0051] G D (t)←G D (t-1)+ΔG D (t) (6);

[0052] In the formula, G D (t-1) represents the parameters of the virtual model at the previous time step.

[0053] In step S22, the reward function is used to guide the reinforcement learning routing algorithm to select the optimal route and spectrum resources:

[0054]

[0055] In the formula, r(t) represents the reward function; α, β, γ, and δ all represent weight coefficients, and α + β + γ + δ = 1; T delay (t) represents the end-to-end transmission delay; P drop (t) represents the packet loss rate; Represents the percentage of remaining energy at a node; α(R) j ) represents the curvature penalty factor, and R j Represents the radius of curvature of node j;

[0056] Within the reinforcement learning framework, spectrum allocation, routing, and curvature are jointly optimized:

[0057]

[0058] S(t+1)=f transition (S(t),A(t),ω(t)) (10);

[0059] Q(S(t),A(t))←Q(S(t),A(t))+α lr [r(t)+γ lf ·Q(S(t+1),A(t+1))-Q(S(t),A(t))] (11);

[0060]

[0061] In the formula, δ i,j (f t ) indicates that in the transmission frequency band f t The overall signal attenuation between node i and node j; λ represents the interference weighting coefficient; I(f t ) indicates that in the transmission frequency band f t Real-time interference intensity; f transition (·) denotes the state transition function; ω(t) denotes the random disturbance of the environment; Q(S(t),A(t)) denotes the Q value of performing action S(t) in the current state S(t); Q(S(t+1),A(t+1)) denotes the Q value of performing the next action A(t+1) in the next state S(t+1); α lr G represents the learning rate; lf Indicates the discount factor; a * (t) represents the optimal action; δ p Indicates the penalty coefficient;

[0062] Then update the routing matrix X based on the updated Q-value strategy. new (t).

[0063] S3. Based on a virtual model, design a dual-link switching module that supports switching between PLC communication and Wi-Fi communication;

[0064] The dual-link switching module mentioned in step S3 is a PLC coupling circuit with a dual-link switching mechanism, thereby achieving signal coupling through an LC parallel resonant network; wherein the cutoff frequency f of the PLC coupling circuit is... c and input impedance Z in The design formula is as follows:

[0065]

[0066] In the formula, L and C represent inductance and capacitance, respectively; R s jωL represents the standard impedance; jωL represents the inductive reactance. Indicates the capacitive reactance;

[0067] The expression for the dual-link switching mechanism is as follows:

[0068] Link selection = argmax(CQI) PLC CQI Wi-Fi (14);

[0069] In the formula, CQI represents the channel quality index; CQI represents noise variance. PLC and CQI Wi-Fi These represent the channel quality indicators for PLC communication and Wi-Fi communication, respectively.

[0070] The following power management unit is designed:

[0071]

[0072] In the formula, V out and V in Output voltage and input voltage respectively, and V in ∈[9,36]V; D represents the duty cycle; η represents the power conversion efficiency; P out and P in These represent output power and input power, respectively.

[0073] S4. Based on the virtual model and dual-link switching module, design multi-protocol conversion and management permissions for APP collaborative control to realize cross-protocol communication and hierarchical control between PLC and APP.

[0074] Step S4 specifically includes the following steps:

[0075] S41. Considering the protocol differences between the tunnel equipment layer and the cloud or mobile terminal, design a dynamic transformation engine based on a finite state machine to realize real-time protocol parsing and data format mapping.

[0076] The protocol mapping expression is as follows:

[0077] M:Modbus RTU→MQTT,M(P)=(T,D,QoS) (16);

[0078] In the formula, M: Modbus RTU represents the source protocol; MQTT represents the target protocol; M(P) represents the protocol mapping function, which is used to convert Modbus data P into MQTT format; T represents the mapped MQTT topic; D represents the data payload; QoS represents the quality of service level.

[0079] The following transmission delay model is designed:

[0080] T conv (i,j)=T parse +T trans +T queue +β(R j )·l i,j (17);

[0081] In the formula, T conv (i,j) represents; T parse T trans and T queue These represent protocol parsing time, transmission time, and queue waiting time, respectively; β(R) j ) represents the curve delay coefficient, and β(R) represents the straight section delay coefficient. j When β(R) = 0, β(R) is at the bend. j ) = 0.001;

[0082] S42. Design a permission level management system and an emergency command preemption mechanism using an instruction priority scheduling algorithm; the expression for the instruction priority scheduling algorithm is as follows:

[0083]

[0084] In the formula, P exec Indicates instruction execution priority; P urgency and P auth These represent the urgency and privilege level of the instruction, respectively; ω1, ω2, and ω3 are all weighting coefficients, and ω1 + ω2 + ω3 = 1;

[0085] Upon receiving an emergency command, the following actions are triggered: interrupting the low-priority communication queue, allocating a dedicated PLC communication link, and increasing the transmission speed to compress the transmission delay to within 50ms.

[0086] Simulation Experiment

[0087] Experimental conditions: Tunnel geometric parameters: Straight section: length 500m, width 8m, height 6m; Curved section: radius of curvature 50m (simulating common small-radius curves in tunnels), central angle 90°; Branch parameters: branch density 0.03 branches / m (3 branches per 100m), branch length 20m. Communication parameters: PLC communication link: carrier frequency 10MHz, transmit power 20dBm, noise power spectral density -120dBm / Hz; WiFi communication link: carrier frequency 2.4GHz, transmit power 15dBm, noise power spectral density -110dBm / Hz. Node configuration: 20 PLC device nodes are deployed, randomly distributed in the tunnel, with node spacing ranging from 5-50m, of which 5 nodes serve as relay nodes. Simulation platform: MATLABR2023b+NS-3 co-simulation. Algorithm implementation: Q-Learning algorithm based on reinforcement learning (including curvature penalty factor), LSTM fault prediction model (input dimension 6, hidden layer nodes 32).

[0088] Based on the above experimental conditions, the following results were obtained using the method described in this invention and the traditional back pressure algorithm.

[0089] Table 1 Comparison of Signal Attenuation

[0090]

[0091]

[0092] Table 2 Comparison of Routing Optimization

[0093]

[0094] Table 3 Dual-link switching effect

[0095]

[0096] As shown in Tables 1-3, the communication reliability of this invention is 100% in straight road scenarios and 98% in curved road scenarios (compared to only 60% for traditional methods), meeting the requirement of ≥95%.

[0097] Real-time performance: The average end-to-end latency of this invention is 150ms (compared to 280ms in traditional methods), and the emergency command latency is compressed to 45ms, meeting the <200ms standard. Interference resistance: Spectrum utilization is improved by 71.1%, and the bit error rate is reduced from 15% to 3.2%, verifying the effectiveness of the dual-link switching and dynamic spectrum allocation described in this invention.

[0098] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for constructing a wireless self-organizing network for tunnel PLC devices that supports APP linkage, characterized in that: Includes the following steps: S1. Obtain three-dimensional point cloud data of the tunnel through laser scanning, extract the set of tunnel geometric parameters, construct a signal comprehensive attenuation model based on the set of tunnel geometric parameters, and obtain a three-dimensional hybrid channel matrix based on the signal comprehensive attenuation model; wherein, the set of tunnel geometric parameters includes the radius of curvature of the tunnel, branch length, branch density, straight distance between nodes, and the height and width of the tunnel; S2. A virtual model is obtained by using a digital twin model to represent multi-dimensional data of tunnel physical topology, communication status, and equipment operation in layers, and a dynamic update mechanism is used to achieve real-time synchronization between the virtual model and the physical topology; the multi-dimensional data in the virtual model includes node set, communication link set, equipment status, communication status, and action decision; S3. Based on a virtual model, design a dual-link switching module that supports switching between PLC communication and Wi-Fi communication; S4. Based on the virtual model and dual-link switching module, design multi-protocol conversion and management permissions for APP collaborative control to realize cross-protocol communication and hierarchical control between PLC and APP.

2. The method for constructing a wireless self-organizing network for tunnel PLC devices supporting APP linkage according to claim 1, characterized in that: Step S1 specifically includes the following steps: S11. Extract the following set of tunnel geometric parameters: tunnel radius of curvature R, branch length L. b Branch density D i,j The straight-line distance l between node i and node j i,j The height H and width W of the tunnel; S12, Based on the set of tunnel geometric parameters M G ={R,L b D i,j ,l i,j Construct a comprehensive signal attenuation model for H and W: d i,j =d B (i,j)+10log 10 (N(f)+1) (1); In the formula, δ i,j δ represents the combined signal attenuation between node i and node j; B (i,j) represents the branch load attenuation between node i and node j; N(f) represents the background noise power spectral density at frequency f; in, d R (i,j)=δ0(i,j)·α(R) N (3); In the formula, δ R (i,j) represents the curvature surface reflection attenuation between node i and node j; δ0(i,j) represents the straight-line foundation attenuation, and Both k1 and k2 represent frequency-dependent attenuation coefficients. When using a PLC communication link, k1 = 0.2 and k2 = 0.01; when using a Wi-Fi communication link, k1 = 0.1 and k2 = 0.

02. cw Let represent the carrier frequency; α(R) represents the curvature attenuation factor, and N represents the number of times the signal is reflected within the curve, and S13. Construct a three-dimensional hybrid channel matrix H = [h] based on the integrated attenuation model. i,j ] n×n Where n×n represents the number of rows and columns of the matrix, h i,j This represents the proportion of signal transmission from node i to node j, and P t and P r These represent the transmit power and receive power, respectively.

3. The method for constructing a wireless self-organizing network for tunnel PLC devices supporting APP linkage according to claim 2, characterized in that: Step S2 specifically includes the following steps: S21. Construct the following virtual model G. D (t): G D (t)=(V D ,ε D ,S D (t),C D (t),A D (t)) (4); In the formula, V D Let V represent the set of nodes, and V D ={v i |i=1,2,…,n},v i The attribute v represents node i. i =(x i ,y i ,z i ,R i ,T i ), x i ,y i ,z i R represents the three-dimensional coordinates of node i; i R represents the radius of curvature of the curve at the location of node i, and when node i is on a straight section, R i =∞,T i T represents the communication type of node i. i =1,2,3 represent PLC communication, Wi-Fi communication, and relay communication, respectively; ε D Denotes the set of communication links, and ε D ={e i,j |i,j∈V D }, e i,j This represents the link attribute from node i to node j, e i,j =(l i,j ,L b D i,j ); S D (t) represents the device state, and S D (t)=(H(t),Z(t),L(t),E(t),P(t)), where H(t) represents the real-time three-dimensional hybrid channel matrix; Z(t), L(t) and E(t) represent the data queue length vector, load vector and remaining energy vector of the real-time node, respectively; and P(t) represents the permission vector. C D (t) represents the communication state, and C D (t)=(H(t),Γ(t),F(t)), where Γ(t) represents the signal-to-noise ratio matrix, Γ(t)=[γ i,j (t)] n×n , γ i,j (t) represents the signal-to-noise ratio from node i to node j at time t, P r,i,j (t) represents the received power, N i,j F(t) represents the noise power, F(t) represents the spectrum allocation matrix, and F(t) = [f alloc (i,j,t)] n×n f alloc (i,j,t) represents the spectrum allocation function from node i to node j at time t, f alloc (i,j,t)∈F={2MHz,5MHz,…,100MHz}, where F represents the set of available frequency bands, and F={2MHz,5MHz,8MHz,…,100MHz}, where the frequency bands increase in MHz increments; A D (t) represents the action decision, and A D (t)=(X(t),π(t)), where X(t) represents the routing matrix, X(t)=[x i,j (t)] n×n x i,j (t) represents an element of the routing matrix, and x i,j (t)∈{0,1},∑ j x i,j (t)≤1, when x i,j When π(t) = 1, it indicates that node i selects node j as the next-hop transmission node, π(t) represents the policy function, and S D (t)×C D (t)→A D (t); S22. Design the update trigger conditions and update mechanism, where the trigger conditions are as follows: ||ΔG D (t)|| F >∈)∨(t mod T sync =0) (5); In the formula, ΔG D (t) represents the parameter change of the virtual model, and ΔG D (t)=(ΔV D (t),Δε D (t),ΔS D (t),ΔC D (t)), ΔV D (t), Δε D (t), ΔS D (t) and ΔC D (t) represents the changes in node, link, device status, and communication status, respectively; ||·|| F denoted by Frobenius norm; ∈ indicates a threshold value, and ∈ ∈ (0.05, 0.1); mod indicates the modulo operation; T sync Indicates the timing synchronization period; The update mechanism is as follows: G D (t)←G D (t-1)+ΔG D (t) (6); In the formula, G D (t-1) represents the parameters of the virtual model at the previous time step.

4. The method for constructing a wireless self-organizing network for tunnel PLC devices supporting APP linkage according to claim 3, characterized in that: In step S21, the node set V is filtered. D The key nodes are identified, and edge computing units are deployed at these key nodes. A pre-trained LSTM neural network model is used to predict faults. The input vector X of the LSTM neural network model is set as follows: t =[δ i,j (t),T delay (t),P drop (t),E res (t),R j (t),L i The prediction formula is as follows: (t)] In the formula, X represents the probability of a failure occurring in the next moment. t ,X t-1 ,…,X t-n Let represent the input vectors at times t, t-1, ..., tn, respectively.

5. The method for constructing a wireless self-organizing network for tunnel PLC devices supporting APP linkage according to claim 3, characterized in that: In step S22, the reward function is used to guide the reinforcement learning routing algorithm to select the optimal route and spectrum resources: In the formula, r(t) represents the reward function; α, β, γ, and δ all represent weight coefficients, and α + β + γ + δ = 1; T delay (t) represents the end-to-end transmission delay; P drop (t) represents the packet loss rate; Represents the percentage of remaining energy at a node; α(R) j ) represents the curvature penalty factor, and R j Represents the radius of curvature of node j; Within the reinforcement learning framework, spectrum allocation, routing, and curvature are jointly optimized: S(t+1)=f transition (S(t),A(t),ω(t)) (10); Q(S(t),A(t))←Q(S(t),A(t))+α lr [r(t)+γ lf ·Q(S(t+1),A(t+1))-Q(S(t),A(t))](11); In the formula, δ i,j (f t ) indicates that in the transmission frequency band f t The overall signal attenuation between node i and node j; λ represents the interference weighting coefficient; I(f t ) indicates that in the transmission frequency band f t Real-time interference intensity; f transition (·) denotes the state transition function; ω(t) denotes the random disturbance of the environment; Q(S(t),A(t)) denotes the Q value of performing action A(t) in the current state S(t); Q(S(t+1),A(t+1)) denotes the Q value of performing the next action A(t+1) in the next state S(t+1); α lr γ represents the learning rate; lf Indicates the discount factor; a * (t) represents the optimal action; δ p Indicates the penalty coefficient; Then update the routing matrix X based on the updated Q-value strategy. new (t).

6. The method for constructing a wireless self-organizing network for tunnel PLC devices supporting APP linkage according to claim 3, characterized in that: The dual-link switching module mentioned in step S3 is a PLC coupling circuit with a dual-link switching mechanism, thereby achieving signal coupling through an LC parallel resonant network; wherein the cutoff frequency f of the PLC coupling circuit is... c and input impedance Z in The design formula is as follows: In the formula, L and C represent inductance and capacitance, respectively; R s jωL represents the standard impedance; jωL represents the inductive reactance. Indicates the capacitive reactance of the capacitor; The expression for the dual-link switching mechanism is as follows: Link selection = argmax(CQI) PLC CQI Wi-Fi (14); In the formula, CQI represents the channel quality index; CQI represents noise variance. PLC and CQI Wi-Fi These represent the channel quality indicators for PLC communication and Wi-Fi communication, respectively. The following power management unit is designed: In the formula, V out and V in Output voltage and input voltage respectively, and V in ∈[9,36]V; D represents the duty cycle; η represents power conversion efficiency; P out and P in These represent output power and input power, respectively.

7. The method for constructing a wireless self-organizing network for tunnel PLC devices supporting APP linkage according to claim 6, characterized in that: Step S4 specifically includes the following steps: S41. Considering the protocol differences between the tunnel equipment layer and the cloud or mobile terminal, design a dynamic transformation engine based on a finite state machine to realize real-time protocol parsing and data format mapping. The protocol mapping expression is as follows: M:Modbus RTU→MQTT, M(P)=(T,D,QoS) (16); In the formula, M: Modbus RTU represents the source protocol; MQTT represents the target protocol; M(P) represents the protocol mapping function, which is used to convert Modbus data P into MQTT format; T represents the mapped MQTT topic; D represents the data payload; QoS represents the quality of service level. The following transmission delay model is designed: T conv (i,j)=T parse +T trans +T queue +β(R j )·l i,j (17); In the formula, T parse T trans and T queue These represent protocol parsing time, transmission time, and queue waiting time, respectively; β(R) j ) represents the curve delay coefficient, and β(R) represents the straight section delay coefficient. j When β(R) = 0, β(R) is at the bend. j ) = 0.001; S42. Design a permission level management system and an emergency command preemption mechanism using an instruction priority scheduling algorithm; the expression for the instruction priority scheduling algorithm is as follows: In the formula, P exec Indicates instruction execution priority; P urgency and P auth These represent the urgency and privilege level of the instruction, respectively; ω1, ω2, and ω3 are all weighting coefficients, and ω1 + ω2 + ω3 = 1; Upon receiving an emergency command, the following actions are triggered: interrupting the low-priority communication queue, allocating a dedicated PLC communication link, and increasing the transmission speed to compress the transmission delay to within 50ms.

Images