A Multidimensional Sensor Network Deployment and Collaborative Optimization Method Based on the N00 Method

Abstract

This invention discloses a method for the deployment and collaborative optimization of multi-dimensional sensor networks based on the N00 method, comprising: S1, collecting and preprocessing spatial data fused from multiple sources in the well; S2, establishing a multi-level spatial topology model based on the preprocessed data using the N00 method; S3, establishing a deployment dynamic weight function, and obtaining a deployment optimization model based on the deployment dynamic weight function and the multi-level spatial topology model; S4, obtaining the deployment optimization model based on the deployment dynamic weight function and the multi-level spatial topology model; and S5, optimizing the deployment optimization model using a deep reinforcement learning algorithm based on Transformer. This invention establishes a more accurate spatial model using multi-source data, optimizes the deployment model using a dynamic weight function to form an adaptive deployment scheme, and further optimizes the deployment optimization model to complete collaborative optimization and provide the optimal deployment scheme.

Description

Technical Field

[0001] This invention relates to the field of multidimensional sensor deployment, and in particular to a method for multidimensional sensor network deployment and collaborative optimization based on the N00 method. Background Technology

[0002] In industrial production processes, multi-dimensional sensor deployment is required, which necessitates the establishment of spatial models. Traditional modeling methods lack multi-source data, leading to inaccurate modeling. While existing technologies offer some simple algorithmic deployment methods, dynamically adjusting sensor layout remains a challenge for the industry when the industrial production environment changes rapidly or multiple constraints (such as coverage, connectivity, and cost) overlap. Summary of the Invention

[0003] The purpose of this invention is to provide a multidimensional sensor network deployment and collaborative optimization method based on the N00 method, which aims to solve the problems of spatial modeling in complex environments and adaptive deployment of sensor networks.

[0004] This invention provides a method for the deployment and collaborative optimization of multi-dimensional sensor networks based on the N00 method, including:

[0005] S1. Collect and preprocess spatial data fused from multiple downhole sources.

[0006] S2. Based on the preprocessed data, establish a multi-level spatial topology model based on the N00 construction method;

[0007] S3. Establish a deployment dynamic weight function, and obtain a deployment optimization model based on the deployment dynamic weight function and the multi-level spatial topology model;

[0008] S4. A deployment optimization model is obtained based on the deployment dynamic weight function and the multi-level spatial topology model;

[0009] S5. Optimize the deployment model using a deep reinforcement learning algorithm based on Transformer.

[0010] The embodiments of this invention use multi-source data to establish a spatial model, resulting in a more accurate model. A dynamic weight function is used to optimize the deployment model, and the Transformer deep reinforcement learning algorithm is used to further optimize the deployment model, forming an adaptive deployment scheme.

[0011] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, and in order to make the above and other objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0012] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0013] Figure 1 This is a flowchart of a multi-dimensional sensor network deployment and collaborative optimization method based on the N00 method according to an embodiment of the present invention;

[0014] Figure 2 This is a specific schematic diagram of spatial modeling for a multi-dimensional sensor network deployment and collaborative optimization method based on the N00 method according to an embodiment of the present invention;

[0015] Figure 3 This is a schematic diagram of point cloud data triangulation for spatial modeling of a multidimensional sensor network deployment and collaborative optimization method based on the N00 method according to an embodiment of the present invention.

[0016] Figure 4 This is a schematic diagram of boundary smoothing processing for spatial modeling of a multidimensional sensor network deployment and collaborative optimization method based on the N00 method according to an embodiment of the present invention;

[0017] Figure 5 This is a schematic diagram of the node relationships in spatial modeling of a multidimensional sensor network deployment and collaborative optimization method based on the N00 method according to an embodiment of the present invention;

[0018] Figure 6 This is a schematic diagram of the NURBS curve for spatial modeling of a multidimensional sensor network deployment and collaborative optimization method based on the N00 method according to an embodiment of the present invention.

[0019] Figure 7 This is a schematic diagram of the design of the dynamic weight function and the sensor deployment strategy of a multi-dimensional sensor network deployment and collaborative optimization method based on the N00 method according to an embodiment of the present invention.

[0020] Figure 8 This is a schematic diagram of sensor network deployment according to an embodiment of the present invention, which illustrates a multi-dimensional sensor network deployment and collaborative optimization method based on the N00 method.

[0021] Figure 9 This is a schematic diagram illustrating the specific deployment of a multi-dimensional sensor network deployment and collaborative optimization method based on the N00 method according to an embodiment of the present invention.

[0022] Figure 10This is a schematic diagram of the multidimensional sensor network collaborative optimization process according to an embodiment of the present invention, which describes a multidimensional sensor network deployment and collaborative optimization method based on the N00 method.

[0023] Figure 11 This is a schematic diagram of the collaborative optimization algorithm of a multi-dimensional sensor network deployment and collaborative optimization method based on the N00 method according to an embodiment of the present invention. Detailed Implementation

[0024] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] Method Implementation Examples

[0026] According to embodiments of the present invention, a spatial modeling method for sensor network deployment is provided, such as... Figure 1 As shown, it specifically includes:

[0027] The N00 method is a pillarless mining technology that achieves efficient and safe coal mining through roof cutting for pressure relief and self-forming roadways. However, the spatial structure of the goaf and roadway-working face under this method is extremely complex, and traditional modeling methods are insufficient to meet the requirements for refined sensor network deployment. Therefore, this invention proposes the following solution.

[0028] like Figure 2 As shown;

[0029] S1. Collect spatial data fused from multiple sources and perform preprocessing;

[0030] High-density point cloud data is acquired from the goaf and roadway-working face using 3D laser scanning (LiDAR) equipment to obtain accurate 3D spatial morphology. An inertial navigation system (INS) is mounted on a mobile device (drone or robot) to acquire spatial position and attitude information, compensating for the limitations of LiDAR in dynamic environments. Simultaneously, total station measurements are used to precisely measure key nodes, obtaining high-precision geographic coordinates as a benchmark for data calibration and verification.

[0031] S1 specifically includes: collecting spatial data from multi-source data fusion, performing timestamp calibration on the spatial data, unifying the spatial data into a coordinate system to obtain point cloud data, and performing denoising and fusion processing on the point cloud data to obtain the first point cloud data.

[0032] Timestamp calibration is performed on data collected from different devices to ensure data fusion under the same time reference. A unified mine coordinate system is established, and data from various sources are transformed to the unified coordinate system to eliminate coordinate differences. An improved adaptive filtering algorithm is used to denoise the point cloud data, eliminating random noise and redundant points generated during the acquisition process and improving data quality.

[0033] The process of denoising and fusing the point cloud data to obtain the first point cloud data specifically includes: using an improved adaptive filtering algorithm to denoise the point cloud data and then performing a weighted average fusion to obtain the first point cloud data. The formula for the improved adaptive filtering algorithm is as follows:

[0034]

[0035] Where, σ i Let σi be the filtering threshold for the i-th point cloud, σ0 be the initial filtering threshold, and ρ be the filtering threshold for the i-th point cloud. i Let ρ be the local point cloud density of the i-th point cloud, ρ0 be the global average point cloud density, and β be the adjustment coefficient.

[0036] The algorithm dynamically adjusts the filtering threshold based on the point cloud density, avoiding the loss of details caused by over-filtering.

[0037] Furthermore, a weighted average fusion method is adopted to fuse LiDAR, INS, and total station data, thereby improving the accuracy and completeness of spatial data.

[0038] The formula for calculating the weighted average fusion weight coefficient is as follows:

[0039]

[0040] Among them, w i Let σ be the weight of the i-th data source. i For the precision of the i-th data source, σ j Let n be the precision of the j-th data source, and n be the number of data sources.

[0041] The weighting coefficient w is calculated using the above weighted average fusion weighting coefficient formula. i By assigning greater weight to higher-quality point clouds from various sensor data sources, high-precision 'fused point cloud data' can be formed.

[0042] The fused high-precision point cloud can effectively reduce the risk of node misalignment caused by measurement errors, providing more reliable raw data support for node matching and curve / surface fitting.

[0043] S2. Based on the preprocessed data, establish a multi-level spatial topology model;

[0044] S2 specifically includes: defining the first node and the first edge based on the first point cloud data, constructing a multi-dimensional topology graph model based on the first node and the first edge, establishing the relationship between different layers, and establishing a multi-level spatial topology structure model based on the relationship.

[0045] Based on the application scenario, new nodes and edges are first defined according to the technological characteristics and actual site conditions of the N00 method. Nodes (V) are defined as key points such as goaf boundary points, roadway intersections, working face endpoints, and support structure locations. Edges (E) are established between nodes according to the actual connection relationships, including straight line segments, curved segments, and curved surface segments, reflecting the continuity and complexity of the spatial structure.

[0046] Secondly, spatial topological relationships are constructed: a three-dimensional topological graph model G = (V, E) is built, where V is the set of nodes and E is the set of edges. The model includes the goaf layer, the roadway-working face layer, and the equipment layer. Relationships between different layers are established, such as the spatial relationship between the goaf and the roadway-working face, and the spatial relationship between equipment and roadways, facilitating the formulation of sensor deployment strategies.

[0047] Finally, the spatial topology is optimized, and an automatic matching algorithm based on three-dimensional spatial density clustering is proposed to address the node mismatch problem caused by measurement errors.

[0048] The process of defining a first node and a first edge based on the first point cloud data, constructing a multi-dimensional topological graph model based on the first node and the first edge, establishing the relationship between different layers, and establishing a multi-level spatial topological structure model based on the relationship specifically includes:

[0049] Based on the first point cloud data, the first node and the first edge are defined, and a multi-dimensional topology graph model is constructed based on the first node and the first edge. Considering the characteristics of complex shape and irregular boundary of a certain layer, the point cloud data of the goaf area is obtained from the first point cloud data, and a boundary modeling method based on the improved α-shape algorithm is used to establish the topology graph model.

[0050] like Figure 3 As shown;

[0051] The method for establishing a topology model using the boundary modeling method based on the improved α-shape algorithm specifically includes: triangulating the point cloud data of the goaf area to generate an initial triangular mesh, introducing an adaptive α parameter, dynamically adjusting the α value according to the point cloud density and spatial distribution, and establishing a topology map;

[0052] The formula for calculating the adaptive α value is as follows:

[0053]

[0054] Where: α i : The α value of the i-th region, α0: the initial α value, ρi ρ represents the point cloud density of the i-th region. max : Maximum point cloud density in the region; γ: Adjustment coefficient, which controls the sensitivity of the α value to density.

[0055] like Figure 4 As shown;

[0056] The topological graph is smoothed using the Laplace smoothing algorithm to obtain the topological graph model. The Laplace smoothing algorithm is as follows:

[0057]

[0058] Among them, P i new : Updated node position, P i old : Original node position, λ: Smoothing factor, ranging from 0 to 1, k is the number of neighboring nodes of node i, P j Let be the coordinates of the j-th neighbor node.

[0059] like Figure 5 As shown;

[0060] Calculate the local density ρ of each node i The formula is as follows, ρ i The number of nodes within a certain radius r;

[0061]

[0062] Where δ() is the Dirac function, N is the total number of nodes, and P i : The coordinates of the i-th node, P j : The coordinates of the j-th node, P i Used to fix a certain node, P j Used to compare all nodes;

[0063] Set density threshold ρ th Nodes in high-density regions are selected; these high-density nodes are then clustered, with the cluster centers serving as new node locations. The clustering node condition formula is as follows:

[0064] ||P i -P j ||≤δ;

[0065] Where δ is the tolerance threshold;

[0066] For curve and surface structures, a multi-level spatial topology model is established using a curve and surface fitting method based on non-uniform rational B-splines.

[0067] like Figure 6 As shown;

[0068] The formula for the NURBS curve is as follows:

[0069]

[0070] Where N i,p (u) is the i-th basis function, w i P represents the weight of the i-th node in the curve and surface structures. i Let be the coordinates of the i-th node of the curve and surface structure, u be a parameter variable defined within the node vector range, n be the number of nodes in the curve and surface structure, and p be the order of the B-spline curve. min ,u max Let ] represent the range of values ​​for the parameter variable u, establish the positional relationships between different layers, and build a multi-level spatial topology model based on these relationships. Typically, the connection between layers is determined by calculating coordinate distances, overlap, and adjacency.

[0071] This invention establishes a multi-level three-dimensional topological model of goaf, roadway-working face and equipment, constructs cross-level correlation relationships, and supports the precise deployment of multi-dimensional sensor networks.

[0072] First, the layers are divided as follows: the goaf layer describes the spatial morphology, boundary features, and internal structural information of the goaf. The roadway-working face layer describes the spatial location, shape, orientation, and topological connections of the roadways and working faces. The equipment layer describes the location, type, function, and attributes of the underground equipment.

[0073] Secondly, cross-layer correlation is performed: by determining the spatial relationship between the goaf, roadways, and working faces, high-risk areas and key monitoring areas are identified. Based on the layout of the equipment in the roadways, the deployment requirements of the equipment monitoring sensors are determined.

[0074] Finally, the data structure of the topology model was designed: an object-oriented data structure was adopted to encapsulate spatial entities and topological relationships at different levels, which facilitates the automated generation of sensor deployment strategies.

[0075] This method can accurately fit complex spatial curves and surfaces, reflecting the actual morphology of the goaf and roadway-working face under the N00 method.

[0076] Based on the construction of a complex spatial topology model of the goaf and roadway-working face under the N00 method, the efficient and accurate deployment of a multi-dimensional sensor network has become crucial to ensuring safe mine production. Because underground operating parameters (such as roof pressure, gas concentration, and equipment density) change in real time, traditional static deployment strategies are ill-suited to the dynamic environment. Therefore, this invention proposes a sensor deployment strategy based on a dynamic weighting function, incorporating real-time operating parameters into the deployment decision to achieve adaptive optimization of the sensor network.

[0077] like Figure 7 As shown;

[0078] A deployment dynamic weight function is established, and a deployment optimization model is obtained based on the deployment dynamic weight function and the multi-level spatial topology model.

[0079] By combining real-time operating parameters with sensor deployment requirements, the deployment priority of sensors can be dynamically adjusted according to the risk level and monitoring needs of different areas.

[0080] Firstly, considering the special working conditions under the N00 method, key parameters that have a significant impact on mine safety are selected, including:

[0081] Top plate pressure (P) top ): Reflects the stability of the goaf and the roof of the roadway.

[0082] Gas concentration (C) gas ): Affects the risk of gas explosion and poisoning.

[0083] Equipment density (D) equip ): Reflects the quantity and importance of equipment in the area, affecting the need for equipment monitoring.

[0084] Temperature change rate (ΔT): serves as an early warning system for the risk of spontaneous combustion and fire.

[0085] The establishment and deployment of the dynamic weight function specifically includes:

[0086] Obtain real-time operating parameters and standardize them using the following formula:

[0087]

[0088] Among them, F k For the k-th real-time operating condition parameter, and These are the minimum and maximum values ​​of the parameter, respectively.

[0089] A dynamic weighting function is established based on real-time operating parameters, as shown in the following formula:

[0090]

[0091] w s (i) represents the dynamic weight of the s-th type of sensor at spatial location i; w 0s α represents the initial weight of the s-th type of sensor, reflecting its importance; sk φ(f) represents the sensitivity coefficient of the s-th type sensor to the k-th operating condition parameter; k(i) is the influence function of the operating parameters, reflecting the form of the influence of the parameters on the weights; β is the adjustment index, which controls the nonlinearity of the weight growth.

[0092] The deployment optimization model derived from the deployment dynamic weight function and multi-level spatial topology model specifically includes: the deployment optimization model derived from the deployment dynamic weight function and multi-level spatial topology model, and the formula of the deployment optimization model is as follows:

[0093]

[0094] Z represents the objective function, which comprehensively considers monitoring effectiveness and deployment cost; N represents the total number of spatial locations to be deployed; S represents the total number of sensor types; and w... s (i) represents the dynamic weight of the s-th type of sensor at spatial location i, x is Let x be a binary variable. If the s-th type of sensor is deployed at spatial location i, then x is =1; otherwise x is =0, c s Let λ be the unit cost of the s-th type of sensor, and λ be the trade-off coefficient that balances monitoring effectiveness with deployment cost.

[0095] Set constraints to ensure that critical areas are effectively covered by the corresponding types of sensors;

[0096] 1. The formula for covering constraints is as follows:

[0097] Wherein: S i Let N(i) be the set of sensor types required for spatial location i; let N(i) be the set of neighborhood locations of spatial location i, and let x be the coverage area of ​​the sensor. js Let x be a binary variable. If the s-th type of sensor is deployed at location j, then x js =1; otherwise x js =0,;

[0098] If the s-th type of sensor effectively covers spatial location i at spatial location j, then Otherwise, it is 0; θ i : Minimum coverage required for spatial location i;

[0099] 2. Connectivity constraints:

[0100] Based on the spatial topology model and the communication range of the sensors, a connectivity matrix between nodes is constructed, with the connectivity constraint condition: network connectivity ≥ δ;

[0101] The network connectivity is calculated using the algebraic connectivity of the minimum cut set or the Laplace matrix, where δ is a preset connectivity threshold.

[0102] 3. Deployment location constraints:

[0103]

[0104] Where, N valid Let i be the set of deployable spatial locations.

[0105] Dynamic weight function w s In (i), f k (i) represents the operating parameters (such as roof pressure, gas concentration, equipment density, etc.) at spatial location i. These operating parameters are closely related to the spatial topology: the roof pressure value is usually measured according to the roadway-working face node or the goaf boundary node (i.e., the stress applied at the topological node). Equipment density D equip The statistics need to be based on the topology and equipment location within the equipment layer. Gas concentration C gas Measurements or estimates also need to be made at the corresponding spatial nodes.

[0106] In summary, "position i is backed by topological node / mesh coordinates, and load case f..." k (i) Obtained from the topology model and real-time monitoring data. The two are combined in the dynamic weighting function, making w... s (i) It combines spatial location dependence and real-time operational condition dependence. The combination of "spatial topology" and "dynamic weights" is mainly reflected in the following three points:

[0107] ① Topology as the basis for positioning: This allows each location i to determine its coordinates and adjacency information within the goaf or roadway; ② Correspondence between working parameters and this location: These working parameters f k (i) Assignment needs to be based on the mapping of nodes / mesh in the topology; ③ Dynamic weights w s (i) It can calculate the deployment priority of sensor type s based on the real-time operating parameters of topological location i.

[0108] Dynamic weight function part w s (i) is responsible for determining the attractiveness (i.e., deployment value) of the sensor s to the topological location i. The spatial topology then ensures "where deployment is allowed and how to maintain network coverage and communication" through coverage constraints N(i) and connectivity constraints.

[0109] "Spatial topology" and "dynamic weights" are combined in the objective function (using "weights" to calculate deployment benefits) and constraints (using "topological adjacency" and network connectivity thresholds to maintain coverage and communication stability). Spatial topology determines "deployable range + coverage / connectivity requirements", while dynamic weights determine "value weights for each location and sensor type". Both are incorporated into the objective function and constraint system, enabling the overall deployment to have "spatiotemporal adaptive" capabilities.

[0110] like Figure 8 and 9 As shown;

[0111] In response to the unique risks in goaf areas, roadways-working faces, and densely equipped areas under the N00 method, a sensor deployment strategy for key areas is proposed to improve the accuracy and real-time performance of safety monitoring.

[0112] Regarding goaf areas, high-precision stress sensors and microseismic monitoring equipment should be prioritized for deployment in high-stress areas of the goaf roof. This is based on real-time roof pressure parameters P. top Dynamically adjust the sensor weights w stress (i) High-stress areas have higher weighting. Gas sensors are deployed in the upper corners and blind alleys where gas tends to accumulate. The sensitivity coefficient α of the gas sensor to gas concentration is... gas The settings are relatively high to ensure priority deployment in high-gas areas.

[0113] In the roadway-working face area, equipment status monitoring sensors, such as temperature sensors and current sensors, are deployed in areas with dense equipment. This is based on the equipment density D. equip Adjust sensor weights w equip (i) Deploy temperature and humidity sensors and hazardous gas sensors in areas with frequent human activity. Adjust the weights of the relevant sensors based on the temperature change rate ΔT and humidity parameters.

[0114] To ensure the integrity of the communication network, relay nodes are deployed in areas with severe signal attenuation to guarantee network connectivity. The deployment weights of relay nodes are adjusted based on signal attenuation parameters in the spatial topology model. Backup nodes are also deployed near critical nodes to improve network robustness and resilience. The locations of backup nodes are planned by analyzing the network's critical nodes (highly concentrated data aggregation points).

[0115] This invention also provides a method for optimizing the deployment of multidimensional sensor networks, namely, an optimization deployment model based on Transformer.

[0116] This paper presents a deep reinforcement learning algorithm (DQN-Trans) based on the Transformer architecture and an improved multi-objective optimization algorithm. It fully utilizes dynamic weight functions and spatial topology models to achieve real-time adaptive adjustment of sensor network deployment and solve for the global optimum.

[0117] The multidimensional sensor network deployment problem under the N00 method is modeled as a sequential decision problem in deep reinforcement learning (DRL). By leveraging the learning ability of agents in dynamic environments, the sensor deployment strategy can be optimized in real time.

[0118] like Figure 10 As shown:

[0119] First, the state space:

[0120] The state vector st is defined as follows: at time step t, the state st includes the current sensor deployment status, real-time operating parameters, and spatial topology information.

[0121] s t =[X t ,F t G t ];

[0122] in:

[0123] Sensor deployment matrix, A sensor of type s was deployed at location i.

[0124] The operating condition parameter matrix contains the real-time operating condition parameters at each location i.

[0125] G t Spatial topology information, including the connection relationships between nodes and edges.

[0126] Next, define the ActionSpace:

[0127] Action set A represents the action set A of an agent that can choose to deploy or remove a sensor of a certain type s at a certain location i at each time step, or leave it unchanged.

[0128] a t =(i,s,a) op );

[0129] Among them, a op ∈{deploy, remove, keep}.

[0130] Simultaneously, a reward function was designed, incorporating a dynamic weight function to balance monitoring effectiveness, deployment costs, and network connectivity.

[0131] R t =ΔZ t -η·Penalty t ;

[0132] Where: ΔZ t =Z t -Z t -1: Optimize the increment of the objective function, Z t Z represents the objective function value at time step t. t The definition is consistent with the aforementioned optimization objective function:

[0133] Penalty: Penalties for violating constraints, including covering constraints and connectivity constraints.

[0134] η: Penalty coefficient, which controls the degree of influence of the penalty term.

[0135] The final goal is to maximize cumulative rewards:

[0136]

[0137] Where π is the strategy, γ∈[0,1) is the discount factor, and T is the total number of time steps.

[0138] Specifically, this includes: optimizing the deployment and model of deep reinforcement learning algorithms based on Transformer.

[0139] The system acquires real-time operating parameters, converts the state vectors of the real-time operating parameters, deployment optimization model, and multi-level spatial topology model into embedding vectors, adds position encoding to the embedding vectors to obtain the embedding position vectors, inputs the embedding position vectors into the multi-head self-attention layer to extract global features, performs nonlinear transformation on the global features based on the feedforward neural network, outputs evaluation values, and selects the one with the highest evaluation value as the optimal deployment model.

[0140] Under the N00 method, a deep Q-network (DQN-Trans) based on the Transformer architecture is introduced. By leveraging the powerful global feature extraction capability of the Transformer, the limitations of traditional DQN in processing large-scale, high-dimensional state spaces are addressed.

[0141] The building blocks of the DQN-Trans model are as follows:

[0142] Input layer: The state vector s t The data is converted into an embedding vector, including the sensor deployment status, operating parameters, and spatial topology information. Location encoding is then added to preserve the positional information of each element in the sequence.

[0143] The Transformer encoder uses a multi-head self-attention layer to extract global features from the input sequence. A feed-forward network is then used to perform a non-linear transformation on the output of the attention layer.

[0144] The action value function Q(s) of the output layer t a t The output (θ) corresponds to each possible action a. t The Q value.

[0145] like Figure 11 As shown;

[0146] The algorithm's execution flow is as follows:

[0147] Step 1: Initialize the parameters θ of the DQN-Trans model.

[0148] Initialize the experience replay pool D.

[0149] Step 2: Interaction and Data Collection;

[0150] For each time step t: obtain the current state s from the environment. t .

[0151] Choose action a according to the greedy strategy. t :

[0152] ∈: Exploration rate; used to determine whether to take a random action (exploration) or take the optimal action given by the network in the current state.

[0153] Perform action a t Receive reward R t and the next state s t+1 R t Instant rewards; consisting of "monitoring effectiveness - cost - penalty (violation of coverage / connectivity, etc.)".

[0154] (s) t ,a t ,R t ,s t+1 Store it in the experience replay pool D.

[0155] Step 3: Model training;

[0156] For each training step:

[0157] Randomly sample a small batch of data from D {(s) j ,a j ,R j ,s j+1 )}.

[0158] Calculate the target value y j :

[0159] a' indicates the next action;

[0160] Here, θ represents the parameters of the target network, which are updated periodically.

[0161] Minimize the loss function L(θ):

[0162] Mini-batch size: denoted as |B|;

[0163] Update the model parameters θ.

[0164] Step 4: Parameter update;

[0165] The sensor deployment strategy is as follows:

[0166] In response to the unique risks in goaf areas, roadways-working faces, and densely equipped areas under the N00 method, a sensor deployment strategy for key areas is proposed to improve the accuracy and real-time performance of safety monitoring.

[0167] Regarding goaf areas, high-precision stress sensors and microseismic monitoring equipment should be prioritized for deployment in high-stress areas of the goaf roof. This is based on real-time roof pressure parameters P. top Dynamically adjust the sensor weights w stress (i) High-stress areas have higher weighting. Gas sensors are deployed in the upper corners and blind alleys where gas tends to accumulate. The sensitivity coefficient α of the gas sensor to gas concentration is... gas The settings are relatively high to ensure priority deployment in high-gas areas.

[0168] In the roadway-working face area, equipment status monitoring sensors, such as temperature sensors and current sensors, are deployed in areas with dense equipment. This is based on the equipment density D. equip Adjust sensor weights w equip (i) Deploy temperature and humidity sensors and hazardous gas sensors in areas with frequent human activity. Adjust the weights of the relevant sensors based on the temperature change rate ΔT and humidity parameters.

[0169] To ensure the integrity of the communication network, relay nodes are deployed in areas with severe signal attenuation to guarantee network connectivity. The deployment weights of relay nodes are adjusted based on signal attenuation parameters in the spatial topology model. Backup nodes are also deployed near critical nodes to improve network robustness and resilience. The locations of backup nodes are planned by analyzing the network's critical nodes (highly concentrated data aggregation points).

[0170] The implementation process is designed as follows:

[0171] Step 1: Obtain the latest operating condition parameter data through the existing monitoring system.

[0172] Step 2: Calculate the weights of each location and type of sensor using a dynamic weighting function.

[0173] Step 3: Solve for the optimal sensor deployment scheme.

[0174] Step 4: Based on the solution results, actually deploy the sensors or adjust the configuration of existing sensors.

[0175] Step 5: Regularly update operating parameters, recalculate weights, adjust deployment plans, and achieve dynamic optimization.

[0176] The beneficial effects of this invention are as follows:

[0177] This invention proposes a three-dimensional spatial topology modeling method based on multi-source data fusion, which solves the modeling problem of complex spatial structures of goaf and roadways under the N00 method.

[0178] By incorporating real-time operating parameters into the sensor deployment strategy, a dynamic weighting function was proposed to achieve adaptive optimization of sensor deployment.

[0179] An optimization model was established that comprehensively considers monitoring effectiveness and deployment costs to optimize sensor deployment.

[0180] For the first time, a Transformer-based deep reinforcement learning algorithm was applied to sensor network optimization, and the DQN-Trans algorithm was proposed, which realizes real-time adjustment and global optimization of sensor deployment strategy.

[0181] 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 the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions to the technical solutions of the embodiments of the present invention do not cause the essence of the corresponding technical solutions to deviate from the scope of the present solution.

Claims

1. A method for deploying and collaboratively optimizing multi-dimensional sensor networks based on the N00 method, characterized in that, include: S1. Collect and preprocess spatial data fused from multiple downhole sources. S2. Based on the preprocessed data, establish a multi-level spatial topology model based on the N00 construction method; S3. Establish a deployment dynamic weight function, and obtain a deployment optimization model based on the deployment dynamic weight function and the multi-level spatial topology model; S4. Optimize the deployment model using deep reinforcement learning algorithms based on Transformer; S2 specifically includes: defining a first node and a first edge based on the first point cloud data; constructing a multi-dimensional topology graph model based on the first node and the first edge; establishing the association relationship between different layers; and establishing a multi-level spatial topology structure model based on the N00 construction method according to the association relationship. The process of defining a first node and a first edge based on the first point cloud data, constructing a multi-dimensional topological graph model based on the first node and the first edge, establishing relationships between different layers, and establishing a multi-level spatial topological structure model based on these relationships specifically includes: Based on the first point cloud data, the first node and the first edge are defined. The first node includes: the boundary point of the goaf, the intersection point of the roadway, the end point of the working face and the location of the support structure. The first edge is the connection between the first nodes. A multi-dimensional topology model is constructed based on the first node and the first edge. Based on the first side, establish the goaf layer, roadway-working face layer and equipment layer. The goaf layer describes the spatial morphology, boundary features and internal structural information of the goaf. The roadway-working face layer describes the spatial location, shape, orientation and topological connection of the roadway and working face. The equipment layer describes the location, type, function and attributes of the underground equipment. Establish the positional relationships between the goaf layer, the roadway-working face layer, and the equipment layer, that is, calculate the coordinate distance, overlap, and adjacency between different layers; Considering the complex shapes and irregular boundaries in each positional relationship, the local density of each node is calculated. The formula is as follows: ； in, Let N be the Dirac function, and N be the number of nodes within the radius r. Let i be the coordinates of the i-th node. Let j be the coordinates of the j-th node; Set density threshold Nodes in high-density regions are selected; these high-density nodes are then clustered, with the cluster centers serving as new node locations. The clustering node condition formula is as follows: ， in, This is the tolerance threshold; In view of the complex shape and irregular boundary of the goaf, the point cloud data of the goaf in the first point cloud data is obtained, and a topological graph model is established by the boundary modeling method based on the improved α-shape algorithm. For the curves and surfaces in various positional relationships, a multidimensional topological structure model is established by using a curve and surface fitting method based on non-uniform rational B-splines. The formula for the NURBS curve is as follows: ， in Let i be the basis function. The weights are the weights of the i-th control point for the curve and surface structures. Control points are a term used in NURBS fitting. Let represent the coordinates of the i-th control point of the curve and surface structure, u be a parameter variable defined within the control point vector range, n be the number of control points of the curve and surface structure, and p be the order of the B-spline curve. The range of values ​​for the parameter variable u; Establish a multi-level spatial topology model based on positional relationships.

2. The method according to claim 1, characterized in that, S1 specifically includes: collecting spatial data from multi-source data fusion in the well, performing timestamp calibration on the spatial data, unifying the spatial data into a coordinate system to obtain point cloud data, and performing denoising and fusion processing on the point cloud data to obtain the first point cloud data.

3. The method according to claim 2, characterized in that, The step of denoising and fusing the point cloud data to obtain the first point cloud data specifically includes: using an improved adaptive filtering algorithm to denoise the point cloud data and then performing a weighted average fusion to obtain the first point cloud data. The formula for the improved adaptive filtering algorithm is as follows: ， in, Let be the filtering threshold for the i-th point cloud. This is the initial filter threshold. Let be the local point cloud density of the i-th point cloud. The global average point cloud density, For adjustment coefficients; The formula for calculating the weight parameters of the weighted average fusion is as follows: ， in, The weight of the i-th data source, For the precision of the i-th data source, Let j be the precision of the j-th data source, and n be the number of data sources, where 1 ≤ j ≤ n.

4. The method according to claim 1, characterized in that, The method for establishing a topological graph model using a boundary modeling approach based on an improved α-shape algorithm specifically includes: Triangulation is performed on the point cloud data of the goaf area to generate an initial triangular mesh. An adaptive α parameter is introduced, and the α value is dynamically adjusted according to the point cloud density and spatial distribution to establish a topology map. The formula for calculating the adaptive α value is as follows: ， in: Let α be the value of the i-th region. The initial value of α, Let be the point cloud density of the i-th region. α represents the maximum point cloud density in the region, and γ is an adjustment coefficient that controls the sensitivity of the α value to the density. The topological graph is smoothed using the Laplace smoothing algorithm to obtain the topological graph model. The Laplace smoothing algorithm is as follows: ， in, For the updated node position, Let be the original node position, λ be a smoothing factor ranging from 0 to 1, and k be the number of neighboring nodes of node i. Let be the coordinates of the j-th neighbor node.

5. The method according to claim 1, characterized in that, The establishment and deployment of the dynamic weight function specifically includes: Obtain real-time downhole operating parameters and standardize them using the following formula: ， in, For the k-th real-time operating condition parameter, and These are the minimum and maximum values ​​of the parameter, respectively. A dynamic weighting function is established based on real-time operating parameters, as shown in the following formula: ， Let be the dynamic weight of the s-th type of sensor at spatial location i; These are the basic weights for the s-th type of sensor; Let be the sensitivity coefficient of the s-th type of sensor to the k-th operating condition parameter; β is the influence function of the operating parameters, reflecting the form of the parameters' influence on the weights; β is the adjustment exponent, controlling the nonlinearity of the weight growth.

6. The method according to claim 5, characterized in that, Specifically, S3 includes: obtaining a deployment optimization model based on the deployment dynamic weight function and the multi-level spatial topology model. The formula for the deployment optimization model is as follows: ， Z represents the objective function for optimization, taking into account both monitoring effectiveness and deployment cost; N represents the total number of spatial locations to be deployed; and S represents the total number of sensor types. Let be the dynamic weight of the s-th type of sensor at spatial location i; Let be a binary variable. If the s-th type of sensor is deployed at spatial location i, then ,otherwise c s denoted as the unit cost of the s-th type sensor; λ is a trade-off coefficient that balances monitoring effectiveness with deployment cost. Set constraints to ensure that critical areas are effectively covered by the appropriate type of sensor; 1. The formula for covering constraints is as follows: ， in: The set of sensor types required for spatial location i; Let be the set of neighborhood locations of spatial location i, and be the coverage area of ​​the sensor; Let be a binary variable. If the s-th type of sensor is deployed at spatial location j, then ,otherwise If the s-th type of sensor effectively covers position i at spatial position j, then ,otherwise ; The minimum coverage required for spatial location i; 2. Connectivity constraints: Based on the spatial topology model and the communication range of the sensors, a connectivity matrix is ​​constructed between nodes, with the connectivity constraint: network connectivity ≥ δ; The network connectivity is calculated using the algebraic connectivity of the minimum cut set or the Laplace matrix, where δ is a preset connectivity threshold.

3. Deployment location constraints: ， in, Deployable spatial locations A set of.

7. The method according to claim 1, characterized in that, The optimized deployment model based on the Transformer-based deep reinforcement learning algorithm specifically includes: The system acquires real-time downhole operating parameters, converts the state vectors of the downhole real-time operating parameters, deployment optimization model, and multi-level spatial topology model into embedding vectors, adds position encoding to the embedding vectors to obtain embedded position vectors, inputs the embedded position vectors into a multi-head self-attention layer to extract global features, performs nonlinear transformation on the global features based on a feedforward neural network, outputs evaluation values, and selects the highest evaluation value as the optimal deployment model.

Images