Coal mine safety production analysis method and system fused with multi-modal data

Abstract

The invention relates to a coal mine safety production analysis method and system fusing multi-modal data, and relates to the technical field of mining management, and the method comprises the steps: calibrating an equipment aging label based on an equipment state ledger; classifying the personnel behavior data of which the service duration is smaller than or equal to a threshold value, and determining first-class, second-class and third-class violation behaviors; randomly configuring coal mine production capacity; traversing first-class, second-class and third-class violation behaviors, constructing index constraints in combination with production capacity and equipment aging labels, retrieving a corresponding coal mine production scheduling sample set, and calculating first-class, second-class and third-class coal mine production accident triggering probabilities; and when the triggering probabilities of the first-class, second-class and third-class coal mine production accidents are all smaller than or equal to a threshold value, carrying out production scheduling management according to the coal mine production capacity. The problems that traditional coal mine production scheduling management only makes a decision from a single dimension, and cannot give consideration to multi-dimensional collaborative requirements of production efficiency and operation safety, so that the production scheduling management lacks accurate matching, is insufficient in fineness, and is difficult to realize comprehensive management are solved.

Description

Technical Field

[0001] The present application relates to the field of mining management technology, and in particular to a coal mine safety production analysis method and system integrating multimodal data. Background Art

[0002] With the advancement of coal mine production technology, optimizing coal mine production scheduling management has become a key foundation for ensuring production safety and improving efficiency. Existing coal mine production scheduling management methods rely solely on a single dimension for decision-making. This approach either ignores equipment status, leading to increased safety risks, or neglects personnel behavior, leading to insufficient production coordination. This approach struggles to balance the multi-dimensional demands of production efficiency and operational safety, and fails to meet the practical requirements for more sophisticated coal mine production management. Summary of the Invention

[0003] In order to solve the above technical problems, this application provides a coal mine production safety analysis method and system that integrates multimodal data, improves the satisfaction of multi-dimensional needs and the precision of coal mine production scheduling management, changes the traditional single-dimensional model, achieves a balance between operational safety and production efficiency, and effectively improves the comprehensiveness of mining production management.

[0004] The embodiments of this application disclose the following technical solutions: In a first aspect, an embodiment of the present application provides a coal mine safety production analysis method that integrates multimodal data, the method comprising: Based on the pre-stored equipment status ledger, calibrate the equipment aging label; Classify the behavioral data of personnel whose service time is less than or equal to the service time threshold to determine Class I, Class II, and Class III violations. Class I violations represent violations with a security incident triggering probability greater than or equal to the triggering probability threshold; Class II violations represent violations with a security incident triggering probability less than the triggering probability threshold and greater than 0; Class III violations represent violations that are inconsistent with standard behavior and have a security incident triggering probability equal to 0; Randomly allocate coal mine production; Traverse the first, second, and third types of violations, respectively combine them with the coal mine production volume and the equipment aging label to build index constraints, retrieve the first, second, and third types of coal mine production scheduling sample sets, calculate the accident sample ratios, and obtain the first, second, and third types of coal mine production accident triggering probabilities; When the probability of triggering a Class I coal mine production accident, the probability of triggering a Class II coal mine production accident, and the probability of triggering a Class III coal mine production accident are all less than or equal to the accident trigger probability threshold, production scheduling management is performed based on the coal mine production volume.

[0005] In a second aspect, an embodiment of the present application provides a coal mine safety production analysis system that integrates multimodal data, the system comprising: Equipment aging calibration module, used to calibrate equipment aging labels based on pre-stored equipment status records; A personnel violation classification module is used to classify the behavioral data of personnel whose service time is less than or equal to the service time threshold, and determine Class I, Class II, and Class III violations. Class I violations represent violation types with a security event triggering probability greater than or equal to the triggering probability threshold, Class II violations represent violation types with a security event triggering probability less than the triggering probability threshold and greater than 0, and Class III violations represent violations that are inconsistent with standard behavior and have a security event triggering probability equal to 0; Randomly configure the output module, used to randomly configure the coal mine production; An accident probability calculation module is used to traverse the first type of violation, the second type of violation, and the third type of violation, respectively combine them with the coal mine production volume and the equipment aging label to construct index constraints, retrieve the first type of coal mine production scheduling sample set, the second type of coal mine production scheduling sample set, and the third type of coal mine production scheduling sample set, respectively calculate the accident sample ratio, and obtain the first type of coal mine production accident triggering probability, the second type of coal mine production accident triggering probability, and the third type of coal mine production accident triggering probability; The production scheduling management module is used to perform production scheduling management according to the coal mine production volume when the probability of triggering a Class I coal mine production accident, the probability of triggering a Class II coal mine production accident, and the probability of triggering a Class III coal mine production accident are all less than or equal to the accident trigger probability threshold.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages: This application proposes a coal mine production safety analysis method and system that integrates multimodal data. Through the coordinated operation of equipment aging label calibration, personnel violation classification, production volume configuration, accident probability calculation and production scheduling management, safe and efficient management and control of coal mine production scheduling are achieved. First, based on the equipment status ledger, the service life and maintenance frequency of each equipment are extracted. The stability probability is generated through the operation stability probability prediction channel of the corresponding model and integrated into the equipment aging label. The personnel behavior data with service life that meets the threshold is classified to determine the first category (safety event trigger probability ≥ threshold), second category (probability < threshold and > 0), and third category (probability = 0) of violation behavior. After randomly configuring the coal mine production volume, the index constraint is constructed by combining the first two. The first category, second category, and third category coal mine production scheduling sample sets are retrieved, and the accident sample ratio is calculated for each category to obtain the three types of accident trigger probabilities. Finally, depending on whether the probability meets the standard, production scheduling management is executed or the production volume is adjusted according to the preset gradient, and the risk is prompted to the user end, forming a closed-loop management process.

[0007] This application's technical solution builds a three-dimensional "production-equipment-personnel" management system by integrating multimodal data. This addresses the problem of traditional coal mine production scheduling management, which relies solely on a single dimension for decision-making, resulting in an inability to balance the multi-dimensional needs of production efficiency and operational safety, and poor scheduling management precision. Through the quantification of equipment risks, hierarchical violation management, and dynamic production adjustments, it not only provides a precise basis for balancing safety and output, but also effectively improves the comprehensiveness of mining production management by balancing production and operational safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0009] Figure 1 A flowchart of a coal mine safety production analysis method integrating multimodal data provided in an embodiment of the present application; Figure 2 A schematic diagram of the structure of a coal mine safety production analysis system that integrates multimodal data provided in an embodiment of the present application.

[0010] In the accompanying drawings, the components represented by the reference numerals are described as follows: Equipment aging calibration module 01, personnel violation classification module 02, random configuration output module 03, accident probability calculation module 04, production scheduling management module 05. DETAILED DESCRIPTION

[0011] This application provides a coal mine production safety analysis method and system that integrates multimodal data, which is used to solve the technical problems in the existing technology that coal mine production scheduling management only considers a single dimension and cannot take into account multi-dimensional requirements such as production efficiency, equipment status and personnel behavior, resulting in poor production scheduling management precision and difficulty in achieving coordinated production and operation safety.

[0012] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without making creative efforts are within the scope of protection of this application.

[0013] In the description of this application, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of the technical features indicated. Therefore, a feature specified as "first" or "second" may explicitly or implicitly include one or more of the described features. In the description of this application, "plurality" means two or more, unless otherwise specifically specified.

[0014] In the description of this application, the term "for example" is used to mean "used as an example, illustration or explanation". Any embodiment described as "for example" in this application is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is given to enable any person skilled in the art to implement and use the present invention. In the following description, details are listed for the purpose of explanation. It should be understood that a person of ordinary skill in the art will recognize that the present invention can be implemented without using these specific details. In other examples, well-known structures and processes will not be elaborated in detail to avoid obscuring the description of the present invention with unnecessary details. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed in this application.

[0015] Example 1, as shown in the attached Figure 1 As shown, the present application provides a coal mine safety production analysis method integrating multimodal data, the method comprising the following steps: S110: Based on the pre-stored equipment status ledger, calibrate the equipment aging label; In the embodiment of the present application, during the equipment aging label calibration stage, in order to ensure that the label can accurately reflect the accident risk of each device, it is necessary to form a comprehensive risk identification through equipment status data extraction and model-specific channel processing.

[0016] Specifically, we first extract the service life and maintenance frequency of the No. 1 equipment in the coal mine production line from the pre-stored equipment status ledger as basic characteristic parameters, and process them through the first operation stability probability prediction channel bound to the equipment model to generate the No. 1 equipment stability probability (that is, the probability of an accident occurring when the equipment is operating).

[0017] Furthermore, the same operation is performed on the Q-th equipment in the coal mine production line until the service life and maintenance frequency of all equipment are extracted, and the corresponding equipment stability probability is generated through the exclusive operation stability probability prediction channel bound to each model.

[0018] Finally, the stability probabilities of equipment numbers 1 to Q are combined to form an equipment aging label. This label integrates the stability probabilities of all equipment numbers in the coal mine production line and comprehensively reflects the degradation trend and risk status of each equipment.

[0019] This step quantifies and integrates equipment accident risks by extracting key status parameters one by one according to the equipment position number and matching them with the model-specific probability prediction channel to generate stable probabilities. This provides accurate basic data support in the equipment dimension for subsequent production scheduling risk assessment based on personnel behavior data and production volume.

[0020] Step S110 of the method provided in the embodiment of the present application includes: Extract the service time and maintenance frequency of the first equipment in the coal mine production line from the equipment status ledger, and process them through the first operation stability probability prediction channel bound to the first equipment model to generate the stability probability of the first equipment. The equipment stability probability represents the probability of an accident occurring during the equipment operation. Until the service life and maintenance frequency of the Q-th equipment in the coal mine production line are extracted from the equipment status ledger, and the stability probability of the Q-th equipment is generated by processing the Q-th operation stability probability prediction channel bound to the Q-th equipment model; The stability probabilities of the first-numbered devices up to the Q-th-numbered devices are added to the device aging tag.

[0021] In an embodiment of the present application, in a scenario where a coal mine production line contains multiple equipment with different models, status parameters and risk characteristics, in order to ensure that the label can comprehensively and accurately reflect the risk status of all equipment in the coal mine production line, it is necessary to generate and integrate stable probabilities for each equipment number to achieve quantitative characterization of the risks of the entire production line equipment.

[0022] In the method provided in the embodiment of the present application, the step of "extracting the service time and maintenance frequency of the first equipment of the coal mine production line from the equipment status ledger, and generating the stability probability of the first equipment through a first operation stability probability prediction channel bound to the model of the first equipment" includes: Retrieve multiple pieces of first-number device model operation record data, wherein any piece of the plurality of first-number device model operation record data includes first-number device service time record data, first-number device maintenance frequency record data, and a device fault trigger flag, where when the flag is equal to 1, it is considered that the device fault state is triggered, and when the flag is equal to 0, it is considered that the device healthy operation state is triggered; Traversing the plurality of first-numbered device model operation record data, and retrieving a plurality of supplementary device fault trigger identifiers of the first-numbered device model based on the first-numbered device service time record data and the first-numbered device maintenance frequency record data as constraints; Counting the proportion of 0 values ​​in the plurality of supplementary device fault trigger identifiers and the device fault trigger identifiers, and setting the result as a label identifying the stability probability of the first-numbered device; Using the label of the stability probability of the first-numbered device as supervision, and the service time record data and maintenance frequency record data of the first-numbered device as input, the multiple operation record data of the first-numbered device model are retrieved, and through machine learning, the first operation stability probability prediction channel is trained and bound to the first-numbered device model.

[0023] In an embodiment of the present application, during the model training stage for generating the equipment stability probability, in order to ensure that the trained first operation stability probability prediction channel can accurately adapt to the characteristics of the first-number equipment model, it is necessary to construct an equipment-specific prediction channel through historical operation data retrieval and supervised learning.

[0024] Specifically, we first retrieve multiple pieces of operation record data for the first-number equipment model from the database. These data contain status parameters such as service time and maintenance frequency, as well as a trigger flag to indicate whether the equipment is faulty (1 for faulty and 0 for healthy), providing basic samples for model training.

[0025] Among them, by traversing historical data and retrieving additional fault identifications based on service time and maintenance frequency as constraints, the sample dimensions can be enriched to ensure the comprehensiveness of the training data.

[0026] Furthermore, the proportion of healthy operating states (i.e., marked as 0) is used as a stable probability label, and machine learning training is performed with state parameters as input, so that the model can accurately map the relationship between equipment status and accident probability, and ultimately form a device-specific prediction channel bound to the equipment model.

[0027] In the embodiment of the present application, when training the first operation stability probability prediction channel, it is necessary to achieve accurate mapping of equipment status and stability probability through multi-source sample fusion, probability label calibration and supervised learning training, so as to provide reliable model support for the subsequent generation of equipment aging labels.

[0028] Specifically, taking the first-number equipment model as an example, we first collect historical operation records including service time, maintenance frequency and fault trigger identification, and retrieve supplementary samples through the same state parameter constraints to expand the amount of training data.

[0029] For example, if a record shows that the equipment service time is 5,000 hours, the maintenance frequency is twice a month, and the trigger flag is 0 (healthy), then other records with the same parameter combination are retrieved as supplementary samples. If 8 out of 10 supplementary samples are marked as 0, the stable probability label in this state is 80%.

[0030] Furthermore, with the stable probability label as the supervision target, the service time and maintenance frequency are used as input features, and the model is trained through the random forest algorithm, so that the model can output the stable probability according to the real-time state parameters.

[0031] Specifically, the collected multiple operation record data of the first-number equipment model are first divided into training set, validation set and test set in a ratio of 8:1:1 to ensure that the data covers scenarios with different service time, maintenance frequency and fault trigger identification, so as to reflect the operation characteristics of this model of equipment in diverse states.

[0032] For example, 1600 training sets, 200 validation sets, and 200 test sets are extracted from 2000 records for model parameter learning, performance evaluation, and generalization ability testing to ensure that the trained model can adapt to the status characteristics of the No. 1 equipment model under different service times and maintenance frequencies.

[0033] Furthermore, when using the training set to train the random forest model, 70% of the sample data is randomly selected from the training set each time, and the service time record data and maintenance frequency record data are used as input features, and the stable probability label is used as the output label and input into the model.

[0034] During training, the initial number of decision trees was set at 30, and model performance was evaluated on the validation set every time five trees were added. After the first round of training, the average error between the predicted stability probability and the actual stability probability label on the validation set was 0.08. When the number of decision trees was increased to 50, this error narrowed to 0.05, indicating that the model's ability to fit the relationship between device status and stability probability is gradually improving.

[0035] To prevent overfitting, an early stopping mechanism was implemented. Training was stopped when the validation set error stopped decreasing for four consecutive rounds (error fluctuation was less than 0.002), indicating model convergence. For example, if the validation set error stabilized at 0.04 and did not decrease for four consecutive rounds during the 25th round of training, the model was considered converged and training was stopped.

[0036] Finally, the trained random forest model can receive the real-time service time and maintenance frequency of the first-number equipment and output the corresponding stability probability, providing a quantitative basis for calculating the stability probability of the first-number equipment. The model serves as the first operation stability probability prediction channel bound to the first-number equipment model.

[0037] For example, if the first-numbered piece of equipment is a coal mine roadheader, and a historical record shows a service life of 8,000 hours and a maintenance frequency of four times per month, the corresponding stability probability label is 65%. After inputting this data into model training, when the roadheader's real-time service life reaches 7,500 hours and a maintenance frequency of three times per month, the trained first-operation stability probability prediction channel processes it and outputs a stability probability of 72%. This accurately reflects the accident risk of equipment operation in this state and provides quantitative data for calibrating equipment aging labels.

[0038] Similarly, for the Q-th equipment in the coal mine production line, the same model training process as the first-number equipment is used to ensure that the Q-th operation stability probability prediction channel can accurately adapt to the characteristics of the equipment model.

[0039] Specifically, we first retrieve multiple pieces of Q-numbered equipment model operation record data from the equipment status ledger. Each record contains its service time record data, maintenance frequency record data, and equipment fault trigger flag (1 for fault, 0 for healthy).

[0040] Furthermore, by traversing these records, supplementary fault trigger identifiers are retrieved with service time and maintenance frequency as constraints, and the proportion of healthy status is calculated as a stable probability label; then, using this label as supervision and service time and maintenance frequency as input, historical data is retrieved to train the Q-th operation stable probability prediction channel through machine learning, and it is bound to the Q-th position number equipment model.

[0041] Furthermore, the stability probabilities of the devices from the first to the Qth position are sequentially integrated to form a device aging label.

[0042] For example, if the Qth device is a coal mine ventilator, and a historical record shows a service life of 12,000 hours and a maintenance frequency of 6 times per month, the corresponding stability probability label is 40%. After model training, when the ventilator's real-time service life is 11,000 hours and the maintenance frequency is 5 times per month, the Qth operation stability probability prediction channel outputs a stability probability of 48%.

[0043] Furthermore, the 72% stability probability of the No. 1 tunnel boring machine, the 48% stability probability of the No. Q ventilator, and the stability probabilities of other equipment (such as 85% of the No. 2 scraper conveyor, 60% of the No. 3 hydraulic support, etc.) are integrated to form an equipment aging label.

[0044] Among them, the equipment aging label centrally presents the quantitative value of the accident risk of equipment in the entire production line, providing unified equipment dimension data support for the subsequent construction of index constraints in combination with personnel violations and coal mine production volume, and realizing the systematic representation of multiple equipment statuses.

[0045] S120: Classifying the behavior data of personnel whose service time is less than or equal to the service time threshold to determine Class I, Class II, and Class III violations, wherein Class I violations represent violations whose security event triggering probability is greater than or equal to the triggering probability threshold, Class II violations represent violations whose security event triggering probability is less than the triggering probability threshold and greater than 0, and Class III violations represent violations that are inconsistent with standard behavior and whose security event triggering probability is equal to 0; In the embodiment of the present application, during the personnel violation classification stage, in order to ensure that the classification results can accurately reflect the safety risk levels of different violations, it is necessary to extract behavioral data, match it with the standard action library, and perform historical accident correlation analysis to achieve accurate risk classification.

[0046] Specifically, first, the first personnel behavior data including the execution position number and execution action is extracted from the personnel behavior data, the execution position number is matched with the pre-built standard action library, and the standard execution action corresponding to the position is obtained, which is used as the basis for judging the compliance of the behavior.

[0047] Furthermore, when the execution action is inconsistent with the standard execution action, the historical job records of the job number are retrieved with the execution action as a constraint, and the proportion of transactions in the historical records that triggered security incidents due to the same or similar illegal actions is counted and set as the security incident triggering probability of the behavior.

[0048] At the same time, based on the trigger probability threshold set by the user end, the first person behavior data is classified.

[0049] Finally, after all personnel behavior data are analyzed, the classification results of three types of violations are output, providing an accurate basis for quantifying personnel-dimensional risks for the subsequent construction of index constraints for production scheduling risk assessment based on equipment aging labels and coal mine production volume.

[0050] Step S120 in the method provided in the embodiment of the present application includes: Extracting first personnel behavior data from the personnel behavior data, wherein the first personnel behavior data includes an execution position number and an execution action; Based on the execution position number, a pre-built standard action library is input to match standard behaviors, wherein the standard behaviors include standard execution actions; When the execution action is inconsistent with the standard execution action, the historical operation records of the execution position number are retrieved based on the execution action as a constraint; Counting the proportion of security incident triggering transactions in the historical job records of the execution post number, and setting it as the security incident triggering probability; Classifying the first personnel behavior data based on the trigger probability threshold preset by the user terminal and the security event trigger probability; When the analysis of all personnel behavior data is completed, the first type of violation, the second type of violation, and the third type of violation are output.

[0051] In the embodiment of the present application, in order to ensure that the classification results can accurately distinguish the safety risk levels of different violations, risk classification must be achieved through behavioral feature extraction, standard library matching and historical accident correlation analysis, providing a risk basis in the personnel dimension for subsequent production scheduling risk assessment.

[0052] Specifically, firstly, the first personnel behavior data including the execution position number and the execution action is extracted from the personnel behavior data to clarify the position attribute and specific operation content of the behavior subject.

[0053] For example, if the first personnel behavior data extracted from the personnel behavior data is "Executing position number 005 (coal mining machine driver position), executing the action 'Failure to check the hydraulic system pressure to the rated value before starting the machine'", it can be clearly determined that the subject of the behavior is the coal mining machine driver position, and the specific operation content is failure to check the hydraulic system pressure to the rated value before starting the machine, laying the foundation for subsequent matching with the standard action library.

[0054] Furthermore, based on the execution position number, a pre-built standard action library is matched to obtain the standard execution action corresponding to the position, which is used as a benchmark for judging the compliance of the behavior.

[0055] For example, the standard action library for the coal mining machine driver position stipulates that "the hydraulic system pressure must be checked to the rated value before starting the machine", and this action is the standard execution action for this position.

[0056] Furthermore, when the execution action is inconsistent with the standard execution action, it is determined to be a potential violation and its security risk needs to be quantified through historical data retrieval.

[0057] At the same time, with the execution action as a constraint, retrieve the historical job records of the same execution position number, count the proportion of transactions in the historical records that triggered security incidents due to the same or similar illegal actions, and set it as the security incident triggering probability.

[0058] For example, if the action of "failing to check the hydraulic system pressure to the rated value before starting the machine" occurs 80 times in the coal mining machine driver's job history record, and 12 of them cause safety incidents related to hydraulic system failure, then the safety incident triggering probability of this action is 15% (12 / 80).

[0059] Furthermore, based on the trigger probability threshold set by the user side (such as the threshold is set to 10%), the first person behavior data is classified: the safety incident triggering probability greater than or equal to 10% is classified as a Class I violation (high risk), less than 10% and greater than 0 is classified as a Class II violation (medium risk), and inconsistent with the standard behavior but has not triggered a safety incident in the historical record (the triggering probability is 0) is classified as a Class III violation (low risk).

[0060] Finally, after all personnel behavior data have completed the above analysis, the classification results of the three types of violations are summarized and output to achieve a systematic risk grading of personnel violations.

[0061] This classification process compares job-specific standards with historical accident data, ensuring that the classification results accurately reflect the actual risk levels of different violations, laying the foundation for the subsequent construction of multi-dimensional production scheduling constraints based on equipment aging labels and coal mine production volume.

[0062] S130: Randomly configure coal mine production; In the embodiment of the present application, during the random configuration stage of coal mine production, in order to ensure that the configuration results can cover risk scenarios under different production loads, it is necessary to achieve diversified settings of production volume through unbiased random sampling and full-range coverage principles, so as to provide comprehensive production dimension data support for subsequent multi-dimensional production scheduling risk assessment.

[0063] Specifically, the range of coal mine production volume must be clarified first. This range is determined based on the design capacity of the coal mine production line, the maximum load of the equipment, and the production threshold specified by safety regulations, covering the entire range from the minimum safe production volume to the maximum design production volume.

[0064] For example, the design production capacity of a coal mine production line is 8,000 tons per day, and the safety regulations limit the minimum production to 3,000 tons per day. The production value range is set to 3,000 tons to 8,000 tons.

[0065] Furthermore, a random sampling method is used to generate several discrete production volume values ​​within the above-mentioned value range to ensure that each value conforms to the operating load characteristics of the production equipment and that the difference between any two values ​​is not less than the preset minimum interval (e.g., 500 tons) to avoid evaluation redundancy caused by overly dense production.

[0066] For example, within the range of 3,000 tons to 8,000 tons, eight production volume values ​​of 3,200 tons, 3,800 tons, 4,500 tons, 5,200 tons, 6,000 tons, 6,800 tons, 7,500 tons, and 8,000 tons are randomly generated.

[0067] Finally, the generated coal mine production volumes are used as random configuration results to form a production volume set. This set covers the possible production load states of the coal mine production line and provides a diverse set of basic production data for subsequent production scheduling risk assessment based on equipment aging labels and personnel violation data.

[0068] This step achieves comprehensive coverage of production load by clarifying the value range and randomly sampling to generate diverse production volume values, providing reliable production dimension support for multi-dimensional risk assessment and helping to improve the adaptability and security of production scheduling plans.

[0069] S140: Traverse the first category of violations, the second category of violations, and the third category of violations, respectively combine them with the coal mine production volume and the equipment aging label to construct index constraints, retrieve the first category of coal mine production scheduling sample set, the second category of coal mine production scheduling sample set, and the third category of coal mine production scheduling sample set, calculate the accident sample ratio respectively, and obtain the first category of coal mine production accident triggering probability, the second category of coal mine production accident triggering probability, and the third category of coal mine production accident triggering probability; In the embodiment of the present application, in order to ensure that the accident risks under different combinations of violations, production loads and equipment status can be reflected in a hierarchical manner, it is necessary to realize the division of three types of production scheduling sample sets and the calculation of accident triggering probabilities by constructing index constraints and retrieval sample sets, so as to provide a decision-making basis for the optimization of the final production scheduling plan.

[0070] Specifically, we first traverse a class of violations, and use the specific manifestations of this class of violations and the randomly configured coal mine production volume as dual constraints, and retrieve the first-level coal mine production sample set that matches the constraint conditions in the historical coal mine production records.

[0071] Furthermore, the sample equipment status ledger of the first coal mine production scheduling sample is extracted from the first-level coal mine production scheduling sample set, and the sample equipment aging label is calibrated according to the same process as the current equipment aging label calibration.

[0072] Among them, when the sample equipment aging label is consistent with the currently generated equipment aging label, the first coal mine production scheduling sample is included in a type of coal mine production scheduling sample set.

[0073] Furthermore, the same process is used to handle Class II and Class III violations. Specifically, for Class II violations, the secondary coal mine production scheduling sample set is retrieved using the secondary coal mine production volume as a constraint. By matching the aging labels of the sample equipment with the aging labels of the current equipment, a Class II coal mine production scheduling sample set is constructed. For Class III violations, a Class III coal mine production scheduling sample set is constructed in the same way.

[0074] Finally, the proportion of samples in which production accidents have occurred in the three types of coal mine production scheduling samples is calculated respectively, which are used as the probability of triggering production accidents in type I coal mines, type II coal mines and type III coal mines respectively.

[0075] This step constructs index constraints by associating the three dimensions of personnel violations, coal mine production, and equipment status, thereby achieving accurate retrieval of historical samples and quantification of accident probabilities under similar risk scenarios, providing a multi-dimensional probabilistic basis for the subsequent formulation of production scheduling strategies based on risk levels.

[0076] Step S140 in the method provided in the embodiment of the present application includes: Extracting a first violation from the class of violations; Retrieving a first-level coal mine production scheduling sample set based on the first violation and the coal mine production volume as constraints; Extracting the sample equipment status ledger of the first coal mine production scheduling sample of the first-level coal mine production scheduling sample set, and calibrating the aging label of the sample equipment; When the sample equipment aging label is the same as the equipment aging label, adding the first coal mine production scheduling sample to the first type of coal mine production scheduling sample set; Among them, the determination process of the second-category coal mine production scheduling sample set and the third-category coal mine production scheduling sample set is the same as that of the first-category coal mine production scheduling sample set.

[0077] In the embodiment of the present application, in order to ensure that the accident risks under different combinations of violations, production loads and equipment status can be reflected in a hierarchical manner, it is necessary to construct three types of production scheduling sample sets through hierarchical retrieval and sample matching to provide a standardized sample basis for subsequent accident triggering probability calculations.

[0078] Specifically, first, for a class of illegal behaviors, a specific first illegal behavior is extracted from the class of behaviors, for example, a specific illegal action with a security event trigger probability ≥ a threshold.

[0079] For example, if a type of violation includes "the coal mining machine driver did not check the hydraulic system pressure to the rated value before starting the machine (the probability of triggering a safety event is 15%, and the threshold is set to 10%)", then this action will be regarded as the first violation.

[0080] Furthermore, taking the first violation and the randomly configured coal mine production volume as dual constraints, all samples that meet the constraints in the historical coal mine production records are retrieved to form a first-level coal mine production sample set.

[0081] For example, if the first violation is "the coal mining machine driver did not check the hydraulic system pressure to the rated value before starting the machine" and the randomly configured coal mine production is 5,000 tons / day, then all samples in the historical production scheduling records of "the coal mining machine driver did not check the hydraulic system pressure to the rated value before starting the machine and the production on that day was 5,000 tons / day" are retrieved, and these samples are aggregated to form a first-level coal mine production scheduling sample set.

[0082] Furthermore, the first coal mine production scheduling sample is selected from the acquired first-level coal mine production scheduling sample set, and its corresponding sample equipment status ledger is extracted. According to the process that is completely consistent with the current equipment aging label calibration, that is, the stability probability is generated and integrated through the operation stability probability prediction channel exclusive to each equipment model, and the sample equipment aging label of the sample is calibrated.

[0083] Among them, if the sample equipment aging label is completely consistent with the currently generated equipment aging label, the first coal mine production scheduling sample is included in the first category coal mine production scheduling sample set.

[0084] For example, a first coal mine production schedule sample is selected from the above-mentioned first-level coal mine production schedule sample set, and its sample equipment status ledger includes the status parameters of equipment such as the No. 1 tunnel boring machine (service time 7,500 hours, maintenance frequency 3 times a month), the No. Q ventilator (service time 11,000 hours, maintenance frequency 5 times a month), etc.

[0085] At the same time, according to the equipment aging label calibration process, stability probabilities (such as 72% for roadheaders and 48% for ventilators) are generated through each dedicated prediction channel and integrated into a sample equipment aging label. If the sample equipment aging label is exactly the same as the current equipment aging label (such as 72% for roadheaders and 48% for ventilators), the sample is added to the Class 1 coal mine production scheduling sample set.

[0086] Similarly, for Class II and Class III violations, the same processing procedures as Class I violations are used to construct Class II coal mine production scheduling sample sets and Class III coal mine production scheduling sample sets respectively.

[0087] Ultimately, the second-category coal mine production scheduling sample set constructed by the same logic (such as the second-category violation sample of "gas detectors failing to carry gas detectors as required") and the third-category coal mine production scheduling sample set (such as the third-category violation sample of "transport workers failing to follow standard routes during operations") together constitute a production scheduling sample system covering all risk levels.

[0088] This process uses parameters in three dimensions, namely, personnel violations, coal mine production volume, and equipment status, as correlation constraints to achieve accurate screening of historical production scheduling samples under similar risk scenarios, ensuring that the three types of sample sets correspond to different combinations of violation risk levels, equipment, and production status, respectively. This provides a standardized and comparable sample basis for the subsequent calculation of the probability of production accident triggering under each category, so that the final probability calculation results can truly reflect the possibility of accidents in different risk scenarios.

[0089] On this basis, the method provided in the embodiment of the present application further includes: When any one of the probability of triggering a Class I coal mine production accident, the probability of triggering a Class II coal mine production accident, and the probability of triggering a Class III coal mine production accident is greater than the accident trigger probability threshold, determining whether the coal mine production volume is less than the production volume threshold; If it is greater than or equal to, based on a preset gradient, the coal mine production is reduced to obtain an updated coal mine production execution cycle; If it is less than that, a production risk warning will be issued to the coal mine production line at the user end.

[0090] In the embodiment of the present application, after the corresponding accident trigger probability is calculated through the production scheduling sample sets of three types of coal mines, in order to achieve dynamic prevention and control of production risks and optimal balance of output, it is necessary to conduct closed-loop control of coal mine production through threshold judgment and step-by-step adjustment to ensure that production continuity is maintained under the premise of controllable risks.

[0091] Specifically, first, the calculated triggering probability of a Class I coal mine production accident, the triggering probability of a Class II coal mine production accident, and the triggering probability of a Class III coal mine production accident are compared with preset accident triggering probability thresholds respectively.

[0092] Among them, when the probability value of triggering any type of coal mine production accident exceeds the threshold, it indicates that the current production combination (corresponding to violations, equipment status and production volume) has a high safety risk, and the production volume adjustment process needs to be started immediately.

[0093] For example, if the preset accident trigger probability threshold is 10%, the calculated trigger probability of a Class I coal mine production accident is 15%, Class II is 8%, and Class III is 5%. Since the Class I probability exceeds the threshold, it indicates that the current production combination has a high safety risk and the production volume adjustment process needs to be started immediately.

[0094] Furthermore, after the production adjustment process is initiated, it is prioritized to determine whether the current coal mine production is less than the production threshold. The production threshold is set based on the coal mine's capacity design and safety regulations. For example, the production threshold of a coal mine is 4,000 tons / day.

[0095] Specifically, if the current production volume is greater than or equal to the production volume threshold, it means that there is room to avoid risks by reducing production. At this time, the production volume is adjusted downward based on the preset gradient, the updated coal mine production volume is generated, and the entire process is re-executed, that is, the equipment aging label is recalibrated, the personnel behavior is classified, the production volume is configured, and the accident trigger probability is calculated, until the accident trigger probability of all categories is lower than the threshold or the production volume has dropped below the threshold.

[0096] For example, if the calculated probability of triggering a Class I coal mine production accident is 18% (the accident trigger probability threshold is 10%), and the current coal mine production is 5,000 tons / day (the production threshold is 4,000 tons / day), because the production is greater than the threshold, the production is reduced by 500 tons / day at a time according to the preset gradient to 4,500 tons / day and the cycle is re-executed; if the probability of triggering a Class II coal mine production accident is still 12% (>10%) after recalculation, it will continue to be reduced to 4,000 tons / day and the cycle will be executed again until the conditions are met.

[0097] On the contrary, if the current production volume is less than the production volume threshold, it means that there is no room for downward adjustment, and continued production may face uncontrollable risks. At this time, a production risk warning needs to be issued on the user side to take targeted measures.

[0098] In the method provided in the embodiment of the present application, the step of “providing a production risk warning to the coal mine production line at the user end if less than” includes: Setting the device stability probability of the first bit number of the device aging tag to 0, and obtaining an updated device aging tag; Traversing the first, second, and third types of violations, respectively, combining them with the production volume threshold and the update equipment aging tag to construct index constraints, and calculating the probability of triggering a first, second, and third type of updated coal mine production accidents; When the trigger probability of the first type of updated coal mine production accident, the trigger probability of the second type of updated coal mine production accident, and the trigger probability of the third type of updated coal mine production accident are all less than or equal to the accident trigger probability threshold, the first number device is added to the risk warning device; Until the analysis of all the position numbered devices is completed, the risk warning device will issue a production risk warning to the coal mine production line at the user end according to the risk warning device.

[0099] In the embodiment of the present application, in a scenario where the coal mine production is less than the production threshold and the probability of accident triggering still exceeds the standard, in order to accurately locate the source of the risk and generate targeted production risk warnings, it is necessary to accurately identify the risk equipment through individual and associated analysis of the equipment risks one by one to ensure that the prompt content focuses on key hidden dangers.

[0100] Specifically, first, for the first device number included in the device aging label, its stability probability is forcibly set to 0 (to simulate the extreme risk state of complete failure of the device), and an updated device aging label is generated.

[0101] For example, if the first-numbered device is a coal mine roadheader, its stability probability in the device aging label is 72%. After it is forcibly set to 0, the stability probability of the device in the updated device aging label becomes 0, and the remaining numbered devices (such as the Q-numbered ventilator 48%) remain unchanged.

[0102] Among them, this step can intuitively assess the impact of a single device on the overall production risk by isolating the risk of the device.

[0103] Furthermore, we traverse the first, second and third category violations, and use the various types of violations, the current production volume threshold and the updated equipment aging label as triple constraints to rebuild the index constraint, retrieve the corresponding updated coal mine production scheduling sample set, and count the proportion of accident samples therein to obtain the first, second and third category updated coal mine production accident triggering probabilities.

[0104] Among them, because the coal mine production is less than the production threshold, the production threshold is fixed as the production load parameter at this time.

[0105] For example, if the production threshold is 4,000 tons / day and the current coal mine production is 3,500 tons / day (less than the threshold), 4,000 tons / day will be used as the production load parameter during retrieval. Combined with the violation of "coal mining machine driver failed to check the hydraulic system pressure" and the updated equipment aging label (the tunnel boring machine stability probability is 0%), the sample set that matches this constraint in the retrieval history is retrieved.

[0106] Furthermore, if the updated coal mine production accident trigger probability is less than or equal to the accident trigger probability threshold, the first device is considered the key factor causing the original risk to exceed the standard, and it is added to the risk warning device list. If the probability still exceeds the standard after the update, it indicates that the device is not a core risk source and is not included in the warning list.

[0107] For example, if the accident trigger probability threshold is 10%, the original Class I coal mine production accident trigger probability is 15% (exceeding the standard), and after setting the stability probability of the first-number tunnel boring machine to 0, the calculated Class I update probability is 8% (≤10%), then the tunnel boring machine is determined to be a key risk source and added to the risk warning equipment list.

[0108] Similarly, the devices with the 2nd to Qth digits are analyzed in sequence according to the above processing flow. That is, the stability probability of each digit of the device is set to 0 to generate the corresponding updated device aging label. The constraint construction and accident trigger probability calculation steps are repeated until the risk test of all digits of the device is completed.

[0109] For example, when analyzing the scraper conveyor No. 2 (original stability probability of 85%), its stability probability is set to 0 to generate a new updated equipment aging label. Combined with the second-class violation of "failure to carry a gas detector as required" and the production volume threshold of 4,000 tons / day, the second-class update probability is recalculated to be 12% (still exceeding the standard), and it is determined that the scraper conveyor is not a core risk source and is not included in the list.

[0110] Finally, a production risk warning is generated on the user side based on the risk warning equipment list. The content includes the equipment position number that needs to be paid special attention to, its current stability probability and the associated risk scenario, providing on-site personnel with clear investigation and intervention targets.

[0111] For example, the risk warning content may be "Risk warning equipment: No. 1 tunnel boring machine (current stability probability 72%). When this equipment is combined with the violation of 'failure to check the hydraulic system pressure' and an output of 4,000 tons / day, it is a key factor leading to an excessive probability of accident triggering. Please shut down the machine for maintenance as a priority."

[0112] This process achieves a dynamic balance between production volume and safety risks through a closed-loop solution of "threshold judgment-gradient adjustment-cyclic verification-risk warning". It not only avoids efficiency losses caused by blind production stoppages, but also prevents risk accumulation through timely intervention, providing a flexible and rigorous management and control mechanism for coal mine safety production.

[0113] S150: When the probability of triggering a Class I coal mine production accident, the probability of triggering a Class II coal mine production accident, and the probability of triggering a Class III coal mine production accident are all less than or equal to the accident triggering probability threshold, production scheduling management is performed according to the coal mine production volume.

[0114] In the embodiment of the present application, in a scenario where the probability of triggering three types of coal mine production accidents is within a safe and controllable range, in order to achieve coordinated optimization of production efficiency and stable equipment operation, it is necessary to use the current coal mine production volume as a benchmark, combine the equipment status and personnel behavior norms to schedule and manage the entire production line, so as to ensure that the operating parameters of each link are accurately matched with the output demand.

[0115] Specifically, first of all, the coal mine production volume after risk verification is used as the core parameter. For the No. 1 equipment in the coal mine production line, according to its stability probability in the equipment aging label, the operation stability probability prediction channel bound to the equipment model is called to generate operating parameters adapted to the current production volume to ensure that the equipment load matches the output target.

[0116] For example, if the first-numbered equipment is a tunnel boring machine, its stability probability in the equipment aging label is 72%, and the current coal mine production is 5,000 tons / day. Through the operation stability probability prediction channel bound to the tunnel boring machine model, the generated operating parameters are "operating power 80%, start and stop interval 5 minutes per hour", so that the tunneling efficiency is adapted to the production demand of 5,000 tons / day.

[0117] Similarly, the same operation is performed on the equipment No. 2 to No. Q in the coal mine production line. That is, based on the stability probability and model characteristics of each equipment, the corresponding operation stability probability prediction channel is used to generate operating parameters adapted to the current production volume, so that the operating status of the equipment in the entire production line forms a dynamic balance with the output demand.

[0118] For example, the second equipment is a scraper conveyor with a stability probability of 85%. Combined with the production volume of 5000 tons / day, the operating parameters of "conveying speed 1.2m / s, coal cleaning once every 30 minutes" are generated through its exclusive operation stability probability prediction channel; the Qth equipment is a ventilator with a stability probability of 48%, corresponding to the generation of "air volume 3000m 3 / h, wind pressure 2500Pa" parameters to ensure that the ventilation efficiency matches the current production scale.

[0119] At the same time, based on the classification results of Class I, Class II, and Class III violations, operational specifications and guidelines that match the current production volume are pushed to personnel in corresponding positions. For example, for positions where Class I violations have occurred, the execution standard prompts for key steps are strengthened to ensure that personnel behavior is consistent with equipment operation and production targets.

[0120] For example, if the first type of violation includes "the coal mining machine driver failed to check the hydraulic system pressure", and the current production volume is 5,000 tons / day, the operating instructions of "checking the hydraulic system pressure once an hour to ensure that the pressure is maintained in the range of 15-20MPa" will be pushed to the coal mining machine driver; for the second type of violation "failing to carry a gas detector as required", the standard prompt of "recording the gas concentration every 2 hours and uploading the detection data simultaneously" will be pushed to the gas detector.

[0121] Ultimately, the operating parameters of each device and personnel operating specifications are integrated into a unified production scheduling management plan to achieve systematic linkage of the entire production line from equipment operation to personnel operations.

[0122] This step builds a "production volume-equipment-personnel" management system by adapting production parameters one by one according to the position number of the equipment and linking personnel behavior standards. It maximizes production efficiency under the current output while ensuring operational safety, and provides a precise management basis for the efficient and stable operation of the coal mine production line.

[0123] The embodiments of the present application achieve the following technical effects through the above specific implementation methods: This application proposes a coal mine production safety analysis method that integrates multimodal data. First, based on the equipment status ledger, the service life and maintenance frequency of each equipment are extracted. The stability probability is generated through the operation stability probability prediction channel bound to the equipment model and integrated into the equipment aging label; the personnel behavior data is classified to determine the first type of violation, the second type of violation and the third type of violation, and the coal mine production volume is randomly allocated; then the three are combined to construct an index constraint, and the first type of coal mine production sample set, the second type of coal mine production sample set and the third type of coal mine production sample set are retrieved to calculate the probability of triggering a first type of coal mine production accident, the second type of coal mine production accident triggering probability and the third type of coal mine production accident triggering probability; finally, according to whether the probability meets the standard, the production scheduling management is executed, or the production volume is adjusted according to the preset gradient, the risk is prompted on the user side and the risk equipment is accurately located, forming a closed-loop management of the entire process. This process realizes the coordinated management and control of multi-dimensional data such as production efficiency, equipment status, and personnel behavior, providing a complete solution for the balance between operation safety and production efficiency.

[0124] The method provided in the embodiment of the present application realizes the fusion and integrated management of multimodal data through the technical solution of "equipment aging label calibration - personnel violation classification - coal mine production volume configuration - accident trigger probability calculation - production scheduling management", and solves the technical problems existing in traditional coal mine production scheduling management, such as insufficient coordination between equipment status, personnel behavior and output demand, resulting in high safety risks and low production efficiency. It provides an accurate and efficient management solution for coal mine production scheduling, which not only ensures maximization of production efficiency under the premise of controllable risks, but also effectively improves the comprehensiveness of mining production management by coordinating multi-dimensional factors.

[0125] Example 2, as shown in the attached Figure 2 As shown, based on the inventive concept of the coal mine production safety analysis method for fusing multimodal data provided in Example 1, this application also provides a coal mine production safety analysis system for fusing multimodal data, specifically including: Equipment aging calibration module 01 is used to calibrate equipment aging labels based on pre-stored equipment status records; Personnel Violation Classification Module 02 is used to classify the behavior data of personnel whose service time is less than or equal to the service time threshold, and determine Class I, Class II, and Class III violations. Class I violations represent violations whose security event triggering probability is greater than or equal to the triggering probability threshold, Class II violations represent violations whose security event triggering probability is less than the triggering probability threshold and greater than 0, and Class III violations represent violations that are inconsistent with standard behavior and whose security event triggering probability is equal to 0; Randomly configure production module 03, used to randomly configure coal mine production; The accident probability calculation module 04 is used to traverse the first type of violation, the second type of violation, and the third type of violation, respectively, and combine them with the coal mine production volume and the equipment aging label to construct an index constraint, retrieve the first type of coal mine production scheduling sample set, the second type of coal mine production scheduling sample set, and the third type of coal mine production scheduling sample set, respectively calculate the accident sample ratio, and obtain the first type of coal mine production accident triggering probability, the second type of coal mine production accident triggering probability, and the third type of coal mine production accident triggering probability; The production scheduling management module 05 is used to perform production scheduling management according to the coal mine production volume when the probability of triggering a Class I coal mine production accident, the probability of triggering a Class II coal mine production accident, and the probability of triggering a Class III coal mine production accident are all less than or equal to the accident trigger probability threshold.

[0126] In one embodiment, the device aging calibration module 01 is further configured to: Extract the service time and maintenance frequency of the first equipment in the coal mine production line from the equipment status ledger, and process them through the first operation stability probability prediction channel bound to the first equipment model to generate the stability probability of the first equipment. The equipment stability probability represents the probability of an accident occurring during the equipment operation. Until the service life and maintenance frequency of the Q-th equipment in the coal mine production line are extracted from the equipment status ledger, and the stability probability of the Q-th equipment is generated by processing the Q-th operation stability probability prediction channel bound to the Q-th equipment model; The stability probabilities of the first-numbered devices up to the Q-th-numbered devices are added to the device aging tag.

[0127] In one embodiment, the personnel violation classification module 02 is further configured to: Extracting first personnel behavior data from the personnel behavior data, wherein the first personnel behavior data includes an execution position number and an execution action; Based on the execution position number, a pre-built standard action library is input to match standard behaviors, wherein the standard behaviors include standard execution actions; When the execution action is inconsistent with the standard execution action, the historical operation records of the execution position number are retrieved based on the execution action as a constraint; Counting the proportion of security incident triggering transactions in the historical job records of the execution post number, and setting it as the security incident triggering probability; Classifying the first personnel behavior data based on the trigger probability threshold preset by the user terminal and the security event trigger probability; When the analysis of all personnel behavior data is completed, the first type of violation, the second type of violation, and the third type of violation are output.

[0128] In one embodiment, the accident probability calculation module 04 is further configured to: Extracting a first violation from the class of violations; Retrieving a first-level coal mine production scheduling sample set based on the first violation and the coal mine production volume as constraints; Extracting the sample equipment status ledger of the first coal mine production scheduling sample of the first-level coal mine production scheduling sample set, and calibrating the aging label of the sample equipment; When the sample equipment aging label is the same as the equipment aging label, adding the first coal mine production scheduling sample to the first type of coal mine production scheduling sample set; Among them, the determination process of the second-category coal mine production scheduling sample set and the third-category coal mine production scheduling sample set is the same as that of the first-category coal mine production scheduling sample set.

[0129] It should be noted that the order in which the embodiments of the present application are presented is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. Furthermore, the foregoing descriptions of specific embodiments of this specification are provided. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order or sequential sequence shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0130] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should be included in the scope of protection of the present application.

[0131] This specification and drawings are merely illustrative of the present application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Obviously, those skilled in the art may make various modifications and variations to this application without departing from the scope of this application. Thus, this application is intended to include such modifications and variations as fall within the scope of this application and its equivalents.

Claims

1. A coal mine safety production analysis method integrating multimodal data, characterized in that: include: Based on the pre-stored equipment status ledger, calibrate the equipment aging label; Classify the behavioral data of personnel whose service time is less than or equal to the service time threshold to determine Class I, Class II, and Class III violations. Class I violations represent violations with a security incident triggering probability greater than or equal to the triggering probability threshold; Class II violations represent violations with a security incident triggering probability less than the triggering probability threshold and greater than 0; Class III violations represent violations that are inconsistent with standard behavior and have a security incident triggering probability equal to 0; Randomly allocate coal mine production; Traverse the first, second, and third types of violations, respectively combine them with the coal mine production volume and the equipment aging label to build index constraints, retrieve the first, second, and third types of coal mine production scheduling sample sets, calculate the accident sample ratios, and obtain the first, second, and third types of coal mine production accident triggering probabilities; When the probability of triggering a Class I coal mine production accident, the probability of triggering a Class II coal mine production accident, and the probability of triggering a Class III coal mine production accident are all less than or equal to the accident trigger probability threshold, production scheduling management is performed based on the coal mine production volume.

2. The method according to claim 1, wherein Based on the pre-stored equipment status ledger, calibrate the equipment aging label, including: Extract the service time and maintenance frequency of the first equipment in the coal mine production line from the equipment status ledger, and process them through the first operation stability probability prediction channel bound to the first equipment model to generate the stability probability of the first equipment. The equipment stability probability represents the probability of an accident occurring during the equipment operation. Until the service life and maintenance frequency of the Q-th equipment in the coal mine production line are extracted from the equipment status ledger, and the stability probability of the Q-th equipment is generated by processing the Q-th operation stability probability prediction channel bound to the Q-th equipment model; The stability probabilities of the first-numbered devices up to the Q-th-numbered devices are added to the device aging tag.

3. The method according to claim 2, wherein The service life and maintenance frequency of the first equipment in the coal mine production line are extracted from the equipment status ledger. The stability probability of the first equipment is generated by processing the first operation stability probability prediction channel bound to the model of the first equipment, including: Retrieve multiple pieces of first-number device model operation record data, wherein any piece of the plurality of first-number device model operation record data includes first-number device service time record data, first-number device maintenance frequency record data, and a device fault trigger flag, where when the flag is equal to 1, it is considered that the device fault state is triggered, and when the flag is equal to 0, it is considered that the device healthy operation state is triggered; Traversing the plurality of first-numbered device model operation record data, and retrieving a plurality of supplementary device fault trigger identifiers of the first-numbered device model based on the first-numbered device service time record data and the first-numbered device maintenance frequency record data as constraints; Counting the proportion of 0 values ​​in the plurality of supplementary device fault trigger identifiers and the device fault trigger identifiers, and setting the result as a label identifying the stability probability of the first-numbered device; Using the label of the stability probability of the first-numbered device as supervision, and the service time record data and maintenance frequency record data of the first-numbered device as input, the multiple operation record data of the first-numbered device model are retrieved, and through machine learning, the first operation stability probability prediction channel is trained and bound to the first-numbered device model.

4. The method according to claim 1, wherein The behavioral data of personnel whose service time is less than or equal to the service time threshold is classified into Category 1, Category 2, and Category 3 violations, including: Extracting first personnel behavior data from the personnel behavior data, wherein the first personnel behavior data includes an execution position number and an execution action; Based on the execution position number, a pre-built standard action library is input to match standard behaviors, wherein the standard behaviors include standard execution actions; When the execution action is inconsistent with the standard execution action, the historical operation records of the execution position number are retrieved based on the execution action as a constraint; Counting the proportion of security incident triggering transactions in the historical job records of the execution post number, and setting it as the security incident triggering probability; Classifying the first personnel behavior data based on the trigger probability threshold preset by the user terminal and the security event trigger probability; When the analysis of all personnel behavior data is completed, the first type of violation, the second type of violation, and the third type of violation are output.

5. The method according to claim 1, wherein Traverse the first, second, and third types of violations, respectively combine them with the coal mine production volume and the equipment aging label to build index constraints, retrieve the first, second, and third types of coal mine production scheduling sample sets, calculate the accident sample ratios, and obtain the first, second, and third types of coal mine production accident triggering probabilities, including: Extracting a first violation from the class of violations; Retrieving a first-level coal mine production scheduling sample set based on the first violation and the coal mine production volume as constraints; Extracting the sample equipment status ledger of the first coal mine production scheduling sample of the first-level coal mine production scheduling sample set, and calibrating the aging label of the sample equipment; When the sample equipment aging label is the same as the equipment aging label, adding the first coal mine production scheduling sample to the first type of coal mine production scheduling sample set; Among them, the determination process of the second-category coal mine production scheduling sample set and the third-category coal mine production scheduling sample set is the same as that of the first-category coal mine production scheduling sample set.

6. The method according to claim 2, wherein Also includes: When any one of the probability of triggering a Class I coal mine production accident, the probability of triggering a Class II coal mine production accident, and the probability of triggering a Class III coal mine production accident is greater than the accident trigger probability threshold, determining whether the coal mine production volume is less than the production volume threshold; If it is greater than or equal to, based on a preset gradient, the coal mine production is reduced to obtain an updated coal mine production execution cycle; If it is less than that, a production risk warning will be issued to the coal mine production line at the user end.

7. The method according to claim 6, wherein If it is less than that, the user will be prompted with production risk warnings for the coal mine production line, including: Setting the device stability probability of the first bit number of the device aging tag to 0, and obtaining an updated device aging tag; Traversing the first, second, and third types of violations, respectively, combining them with the production volume threshold and the update equipment aging tag to construct index constraints, and calculating the probability of triggering a first, second, and third type of updated coal mine production accidents; When the trigger probability of the first type of updated coal mine production accident, the trigger probability of the second type of updated coal mine production accident, and the trigger probability of the third type of updated coal mine production accident are all less than or equal to the accident trigger probability threshold, the first number device is added to the risk warning device; Until the analysis of all the position numbered devices is completed, the risk warning device will issue a production risk warning to the coal mine production line at the user end according to the risk warning device.

8. A coal mine safety production analysis system integrating multimodal data is characterized by: The system is used to execute the coal mine production safety analysis method for fusing multimodal data according to any one of claims 1 to 7, and the system includes: Equipment aging calibration module, used to calibrate equipment aging labels based on pre-stored equipment status records; A personnel violation classification module is used to classify the behavioral data of personnel whose service time is less than or equal to the service time threshold, and determine Class I, Class II, and Class III violations. Class I violations represent violation types with a security event triggering probability greater than or equal to the triggering probability threshold, Class II violations represent violation types with a security event triggering probability less than the triggering probability threshold and greater than 0, and Class III violations represent violations that are inconsistent with standard behavior and have a security event triggering probability equal to 0; Randomly configure the output module, used to randomly configure the coal mine production; An accident probability calculation module is used to traverse the first type of violation, the second type of violation, and the third type of violation, respectively combine them with the coal mine production volume and the equipment aging label to construct index constraints, retrieve the first type of coal mine production scheduling sample set, the second type of coal mine production scheduling sample set, and the third type of coal mine production scheduling sample set, respectively calculate the accident sample ratio, and obtain the first type of coal mine production accident triggering probability, the second type of coal mine production accident triggering probability, and the third type of coal mine production accident triggering probability; The production scheduling management module is used to perform production scheduling management according to the coal mine production volume when the probability of triggering a Class I coal mine production accident, the probability of triggering a Class II coal mine production accident, and the probability of triggering a Class III coal mine production accident are all less than or equal to the accident trigger probability threshold.

Images