A method and system for generating a running slurry plugging strategy based on coal mine separation layer grouting

Abstract

The application discloses the coal mine intelligent grouting control and scheduling technical field, disclose a kind of based on coal mine separation layer grouting's running grout plugging strategy generation method and system, including: using multi-source sensing equipment to collect grouting state parameters, construct state representation vector and establish weighted Petri net model;Based on historical state representation vector, extract long-term trend component and short-term residual component, and generate prediction state representation vector by fusion;System fusion current state and prediction state information, calculate each transfer site scheduling score, select the highest score transfer site to execute plugging operation;Based on grouting feedback information dynamic correction state vector and update weight function, realize the adaptive evolution of scheduling logic.Improve the accuracy and execution efficiency of plugging decision, effectively deal with the resource coordination and dynamic control problem under complex grouting working condition.

Description

Technical Field

[0001] This invention relates to the field of intelligent grouting control and scheduling technology in coal mines, specifically to a method and system for generating grout leakage sealing strategies based on coal mine delamination grouting. Background Technology

[0002] During coal mining, complex geological structures and frequent mining disturbances easily create hazardous areas such as rock strata delamination and fracture channels. If grouting is not timely or the method is inappropriate, the grout may run out of the intended sealing area along the fractures, resulting in "grout leakage." This not only wastes materials and increases treatment costs but also affects subsequent roadway support and safe production, and may even induce disasters such as water inrush and roof collapse. Currently, coal mine delamination grouting mainly relies on manual experience to set grouting parameters and plan sealing paths. It lacks effective modeling and prediction methods for complex geological changes and dynamic scheduling states, making it difficult to cope with sealing scheduling problems under the interaction of multiple factors. This exhibits limitations such as low level of intelligence, slow response, and rigid scheduling.

[0003] Existing research has attempted to introduce rule bases, expert systems, or simple logic models to assist in the optimization of grouting scheduling. However, these approaches suffer from problems such as rule lag, inability to adapt to changes in the field, and a lack of dynamic modeling capabilities for the evolution of the grouting process. In recent years, Petri nets have been gradually applied to fields such as manufacturing execution and network control due to their excellent visualization modeling and scheduling logic expression capabilities. However, when faced with complex evolution of closure states and dynamic changes in resource constraints, the static structure and fixed scheduling strategies of traditional Petri net models are insufficient to support flexible closure scheduling under complex operating conditions.

[0004] Therefore, there is an urgent need to construct a new blocking decision-making method that integrates state perception, trend prediction and adaptive evolution mechanism to achieve accurate identification of grout leakage risk and intelligent dynamic scheduling of blocking paths, so as to improve the safety, reliability and resource allocation efficiency of coal mine grouting management. Summary of the Invention

[0005] In view of the above-mentioned problems, the present invention is proposed.

[0006] Therefore, the technical problem solved by this invention is that existing coal mine delamination grouting processes suffer from inaccurate identification of grout leakage risks, untimely blocking and scheduling responses, and a lack of dynamic adaptive mechanisms.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a method for generating a grout leakage sealing strategy based on coal mine delamination grouting, comprising: collecting a set of grouting state parameters through a multi-source sensing device, and constructing a state representation vector based on the set of grouting state parameters;

[0008] Based on the state representation vector, a weighted Petri net model is constructed;

[0009] Based on the historical state representation vector, the long-term trend component and the short-term residual component are extracted and fused to generate the predicted state representation vector.

[0010] Using the predicted state representation vector and the state representation vector as input, the scheduling score for each transition bit is calculated.

[0011] Select the transfer bit with the highest scheduling score and execute the blocking scheduling operation;

[0012] Collect execution feedback information of blocking tasks, update the state representation vector and weighted Petri net model, and realize the adaptive evolution of scheduling logic.

[0013] As a preferred embodiment of the method for generating a grout leakage sealing strategy based on coal mine delamination grouting as described in this invention, the grouting state parameter set includes grouting process parameters, geological structure parameters, grouting material inventory information, grouting equipment operating status, operator scheduling status, and historical sealing data.

[0014] The grouting state parameter set is preprocessed, and after preprocessing, the state representation vector of the node is constructed based on the grouting state parameter set;

[0015] The state representation vector of the constructed node includes: establishing parameter combination judgment conditions based on the grouting state parameter set, and classifying the corresponding features into different levels according to the assignment rules; and combining the discrete feature values ​​of grout run risk, the discrete feature values ​​of sealing complexity, the discrete feature values ​​of resource capacity, and the discrete feature values ​​of operational availability to form the state representation vector of the node.

[0016] The assignment rules include: when all conditions are met simultaneously, the value is 1; when some conditions are met, the value is 0.5; and when no conditions are met, the value is 0.

[0017] The assignment of discrete feature values ​​includes performing a judgment based on the combined threshold conditions set by the grouting state parameter set, and assigning values ​​to the judgment results according to the assignment rules.

[0018] As a preferred embodiment of the grout leakage sealing strategy generation method based on coal mine delamination grouting described in this invention, the weighted Petri net model is based on the state representation vector of the nodes, and constructs a dynamic mapping relationship G = (S, T, E, Token, W) between the task state and behavior transition, where S represents the set of state bits; T represents the set of transition bits; E represents the set of directed edges; Token represents the dynamic tagging information corresponding to the state bits; and W represents the set of weight functions.

[0019] As a preferred embodiment of the grout leakage sealing strategy generation method based on coal mine delamination grouting described in this invention, wherein: during the operation of the weighted Petri net model, the dynamic tagging information Token(S) of each state bit in the state bit set S is used... i ), and in conjunction with the weight function set W, calculate the transition bit T corresponding to the state bit. j Scheduling score DS(T) j The formula is expressed as:

[0020]

[0021] Among them, W ij Indicates the status bit S i For the transfer bit T j The influence strength, and j represents the index; F(M(P) i )) is a correction function, which represents a weighted fusion of the state representation vector, including the slurry run risk value, the complexity of the plugging operation, the resource capability score, and the operational availability score;

[0022] Based on the scheduling scores of each transfer bit, within the current scheduling period, the transfer bit T with the highest score is selected from the transfer bit set T. * =argmax(DS(T) j This triggers the corresponding blocking action; where T * This represents the optimal transfer selected for execution within the current scheduling period; argmax represents finding the transfer bit with the highest score from all transfer bits; DS(T) j ) indicates the transfer bit T j The scheduling score;

[0023] When the status bit S exists i When the corresponding node has insufficient resources or its scheduling score is lower than the set execution threshold for m consecutive scheduling cycles, and no connected transition bit is triggered, a bridging transition bit T is automatically generated. bridge Connection status bit S m With status bit S i ; and a bridging path S is added to the weighted Petri net model. m →T bridge →S i ; Collect feedback information during the sealing process to correct the state representation vector of the corresponding node; update the weight function set W, increase the weight for successful paths and decrease the weight for failed paths; the feedback information includes the deviation between the actual grouting volume and the predicted value, the regional pressure recovery rate, the operation response delay and the equipment stability.

[0024] Every N scheduling cycles, the average trigger success rate of each transfer bit in the most recent L scheduling cycles is calculated; when the activity of each transfer bit is lower than the activity threshold, or the average trigger success rate of each transfer bit is lower than the set success rate threshold in c consecutive scheduling cycles, the transfer bit is removed from the transfer set.

[0025] As a preferred embodiment of the grouting strategy generation method based on coal mine delamination grouting described in this invention, the extraction of long-term trend components and short-term residual components includes: constructing a fixed-length state vector sequence based on the historical state representation vector of a node within each scheduling cycle; performing a sliding weighted average processing on the state vector sequence to obtain the long-term trend components of each feature dimension; and defining the difference between the current state vector and the long-term trend components as the short-term residual components.

[0026] The process of generating the fusion prediction state representation vector includes: calculating fusion weight coefficients based on the absolute values ​​of the long-term trend component and the short-term residual component in each feature dimension; using the fusion weight coefficients, performing a linear combination on the long-term trend component and the short-term residual component in each feature dimension to obtain a fusion prediction value; concatenating all fusion prediction values ​​in the order of feature dimensions to form a fusion prediction vector, and inputting the fusion prediction vector into a preset fully connected layer to generate the prediction state representation vector for the next scheduling cycle.

[0027] As a preferred embodiment of the grouting blocking strategy generation method based on coal mine delamination grouting described in this invention, the calculation of the scheduling score of each transfer position includes, in each scheduling cycle, obtaining the current state representation vector and the corresponding predicted state representation vector of each node, and constructing a fused state representation vector based on a weighted fusion method.

[0028] The scheduling score of each transition bit is calculated by utilizing the weight function relationship between each state bit and each transition bit. The score of each transition bit is calculated by the fused state representation vector corresponding to the state bit connected to the transition bit and the weight function. The scheduling scores of the transition bits are sorted, and the transition bit with the highest score is selected as the blocking action to be executed in the current scheduling cycle.

[0029] As a preferred embodiment of the grouting blocking strategy generation method based on coal mine delamination grouting described in this invention, the updating of the state representation vector and weighted Petri net model includes: after completing the blocking operation, collecting feedback information, correcting the state representation vector of the node, and updating the weight function set in the weighted Petri net model, increasing the weight of successful scheduling paths and decreasing the weight of failed scheduling paths; when a state bit is not scheduled within a consecutive preset scheduling period, and the scheduling scores of all transition bits connected to the state bit are lower than the score threshold, a bridging mechanism is triggered to generate bridging transition bits and bridging edges, connecting resource-rich state bits with unscheduled state bits, and constructing temporary bridging paths.

[0030] A system for generating a grout leakage sealing strategy based on coal mine delamination grouting, wherein: a data module collects a set of grouting state parameters through multi-source sensing devices, and constructs a state representation vector based on the set of grouting state parameters;

[0031] The model module constructs a weighted Petri net model based on the state representation vector;

[0032] The prediction module extracts long-term trend components and short-term residual components based on the historical state representation vector, and fuses them to generate a predicted state representation vector.

[0033] The scoring module takes the predicted state representation vector and the state representation vector as input to calculate the scheduling score for each transition bit.

[0034] The execution module selects the transfer bit with the highest scheduling score and performs the blocking scheduling operation.

[0035] The feedback module collects execution feedback information of the blocking task, updates the state representation vector and the weighted Petri net model, and realizes the adaptive evolution of the scheduling logic.

[0036] A computer device includes: a memory and a processor; the memory stores a computer program, characterized in that: when the processor executes the computer program, it implements the steps of the method described in any one of the present invention.

[0037] A computer-readable storage medium having a computer program stored thereon, characterized in that: when the computer program is executed by a processor, it implements the steps of the method described in any one of the present invention.

[0038] The beneficial effects of this invention are as follows: The method for generating a grout-blocking strategy based on coal mine delamination grouting provided by this invention constructs a state representation vector by fusing multi-source sensing data, and achieves dynamic mapping between nodes and behavioral paths by combining weighted Petri net modeling. It predicts the state representation vector based on historical state representation vectors, and calculates transition scores based on the fused state within the scheduling cycle and optimizes the execution path, effectively improving the real-time performance and accuracy of grouting scheduling. After scheduling, task feedback information is collected to dynamically correct the state vector and weight function, achieving continuous evolution and adaptive optimization of the structure, significantly enhancing the system's grouting response capability and resource regulation efficiency under complex geological conditions. Attached Figure Description

[0039] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1 The first embodiment of the present invention provides an overall flowchart of a method for generating a grout leakage sealing strategy based on coal mine delamination grouting. Detailed Implementation

[0041] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0042] Example 1, referring to Figure 1 As an embodiment of the present invention, a method for generating a grout leakage sealing strategy based on coal mine delamination grouting is provided, comprising:

[0043] S1: Collect a set of grouting state parameters through multi-source sensing devices, and construct a state representation vector based on the set of grouting state parameters.

[0044] During system operation, the set of nodes for sealing and scheduling is first identified based on mine construction drawings, geological structure models, historical grouting records, and real-time monitoring data, denoted as:

[0045] P = {P1, P2, ..., P} n}

[0046] Each node P iFor a specific spatial grouting unit, it can be identified from the following sources: (1) the coordinates of the grouting holes laid down underground and the tunnel distribution information in the design drawings; (2) the fracture zones, fault intersection zones, and hydraulic anomaly zones identified in the geological model; (3) typical grout leakage areas that appear multiple times in historical sealing data; and (4) the cluster points of areas where pressure changes and flow increases occur in the current monitoring and scheduling cycle. The system summarizes the above sources to form a node set P, which serves as the basic target object for scheduling and modeling.

[0047] During system operation, a set of grouting status parameters is collected by various sensing devices deployed underground in the mine and at the ground control terminal. The set of grouting status parameters includes: grouting process parameters, geological structure parameters, grouting material inventory information, grouting equipment operating status, operator scheduling status, and historical sealing data.

[0048] Grouting process parameters, including grouting pressure and grouting flow rate, are collected in real time by pressure sensors and flow sensors connected to the grouting pump, respectively; geological structure parameters, including rock fracture rate and porosity, are obtained from geological exploration data or sonic logging devices; grouting material inventory information is obtained in real time through an IoT storage module; grouting equipment operating status, including grouting pump start / stop status, current load, and equipment fault codes, are obtained from the PLC or DCS control interface; operator scheduling status is determined through linkage between the downhole positioning system and the task dispatch system to determine the availability and geographical distribution of current operators; historical plugging data, including plugging success rate, execution records in similar scenarios, plugging time, and effect evaluation indicators, are obtained from the grouting construction database.

[0049] After preprocessing the grouting state parameter set, the system constructs four types of state characteristic values ​​for nodes based on the normalized parameter information: grout leakage risk, sealing complexity, resource capacity, and operational availability. All characteristic values ​​are expressed using a three-valued discrete quantization method ({0, 0.5, 1}). These are used to characterize the state levels of different sealing nodes. The determination method for each characteristic value is as follows:

[0050] The risk of grout leakage is determined based on three types of indicators: grouting process parameters, geological structure parameters, and historical sealing data. The system determines whether the following three conditions are met simultaneously: the rate of change of grouting pressure is lower than a set threshold, the real-time flow rate is higher than the corresponding threshold of the historical average, and the historical average grouting value exceeds a reference benchmark; or the sum of fracture rate and porosity exceeds a preset threshold and the number of historical sealing failures exceeds a limited number. If any complete combination of conditions is met, the risk of grout leakage is assigned a value of 1; if only one of the above conditions is met, the value is assigned 0.5; otherwise, the value is assigned 0.

[0051] The complexity of the sealing process is determined based on geological structure parameters and historical sealing data. The system judges whether the following three conditions are met simultaneously: the number of lithological types reaches the set complexity standard, the number of small-angle fractures in the fracture intersection distribution exceeds the set number, the average sealing time exceeds the limit, or the sealing success rate is lower than the set level. If any complete combination of conditions is met, the sealing complexity is assigned a value of 1; if only one condition is met, the value is assigned 0.5; otherwise, the value is assigned 0.

[0052] Determining resource capacity depends on grouting material inventory information and the operating status of grouting equipment. The system determines whether the following two conditions are met simultaneously: the ratio of material inventory to demand is higher than a set ratio, and the equipment load is within a set range with no current fault information. If both conditions are met, the resource capacity is assigned a value of 1; if only one condition is met, the value is assigned 0.5; otherwise, the value is assigned 0.

[0053] The availability of the operation is determined by the scheduling status of the operators and the response indicators in the historical blocking data. The system determines whether the following two conditions are met simultaneously: the ratio of the number of currently available operators to the scheduling demand is greater than a set standard, and the historical average scheduling response latency is lower than a limited time threshold. If both conditions are met, the operation availability is assigned a value of 1; if only one condition is met, the value is assigned 0.5; otherwise, the value is assigned 0.

[0054] Based on the above four types of state feature values, a state representation vector is constructed to describe the node state, denoted as:

[0055] M(P i )=[R i D i C i A i ]

[0056] Among them, R i Represents node P i The discrete characteristic value of the grout leakage risk is obtained by weighted fusion calculation based on the pressure fluctuation coefficient calculated from pressure sensor data, the rock fracture rate at the location, and the historical sealing failure records corresponding to that node; D i Represents node P i The discrete characteristic value of the sealing complexity is obtained by weighting the structural difficulty level marked on the structural design drawings based on the construction environment parameters and structural layout characteristics of the area where it is located; C i Represents node P i The discrete characteristic values ​​of resource capacity are obtained by normalizing and superimposing three resource indicators based on the remaining slurry volume, the number of dispatchable grouting pumps, and the pipeline network accessibility; A i Represents node P iThe discrete characteristic value of the availability of the operation is generated by combining the real-time number of on-duty personnel and the online status of the grouting equipment with the scheduling plan. M(P) i ) represents node P i State representation vector; P i This represents the i-th grouting node, where i represents the node index.

[0057] By introducing multi-source sensing devices, the system can comprehensively collect core parameters during the grouting process, including grouting pressure, flow rate, rock stratum structure characteristics, material inventory, and operational status, greatly enhancing the system's ability to fully perceive the actual working conditions at the construction site. Based on this, the system constructs a standardized set of grouting status parameters and achieves high data consistency through time alignment, normalization, and outlier correction, significantly improving data quality and real-time performance. The final generated status representation vector covers multi-dimensional indicators such as grout leakage risk, sealing complexity, resource capacity, and operational availability, providing a structured and high-precision foundation for subsequent modeling and scheduling.

[0058] Most existing grouting scheduling schemes rely on manual experience or single sensor parameters, making it difficult to effectively identify key nodes under complex geological conditions. Furthermore, they suffer from low state awareness and lagging data updates. In contrast, by integrating drawing design, real-time monitoring, and historical data, this new method comprehensively identifies grouting scheduling targets from both spatial distribution and risk trend perspectives, forming a dynamic node set. Based on standardized processing results, a high-dimensional state vector is constructed, achieving comprehensive quantification and accurate modeling of the grouting unit's state. This effectively overcomes the traditional bottlenecks of "fuzzy node target identification" and "discrete parameters preventing modeling."

[0059] By constructing multidimensional state representation vectors for nodes, key indicators such as pressure fluctuation coefficient, fracture rate, structural complexity, resource supply capacity, and operational accessibility are systematically integrated. Compared to traditional schemes that rely solely on local parameters or static models, this approach significantly improves the completeness, interpretability, and computability of node state modeling. Furthermore, the state vectors serve as direct inputs for subsequent Petri net modeling and scheduling decisions, effectively constructing a closed-loop control chain from "perception—modeling—decision-making," thus driving a fundamental transformation of the grouting scheduling system from experience-driven to data-driven.

[0060] S2: Construct a weighted Petri net model based on the state representation vector.

[0061] The completed node state representation vector M(P) i )=[R i D i C i A iAfter constructing the state representation vector, a weighted Petri net model for blocking task modeling and scheduling control is further constructed based on the state representation vector. The weighted Petri net model is denoted as G = (S, T, E, Token, W), where S represents the set of state bits, describing the current state of a node; T represents the set of transition bits, describing behavioral events during the blocking scheduling process; E represents the set of directed edges, defining the logical connections and resource dependencies between state bits and transition bits; Token represents the dynamic tagging information corresponding to the state bits; and W represents the set of weight functions, defining the transition triggering conditions, resource consumption relationships, and scheduling priority calculation methods.

[0062] The system is based on a set of nodes P, and for each node P i Create the corresponding status bit S i And use the state representation vector as the state bit token:

[0063] Token(S i )=M(P i )=[R i D i C i A i ]

[0064] Among them, S i This represents the i-th state bit, corresponding to node P. i M(P) i ) represents node P i State representation vector; Token(S) i ) represents the state representation vector M(P) i ) as state bit S i The mark.

[0065] To characterize the dynamic impact of node states on scheduling behavior, based on the aforementioned state representation vector M(P) i ) and transfer behavior unit T j Based on the relationships between them, construct a global state weight matrix W, whose matrix elements W i,j Represents node P i Current state in relation to scheduling behavior T j Influence intensity; W i,j The formula is expressed as:

[0066] W i,j =f(M(P) i ),T j )

[0067] Where f(·) represents the scheduling weight calculation function, which maps the influence intensity according to the state and behavior; T j This represents the j-th transition behavior unit, such as "start grouting" or "call resources".

[0068] The global state weight matrix dynamically expresses the scheduling impetus of each node for a specific transition action, i.e., the scheduling priority tension.

[0069] The system constructs a corresponding set of transition bits T based on the possible behaviors of the blocking operation (such as initiating blocking, resource allocation, and blocking interruption). j However, unlike traditional methods that statically determine whether a transition has been triggered, this method defines a scheduling score function DS(T) for each transition bit. j The scoring function is dynamically calculated based on the weight matrix.

[0070]

[0071] Among them, DS(T) j ) indicates the transfer bit T j The scheduling score.

[0072] In the scoring calculation process, in order to represent the multidimensional state vector M(P) i The mapping is transformed into a single-valued result that can be used for transfer scoring, and a correction function F(M(P) is introduced. i )).

[0073] F(M(P i The state representation vector, including slurry spill risk value, plugging operation complexity, resource capability score, and operational availability score, is weighted and fused using the following formula:

[0074] F(M(P i ))=α1·R i +α2·D i +α3·C i +α4·A i

[0075] Where α1, α2, α3, and α4 represent weight coefficients; the correction function is used to compress multidimensional state information into scheduling score input, ensuring that the scoring mechanism is feasible and the parameters are controllable.

[0076] All transfer bits are sorted according to their contention score, and the system executes the transfer bit with the highest score in each scheduling cycle:

[0077] T * =argmaxDS(T j )

[0078] Among them, T * This represents the optimal transfer selected for execution within the current scheduling period; argmax represents finding the transfer with the highest score from all transfer bits; DS(T) j ) indicates the transfer bit T j The scheduling score.

[0079] To establish the connection structure between state bits and transition bits, the system associates state tokens with transition logic through a set of directed edges E. Each edge is assigned a weight value, representing the corresponding resource consumption or state threshold value. Furthermore, the system incorporates an adaptive bridging mechanism. When a node's state bit S... i When a node is in a low scheduling priority state due to insufficient resources or deterioration for an extended period, the system automatically triggers bridging rules and dynamically generates a temporary transition bit T connecting the node to surrounding nodes with sufficient resources. bridge With the corresponding connecting edge E bridge The cost and priority of bridging resource calls are dynamically adjusted by a set of weight functions W, causing the system structure to evolve dynamically with state feedback. The formula is as follows:

[0080] S m →T bridge →S i

[0081] Among them, S m Indicates an auxiliary node with sufficient resources or in good condition; T bridge This indicates a dynamically generated bridge transfer used to temporarily support blocked resources; S i This indicates that the node is currently experiencing resource shortages or scheduling bottlenecks.

[0082] After completing the above transfer operations, the system enters the status feedback phase. At this time, feedback information during the sealing process is collected, including: the deviation between the actual grouting volume and the estimated value; the pressure recovery rate of the grouting area; the actual response time of the operators; and the operational stability of the equipment.

[0083] The system uses this feedback information to correct the state representation vector of the corresponding node:

[0084] M(P i )←M(P i )+ΔM(P i )

[0085] Among them, M(P i ) represents node P i The current state representation vector; ΔM(P) i ) represents the state correction amount calculated based on feedback information.

[0086] The weight function set W is updated synchronously. The weights of the corresponding items in the scoring function are increased for paths with stable scheduling, and scheduling penalties are imposed on paths with frequent failures, thereby improving the robustness and structural evolution capability of the scheduling strategy.

[0087] To ensure the long-term self-evolution capability of the structure, the system performs a structural stability assessment every N scheduling cycles. At this assessment node, the system evaluates the stability of each transition bit T. j By reviewing the performance over the most recent k consecutive scheduling cycles, the average trigger success rate is calculated:

[0088]

[0089] in, Indicates the transfer bit T j The average trigger success rate over the most recent L scheduling cycles; L represents the number of scheduling cycles used to calculate the average, i.e., the width of the sliding window; t0 represents the starting scheduling cycle number used for backtracking statistics. This indicates that in the t-th scheduling period, the transfer bit T j The actual success rate of triggering.

[0090] For activity count(T) j Below the activity threshold θ active or transfer success rate A transition path whose success rate is below the success rate threshold θ for c consecutive scheduling cycles is identified as an inefficient path and automatically removed from the transition path (succ).

[0091] Set T←T\{T j This improves the overall scheduling efficiency and structural compactness of the network.

[0092] By constructing a weighted Petri net model based on state representation vectors, a dynamic mapping relationship between node states and scheduling behaviors was established, enabling accurate modeling of grouting and plugging tasks in multi-node, multi-constraint environments. Utilizing dynamic weighting functions and state labeling mechanisms, the system can calculate the scoring priority of each scheduling path in real time, ensuring the execution of the optimal transfer operation within each scheduling cycle. This effectively improves plugging response efficiency, resource allocation rationality, and the intelligence level of scheduling decisions. Simultaneously, the introduction of structural evolution and bridging mechanisms enables the model to possess self-evolutionary capabilities adapting to dynamic geological conditions and resource environments, significantly enhancing the robustness and long-term availability of the scheduling system in complex coal mine environments.

[0093] Unlike traditional methods that use static state descriptions and fixed scheduling rules, this approach introduces a state vector-driven mechanism and dynamic calculation logic for the scoring function, enabling state bits to have a quantifiable and updatable dynamic impact on scheduling transitions. Furthermore, unlike existing methods with fixed processes and immutable structures, this solution uses a bridging mechanism to achieve structural bypasses of scheduling bottleneck state bits, dynamically generating auxiliary transitions and connection edges. This effectively solves the problem of resource bottleneck nodes failing to schedule, leading to blocking task failures, and enhances the model's topological adaptability and response flexibility.

[0094] S3: Based on the historical state representation vector, extract the long-term trend component and the short-term residual component, and fuse them to generate the predicted state representation vector.

[0095] The system first processes each state bit S in the state bit set S. i The corresponding historical state representation vectors are sampled to construct a sequence of historical state representation vectors within a time window k:

[0096] X i =[M (t-k+1) (P i ),M (t-k+2) (P i ),…,M (t) (P i )]

[0097] in, Indicates node P during the t-th scheduling period i The state representation vector; k is the window length, representing the number of time steps involved in the prediction. X i Represents node P i Historical state representation vector sequence; M (t) (P i ) represents node P at time t in the scheduling period. i The state representation vector; t represents the current scheduling cycle, and k represents the time window length, i.e., the number of historical scheduling cycles involved in the prediction.

[0098] To fully explore the temporal evolution information in this historical state sequence, the original state sequence is decomposed into long-term trend components and short-term residual components.

[0099] Long-term trend component: Represents the overall trend of each feature dimension over time, used to depict the continuous changing trend of node states.

[0100] Short-term residual components: represent rapid fluctuations relative to the trend components, reflecting abnormal responses or sudden disturbances in grouting behavior.

[0101] Trend extraction of the state sequence is performed based on the weighted moving average method. For each state vector dimension p, a weighted long-term trend component is defined. for:

[0102]

[0103] in, Represents node P iWithin a scheduling period τ, the weighted long-term trend component on the p-th dimension; τ represents the scheduling period index; i represents the node number, ranging from [1, N], where N is the total number of nodes. p represents the p-th dimension in the state representation vector, ranging from [1, d], where d is the number of dimensions of each state vector; k represents the length of the time window; j represents the number of historical offset steps within the time window. t represents the time index of the current scheduling period. Represents node P i At time t, the actual state value in the p-th dimension; w j This represents the weighting coefficient corresponding to time step tj within the sliding window.

[0104] The short-run residual components are:

[0105] in, Represents node P i The short-term residual component in the p-th dimension at time t.

[0106] Long-term trend components and short-term residual components The two types of feature inputs will be fed into the prediction model to infer the state evolution trend.

[0107] To avoid representation shift caused by directly splicing trends and residuals, a long-term trend component is used instead. With short-term residual components By constructing adaptive fusion coefficients based on the relative magnitudes across each feature dimension, and achieving dimension-wise fusion through linear weighting, the stability and response accuracy of the prediction vector are improved.

[0108] First, define the fusion coefficient α on each dimension p. p This coefficient measures the dominance of a trend in the p-th dimension, and it is calculated as follows:

[0109]

[0110] Where ε is a small constant to prevent the denominator from being zero. p This represents the fusion coefficient in the p-th dimension.

[0111] Subsequently, the system performs a fusion operation on each feature dimension to obtain the value of each dimension of the predicted vector:

[0112]

[0113] in, Represents node P i The predicted fusion state value of the p-th dimension at time t+1 of the next scheduling period.

[0114] The above operation is equivalent to adjusting the degree of residual introduction based on the trend in each dimension, thereby achieving a dynamic balance between trend and resistance.

[0115] Finally, the fused values ​​of each dimension constitute the predicted representation vector of node Pi:

[0116]

[0117] Input the fused prediction vector into the fully connected mapping layer, and the output is a prediction state representation vector:

[0118]

[0119] in, Represents node P i The fused prediction representation vector in the time scheduling period t+1; (P i ) represents node P i The predicted state representation vector in scheduling period t+1; t represents the time step of the current scheduling period; W f b represents the weight matrix of the fully connected mapping layer; f This represents the bias vector of the fully connected mapping layer.

[0120] The fully connected mapping layer adopts a single-layer perceptron structure, including an input dimension of d and an output dimension of d. ′ Where d is the dimension of the fused prediction vector, d ′ To maintain consistency in the dimension of the predicted state representation vector, d′ = d is typically set. This mapping layer employs a linear transformation with a bias term and is superimposed with the nonlinear activation function ReLU (RectifiedLinearUnit) to enhance the model's fitting ability and its ability to express nonlinear trends. Its mapping function form is:

[0121]

[0122] Where W∈R represents the weight matrix of the fully connected mapping layer, and b∈R d′ Represents the bias vector; For node P i The fused prediction vector in scheduling period t+1 This is the predicted state representation vector output by the mapping. The weight matrix and bias vector are trained using a supervised learning method driven by historical state sequences and feedback results. The training objective is to minimize the mean squared error (MSE) between the predicted state and the actual state, thereby enhancing prediction accuracy and model generalization ability.

[0123] In the state prediction process, the historical state representation vector sequence of nodes is decomposed into long-term trend components and short-term residual components. This effectively characterizes the dual factors of "gradual change characteristics" and "abrupt disturbances" in the scheduling evolution process, overcoming the fuzzy representation problem caused by using only the original state sequence or simple sliding window prediction in existing methods. The long-term trend component extracts the stable trend of indicators in each dimension through weighted moving average, accurately reflecting the evolutionary main line of the blocking state, such as grouting risk, resource occupation, and operational capacity. The short-term residual component, as a high-frequency response of trend offset, characterizes abnormal events and short-term fluctuations, enabling the system to have sensitive identification capabilities for nonlinear disturbances, thereby effectively improving the prediction accuracy and scheduling robustness under dynamic operating conditions.

[0124] To further overcome the "scale imbalance" problem in the fusion of trend and residual components, the prediction mechanism introduces an adaptive fusion coefficient design based on relative amplitude. This coefficient is dynamically generated according to the ratio of the absolute values ​​of the trend component and the residual component, reflecting the balance between trend dominance and disturbance intensity in each dimension within the current scheduling period. Unlike existing splicing or fixed-weighted fusion methods, this structure can increase the sensitivity weight of the residual component during state transitions and strengthen the trend-dominant term during stable states, thus maintaining both anti-interference capability and responsiveness in the predicted values, achieving a "steady-state-disturbance self-adjustment" mechanism for the prediction representation.

[0125] The fused vectors are input to a fully connected mapping layer, further enhancing the coupling modeling capability between multi-dimensional predicted states. This ensures that the system's output predicted state representation not only retains the dynamic evolution information of each feature dimension but also constructs a nonlinear interaction structure between dimensions. This approach effectively overcomes the structural limitations of existing methods based on direct fitting and fixed threshold prediction, solves the problems of delayed and insensitive scheduling path selection under complex conditions, and significantly enhances the prior accuracy of the subsequent scheduling and scoring module and the adaptive capability of Petri net structure updates.

[0126] S4: Using the predicted state representation vector and the state representation vector as input, calculate the scheduling score for each transition bit.

[0127] Within each scheduling period t, the system uses all nodes P in the node set P. i The current state representation vector M (t) (P i and the corresponding predicted state representation vector Using the input as input, calculate each transition bit T in the transition bit set T. j The scheduling score is used to evaluate its trigger priority within the current scheduling cycle. First, a trend correction function is defined. Used to fuse the predicted state representation vector with the state representation vector:

[0128]

[0129] Among them, M (t) (P i ) represents the state representation vector in the current scheduling period t; α∈[0,1] represents the state fusion coefficient, initially set to α=0.6.

[0130] Based on the above fusion results, the system calculates the transition bit T by combining the weight relationship between the state bits and the transition bits. j Scheduling scoring function:

[0131]

[0132] Among them, DS (t) (T j ) indicates the transfer bit T j The score value in the current scheduling cycle; S i Indicates the relationship with node P i The corresponding status bit.

[0133] By introducing a weighted fusion mechanism of the predicted state representation vector and the current state representation vector in each scheduling cycle, a fused state representation vector is constructed and used as input to participate in the scheduling score calculation of each transition bit. This enables a feedforward response to the future state trend of nodes, enhances the scheduling strategy's comprehensive perception of the complexity of blocking tasks, resource fluctuations, and operational timeliness, and thus improves the foresight and accuracy of scheduling decisions.

[0134] Compared to traditional scoring methods that rely on static rules or historical averages, this method integrates the current state with the predicted trend to construct a trend correction function, dynamically adjusting the state input of each node in the scoring mechanism. This allows the scheduling score to not only reflect the current resource status but also the direction of trend changes, significantly improving the adaptability and response efficiency of scheduling behavior in dynamic environments.

[0135] This mechanism breaks through the technical bottlenecks of existing blocking scheduling, such as "inability to integrate trend changes", "lag in scheduling response" and "rigid fixed scoring model". It realizes the trend adaptive adjustment of the state representation vector and the dynamic evolution of the scheduling scoring mechanism, providing an intelligent and controllable scheduling strategy foundation for high-risk, multi-path blocking tasks under complex working conditions.

[0136] S5: Select the transition bit with the highest scheduling score and execute the blocking scheduling operation; collect the execution feedback information of the blocking task, update the state representation vector and the weighted Petri net model, and realize the adaptive evolution of the scheduling logic.

[0137] Based on the ranking of all scores, the transition bit with the highest score is selected as the action to be executed in the current scheduling cycle:

[0138]

[0139] Among them, T *(t) The target transition bit with the highest score in the t-th scheduling cycle is represented; argmax represents the variable that maximizes the value of the subsequent function; T j DS represents the j-th candidate shift bit in the shift bit set; (t) (T j ) represents the transfer bit T in the t-th scheduling cycle. j The scheduling score.

[0140] To improve the scheduling response capability of inactive nodes, if a certain state bit S k If a node has not been scheduled for m consecutive scheduling cycles and all the scores of its connected transition bits are below a set score threshold, a bridging mechanism is automatically triggered to dynamically generate a temporary bridging path.

[0141]

[0142] By introducing temporary transfer bits With bridge side E bridge Connection status bit S r With S k Based on the set of weight functions W, the system dynamically calculates the calling cost and priority of bridging paths, thereby achieving dynamic decoupling and adaptive evolution of the network topology. After the scheduling behavior is completed, the system synchronously collects feedback information on the blocking task, including the deviation between the grouting volume and the target value, the pressure recovery rate of the grouting area, the response time of the operators, and the stability of equipment operation.

[0143] Based on the feedback information, the system corrects the state representation vector of the corresponding node:

[0144] M (t+1) (P k )←M (t) (P k )+ΔM(P k )

[0145] Among them, S r Auxiliary state bit indicating sufficient resources and an active state; S represents the temporary bridging transfer bit dynamically introduced during scheduling period t; k This indicates a target state bit that has been continuously not scheduled or has a consistently low score; M (t) (P k ) represents node P k The state representation vector in the current scheduling period t; ΔM(P) k ) represents node P calculated based on feedback information. k State correction increment; M (t+1)(P k ) represents the updated node P in the (t+1)th scheduling cycle. k The new state representation vector.

[0146] Simultaneously, the weight function set W is updated to increase the weight of well-performing paths in the scoring function and reduce the scheduling priority of repeatedly failing paths, thereby enhancing the stability of the overall scheduling strategy and the adaptive capability of the structure. To further support long-term optimization of the network structure, the system performs a structural evaluation every N scheduling cycles, including: node activity statistics; transition trigger success rate analysis; and bridging mechanism call frequency analysis.

[0147] Node activity statistics: Evaluate the frequency of each status bit participating in scheduling in the most recent k periods. When it is lower than the activity threshold, it is marked as a candidate for deletion.

[0148] Transfer bit trigger success rate analysis: Calculate the ratio of the actual number of successful executions of the transfer bit to the total number of triggers. If the ratio is lower than the set success rate threshold for a continuous period, reduce its scheduling priority or remove it.

[0149] Bridge mechanism call frequency analysis: When a certain transfer path is frequently triggered by bridging operations, it indicates that the original structure has failed and the original path structure needs to be replaced or reconstructed.

[0150] Based on the feedback results of scheduling execution, the state representation vector is dynamically corrected and the weighted Petri net model structure is updated to achieve continuous optimization and evolution of the blocking scheduling logic, thereby enhancing the system scheduling stability and resource regulation intelligence in complex environments.

[0151] After completing the state representation vector correction and weighted Petri net weight update based on scoring feedback, the "state bit-transition bit-feedback score" trajectory information of each round of scheduling is further recorded, and a historical behavior trajectory graph is constructed. The trajectory graph adopts a directed graph storage method based on a hash index structure, using the state bit-transition bit pair as the directed edge index key in the graph, and associating the scoring result, execution feedback, and timestamp information of each round of execution to form a path behavior record pool. Every T... eval In each scheduling cycle, the system performs cluster analysis on the path data in the trajectory map to assess its performance. Clustering features include: time-series fluctuations in score values, path trigger frequency, mean feedback deviation, and stability. Through joint evaluation using K-Means clustering and the Local Anomaly Factor (LOF) algorithm, the following two types of behavioral paths are extracted:

[0152] High-efficiency trajectory: The score is stable over a long period of time, the trigger frequency is high, and the feedback effect is good.

[0153] Failure trajectory: drastic fluctuations in scores, continuous deviations from expected goals, and low success rate.

[0154] For the transition path corresponding to the failed trajectory, the following strategy is implemented: reduce its base weight in the scoring function and add it to the disabled candidate pool, in consecutive N... fail After a failure, temporary removal is performed. For efficient trajectory paths, an additional incentive weight Δw is added to the scoring function. eff Furthermore, its priority in the candidate transition bit sorting is adjusted to form a strategic scheduling guide. The above processing further breaks the static constraints of the weighted Petri net path structure, enabling the system to self-evolve and optimize its structure based on historical effects, thereby improving scheduling adaptability and efficiency in complex environments.

[0155] During the blocking strategy generation process, the scoring-driven mechanism calculates the scheduling score of each transfer path and selects the path with the highest score as the current scheduling action, which significantly improves the matching degree of scheduling behavior to the current state. Simultaneously, a bridging mechanism is designed to automatically trigger when a node with a persistently low score or has not been scheduled for a period of time, generating a temporary bridging path connecting "resource-rich state bits" and "inactive state bits." This solves the problem of resource scheduling deadlock in low-activity areas in traditional Petri net structures, enhancing path flow and structural flexibility during the blocking task execution process. This strategy effectively avoids the "scheduling starvation" phenomenon and overcomes the problem of path reconfiguration in existing static topology scheduling.

[0156] After the scheduling action is completed, the system collects multi-dimensional feedback information on the grouting behavior, including key parameters such as grouting volume deviation, pressure recovery speed, operation response delay, and equipment stability. Based on this, the relevant state representation vectors are corrected, and the path weights in the weighted Petri net are adjusted simultaneously. Compared with traditional blocking scheduling strategies based on static rules, this structure achieves continuous parameter self-correction based on actual task execution feedback. It can dynamically strengthen high-quality paths and suppress inefficient paths, effectively cope with the interference of uncertain factors such as changes in geological conditions and resource fluctuations, thereby enhancing the stability and accuracy of the overall blocking system.

[0157] After each round of scheduling, the system records the execution trajectory of "state bit → transition bit → feedback score," constructing a behavioral trajectory graph. Within the structural evaluation cycle, it performs cluster analysis on the trajectory execution effects to identify "repeated failure trajectories" and "efficient trajectories." By disabling or downgrading failed paths and applying incentive scores to high-quality paths, a self-evolutionary mechanism based on execution history is formed. This path evolution process does not rely on a pre-set fixed graph structure but actively adjusts the topology connections based on scheduling effects. This significantly breaks through the static limitations of traditional Petri nets in path structure, endowing the model with the ability to continuously adapt to dynamic task distributions, demonstrating strong robustness and intelligent evolutionary capabilities.

[0158] Example 2, an embodiment of the present invention, provides a grout leakage sealing strategy generation system based on coal mine delamination grouting, comprising:

[0159] The data module collects a set of grouting state parameters through multi-source sensing devices and constructs a state representation vector based on the set of grouting state parameters.

[0160] The model module constructs a weighted Petri net model based on the state representation vector.

[0161] The prediction module extracts long-term trend components and short-term residual components based on the historical state representation vector, and then fuses them to generate a predicted state representation vector.

[0162] The scoring module takes the predicted state representation vector and the state representation vector as input and calculates the scheduling score for each transition bit.

[0163] The execution module selects the transfer bit with the highest scheduling score and performs the blocking scheduling operation.

[0164] The feedback module collects execution feedback information of the blocking task, updates the state representation vector and the weighted Petri net model, and realizes the adaptive evolution of the scheduling logic.

[0165] Example 3, an embodiment of the present invention, differs from the previous two embodiments in that:

[0166] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0167] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device).

[0168] "Computer-readable medium" can be any means that can contain, store, communicate, propagate or transmit programs for use by or in conjunction with an instruction execution system, apparatus or device.

[0169] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0170] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0171] Example 4 is an embodiment of the present invention, which provides a method and system for generating a grout leakage sealing strategy based on coal mine delamination grouting. In order to verify the beneficial effects of the present invention, a simulation experiment is conducted for scientific demonstration.

[0172] The W1207 working face in a certain mining area was selected as the experimental site. This working face has a typical delamination development zone, is prone to grout leakage during grouting, and has the characteristics of complex geological structure and multi-point sealing requirements, making it suitable for verifying the scheduling capability and reliability of the scheme of the present invention.

[0173] Before the experiment began, six sets of multi-source sensing devices were deployed at the W1207 working face, including grouting flow meters, downhole ground pressure sensors, grouting pressure sensors, material inventory detection units, electric grouting machine status detection modules, and a personnel positioning system. These devices collected data in real time, including grouting pressure, rock stress, grouting rate, equipment load, personnel configuration, and remaining materials, to construct a grouting status parameter set. After normalization and feature dimensionality reduction analysis of this parameter set, a four-dimensional state representation vector was generated, representing grout leakage risk, sealing complexity, resource capacity, and operational availability.

[0174] A weighted Petri net structure centered on nodes is constructed, encoding the dynamic evolution relationship between state positions and transition positions into a set of weight functions. Using scheduling cycles as units, the system performs a weighted moving average on the sequence of historical state representation vectors of nodes to extract long-term trend components and calculates the residuals between these components and the actual state values ​​to obtain short-term perturbation characteristics. By calculating the relative amplitudes of the two types of components, the fusion coefficient for each dimension is obtained, leading to the fused predicted state vector for the next cycle.

[0175] The scheduling scoring function calculates a weighted score for each transition position based on the fused state input consisting of the current state vector and the predicted vector, and selects the transition position with the highest score to trigger the corresponding blocking action. After the scheduling execution is completed, the system collects feedback information such as the actual grouting volume, pressure recovery rate, response delay, and equipment stability, and adjusts the original state representation vector and Petri net structure weights accordingly.

[0176] The entire experiment consisted of 20 scheduling cycles to ensure that the system strategy performed adequately under different operating conditions.

[0177] During the experiment, a total of 9 nodes were identified, involving 26 valid transfer paths. In the initial stage (cycles 1 to 5), some nodes failed to block the blockade and had scores below 0.35. Node 3, in particular, had scores of only 0.28 and 0.31 in the first two cycles and was ultimately not triggered. Through trend and residual prediction mechanisms, starting from cycle 6, the score of node 3 rose to 0.46, and it was selected for blocking in cycle 8. The actual grouting volume deviation was controlled within ±3.2%, and the regional pressure recovery time was shortened to 81.5% of the original.

[0178] During continuous scheduling, the system automatically identified and disabled four transfer paths with high failure rates. After the score was below 0.4 for three consecutive cycles, the bridging mechanism was triggered. Node 6 successfully established a temporary bridging path with Node 2. Grouting scheduling was completed through this path in the 13th cycle. The operation response latency decreased by 26.4% compared to the initial stage, and the equipment load stability improved by 9.8%.

[0179] During the 15th to 20th cycles, the system entered a stable scheduling phase, with the average success rate of triggering transfer paths per cycle increasing to 92.3%, a significant difference from the 71.6% in the initial phase.

[0180] Experimental results show that the slurry blocking strategy generation method constructed in this invention is superior to existing solutions in terms of dynamic scheduling efficiency, adaptive adjustment capability, and blocking behavior response accuracy.

[0181] First, the decomposition mechanism based on long-term trends and short-term residuals can more accurately characterize the state evolution trend of nodes, avoiding scheduling errors caused by traditional methods that rely solely on the current state. In the experiment, several nodes with initially low scores were gradually re-identified and triggered under the guidance of predicted trends, preventing resource idleness.

[0182] Secondly, the integration of predicted state vectors and real-time states to construct the scheduling score input significantly improves the accuracy of transfer selection. After incorporating prediction logic into the scoring model, the actual grouting deviation of the blocking action decreased from an average of ±7.8% to ±3.1%, and the scheduling response delay decreased by an average of 23.7%, indicating that the scoring mechanism is more forward-looking and real-time.

[0183] Furthermore, by collecting scheduling feedback information, the Petri net model achieves adaptive evolution, enabling the system to possess dynamic learning and structural reconstruction capabilities. With the assistance of bridging mechanisms and failed path elimination strategies, the model can gradually optimize the path structure, forming a scheduling network topology that is more adapted to the current environment, effectively avoiding redundant computation and inefficient scheduling caused by structural rigidity.

[0184] Overall, this approach overcomes the problem that traditional static sealing strategies cannot respond to changes in on-site disturbances in a timely manner. It constructs an intelligent scheduling mechanism that integrates predictability, self-adjustment, and evolution. The technical approach is innovative, and the practical effect has engineering promotion value. It can significantly improve the intelligence level and operational stability of coal mine delamination grouting sealing systems.

[0185] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for generating a grout leakage sealing strategy based on coal mine delamination grouting, characterized in that, include: A set of grouting state parameters is collected by multi-source sensing devices, and a state representation vector is constructed based on the set of grouting state parameters. Based on the state representation vector, a weighted Petri net model is constructed; Based on the historical state representation vector, the long-term trend component and the short-term residual component are extracted and fused to generate the predicted state representation vector. Using the predicted state representation vector and the state representation vector as input, the scheduling score for each transition bit is calculated. Select the transfer bit with the highest scheduling score and execute the blocking scheduling operation; Collect execution feedback information of blocking tasks, update the state representation vector and weighted Petri net model, and realize the adaptive evolution of scheduling logic.

2. The method for generating a grout leakage sealing strategy based on coal mine delamination grouting as described in claim 1, characterized in that: The grouting status parameter set includes grouting process parameters, geological structure parameters, grouting material inventory information, grouting equipment operating status, operator scheduling status, and historical sealing data; The grouting state parameter set is preprocessed, and after preprocessing, the state representation vector of the node is constructed based on the grouting state parameter set; The state representation vector of the constructed node includes: establishing parameter combination judgment conditions based on the grouting state parameter set, and classifying the corresponding features into different levels according to the assignment rules; and combining the discrete feature values ​​of grout run risk, the discrete feature values ​​of sealing complexity, the discrete feature values ​​of resource capacity, and the discrete feature values ​​of operational availability to form the state representation vector of the node. The assignment rules include: when all conditions are met simultaneously, the value is 1; when some conditions are met, the value is 0.5; and when no conditions are met, the value is 0. The assignment of discrete feature values ​​includes performing a judgment based on the combined threshold conditions set by the grouting state parameter set, and assigning values ​​to the judgment results according to the assignment rules.

3. The method for generating a grout leakage sealing strategy based on coal mine delamination grouting as described in claim 2, characterized in that: The weighted Petri net model is based on the state representation vector of the nodes, and constructs a dynamic mapping relationship G = (S, T, E, Token, W) between the task state and the behavior transition, where S represents the set of state bits; T represents the set of transition bits; E represents the set of directed edges; Token represents the dynamic tag information corresponding to the state bits; and W represents the set of weight functions.

4. The method for generating a grout leakage sealing strategy based on coal mine delamination grouting as described in claim 3, characterized in that: During the operation of the weighted Petri net model, the dynamic tagging information Token(S) of each state bit in the state bit set S is used. i ), and in conjunction with the weight function set W, calculate the transition bit T corresponding to the state bit. j Scheduling score The formula is expressed as: Among them, W ij Indicates the status bit S i For the transfer bit T j The influence strength, i and j represent indices; F(M(P) i )) is a correction function, which represents a weighted fusion of the state representation vector, including the slurry run risk value, the complexity of the plugging operation, the resource capability score, and the operational availability score; Based on the scheduling scores of each transfer bit, within the current scheduling period, the transfer bit T with the highest score is selected from the transfer bit set T. * =argmax(DS(T) j This triggers the corresponding blocking action; where T * This represents the optimal transfer selected for execution within the current scheduling period; argmax represents finding the transfer bit with the highest score from all transfer bits; DS(T) j ) indicates the transfer bit T j The scheduling score; When the status bit S exists i When the corresponding node has insufficient resources or its scheduling score is lower than the set execution threshold for m consecutive scheduling cycles, and no connected transition bit is triggered, a bridging transition bit T is automatically generated. bridge Connection status bit S m With status bit S i ; and a bridging path S is added to the weighted Petri net model. m →T bridge →S i ; Collect feedback information during the sealing process to correct the state representation vector of the corresponding node; update the weight function set W, increase the weight for successful paths and decrease the weight for failed paths; the feedback information includes the deviation between the actual grouting volume and the predicted value, the regional pressure recovery rate, the operation response delay and the equipment stability. Every N scheduling cycles, the average trigger success rate of each transfer bit in the most recent L scheduling cycles is calculated; when the activity of each transfer bit is lower than the activity threshold, or the average trigger success rate of each transfer bit is lower than the set success rate threshold in c consecutive scheduling cycles, the transfer bit is removed from the transfer set.

5. The method for generating a grout leakage sealing strategy based on coal mine delamination grouting as described in claim 4, characterized in that: The extraction of long-term trend components and short-term residual components includes: constructing a fixed-length state vector sequence based on the historical state representation vector of the node within each scheduling cycle; performing a sliding weighted average on the state vector sequence to obtain the long-term trend components of each feature dimension; and defining the difference between the current state vector and the long-term trend components as the short-term residual components. The process of generating the fusion prediction state representation vector includes: calculating fusion weight coefficients based on the absolute values ​​of the long-term trend component and the short-term residual component in each feature dimension; using the fusion weight coefficients, performing a linear combination on the long-term trend component and the short-term residual component in each feature dimension to obtain a fusion prediction value; concatenating all fusion prediction values ​​in the order of feature dimensions to form a fusion prediction vector, and inputting the fusion prediction vector into a preset fully connected layer to generate the prediction state representation vector for the next scheduling cycle.

6. The method for generating a grout leakage sealing strategy based on coal mine delamination grouting as described in claim 5, characterized in that: The calculation of the scheduling score for each transition bit includes obtaining the current state representation vector and the corresponding predicted state representation vector for each node in each scheduling cycle, and constructing a fused state representation vector based on a weighted fusion method. The scheduling score of each transition bit is calculated by utilizing the weight function relationship between each state bit and each transition bit. The score of each transition bit is calculated by the fused state representation vector corresponding to the state bit connected to the transition bit and the weight function. The scheduling scores of the transition bits are sorted, and the transition bit with the highest score is selected as the blocking action to be executed in the current scheduling cycle.

7. The method for generating a grout leakage sealing strategy based on coal mine delamination grouting as described in claim 6, characterized in that: The update of the state representation vector and weighted Petri net model includes: after completing the blocking operation, collecting feedback information, correcting the state representation vector of the node, and updating the set of weight functions in the weighted Petri net model, increasing the weight of successful scheduling paths and decreasing the weight of failed scheduling paths; when a state bit is not scheduled within a consecutive preset scheduling period, and the scheduling scores of all transition bits connected to the state bit are lower than the score threshold, a bridging mechanism is triggered to generate bridging transition bits and bridging edges, connect resource-rich state bits with unscheduled state bits, and construct a temporary bridging path.

8. A system for generating a grout leakage sealing strategy based on coal mine delamination grouting, employing the method described in any one of claims 1-7, characterized in that: The data module collects a set of grouting state parameters through multi-source sensing devices and constructs a state representation vector based on the set of grouting state parameters. The model module constructs a weighted Petri net model based on the state representation vector; The prediction module extracts long-term trend components and short-term residual components based on the historical state representation vector, and fuses them to generate a predicted state representation vector. The scoring module takes the predicted state representation vector and the state representation vector as input to calculate the scheduling score for each transition bit. The execution module selects the transfer bit with the highest scheduling score and performs the blocking scheduling operation. The feedback module collects execution feedback information of the blocking task, updates the state representation vector and the weighted Petri net model, and realizes the adaptive evolution of the scheduling logic.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the method for generating a grout leakage sealing strategy based on coal mine delamination grouting as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the method for generating a grout leakage sealing strategy based on coal mine delamination grouting as described in any one of claims 1 to 7.

Images