Multi-mechanism collaborative management system for surface mine blasting

Abstract

The invention belongs to the technical field of mine operation management. The surface mine blasting multi-mechanism collaborative management system comprises a plan making module, an information acquisition module, a central control module and a network transmission module. According to the method, intelligent cognition and reasoning of the whole blasting process of the surface mine are achieved by constructing the semantic network, cross-link and multi-factor coupled composite safety risks which are difficult to find through a traditional method can be accurately recognized, and early warning information containing semantic interpretation is generated; therefore, safety management is improved from passive response to active early warning and intelligent intervention. And meanwhile, the central control module automatically sends a cooperative control instruction to a related organization terminal, so that information barriers among the mechanisms can be broken, full-chain efficient cooperation from plan making to operation execution is realized, and the overall safety management level and cooperation efficiency of the blasting operation are further improved.

Description

Technical Field

[0001] This invention relates to the field of mine operation management technology, specifically to a multi-agency collaborative management system for blasting in open-pit mines. Background Technology

[0002] Open-pit mine blasting operations are complex systems engineering projects involving multiple professional fields such as geology, surveying, design, drilling, blasting, and safety. This process is typically completed collaboratively by multiple organizations, including the production operations department, technical quality department, safety and environmental protection department, and equipment operation department. A typical process includes: the client providing the working face; the technical quality department measuring the working face and designing the borehole layout; the equipment operation department executing the drilling operation; the technical quality department inspecting the boreholes; the explosives preparation station providing blasting materials; and finally, the production operations department carrying out the blasting operation. Throughout this process, the safety and environmental protection department is responsible for safety supervision of the entire process.

[0003] Traditional management models rely on customized management software, telephones, walkie-talkies, and paper documents for information transmission and collaboration, making true collaborative management difficult. While some management systems can now achieve basic data recording and process advancement, they suffer from significant shortcomings in core collaborative efficiency and dynamic control of safety risks. These shortcomings primarily manifest in: a lack of deep integration and intelligent analysis of information from the entire process and multiple sources, hindering the proactive identification of complex safety risks arising from the coupling of risk factors across multiple stages. For example, they cannot establish the intrinsic correlation between "drilling depth deviation," "specific rock strata conditions," and "charge design," thus failing to provide early warnings of potential risks such as "foundation defects," "flying rocks," or "excessive blasting vibration" before blasting operations.

[0004] Therefore, how to achieve efficient collaboration and dynamic safety management of all participating institutions in order to improve mine production efficiency and protect personnel safety is a technical problem that urgently needs to be solved. Summary of the Invention

[0005] The purpose of this invention is to provide a multi-mechanism collaborative management system for blasting in open-pit mines, computer equipment, computer-readable storage media, and computer program products to solve the problems mentioned in the background art.

[0006] This invention provides a multi-mechanism collaborative management system for blasting in open-pit mines. The system includes a planning module, an information acquisition module, a central control module, and a network transmission module; wherein, The planning module is configured to: generate a blasting operation plan based on user input information; The information acquisition module is configured to: collect multi-source status data in real time from multiple organizational terminals in the blasting operation process; The central control module is configured to: construct and maintain a semantic network representing knowledge in the blasting operation domain, wherein the semantic network describes the intrinsic relationships between operation elements through entity nodes and semantic relation edges; The multi-source state data acquired by the information acquisition module is dynamically mapped to the corresponding nodes and edges of the semantic network, and its attribute state is updated. Based on the updated semantic network, the composite security risks coupled from multiple links are identified and evaluated in real time through graph traversal and logical reasoning. Early warning information containing semantic explanations is generated according to the composite security risks, and collaborative control instructions are automatically sent to the relevant organizational terminals. The network transmission module is configured to establish a communication connection between the central control module and the terminals of each organization to enable the transmission of data and instructions.

[0007] The present invention also provides a computer device for use in a system as described in any of the preceding claims, including a memory, a processor, and a computer program or instructions stored in the memory.

[0008] The present invention also provides a computer-readable storage medium for use in a system as described in any of the preceding claims, storing a computer program or instructions.

[0009] The present invention also provides a computer program product applicable to a system as described in any of the preceding claims, comprising a computer program or instructions.

[0010] This invention achieves intelligent cognition and reasoning of the entire open-pit mine blasting process by constructing a semantic network. It can accurately identify complex safety risks that are difficult to detect by traditional methods, involving multiple links and factors, and generate early warning information with semantic interpretation. This transforms safety management from passive response to proactive early warning and intelligent intervention. At the same time, by automatically issuing collaborative control commands to relevant organizational terminals through the central control module, information barriers between organizations can be broken down, achieving efficient collaboration across the entire chain from planning to execution. This improves the overall safety management level and collaborative efficiency of blasting operations. Attached Figure Description

[0011] Figure 1 This is a schematic diagram of the structure of an open-pit mine geological environment simulation and early warning system based on a dynamic evolution model, as disclosed in an embodiment of the present invention. Figure 2 This is another structural schematic diagram of an open-pit mine geological environment simulation and early warning system based on a dynamic evolution model disclosed in an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of an improved time series prediction model disclosed in an embodiment of the present invention. Detailed Implementation

[0012] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0013] Please see Figure 1 This invention provides a multi-mechanism collaborative management system 100 for open-pit mine blasting, the system comprising a planning module 101, an information acquisition module 102, a central control module 103, and a network transmission module 104; wherein, The planning module 101 is configured to: generate a blasting operation plan based on user input information; Specifically, the planning module 101 receives blasting task instructions and initial parameters input by users (such as mine production planners). Blasting task instructions include, but are not limited to, the location coordinates of the blasting working face, the expected blasting volume, and project schedule requirements; initial parameters include, but are not limited to, specific safety and environmental constraints, such as the need to control vibrations near important facilities.

[0014] This module has a built-in operation plan generation logic that can automatically generate a structured and executable blasting operation plan based on the above input information, including the blasting operation target, scope, preliminary timing and key nodes.

[0015] The information acquisition module 102 is configured to: collect multi-source status data in real time from multiple organizational terminals in the blasting operation process; Specifically, the information acquisition module 102 is responsible for establishing data connections with terminals deployed in multiple organizations such as the production operation department, technical quality department, safety and environmental protection department, equipment operation department, explosive preparation station and pyrotechnics warehouse, so as to collect multi-source status data from these terminals in real time at high frequency, which characterize the progress status of each link.

[0016] Multi-source status data includes, for example, "working face measurement results," "hole layout design drawings," and "hole inspection pass rate" from the Technical Quality Department terminal; "drilling rig operating parameters" and "drilling depth and angle" from the Equipment Operation Department terminal; "on-site safety inspection records" and "environmental monitoring data" from the Safety and Environmental Protection Department terminal; and "explosives requisition and inventory information" from the explosives preparation station terminal. After preliminary cleaning and formatting of the above raw data, high-quality, standardized data is transmitted to the central control module.

[0017] The central control module 103 is configured to: construct and maintain a semantic network representing knowledge in the blasting operation domain, wherein the semantic network describes the intrinsic relationships between operation elements through entity nodes and semantic relation edges; The multi-source state data acquired by the information acquisition module is dynamically mapped to the corresponding nodes and edges of the semantic network, and its attribute state is updated. Based on the updated semantic network, the composite security risks coupled from multiple links are identified and evaluated in real time through graph traversal and logical reasoning. Early warning information containing semantic explanations is generated according to the composite security risks, and collaborative control instructions are automatically sent to the relevant organizational terminals. Specifically, the central control module 103 is the intelligent hub of the system of the present invention, and its functions are implemented as follows: Within this module, a semantic network is constructed and continuously maintained to deeply characterize knowledge in the field of open-pit blasting operations. This semantic network abstracts physical entities (such as "drill hole B01", "explosive TNT", "safety officer Zhang San") and abstract concepts (such as "hole layout design D001", "safety procedure S_R01") into entity nodes, and precisely describes the intrinsic relationships between them through semantic relation edges (such as hasPart, followedBy, uses, violates, locatedIn, require) to form a machine-understandable blasting operation knowledge graph.

[0018] Meanwhile, this module continuously receives real-time data streams from the information acquisition module 102 and dynamically maps them to the corresponding nodes and edges in the semantic network, thereby updating their attribute states. For example, when borehole inspection data is received, this module finds the corresponding "drilling" node, updates its "hole depth" attribute from "design value" to "measured value", and updates its "status" attribute to "qualified" or "unqualified".

[0019] Next, based on the updated semantic network, the module launches a graph traversal algorithm and a logical reasoning engine. In this process, it actively explores cross-node and cross-stage related paths to identify and assess complex safety risks coupled from multiple risk factors in real time. For example, it infers that "Drilling B01 is not up to standard" + "Drilling B01 has Part blasting design D001" + "Blasting design D001 uses ANFO explosive" + "ANFO explosive requires a specific hole density" -> "Current hole integrity is at risk" -> "May cause foundation or flyrock risks".

[0020] Upon identifying a complex safety risk, a warning message containing semantic explanations is generated. This warning message clearly elucidates the root cause of the risk, its transmission path, and potential consequences. For example, "Warning: Due to the excessive depth deviation of borehole No. 5, the integrity of the charge structure in its blasting design has been compromised, violating the emulsion explosive usage specifications and posing a risk of flyrock." Simultaneously, the module automatically sends collaborative control commands to relevant organizational terminals, such as sending a "Stop Blasting Operations" command to the production operations department terminal and pushing the "Please Review Blasting Design D001" task to the technical quality department terminal.

[0021] The network transmission module 104 is configured to establish a communication connection between the central control module and the terminals of each organization to realize the transmission of data and instructions.

[0022] Specifically, the network transmission module 104 can employ reliable communication protocols (such as TCP / IP, MQTT, etc.) to establish a stable and secure communication connection between the central control module 103 and all distributed organizational terminals. This module is responsible for the encoding, encapsulation, and transmission of all uplink data (from terminal to center) and downlink instructions (from center to terminal), ensuring low-latency and high-reliability information transmission in complex mining industrial environments, and providing basic communication support for real-time collaboration among multiple organizations.

[0023] This invention achieves intelligent cognition and reasoning of the entire open-pit mine blasting process by constructing a semantic network. It can accurately identify complex safety risks that are difficult to detect by traditional methods, involving multiple links and factors, and generate early warning information with semantic interpretation. This transforms safety management from passive response to proactive early warning and intelligent intervention. At the same time, by automatically issuing collaborative control commands to relevant organizational terminals through the central control module, information barriers between organizations can be broken down, achieving efficient collaboration across the entire chain from planning to execution. This improves the overall safety management level and collaborative efficiency of blasting operations.

[0024] As an example, please refer to Figure 2 The organizational terminals include the production operations department, technical quality department, safety and environmental protection department, equipment operation department, explosives preparation station, and pyrotechnics warehouse terminals.

[0025] In this embodiment, the terminals of at least the five core departments mentioned above that are connected and coordinated by the present invention are specifically configured as follows: Production Operations Terminal: This terminal is the final execution unit of the blasting operation plan. It receives blasting construction instructions and collaborative control instructions (such as "pause operation" or "start operation") from the central control module 103, and is responsible for providing feedback on the on-site operation progress and status.

[0026] Technical Quality Department Terminal: This terminal is the source of technical solution design and quality control. It is responsible for uploading key data related to blasting design and quality acceptance, including but not limited to "working face measurement results", "hole layout design drawings", and "hole inspection results", and receiving collaborative instructions from the central control module 103 for design review, scheme optimization and other tasks.

[0027] Safety and Environmental Protection Department Terminal: This terminal is the core of the entire process safety supervision. It is used to input supervision information such as "on-site safety inspection records" and "environmental monitoring data," and serves as the highest priority early warning information receiving terminal. When the central control module 103 identifies a major safety risk, this terminal will receive an immediate alarm so that safety management personnel can take emergency intervention measures.

[0028] Equipment Operations Terminal: This terminal is closely linked to the on-site drilling operation. It is responsible for uploading raw data reflecting the quality of the drilling operation, such as "drilling rig working parameters" and "drilling depth and angle," in real time, providing the semantic network with real-time input of the drilling process.

[0029] Explosives preparation station and pyrotechnics warehouse terminal: This terminal is the management and supply node for blasting materials. It is responsible for providing "pyrotechnics requisition and inventory information" to ensure that the flow information of dangerous materials such as explosives and detonators is included in the system monitoring in real time and matched and verified with blasting design requirements to prevent misuse of materials or management loopholes.

[0030] As an example, the central control module 103 is configured to maintain the semantic network using the following strategy: Based on a multi-dimensional confidence evaluation system that includes data source reliability, time decay factor and expert feedback, a dynamically updated comprehensive confidence score is calculated for entity nodes and relation edges in the semantic network. When a semantic conflict is detected or the comprehensive confidence score is lower than the threshold, graph-based similarity analysis and inference path probability assessment are used to identify and resolve logical contradictions in the semantic network. Furthermore, by continuously analyzing historical assignment data through a sliding time window incremental learning algorithm, and dynamically adjusting the weight coefficients of relation edges and node attribute values ​​in the semantic network according to the gradual changing trend of data features, the knowledge base is autonomously updated.

[0031] In this implementation, to ensure the reliability and consistency of knowledge within the semantic network, the present invention dynamically calculates the comprehensive confidence level for each entity node (e.g., "drill hole B01") and each semantic relation edge (e.g., "drill hole B01 hasPart blasting design D001") in the semantic network. This comprehensive confidence level is derived based on a multi-dimensional confidence evaluation system, which mainly includes: Data source reliability: Data from high-precision measuring instruments or authoritative technical specifications receives a higher reliability score. The calculation formula is, for example: in, , where represents the weighting coefficients for each dimension; The baseline reliability is based on the inherent type of the data source and is preset according to the baseline reliability. For example, the baseline value of authoritative data sources (such as calibrated measuring instruments or official technical specifications) is higher. Historical error count: the number of confirmed error records in the historical data of this data source; This represents the total amount of historical data, specifically the total number of data entries reported by this data source within the statistical period. An environmental stability score is assigned to assess the impact of the data collection environment. The score is calculated based on environmental sensor data (such as vibration, temperature, and humidity). The more stable the environment, the higher the score.

[0032] Time decay factor: Newly generated data has a higher weighting for timeliness than older historical data, reflecting the most recent changes in the field conditions. Its calculation formula is, for example: in, This is the time decay amount, the difference between the current time and the time when the data was generated or the time when it was last validly verified; This is the decay rate constant, which is a constant greater than 0. It controls how quickly the information value decays, and different values ​​can be set for data with different timeliness requirements.

[0033] Expert feedback: When security personnel or technicians confirm or reject system alerts, this feedback directly affects the confidence level of the relevant knowledge.

[0034] When new field data is input into the system, causing logical contradictions in the semantic network (e.g., the same operational step is inferred as both "safe" and "violation"), the system initiates a resolution process. Specifically, through graph-based similarity analysis (comparing the closeness of the connections between contradictory parties in the knowledge graph) and inference path probability assessment (analyzing the reliability of the inference chain leading to the contradictory conclusion), the system automatically identifies the most likely source of error, resolves the contradictions, and maintains the consistency of the network logic. When the overall confidence level of any knowledge falls below a preset threshold, it is automatically marked as unreliable and temporarily excluded from the inference process.

[0035] Furthermore, to enable the semantic network to learn and evolve adaptively, this invention does not process all historical data at once, but instead employs a sliding time window to continuously monitor job data within a recent period. This approach is computationally efficient and can keenly capture the latest technological developments and changes in job patterns.

[0036] Specifically, by analyzing data within a time window, gradual trends in certain data characteristics can be identified. For example, it might be discovered that the drilling efficiency of a new type of drilling rig is consistently and stably superior to that of an older model, or that the blastability assessment parameters of a certain rock formation change slowly with the seasons. Based on the identified trends, the weight coefficients of the relationship edges and the node attribute values ​​in the semantic network are dynamically adjusted. For example, the weight of the relationship between the new drilling rig and high efficiency can be gradually increased, or the blastability attribute values ​​of a specific rock formation can be updated. It is understood that the above process is continuous, smooth, and automatic, thereby achieving autonomous updates to the knowledge base and ensuring that the semantic network always reflects the most effective operational practices and the latest field knowledge.

[0037] This implementation method effectively ensures the consistency and reliability of semantic network knowledge by introducing a multi-dimensional confidence evaluation and conflict resolution mechanism, thereby improving the accuracy of risk identification and decision support. At the same time, incremental learning based on sliding time windows enables the system to autonomously adapt to the gradual evolution of the operating environment and technology, realize the continuous optimization and dynamic evolution of the knowledge base, and significantly enhance the long-term applicability and intelligence level of the system.

[0038] As an example, based on the updated semantic network, complex security risks resulting from the coupling of risks at multiple stages are identified and assessed in real time through graph traversal and logical reasoning, including: By performing graph traversal based on temporal and dependency edges in the semantic network, the complete transmission link from the initial abnormal node to the potentially affected node is identified, thus forming a risk propagation path. When multiple risk propagation paths intersect in the semantic network, a composite security risk is identified, and a risk coupling factor between the paths is calculated. The risk coupling factor is determined based on the product of the path distance between the intersecting nodes, the weight of the relation edge, the severity of the anomaly, and whether a risk loop is formed. Based on the topology of each risk propagation path and the strength of the risk coupling factors, a multi-dimensional risk assessment matrix is ​​generated, which includes risk level, scope of impact, and priority of response.

[0039] In this implementation, when the state of any entity node (such as a borehole or equipment) in the semantic network is marked as abnormal, the graph is automatically traversed starting from the initial abnormal node, along the temporal relationship edge (followedBy) representing the order of the operation process and the dependency relationship edge (requires) representing the conditional dependency.

[0040] Understandably, the above process aims to identify all downstream or related nodes that may be directly or indirectly affected by the initial anomalous node, thus outlining one or more complete risk transmission chains. For example, discovering the anomaly of "insufficient drilling depth" will affect subsequent "charge operations" through the followedBy relationship, and ultimately jeopardize "blasting effectiveness and safety" through the requires relationship. The collection of these chains constitutes the risk propagation path, used to reveal the potential scope of risk spread.

[0041] Similarly, all discovered risk propagation paths are continuously monitored. When two or more risk propagation paths intersect in the semantic network (i.e., share one or more common nodes), it is determined that these originally independent risks have interacted and formed a composite security risk.

[0042] To quantify the intensity of this interaction, the risk coupling factor is further calculated based on at least path distance, relation edge weights, anomaly severity, and the formation of risk loops. The specific meanings of each factor are as follows: path distance: the position of the converging node in each propagation path; the closer the distance, the more direct the impact and the stronger the coupling; relation edge weights: the confidence and strength of semantic relationships on the path; the higher the weight, the greater the coupling contribution; anomaly severity: the severity level of the anomaly at the source of each path; risk loop formation: if these risk propagation paths can form a closed loop, it means that the risk is self-reinforcing and amplified cyclically within this loop. In this case, a coefficient greater than 1 will significantly increase the value of the coupling factor.

[0043] Risk coupling factors The calculation formula is, for example: in, The number of intersecting risk transmission paths, The average weight of the key semantic relation edges on path i reflects the logical confidence of the path. The higher the weight, the greater the possibility of risk transmission. The severity level of the source anomaly of path i is represented by the severity level. The higher the severity level, the greater the contribution to the composite risk. The number of hops from the abnormal node to the intersection node is normalized. The shorter the distance, the more direct the risk transmission and the greater the impact. This is a preset gain coefficient used to quantify the additional amplification effect brought about by the risk closure loop; This is a Boolean indicator function, which is 1 when a closed loop of risk propagation path is detected, and 0 otherwise.

[0044] Using the topology of each risk propagation path (such as path depth, breadth, and loop conditions) and the strength of the calculated risk coupling factors as core inputs, a multi-dimensional risk assessment matrix is ​​generated through a built-in evaluation algorithm. This matrix includes: risk level, the overall level of risk after comprehensive quantification; scope of impact, clearly identifying the affected operational processes, equipment, and areas; and response priority, providing the order of action based on the speed of risk evolution and potential consequences.

[0045] This implementation method achieves accurate quantitative assessment of complex security risks by constructing risk propagation paths and calculating risk coupling factors. It can promptly identify cross-stage and multi-factor coupled risks that are difficult to identify using traditional methods. At the same time, by introducing a risk closed-loop amplification mechanism, it enhances the system's early warning sensitivity to self-reinforcing risks.

[0046] As an example, graph traversal based on temporal and dependency edges in a semantic network identifies the complete propagation path from the initial anomalous node to potentially influential nodes, forming a risk propagation path, including: When traversing the graph along semantic relation edges, different types of semantic relation edges are assigned propagation probability weights based on historical data statistics, while dynamic resistance factors based on real-time status are set for key security control nodes in the semantic network. The path confidence of each potential risk propagation path is calculated by combining the propagation probability weight and the dynamic resistance factor. Only risk propagation paths with a path confidence higher than a preset threshold are retained for subsequent composite security risk assessment.

[0047] In this implementation, traditional graph traversal-based risk propagation analysis relies solely on network topology, which has two significant drawbacks: First, it cannot distinguish the differences in risk propagation capabilities of different types of semantic relationships in actual operations, potentially misjudging theoretically connected but practically low-probability paths as major threats; second, it ignores the dynamic blocking effect of security control nodes in the system, failing to reflect the true situation of risk propagation under real-time operating conditions. Therefore, this invention introduces a propagation probability based on historical statistics and a dynamic resistance mechanism based on real-time status, replacing theoretical deduction with engineering empirical verification for risk propagation path identification, thereby significantly improving the accuracy of path discovery.

[0048] Specifically, by analyzing accumulated historical operational data, a unique propagation probability weight is established for each type of semantic relationship edge. For example, data mining based on the past 200 blasting operations revealed that "hole layout deviation" affects "charge calculation" with a probability as high as 92% through the `requires` relationship, while affecting "nearby equipment safety" with a probability of only 15% through the `locatedIn` relationship. Therefore, the propagation probability weight for the `requires` dependency relationship is set to 0.9, and the propagation probability weight for the `locatedIn` positional relationship is set to 0.3. These weight values ​​are derived by analyzing the statistical impact paths of similar abnormal nodes in historical operational data, reflecting the risk transmission capability of different types of relationships in actual operations.

[0049] Simultaneously, a state-resistance mapping relationship is established for key nodes in the semantic network that have risk control functions. For example, the resistance factor of the "safety acceptance" node is 0.9 when it is in the "verified" state, 0.5 when it is in the "in progress" state, and 0.1 when it is in the "not started" state; the resistance factor of the "dust removal equipment" node is 0.8 when it is operating normally, and drops to 0.2 when it is in a fault state. It can be understood that these factors can be updated in real time based on IoT monitoring data, accurately reflecting the current risk resistance capability of the system.

[0050] Finally, the confidence propagation algorithm is used to calculate the propagation probability weights sequentially along each potential path. The product of the two and the dynamic resistance factor The cumulative decay, calculated using a formula such as: Path confidence The calculation formula is as follows: in: This represents the total number of semantic relation edges along this path. This represents the total number of critical security control nodes along this path.

[0051] Taking three potential paths A, B, and C as examples, their calculated path confidence scores are 0.85, 0.45, and 0.72, respectively. When the preset threshold is set to 0.65, path B (0.45 < 0.65) will be automatically filtered out, and only paths A and C will be retained as valid risk propagation paths for subsequent analysis.

[0052] This implementation method constructs a quantitative risk propagation model by introducing a propagation probability weight based on historical statistics and a dynamic resistance factor based on real-time status, upgrading path identification from static topology analysis to dynamic empirical evaluation. Furthermore, through path confidence calculation and threshold screening mechanisms, low-probability propagation paths are effectively filtered out, enabling risk assessment to focus on high-credibility risk transmission links, thereby improving the accuracy and practicality of risk warning.

[0053] As an example, setting a dynamic resistance factor based on real-time state for key security control nodes in a semantic network includes: The probability of security control failure of key security control nodes in the semantic network in the current time period is predicted based on the time-series prediction model. The prediction model is pre-constructed based on historical state data, and its prediction comprehensively considers the running time of the equipment, maintenance cycle, environmental stress and historical failure modes. The dynamic resistance factor of key safety control nodes is dynamically adjusted based on the probability of safety control failure.

[0054] The aforementioned implementation methods set the dynamic resistance factor for critical safety control nodes based on their current operating status, failing to fully consider the potential inaccuracy risks caused by factors such as equipment aging, maintenance delays, or accumulated environmental stress. This static assessment method cannot accurately reflect the potential decline in control effectiveness of critical safety control nodes during actual operation, leading to biases in the prediction of risk propagation paths. Therefore, this invention introduces a time-series prediction-based inaccuracy probability assessment mechanism to achieve proactive adjustment of the dynamic resistance factor.

[0055] Specifically, the time-series prediction model employs an encoder-decoder architecture based on an attention mechanism. The encoder consists of multiple layers of gated recurrent units, used to extract temporal features from the historical state data of nodes; the decoder focuses on the influencing factors of key time steps through an attention mechanism, outputting the probability value of inaccuracy within the future time window. The input features of the time-series prediction model include, but are not limited to: the ratio of equipment runtime to design life, the time interval from the last maintenance operation to the present, the cumulative duration of environmental stress parameters exceeding the standard, and historical inaccuracy records and failure modes of similar equipment.

[0056] As an improvement, please refer to Figure 3 The temporal prediction model of this invention adopts an encoder-dynamic context fusion layer-decoder architecture based on an attention mechanism. That is, a dynamic context fusion layer is added between the encoder and the decoder as a connecting layer. This connecting layer dynamically adjusts the context information passed to the decoder by identifying and analyzing the temporal pattern features of the input sequence in real time, so that the model can adaptively focus on the most critical historical information.

[0057] 1. Encoder: Composed of bidirectional gated loop units, its task is to read and understand the input historical state data sequence. The process steps the sequence by time step, combining the input of each time step (such as the ratio of equipment runtime to environmental stress exceedance duration) with the hidden state of the previous time step, updating the current hidden state, and finally outputting a hidden state sequence containing information from all time steps. Each of them Each contains sequence information up to the i-th time step.

[0058] 2. Dynamic Context Fusion Layer: This layer includes a temporal pattern feature extraction module, a pattern-aware attention module, and a context vector synthesis module. The temporal pattern feature extraction module is responsible for real-time computation of temporal pattern features from the hidden state sequence output by the encoder to guide the attention mechanism. The pattern-aware attention module uses the pattern feature vectors to dynamically adjust the attention weight allocation strategy. The context vector synthesis module performs the final information aggregation based on the attention weights. The workflow of this layer is as follows: a) Temporal pattern recognition: Analyzing the hidden state sequence output by the encoder. Real-time computation of key time-series pattern features, including: Trend stability By calculating the slope or variance of the linear regression sequence, it can be determined whether the equipment condition is stable, slowly deteriorating, or rapidly deteriorating.

[0059] Periodic intensity The analysis focuses on whether there are periodic fluctuations in the sequence related to maintenance cycles and production shifts, and the extent of these fluctuations.

[0060] mutation point density The number of times a statistical sequence undergoes drastic changes within a recent window reflects the frequency of shocks experienced by the equipment.

[0061] b) Dynamic attention weight calculation: Traditional attention mechanisms rely solely on the previous hidden state of the decoder. With each hidden state of the encoder The degree of correlation. This invention dynamically adjusts attention calculation by introducing temporal pattern features. The specific calculation process is as follows: Attention energy value calculation: ,in, For the decoder at time step The hidden state, For the encoder at time step The hidden state, This represents the feature vector of the time series pattern. , This is the weight matrix. For the weight vector, This is a bias term.

[0062] Attention weight calculation: The energy values ​​are converted into a normalized attention weight distribution using the softmax function to ensure... The specific impact of time series pattern characteristics is as follows: when trend stability At lower levels, , The item increases the attention energy value of the recent time step; when the periodic intensity When the density is high, increase attention to historical time steps; when the density of mutation points is high... When the value is large, increase the attention weight for time steps with abnormal fluctuations.

[0063] c) Context vector synthesis: utilizing calculated dynamic attention weights For the encoder hidden state sequence Perform a weighted summation to generate a pattern-aware context vector. .

[0064] ,in: In the decoder time step For encoder time steps Attention weights; and These are learnable weight matrices and bias vectors used to train the encoder's hidden state before synthesis. A linear transformation is performed to align the hidden state spaces of the encoder and decoder and extract more relevant features; It is the decoder at the time step The received, dynamically generated context vector is no longer a simple compression of all historical information, but rather a vector based on the temporal pattern of the current sequence. and the current state of the decoder The most critical information, after dynamic filtering and focusing, represents the most important information summary in the historical sequence for the current prediction step.

[0065] Based on the predicted probability of inaccuracy The dynamic resistance factor is determined using the following piecewise function. :when When the node is deemed to have good reliability, it is set... ;when When a node is deemed to be at risk of inaccuracy, a setting is made. ;when At that time, the risk of inaccurate judgment nodes is relatively high, so setting .

[0066] Understandably, it can also record the trend of resistance factor changes at each node, and when a continuous decrease in resistance factor is detected, it can automatically send preventive maintenance suggestions to the equipment operation department terminal.

[0067] This implementation method predicts the probability of failure of key safety control nodes, thereby enabling dynamic adjustment of the dynamic resistance factor nodes of key safety control nodes. This allows the system to proactively reflect the decay trend of node control effectiveness. Therefore, this implementation method can effectively solve potential risks such as equipment aging and maintenance delays that traditional static assessment methods cannot foresee, and thus significantly improve the accuracy of risk propagation path prediction and the timeliness of early warning.

[0068] As an example, generating warning information containing semantic interpretation based on the composite security risk includes: Extract entity attributes, relationship paths, and risk transmission logic related to complex security risks from the semantic network, including risk source node attributes, key node sequences on the propagation path, risk coupling mechanisms, and operational processes that may be affected. Based on the risk assessment matrix and semantic network topology, early warning information containing multiple elements is automatically generated and pushed to the corresponding terminals.

[0069] In this implementation, the risk source node (such as a borehole marked "unqualified") and its associated key nodes (such as the relevant blasting design and the type of explosives used) are located, and their core attributes are extracted. For example, the "designed hole depth", "measured hole depth", and "deviation value" of the borehole are extracted; the "design number" and "total charge" of the blasting design are extracted; and the "type" and "sensitivity" of the explosives are extracted.

[0070] By tracing back the risk propagation path discovered during the risk assessment phase, the key node sequences constituting this risk propagation path and the semantic relationship edges connecting them are extracted. For example, the extracted path sequence is: [Drill Hole B01] - (hasPart) -> [Blasting Design D001] - (uses) -> [Explosive ANFO] - (requires) -> [Specific Hole Density]. Simultaneously, the risk coupling mechanism is analyzed and extracted, for example: "Due to the depth deviation of Drill Hole B01, the integrity of the charge structure of its associated blasting design D001 is compromised, thereby violating the specific hole density condition required by the explosive ANFO."

[0071] Based on the topology of the semantic network, the affected upstream and downstream operations are analyzed. For example, it is determined that the risk not only affects the current charging operation, but may also ultimately affect the blasting effect (the emergence of a foundation) and the efficiency of post-blast loading.

[0072] After information extraction, the extracted semantic information is filled into a pre-defined, structured warning template, automatically generating a complete and easily understandable natural language description of the warning message. This warning message includes at least: Risk characterization: clearly define the risk type (e.g., "charge structure integrity risk") and level (e.g., "high risk"); Root cause and path analysis: clearly explain the source of the risk ("15% depth deviation in borehole No. 5") and the propagation chain; Impact and Consequences: Describe the possible direct consequences ("may cause flying rock accidents") and indirect impacts ("affects blasting effectiveness and increases secondary processing costs"). Recommended action: Provide specific and prioritized action measures ("Immediately suspend the loading operation and request the Technical Quality Department to review the blasting design D001").

[0073] Finally, based on the final risk level determined by the risk assessment matrix, a tiered push strategy is implemented to ensure that information reaches the most relevant personnel, avoiding alert fatigue or information omissions. A specific push mechanism is as follows: Level 1 Warning (High Risk): Pushed to all relevant terminals (production, technology, security, equipment, etc.) and requires mandatory confirmation, initiating the highest level of response.

[0074] Level 2 warning (medium risk): The warning is mainly pushed to the terminals of management departments and technical leaders, who then make decisions and distribute tasks.

[0075] Level 3 Early Warning (Low Risk): Pushed to the terminals of directly relevant work teams for on-site reminders and process control.

[0076] This invention also provides a computer device for use in a system as described in any of the preceding claims, including a memory, a processor, and a computer program or instructions stored in the memory.

[0077] This invention also provides a computer-readable storage medium for use in a system as described in any of the preceding claims, storing a computer program or instructions.

[0078] This invention also provides a computer program product applied to a system as described in any of the preceding claims, comprising a computer program or instructions.

[0079] The above description is only a part of the embodiments of this application and does not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A multi-mechanism collaborative management system for blasting in open-pit mines, characterized in that, The system includes a planning module, an information collection module, a central control module, and a network transmission module; wherein, The planning module is configured to: generate a blasting operation plan based on user input information; The information acquisition module is configured to: collect multi-source status data in real time from multiple organizational terminals in the blasting operation process; The central control module is configured to: construct and maintain a semantic network representing knowledge in the blasting operation domain, wherein the semantic network describes the intrinsic relationships between operation elements through entity nodes and semantic relation edges; The multi-source state data acquired by the information acquisition module is dynamically mapped to the corresponding nodes and edges of the semantic network, and its attribute state is updated. Based on the updated semantic network, the composite security risks coupled from multiple links are identified and evaluated in real time through graph traversal and logical reasoning. Early warning information containing semantic interpretation is generated according to the composite security risks, and collaborative control instructions are automatically sent to the relevant organizational terminals. The network transmission module is configured to establish a communication connection between the central control module and the terminals of each organization to enable the transmission of data and instructions.

2. The multi-mechanism collaborative management system for blasting in open-pit mines according to claim 1, characterized in that: The organizational terminals include the production operations department, technical quality department, safety and environmental protection department, equipment operation department, explosives preparation station, and pyrotechnics warehouse.

3. The multi-mechanism collaborative management system for blasting in open-pit mines according to claim 1, characterized in that: The central control module is configured to maintain the semantic network using the following strategy: Based on a multi-dimensional confidence evaluation system that includes data source reliability, time decay factor and expert feedback, a dynamically updated comprehensive confidence score is calculated for entity nodes and relation edges in the semantic network. When a semantic conflict is detected or the comprehensive confidence score is lower than the threshold, graph-based similarity analysis and inference path probability assessment are used to identify and resolve logical contradictions in the semantic network. Furthermore, by continuously analyzing historical assignment data through a sliding time window incremental learning algorithm, and dynamically adjusting the weight coefficients of relation edges and node attribute values ​​in the semantic network according to the gradual changing trend of data features, the knowledge base is autonomously updated.

4. The multi-mechanism collaborative management system for blasting in open-pit mines according to claim 3, characterized in that: Based on the updated semantic network, complex security risks resulting from the coupling of risks at multiple stages are identified and assessed in real time through graph traversal and logical reasoning, including: By performing graph traversal based on temporal and dependency edges in the semantic network, the complete transmission link from the initial abnormal node to the potentially affected node is identified, thus forming a risk propagation path. When multiple risk propagation paths intersect in the semantic network, a composite security risk is identified, and a risk coupling factor between the paths is calculated. The risk coupling factor is determined based on the product of the path distance between the intersecting nodes, the weight of the relation edge, the severity of the anomaly, and whether a risk loop is formed. Based on the topology of each risk propagation path and the strength of the risk coupling factors, a multi-dimensional risk assessment matrix is ​​generated, which includes risk level, scope of impact, and priority of response.

5. The multi-mechanism collaborative management system for blasting in open-pit mines according to claim 4, characterized in that: Graph traversal is performed based on temporal and dependency edges in the semantic network to identify the complete propagation link from the initial anomalous node to potentially influential nodes, forming a risk propagation path, including: When traversing the graph along semantic relation edges, different types of semantic relation edges are assigned propagation probability weights based on historical data statistics, while dynamic resistance factors based on real-time status are set for key security control nodes in the semantic network. The path confidence of each potential risk propagation path is calculated by combining the propagation probability weight and the dynamic resistance factor. Only risk propagation paths with a path confidence higher than a preset threshold are retained for subsequent composite security risk assessment.

6. The multi-mechanism collaborative management system for blasting in open-pit mines according to claim 5, characterized in that: To set dynamic resistance factors based on real-time state for key security control nodes in the semantic network, including: The probability of security control failure of key security control nodes in the semantic network in the current time period is predicted based on the time-series prediction model. The prediction model is pre-constructed based on historical state data, and its prediction comprehensively considers the running time of the equipment, maintenance cycle, environmental stress and historical failure modes. The dynamic resistance factor of key safety control nodes is dynamically adjusted based on the probability of safety control failure.

7. The multi-mechanism collaborative management system for blasting in open-pit mines according to claim 1, characterized in that: The step of generating warning information containing semantic interpretation based on the composite security risks includes: Extract entity attributes, relationship paths, and risk transmission logic related to complex security risks from the semantic network, including risk source node attributes, key node sequences on the propagation path, risk coupling mechanisms, and operational processes that may be affected. Based on the risk assessment matrix and semantic network topology, early warning information containing multiple elements is automatically generated and pushed to the corresponding terminals.

8. A computer device, applied to the system according to any one of claims 1-7, characterized in that, This includes a memory, a processor, and computer programs or instructions stored in the memory.

9. A computer-readable storage medium, applied to the system according to any one of claims 1-7, characterized in that, It stores computer programs or instructions.

10. A computer program product, applied to the system according to any one of claims 1-7, characterized in that, This includes computer programs or instructions.

Images