Coal mine production business optimization method based on industrial knowledge graph

Abstract

The invention discloses a coal mine production business optimization method based on an industrial knowledge graph, and relates to the field of coal mine production, and the method comprises the steps: constructing a coal mine production knowledge graph according to historical ledger data; extracting prior constraints, and performing causal structure learning to obtain a coal mine production causal diagram; constructing a variable influence structure equation set according to the causal diagram; carrying out anti-fact simulation, calculating key indexes under different intervention behaviors, and taking the intervention behaviors meeting key index thresholds as feasible scheme domains; and performing multi-target optimization on the feasible scheme domain to obtain an optimization scheme set, and writing the scheme conforming to the comprehensive benefit score threshold value as a configuration scheme back to the knowledge graph. The influence of different intervention measures on safety, yield and cost can be evaluated, so that managers can solve the core contradiction between safety and benefit in coal mine production, the optimal solution of production efficiency and cost control is found on the premise that safety is guaranteed, and fine control over coal mine production business risks is facilitated.

Description

A Coal Mine Production Operation Optimization Method Based on Industrial Knowledge Graph Technical Field

[0001] The present invention relates to the technical field of coal mine production optimization, and in particular to an industrial knowledge graph-based method for optimizing coal mine production operations. Background Technology

[0002] Coal mine production technology is a complex technology that involves a large amount of data collection, analysis and decision-making. In the process of coal mine production, production managers need to consider many factors, such as safety, production costs and production efficiency.

[0003] Coal mine production mainly relies on manual monitoring and experience-based judgment, which leads to inefficiency and inaccurate decision-making. Furthermore, the complex underground environment and production conditions can easily affect the production process, making it impossible to achieve the production managers' expected goals and requiring adjustments to production parameters. Currently, there is a lack of intelligent analysis methods to help producers analyze production conditions and predict potential risks, making it impossible to quickly provide producers with constructive decision-making advice. Summary of the Invention

[0004] The technical problem solved by this invention is that the existing technology lacks an intelligent analysis system that helps producers analyze production conditions and predict potential risks, and cannot quickly provide producers with constructive decision-making opinions.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a coal mine production business optimization method based on an industrial knowledge graph, comprising: step S100: acquiring historical ledger data of coal mine production, and constructing a coal mine production knowledge graph based on the historical ledger data; step S200: extracting prior constraints from the coal mine production knowledge graph, and using a causal constraint algorithm to perform causal structure learning to obtain a coal mine production causal graph; step S300: constructing a set of variable influence structure equations based on the coal mine production causal graph; step S400: performing counterfactual simulation based on the set of variable influence structure equations, calculating key indicators under different intervention behaviors, and taking intervention behaviors that meet the threshold of key indicators as feasible solution domains; step S500: performing multi-objective optimization on the feasible solution domains to obtain an optimized solution set, calculating the comprehensive benefit score of the optimized solution set, and taking the solution that meets the comprehensive benefit score threshold as a configuration solution; step S600: adding the configuration solution to the graph to obtain an updated coal mine production knowledge graph.

[0006] Preferably, step S100 specifically includes: Step S110: acquiring historical ledger data of coal mine production, and dividing the historical ledger data into structured text and unstructured text; Step S120: extracting entities, relationships, and attributes from the structured text through rule mapping, extracting entities, relationships, and attributes from the unstructured text through semantic recognition technology, and aligning the same entities in different expression forms using OpenEA to generate structured triples; wherein, the structured text includes production reports, equipment ledgers, resource consumption lists, environmental monitoring data, and spreadsheets, and the unstructured text includes shift handover records, maintenance records, meeting minutes, and accident investigation reports. The entities include equipment, production, safety, personnel, and resources; the relationships include location, belonging, participation, causing, and adjacency; the equipment category includes equipment number, equipment name, equipment operating parameters, and equipment maintenance work order; the production category includes mining location, working face, footage, daily output, operating rate, and cycle number; the safety category includes gas concentration, ventilation system monitoring data, and roof pressure monitoring data; the personnel category includes the number of people going down the mine, personnel positions, and personnel ages; the resource consumption category includes support materials, accessory materials, water resources, and electricity resources; Step S130: Import the structured ternary into the graph database for storage to generate a coal mine production knowledge graph.

[0007] Preferably, step S200, extracting prior constraints from the coal mine production knowledge graph, specifically includes: Step S210: Based on the coal mine production knowledge graph, extract prior constraints, which include allowed edges, prohibited edges, and preferred edges; allowed edges represent causal relationships that directly or indirectly exist in the coal mine production knowledge graph, prohibited edges represent causal relationships that cannot exist in the coal mine production knowledge graph, and preferred edges represent causal relationships with a clear temporal order; Step S220: By setting the logical hierarchy sequence of variables, allowed edges are transformed into constraints that restrict the direction of causal search, allowing only the detection from earlier variables to later variables; variables that are prohibited from connection are designated as detection rules to be skipped, directly excluding unreasonable causal relationship searches; by explicitly specifying directed edges that must exist, preferred edges are transformed into mandatory assertions on the algorithm, locking in the core causal path.

[0008] Preferably, in step S200, the causal structure learning using the causal constraint algorithm to obtain the coal mine production causal graph specifically includes: Step S230: Organizing the historical ledger data into a structured numerical matrix X, with the expression: Where n represents the number of samples and p represents the number of variables; Step S240: Input the prior constraints after format conversion into the FCI causal discovery algorithm, perform conditional independence test on the structured numerical matrix, output a partial directed acyclic graph, orient the unoriented edges in the partial directed acyclic graph through V-structure rules and experience in the coal mine production field, and obtain a directed acyclic graph as the causal graph of coal mine production.

[0009] Preferably, step S300 specifically includes: Step S310: taking the nodes in the cause-effect graph of the coal mine production business as endogenous variables, constructing a structural equation for each endogenous variable, and generating a variable influence structural equation, the expression of which is: ;in, Endogenous variables in a causal graph of coal mine production The set of parent nodes, Represents the set of parameters to be estimated. Indicates the error term. The structure function is represented; Step S320: Using the structure constraints of the structured numerical matrix and the coal mine production cause-effect graph, the parameters in each equation are estimated using the maximum likelihood method; Step S330: Residual analysis is performed on each equation to ensure that the error of each equation is within the first allowable threshold; Step S340: All variable influence structure equations are integrated to obtain a variable influence structure equation set.

[0010] Preferably, step S400 specifically includes: step S410: selecting intervention variables and setting intervention values; step S420: calculating the counterfactual value of each endogenous variable according to the topological order of the coal mine production causal diagram; step S430: performing repeated calculations using the Monte Carlo method to obtain the probability distribution of the endogenous variables, the expression of which is: ;in, Represents the outcome variable. Indicates mandatory intervention variable , will variables The value is set to Z represents the set of other influencing factors. Represents a probability distribution. This represents the summation over all possible influencing factors; Step S440: Based on the probability distribution, calculate the expected value of each outcome variable.

[0011] Preferably, step S400 further includes: step S450: based on coal mine production experience, select multiple intervention variables for multi-scenario simulation, repeat steps S410-S440, and obtain multi-scenario simulation results; step S460: integrate the multi-scenario simulation results into a multi-dimensional optimization matrix, wherein the rows of the multi-dimensional optimization matrix represent different intervention behaviors, the columns represent the mathematical expectation of the intervention result variables, and the mathematical expectation is recorded as a key indicator; step S470: based on a pre-set threshold group, exclude scenarios that do not meet the threshold group to obtain a feasible solution domain, wherein the threshold group includes a production cost threshold, a daily output threshold, and a safety risk threshold.

[0012] Preferably, step S500 specifically includes: step S510: taking the values ​​of the intervention variables in the feasible solution domain as decision variables, denoted as vector X, with the expression: ;in, Indicates the first There are several adjustable parameters, where m represents the total number of decision variables; Step S520: With maximizing daily output, minimizing safety risks, and minimizing costs as production optimization objectives, construct a production optimization function, the expression of which is: ; ; ; ;in, Represents the mathematical expectation. Indicates the probability of an event occurring; Step S530: Use a multi-objective particle swarm optimization algorithm to solve the coal mine production optimization function, initialize the particle swarm, and the position of each particle represents a set of intervention schemes. During the iteration process, the particle updates its velocity and position based on its individual historical best position and global best position. The fitness of each particle is calculated by a fast counterfactual evaluator. The values ​​of the decision variables corresponding to the particle positions are substituted into the structure function corresponding to the variable influence structure equation to calculate the mathematical expectation of the structure function value. Based on the production optimization function, the mathematical expectation is directly used as the fitness value. After the iteration converges, the Pareto optimal solution set is output. The schemes corresponding to the Pareto optimal solution set are the optimization scheme set.

[0013] Preferably, step S500 further includes: step S540: performing a comprehensive evaluation on each scheme in the set of optimization schemes, calculating the safety benefit index, production benefit index and cost control index, and weighting the safety benefit index, production benefit index and cost control index to obtain a comprehensive benefit score; ; ; ; Step S550: Based on coal mine production experience, feasible solutions that meet the preset comprehensive benefit scoring threshold are coded to obtain configuration schemes, which include equipment configuration, personnel configuration and production environment configuration.

[0014] Preferably, step S600 specifically includes: extracting the scheme components from the configuration scheme, and adding the scheme components as new entities to the coal mine production knowledge graph; the scheme components include specific equipment, process parameters, and personnel teams.

[0015] The beneficial effects of this invention are as follows: This application integrates multi-source heterogeneous data through industrial knowledge graphs and uses causal discovery technology to extract the driving relationships between variables, constructing an interpretable causal graph for coal mine production. Based on this causal graph, counterfactual simulation and multi-objective optimization are performed, which can systematically evaluate the impact of different intervention measures on safety, output, and cost. This provides decision support for optimizing coal mine production operations, enabling managers to resolve the core contradiction between safety and efficiency in coal mine production. Under the premise of ensuring safety, it can find the optimal solution for production efficiency and cost control, which is conducive to promoting the refined control of risks in coal mine production operations. Attached Figure Description

[0016] Figure 1 is a schematic diagram of the basic process of a coal mine production business optimization method based on industrial knowledge graph provided by an embodiment of the present invention. Detailed Implementation

[0017] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0018] Referring to Figure 1, this embodiment of the present invention provides a method for optimizing coal mine production operations based on an industrial knowledge graph, comprising: Step S100: acquiring historical ledger data of coal mine production and constructing a coal mine production knowledge graph based on the historical ledger data; Step S200: extracting prior constraints from the coal mine production knowledge graph and performing causal structure learning using a causal constraint algorithm to obtain a coal mine production causal graph; Step S300: constructing a set of variable influence structural equations based on the coal mine production causal graph; Step S400: performing counterfactual simulation based on the set of variable influence structural equations to calculate key indicators under different intervention behaviors, and taking intervention behaviors that meet the threshold of key indicators as feasible solution domains; Step S500: performing multi-objective optimization on feasible solution domains to obtain an optimized solution set, calculating the comprehensive benefit score of the optimized solution set, and taking the solutions that meet the comprehensive benefit score threshold as configuration solutions; Step S600: adding the configuration solutions to the graph to obtain an updated coal mine production knowledge graph.

[0019] Step S100 specifically includes: Step S110: Obtain historical ledger data of coal mine production, and divide the historical ledger data into structured text and unstructured text; Step S120: Extract entities, relationships, and attributes from the structured text through rule mapping, and extract entities, relationships, and attributes from the unstructured text through semantic recognition technology. In this embodiment, the following unstructured text is used: "On the morning shift of March 1, 2024, the maintenance team conducted routine maintenance on the MC101 coal mining machine at the 101 working face. It was found that the vibration value of the cutting section bearing exceeded the standard, reaching 8.5 mm / s. The bearing was immediately replaced, and the vibration value dropped to 2.1 mm / s after replacement." From the data "Maintenance took 3.5 hours, affecting the shift's output by approximately 300 tons," the following triples were extracted: <Working Face 101, Daily Output, 8500 tons>, <Fault Vibration MC101, Occurred at, Coal Mining Machine MC101>. OpenEA was used to align the same entities with different representations, generating structured triples. In coal mine production-related data, there are numerous homonyms and synonyms, such as No. 1 Coal Mining Machine, Coal Mining Machine 1, and No. 1 Coal Mining Equipment, all representing the same entity. After entity alignment, they were unified as "No. 1 Coal Mining Machine," thus avoiding chaotic entity relationships in the subsequent knowledge graph.

[0020] The structured text includes production reports, equipment ledgers, resource consumption lists, environmental monitoring data, and spreadsheets; the unstructured text includes shift handover records, maintenance records, meeting minutes, and accident investigation reports. Entities include equipment, production, safety, personnel, and resources. Relationships include location, belonging, participation, cause, and adjacency. Equipment includes equipment number, equipment name, equipment operating parameters, and equipment maintenance work orders. Production includes mining location, working face, footage, daily output, operating rate, and cycle count. Safety includes gas concentration, ventilation system monitoring data, and roof pressure monitoring data. Personnel includes the number of people going down the mine, personnel positions, and personnel ages. Resource consumption includes support materials, accessory materials, water resources, and electricity resources. Step S130: Import the structured ternary sets into the graph database for storage to generate a coal mine production knowledge graph.

[0021] Step S100 integrates the "data silos" scattered across different systems and formats into an interconnected semantic network, providing a foundation for global analysis. Through entity alignment and relation extraction, the raw data is transformed into a set of triples consisting of "entity-relationship-attribute". This is more standardized than directly using the original data table, and can be more effectively input into subsequent causal constraint algorithms, helping the algorithms reduce the search space and improve the accuracy of causal graph learning.

[0022] In step S200, extracting prior constraints from the coal mine production knowledge graph specifically includes: Step S210: Based on the coal mine production knowledge graph, extract prior constraints, which include permissible edges, prohibited edges, and preferred edges; permissible edges represent causal relationships that directly or indirectly exist between entities in the coal mine production knowledge graph, prohibited edges represent causal relationships that cannot exist in the coal mine production knowledge graph, and preferred edges represent causal relationships with a clear temporal order; in this embodiment, the knowledge graph contains triples <gas concentration, leading to, explosion accident>, <insufficient ventilation, leading to, increased gas concentration>, <checking the ventilation system, prior to, starting the coal mining equipment>, and <wearing an isolation respirator, preventing, poisoning accident>, where there is a direct causal relationship between gas concentration and explosion accident, so "gas concentration → explosion accident" can be extracted as a permissible edge; based on common knowledge in the field of coal mine production, it can be deduced that there is an indirect causal relationship between insufficient ventilation and explosion accident, so "insufficient ventilation → explosion accident" can also be used as a permissible edge. Wearing an isolation respirator can prevent poisoning incidents; therefore, "wearing an isolation respirator → poisoning accident" is a prohibited edge. "Check ventilation" takes precedence over "start equipment." Thus, "check ventilation system → start coal mining equipment" is a preferred edge, forcibly set as a directed edge in the initial completely undirected graph, and cannot be deleted or reversed in subsequent conditional independence checks.

[0023] Step S220: By setting the logical hierarchy sequence of variables, the allowed edges are transformed into constraints that restrict the direction of causal search, allowing only the detection from the first variable to the subsequent variable; the prohibited connections are specified as the detection rules to be skipped, directly excluding unreasonable causal relationship searches; by explicitly specifying the directed edges that must exist, the preferred edges are transformed into mandatory assertions on the algorithm, locking in the core causal path.

[0024] Coal mine production involves hundreds of variables, and the search space for causal discovery without any constraints is exponential. In this embodiment, prior constraints reduce the search space by several orders of magnitude and incorporate expert experience and conventional knowledge in the field of coal mine production, thereby improving the reliability of the causal graph of coal mine production.

[0025] In step S200, the causal structure learning is performed using the causal constraint algorithm to obtain the causal graph of coal mine production. Specifically, step S230 involves organizing the historical ledger data into a structured numerical matrix X, with the expression: Where n represents the number of samples and p represents the number of variables; Step S240: Input the prior constraints after format conversion into the FCI causal discovery algorithm, perform conditional independence test on the structured numerical matrix, output a partial directed acyclic graph, orient the unoriented edges in the partial directed acyclic graph through V-structure rules and experience in the coal mine production field, and obtain a directed acyclic graph as the causal graph of coal mine production.

[0026] In this embodiment, three undirected edges connect four variables: ventilation volume-gas concentration, gas concentration-production shutdown, and equipment vibration-gas concentration, forming a chain structure. Ventilation volume and production shutdown are related under no-load conditions, but independent when a given gas concentration is present. Gas concentration is the intermediate transmission node, and the orientation is: ventilation volume → gas concentration → production shutdown. Based on experience in coal mine production, equipment vibration does not directly affect gas concentration, but gas concentration does not affect equipment vibration. Historical data shows that equipment with high vibration often has poor ventilation, indicating that equipment vibration is related to ventilation volume. Since "ventilation volume → gas concentration" is already oriented, the remaining edges are oriented as: equipment vibration → ventilation volume, resulting in the final orientation: equipment vibration → ventilation volume → gas concentration → production shutdown.

[0027] This step generates a causal graph of coal mine production, laying a solid structural foundation for subsequent structural equation modeling and counterfactual simulation. It makes it possible to quantitatively extrapolate "how B will change if A is changed," thereby transforming a static knowledge graph into a dynamic decision-making system and achieving a key leap from data association to causal intervention.

[0028] Step S300 specifically includes: Step S310: Taking the nodes in the cause-effect graph of coal mine production operations as endogenous variables, constructing a structural equation for each endogenous variable, and generating a variable influence structural equation. The expression of the variable influence structural equation is: ;in, Endogenous variables in a causal graph of coal mine production The set of parent nodes, Represents the set of parameters to be estimated. Indicates the error term. The structure function is represented; Step S320: Using the structural constraints of the structured numerical matrix and the cause-effect graph of coal mine production, the parameters in each equation are estimated using the maximum likelihood method; Step S330: Residual analysis is performed on each equation to ensure that the error of each equation is within the first allowable threshold; Step S340: All variable influence structure equations are integrated to obtain a set of variable influence structure equations.

[0029] In this embodiment, the arrow in the causal graph representing "ventilation volume → gas concentration" is transformed into the equation Gas Concentration = 0.8 - 0.05 × Ventilation Volume + ε, clearly stating that "for every 10 units increase in ventilation volume, the gas concentration decreases by 0.5 units." The parameters estimated using the maximum likelihood method provide the specific impact of each causal variable on the outcome variable. The integrated set of equations constitutes a complete coal mine production simulation, which can be used to predict the effects of interventions and provides a computational basis for subsequent counterfactual simulations. Transforming the causal graph into a computable set of equations, and converting static node-edge relationships into dynamic mathematical functions, makes causal relationships computable and quantifiable.

[0030] Step S400 specifically includes: Step S410: Selecting intervention variables and setting intervention values; Step S420: Calculating the counterfactual value of each endogenous variable according to the topological order of the coal mine production causal diagram; Step S430: Performing multiple repeated calculations using the Monte Carlo method to obtain the probability distribution of the endogenous variables, the expression of which is: ;in, Represents the outcome variable. Indicates mandatory intervention variable , will variables The value is set to , This represents the set of other influencing factors. Represents a probability distribution. This formula represents the summation over all possible influencing factors; it calculates the various possible structures and probabilities of Y under intervention after considering all possible combinations of influencing factors.

[0031] Step S440: Calculate the expected value of each outcome variable based on the probability distribution.

[0032] In this embodiment, the current ventilation rate of working face A in the mine is 1000 m³ / min. The intervention is to reduce the ventilation rate to 800 m³ / min. The calculated counterfactual values ​​are: methane concentration of 0.82%, daily output of 88 tons, and electricity cost of 0.8 yuan / ton. Based on the above data, the safety risks and production costs under the intervention can be predicted.

[0033] This embodiment transforms the effectiveness of intervention measures from qualitative judgment to probabilistic and quantitative assessment, providing a scientific basis for high-risk decision-making. By employing the Monte Carlo method and conducting multiple simulations, the influence of all unobserved factors Z is quantified, resulting in a complete probability distribution of the results rather than a single numerical value. This truly reflects the comprehensive impact of uncertainties such as geological conditions, equipment status, and personnel operation.

[0034] Step S400 further includes: Step S450: Based on coal mine production experience, select multiple intervention variables for multi-scenario simulation, repeat steps S410-S440, and obtain multi-scenario simulation results; Step S460: Integrate the multi-scenario simulation results into a multi-dimensional optimization matrix, where the rows of the multi-dimensional optimization matrix represent different intervention behaviors, and the columns represent the mathematical expectation of the intervention result variables, and the mathematical expectation is recorded as a key indicator; Step S470: Based on a pre-set threshold group, exclude scenarios that do not meet the threshold group to obtain the feasible solution domain, where the threshold group includes production cost threshold, daily output threshold, and safety risk threshold.

[0035] This step involves screening solutions based on experience and knowledge in coal mine production. In this embodiment, the gas concentration threshold is specified as 80%. Since the gas concentration cannot exceed 80%, changing the ventilation rate from 1000 m³ / min to 800 m³ / min would be considered an action that does not meet production conditions. Eliminating such intervention solutions that do not meet production conditions narrows the decision-making scope, allowing managers to focus on feasible and effective solutions and improving decision-making efficiency.

[0036] In this embodiment, by selecting multiple intervention variables for multi-scenario simulation, various possible combinations of interventions can be systematically explored, avoiding the omission of potential effective solutions. In coal mine production, multiple variables such as ventilation volume, maintenance frequency, and support density are adjustable. Multi-scenario simulation helps managers fully understand the effects of different strategies. By setting thresholds such as production cost threshold, daily output threshold, and safety risk threshold, it is ensured that the selected solutions meet both safety requirements and economic constraints, achieving a balance between safety and efficiency.

[0037] Step S500 specifically includes: Step S510: Taking the values ​​of the intervention variables in the feasible solution domain as decision variables, denoted as vector X, with the expression: ;in, Indicates the first There are several adjustable parameters, such as ventilation volume and maintenance cycle, where m represents the total number of decision variables; Step S520: With maximizing daily output, minimizing safety risks, and minimizing costs as production optimization objectives, construct a production optimization function. The expression of the production optimization function is: ; ; ; Where E represents the expected value and P represents the probability of the event occurring; Step S530: The multi-objective particle swarm optimization algorithm is used to solve the coal mine production optimization function. The particle swarm is initialized, and the position of each particle represents a set of intervention schemes. During the iteration process, the particle updates its velocity and position according to its individual historical best position and global best position. The fitness of each particle is calculated by a fast counterfactual evaluator. The values ​​of the decision variables corresponding to the particle positions are substituted into the structure function corresponding to the variable influence structure equation to calculate the expected value of the structure function value. Based on the production optimization function, the expected value is directly used as the fitness value. After the iteration converges, the Pareto optimal solution set is output. The schemes corresponding to the Pareto optimal solution set are the optimization scheme set.

[0038] In this embodiment, the calculated mathematical expectation represents the performance of each intervention scheme across multiple objectives: safety risk, daily output, and cost. Coal mine production typically requires trade-offs among multiple objectives, such as increasing output, reducing safety risk, and controlling costs. Steps S510-S530 construct a multi-objective optimization function that simultaneously considers these objectives and finds a set of equilibrium solutions, i.e., the Pareto optimal solution set, providing decision-makers with multiple choices.

[0039] Step S500 further includes: Step S540: Perform a comprehensive evaluation on each scheme in the optimization scheme set, calculate the safety benefit index, production benefit index and cost control index, and weight the safety benefit index, production benefit index and cost control index to obtain a comprehensive benefit score; ; ; ; In this embodiment, the weighting coefficients are set based on domain experience. Safety is the most important factor in coal mine production, and safety cannot be sacrificed for output. Therefore, the safety risk weighting coefficient is set to a relatively high proportion of 0.5, while the other two are set to lower proportions. The production efficiency index weight is 0.3, and the cost control weight is set to 0.2.

[0040] The weighted comprehensive score avoids one-sided optimization, transforming complex multi-objective optimization into quantifiable decisions, providing managers with a basis for decision-making while ensuring a safety baseline. It identifies solutions with higher comprehensive benefit scores as compromises, ensuring the final configuration achieves the optimal balance between safety, efficiency, and economy, thus bridging the gap between theoretical deduction and on-site implementation.

[0041] Step S550: Based on coal mine production experience, feasible solutions that meet the preset comprehensive benefit scoring threshold are coded to obtain configuration schemes, which include equipment configuration, personnel configuration and production environment configuration.

[0042] In this embodiment, the current ventilation volume of working face A in the mine is 1000 m³ / min. The intervention is to increase the ventilation volume to 1200 m³ / min. The calculated counterfactual values ​​are: gas concentration of 0.75%, daily output of 100 tons, and electricity cost of 0.9 yuan / ton. These counterfactual values ​​are all indicators that meet the production conditions. Furthermore, by evaluating the comprehensive benefits of gas concentration, daily output, and electricity cost, it is proven that the comprehensive benefits assessment reaches the preset threshold. Therefore, "increasing the ventilation volume to 1200 m³ / min" is selected as a configurable option.

[0043] This embodiment demonstrates how quantitative predictions derived from causal simulation can be transformed into configuration instructions that take into account multiple objectives and can directly guide on-site production, thus realizing a closed loop of data-driven decision-making.

[0044] Step S600 specifically includes: extracting the scheme components from the configuration scheme and adding the scheme components as new entities to the coal mine production knowledge graph; the scheme components include specific equipment, process parameters, and personnel teams.

[0045] This application integrates multi-source heterogeneous data through industrial knowledge graphs and uses causal discovery technology to extract the driving relationships between variables, constructing an interpretable causal graph for coal mine production. Based on this causal graph, counterfactual simulation and multi-objective optimization are performed, which can systematically evaluate the impact of different intervention measures on safety, output, and cost. This provides decision support for optimizing coal mine production operations, enabling managers to resolve the core contradiction between safety and efficiency in coal mine production. Under the premise of ensuring safety, it can find the optimal solution for production efficiency and cost control, which is conducive to promoting the refined control of risks in coal mine production operations.

[0046] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means that implement the functions specified in one or more flowcharts and / or one or more block diagrams.

[0047] 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 protection scope of the present invention.

Claims

1. A method for optimizing coal mine production operations based on industrial knowledge graphs, characterized in that, include: Step S100: Obtain historical ledger data of coal mine production, and construct a coal mine production knowledge graph based on the historical ledger data; Step S200: Extract prior constraints from the coal mine production knowledge graph, and use a causal constraint algorithm to learn the causal structure to obtain a coal mine production causal graph; Step S300: Construct a set of variable influence structural equations based on the coal mine production causal graph; Step S400: Perform counterfactual simulation based on the set of variable influence structural equations to calculate key indicators under different intervention behaviors, and take intervention behaviors that meet the key indicator thresholds as feasible solution domains; Step S500: Perform multi-objective optimization on the feasible solution domains to obtain an optimized solution set, calculate the comprehensive benefit score of the optimized solution set, and take the solutions that meet the comprehensive benefit score thresholds as configuration solutions; Step S600: Add the configuration solutions to the graph to obtain an updated coal mine production knowledge graph.

2. The coal mine production operation optimization method based on industrial knowledge graph as described in claim 1, characterized in that, Step S100 specifically includes: Step S110: Obtaining historical ledger data of coal mine production, and dividing the historical ledger data into structured text and unstructured text; Step S120: Extracting entities, relationships, and attributes from the structured text through rule mapping, extracting entities, relationships, and attributes from the unstructured text through semantic recognition technology, and aligning the same entities in different expression forms using OpenEA to generate structured triples; wherein, the structured text includes production reports, equipment ledgers, resource consumption lists, environmental monitoring data, and spreadsheets, and the unstructured text includes shift handover records, maintenance records, meeting minutes, and accident investigation reports; The entities include equipment, production, safety, personnel, and resources; the relationships include location, belonging, participation, causing, and adjacency; the equipment category includes equipment number, equipment name, equipment operating parameters, and equipment maintenance work order; the production category includes mining location, working face, footage, daily output, operating rate, and cycle number; the safety category includes gas concentration, ventilation system monitoring data, and roof pressure monitoring data; the personnel category includes the number of people going down the mine, personnel positions, and personnel ages; the resource consumption category includes support materials, accessory materials, water resources, and electricity resources; Step S130: Import the structured ternary into the graph database for storage to generate a coal mine production knowledge graph.

3. The coal mine production operation optimization method based on industrial knowledge graph as described in claim 1, characterized in that, In step S200, extracting prior constraints from the coal mine production knowledge graph specifically includes: Step S210: Based on the coal mine production knowledge graph, extract prior constraints, which include allowed edges, prohibited edges, and preferred edges; allowed edges represent causal relationships that directly or indirectly exist in the coal mine production knowledge graph, prohibited edges represent causal relationships that cannot exist in the coal mine production knowledge graph, and preferred edges represent causal relationships with a clear temporal order; Step S220: By setting the logical hierarchy sequence of variables, allowed edges are transformed into constraints that restrict the direction of causal search, allowing only the verification from earlier variables to later variables; variables that are prohibited from connection are designated as verification rules to be skipped, directly excluding unreasonable causal relationship searches; by explicitly specifying the directed edges that must exist, preferred edges are transformed into mandatory assertions on the algorithm, locking in the core causal path.

4. The coal mine production operation optimization method based on industrial knowledge graph as described in claim 1, characterized in that, In step S200, the causal structure learning is performed using a causal constraint algorithm to obtain the causal graph of coal mine production. Specifically, step S230 involves organizing the historical ledger data into a structured numerical matrix X, with the expression: ;in, Indicates the number of samples. The number of variables is represented; Step S240: Input the prior constraints after format conversion into the FCI causal discovery algorithm, perform conditional independence test on the structured numerical matrix, output a partial directed acyclic graph, and orient the unoriented edges in the partial directed acyclic graph through V-structure rules and experience in the coal mine production field to obtain a directed acyclic graph as the causal graph of coal mine production.

5. The coal mine production operation optimization method based on industrial knowledge graph as described in claim 4, characterized in that, Step S300 specifically includes: Step S310: Taking the nodes in the cause-effect graph of the coal mine production business as endogenous variables, constructing a structural equation for each endogenous variable, and generating a variable influence structural equation, the expression of which is: ;in, Endogenous variables in a causal graph of coal mine production The set of parent nodes, Represents the set of parameters to be estimated. Indicates the error term. The structure function is represented; Step S320: Using the structure constraints of the structured numerical matrix and the coal mine production cause-effect graph, the parameters in each equation are estimated using the maximum likelihood method; Step S330: Residual analysis is performed on each equation to ensure that the error of each equation is within the first allowable threshold; Step S340: All variable influence structure equations are integrated to obtain a variable influence structure equation set.

6. The coal mine production operation optimization method based on industrial knowledge graph as described in claim 1, characterized in that, Step S400 specifically includes: Step S410: Selecting intervention variables and setting intervention values; Step S420: Calculating the counterfactual value of each endogenous variable according to the topological order of the coal mine production causal diagram; Step S430: Performing repeated calculations using the Monte Carlo method to obtain the probability distribution of the endogenous variables, the expression of which is: ;in, Represents the outcome variable. Indicates mandatory intervention variable , will variables The value is set to ， This represents the set of other influencing factors. Represents a probability distribution. Step S440: Sum all possible influencing factors; Based on the probability distribution, calculate the expected value of each outcome variable.

7. The coal mine production operation optimization method based on industrial knowledge graph as described in claim 6, characterized in that, Step S400 further includes: Step S450: Based on coal mine production experience, select multiple intervention variables for multi-scenario simulation, repeat steps S410-S440, and obtain multi-scenario simulation results; Step S460: Integrate the multi-scenario simulation results into a multi-dimensional optimization matrix, where the rows of the multi-dimensional optimization matrix represent different intervention behaviors, and the columns represent the mathematical expectations of the intervention results variables, and the mathematical expectations are recorded as key indicators; Step S470: Based on a pre-set threshold group, exclude scenarios that do not meet the threshold group to obtain a feasible solution domain, where the threshold group includes a production cost threshold, a daily output threshold, and a safety risk threshold.

8. The coal mine production operation optimization method based on industrial knowledge graph as described in claim 1, characterized in that, Step S500 specifically includes: Step S510: Taking the values ​​of the intervention variables in the feasible solution domain as decision variables, denoted as vector X, with the expression: ;in, Indicates the first There are several adjustable parameters, where m represents the total number of decision variables; Step S520: With maximizing daily output, minimizing safety risks, and minimizing costs as production optimization objectives, construct a production optimization function, the expression of which is: ； ； ； ;in, Represents the mathematical expectation. Indicates the probability of an event occurring; Step S530: Use a multi-objective particle swarm optimization algorithm to solve the coal mine production optimization function, initialize the particle swarm, and the position of each particle represents a set of intervention schemes. During the iteration process, the particle updates its velocity and position based on its individual historical best position and global best position. The fitness of each particle is calculated by a fast counterfactual evaluator. The values ​​of the decision variables corresponding to the particle positions are substituted into the structure function corresponding to the variable influence structure equation to calculate the mathematical expectation of the structure function value. Based on the production optimization function, the mathematical expectation is directly used as the fitness value. After the iteration converges, the Pareto optimal solution set is output. The schemes corresponding to the Pareto optimal solution set are the optimization scheme set.

9. The coal mine production operation optimization method based on industrial knowledge graph as described in claim 1, characterized in that, The step S500 further includes: step S540: comprehensively evaluate each scheme in the set of optimization schemes, calculate the safety benefit index, production benefit index and cost control index, and weight the safety benefit index, production benefit index and cost control index to obtain a comprehensive benefit score; ； ； ； Step S550: Based on coal mine production experience, feasible solutions that meet the preset comprehensive benefit scoring threshold are coded to obtain configuration schemes, which include equipment configuration, personnel configuration and production environment configuration.

10. The coal mine production operation optimization method based on industrial knowledge graph as described in claim 1, characterized in that, Step S600 specifically includes: extracting the scheme components from the configuration scheme, and adding the scheme components as new entities to the coal mine production knowledge graph; the scheme components include specific equipment, process parameters, and personnel teams.

Images