Intelligent Ventilation Method and System for Gradient Pressure Regulation of Underground Cavern Groups Based on Hypergraph Networks

Abstract

The present invention discloses an intelligent ventilation method and system for gradient pressure regulation of underground cavern groups based on a hypergraph network, including establishing a hypergraph structure of the ventilation system, adopting a convolutional neural directed hypergraph aggregation method to convert monitoring data into input parameters for solving the required air volume of pipe segments, and endowing the model with more effective joint regulation features; adopting a ventilation network calculation technology and a response efficiency joint regulation method to solve the optimal ventilation network calculation results; in terms of the joint regulation control strategy, proposing ventilation control target curves under different target requirements, establishing a control strategy based on a deep deterministic policy gradient reinforcement learning algorithm, combining the deviation between the calculated data of each node and the target control data, establishing a standard library, a criterion library, a rule library, etc., to form a coupled hierarchical joint control instruction of fan-air valve-air door from power to the end; in terms of the ventilation control system and equipment, adopting an "edge-cloud" architecture, including a ventilation monitoring system, a network analysis system and a control optimization system, to realize the dynamic regulation closed loop of the intelligent ventilation system.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent ventilation for tunnel engineering, and particularly relates to an intelligent ventilation method and system for gradient pressure regulation of underground cavern groups based on a hypergraph network. Background Technique

[0002] Intelligent demand-based ventilation is the basis for ensuring the environmental safety of underground cavern groups and is also one of the core technologies for the high-quality construction and development of hydropower projects. With the improvement of the level of intelligent construction of projects, higher requirements are also put forward for intelligent ventilation and control under dynamic demands, involving multiple disciplinary fields such as space science, information science, construction technology, basic theories of mathematics, and fluid mechanics. Among them, ventilation network calculation is the underlying core technology of intelligent ventilation and control, and it is the key to being able to analyze monitoring data in real time and accurately, and provide guidance for the fault diagnosis and disaster prediction of the ventilation system based on the spatial changes of the construction progress, construction personnel and activities, abnormal fires, etc.

[0003] Regarding the theoretical research of ventilation networks, since the Cross algorithm proposed in 1936, the coupled heat flow ventilation theory, the concepts of true and false topological networks, and fuzzy optimization mathematical models have emerged successively. Now it has been relatively mature. In terms of calculation methods, considering the influence of natural wind pressure, a real-time ventilation calculation based on thermodynamics is proposed; analyzing the smoke distribution in the three-dimensional structure of subway stations and tunnel troughs under emergency ventilation, the influence of fire wind pressure on ventilation calculation and the effectiveness of smoke control under network-type smoke flow paths; coupling multiple objectives such as ventilation efficiency and energy consumption, and adjusting and controlling through methods such as genetic algorithms and particle swarm algorithms to optimize the energy consumption of the ventilation system, improve the overall convergence of the calculation, and for the adjustment stability, a dynamic evaluation model for the stability and reliability of mine ventilation systems is constructed based on system dynamics and attribute mathematics theory. In terms of intelligent control, a multi-parameter environmental monitoring system is built, paying attention to the combination with on-site environment monitoring to improve the accuracy and engineering applicability of simulation results, establishing an air environmental quality prediction based on CFD simulation, and building a three-dimensional visual emergency decision-making platform.

[0004] In summary, in recent years, domestic and foreign scholars have achieved certain results in using ventilation network calculation and intelligent technology to solve the adjustment calculation problem of the dynamically changing required air volume of ventilation network branches, and have gradually applied key intelligent technologies and platform facilities such as on-line monitoring, optimized layout, early warning, fault diagnosis, and closed-loop control in engineering. However, with the upgrade of the complexity degree of the ventilation system structure branches and the improvement of the overall stability requirements during the ventilation system adjustment process, the traditional network calculation cannot meet the accuracy requirements of ventilation parameters, and the numerical simulation method has great limitations in terms of computing resources for real-time calculation of ventilation parameters of large-scale projects, and cannot provide precise intelligent dynamic feedback control for the ventilation system. Therefore, there are still challenges in the real-time performance, stability, and engineering applicability of ventilation system adjustment.

[0005] With the increasing demands for big data and real-time computing, graph theory and hypergraphs have been widely applied in fields such as water conservancy, mining, and power systems, making the modeling, storage, and analysis of complex relational data more efficient and convenient. This includes the parallelization of saturation sorting solutions using breadth-first and depth-first search algorithms in graph theory. In hypergraph research, methods for identifying hypernetwork nodes using information entropy have been proposed, and in terms of power prediction in power systems, a dynamic hypergraph representation method for wind turbines based on wake correlation along the time dimension has been proposed. However, there is still little research on using hypergraph structures to solve ventilation regulation problems, including the selection of nodes, edges, and hyperedges, and the determination of characteristic parameters. No relevant results on using hypergraph structures to complete real-time calculation and coordinated control of ventilation systems have been reported. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to provide an intelligent ventilation method and system for gradient pressure regulation of underground cavern groups based on hypergraph networks. Research is carried out on issues such as the in-depth utilization of complex structural environment monitoring data of underground engineering cavern groups, the coupled closed-loop control technology of fans and dampers, the matching judgment strategy for air supply on demand, and the engineering air volume adjustment effect test guided by demand and energy consumption. Principles such as on-demand supply of air volume load, precise calculation of zonal regulation, and closed-loop feedback control based on deep deterministic policy gradient reinforcement learning are defined. A ventilation system graph and hypergraph structure, a coordinated adjustment method for response efficiency indicators, a feedforward control algorithm based on deep deterministic policy gradient reinforcement learning, and an intelligent ventilation system device and platform based on "edge-cloud-end" are established.

[0007] The technical solution of the present invention is as follows:

[0008] An intelligent ventilation method for gradient pressure regulation of underground cavern groups based on hypergraph networks, comprising the following steps:

[0009] (1) Construction of ventilation system graph and hypergraph structure, comprising the following steps:

[0010] S1. Construction of hypergraph structure: First, construct the basic graph structure of the ventilation system, and then introduce the hypergraph concept to characterize the spatial fluid association in the ventilation regulation process, and use directed hyperedges to describe the directed relationship between two node sets;

[0011] S2. Hypergraph matrix: Use the hypergraph matrix to define the relationship between the node set and the hyperedge set. The expression of the hypergraph matrix structure is as follows:

[0012] G = {V(H), E(H), V(D), E(D), X, W},

[0013] where V(D) represents the node set of the ordinary graph D, and V(D) = {v1,..., v n}\(n\) nodes are included in graph \(D\), \(E(D)\) represents the edge set of the ordinary graph \(D\), and \(E(D)=\{e_1,\cdots,e\}\) m}\(m\) edges are included in graph \(D\), \(V(H)\) represents the node set of the hypergraph \(H\) and \(V(H)=\{v_1,\cdots,v\}\) n}\(E(H)\) is the hyperedge set of the hypergraph \(H\) and \(E(H)=\{E_1,\cdots,E\}\) k}\(X\) represents the feature matrix of nodes, and \(W\) represents the weight matrix;

[0014] Each hyperedge in the hyperedge set \(E(H)\) describes the relationship between k \(T\) k and \(H\) k two subsets: \(E\) k \(=(T\) k )

[0015] where \(T\) k represents the tail set of the directed-edge hyperedge and \(T\) k \(=\{v\) T1 ,\cdots,v\) Tk \}, \(H\) k represents the head set of the directed-edge hyperedge and \(H\) k \(=\{v\) H1 ,\cdots,v\) Hk \};

[0016] The \(T\) k and \(H\) k are subsets of \(E(D)\), and the hypergraph incidence matrix is shown as follows:

[0017] \(B=(b\) ij ) m*n

[0018]

[0019] where \(B\) represents the hypergraph incidence matrix, \(b\) ij represents the relationship between the node \(v\) i and the hyperedge \(E\) j , \(m\) represents the number of nodes in the hypergraph \(H\), \(n\) represents the number of hyperedges in the hypergraph \(H\), \(v\) i represents the \(i\)-th node in the hypergraph \(H\), \(H\) j represents the head set of the hyperedge \(E\) j , \(T\) j represents the tail set of the hyperedge \(E\) j , \(E\) j represents the \(j\)-th hyperedge set in the hypergraph \(H\);

[0020] S3. Convolution algorithm and hypergraph aggregation: The convolution algorithm is combined with the hypergraph aggregation process. The aggregation process includes: node information aggregation and hyperedge information aggregation. The node information aggregation is specifically as follows: according to the joint adjustment sensitivity, the feature vectors of all neighboring nodes of each node are weighted summed to form the head set and tail set representation vectors, and the node information is aggregated through the gate mechanism to finally obtain the representation vector of the directed hyperedge; the hyperedge information aggregation is specifically as follows: all directed hyperedges with the same head set are merged to obtain a new directed hyperedge, and the representation vectors of all directed hyperedges with the same head set are weighted summed to obtain the final output vector of the corresponding node. Finally, the ventilation control hypergraph model is constructed;

[0021] (2) The optimized monitoring layout of the collective coverage model includes the following steps:

[0022] S4. Construction of a set coverage model, including: determination of a target node set and construction of a response node combination. The target node set is determined by querying the performance attributes of all nodes in the ventilation adjustment map, determining the normal, abnormal, and warning categories, and screening out target nodes to be adjusted that exceed the deviation range; the response node combination is constructed by selecting the air volume-wind resistance response efficiency in the data association between nodes as the adjustment basis, using a multi-level relationship query method based on the output vector matrix of the corresponding node in the node information fusion process, retrieving nodes with data associations with the target node set, eliminating nodes that do not meet the adjustable attributes and positive and negative correlation attributes, and constructing a response node set.

[0023] S5. Optimization of the layout of monitoring points, including the following steps:

[0024] Step A: By setting the release point where the concentration of harmful gases suddenly changes, the node range affected by the release point within the monitoring response time is calculated to form a release set;

[0025] Step B: define potential monitoring points, calculate the effective monitoring range of potential monitoring points, establish the effective range set of potential monitoring points, and construct the monitoring point hypergraph element;

[0026] Step C: Use the greedy algorithm to solve the set coverage model and select the minimum number of monitoring points to ensure that there is at least one monitoring point within the effective response range of each release point;

[0027] (3) The ventilation network solution and response efficiency joint adjustment method includes the following steps:

[0028] S6. Gradient voltage regulation model construction, including the following steps:

[0029] Step D: Define the air volume-wind resistance response efficiency attenuation rate, quantify the impact of wind resistance changes on air volume, fit the air volume-wind resistance response efficiency curve through a power function, and calculate the response efficiency attenuation rate;

[0030] Step E: Set the threshold of the wind resistance value, determine the sensitive area of the air volume adjustment, and refine the adjustment plan to specific branches according to the air volume adjustment plan;

[0031] S7. Optimization of the air volume adjustment plan, including the following steps:

[0032] Step F: Screen the response nodes, select the set of nodes participating in the adjustment according to the sorting, and establish an air volume adjustment plan within the response efficiency constraint;

[0033] Step G: If the plan does not meet the target requirements, select more response nodes to update the air volume adjustment plan until the air volume deviation of the target node is reduced to within the deviation threshold;

[0034] Step H: Select new nodes to participate in the adjustment in each iteration to form a new air volume adjustment plan, and finally output a plan that meets the requirements of the combined adjustment sensitivity;

[0035] (4). Control strategy based on the deep deterministic policy gradient reinforcement learning algorithm: Based on the control principles of supplying air volume on demand and individualization of each zone, use the deep deterministic policy gradient reinforcement learning algorithm to construct a control strategy, and conduct individualized control of different air supply demands by simulating each zone in the environment;

[0036] (5). Response efficiency analysis and adjustment, including the following steps:

[0037] S8. Construction of the air volume - wind resistance response efficiency matrix, including the following steps:

[0038] Step I: Define the air volume - wind resistance response efficiency, calculate the relationship between the air volume change and the wind resistance disturbance between nodes to form a response efficiency matrix, and use the iterative method to continuously approximate the optimal solution until the air volume - wind resistance response efficiency matrix of the entire ventilation network is obtained;

[0039] Step J: Construct an iterative sequence, recalculate the response efficiency after each air volume adjustment until the calculation accuracy requirement is met;

[0040] S9. Fan power and wind resistance response efficiency, including the following steps:

[0041] Step K: Analyze the response relationship between the fan power and the wind resistance, define the sensitivity of the fan power change to the wind resistance change to form a fan power - wind resistance response efficiency matrix, and select the branch with low power sensitivity as the wind resistance adjustment branch;

[0042] Step L: Analyze the second - order air network response efficiency, quantify the mutual influence between the adjustment branches, construct a three - dimensional matrix, calculate the cross - relationship between different branches, and ensure the coordination of each branch during the adjustment process.

[0043] Furthermore, the method steps for constructing the basic graph structure of the ventilation system are as follows:

[0044] 1. Adopt a construction method based on Python and Neo4j to construct a ventilation regulation graph database. Use a Python program to call the data interface API to obtain ventilation data, and preprocess the data into the form of "unit - association - feature" triples. Represent the physical space connection formed by the fluid flowing into or out of the cross-section with the flow path relationship, that is, the basic association; represent the main associations reflecting the coupling influence law of ventilation characteristics derived from the spatial relationship with the data relationship.

[0045] 2. Use py2neo to establish a connection between the Python program and Neo4j, batch process the construction instructions of the graph database, define associations and allocate parameters using Cypher query statements. In the Neo4j graph database, use any node as the starting point l1 of the loop, find all relationship units that have a basic association with l1, and use the graph matching ability to find the shortest path in the relationship units as the next loop node l2.

[0046] 3. Use Cypher statements to find the shortest path among the nodes that have a basic association with l2 as the next loop node l3, repeat the query process, traverse the nodes in turn, and construct a loop set L.

[0047] Furthermore, the expression of the head set representation vector h H in step S3 is:

[0048] h H = (H H ⊙ W H )X,

[0049] where h H represents the representation vector of the head set of the hyperedge, H H represents the feature matrix of the head set nodes in the hyperedge, W H represents the weight matrix of the head set, and X represents the feature matrix of the nodes;

[0050] The expression of the tail set representation vector h T in step S3 is:

[0051] h T = (H T ⊙ W T )X,

[0052] where h T represents the representation vector of the tail set of the hyperedge, H T represents the feature matrix of the tail set nodes in the hyperedge, W T represents the weight matrix of the tail set, and X represents the feature matrix of the nodes;

[0053] The representation vector h of the directed hyperedge in step S3 εi has the following expression:

[0054] h ε = σ(h H ) ⊙ h T ,

[0055] where h ε represents the representation vector of the directed hyperedge, h H represents the representation vector of the head set of the hyperedge, h T represents the representation vector of the tail set of the hyperedge, and σ represents the activation function;

[0056] The expression of the final output vector in step S3 is:

[0057] h = (H H ) ⊙ W H )h ε ,

[0058] where h represents the node output vector after aggregating hyperedge information, H H represents the feature matrix of the head set nodes in the hyperedge, W H represents the weight matrix of the head set, and h ε represents the representation vector of the directed hyperedge;

[0059] The expression of the ventilation regulation hypergraph model in step S3 is:

[0060] H ε = {V(H), (H H ) ⊙ W H )h ε},

[0061] where H ε represents the set of basic nodes with a joint adjustment sensitivity greater than ε, ε represents the sensitivity threshold, V(H) represents the node set of the hypergraph H, H H represents the feature matrix of the head set nodes in the hyperedge, W H represents the weight matrix of the head set, and h ε represents the representation vector of the directed hyperedge.

[0062] Furthermore, the release point in step S5 is the node with the requirement for installing harmful gas concentration monitoring, and the expression of the release-related set in step S5 is:

[0063]

[0064]

[0065]

[0066] Among them, represents the set of involved release, represents the set of nodes affected by the released concentration, represents the set of release points, v j represents the node j in the hypergraph H, t max represents the response time, v n represents the nth release point;

[0067] If there are k nodes with sudden changes in the released gas concentration in the node set V(D) of the ordinary graph D, then construct the hyperedge set of release points:

[0068]

[0069] Among them, E 0 (H) represents the hyperedge set of release points, represents the edge set associated with the ith release point, represents the edge set associated with the kth release point, where k represents the total number of release points with sudden changes in the harmful gas concentration; the potential monitoring points in step S5 are the nodes equipped with wind pressure monitoring devices, and the expression for the effective monitoring range of the potential monitoring points is:

[0070]

[0071] Among them, represents the set of effective monitoring ranges of potential monitoring points, v k represents the nodes within the effective range, N j represents the potential monitoring point, t max represents the monitoring response time;

[0072] The expression for the hypergraph elements of the monitoring points in step S5 is:

[0073]

[0074] Among them, E t (H) represents the hyperedge set of potential monitoring points, represents the effective monitoring range edge set of the jth potential monitoring point, where k represents the total number of potential monitoring points;

[0075] According to the response requirements, there is at least one potential monitoring point within the effective response range of the release point, then establish the goal of minimizing the total number of monitoring points under the constraint of monitoring point layout:

[0076]

[0077]

[0078]

[0079] Among them, z represents the number of monitoring deployments, and x j represents the deployment variable, that is, the potential monitoring point N j whether to set the monitoring deployment, n represents the total number of potential monitoring points, represents the potential monitoring point N j the set of nodes that can be covered;

[0080] The specific solution of the set covering model by the greedy algorithm is as follows: Compile and run on the Pycharm platform using the Python programming language to realize the transformation of the above target expression; First, find the set E t (H) in the hyperedge set E of potential monitoring points that contains the nodes most affected by the release concentration 0 the elements in (H) corresponding to the head set and tail set relationship in the release set in the release point set the monitoring points {v H1 ,..., v Hi} as the preferred monitoring deployment positions and put them into the response node set with an empty original value; Secondly, delete the node elements {v H1 ,..., v Hi} that have been put into the response node set from the node set V(H) of the hypergraph H, and delete the elements related to the head set representation vector from the hyperedge set E t (H) of potential monitoring points Among them, represents the edge set in the hyperedge set E t (H) of potential monitoring points that is affected by the release point i; Thirdly, find the set of nodes that are most affected by the release concentration in the remaining hyperedge set E t (H) of potential monitoring points, and repeat the above steps until the hyperedge set E 0 (H) of release points is an empty set; Finally, obtain the elements in the response node set that represent the optimized deployment quantity and positions of potential monitoring points.

[0081] Furthermore, the definition of the air volume - air resistance response efficiency decay rate in step S6 is: the degree of change in the air volume - air resistance response efficiency of a branch caused by the change in the branch air resistance, and the expression of the air volume - air resistance response efficiency decay rate is:

[0082]

[0083] Among them, t ij represents the air volume - air resistance response efficiency decay rate, d ij represents the air volume - air resistance response efficiency, r j represents the air resistance value of response node j, and △r j represents the air resistance rj A small perturbation, q i represents the wind volume at the response node j;

[0084] The expression for the power function fitting the response efficiency curve of air volume-wind resistance in step S6 is:

[0085]

[0086] Among them, d ij represents the air volume-wind resistance response efficiency, r j represents the wind resistance value of the response node j, a, b and c are all constants greater than zero;

[0087] The calculation formula of the attenuation rate of the response efficiency in step S6 is:

[0088]

[0089] Among them, t ij represents the air volume-wind resistance response efficiency attenuation rate and t ij Always negative, that is, d ij With the j The value of r decreases with the increase of j represents the wind resistance value of the response node j, where a and b are both constants greater than zero;

[0090] The expression of the threshold value of the wind resistance value in step S6 is:

[0091]

[0092] Among them, r jmax Indicates the threshold value of wind resistance, r j0 represents the initial value of wind resistance, a, b and c are all constants greater than zero, α represents the sensitivity threshold and 0<α<0.05.

[0093] Furthermore, the method steps for constructing the control strategy using the deep deterministic policy gradient reinforcement learning algorithm are as follows:

[0094] Step 1: Define the state space: The state space is defined by the target branch parameter T t , adjust the branch parameter R t , power branch parameter P t The fan operating voltage, current and frequency are determined, and the function expression of the state space is as shown in the following formula:

[0095] s t =[T t , R t , P t ],

[0096] Among them, T t ={q1，t , q 2，t ,...q n，t , p 1，t , p 2，t ,...p n，t} represents the air volume and air pressure parameters of the target branch at time t, R t = {θ 1，t , θ 2，t ,...θ n，t} represents the damper opening in the regulating branch at time t, P t = {I 1，t , I 2，t ,...I n，t , f 1，t , f 2，t ,...f n，t} represents the fan current and frequency parameters of the power branch at time t;

[0097] Step 2: Define the action space: Execute corresponding actions according to the observed environmental parameters. The action space is determined by the damper opening, fan current, and frequency to be adjusted. The functional representation of the action space is as follows:

[0098] a t = [Δθ t , ΔI t , Δf t ,

[0099] where Δθ t = {Δθ 1，t , Δθ 2，t ,...Δθ n，t} represents the change in damper opening in the regulating branch at time t, and ΔI t = {ΔI 1，t , ΔI 2，t ,...ΔI n，t , Δf 1，t , Δf 2，t ,...Δf n，t} represents the change in fan current and frequency parameters of the power branch at time t;

[0100] Step 3: Define the RL reward: On the premise of meeting the target air volume load, minimize the ventilation energy consumption. In terms of the on-demand ventilation revenue reward, reduce the deviation between the supply air volume and the demand air volume, and satisfy:

[0101]

[0102] where Q n，t is the supply air volume of branch n at time t, is the demand air volume, d Q，t is the deviation between the supply air volume and the demand air volume, dQ，t,max The maximum threshold of the deviation between the air supply volume and the required air volume;

[0103] Set the long-term income ratio as λ1; in terms of the energy consumption income reward, control the damper opening to be greater than 70%, and the fan operates in a stable working area, that is, the working pressure is less than 90% of the maximum pressure, and the operating efficiency is greater than 60%, satisfying:

[0104] θ t ≥θ t，min ,Q f，min ≤Q f ≤Q f，max ,

[0105] where θ t is the damper opening, θ t，min is the minimum threshold of the damper opening, Q f is the fan air volume, Q f,min and Q f,max are respectively the minimum and maximum thresholds of the fan air volume range;

[0106] Set the long-term income ratios as λ2 and λ3, then the reward function is shown as the following formula:

[0107] r t (s t ,a t )=[R Q ,R θ ,R f ,

[0108] where R Q =λ1c r，Q ,c r，Q represents the air volume reward coefficient, R θ =λ2c r，θ ,c r，θ represents the damper reward coefficient, R θ =λ3c r，f ,c r，f represents the fan reward coefficient;

[0109] The expression of the air volume reward coefficient is:

[0110]

[0111] The expression of the damper reward coefficient is:

[0112]

[0113] The expression of the fan reward coefficient is:

[0114]

[0115] Step 4: Establish an Actor-Critic architecture, where the Actor-Critic architecture includes: an Actor current network function with parameter θ u : u = (s, a|θ u ), where θ u represents the parameter set of the Actor network, s represents the current state of the network, and a represents the action selected based on the current state; a Critic current network function with parameter θ Q : Q = (s, a|θ Q ), where θ Q represents the parameter set of the Critic network, s represents the state received by the action network, and a represents a deterministic action generated based on the received state; and an Actor-Critic corresponding target network with parameters θ u → θ u′ and θ Q → θ Q′ , where θ u′ represents the parameter set of the target Actor network, and θ Q′ represents the parameter set of the target Critic network;

[0116] Step 5: Initialize the Actor-Critic network:

[0117] Set the learning rates of the Actor network and the Critic network, and set the update smoothing coefficient of the target Q network; set the number of neurons in the Actor network and the Critic network architectures, set the capacity of the experience pool and the number of sampling pools for parameter update;

[0118] Step 6: Actor-Critic network update loop:

[0119] Based on the output of the current environmental state, the DDPG agent executes an action and obtains the next environmental state and the RL reward. At the same time, add action perturbation N t , environmental state s t , execute action a t , RL reward r t and the next environmental state s t+1 to form a sample and store it in the experience pool. Randomly select data {s t , a t , r t , s t+1} to the Actor-Critic current network and update the network parameters:

[0120] θ Q′ = τθ Q + (1 - τ)θ Q′ ,

[0121] Among them, θ Q represents the set of parameters of the Critic network, and θ Q′ represents the set of parameters of the target Critic network, and τ represents the factor controlling the update speed of the target network parameters;

[0122] θ u′ = τθ u + (1 - τ)θ u′ ,

[0123] Among them, θ u represents the set of parameters of the Actor network, and θ u′ represents the set of parameters of the target Actor network, and τ represents the factor controlling the update speed of the target network parameters;

[0124] Step 7: Continuously update the process of Step 6 until the policy network learns a good enough policy.

[0125] Furthermore, the definition of the air volume - air resistance response efficiency in Step S8 is: the degree of influence of the branch air resistance on the node pressure or the branch air volume, and the expression of the air volume - air resistance response efficiency is:

[0126]

[0127] Among them, d ij represents the air volume - air resistance response efficiency, r j represents the air resistance value of the response node j, q i represents the air volume at the target node i, △r j represents the small perturbation of the air resistance r j , and △q i represents the change in the air volume at the target node i;

[0128] The expression of the air volume - air resistance response efficiency matrix in Step S8 is:

[0129]

[0130] Among them, D Q represents the air volume - air resistance response efficiency matrix, d ij represents the air volume - air resistance response efficiency, q k represents the air volume at the response node j, r n represents the air resistance value of the nth response node, d jn represents the air resistance r of the response node n n on the air volume q of the response node j itself j , and n represents the number of response nodes;

[0131] The expression of the iteration sequence in Step S8 is:

[0132]

[0133]

[0134] Among them, represents the air volume - air resistance response efficiency in the k - th iteration, represents the air volume value of the target node i at the k - th iteration, represents the air volume value of the target node i at the (k - 1) - th iteration, r j (k) represents the air resistance value of the response node j at the k - th iteration, r j (k-1) represents the air resistance value of the response node j at the (k - 1) - th iteration, dr j (k) represents the air resistance perturbation, that is, the change amount of the air resistance of the response node j in the k - th iteration, dr j (k-1) represents the change amount of the air resistance of the response node j in the (k - 1) - th iteration, and γ represents the acceleration factor;

[0135] The expression of the calculation accuracy is:

[0136]

[0137] Among them, represents the air volume - air resistance response efficiency in the k - th iteration, represents the air volume - air resistance response efficiency in the (k - 1) - th iteration, and ε represents the calculation accuracy.

[0138] Furthermore, the definition of the fan power - air resistance response efficiency in step S9 is: the influence degree of the change of the fan output power on the change of the branch air resistance, and the expression of the fan power - air resistance response efficiency in step S9 is:

[0139]

[0140] Among them, dN ij represents the fan power - air resistance response efficiency, ΔN fi represents the change of the fan output power of the target node i, Δr j represents the change amount of the air resistance of the response node j, N fi represents the fan output power at the target node i, r j represents the air resistance value at the response node j;

[0141] For the fan power - air resistance response efficiency dN ij , the fan output power N fi satisfies the product relationship between the fan air pressure and the fan air volume:

[0142] N fi = h f q i ,

[0143] Among them, N fi represents the fan output power at the target node i, h f represents the fan air pressure, q i represents the air volume value at the target node i;

[0144] The fan air pressure h f satisfies the least squares fitting relationship of a unary polynomial. The degree of the polynomial takes the value of 3. Then the fan output power characteristic curve is shown as follows:

[0145]

[0146] Among them, N fi represents the fan output power at the target node i, q i represents the air volume value at the target node i, a0 represents the constant term, that is, the fan power change rate when the air volume value q i is zero. a1, a2, and a3 all represent the coefficients of the polynomial, describing the non-linear influence of the air volume on the fan power;

[0147] Take the partial derivative with respect to the air volume q i :

[0148]

[0149] Among them, N fi represents the fan output power at the target node i, q i represents the air volume value at the target node i, a0 represents the constant term, that is, the fan power change rate when the air volume value q i is zero. a1, a2, and a3 all represent the coefficients of the polynomial, describing the non-linear influence of the air volume on the fan power;

[0150] The expression of the fan power-air resistance response efficiency matrix in step S9 is:

[0151]

[0152]

[0153] Among them, D N represents the fan power-air resistance response efficiency matrix, dN ij represents the fan power-air resistance response efficiency, N fn represents the fan output power at the response node n, r n represents the air resistance value of the nth response node, d nn represents the air resistance r of the response node nn The influence degree on the output power N of the fan of the response node n fn , ΔN fi Indicates the change in the output power of the fan at the target node i, Δr j Indicates the change in the air resistance of the response node j, d ij Indicates the air volume-air resistance response efficiency;

[0154] The definition of the second-order air network response efficiency in the step S9 is: adjusting the branch e k The influence of the change in air resistance on the air volume of other branches e of the air network i , and the expression of the second-order air network response efficiency is:

[0155]

[0156] Among them, dd ijk Indicates the second-order air network response efficiency, d ij Indicates the air volume-air resistance response efficiency, r k Indicates the air resistance at node k, q i Indicates the air volume value at the target node i, r j Indicates the air resistance value at the response node j;

[0157] The three-dimensional matrix in the step S9 is a second-order air network response efficiency matrix, and the expression of the second-order air network response efficiency matrix is:

[0158] DD(i,j,k) = dd ijk ,

[0159] Among them, DD(i,j,k) is the second-order air network response efficiency matrix, dd ijk Indicates the second-order air network response efficiency. When j = k, it represents the influence of branch adjustment on its own air volume, that is, the effect efficiency decay rate of the adjustment branch. When j ≠ k, it represents the magnitude of the mutual influence between branches. If two adjustment branches e k and e j , meet the conditions: d ij > 0, d ik > 0, d ijk > 0, then it means that the two adjustment branches promote each other for the branch e i . If the conditions are met: d ij > 0, d ik > 0, d ijk < 0, then it means that the two adjustment branches inhibit each other for the branch e i ;

[0160] Taking the second derivative of the loop wind pressure equation system vector with respect to the air resistance vector:

[0161]

[0162] Among them, (R, Q y (R)) indicates that the free variables of the loop air pressure equation set are the air resistance vector R and the co-tree air volume vector Q y , represents the Jacobian matrix of the loop air pressure equation set with respect to air resistance, denoted as f1′, represents the Jacobian matrix of the loop air pressure equation set with respect to the co-tree air volume, denoted as f2′, is the Jacobian gradient algorithm matrix of the co-tree air volume with respect to air resistance, denoted as f3′;

[0163] Let Then the three-dimensional matrix analytical formula of the second-order air network response efficiency is shown as follows:

[0164]

[0165] Among them, represents the second derivative of the co-tree air volume vector Q y with respect to the air resistance vector R, f 11 ″ represents the second-order Jacobian matrix of the loop air pressure equation with respect to air resistance, f 12 ″ represents the mixed second-order Jacobian matrix of the loop air pressure equation with respect to air resistance and co-tree air volume, f 21 ″ represents the mixed second-order Jacobian matrix of the air pressure equation with respect to co-tree air volume and air resistance, f 22 ″ represents the second-order Jacobian matrix of the loop air pressure equation with respect to co-tree air volume.

[0166] An intelligent ventilation system for gradient pressure regulation of underground cavern groups includes a ventilation monitoring system, a network analysis system, a control optimization system and a visualization system. The network analysis system is connected to the ventilation monitoring system and the control optimization system respectively, and the visualization system is connected to the network analysis system. The ventilation monitoring system includes: an integrated air environment monitoring device, an air volume and air pressure monitoring device and a fan operation parameter monitoring device. The integrated air environment monitoring device is arranged in the tunnel, the air volume and air pressure monitoring device is arranged inside the ventilation duct, and the fan operation parameter monitoring device is arranged near the fan and connected to the network analysis system; the network analysis system includes: an integrated intelligent fan control cabinet and a cloud server, and the integrated intelligent fan control cabinet is connected to the ventilation monitoring system respectively. The system is connected to the control optimization system, and the cloud server is connected to the integrated intelligent fan control cabinet and the visualization system respectively. The integrated intelligent fan control cabinet includes: a data acquisition module, a data processing module, a calculation and analysis module and a communication module. The data acquisition module is connected to the calculation and analysis module, the calculation and analysis module is connected to the data processing module, and the communication module is connected to the data processing module and the cloud server respectively; the control optimization system includes: an intelligent air valve and a variable frequency fan control cabinet, the intelligent air valve is arranged on the ventilation duct, and the variable frequency fan control cabinet is arranged near the fan, the intelligent air valve and the variable frequency fan control cabinet are both connected to the network analysis system, the opening range of the intelligent air valve is not less than 70%, and the operating efficiency of the fan is greater than 60%.

[0167] The beneficial effects of the present invention are as follows: a hypergraph network-based gradient pressure-regulated intelligent ventilation method and system for underground cavern groups of the present invention adopts regional model theory, proposes node extraction principles and methods for one-dimensional tube bundle fluid and three-dimensional spatial flow field unit structure, and realizes the construction of different ventilation fluid regional graphs and hypergraph network structures of the ventilation system by establishing a virtual branch field network node coupling solution; on this basis, a set coverage model is established to optimize the monitoring layout, revealing the coupling linkage effect between ventilation branches and loops, and a resistance regulation solution method based on joint regulation sensitivity is proposed. With the target unit air regulation efficiency and overall stability as the core indicators, threshold parameters are set and the regulation scheme is optimized; a reinforcement learning closed-loop feed-back control regulation algorithm based on DDPG is established to obtain the optimal air volume regulation strategy in a dynamic environment; an intelligent ventilation system platform based on the "end-edge-cloud" architecture is developed, which has functions such as real-time solution, remote control and data analysis, and a multi-parameter integrated air environment monitoring device and an intelligent fan control cabinet are developed to realize the comprehensive perception, analysis, control and 3D display and early warning of air environment parameters and ventilation system operating status, thereby improving the intelligent regulation and management capabilities of construction ventilation. BRIEF DESCRIPTION OF THE DRAWINGS

[0168] Figure 1It is a schematic structural diagram of an intelligent ventilation system for gradient pressure regulation of underground cavern groups based on a hypergraph network according to the present invention.

[0169] Figure 2 It is a response efficiency air control joint debugging optimization step diagram of an intelligent ventilation method for gradient pressure regulation of underground cavern groups based on a hypergraph network according to the present invention.

[0170] Figure 3 It is a schematic field-network structure diagram of an intelligent ventilation method for gradient pressure regulation of underground cavern groups based on a hypergraph network according to the present invention.

[0171] Figure 4 It is a basic diagram structure construction and Cypher statement query flowchart of a ventilation system of an intelligent ventilation method for gradient pressure regulation of underground cavern groups based on a hypergraph network according to the present invention.

[0172] Figure 5 It is a schematic diagram of directed hypergraph structure aggregation of an intelligent ventilation method for gradient pressure regulation of underground cavern groups based on a hypergraph network according to the present invention.

[0173] Figure 6 It is a schematic diagram of the user interface of an intelligent ventilation software platform of an intelligent ventilation system for gradient pressure regulation of underground cavern groups based on a hypergraph network according to the present invention. Detailed implementation manners

[0174] As Figures 1-6 shown, an intelligent ventilation method for gradient pressure regulation of underground cavern groups based on a hypergraph network includes the following steps:

[0175] (1). Construction of the ventilation system diagram and the hypergraph structure, including the following steps:

[0176] S1. Construction of the hypergraph structure: First, construct the basic diagram structure of the ventilation system, and then introduce the hypergraph concept to represent the spatial fluid association in the ventilation regulation process, and use directed hyperedges to describe the directed relationship between two node sets;

[0177] S2. Hypergraph matrix: Use the hypergraph matrix to define the relationship between the node set and the hyperedge set, and the expression of the hypergraph matrix structure is as follows:

[0178] G = {V(H), E(H), V(D), E(D), X, W},

[0179] wherein, V(D) represents the node set of the ordinary graph D, and V(D) = {v1,..., v n} contains n nodes in the graph D, E(D) represents the edge set of the ordinary graph D, and E(D) = {e1,..., e m} contains m edges in the graph D, V(H) represents the node set of the hypergraph H and V(H) = {v1,..., v n}, E(H) is the hyperedge set of the hypergraph H and E(H)={E1,…,E k}, X represents the feature matrix of the node, W represents the weight matrix;

[0180] Each hyperedge in the hyperedge set E(H) describes T k and H k The relationship between two subsets: E k =(T k , H k ),

[0181] Among them, T k represents the tail set of directed edge superedges and T k ={v T1 ,...,v Tk}, H k represents the head set of the edge hyperedge and H k ={v H1 ,...,v Hk};

[0182] The T k and H k is a subset of E(D), and the hypergraph incidence matrix is as follows:

[0183] B=(b ij ) m*n

[0184]

[0185] Among them, B represents the hypergraph association matrix, b ij Represents node v i With the hyperedge E j The relationship between them, m represents the number of nodes in the hypergraph H, n represents the number of hyperedges in the hypergraph H, v i represents the i-th node in the hypergraph H, H j Represents the hyperedge E j The head set, T j Represents the hyperedge E j The last set, E j represents the jth hyperedge set in the hypergraph H;

[0186] S3. Convolution algorithm and hypergraph aggregation: The convolution algorithm is combined with the hypergraph aggregation process. The aggregation process includes: node information aggregation and hyperedge information aggregation. The node information aggregation is specifically as follows: according to the joint adjustment sensitivity, the feature vectors of all neighboring nodes of each node are weighted summed to form the head set and tail set representation vectors, and the node information is aggregated through the gate mechanism to finally obtain the representation vector of the directed hyperedge; the hyperedge information aggregation is specifically as follows: all directed hyperedges with the same head set are merged to obtain a new directed hyperedge, and the representation vectors of all directed hyperedges with the same head set are weighted summed to obtain the final output vector of the corresponding node. Finally, the ventilation control hypergraph model is constructed;

[0187] (2) The optimized monitoring layout of the collective coverage model includes the following steps:

[0188] S4. Construction of a set coverage model, including: determination of a target node set and construction of a response node combination. The target node set is determined by querying the performance attributes of all nodes in the ventilation adjustment map, determining the normal, abnormal, and warning categories, and screening out target nodes to be adjusted that exceed the deviation range; the response node combination is constructed by selecting the air volume-wind resistance response efficiency in the data association between nodes as the adjustment basis, using a multi-level relationship query method based on the output vector matrix of the corresponding node in the node information fusion process, retrieving nodes with data associations with the target node set, eliminating nodes that do not meet the adjustable attributes and positive and negative correlation attributes, and constructing a response node set.

[0189] S5. Optimization of the layout of monitoring points, including the following steps:

[0190] Step A: By setting the release point where the concentration of harmful gases suddenly changes, the node range affected by the release point within the monitoring response time is calculated to form a release set;

[0191] Step B: define potential monitoring points, calculate the effective monitoring range of potential monitoring points, establish the effective range set of potential monitoring points, and construct the monitoring point hypergraph element;

[0192] Step C: Use the greedy algorithm to solve the set coverage model and select the minimum number of monitoring points to ensure that there is at least one monitoring point within the effective response range of each release point;

[0193] (3) The ventilation network solution and response efficiency joint adjustment method includes the following steps:

[0194] S6. Gradient voltage regulation model construction, including the following steps:

[0195] Step D: Define the air volume-wind resistance response efficiency attenuation rate, quantify the impact of wind resistance changes on air volume, fit the air volume-wind resistance response efficiency curve through a power function, and calculate the response efficiency attenuation rate;

[0196] Step E: Set the threshold of the wind resistance value, determine the sensitive area of the air volume adjustment, and refine the adjustment plan to specific branches according to the air volume adjustment plan;

[0197] S7. Optimization of the air volume adjustment plan, including the following steps:

[0198] Step F: Screen the response nodes, select the set of nodes participating in the adjustment according to the sorting, and establish an air volume adjustment plan within the response efficiency constraint;

[0199] Step G: If the plan does not meet the target requirements, select more response nodes to update the air volume adjustment plan until the air volume deviation of the target node is reduced to within the deviation threshold;

[0200] Step H: Select new nodes to participate in the adjustment each time of iteration to form a new air volume adjustment plan, and finally output a plan that meets the requirements of the combined adjustment sensitivity;

[0201] (4). Control strategy based on the deep deterministic policy gradient reinforcement learning algorithm: According to the control principle of supplying air volume on demand and personalization for each zone, use the deep deterministic policy gradient reinforcement learning algorithm to construct the control strategy, and conduct personalized control for different air supply demands by simulating each zone in the environment;

[0202] (5). Response efficiency analysis and adjustment, including the following steps:

[0203] S8. Construction of the air volume - wind resistance response efficiency matrix, including the following steps:

[0204] Step I: Define the air volume - wind resistance response efficiency, calculate the relationship between the air volume change and the wind resistance disturbance between nodes to form a response efficiency matrix, and use the iterative method to continuously approach the optimal solution. Finally, obtain the air volume - wind resistance response efficiency matrix of the entire ventilation network;

[0205] Step J: Construct an iterative sequence, recalculate the response efficiency after each air volume adjustment until the calculation accuracy requirement is met;

[0206] S9. Fan power and wind resistance response efficiency, including the following steps:

[0207] Step K: Analyze the response relationship between the fan power and the wind resistance, define the sensitivity of the fan power change to the wind resistance change to form a fan power - wind resistance response efficiency matrix, and select the branch with low power sensitivity as the wind resistance adjustment branch;

[0208] Step L: Analyze the second - order wind network response efficiency, quantify the mutual influence between the adjustment branches, construct a three - dimensional matrix, calculate the cross - relationship between different branches, and ensure the coordination of each branch during the adjustment process.

[0209] Preferably, the method steps for constructing the basic diagram structure of the ventilation system are as follows:

[0210] 1. Adopt a construction method based on Python and Neo4j to construct a ventilation regulation diagram database. Use a Python program to call the data interface API to obtain ventilation data, and preprocess the data into the form of "unit - association - feature" triples. Represent the physical space connection formed by the fluid flowing into or out of the cross-section with the flow path relationship, that is, the basic association; represent the main associations reflecting the coupling influence law of ventilation characteristics derived from the spatial relationship with the data relationship.

[0211] 2. Use py2neo to establish a connection between the Python program and Neo4j, batch process the construction instructions of the graph database, use Cypher query statements to define associations and allocate parameters. In the Neo4j graph database, use any node as the starting point l1 of the loop, find all relationship units that have a basic association with l1, and use the graph matching ability to find the shortest path in the relationship units as the next loop node l2.

[0212] 3. Use Cypher statements to find the shortest path among the nodes that have a basic association with l2 as the next loop node l3, repeat the query process, traverse the nodes in turn, and construct a loop set L.

[0213] Furthermore, the expression of the head set representation vector h H in step S3 is:

[0214] h H = (H H ⊙ W H )X,

[0215] where h H represents the representation vector of the head set of the hyperedge, H H represents the feature matrix of the head set nodes in the hyperedge, W H represents the weight matrix of the head set, and X represents the feature matrix of the node;

[0216] The expression of the tail set representation vector h T in step S3 is:

[0217] h T = (H T ⊙ W T )X,

[0218] where h T represents the representation vector of the tail set of the hyperedge, H T represents the feature matrix of the tail set nodes in the hyperedge, W T represents the weight matrix of the tail set, and X represents the feature matrix of the node;

[0219] The representation vector h of the directed hyperedge in step S3 εi has the following expression:

[0220] h ε = σ( h H ) ⊙ h T ,

[0221] where h ε represents the representation vector of the directed hyperedge, h H represents the representation vector of the head set of the hyperedge, hT represents the representation vector of the tail set of the hyperedge, and σ represents the activation function;

[0222] The expression of the final output vector in step S3 is:

[0223] h = (H H ⊙ W H ) h ε ,

[0224] where h represents the node output vector after aggregating hyperedge information, H H represents the feature matrix of the head set nodes in the hyperedge, W H represents the weight matrix of the head set, and h ε represents the representation vector of the directed hyperedge;

[0225] The expression of the ventilation adjustment hypergraph model in step S3 is:

[0226] H ε = {V(H), (H H ⊙ W H ) h ε},

[0227] where H ε represents the set of basic nodes with a co - adjustment sensitivity greater than ε, ε represents the sensitivity threshold, V(H) represents the node set of the hypergraph H, H H represents the feature matrix of the head set nodes in the hyperedge, W H represents the weight matrix of the head set, and h ε represents the representation vector of the directed hyperedge.

[0228] Furthermore, the release point in step S5 is the node with the requirement for installing harmful gas concentration monitoring, and the expression of the release - related set in step S5 is:

[0229]

[0230]

[0231]

[0232] Among them, represents the set of release-related nodes, T i 0 represents the set of nodes affected by the released concentration, represents the set of release points, v j represents node j in the hypergraph H, t max represents the response time, v n represents the nth release point;

[0233] If there are k nodes with sudden changes in the released gas concentration in the node set V(D) of the ordinary graph D, then construct the hyperedge set of release points:

[0234]

[0235] Among them, E 0 (H) represents the hyperedge set of release points, represents the edge set associated with the ith release point, represents the edge set associated with the kth release point, where k represents the total number of release points with sudden changes in the concentration of harmful gases; the potential monitoring points in step S5 are the nodes equipped with wind pressure monitoring devices, and the expression for the effective monitoring range of the potential monitoring points is:

[0236]

[0237] Among them, represents the set of effective monitoring ranges of potential monitoring points, v k represents the nodes within the effective range, N j represents the potential monitoring point, t max represents the monitoring response time;

[0238] The expression for the hypergraph elements of the monitoring points in step S5 is:

[0239]

[0240] Among them, E t (H) represents the hyperedge set of potential monitoring points, represents the effective monitoring range edge set of the jth potential monitoring point, where k represents the total number of potential monitoring points;

[0241] According to the response requirements, there is at least one potential monitoring point within the effective response range of the release point, then establish the goal of minimizing the total number of monitoring points under the monitoring point layout constraints:

[0242]

[0243]

[0244]

[0245] Among them, z represents the number of monitoring deployments, and x j represents the deployment variable, that is, whether the potential monitoring point N j sets the monitoring deployment, n represents the total number of potential monitoring points, represents the potential monitoring point N j the set of nodes that can be covered;

[0246] The specific solution of the set covering model by the greedy algorithm is as follows: Compile and run on the Pycharm platform using the Python programming language to achieve the transformation of the above target expression; First, find the set E t (H) in the hyperedge set E of potential monitoring points that contains the nodes most affected by the release concentration 0 the elements in (H) corresponding to the head set and tail set relationship in the release set Put the monitoring points {v in the release point set H1 ,..., v Hi} as the preferred monitoring deployment positions and put them into the response node set with an empty original value; Second, delete the node elements {v H1 ,..., v Hi} that have been put into the response node set from the node set V(H) of the hypergraph H, and delete the elements related to the head set representation vector from the hyperedge set E t (H) of potential monitoring points Among them, represents the edge set in the hyperedge set E t (H) of potential monitoring points affected by the release point i; Third, find the set of nodes most affected by the release concentration in the remaining hyperedge set E t (H) of potential monitoring points, and repeat the above steps until the hyperedge set E 0 (H) of the release point is an empty set; Finally, obtain the optimized deployment quantity and position of the potential monitoring points represented by the elements in the response node set.

[0247] Preferably, the definition of the air volume-air resistance response efficiency decay rate in step S6 is: the degree of change in the air volume-air resistance response efficiency of the branch caused by the change in the branch air resistance, and the expression of the air volume-air resistance response efficiency decay rate is:

[0248]

[0249] Among them, t ij represents the air volume-air resistance response efficiency decay rate, d ij represents the air volume-air resistance response efficiency, r j represents the air resistance value of the response node j, and △rj Indicates wind resistance r j A small perturbation, q i represents the wind volume at the response node j;

[0250] The expression for the power function fitting the response efficiency curve of air volume-wind resistance in step S6 is:

[0251]

[0252] Among them, d ij represents the air volume-wind resistance response efficiency, r j represents the wind resistance value of the response node j, a, b and c are all constants greater than zero;

[0253] The calculation formula of the attenuation rate of the response efficiency in step S6 is:

[0254]

[0255] Among them, t ij represents the air volume-wind resistance response efficiency attenuation rate and t ij Always negative, that is, d ij With the j The value of r decreases with the increase of j represents the wind resistance value of the response node j, where a and b are both constants greater than zero;

[0256] The expression of the threshold value of the wind resistance value in step S6 is:

[0257]

[0258] Among them, r jmax Indicates the threshold value of wind resistance, r j0 represents the initial value of wind resistance, a, b and c are all constants greater than zero, α represents the sensitivity threshold and 0<α<0.05.

[0259] Preferably, the method steps for constructing a control strategy using the deep deterministic policy gradient reinforcement learning algorithm are:

[0260] Step 1: Define the state space: The state space is defined by the target branch parameter T t , adjust the branch parameter R t , power branch parameter P t The fan operating voltage, current and frequency are determined, and the function expression of the state space is as shown in the following formula:

[0261] s t =[T t , R t , P t ],

[0262] Among them, Tt = {q 1，t , q 2，t , …q n，t , p 1，t , p 2，t ,...p n，t} represents the air volume and air pressure parameters of the target branch at time t, R t = {θ 1，t , θ 2，t ,...θ n，t} represents the damper opening in the adjustment branch at time t, P t = {I 1，t , I 2，t , …I n，t , f 1，t , f 2，t , …f n，t} represents the fan current and frequency parameters of the power branch at time t;

[0263] Step 2: Define the action space: Execute corresponding actions according to the observed environmental parameters. The action space is determined by the damper opening to be adjusted, the fan current, and the frequency. The functional representation of the action space is as shown in the following formula:

[0264] a t = [Δθ t , ΔI t , Δf t

[0265] where Δθ t = {Δθ 1，t , Δθ 2，t ,...Δθ n，t} represents the change in the damper opening in the adjustment branch at time t, and ΔI t = {ΔI 1，t , ΔI 2，t ,...ΔI n，t , Δf 1，t , Δf 2，t ,...Δf n，t} represents the change in the fan current and frequency parameters of the power branch at time t;

[0266] Step 3: Define the RL reward: On the premise of meeting the target air volume load, minimize the ventilation energy consumption as much as possible. In terms of the on-demand ventilation revenue reward, reduce the deviation between the supply air volume and the required air volume to meet:

[0267]

[0268] where Q n，t is the supply air volume of branch n at time t, is the required air volume, and d Q，t ​There is a deviation between the air supply volume and the required air volume, d Q，t,max The maximum threshold value of the deviation between the air supply volume and the required air volume;

[0269] Set the long-term revenue ratio to λ1; in terms of the energy consumption revenue reward, control the damper opening to be greater than 70%, and the fan operates in a stable working area, that is, the working pressure is less than 90% of the maximum pressure, and the operating efficiency is greater than 60%, satisfying:

[0270] θ t ≥θ t，min ,Q f，min ≤Q f ≤Q f，max ,

[0271] where, θ t is the damper opening, θ t，min is the minimum threshold value of the damper opening, Q f is the fan air volume, Q f，min and Q f，max are respectively the minimum and maximum threshold values of the fan air volume range;

[0272] Set the long-term revenue ratios to λ2 and λ3, then the reward function is shown as the following formula:

[0273] r t (s t ,a t )=[R Q ,R θ ,R f ,

[0274] where, R Q =λ1c r，Q ,c r,Q represents the air volume reward coefficient, R θ =λ2c r，θ ,c r，θ represents the damper reward coefficient, R θ =λ3c r，f ,c r，f represents the fan reward coefficient;

[0275] The expression of the air volume reward coefficient is:

[0276]

[0277] The expression of the damper reward coefficient is:

[0278]

[0279] The expression of the fan reward coefficient is:

[0280]

[0281] Step 4: Establish an Actor-Critic architecture, where the Actor-Critic architecture includes: an Actor current network function with parameters θ u : u = π(s, a|θ u ), where θ u represents the parameter set of the Actor network, s represents the current state of the network, and a represents the action selected based on the current state; a Critic current network function with parameters θ Q : Q = Q(s, a|θ Q ), where θ Q represents the parameter set of the Critic network, s represents the state received by the action network, and a represents a deterministic action generated based on the received state; and Actor-Critic corresponding target networks with parameters θ u → θ u′ and θ Q → θ Q′ , where θ u′ represents the parameter set of the target Actor network, and θ Q′ represents the parameter set of the target Critic network;

[0282] Step 5: Initialize the Actor-Critic network:

[0283] Set the learning rates of the Actor network and the Critic network, and set the update smoothing coefficient of the target Q network; set the number of neurons in the Actor network and the Critic network architectures, set the capacity of the experience pool and the number of sampling pools for parameter update;

[0284] Step 6: Actor-Critic network update loop:

[0285] Based on the current environmental state output, the DDPG agent executes an action and obtains the next environmental state and the RL reward. At the same time, add action perturbation N t , environmental state s t , execute action a t , RL reward r t and the next environmental state s t+1 to form a sample and store it in the experience pool. Randomly select data {s t , a t , r t , s t+1} to the Actor-Critic current network and update the network parameters:

[0286] θ Q′ = τθ Q + (1 - τ)θ Q′ ,

[0287] Among them, θ Q represents the parameter set of the Critic network, and θ Q′ represents the parameter set of the target Critic network, and τ represents the factor controlling the update speed of the target network parameters;

[0288] θ u′ = τθ u +(1 - τ)θ u ′,

[0289] Among them, θ u represents the parameter set of the Actor network, and θ u′ represents the parameter set of the target Actor network, and τ represents the factor controlling the update speed of the target network parameters;

[0290] Step 7: Continuously update the process of Step 6 until the policy network learns a good enough policy.

[0291] Preferably, the definition of the air volume - air resistance response efficiency in Step S8 is: the influence degree of the branch air resistance on the node pressure or the branch air volume, and the expression of the air volume - air resistance response efficiency is:

[0292] [[ID=***]]

[0293] Among them, d ij represents the air volume - air resistance response efficiency, r j represents the air resistance value of the response node j, q i represents the air volume at the target node i, △r j represents the small perturbation of the air resistance r j , and △q i represents the change in the air volume at the target node i;

[0294] The expression of the air volume - air resistance response efficiency matrix in Step S8 is:

[0295]

[0296] Among them, D Q represents the air volume - air resistance response efficiency matrix, d ij represents the air volume - air resistance response efficiency, q j represents the air volume at the response node j, r n represents the air resistance value of the nth response node, d [[ID=***]] jn represents the influence degree of the air resistance r n of the response node n on its own air volume q j at the response node j, and n represents the number of response nodes;

[0297] Note: There are two tags with "***" in the original text which seem to be incomplete or incorrect in terms of the provided rules. I've translated the text as accurately as possible based on the given content.The expression of the iterative sequence in step S8 is as follows:

[0298]

[0299] dr j (k) = γdr j (k-1) ,

[0300] where, represents the air volume - air resistance response efficiency in the k - th iteration, represents the air volume value of the target node i at the k - th iteration, represents the air volume value of the target node i at the (k - 1)-th iteration, r j (k) represents the air resistance value of the response node j at the k - th iteration, r j (k-1) represents the air resistance value of the response node j at the (k - 1)-th iteration, dr j (k) represents the air resistance perturbation, that is, the change in the air resistance of the response node j in the k - th iteration, dr j (k-1) represents the change in the air resistance of the response node j in the (k - 1)-th iteration, and γ represents the acceleration factor;

[0301] The expression of the calculation accuracy is:

[0302]

[0303] where, represents the air volume - air resistance response efficiency in the k - th iteration, represents the air volume - air resistance response efficiency in the (k - 1)-th iteration, and ε represents the calculation accuracy.

[0304] Preferably, the definition of the fan power - air resistance response efficiency in step S9 is: the influence degree of the change in the fan output power on the change in the branch air resistance. The expression of the fan power - air resistance response efficiency in step S9 is:

[0305]

[0306] where, dN ij represents the fan power - air resistance response efficiency, ΔN fi represents the change in the fan output power of the target node i, Δr j represents the change in the air resistance of the response node j, N fi represents the fan output power at the target node i, r j represents the air resistance value at the response node j;

[0307] For the fan power-air resistance response efficiency dN ij The output power N of the fan fi satisfies the product relationship between the fan air pressure and the fan air volume:

[0308] N fi = h f q i ,

[0309] wherein, N fi represents the output power of the fan at the target node i, h f represents the fan air pressure, and q i represents the air volume value at the target node i;

[0310] The fan air pressure h f satisfies the least squares fitting relationship of a unary polynomial. The degree of the polynomial takes the value of 3. Then the fan output power characteristic curve is shown as follows:

[0311]

[0312] wherein, N fi represents the output power of the fan at the target node i, q i represents the air volume value at the target node i, a0 represents the constant term, that is, the fan power change rate when the air volume value q i is zero, and a1, a2, and a3 all represent the coefficients of the polynomial, describing the non-linear influence of the air volume on the fan power;

[0313] Take the partial derivative with respect to the air volume q i :

[0314]

[0315] wherein, N fi represents the output power of the fan at the target node i, q i represents the air volume value at the target node i, a0 represents the constant term, that is, the fan power change rate when the air volume value q i is zero, and a1, a2, and a3 all represent the coefficients of the polynomial, describing the non-linear influence of the air volume on the fan power;

[0316] The expression of the fan power-air resistance response efficiency matrix in the step S9 is:

[0317]

[0318]

[0319] wherein, D N represents the fan power-air resistance response efficiency matrix, dN ij represents the fan power-air resistance response efficiency, Nfn represents the output power of the fan at the response node n, r n represents the wind resistance value of the nth response node, d nn represents the wind resistance r of the response node n n influence degree on the output power N of the fan at the response node n fn , ΔN fi represents the change in the output power of the fan at the target node i, Δr j represents the change amount of the wind resistance of the response node j, d ij represents the air volume - wind resistance response efficiency;

[0320] The definition of the second - order wind network response efficiency in the step S9 is: adjusting the branch e k influence of the wind resistance change on the air volume of other branches e of the wind network i The expression of the second - order wind network response efficiency is:

[0321]

[0322] where, dd ijk represents the second - order wind network response efficiency, d ij represents the air volume - wind resistance response efficiency, r k represents the wind resistance at the node k, q i represents the air volume value at the target node i, r j represents the wind resistance value at the response node j;

[0323] The three - dimensional matrix in the step S9 is the second - order wind network response efficiency matrix, and the expression of the second - order wind network response efficiency matrix is:

[0324] DD(i,j,k) = dd ijk ,

[0325] where, DD(i, j, k) is the second - order wind network response efficiency matrix, dd ijk represents the second - order wind network response efficiency. When j = k, it represents the influence of the branch adjustment on its own air volume, that is, the attenuation rate of the effect efficiency of the adjustment branch. When j ≠ k, it represents the magnitude of the mutual influence between branches. If two adjustment branches e k and e j , meet the conditions: d ij > 0, d ik > 0, d ijk > 0, then it means that the two adjustment branches promote each other for the branch e i . If they meet the conditions: d ij > 0, d ik > 0, d ijk < 0, then it means that the two adjustment branches inhibit each other for the branch e i ;

[0326] Take the second derivative of the vector of the loop air pressure equations with respect to the air resistance vector:

[0327]

[0328] where, (R, Q y (R)) indicates that the free variables of the loop air pressure equations are the air resistance vector R and the co-tree air volume vector Q y , represents the Jacobian matrix of the loop air pressure equations with respect to the air resistance, denoted as f1′, represents the Jacobian matrix of the loop air pressure equations with respect to the co-tree air volume, denoted as f2′, is the Jacobian gradient algorithm matrix of the co-tree air volume with respect to the air resistance, denoted as f3′;

[0329] Let Then the analytical formula of the three-dimensional matrix of the second-order air network response efficiency is shown as follows:

[0330]

[0331] where, represents the second derivative of the co-tree air volume vector Q y with respect to the air resistance vector R, f 11 ″ represents the second-order Jacobian matrix of the loop air pressure equation with respect to the air resistance, f 12 ″ represents the mixed second-order Jacobian matrix of the loop air pressure equation with respect to the air resistance and the co-tree air volume, f 21 ″ represents the mixed second-order Jacobian matrix of the air pressure equation with respect to the co-tree air volume and the air resistance, f 22 ″ represents the second-order Jacobian matrix of the loop air pressure equation with respect to the co-tree air volume.

[0332] An intelligent ventilation system for gradient pressure regulation of an underground cavern group, comprising a ventilation monitoring system, a network analysis system (analyzing control parameters such as the air volume, air pressure, air resistance, and fan operation characteristic curve of the ventilation network branches), a control optimization system (responsible for adjusting the operation parameters and resistance of the ventilation equipment according to the calculated data to achieve real-time control of the air state and air pressure in the roadway. Combining the calculation results of the data analysis module with the operation parameters of the ventilation system equipment in the tunnel, the control module issues instructions according to the decision results, adjusts the fan and air valve to execute actions through data transmission, and at the same time feeds back the execution data), and a visualization system (the visualization display system is based on the three-dimensional solid modeling of the engineering cavern group, visualizes the parameter data such as the ventilation environment, ventilation fluid, and operation of ventilation equipment and facilities, and displays the optimization of system energy consumption, ventilation environment warning, and ventilation equipment fault information). The network analysis system is respectively connected to the ventilation monitoring system and the control optimization system, the visualization system is connected to the network analysis system, and the ventilation monitoring system includes: an integrated air environment monitoring device (collecting pollutant concentration data such as temperature, humidity, carbon dioxide, carbon monoxide, oxygen, dust, and harmful gases), an air volume and air pressure monitoring device, and a fan operation parameter monitoring device (reading the state parameters such as the operating current, voltage, power, frequency, and control mode of the fan to achieve real-time collection and upload of the air environment and ventilation system parameters). The integrated air environment monitoring device is arranged in the tunnel, the air volume and air pressure monitoring device is arranged inside the ventilation duct, and the fan operation parameter monitoring device is arranged near the fan and connected to the network analysis system; the network analysis system includes: an integrated intelligent fan control cabinet and a cloud server. The integrated intelligent fan control cabinet is respectively connected to the ventilation monitoring system and the control optimization system (the integrated intelligent fan control cabinet uses a wired or wireless connection method with an industrial general standard interface). The cloud server is respectively connected to the integrated intelligent fan control cabinet and the visualization system. The integrated intelligent fan control cabinet includes: a data acquisition module, a data processing module (responsible for processing the collected data, including tasks such as data cleaning, conversion, integration, storage, and transmission), a calculation and analysis module (responsible for calculating the air volume and air pressure of each branch in the tunnel in real time, estimating the ventilation load based on the parameter data in the database, and solving the parameters of each branch and node of the ventilation network in real time according to the principle of on-demand ventilation.Meanwhile, it has functions such as abnormal warning of air environment data, energy consumption diagnosis of ventilation system, and fault diagnosis and warning of fan operation, etc., and a communication module (transmits the collected data to the network calculation and display platform through 5G wireless transmission or wired transmission technology). The data acquisition module is connected to the calculation and analysis module, the calculation and analysis module is connected to the data processing module, and the communication module is respectively connected to the data processing module and the cloud server. The control optimization system includes: an intelligent air valve and a variable-frequency fan control cabinet (receives the instructions formed by the cloud ventilation network calculation results, and completes the actions of changing the air valve opening and the fan operation state through a proportional-integral-derivative controller PID, ensuring that the deviation between the required air volume and the target air volume is less than 5%). The intelligent air valve is arranged on the ventilation duct, the variable-frequency fan control cabinet is arranged near the fan, both the intelligent air valve and the variable-frequency fan control cabinet are connected to the network analysis system, the opening range of the intelligent air valve is not less than 70%, and the operating efficiency of the fan is greater than 60%.

[0333] Although the specific implementation manners of the present invention have been introduced above in conjunction with the accompanying drawings, it does not limit the protection scope of the present invention. Those of ordinary skill in the art should understand that any non-creative modifications or equivalent replacements to the technical solutions of the present invention, without departing from the protection scope of the present technical solution, should be covered within the scope of the claims of the present invention.

Claims

1. An intelligent ventilation method for gradient pressure regulation of underground cavern groups based on hypergraph networks, characterized in that, It includes the following steps: (1) Construction of ventilation system diagram and hypergraph structure, including the following steps: S1. Construction of hypergraph structure: First, construct the basic graph structure of the ventilation system. Then, introduce the hypergraph concept to represent the spatial fluid association in the ventilation regulation process, and use directed hyperedges to describe the directed relationship between two node sets; S2. Hypergraph matrix: Use the hypergraph matrix to define the relationship between the node set and the hyperedge set. The expression of the hypergraph matrix is as follows: G = {V(H), E(H), V(D), E(D), X, W}, Among them, V(D) represents the node set of the ordinary graph D, and V(D) = {v1,..., v n}, which contains n nodes in graph D. E(D) represents the edge set of the ordinary graph D, and E(D) = {e1,..., e m}, which contains m edges in graph D. V(H) represents the node set of the hypergraph H and V(H) = {v1,..., v n}, E(H) is the hyperedge set of the hypergraph H and E(H) = {E1,..., E k}, X represents the feature matrix of nodes, and W represents the weight matrix; Each hyperedge description T in the hyperedge set E(H) k and H k The relationship between two subsets: E k =(T k , H k ), Among them, T k represents the tail set of the directed-edge hyperedge and T k = {v T1 ,..., v Tk}, H k represents the head set of the directed-edge hyperedge and H k = {v H1 ,..., v Hk}; The said T k and H k are subsets of E(D), and the hypergraph incidence matrix is shown as follows: B = (b ij ) m*n Among them, B represents the hypergraph incidence matrix, and b ij represents the relationship between node v i and hyperedge E j . m represents the number of nodes in hypergraph H, n represents the number of hyperedges in hypergraph H, v i represents the i-th node in hypergraph H, and H j represents the head set of hyperedge E j , and T j represents the tail set of hyperedge E j . E j represents the j-th hyperedge set in hypergraph H; S3. Convolution algorithm and hypergraph aggregation: Combine the convolution algorithm with the hypergraph aggregation process. The aggregation process includes: node information aggregation and hyperedge information aggregation. The node information aggregation is specifically: According to the joint debugging sensitivity, weighted sum the feature vectors of all neighboring nodes of each node to form the head set and tail set representation vectors, and aggregate the node information through the gate mechanism to finally obtain the representation vector of the directed hyperedge; The hyperedge information aggregation is specifically: Combine all directed hyperedges with the same head set to obtain a new directed hyperedge, weighted sum the representation vectors of all directed hyperedges with the same head set to obtain the final output vector of the corresponding node. Finally, construct the ventilation regulation hypergraph model; (2) Optimization monitoring layout of the set covering model, including the following steps: S4. Construction of the set covering model, including: determination of the target node set and construction of the response node combination. The determination of the target node set is specifically: By querying the performance attributes of all nodes in the ventilation regulation graph database, judge the normal, abnormal, and warning categories, and screen out the target nodes to be adjusted that exceed the deviation range; The construction of the response node combination is specifically: Select the air volume - air resistance response efficiency in the data association between nodes as the adjustment basis, and according to the output vector matrix of the corresponding node in the node information fusion process, use the multi - level relationship query method to retrieve the nodes that have data association with the target node set, and eliminate the nodes that do not meet the adjustable attributes and positive and negative correlation attributes to construct the response node set; S5. Optimization layout of monitoring points, including the following steps: Step A: By setting the release points of sudden changes in harmful gas concentration, calculate the node range affected by the release points during the monitoring response time to form the affected set; Step B: Define potential monitoring points, calculate the effective monitoring range of potential monitoring points, establish the set of effective ranges of potential monitoring points, and construct hypergraph elements of monitoring points; Step C: Use the greedy algorithm to solve the set covering model, select the least number of monitoring points for layout to ensure that there is at least one monitoring point within the effective response range of each release point; (3) Ventilation network calculation and joint debugging method of response efficiency, including the following steps: S6. Construction of gradient pressure regulation model, including the following steps: Step D: Define the attenuation rate of air volume - air resistance response efficiency, quantify the impact of air resistance change on air volume, fit the response efficiency curve of air volume - air resistance through a power function, and calculate the attenuation rate of response efficiency; Step E: Set the threshold of air resistance value, determine the sensitive area of air volume adjustment, and refine the adjustment plan to specific branches according to the air volume adjustment plan; S7. Optimization of the air volume adjustment plan, including the following steps: Step F: Screen the response nodes, select the set of nodes participating in the regulation according to the sorting, and establish a air volume regulation plan within the response efficiency constraint; Step G: If the plan does not meet the target requirements, select more response nodes to update the air volume regulation plan until the air volume deviation of the target node is reduced to within the deviation threshold; Step H: Select new nodes to participate in the regulation in each iteration to form a new air volume regulation plan, and finally output a plan that meets the requirements of the joint regulation sensitivity; (4) Control strategy based on the deep deterministic policy gradient reinforcement learning algorithm: According to the control principle of supplying air volume as needed and personalized zoning, use the deep deterministic policy gradient reinforcement learning algorithm to construct a control strategy, and conduct personalized control of different air supply demands by simulating each area in the environment; (5) Response efficiency analysis and regulation, including the following steps: S8. Construction of the air volume - air resistance response efficiency matrix, including the following steps: Step I: Define the air volume - air resistance response efficiency, calculate the relationship between the air volume change and the air resistance disturbance between nodes, form a response efficiency matrix, and use the iterative method to continuously approximate the optimal solution, and finally obtain the air volume - air resistance response efficiency matrix of the entire ventilation network; Step J: Construct an iterative sequence, recalculate the response efficiency after each air volume regulation until the calculation accuracy requirement is met; S9. Fan power and air resistance response efficiency, including the following steps: Step K: Analyze the response relationship between the fan power and the air resistance, define the sensitivity of the fan power change to the air resistance change, form a fan power - air resistance response efficiency matrix, and select the branch with low power sensitivity as the air resistance regulation branch; Step L: Analyze the second - order air network response efficiency, quantify the mutual influence between the regulation branches, construct a three - dimensional matrix, and calculate the cross - correlation between different branches to ensure the coordination of each branch during the regulation process.

2. The intelligent ventilation method for gradient pressure regulation of underground cavern groups based on a hypergraph network according to claim 1, wherein The method steps for constructing the basic graph structure of the ventilation system are as follows: First, adopt a construction method based on Python and Neo4j to construct a ventilation regulation graph database. Use the Python program to call the data interface API to obtain ventilation data, and pre - process the data into the form of "unit - association - feature" triples. Represent the physical space connection formed by the fluid inflow section or outflow section with the flow path relationship, that is, the basic association; represent the main associations reflecting the coupling influence law of ventilation characteristics derived from the spatial relationship with the data relationship; Second, use py2neo to establish a connection between the Python program and Neo4j, batch - process the construction instructions of the graph database, use Cypher query statements to define associations and assign parameters. In the Neo4j graph database, take any node as the starting point l1 of the loop, find all relationship units that have a basic association with l1, and use the graph matching ability to find the shortest path in the relationship units as the next loop node l2; Third, use Cypher statements to find the shortest path among the nodes that have a basic association with l2 as the next loop node l3, repeat the query process, traverse the nodes in turn, and construct a loop set L.

3. The intelligent ventilation method for gradient pressure regulation of underground cavern groups based on a hypergraph network according to claim 1, wherein The head set characterization vector h of step S3 H has the following expression: h H = (H H ⊙ W H ) X, Among them, h H represents the representation vector of the head set of the hyperedge, and H H represents the feature matrix of the head set nodes in the hyperedge, and W H represents the weight matrix of the head set; X represents the feature matrix of the nodes. The tail set characterization vector h of step S3 T has the following expression: h T = (H T ⊙W T ) X, Among them, h T represents the representation vector of the tail set of the hyperedge, h T represents the feature matrix of the tail set nodes in the hyperedge, W T represents the weight matrix of the tail set; X represents the feature matrix of the nodes; The representation vector h of the directed hyperedge in step S3 εi has the following expression: h ε = σ(h u ) ⊙ h T , Among them, h ε represents the representation vector of the directed hyperedge, h H represents the representation vector of the head set of the hyperedge, h T represents the representation vector of the tail set of the hyperedge, and σ represents the activation function; The expression of the final output vector in step S3 is: h = (H H ⊙W H )h ε , Among them, h represents the node output vector after aggregating hyper-edge information, and H H represents the feature matrix of the head-set nodes in the hyper-edge, and W H represents the weight matrix of the head-set, and h ε represents the representation vector of the directed hyper-edge; The expression of the ventilation regulation hypergraph model in step S3 is: H ε = {V(H), (H H ⊙W H )h ε}, Among them, H ε represents the set of basic nodes whose joint debugging sensitivity is greater than ε, where ε represents the sensitivity threshold, V(H) represents the node set of the hypergraph H, and H H represents the feature matrix of the head set nodes in the hyperedge, and W H represents the weight matrix of the head set, and h ε represents the characterization vector of the directed hyperedge.

4. A gradient pressure regulating intelligent ventilation method for underground cavern groups based on a hypergraph network according to claim 1, characterized in that The release point in step S5 is a node with the need to install a harmful gas concentration monitor, and the expression of the release set in step S5 is: Among them, represents the set of involved release, represents the set of nodes affected by the released concentration, represents the set of release points, v j represents node j in the hypergraph H, t max represents the response time, v n represents the nth release point; If there are k nodes with sudden changes in the released gas concentration in the node set V(D) of the ordinary graph D, then construct the release point hyperedge set: Among them, E 0 (H) represents the hyper - edge set of the release points, represents the edge set associated with the i - th release point, represents the edge set associated with the k - th release point, where k represents the total number of release points where the concentration of harmful gases mutates; the potential monitoring points in step S5 are the nodes equipped with wind pressure monitoring devices, and the expression for the effective monitoring range of the potential monitoring points is: Among them, represents the set of effective monitoring ranges of potential monitoring points, v k represents the nodes within the effective range, N j represents the potential monitoring points, t max represents the monitoring response time; The expression of the monitoring point hypergraph element in step S5 is: Among them, E t (H) represents the hyperedge set of potential monitoring points, represents the effective monitoring range edge set of the j-th potential monitoring point, and k represents the total number of potential monitoring points; According to the response requirements, there is at least one potential monitoring point within the effective response range of the release point, then establish the goal of minimizing the total number of monitoring points under the monitoring point layout constraints: Among them, z represents the number of monitoring layouts, and x j represents the layout variable, that is, the potential monitoring point N j indicating whether to set up a monitoring layout, n represents the total number of potential monitoring points, representing the potential monitoring point N j the set of nodes that can be covered; The specific solution of the greedy algorithm for the set covering model is as follows: Compile and run on the Pycharm platform using the Python programming language to achieve the transformation of the above target expression; First, find the set of elements in the hyperedge set E t (H) of potential monitoring points that are affected by the most released concentrations, which is the set E 0 (H); Corresponding to the head-set and tail-set relationship in the involved release set , take the monitoring points {v in the release point set H1 ,..., v Hi} as the preferred layout positions for monitoring, and put them into the response node set with an empty original value; Secondly, delete the node elements {v H1 ,..., v Hi} that have been put into the response node set from the node set V(H) of the hypergraph H, and delete the elements related to the head-set representation vector from the hyperedge set E t (H) of potential monitoring points Among them, represents the edge set in the hyperedge set E t (H) of potential monitoring points that is affected by the involved release point i; Thirdly, find the set of elements in the remaining hyperedge set E t (H) of potential monitoring points that are affected by the involved release concentration, and repeat the above steps until the hyperedge set E 0 (H) of the release point is an empty set; Finally, obtain the optimized layout quantity and position of the potential monitoring points represented by the elements in the response node set.

5. The intelligent ventilation method for gradient pressure regulation of underground cavern groups based on a hypergraph network according to claim 1, characterized in that The definition of the air volume-air resistance response efficiency decay rate in step S6 is: the degree of change in the air volume-air resistance response efficiency of a branch caused by the change in the branch air resistance, and the expression of the air volume-air resistance response efficiency decay rate is: Among them, t ij represents the attenuation rate of the air volume-air resistance response efficiency, d ij represents the air volume-air resistance response efficiency, r j represents the air resistance value of the response node j, Δr j represents the small perturbation of the air resistance r j , q i represents the air volume at the response node j; the expression for fitting the response efficiency curve of the air volume-air resistance with a power function in the step S6 is: Among them, d ij represents the air volume-air resistance response efficiency, r j represents the air resistance value of response node j, and a, b, and c are all constants greater than zero; The calculation formula of the decay rate of the response efficiency in step S6 is: Among them, t ij represents the attenuation rate of the air volume - air resistance response efficiency, and t ij is always negative, that is, d ij decreases as r j increases, where r j represents the air resistance value of the response node j, and both a and b are constants greater than zero; The expression of the threshold value of the air resistance value in step S6 is: where r jmax represents the threshold value of the wind resistance, r j0 represents the initial value of the wind resistance, a, b, and c are all constants greater than zero, α represents the sensitivity threshold and 0 < α < 0.

05.

6. The intelligent ventilation method for gradient pressure regulation of underground cavern groups based on a hypergraph network according to claim 1, characterized in that The definition of the air volume-air resistance response efficiency in step S8 is: the degree of influence of the branch air resistance on the node pressure or the branch air volume, and the expression of the air volume-air resistance response efficiency is: Among them, d ij represents the air volume - air resistance response efficiency, r j represents the air resistance value of response node j, q i represents the air volume at target node i, Δr j represents a small perturbation of the air resistance r j , and Δq i represents the change in the air volume at target node i; The expression of the air volume-air resistance response efficiency matrix in step S8 is: Among them, D Q represents the air volume - air resistance response efficiency matrix, and d ij represents the air volume - air resistance response efficiency, q j represents the air volume at response node j, r n represents the air resistance value of the nth response node, and d jn represents the air resistance r of response node n n represents the influence degree on the air volume q of response node j itself j , and n represents the number of response nodes; The expression of the iteration sequence in step S8 is: dr j (k) = γ dr j (k-1) , Among them, d ij (k) represents the air volume - air resistance response efficiency in the k-th iteration, represents the air volume value of the target node i at the k-th iteration, represents the air volume value of the target node i at the (k - 1)-th iteration, r j (k) represents the air resistance value of the response node j at the k-th iteration, r j (k-1) represents the air resistance value of the response node j at the (k - 1)-th iteration, dr j (k) represents the air resistance perturbation, that is, the change in the air resistance of the response node j in the k-th iteration, dr j (k-1) represents the change in the air resistance of the response node j in the (k - 1)-th iteration, and γ represents the acceleration factor; The expression of the calculation accuracy is: |d ij (k) -d ij (k-1) | ≤ ε, where, d ij (k) represents the air volume-air resistance response efficiency in the k-th iteration, d ij (k-1) represents the air volume-air resistance response efficiency in the (k - 1)-th iteration, and ε represents the calculation accuracy.

7. A gradient pressure regulating intelligent ventilation method for underground cavern groups based on a hypergraph network according to claim 1, characterized in that, The definition of the fan power-air resistance response efficiency in step S9 is: the degree of influence of the change in the fan output power on the change in the branch air resistance, and the expression of the fan power-air resistance response efficiency in step S9 is: Among them, dN ij represents the fan power - wind resistance response efficiency, ΔN fi represents the change in the fan output power of the target node i, Δr j represents the change in the wind resistance of the response node j, N fi represents the fan output power at the target node i, r j represents the wind resistance value at the response node j; For the fan power-air resistance response efficiency dN ij , the fan output power N fi satisfies the product relationship between the fan air pressure and the fan air volume: N fi = h f q i , Among them, N fi represents the fan output power at the target node i, h f represents the fan air pressure, q i represents the air volume value at the target node i; Fan air pressure h f Satisfies the least squares fitting relationship of a unary polynomial. The degree of the polynomial takes the value of 3. Then the fan output power characteristic curve is shown as follows: Among them, N fi represents the fan output power at the target node i, and q i represents the air volume value at the target node i. a0 represents the constant term, that is, the fan power change rate when the air volume value q i is zero. a1, a2, and a3 all represent the coefficients of the polynomial, describing the non-linear influence of the air volume on the fan power; Partial derivative with respect to the air volume q i Find the partial derivative: Among them, N fi represents the output power of the fan at the target node i, and q i represents the air volume value at the target node i. a0 represents the constant term, that is, the power change rate of the fan when the air volume value q i is zero. a1, a2, and a3 all represent the coefficients of the polynomial, describing the non-linear influence of the air volume on the fan power; The expression of the fan power-air resistance response efficiency matrix in step S9 is: Among them, D N represents the fan power-air resistance response efficiency matrix, dN ij represents the fan power-air resistance response efficiency, N fn represents the fan output power at the response node n, r n represents the air resistance value of the nth response node, d nn represents the air resistance r of the response node n n the influence degree on the fan output power N of the response node n fn , ΔN fi represents the change in the fan output power of the target node i, Δr j represents the change amount of the air resistance of the response node j, d ij represents the air volume-air resistance response efficiency; The definition of the second-order air network response efficiency in step S9 is: adjusting branch e k The impact of the air resistance change on the air volume of other branches e i in the air network. The expression of the second-order air network response efficiency is: Among them, dd ijk represents the second-order wind network response efficiency, d ij represents the air volume-air resistance response efficiency, r k represents the air resistance at node k, q i represents the air volume value at the target node i, r j represents the air resistance value at the response node j; The three-dimensional matrix in step S9 is the second-order wind network response efficiency matrix, and the expression of the second-order wind network response efficiency matrix is: DD(i, j, k) = dd ijk , Among them, DD(i, j, k) is the second-order wind network response efficiency matrix, and dd ijk represents the second-order wind network response efficiency. When j = k, it represents the influence of branch regulation on its own air volume, that is, the effect efficiency attenuation rate of the regulation branch. When j ≠ k, it represents the magnitude of the mutual influence between branches. If two regulation branches e k and e j , satisfy the conditions: d ij > 0, d ik > 0, d ijk > 0, then it means that the two regulation branches promote each other for branch e i . If they satisfy the conditions: d ij > 0, d ik > 0, d ijk < 0, then it means that the two regulation branches inhibit each other for branch e i ; Take the second derivative of the loop wind pressure equation system vector with respect to the air resistance vector: Among them, (R, Q y (R)) indicates that the free variables of the loop air pressure equation set are the air resistance vector R and the co-tree air volume vector Q y , represents the Jacobian matrix of the loop air pressure equation set with respect to the air resistance, denoted as f1′, represents the Jacobian matrix of the loop air pressure equation set with respect to the co-tree air volume, denoted as f2′, is the Jacobian gradient algorithm matrix of the co-tree air volume with respect to the air resistance, denoted as f3′; Let Then the three-dimensional matrix analytical formula for the second-order wind network response efficiency is shown as follows: Among them, represents the co-tree air volume vector Q y the second derivative of the air resistance vector R, f″ 11 represents the second-order Jacobian matrix of the loop air pressure equation with respect to air resistance, f″ 12 represents the mixed second-order Jacobian matrix of the loop air pressure equation with respect to air resistance and co-tree air volume, f″ 21 represents the mixed second-order Jacobian matrix of the air pressure equation with respect to co-tree air volume and air resistance, f″ 22 represents the second-order Jacobian matrix of the loop air pressure equation with respect to co-tree air volume.

8. An intelligent ventilation system for gradient pressure regulation of an underground cavern group for performing the method described in claim 1, characterized in that, It includes a ventilation monitoring system, a network analysis system, a control optimization system and a visualization system. The network analysis system is respectively connected to the ventilation monitoring system and the control optimization system, and the visualization system is connected to the network analysis system. The ventilation monitoring system includes: an integrated air environment monitoring device, an air volume and air pressure monitoring device, and a fan operation parameter monitoring device. The integrated air environment monitoring device is arranged in the tunnel to monitor air environment parameters. The air volume and air pressure monitoring device is arranged inside the ventilation duct to collect air volume and air pressure data to support the construction of a hypergraph structure and the optimal layout of monitoring points. The fan operation parameter monitoring device is arranged near the fan and connected to the network analysis system to monitor the fan operation status. The network analysis system includes: an integrated intelligent fan control cabinet and a cloud server. The integrated intelligent fan control cabinet is respectively connected to the ventilation monitoring system and the control optimization system, and the cloud server is respectively connected to the integrated intelligent fan control cabinet and the visualization system. The integrated intelligent fan control cabinet includes: a data acquisition module, a data processing module, a calculation and analysis module, and a communication module. The data acquisition module is connected to the calculation and analysis module to collect data from the ventilation monitoring system. The calculation and analysis module is connected to the data processing module to perform the construction of the ventilation system diagram and the solution of the ventilation network based on the hypergraph network. The communication module is respectively connected to the data processing module and the cloud server. The control optimization system includes: an intelligent air valve and a variable-frequency fan control cabinet. The intelligent air valve is arranged on the ventilation duct and connected to the network analysis system to adjust the air volume and air resistance according to the ventilation network solution result. The opening range of the intelligent air valve is not less than 70%. The variable-frequency fan control cabinet is arranged near the fan and connected to the network analysis system to adjust the air volume and air resistance according to the ventilation network solution result. The operating efficiency of the fan is greater than 60%.

Images