Coal mine comprehensive support capability dynamic coupling early warning method and system

Abstract

The invention provides a coal mine comprehensive guarantee capability dynamic coupling early warning method and system, and relates to the technical field of coal mine safety production assessment, and the method comprises the steps: constructing a multi-dimensional fusion assessment system comprising a first-level index and a second-level index according to a coal mine comprehensive guarantee demand and an assessment target; collecting corresponding multi-source heterogeneous index data under each dimension in real time; a two-stage dynamic weight distribution mechanism is adopted, an improved entropy weight method and a working condition sensing factor are combined, and the final weight of each index at the current moment is calculated; constructing a four-dimensional feature vector and forming a multi-dimensional tensor, calculating a comprehensive score of each dimension, and generating a coal mine comprehensive support capability index through weighted fusion and Sigmoid function mapping; and performing graded early warning according to the comprehensive support capability index of the coal mine and outputting an early warning signal. According to the method, dynamic coupling evaluation of multi-dimensional risks such as coal mine technology, safety, personnel and system can be realized, comprehensiveness and real-time performance of coal mine safety guarantee are improved, and closed-loop management of risk early warning and resource configuration is effectively supported.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of coal mine safety production evaluation, and particularly relates to a coal mine comprehensive guarantee capability dynamic coupling early warning method and system. BACKGROUND

[0002] As an important pillar of national energy supply, coal mines are widely used in power, metallurgy, chemical industry and other industrial fields. With the increase of mining depth and the development of production intensification, the underground environment presents complex risk characteristics of multi-factor superposition such as gas accumulation, water inrush, roof pressure, and the dynamic coupling effect between technical equipment, personnel operation, management system and other elements is increasingly significant. In related technologies, a basic framework of coal mine safety evaluation is constructed through the collaborative work of multiple systems such as equipment state monitoring, safety monitoring system, human resource management and enterprise resource planning. Specifically, the evaluation system covers the whole process from data collection, index construction, weight distribution to risk fusion, including key links such as sensor deployment, multi-source data integration, information entropy analysis and dynamic modeling.

[0003] However, in the existing coal mine comprehensive guarantee evaluation method, a single dimension (such as equipment state or accident statistics) is directly used for evaluation, and the interactive influence of human factors and management factors is not fully considered, which may lead to one-sidedness of risk identification and formation of evaluation blind area. Specifically, the traditional method lacks effective correlation modeling between technology, safety, personnel, system and other dimensions, and it is difficult to reveal the risk transmission path, thereby affecting the accuracy of risk early warning and the scientificity of resource allocation. In addition, the existing technology usually determines the index weight based on expert experience, without dynamic adjustment combined with data fluctuation characteristics and working condition changes, resulting in that the evaluation result lags behind the actual risk evolution, and cannot meet the needs of real-time, intelligent and closed-loop management of high-yield and high-efficiency mines. SUMMARY

[0004] The present application aims to at least partially solve one of the technical problems in the related art.

[0005] To this end, a first object of the present application is to propose a coal mine comprehensive guarantee capability dynamic coupling early warning method.

[0006] A second object of the present application is to propose a coal mine comprehensive guarantee capability dynamic coupling early warning system.

[0007] To achieve the above-mentioned purpose, a coal mine comprehensive guarantee capability dynamic coupling early warning method is proposed in the first aspect of the present application, comprising:

[0008] S1, a multi-dimensional fusion evaluation system including four dimensions of technical performance, safety performance, personnel management and system guarantee is constructed, and secondary indexes under each dimension are determined;

[0009] S2, real-time collection of multi-source heterogeneous data corresponding to each dimension, including collection of technical performance data through Internet of Things sensors, acquisition of safety performance data through a safety monitoring system, extraction of personnel management data from an HR system, and extraction of system guarantee data from an ERP log analysis system;

[0010] S3, using a two-stage dynamic weight distribution mechanism, combining an improved entropy weight method and a working condition perception factor to calculate the maximum weight of each index at the current time, wherein the improved entropy weight method adjusts the size of the sliding window based on the data fluctuation rate, and corrects the entropy value through correlation degree, and the working condition perception factor dynamically adjusts the weight according to the safety risk value and the equipment state value;

[0011] S4, based on the maximum weight and the real-time data of each dimension, a four-dimensional feature vector is constructed and a multi-dimensional tensor is formed, the comprehensive score of each dimension is calculated, and a coal mine comprehensive guarantee capability index is generated through weighted fusion and Sigmoid function mapping;

[0012] S5, according to the coal mine comprehensive guarantee capability index, a graded early warning output early warning signal is output to support the active prevention and control of coal mine safety risk and the optimization of resource allocation.

[0013] In an embodiment of the present application, the multi-dimensional fusion evaluation system comprising four dimensions of technical performance, safety performance, personnel management and system guarantee is constructed, and the secondary indicators under each dimension are determined, which further comprises:

[0014] S11, the technical performance dimension includes 8 secondary indicators of equipment failure frequency, equipment average failure-free time, equipment automation degree, equipment automation control accuracy, system response time, man-machine interaction friendliness, data coverage comprehensiveness, data real-time and accuracy;

[0015] S12, the safety performance dimension includes 8 secondary indicators of continuous safe production days, safety accident rate, injury incidence, pneumoconiosis detection rate, hidden danger investigation coverage rate, hidden danger rectification timeliness, equipment maintenance compliance rate, and underground gas and dust concentration compliance rate;

[0016] S13, the personnel management dimension includes 10 secondary indicators of personnel allocation adequacy, professional personnel proportion, department setting scientificity, post configuration rationality, technical training coverage rate, training examination qualification rate, employee salary and welfare satisfaction, employee working environment satisfaction, employee vacation welfare satisfaction, and employee mental health index;

[0017] S14, the system guarantee dimension includes safety responsibility system perfection degree, safety inspection system execution degree, production plan formulation rationality, production process control strictness, material procurement rationality, inventory management effectiveness, emergency rescue team construction perfection degree, emergency material reserve sufficiency, personnel recruitment and selection system scientificity, personnel performance appraisal system fairness, personnel incentive and care system effectiveness, 11 secondary indexes.

[0018] In an embodiment of the application, the real-time collection of multi-source heterogeneous data corresponding to each dimension includes collecting technical performance data through Internet of Things sensors, obtaining safety performance data from a safety monitoring system, extracting personnel management data from an HR system, and analyzing system guarantee data from an ERP log.

[0019] S21, when the system guarantee data from the ERP log is analyzed, natural language processing technology is used to extract key indicators in unstructured data such as safety responsibility system perfection degree, safety inspection system execution degree, production plan formulation rationality, production process control strictness, material procurement rationality, inventory management effectiveness, emergency rescue team construction perfection degree, emergency material reserve sufficiency, personnel recruitment and selection system scientificity, personnel performance appraisal system fairness, and personnel incentive and care system effectiveness.

[0020] S22, when the personnel management data is extracted from the HR system, a machine learning model is used to predict and analyze the mental health index of employees, and a personnel risk index is generated by combining historical training records and post adaptation degree, and the personnel adequacy rate, professional personnel proportion, department setting scientificity, post configuration rationality, technical training coverage rate, training examination qualification rate, employee salary and welfare satisfaction, employee working environment satisfaction, and employee vacation welfare satisfaction are quantified to obtain quantified index data.

[0021] In an embodiment of the application, the two-stage dynamic weight distribution mechanism is used to calculate the maximum weight of each index at the current time by combining the improved entropy weight method and the working condition perception factor.

[0022] S31, in the improved entropy weight method, the size of the sliding window is automatically adjusted according to the index fluctuation rate.

[0023] S32, the working condition perception factor includes a risk transmission intensity factor and a device state factor, wherein the risk transmission intensity factor is calculated by weighting the abnormality degree of safety indexes, and the device state factor is generated based on real-time data of device operating state and failure frequency.

[0024] In an embodiment of the present application, the four-dimensional feature vector is constructed based on the real-time data of each dimension and the maximum weight, a multi-dimensional tensor is formed, the comprehensive score of each dimension is calculated, and the coal mine comprehensive support capability index is generated through weighted fusion and Sigmoid function mapping, and the method further comprises:

[0025] S41, when the four-dimensional feature vector is constructed, a tensor decomposition method is used to extract the coupling characteristics between dimensions to identify the cross-dimensional risk transmission path;

[0026] S42, the parameters of the Sigmoid function mapping are optimized online according to the historical evaluation results and actual accident data to improve the early warning accuracy.

[0027] In an embodiment of the present application, the method further comprises:

[0028] S6, a risk evolution prediction model is generated according to the coal mine comprehensive support capability index and historical trend data, and the early warning level and emergency response strategy are dynamically adjusted based on the prediction results.

[0029] To achieve the above purpose, a coal mine comprehensive support capability dynamic coupling early warning system is provided in the second aspect of the present application, which comprises:

[0030] An evaluation system construction module is configured to construct a multi-dimensional fusion evaluation system comprising four dimensions of technical performance, safety performance, personnel management and system guarantee, and determine the secondary indicators under each dimension;

[0031] A data acquisition module is configured to acquire real-time multi-source heterogeneous data corresponding to each dimension, including acquiring technical performance data through Internet of Things sensors, acquiring safety performance data through safety monitoring systems, extracting personnel management data from HR systems, and parsing system guarantee data from ERP logs;

[0032] A weight calculation module is configured to use a two-stage dynamic weight distribution mechanism to calculate the maximum weight of each indicator at the current time by combining an improved entropy weight method and a working condition perception factor, wherein the improved entropy weight method adjusts the size of the sliding window based on the data fluctuation rate and corrects the entropy value through correlation degree, and the working condition perception factor dynamically adjusts the weight according to the safety risk value and the equipment state value;

[0033] A comprehensive index generation module is configured to construct a four-dimensional feature vector and form a multi-dimensional tensor based on the maximum weight and real-time data of each dimension, calculate the comprehensive score of each dimension, and generate a coal mine comprehensive support capability index through weighted fusion and Sigmoid function mapping;

[0034] An early warning output module is configured to perform hierarchical early warning according to the comprehensive support capability index and output an early warning signal to support proactive prevention and control of coal mine safety risks and optimization of resource allocation.

[0035] The method and system of the embodiment of the present application can realize dynamic coupling evaluation of four dimensions of technology, safety, personnel and system, comprehensively reflect real-time risk situation of comprehensive guarantee capability of coal mines, improve accuracy and timeliness of evaluation, and effectively support active prevention and control of safety risk and optimization of resource allocation.

[0036] Additional aspects and advantages of the present application will be described in the following description, will become apparent from the following description, or will be learned by practice of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0037] The above and / or additional aspects and advantages of the present application will become apparent and be readily understood from the following description, taken in conjunction with the accompanying drawings, in which:

[0038] Figure 1 A flowchart of a coal mine comprehensive guarantee capability dynamic coupling early warning method provided for an embodiment of the present application is shown in

[0039] Figure 2 A structural diagram of a coal mine comprehensive guarantee capability dynamic coupling early warning system provided for an embodiment of the present application is shown in DETAILED DESCRIPTION

[0040] It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict. The present application will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.

[0041] In order to enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should belong to the protection scope of the present application.

[0042] A coal mine comprehensive guarantee capability dynamic coupling early warning method and system according to an embodiment of the present application will be described below with reference to the accompanying drawings.

[0043] Figure 1 A flowchart of a coal mine comprehensive guarantee capability dynamic coupling early warning method according to an embodiment of the present application is shown in Figure 1 as shown, comprising:

[0044] S1, a multi-dimensional fusion evaluation system containing four dimensions of technical performance, safety performance, personnel management and system guarantee is constructed, and secondary indexes under each dimension are determined.

[0045] Specifically, the step aims to build a multi-dimensional fusion evaluation system covering four dimensions of technical performance, safety performance, personnel management, and system guarantee, and to refine a number of secondary indicators with quantifiable characteristics under each dimension, thereby providing structured and systematic data support for dynamic evaluation of the comprehensive guarantee capability of coal mines. In some implementations, the system, based on the multi-element coupling characteristics of coal mine safety production, adopts a hierarchical index design method to combine macro management dimensions with micro technical parameters, forming an evaluation framework that is comprehensive and logically clear.

[0046] At the parameter index level, each secondary indicator has a clear quantification standard and data collection frequency. For example, the "equipment failure frequency" is collected once an hour in terms of the number of failures per unit time, and the "gas concentration" is set to 0.5% CH4 according to the "Coal Mine Safety Monitoring System and Detection Instrument Use Management Specification" (AQ 1029-2019), exceeding which triggers the early warning mechanism. In addition, during the index standardization process, positive and negative indicators are classified to ensure that indicators of different natures are compared and integrated under a unified dimension.

[0047] In application scenarios, the evaluation system can be embedded in the comprehensive management system of coal mines to provide dynamic evaluation results for management and assist in developing risk control strategies and resource optimization allocation schemes. At the technical effect level, the step effectively solves the problems of dimension fragmentation and rigid weight in traditional evaluation methods by building a multi-dimensional index system, laying a solid foundation for subsequent dynamic weight allocation and multi-dimensional tensor fusion evaluation, and significantly improving the comprehensiveness and scientificity of coal mine safety guarantee capability evaluation.

[0048] Specifically, according to the comprehensive guarantee requirements and evaluation goals of coal mines, an index system for evaluating the comprehensive guarantee capability of coal mines is determined; the index system includes primary indicators and secondary indicators, the primary indicators are used to reflect the main characteristics of the comprehensive guarantee capability of coal mines, and the secondary indicators are used to specifically subdivide the primary indicators.

[0049] The primary indicators of this embodiment include four dimensions: technical performance, safety performance, personnel management, and system guarantee, and the secondary indicators include 37 sub-indicators. The primary indicators include four: technical performance indicators, safety performance indicators, personnel and organization management indicators, and system guarantee indicators.

[0050] The secondary indicators include 37, including: 8 technical performance secondary indicators, 8 safety performance secondary indicators, 10 personnel and organization management secondary indicators, and 11 system guarantee secondary indicators.

[0051] Further, S1 includes:

[0052] S11, the technical performance dimension includes eight secondary indicators of equipment failure frequency, equipment average failure-free time, equipment automation degree, equipment automation control accuracy, system response time, human-computer interaction friendliness, data coverage comprehensiveness, data real-time and accuracy.

[0053] Specifically, this step involves eight secondary indicators of the "technical performance dimension" in the coal mine comprehensive support capability evaluation system, including equipment failure frequency, equipment average failure-free time, equipment automation degree, equipment automation control accuracy, system response time, human-computer interaction friendliness, data coverage comprehensiveness, data real-time and accuracy. These indicators together constitute the basis for quantitative evaluation of the running state of the coal mine technical system and are the key input elements for multi-dimensional fusion dynamic evaluation.

[0054] In terms of technical implementation, equipment failure frequency is quantified in units of "times per thousand hours" by real-time collection of equipment running state data through mine Internet of Things sensors (such as PLC, SCADA system), combined with fault diagnosis algorithms (such as threshold-based anomaly detection or LSTM time series prediction model) to count the number of faults per unit time. Automation control accuracy is represented in the form of "percentage" by comparing the deviation rate of the set value and the actual execution value of the control system, combined with a PID control precision evaluation model, reflecting the response accuracy of the automation system.

[0055] Equipment average failure-free time (MTBF) is calculated by fusing historical running data and real-time monitoring data, fitted using an exponential distribution or Weibull distribution model, with units of "hours". System response time is calculated by collecting the time interval from the issuance of a control command to the completion of the response of the actuator, usually required to be less than 200ms to meet real-time control requirements. Equipment maintenance period is predicted based on a preventive maintenance strategy, combined with a device aging model (such as the Arrhenius model), with units of "days". Equipment redundancy is evaluated by the proportion of redundant configuration (such as dual hot standby, three out of two voting system), commonly using redundancy coefficient (RDC). Equipment update rate is calculated based on the matching degree of equipment service life and update plan, with "updated equipment number / total equipment number" as the calculation method. Equipment operating efficiency is calculated by the ratio of effective output to energy consumption per unit time, usually using industrial standard models such as COP (Coefficient of Performance) or OEE (Overall Equipment Effectiveness).

[0056] In terms of parameter setting, each indicator needs to be standardized according to the "Coal Mine Safety Regulations" (AQ 1051-2008) and "Coal Mine Mechanical and Electrical Equipment Integrity Standard" (MT / T 1001-2006) to ensure data comparability and evaluation consistency. At the same time, to meet the needs of dynamic evaluation, some indicators (such as system response time) need to set sliding time window (such as 24 hours) for real-time updating.

[0057] This step plays a role in data foundation construction and quantitative analysis in the overall evaluation system, providing key inputs for subsequent dynamic weight allocation and multi-dimensional tensor fusion, thus realizing real-time and accurate evaluation of coal mine technical support capability and improving the safety and stability of mine operation.

[0058] S12, the safety performance dimension includes 8 secondary indicators of continuous safe production days, safety accident rate, injury incidence, pneumoconiosis detection rate, hidden danger investigation coverage rate, hidden danger rectification timeliness rate, equipment maintenance compliance rate, and underground gas and dust concentration compliance rate.

[0059] Specifically, this step involves the selection and evaluation of 8 secondary indicators in the safety performance dimension, including continuous safe production days, safety accident rate, injury incidence, pneumoconiosis detection rate, hidden danger investigation coverage rate, hidden danger rectification timeliness rate, equipment maintenance compliance rate, and underground gas and dust concentration compliance rate. These indicators together form a quantitative evaluation system of coal mine safety performance, reflecting the real-time status of the mine in terms of physical environment, safety management, and emergency response.

[0060] In terms of technical implementation, this step extracts real-time data from coal mine safety monitoring systems, hidden danger management systems, training record systems, and dispatch logs through multi-source heterogeneous data fusion technology. For example, gas concentration data is collected by methane sensors (such as KJ90NB) in real time, with a sampling frequency of 1 per second and data format complying with AQ 6201-2006 "General Technical Requirements for Coal Mine Safety Monitoring Systems". Dust concentration is measured by laser dust sensors (such as PM2.5 / PM10 dual-channel equipment) with a sampling period of 5 minutes and data unit of mg / m 3 , complying with GB 50417-2019 "Technical Specification for Prevention and Control of Dust in Coal Mines". Roof pressure data is collected by hydraulic support pressure sensors or roof separation meters with a sampling frequency of 1 per minute and data unit of MPa.

[0061] At the parameter index level, each index is provided with a safety threshold and an abnormality determination standard. For example, the threshold of gas concentration is 0.8%, and exceeding this value triggers a warning; the hidden danger rectification rate requires completion within 72 hours, and a rectification rate lower than 80% is considered abnormal. The accident response time is measured by the time interval from the occurrence of the accident to the start of emergency disposal, and a response time exceeding 15 minutes is considered as response lag.

[0062] At the application scenario level, this step is applicable to real-time safety monitoring systems in coal mines and can be integrated into a comprehensive automation platform for mine to dynamically evaluate the safety status of the mine and assist the dispatch center in risk warning and emergency decision-making. By incorporating safety performance indicators into a multi-dimensional tensor fusion model, cross-dimensional risk transmission analysis can be achieved, improving the comprehensiveness and accuracy of the evaluation.

[0063] The technical value of this step lies in the scientific selection of representative safety performance indicators combined with a dynamic weight mechanism, which can effectively capture subtle changes in the safety status of the mine and provide key inputs for subsequent comprehensive support capability evaluation, thereby achieving real-time and accurate identification and early warning of safety risks in coal mines.

[0064] S2, real-time collection of multi-source heterogeneous data corresponding to each dimension, including collection of technical performance data through Internet of Things sensors, acquisition of safety performance data through safety monitoring systems, extraction of personnel management data from HR systems, and analysis of system support data from ERP logs.

[0065] Specifically, this step is "real-time collection of multi-source index data", which is a key data input link in the multi-dimensional fusion-based dynamic evaluation method for comprehensive support capability of coal mines. The technical implementation principle is to use heterogeneous data integration technology to synchronously obtain structured and unstructured data from multiple subsystems of coal mine production and operation, build a unified data collection framework, and provide real-time, accurate, and comprehensive data support for subsequent dynamic evaluation.

[0066] At the technical implementation level, this step adopts a distributed data collection architecture, combining edge computing and centralized data processing mechanisms. Technical performance data is collected in real time by mine IoT sensors (compliant with AQ 6201-2006 "General Technical Requirements for Coal Mine Safety Monitoring Systems") deployed on key equipment in the mine (such as coal mining machines, conveyors, ventilation systems, etc.). The data frequency is usually set to 1 per second to 1 per minute, with specific adjustments based on index sensitivity. Safety performance data is obtained by accessing the mine safety monitoring system (such as the KJ70X safety monitoring system), acquiring key parameters such as gas concentration, dust concentration, temperature, and pressure. The data collection period is generally seconds, ensuring rapid response to sudden risks. Personnel management data is provided by the HR system, including training coverage, job suitability, mental health index, etc. API interface or database direct connection is used for batch extraction on a daily or hourly basis. System guarantee data is extracted from ERP system logs, such as emergency material reserve adequacy and system implementation deviation rate. ETL tools are usually used for structured processing, with a log analysis frequency of once every 24 hours.

[0067] At the parameter index level, this step involves key parameters such as multi-source data collection frequency, data format standardization, and interface protocol compatibility. For example, sensor data must meet the requirements of MT / T 1004-2006 "General Technical Conditions for Coal Mine Underground Workplace Environment Parameter Monitoring System", and safety monitoring system data must comply with the monitoring standards in GB 50383-2016 "Coal Mine Safety Regulations". During data standardization, Z-score standardization or Min-Max normalization methods are used to ensure that different dimension data is processed under a unified dimension.

[0068] At the application scenario level, this step is widely used in coal mine underground production scheduling centers, safety monitoring platforms, and management decision systems, realizing real-time sensing and data integration of mine operation status. Through synchronous collection of multi-source heterogeneous data, the system can comprehensively reflect the operation status of the mine in the technical, safety, personnel, and system dimensions, providing real-time input for dynamic evaluation models.

[0069] The technical effect of this step is that by building a unified data collection interface and standardized processing process, it effectively solves the problem of scattered data sources and outdated updates in traditional evaluation methods, laying a solid data foundation for subsequent dynamic weight allocation and multi-dimensional fusion evaluation, significantly improving the real-time and accuracy of coal mine comprehensive support capability evaluation.

[0070] In one embodiment of the present application, the device failure frequency, device average failure-free time, device automation level, device automation control accuracy, system response time, human-machine interaction friendliness, data coverage comprehensiveness, and data real-time and accuracy indicators of the technical performance dimension are quantified.

[0071] Wherein, the equipment failure frequency: equipment failure frequency = (the number of equipment failures / total equipment running time) x 100%.

[0072] Wherein, the average equipment failure-free time: average equipment failure-free time (MTBF), MTBF = (total equipment running time / equipment failure number).

[0073] Wherein, the degree of equipment automation: degree of equipment automation = (the number of equipment automation functions / total number of equipment functions) x 100%.

[0074] Wherein, the accuracy of equipment automation control: the accuracy of equipment automation control = (the number of accurate control instructions / total control instruction number) x 100%.

[0075] Wherein, the system response time: system response time = (system return result time - system request receiving time).

[0076] Wherein, the friendliness of human-computer interaction: the friendliness of human-computer interaction = (user satisfaction score + interaction efficiency score) / 2.

[0077] Wherein, the comprehensiveness of data coverage: the comprehensiveness of data coverage = (the number of actual collected data types / the total number of required collected data types) x 100%.

[0078] Wherein, the real-time and accuracy of data: data real-time = (the proportion of data update frequency meeting the requirements) x 100%. Data accuracy = (the number of correct and complete data records / total record number) x 100%. Comprehensive score = (data real-time score) x 50% + (data accuracy score) x 50%.

[0079] By quantifying the equipment failure frequency, the average equipment failure-free time, the degree of equipment automation, the accuracy of equipment automation control, the system response time, the friendliness of human-computer interaction, the comprehensiveness of data coverage, and the real-time and accuracy of data of the technical performance dimension, the quantification results corresponding to each index are obtained.

[0080] In an embodiment of the present application, eight indexes of continuous safe production days, safety accident rate, injury rate, pneumoconiosis detection rate, hidden danger investigation coverage rate, hidden danger rectification timeliness rate, equipment maintenance compliance rate, and underground gas and dust concentration compliance rate of the safety performance dimension are quantified.

[0081] Wherein, the continuous safe production days: continuous safe production days = (the number of days from the current date to the last safety accident date).

[0082] Wherein, the safety accident rate: safety accident rate = (the number of safety accidents / total production activity number) x 100%.

[0083] Wherein, the incidence of industrial injury: incidence of industrial injury = (number of industrial injury accidents / total number of employees) x 100%.

[0084] Wherein, the detection rate of pneumoconiosis: detection rate of pneumoconiosis = (number of cases of pneumoconiosis detected / total number of employees) x 100%.

[0085] Wherein, the coverage rate of hidden danger investigation: coverage rate of hidden danger investigation = (number of areas or equipment already investigated / total number of areas or equipment) x 100%.

[0086] Wherein, the timely rate of hidden danger rectification: timely rate of hidden danger rectification = (number of hidden dangers that are rectified on time / total number of hidden dangers found) x 100%.

[0087] Wherein, the equipment maintenance and maintenance compliance rate: equipment maintenance and maintenance compliance rate = (actual number of maintenance and maintenance completed / planned number of maintenance and maintenance) x 100%.

[0088] Wherein, the compliance rate of underground gas and dust concentration: compliance rate of underground gas and dust concentration = (number of detection results meeting safety standards / total number of detections) x 100%.

[0089] By quantifying the continuous safety production days, safety accident rate, industrial injury rate, pneumoconiosis detection rate, hidden danger investigation coverage rate, hidden danger rectification timely rate, equipment maintenance and maintenance compliance rate, and underground gas and dust concentration compliance rate of the safety performance dimension, the quantification results corresponding to each index are obtained.

[0090] Further, S2 comprises:

[0091] S21, when the ERP log analysis system guarantee data, the key indicators in the unstructured data of safety responsibility system perfection degree, safety inspection system execution degree, production plan making rationality, production process control strictness, material procurement rationality, inventory management effectiveness, emergency rescue team construction perfection degree, emergency material reserve sufficiency, personnel recruitment and selection system scientificity, personnel performance appraisal system fairness, and personnel incentive and care system effectiveness are extracted by using natural language processing technology.

[0092] Specifically, in some implementations, when the ERP log analysis system guarantee data, the key indicators in the unstructured text, such as system execution record, emergency material reserve sufficiency, safety responsibility implementation rate, safety investment proportion, safety examination compliance rate, etc., are extracted by using natural language processing (NLP) technology, which is the core link of data collection of the system guarantee layer of the present application. This step realizes intelligent analysis and structured output of unstructured log data in the ERP system through text preprocessing, feature extraction and semantic recognition techniques.

[0093] Specifically, the technical implementation level includes: first, the ERP log is segmented and tagged. A Chinese segmentation model based on rules and statistics (such as jieba or HanLP) is used, combined with a coal industry terminology dictionary for domain adaptive optimization to improve the accuracy of key term recognition. Second, through named entity recognition (NER) technology, the system identifies institutional names, performers, execution times, material types, and quantities of other entity information in the log. Further, using dependency syntax analysis and semantic role labeling (SRL) technology, the system identifies the logical relationship between verbs such as "reserve", "implement", "invest", and "examine" and their corresponding objects in the text, thereby extracting specific institutional execution behaviors and quantitative indicators.

[0094] At the parameter index level, the key parameters involved in this step include: recall rate and precision rate of the segmentation model (usually requiring F1 value ≥ 0.85), accuracy of entity recognition (requiring ≥ 90%), and confidence threshold of semantic relationship extraction (generally set to 0.75 or higher). At the same time, for indicators such as emergency material reserve adequacy, the system needs to combine pre-set reserve standards (such as Article 135 of the Coal Mine Safety Regulations on emergency material allocation requirements) for compliance judgment.

[0095] In application scenarios, this step can be deployed in the log analysis module of the coal mine ERP system to process text data such as operation logs related to institutional execution, approval records, and material in-out warehouse logs in real time, providing data support for the 11 secondary indicators of the institutional guarantee layer. Through this step, the system can automatically identify weak links in institutional execution, assisting management personnel in optimizing and tracing the responsibility of the system.

[0096] In one embodiment of the present application, the quantification process of the safety responsibility system perfection degree, the safety inspection system execution degree, the production plan formulation rationality, the production process control strictness, the material procurement rationality, the inventory management effectiveness, the emergency rescue team construction perfection degree, the emergency material reserve adequacy, the personnel recruitment and selection system scientificity, the personnel performance evaluation system fairness, and the personnel incentive and care system effectiveness is as follows:

[0097] Among them, the safety responsibility system perfection degree: each element is scored, with a full score of 10 points. For example, the responsibility subject explicitness can be scored according to the clarity of job responsibilities, the responsibility content integrity can be scored according to the coverage, and the responsibility accountability mechanism perfection degree can be scored according to the rationality of measures. Finally, the total score is calculated to evaluate the overall perfection degree of the safety responsibility system.

[0098] The safety inspection system implementation degree: the implementation degree score can be calculated by the following formula: implementation degree score = (actual inspection times / planned inspection times) x 30% + (inspection coverage ratio) x 30% + (rectification completion rate) x 40%. Among them, the inspection coverage ratio refers to the proportion of the number of actual inspection departments or areas to the total number of planned inspections, and the rectification completion rate refers to the proportion of the number of problems rectified to the total number of problems found.

[0099] Among them, the production plan rationality: plan achievement rate = (actual completed output / plan output) x 100%. The balance of the production plan can be evaluated by calculating the production load balance index, for example, by analyzing the workload fluctuation of each production link.

[0100] Among them, the production process control strictness: the strictness comprehensive score can be calculated by the following formula: strictness comprehensive score = (quality inspection batch qualified rate) x 40% + (process parameter control precision score) x 30% - (production safety accident times) x 30%. Among them, the process parameter control precision score can be scored according to the deviation range of actual parameters and standard values.

[0101] Among them, the rationality of material procurement: procurement cost saving rate = [(budget procurement cost-actual procurement cost) / budget procurement cost] x 100%. On-time delivery rate = (on-time delivery procurement batches / total procurement batches) x 100%.

[0102] Among them, the effectiveness of inventory management: inventory turnover rate = (sales cost / average inventory amount) x 100%. Inventory accuracy rate = (inventory account balance quantity consistent with actual inventory quantity) x 100%.

[0103] Among them, the perfection degree of emergency rescue team construction: each aspect can be scored, such as personnel allocation score, training score and exercise score, and finally the comprehensive score is calculated. The personnel allocation score can be scored according to the proportion of professional rescue personnel and whether the number of personnel meets the requirements of enterprise scale and risk level; the training score can be scored according to the number of annual training times, training content coverage and training examination pass rate; the exercise score can be scored according to the number of actual combat exercises and exercise effect evaluation score.

[0104] Among them, the sufficiency of emergency material reserves: the calculation formula is: material reserve satisfaction rate = (actual reserve material quantity / demand list material quantity) x 100%. The proportion of expired materials = (expired or damaged material quantity / total reserve material quantity) x 100%. The comprehensive score can be calculated by the following formula: comprehensive score = (material reserve satisfaction rate) x 60% - (expired material proportion) x 40%.

[0105] Among them, the scientificity of personnel recruitment and selection system: the calculation formula is: the recruitment rate=(the number of recruited employees / the number of applicants) x 100%. The retention rate of recruited employees=(the number of recruited employees still in service after the probation period / the total number of recruited employees) x 100%. The qualification rate of new employees during the probation period=(the number of new employees passing the probation period examination / the total number of new employees) x 100%. The comprehensive score can be calculated by the following formula: comprehensive score=(recruitment rate) x 30%+(recruitment retention rate) x 30%+(new employee probation qualification rate) x 40%.

[0106] Among them, the fairness of personnel performance evaluation system: the calculation formula is: the matching degree score of evaluation index=(the number of evaluation indexes matched with the job responsibilities / the total number of evaluation indexes) x 100%. The transparency score of evaluation process=(the degree of openness of evaluation process+the integrity of evaluation record) / 2. The employee satisfaction score=(the number of employees satisfied with the evaluation results / the total number of employees) x 100%. The comprehensive score can be calculated by the following formula: comprehensive score=(evaluation index matching degree score) x 40%+(evaluation process transparency score) x 30%+(employee satisfaction score) x 30%.

[0107] Among them, the effectiveness of personnel incentive and care system: the calculation formula is: the diversity score of incentive measures=(the number of types of incentive methods / the total number of available incentive methods) x 100%. The employee performance improvement rate=(the number of employees with improved performance / the total number of employees) x 100%. The employee retention rate=(the number of employees still in service within a certain period / the number of employees at the beginning of the period) x 100%. The employee satisfaction score=(the number of employees satisfied with the incentive and care measures / the total number of employees) x 100%. The comprehensive score can be calculated by the following formula: comprehensive score=(incentive measure diversity score) x 30%+(employee performance improvement rate) x 30%+(employee retention rate) x 20%+(employee satisfaction score) x 20%.

[0108] By scoring each index in safety responsibility system perfection degree, safety inspection system implementation degree, production plan rationality, production process control strictness, material procurement rationality, inventory management effectiveness, emergency rescue team construction perfection degree, emergency material reserve sufficiency, personnel recruitment and selection system scientificity, personnel performance evaluation system fairness, and personnel incentive and care system effectiveness, the quantitative results of each index are finally obtained.

[0109] On the level of technical effect, this step effectively solves the problems of low efficiency and easy omission of traditional manual extraction of system data, improves the real-time and accuracy of system guarantee data, provides high-quality input for subsequent dynamic weight calculation and four-dimensional tensor fusion evaluation, and enhances the comprehensiveness and scientificity of comprehensive coal mine guarantee capability evaluation.

[0110] S22, when the HR system extracts personnel management data, a machine learning model is used to predict and analyze the mental health index of employees, and a personnel risk index is generated by combining historical training records and job adaptation degree, and the personnel staffing adequacy rate, the proportion of professional personnel, the scientificity of department setting, the rationality of post configuration, the technical training coverage rate, the training examination qualification rate, the employee salary and welfare satisfaction, the employee working environment satisfaction, and the employee vacation welfare satisfaction are quantified to obtain quantified index data.

[0111] Specifically, in the extraction process of personnel management data, the HR system uses a machine learning model to predict and analyze the mental health index of employees, and generates a personnel risk index by combining historical training records and job adaptation degree. This step is the core link of the personnel management dimension evaluation in the coal mine comprehensive support capability dynamic evaluation system, aiming to identify potential personnel risks through intelligent means and improve the foresight and accuracy of mine safety management.

[0112] In some implementations, the prediction model of the mental health index can be constructed based on a supervised learning algorithm, such as Random Forest or Long Short-Term Memory Network (LSTM), and the input features include employee emotional state questionnaire scores, attendance rate, working hours, psychological intervention records, and other structured data. The model training data set needs to cover at least 12 months of historical data, and feature engineering processing such as missing value filling, standardization, and feature encoding is performed to improve the model generalization ability. The prediction output is a mental health index of 0-100, and employees with a score below 60 are marked as potential psychological risk individuals.

[0113] Further, the system multi-dimensionally fuses the employee mental health index with historical training records (such as safety training completion rate, retraining cycle, examination results) and job adaptation degree (such as job skill matching degree, working years, accident participation records) to construct a personnel risk index. Among them, the training records adopt time series weighted average method, and the weight decays with time (such as adopting exponential decay factor α=0.95) to reflect the influence of the latest training effect on the risk. The job adaptation degree is calculated by matching the job competency model and the employee performance data, and the output is an adaptation degree score of 0-1.

[0114] Further, the personnel staffing adequacy rate, the proportion of professional personnel, the scientificity of department setting, the rationality of post configuration, the technical training coverage rate, the training examination qualification rate, the employee salary and welfare satisfaction, the employee working environment satisfaction, and the employee vacation welfare satisfaction are quantified:

[0115] Among them, the personnel staffing adequacy rate: personnel staffing adequacy rate = (actual on-duty personnel quantity / post demand personnel quantity) x 100%.

[0116] Among them, the proportion of professionals: the proportion of professionals = (the number of professionals / the total number of employees) x 100%.

[0117] Among them, the scientificity of department setting: the score of department responsibility explicitness = (the number of departments with explicit responsibilities / the total number of departments) x 100%. The score of department cooperation efficiency = (the success rate of cross-department project completion) x 100%. The score of department setting support degree = (the number of departments with smooth business processes / the total number of departments) x 100%. The comprehensive score = (the score of department responsibility explicitness) x 40% + (the score of department cooperation efficiency) x 30% + (the score of department setting support degree) x 30%.

[0118] Among them, the rationality of post configuration: the score of post responsibility matching degree = (the number of posts with matching post responsibilities and employee capabilities / the total number of posts) x 100%. The score of post workload balance degree = (the number of posts with balanced workloads / the total number of posts) x 100%. The score of post setting support degree = (the number of posts with smooth business processes / the total number of posts) x 100%. The comprehensive score = (the score of post responsibility matching degree) x 40% + (the score of post workload balance degree) x 30% + (the score of post setting support degree) x 30%.

[0119] Among them, the coverage rate of technical training: the coverage rate of technical training = (the number of employees who have received technical training / the total number of employees) x 100%.

[0120] Among them, the qualified rate of training examination: the qualified rate of training examination = (the number of employees who have passed the training examination / the number of employees who have participated in the training examination) x 100%.

[0121] Among them, the satisfaction degree of employee salary and welfare: the satisfaction degree of employee salary and welfare = (the number of employees who are satisfied with salary and welfare / the total number of employees) x 100%.

[0122] Among them, the satisfaction degree of employee work environment: the satisfaction degree of employee work environment = (the number of employees who are satisfied with the work environment / the total number of employees) x 100%.

[0123] Among them, the satisfaction degree of employee vacation benefits: the satisfaction degree of employee vacation benefits = (the number of employees who are satisfied with vacation benefits / the total number of employees) x 100%.

[0124] By quantifying the personnel adequacy rate, the proportion of professionals, the scientificity of department setting, the rationality of post configuration, the coverage rate of technical training, the qualified rate of training examination, the satisfaction degree of employee salary and welfare, the satisfaction degree of employee work environment, and the satisfaction degree of employee vacation benefits, the corresponding quantitative results for each indicator are obtained.

[0125] This step realizes the quantitative evaluation of personnel psychological state and training effect by introducing machine learning and multi-dimensional data fusion technology, provides key input for subsequent dynamic weight distribution and four-dimensional tensor fusion evaluation, significantly improves the comprehensiveness and real-time performance of coal mine comprehensive support capability evaluation, and has important engineering application value.

[0126] S3, a two-stage dynamic weight distribution mechanism is adopted, improved entropy weight method and working condition perception factor are combined, the maximum weight of each index at the current time is calculated, wherein the improved entropy weight method is based on data fluctuation rate to adaptively adjust the size of the sliding window, and the entropy value is corrected through the correlation degree, and the working condition perception factor is dynamically adjusted according to the safety risk value and the equipment state value.

[0127] Specifically, the two-stage dynamic weight distribution mechanism proposed by the application is an intelligent weight adjustment strategy based on the combination of improved entropy weight method and working condition perception factor, aiming to realize real-time and adaptive optimization of the weight of each index in coal mine comprehensive support capability evaluation. The mechanism effectively solves the problems of strong subjectivity and slow response in the weight setting of the traditional evaluation method by combining data driving and working condition perception.

[0128] In the first stage, the initial weight calculation based on the improved entropy weight method includes four sub-steps of data standardization processing, sliding window entropy value calculation, correlation degree modified entropy value and initial weight generation.

[0129] In the second stage, the working condition perception factor dynamically adjusts the weight according to the safety risk value and the equipment state value of the current mine. Among them, the risk transmission intensity factor is calculated by weighting the abnormal degree of safety indexes, and the equipment state factor is based on the real-time change of equipment operation parameters. The two factors are synthesized with a default weight of 0.5 to form a dynamic adjustment coefficient. The maximum weight is fused by the exponential smoothing method and the basic weight, and the maximum allowed transformation step (default 0.2) is set to prevent weight mutation, and the stability and continuity of the evaluation result are ensured.

[0130] This step plays a core regulation role in the whole evaluation system. Through the dynamic weight mechanism, the evaluation model can respond to the changes of the mine running state in real time, improve the timeliness and accuracy of the evaluation result, and provide a scientific basis for risk warning and resource scheduling.

[0131] The mechanism combines the improved entropy weight method and the working condition perception factor, realizes the adaptive distribution of the maximum weight of the secondary index at any time by combining data driving and environment perception, and improves the real-time performance and accuracy of the evaluation model.

[0132] Specifically, the initial weight calculation based on the improved entropy weight method of the application includes four sub-steps of data standardization processing, sliding window entropy value calculation, correlation degree modified entropy value, and initial weight generation, which are represented as follows:

[0133] Specifically, the data standardization process is as follows:

[0134] First, input: original index matrix X = [x ij ] m×n , where m is the number of time samples, n is the number of indicators, x ij represents the value of the jth indicator of the ith time sample.

[0135] Then, according to the type of indicators (positive indicators or negative indicators), the secondary indicators are standardized:

[0136] Positive indicators (indicators with larger values are better, such as equipment average failure-free time):

[0137]

[0138] Negative indicators (indicators with smaller values are better, such as equipment failure frequency):

[0139]

[0140] In the formula: x ij ′: the normalized value of the secondary indicator;

[0141] Finally, output: standardized matrix X′ = [x ij ′] m×n ,x ij ′∈[0,1].

[0142] Further, S3 includes:

[0143] S31, in the improved entropy weight method, the size of the sliding window is automatically adjusted according to the indicator volatility.

[0144] Specifically, in the dynamic evaluation method of the present application, the size of the sliding window is automatically adjusted according to the indicator volatility, aiming to improve the adaptability of weight calculation to dynamic change environment. This step introduces the indicator volatility as the basis for adjusting the size of the window, so that the entropy value calculation can more accurately reflect the uncertainty of the current data, thereby enhancing the timeliness and accuracy of the evaluation results.

[0145] Further, this step needs to be combined with timestamp information in the implementation process to ensure the continuity and integrity of the data window. The update frequency of the sliding window can be set to once an hour to adapt to the real-time requirements of the coal mine production environment. At the same time, in order to avoid waste of computing resources due to too large window size, or insufficient data representativeness due to too small window size, the value of C can be set differently according to different types of indicators, for example, the value of C can be appropriately increased for safety indicators (such as gas concentration) to enhance stability, and the value of C can be reduced for management indicators (such as training coverage) to improve response speed.

[0146] The technical scheme has important value in the coal mine comprehensive guarantee capability evaluation system. Through dynamic adjustment of the window size, the change characteristics of different indexes under different working conditions can be effectively responded, the sensitivity and adaptability of the entropy weight method to data uncertainty are improved, and a more reliable foundation is provided for subsequent dynamic weight allocation and multi-dimensional fusion evaluation. The introduction of the step solves the problem of evaluation lag of the traditional static window entropy weight method in the face of complex and variable underground environment, and enhances the real-time perception and response ability of the system to the risk situation.

[0147] Specifically, the sliding window entropy value is calculated as follows:

[0148] First, define the time window: for the current time t, select the time window size τ, then the samples in the window are i=t-τ+1, t-τ+2, …, t.

[0149] Second, the window size τ is automatically adjusted according to the volatility of the index:

[0150]

[0151] In the formula: σ j : indicates the standard deviation of index j in the recent period (such as the past 24 hours), reflecting the fluctuation of the index;

[0152] C: constant, used to adjust the length of the window size (usually 10); τ j : sliding window size of index j;

[0153] Then, calculate the probability distribution of each index in the window:

[0154]

[0155] Finally, calculate the entropy value of index j at time t:

[0156]

[0157] Further, the correlation corrected entropy value is:

[0158] First, calculate the correlation coefficient matrix R=[r jl ] n×n :

[0159]

[0160] Where: r jl is the correlation coefficient of index j and index l, is the mean of index j;

[0161] Second, calculate the correlation influence factor of each index j:

[0162]

[0163] wherein, δ jl is an indicator function:

[0164]

[0165] Finally, the entropy value is corrected:

[0166]

[0167] Further, the initial weight is generated:

[0168] First, the corrected information utility value is calculated:

[0169]

[0170] Second, the basic weight is obtained by normalization:

[0171]

[0172] S32, the working condition perception factor includes a risk transmission intensity factor and a device state factor, wherein the risk transmission intensity factor is calculated by weighting the abnormality degree of safety indicators, and the device state factor is generated based on real-time data of device operating state and failure frequency.

[0173] Specifically, in the dynamic evaluation method of the present application, the "working condition perception factor calculation" is a key link to realize dynamic weight adjustment of the multi-dimensional fusion evaluation system. Through the construction of two core factors, the risk transmission intensity factor and the device state factor, the real-time perception of the current operating state of the coal mine is realized, and the weights of each evaluation indicator are adjusted accordingly, so as to improve the timeliness and accuracy of the evaluation result.

[0174] In some implementations, the risk transmission intensity factor is calculated by weighting the abnormality degree of safety indicators. Specifically, for the 8 indicators in the safety performance layer (such as gas concentration, dust concentration, roof displacement, etc.), the system first calculates the deviation degree of each indicator from its respective safety threshold to form an abnormality degree vector.

[0175] In the second phase of the embodiment of the present application, the working condition perception factor dynamically adjusts the weight according to the current mine safety risk value and the device state value. The main dynamic weight correction includes working condition perception factor calculation, dynamic adjustment coefficient synthesis and final weight generation. By introducing the real-time working condition factor, the weight is self-adaptively adjusted according to the state of the mine. The specific steps are as follows:

[0176] First, the working condition perception factor is calculated:

[0177] Risk transmission intensity factor:

[0178]

[0179] wherein: S r : security risk value, calculated by weighting the abnormal degree of security class indicators, as follows:

[0180]

[0181] wherein: S safety is the security performance indicator set, θ j is the security threshold value of indicator j

[0182] weighting coefficient (default is 0.5);

[0183] Device status factor:

[0184]

[0185] Secondly, dynamic adjustment coefficient synthesis:

[0186] The above two factors are weighted into the dynamic adjustment coefficient:

[0187] λ t = α·λ risk + βλ device (15)

[0188] wherein: α, β are the synthesis weights (default α = 0.7, β = 0.3);

[0189] Finally, the final weight generation: using the exponential smoothing method, combined with the weight W j (t-1) of the last moment and the basic weight W j 0 of the current moment, the final weight of the current moment is calculated:

[0190] W j (t) = (1-λ t )·W j (t-1) + λ t ·W j 0 (16)

[0191] Additional weight change rate protection:

[0192] When the adjacent moment weight changes too much, start smoothing processing:

[0193] If |W j (t) -W j (t-1) |>Δ max , then let Wj (t) = W j (t-1) + sign(W j (t) - W j (t-1) ) · Δ max .

[0194] where: Δ max is the maximum allowed transformation step (default is 0.2).

[0195] S4, based on the maximum weight and real-time data of each dimension, constructs a four-dimensional feature vector and forms a multi-dimensional tensor, calculates the comprehensive score of each dimension, and generates a coal mine comprehensive support capability index through weighted fusion and Sigmoid function mapping.

[0196] Specifically, this step is the key fusion and mapping link in the coal mine comprehensive support capability dynamic evaluation method, and its technical implementation principle is based on multi-dimensional feature vector and tensor modeling, combined with dynamic weight and nonlinear function mapping, to realize the quantitative evaluation of coal mine comprehensive support capability. The specific operation mode is as follows:

[0197] In some implementations, the system first extracts the real-time data of the four first-level indicators (technical performance T, safety performance S, personnel management H, and system guarantee P) respectively, and constructs them into four-dimensional feature vectors. The feature vector of each dimension is obtained by weighted summation of multiple second-level indicators under that dimension, and the weight is the maximum weight output by the dynamic weighting engine.

[0198] The technical effect of this step is to convert multi-source heterogeneous data into unified evaluation indicators through dynamic weighting and nonlinear mapping, providing quantitative basis for subsequent hierarchical early warning and decision support. This method effectively overcomes the problem of dimension fragmentation and weight rigidity in traditional evaluation, improves the real-time and accuracy of the evaluation results, and is suitable for dynamic monitoring and risk warning of mine safety state.

[0199] Further, S4 includes:

[0200] S41, when constructing the four-dimensional feature vector, a tensor decomposition method is used to extract the coupling features between dimensions to identify the cross-dimensional risk transmission path.

[0201] Specifically, the four-dimensional feature vector is constructed using a tensor decomposition method, which aims to extract the coupling features between the four dimensions of technical performance (T), safety performance (S), personnel management (H), and system guarantee (P), thereby identifying the cross-dimensional risk transmission path. The core of this step is to construct multi-source heterogeneous real-time data into a four-dimensional tensor, and to mine the nonlinear interaction between dimensions through high-order tensor decomposition techniques such as CP decomposition or Tucker decomposition.

[0202] This step can be deployed in a real-time evaluation system of a mine in practical applications, interfaces with multi-source data acquisition modules, and realizes dynamic identification of risk transmission paths. For example, when the frequency of equipment failure (technical dimension) increases, if the coupling characteristics of personnel response delay (personnel management dimension) and the absence of emergency procedures (institutional guarantee dimension) are found at the same time, it can be determined that there is cross-dimensional risk transmission, thereby triggering the early warning mechanism.

[0203] By extracting the coupling characteristics through tensor decomposition, compared with the traditional linear weighting method, the interaction mechanism of multi-dimensional risks can be more accurately revealed, the prediction ability and response sensitivity of the evaluation system are improved, and key support is provided for the proactive prevention and control of coal mine safety.

[0204] S42, the parameters of the Sigmoid function mapping are optimized online according to historical evaluation results and actual accident data to improve the warning accuracy.

[0205] Specifically, in some implementations, the technical implementation principle of optimizing the parameters of the Sigmoid function mapping according to the historical evaluation results and the actual accident data to improve the warning accuracy is based on a nonlinear normalization and dynamic parameter adjustment mechanism. The core of this step is to update the parameters (such as the slope coefficient and the offset) of the Sigmoid function online through a real-time feedback mechanism, so as to more accurately map the four-dimensional fused comprehensive guarantee capability score to the warning score interval of 0-100, and enhance the sensitivity and discrimination ability of the system to the risk state of the coal mine. The multi-dimensional tensor fusion evaluation of the embodiments of the present application mainly includes the following sub-steps: multi-dimensional feature vector construction, calculation of the comprehensive score of each dimension, and weighted fusion of the four dimensions.

[0206] First, multi-dimensional feature vector construction:

[0207] The four primary indicators are regarded as four dimensions to construct a four-dimensional tensor:

[0208] V = [T, S, H, P] (17)

[0209] Where: T is technical performance, S is safety performance, H is personnel management, and P is institutional guarantee

[0210] Then, calculate the comprehensive score of each dimension (weighted sum of the secondary indicators in each dimension):

[0211] Technical performance score:

[0212] T = ∑ j∈T W j (t) ·x tj ′ (18)

[0213] Similarly, the safety performance score S, the personnel management score H, and the system guarantee score P are calculated.

[0214] Finally, the four dimensions are weighted and fused:

[0215] Q = ω T · σ(T) + ω S · σ(S) + ω H · σ(H) + ω P · σ(P) (19)

[0216] where: ω T = 0.3, ω S = 0.4, ω H = 0.2, ω P = 0.1, weighting coefficients (adjustable).

[0217] σ(S·): Sigmoid function, used to normalize the score to 0-1, and then mapped to 0-100 points.

[0218] S5, according to the comprehensive guarantee capability index, a graded warning is given, and a warning signal is output to support the proactive prevention and control of coal mine safety risks and the optimization of resource allocation.

[0219] Specifically, this step is a grading warning mechanism in the coal mine comprehensive guarantee capability dynamic evaluation method based on multi-dimensional fusion. The technical implementation principle is based on the quantitative results of the comprehensive guarantee capability index. By setting a grading threshold, the evaluation results are mapped to green, yellow, orange, or red warning signals, thereby achieving proactive identification and response to coal mine safety risks. In some implementations, the Sigmoid function is used to normalize the comprehensive score after four-dimensional fusion, mapping it to the standardized interval of 0-100 points, making it easier for subsequent warning grading.

[0220] The specific operation mode is as follows: First, the system calculates the comprehensive guarantee capability index at the current time through the four-dimensional fusion evaluation module. This index is composed of the weighted scores of the technical performance, safety performance, personnel management, and system guarantee. The weights of each dimension are adjusted in real time by the dynamic weighting engine according to the improved entropy weight method and the working condition perception factor. Subsequently, the system inputs the index into the Sigmoid function for nonlinear mapping, with the formula being S = 1 / (1 + e^(- (S - S0) / ω)), where S is the fused original score, S0 is the adjustment coefficient, and ω is usually taken as 0.05-0.1 to control the steepness of the output curve. The mapping result is a standardized comprehensive guarantee capability score of 0-100 points.

[0221] According to the industry safety standards and historical accident data statistics, the system sets four levels of early warning thresholds: a score of 90 or more is green early warning, indicating excellent protection ability and extremely low risk; 75-89 is yellow early warning, indicating that there is potential risk and monitoring needs to be strengthened; 60-74 is orange early warning, indicating that the protection ability is declining and emergency response needs to be started; and a score of less than 60 is red early warning, indicating that the system is in a high-risk state and intervention measures should be taken immediately. The multi-dimensional tensor fusion evaluation of the embodiment of the application also includes hierarchical early warning.

[0222] Illustratively, hierarchical early warning is graded according to the comprehensive score:

[0223] Q>85: green;

[0224] 70Q<85: yellow;

[0225] 60Q<70: orange;

[0226] Q<60: red.

[0227] This step is deployed in the coal mine safety monitoring center in actual application, and combines real-time data flow and historical evaluation model to realize dynamic perception of the overall safety situation of the mine. Its technical value lies in providing intuitive risk level identification for safety management personnel through quantitative evaluation and hierarchical early warning, supporting resource allocation optimization and risk pre-control strategy formulation, and effectively improving the intelligent and forward-looking level of coal mine safety production.

[0228] The coal mine comprehensive protection ability dynamic coupling early warning method of the embodiment of the application can realize multi-dimensional dynamic evaluation of the comprehensive protection ability of the coal mine, effectively improve the real-time and accuracy of risk early warning, and support active prevention and control of safety incidents and optimization of management decision-making.

[0229] To realize the above-mentioned embodiments, as Figure 2 shown, the coal mine comprehensive protection ability dynamic coupling early warning system 10 is also provided in the embodiment, which comprises:

[0230] The evaluation system construction module 100 is used to construct a multi-dimensional fusion evaluation system comprising four dimensions of technical performance, safety performance, personnel management and system protection, and to determine the secondary indicators under each dimension;

[0231] The data acquisition module 200 is used to acquire multi-source heterogeneous data corresponding to each dimension in real time, including acquiring technical performance data through Internet of Things sensors, acquiring safety performance data through safety monitoring systems, extracting personnel management data through HR systems, and analyzing system protection data through ERP logs;

[0232] The weight calculation module 300 is used to calculate the maximum weight of each index at the current time by adopting a two-stage dynamic weight distribution mechanism, combining an improved entropy weight method and a working condition perception factor, wherein the improved entropy weight method is based on adaptive adjustment of a sliding window size based on data fluctuation rate, and the entropy value is corrected through correlation degree, and the working condition perception factor is dynamically adjusted according to the safety risk value and the equipment state value.

[0233] The comprehensive index generation module 400 is used to construct a four-dimensional feature vector and form a multi-dimensional tensor based on the maximum weight and real-time data of each dimension, calculate the comprehensive score of each dimension, and generate a coal mine comprehensive support capability index through weighted fusion and Sigmoid function mapping.

[0234] The early warning output module 500 is used to perform hierarchical early warning according to the comprehensive support capability index, and output an early warning signal to support active prevention and control of coal mine safety risks and optimization of resource allocation.

[0235] Further, the evaluation system construction module is also used to:

[0236] The technical performance dimension includes 8 secondary indexes of device failure frequency, device average failure-free time, device automation degree, device automation control accuracy, system response time, man-machine interaction friendliness, data coverage comprehensiveness, data real-time and accuracy.

[0237] The safety performance dimension includes 8 secondary indexes of continuous safe production days, safety accident rate, work injury rate, pneumoconiosis detection rate, hidden danger investigation coverage rate, hidden danger rectification timeliness, equipment maintenance compliance rate, and underground gas and dust concentration compliance rate.

[0238] The personnel management dimension includes 10 secondary indexes of personnel allocation adequacy, professional personnel proportion, department setting scientificity, post configuration rationality, technical training coverage rate, training examination qualification rate, employee salary and welfare satisfaction, employee working environment satisfaction, employee vacation welfare satisfaction, and employee mental health index.

[0239] The system guarantee dimension includes 11 secondary indexes of safety responsibility system perfection, safety inspection system implementation, production plan rationality, production process control strictness, material procurement rationality, inventory management effectiveness, emergency rescue team construction perfection, emergency material reserve adequacy, personnel recruitment and selection system scientificity, personnel performance appraisal system fairness, and personnel incentive and care system effectiveness.

[0240] Further, the data acquisition module is also used to:

[0241] The ERP log analysis system extracts key indicators in unstructured data such as safety responsibility system perfection, safety inspection system execution, production plan rationality, production process control strictness, material procurement rationality, inventory management effectiveness, emergency rescue team construction perfection, emergency material reserve sufficiency, personnel recruitment and selection system scientificity, personnel performance evaluation system fairness, and personnel incentive and care system effectiveness when guaranteeing data;

[0242] The HR system extracts personnel management data, uses a machine learning model to predict and analyze the mental health index of employees, combines historical training records and post adaptation degree to generate a personnel risk index, and quantifies personnel adequacy rate, professional personnel proportion, department setting scientificity, post configuration rationality, technical training coverage rate, training examination qualification rate, employee salary and welfare satisfaction, employee working environment satisfaction, and employee vacation welfare satisfaction to obtain quantified index data.

[0243] Further, the weight calculation module is also used for:

[0244] In the improved entropy weight method, the size of the sliding window is automatically adjusted according to the index fluctuation rate.

[0245] The working condition perception factor includes a risk transmission intensity factor and a device state factor, wherein the risk transmission intensity factor is calculated by weighting the abnormal degree of safety indexes, and the device state factor is generated based on real-time data of device operating state and failure frequency.

[0246] Further, the comprehensive index generation module is also used for:

[0247] When constructing the four-dimensional feature vector, a tensor decomposition method is used to extract the coupling features between dimensions to identify cross-dimensional risk transmission paths.

[0248] The parameters of the Sigmoid function mapping are optimized online based on historical evaluation results and actual accident data to improve the accuracy of early warning.

[0249] Further, it also includes:

[0250] A prediction adjustment module is used to generate a risk evolution prediction model based on the coal mine comprehensive guarantee capability index and historical trend data, and dynamically adjust the early warning level and emergency response strategy based on the prediction result.

[0251] The coal mine comprehensive guarantee capability dynamic coupling early warning system of the embodiment of the present application can realize multi-dimensional dynamic evaluation of the comprehensive guarantee capability of coal mines, effectively improve the real-time performance and accuracy of risk early warning, and support active prevention and control of safety incidents and optimization of management decision-making.

[0252] In the description of the specification, the description of the terms "one embodiment", "some embodiments", "an example", "a specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In the description of the specification, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any appropriate manner in any one or more embodiments or examples. In addition, the person skilled in the art can combine and combine the different embodiments or examples described in the specification and the features of the different embodiments or examples, without contradiction.

[0253] In addition, the terms "first", "second" are only for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the technical features indicated. Therefore, the features defined with "first", "second" can explicitly or implicitly include at least one of the features. In the description of the present application, the meaning of "a plurality of" is at least two, for example, two, three, etc., unless otherwise specifically limited.

Claims

1. A coal mine comprehensive support capability dynamic coupling early warning method, characterized in that, The method comprises the following steps: S1, constructing a multi-dimensional fusion evaluation system comprising technical performance, safety performance, personnel management and system guarantee four dimensions, and determining the secondary indicators under each dimension; S2, collecting real-time multi-source heterogeneous data corresponding to each dimension, including collecting technical performance data through Internet of Things sensors, obtaining safety performance data through safety monitoring systems, extracting personnel management data from HR systems, and analyzing system guarantee data from ERP logs; S3, using a two-stage dynamic weight distribution mechanism, combining improved entropy weight method and working condition perception factor to calculate the maximum weight of each indicator at the current time, wherein the improved entropy weight method adjusts the sliding window size based on the data fluctuation rate, and corrects the entropy value through correlation degree, and the working condition perception factor dynamically adjusts the weight according to the safety risk value and the equipment state value; S4, based on the maximum weight and real-time data of each dimension, constructing a four-dimensional feature vector and forming a multi-dimensional tensor, calculating the comprehensive score of each dimension, and generating a coal mine comprehensive guarantee capability index through weighted fusion and Sigmoid function mapping; S5, according to the coal mine comprehensive guarantee capability index, the grading early warning output early warning signal, in order to support the active prevention and control of coal mine safety risk and the optimization of resource allocation.

2. The method of claim 1, wherein, The method for constructing a multi-dimensional fusion evaluation system comprising technical performance, safety performance, personnel management and system guarantee four dimensions, and determining the secondary indicators under each dimension, further comprises: S11, the technical performance dimension includes 8 secondary indicators of equipment failure frequency, equipment average failure-free time, equipment automation degree, equipment automation control accuracy, system response time, man-machine interaction friendliness, data coverage comprehensiveness, data real-time and accuracy; S12, the safety performance dimension includes 8 secondary indicators of continuous safe production days, safety accident rate, injury rate, pneumoconiosis detection rate, hidden danger investigation coverage rate, hidden danger rectification timeliness, equipment maintenance compliance rate, and underground gas and dust concentration compliance rate; S13, the personnel management dimension includes 10 secondary indicators of personnel allocation adequacy, professional personnel proportion, department setting scientificity, post configuration rationality, technical training coverage rate, training examination qualification rate, employee salary and welfare satisfaction, employee working environment satisfaction, employee vacation welfare satisfaction, and employee mental health index; S14, the system guarantee dimension includes 11 secondary indicators of safety responsibility system perfection, safety inspection system implementation, production plan rationality, production process control strictness, material procurement rationality, inventory management effectiveness, emergency rescue team construction perfection, emergency material reserve adequacy, personnel recruitment and selection system scientificity, personnel performance appraisal system fairness, and personnel incentive and care system effectiveness.

3. The method of claim 1, wherein, The method for collecting real-time multi-source heterogeneous data corresponding to each dimension, including collecting technical performance data through Internet of Things sensors, obtaining safety performance data through safety monitoring systems, extracting personnel management data from HR systems, and analyzing system guarantee data from ERP logs, further comprises: S21, when the ERP log analysis system guarantees data, the natural language processing technology is adopted to extract the key indexes in the unstructured data of safety responsibility system perfection degree, safety inspection system execution degree, production plan making rationality, production process control strictness, material procurement rationality, inventory management effectiveness, emergency rescue team construction perfection degree, emergency material reserve sufficiency, personnel recruitment and selection system scientificity, personnel performance examination system fairness, and personnel incentive and care system effectiveness; S22, when the HR system extracts personnel management data, a machine learning model is used to predict and analyze the employee mental health index, and a personnel risk index is generated in combination with historical training records and post adaptation degree, and the personnel allocation sufficiency rate, professional personnel proportion, department setting scientificity, post configuration rationality, technical training coverage rate, training examination qualification rate, employee salary and welfare satisfaction, employee working environment satisfaction, and employee vacation welfare satisfaction are quantified to obtain quantified index data.

4. The method of claim 1, wherein, The two-stage dynamic weight distribution mechanism is adopted, the improved entropy weight method and the working condition perception factor are combined, the maximum weight of each index at the current time is calculated, and the maximum weight further includes: S31, in the improved entropy weight method, the size of the sliding window is automatically adjusted according to the index fluctuation rate; S32, the working condition perception factor includes a risk transmission intensity factor and a device state factor, wherein the risk transmission intensity factor is weighted calculated according to the abnormal degree of safety indexes, and the device state factor is generated based on real-time data of device running state and fault frequency.

5. The method of claim 1, wherein, Based on the maximum weight and real-time data of each dimension, a four-dimensional feature vector is constructed to form a multi-dimensional tensor, the comprehensive score of each dimension is calculated, and a coal mine comprehensive support capability index is generated through weighted fusion and Sigmoid function mapping, and the coal mine comprehensive support capability index further includes: S41, when the four-dimensional feature vector is constructed, a tensor decomposition method is used to extract the coupling characteristics between dimensions to identify the cross-dimensional risk transmission path; S42, the parameters of the Sigmoid function mapping are optimized online according to historical evaluation results and actual accident data to improve the warning accuracy.

6. The method of claim 1, wherein, Further includes: S6, according to the coal mine comprehensive support capability index and historical trend data, a risk evolution prediction model is generated, and the warning level and emergency response strategy are dynamically adjusted based on the prediction result.

7. A coal mine comprehensive support capability dynamic coupling early warning system, characterized in that, It includes: An evaluation system construction module is used to construct a multi-dimensional fusion evaluation system including four dimensions of technical performance, safety performance, personnel management and system guarantee, and to determine the secondary indexes under each dimension; A data acquisition module is used to acquire multi-source heterogeneous data corresponding to each dimension in real time, including acquiring technical performance data through Internet of Things sensors, acquiring safety performance data through safety monitoring systems, extracting personnel management data from HR systems, and extracting system guarantee data from ERP log analysis systems; The weight calculation module is configured to calculate the maximum weight of each index at the current time by using a two-stage dynamic weight allocation mechanism, combining an improved entropy weight method and a working condition perception factor, wherein the improved entropy weight method is based on adaptive adjustment of a sliding window size based on data volatility, and the entropy value is corrected by correlation degree, and the working condition perception factor dynamically adjusts the weight according to the safety risk value and the equipment state value; The comprehensive index generation module is configured to construct a four-dimensional feature vector and form a multi-dimensional tensor based on the maximum weight and real-time data of each dimension, calculate the comprehensive score of each dimension, and generate a coal mine comprehensive support capability index by weighted fusion and Sigmoid function mapping; The early warning output module is configured to perform hierarchical early warning according to the comprehensive support capability index and output an early warning signal to support active prevention and control of coal mine safety risks and optimization of resource allocation.

8. The system of claim 7, wherein, The evaluation system construction module is further configured to: The technical performance dimension includes eight secondary indexes of device failure frequency, device average failure-free time, device automation degree, device automation control accuracy, system response time, man-machine interaction friendliness, data coverage comprehensiveness, data real-time and accuracy; The safety performance dimension includes eight secondary indexes of continuous safe production days, safety accident rate, injury incidence, pneumoconiosis detection rate, hidden danger investigation coverage rate, hidden danger rectification timeliness, equipment maintenance compliance rate, and underground gas and dust concentration compliance rate; The personnel management dimension includes ten secondary indexes of personnel allocation adequacy, professional personnel proportion, department setting scientificity, post configuration rationality, technical training coverage rate, training examination qualification rate, employee salary and welfare satisfaction, employee working environment satisfaction, employee vacation welfare satisfaction, and employee mental health index; The system guarantee dimension includes eleven secondary indexes of safety responsibility system perfection, safety inspection system execution, production plan formulation rationality, production process control strictness, material procurement rationality, inventory management effectiveness, emergency rescue team construction perfection, emergency material reserve adequacy, personnel recruitment and selection system scientificity, personnel performance appraisal system fairness, and personnel incentive and care system effectiveness.

9. The system of claim 7, wherein, The data collection module is further configured to: When the ERP log analysis system guarantee data is analyzed, the natural language processing technology is used to extract key indicators in unstructured data of safety responsibility system perfection, safety inspection system execution, production plan formulation rationality, production process control strictness, material procurement rationality, inventory management effectiveness, emergency rescue team construction perfection, emergency material reserve adequacy, personnel recruitment and selection system scientificity, personnel performance appraisal system fairness, and personnel incentive and care system effectiveness. When the HR system extracts personnel management data, a machine learning model is used to predict and analyze the mental health index of employees, and a personnel risk index is generated in combination with historical training records and job suitability. The personnel adequacy rate, the proportion of professionals, the scientific nature of department setting, the rationality of job allocation, the technical training coverage rate, the training examination qualification rate, the employee salary and welfare satisfaction, the employee work environment satisfaction, and the employee vacation welfare satisfaction are quantified to obtain quantified index data.

10. The system of claim 7, wherein, The weight calculation module is further configured to: In the improved entropy weight method, the size of the sliding window is automatically adjusted according to the index fluctuation rate; The working condition perception factor includes a risk transmission intensity factor and a device state factor, wherein the risk transmission intensity factor is calculated by weighting the abnormal degree of safety indicators, and the device state factor is generated based on real-time data of device operating state and failure frequency.

11. The system of claim 7, wherein, The comprehensive index generation module is further configured to: When constructing the four-dimensional feature vector, a tensor decomposition method is used to extract the coupling features between dimensions to identify cross-dimensional risk transmission paths; The parameters of the Sigmoid function mapping are optimized online according to historical evaluation results and actual accident data to improve the accuracy of early warning.

12. The system of claim 7, wherein, Further comprising: A prediction adjustment module is configured to generate a risk evolution prediction model based on the coal mine comprehensive support capability index and historical trend data, and dynamically adjust the early warning level and emergency response strategy based on the prediction results.

Images