Work environment safety analysis and evaluation method and device, computer equipment and medium

Abstract

This invention relates to the field of knowledge graph technology, and discloses a method, apparatus, computer equipment, and medium for workplace safety analysis and evaluation. The method includes: establishing a hierarchical knowledge base of workplace elements and establishing entity relationships between its levels to generate an initial knowledge graph; comparing pre-acquired indicator data of workplace elements with the hierarchical knowledge base to identify risk indicators, simultaneously calculating the risk values ​​of the risk indicators, and constructing a safety risk analysis and evaluation set based on the risk indicators and their corresponding risk values; constructing a risk knowledge graph for analyzing and evaluating workplace safety based on the safety risk analysis and evaluation set and the initial knowledge graph, and performing workplace safety analysis and evaluation based on the risk knowledge graph. This invention can be effectively applied to the safety analysis and evaluation of workplace environments in different industries, providing a scientific method and effective means to ensure workplace safety.

Description

Technical Field

[0001] This invention relates to the field of knowledge graph technology, specifically to methods, apparatus, computer equipment, and media for workplace safety analysis and evaluation. Background Technology

[0002] The work environment is a crucial factor affecting safe production. Work environment elements include air quality, lighting, temperature, humidity, noise, vibration, dust, radiation, and toxic or hazardous substances. Currently, the analysis and evaluation of work environment factors are incomplete and inaccurate, control standards are scattered or even missing, and safety control measures lack systematicity and specificity. Production accidents caused by unsafe work environments occur frequently. Therefore, employing knowledge graph technology to establish scientific methods and means to comprehensively and scientifically evaluate work environment safety risks, ensuring that production operations comply with laws and regulations, is of great significance for improving the level of safe production.

[0003] Using knowledge graph technology for workplace safety analysis and evaluation faces technical challenges such as rigid knowledge graph architecture and inaccurate risk indicator assessment. Summary of the Invention

[0004] In view of this, the present invention provides a method, apparatus, computer equipment and medium for workplace safety analysis and evaluation, in order to solve the technical problems of rigid knowledge graph architecture and inaccurate risk indicator judgment in workplace scenarios, and to provide an effective method and means for conducting workplace safety analysis and evaluation.

[0005] In a first aspect, the present invention provides a method for analyzing and evaluating workplace safety, the method comprising:

[0006] Establish a hierarchical knowledge base for operational environment elements, and establish entity relationships between its levels to generate an initial knowledge graph;

[0007] Risk indicators are identified by comparing the pre-acquired indicator data of the work environment elements with the hierarchical knowledge base of the work environment elements. At the same time, the risk values ​​of the risk indicators are calculated, and a safety risk analysis and evaluation set is constructed based on the risk indicators and their corresponding risk values.

[0008] Based on the safety risk analysis and evaluation set and the initial knowledge graph, a risk knowledge graph is constructed to analyze and evaluate the safety of the work environment, and the work environment is analyzed and evaluated based on the risk knowledge graph.

[0009] This invention provides a method for safety analysis and evaluation of the work environment. By constructing a multi-level knowledge base and integrating multi-source data, it accurately identifies risk indicators by comparing the indicator data of pre-acquired work environment elements with the hierarchical knowledge base, and calculates the risk values ​​of the risk indicators. Finally, it forms a risk knowledge graph that can integrate information from multiple aspects. This method can be effectively applied to the safety evaluation of work environments in different industries, improving the accuracy and universality of safety evaluation, and solving the problems of rigid knowledge graph architecture and inaccurate risk indicator judgment in work environment scenarios.

[0010] In one optional implementation, a hierarchical knowledge base of operational environment elements is established, and entity relationships are established between its levels to generate an initial knowledge graph, including:

[0011] Obtain indicator data, work standards, influencing factors, and case studies of work accidents related to the work environment.

[0012] The system is structured as follows: an indicator data layer based on indicator data of work environment elements; a work standard layer based on work standards of work environment elements; an impact analysis layer based on factors influencing work environment elements; and an accident case layer based on work accident cases.

[0013] The data and entity relationship network in the indicator data layer, operation standard layer, impact analysis layer and accident case layer are filtered, and the data is stored in a relational database and the entity relationship network is stored in a graph database. A hierarchical knowledge base is generated based on the stored relational database and the stored graph database.

[0014] The semantic similarity of entities across levels in the hierarchical knowledge base is calculated using a pre-trained model, and entities with semantic similarity exceeding a preset threshold are associated across levels to generate an initial knowledge graph.

[0015] This invention provides a method for workplace safety analysis and evaluation. By clearly defining the functions and storage content of different levels, it systematically classifies and structures heterogeneous data from multiple sources. A pre-trained model calculates the semantic similarity between entities at each level, associating entities with similarity greater than a similarity threshold. This optimizes the logical hierarchy of knowledge representation, significantly improving the accuracy and response speed of data retrieval. It enables a more comprehensive and accurate safety evaluation of the workplace environment, effectively addressing the problem of weak semantic connections between levels. The interconnected levels form a comprehensive and systematic hierarchical knowledge base, providing rich and accurate data support for workplace safety analysis and evaluation, and improving data utilization.

[0016] In one optional implementation, the indicator data of the work environment elements includes real-time data of the work environment elements, which is obtained in the following manner:

[0017] In the scenario of monitoring the work environment, a sensor network is deployed in a distributed manner, and real-time data of work environment elements are collected from multiple sensors in the sensor network. The real-time data of work environment elements includes timestamps, sensor measurements, and measurement locations.

[0018] Using a preset reference time start point or the initial acquisition time of the first sensor as a benchmark, the timestamps of the other sensors are offset and compensated to generate real-time data of the working environment elements under a unified time benchmark.

[0019] If there are missing values ​​in the real-time data of the work environment elements, supplementary values ​​are calculated by bidirectional linear interpolation based on the data points before and after the missing location to generate complete real-time data of the work environment elements.

[0020] This invention provides a method for analyzing and evaluating workplace safety, achieving accurate temporal alignment and fusion of operational data, and supplementing missing values ​​in real-time data of workplace elements through bidirectional linear interpolation. This enables workplace data to more accurately reflect the actual situation of the workplace, effectively solving the problems of incomplete data and inconsistent timing, and improving data quality and usability.

[0021] In one alternative implementation, the operational standards layer includes a safety range; the operational environment index data also includes historical data on operational environment elements.

[0022] Risk indicators are identified by comparing pre-acquired indicator data of operational environment elements with a hierarchical knowledge base. Simultaneously, the risk values ​​of these indicators are calculated, and a safety risk analysis and evaluation set is constructed based on the risk indicators and their corresponding risk values, including:

[0023] By comparing the real-time data of the work environment elements with the safety range of the work standard layer, the real-time data of the work environment elements that are not within the safety range are identified as risk indicators.

[0024] Based on historical data of work environment elements, an attention mechanism is used to assign two-dimensional weights to risk indicators, and a weighted calculation method is used to calculate the risk value corresponding to the risk indicator based on the two-dimensional weights.

[0025] Establish entity associations between risk indicators and their corresponding risk values, and construct a safety risk analysis and evaluation set based on these entity associations. Store the safety risk analysis and evaluation set in the indicator data layer.

[0026] The present invention provides a method for analyzing and evaluating the safety of the work environment, which can more accurately identify the risk indicators of the work environment, effectively solves the problems of difficulty in dynamically allocating weights and accurately identifying the risk indicators of the work environment, and improves the accuracy and reliability of safety evaluation.

[0027] In one optional implementation, the two-dimensional weighting includes historical data attention weighting and real-time data attention weighting of the work environment elements.

[0028] Based on historical data of work environment elements, an attention mechanism is used to assign two-dimensional weights to risk indicators, and a weighted calculation method is used to calculate the risk value corresponding to the risk indicator based on the two-dimensional weights, including:

[0029] The attention weights of historical data for work environment elements are calculated based on a pre-trained gated recurrent unit model.

[0030] Calculate the degree of anomaly and rate of change of risk indicators, calculate the real-time emergency score based on the degree of anomaly and rate of change of risk indicators, and calculate the real-time data attention weight of the work environment elements based on the real-time emergency score.

[0031] The comprehensive attention weight of the risk indicator is calculated based on the attention weight of historical data and the attention weight of real-time data of work environment elements.

[0032] The risk value corresponding to the risk indicator is calculated using a weighted calculation method based on the comprehensive attention weight of the risk indicator.

[0033] This invention provides a method for workplace safety analysis and evaluation. It utilizes recurrent neural networks to mine the temporal dependencies of historical data, capturing the long-term evolution patterns of risk indicators. This avoids the limitations of relying solely on real-time data. A real-time emergency scoring mechanism combines the anomaly degree and rate of change of risk indicators to dynamically reflect the immediate urgency of workplace risks. The dual-dimensional weight fusion ensures that risk value calculation includes both "long-term risk accumulation" and "short-term sudden risks," more closely reflecting the complexity of actual workplace environments. An attention mechanism automatically identifies the most critical features in historical data that impact risk, avoiding the neglect of key information by traditional equal-weighted average or fixed-weight models, thus providing decision support for enterprise safety production management.

[0034] In one optional implementation, a risk knowledge graph for analyzing and evaluating workplace safety is constructed based on a safety risk analysis and evaluation set and an initial knowledge graph, including:

[0035] Semantically match each risk indicator in the security risk analysis and evaluation set with the entity nodes in the indicator data layer, and establish the association between the risk indicator and the entity nodes in its respective layer.

[0036] The visualization attributes of risk indicators are adjusted according to the magnitude of the risk value. The visualization attributes include color and size.

[0037] Based on the pre-trained model, the semantic similarity between risk indicators and entity nodes in the operation standard layer, impact analysis layer, and accident case layer is calculated, and cross-level association is established based on the relationship between semantic similarity and preset thresholds.

[0038] A risk knowledge graph for evaluating workplace safety is constructed based on cross-level relationships.

[0039] This invention provides a method for workplace safety analysis and evaluation. It associates real-time risk indicators from a safety risk analysis and evaluation dataset with different types of data in an initial knowledge graph, including indicator data layers, work standard layers, impact analysis layers, and accident case layers. This association extends beyond historical monitoring data in the indicator data layer; it also links to occupational exposure limits in the work standard layer, poisoning hazards in the impact analysis layer, and similar accident cases in the accident case layer, forming a complete knowledge chain. By establishing semantic relationships between entities, fragmented safety data is transformed into structured knowledge. Presented in a graph structure, it clearly displays the source, impact, and countermeasures of risk indicators, facilitating understanding and analysis. The visualization attributes of risk indicators are adjusted according to their risk values. This intuitive visualization allows safety managers to quickly locate high-risk areas and key risk indicators, enabling timely action. Utilizing a pre-trained model to calculate semantic similarity and establish cross-level relationships reveals implicit connections between risk indicators and other knowledge entities.

[0040] In one optional implementation, the semantic similarity between risk indicators and entity nodes in the operation standard layer, impact analysis layer, and accident case layer is calculated based on a pre-trained model, and cross-level associations are established based on the relationship between semantic similarity and preset thresholds, including:

[0041] The semantic similarity between risk indicators and entities in the operation standard layer is calculated using a pre-trained model to establish compliance or violation relationships.

[0042] Based on the semantic matching of risk indicators, the potential consequences described in the impact analysis layer are formed to form a risk propagation path and construct a causal relationship between risk and consequences.

[0043] The current risk indicators are matched with historical accident cases at the accident case layer in multiple dimensions. When the matching degree exceeds the preset threshold, a similarity relationship is established and historical handling measures are associated to form risk prediction and handling suggestions. The multi-dimensional matching includes indicator type similarity matching, numerical range similarity matching, and time pattern similarity matching.

[0044] This invention provides a method for workplace safety analysis and evaluation. By using a pre-trained model to calculate the semantic similarity between risk indicators and entities at the operational standards layer, it can accurately identify the compliance or violation relationships between risk indicators and regulatory standards. Semantic matching is used to uncover potential connections between different levels of knowledge, organically integrating scattered regulatory, impact analysis, and accident case knowledge. Based on semantic matching, risk propagation paths are formed, constructing causal relationships between risks and consequences, and enabling a systematic analysis of risk impacts. By matching current risk indicators with historical cases at the accident case layer in multiple dimensions such as indicator type, numerical range, and time pattern, it is possible to predict risk development trends based on historical experience.

[0045] Secondly, the present invention provides a work environment safety analysis and evaluation device, the device comprising:

[0046] The initialization module is used to establish a hierarchical knowledge base for the operating environment, establish entity relationships between its levels, and generate an initial knowledge graph.

[0047] The attention module is used to compare pre-acquired operational environment indicator data with a hierarchical knowledge base to identify risk indicators, calculate the risk value of the risk indicators, and construct a safety risk analysis and evaluation set based on the risk indicators and their corresponding risk values.

[0048] The safety evaluation module is used to construct a risk knowledge graph for analyzing and evaluating the safety of the work environment based on the safety risk analysis and evaluation set and the initial knowledge graph, and to perform safety analysis and evaluation of the work environment based on the risk knowledge graph.

[0049] Thirdly, the present invention provides a computer device, including: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the computer instructions to perform the work environment safety analysis and evaluation method of the first aspect or any corresponding embodiment described above.

[0050] Fourthly, the present invention provides a computer-readable storage medium storing computer instructions for causing a computer to execute the work environment safety analysis and evaluation method of the first aspect or any corresponding embodiment described above.

[0051] Fifthly, the present invention provides a computer program product, including computer instructions, which are used to cause a computer to execute the working environment safety analysis and evaluation method described in the first aspect or any corresponding embodiment thereof. Attached Figure Description

[0052] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0053] Figure 1 This is a flowchart illustrating the work environment safety analysis and evaluation method according to an embodiment of the present invention;

[0054] Figure 2 This is a flowchart illustrating another method for analyzing and evaluating workplace safety according to an embodiment of the present invention;

[0055] Figure 3 This is a flowchart illustrating another method for analyzing and evaluating work environment safety according to an embodiment of the present invention;

[0056] Figure 4 This is a flowchart illustrating a method for analyzing and evaluating the safety of a work environment according to an embodiment of the present invention.

[0057] Figure 5 This is a flowchart illustrating another method for analyzing and evaluating workplace safety according to an embodiment of the present invention;

[0058] Figure 6 This is a schematic diagram of a risk knowledge graph for evaluating workplace safety according to an embodiment of the present invention;

[0059] Figure 7 This is a structural block diagram of the work environment safety analysis and evaluation device according to an embodiment of the present invention;

[0060] Figure 8 This is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. Detailed Implementation

[0061] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0062] The field of work safety involves multi-source, heterogeneous data, including monitoring data, hazard rectification information, laws and regulations, and accident case studies. This data encompasses structured, semi-structured, and unstructured types. Due to the lack of effective data integration mechanisms, low data utilization is a common problem in actual production processes.

[0063] Currently, to address the issue of low data utilization, some scholars have proposed and established knowledge graph construction methods and related invention patents in relevant fields. Some technical solutions involve processing historical construction data to build an initial knowledge graph, processing power construction site data to identify existing risks, formulating corresponding control measures, and dynamically updating the initial knowledge graph. Other solutions involve preprocessing real-time and historical data to construct a knowledge graph, performing anomaly detection, accident prediction, and risk assessment on real-time data, generating early warning information, and developing emergency measures. However, knowledge graphs face semantic confusion issues caused by "data silos" and "flat" architectures when processing multi-source heterogeneous data, severely restricting dynamic retrieval efficiency. Using knowledge graph technology for operational environment safety analysis and evaluation also faces technical challenges such as rigid knowledge graph architecture and inaccurate risk indicator judgments.

[0064] This invention provides a method for analyzing and evaluating workplace safety. It integrates multi-source data through a multi-level knowledge base and uses an attention mechanism for weight allocation to construct a risk knowledge graph that can integrate information from multiple aspects.

[0065] According to an embodiment of the present invention, a method for analyzing and evaluating the safety of the work environment is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0066] This embodiment provides a method for analyzing and evaluating the safety of the work environment, which can be used with the aforementioned computer equipment. Figure 1 This is a flowchart of the work environment safety analysis and evaluation method according to an embodiment of the present invention, such as... Figure 1 As shown, the process includes the following steps:

[0067] Step S101: Establish a hierarchical knowledge base for the elements of the work environment, and establish entity relationships between its levels to generate an initial knowledge graph.

[0068] Specifically, a hierarchical knowledge base is a management approach that organizes knowledge bases according to different hierarchical structures, aiming to improve the efficiency of knowledge management and the scalability of the system. Hierarchical knowledge bases are typically divided into multiple layers based on factors such as the level of abstraction, data sources, and application scenarios. A knowledge graph is a semantic network used to describe the relationships between entities; it is a semi-structured data representation method used to describe entities, attributes, and the relationships between entities.

[0069] The hierarchical knowledge base in this embodiment includes an indicator data layer that stores real-time data and historical data of work environment elements; an operation standard layer that stores the names, effective dates, and corresponding safety ranges of regulations and standards; an impact analysis layer that stores the impact of work environment indicators on workers, on production, and the factors affecting work environment indicators; and an accident case layer that stores accident causes, characteristics of work environment indicator data, and improvement measures.

[0070] The factors of the working environment include air, lighting, temperature, humidity, noise, vibration, dust, radiation, and toxic and harmful substances.

[0071] The data of each level above is stored in a relational database, and the entity relationship network is stored in a graph database, forming cross-level entities. The entities of the cross-level entities are then linked across levels through a pre-trained model to generate an initial knowledge graph for each work environment element.

[0072] Step S102: Compare the pre-acquired indicator data of the work environment elements with the hierarchical knowledge base of the work environment elements to identify risk indicators, calculate the risk values ​​of the risk indicators, and construct a safety risk analysis and evaluation set based on the risk indicators and their corresponding risk values.

[0073] Specifically, the work environment index data includes real-time data and historical data of the work environment elements recorded above.

[0074] Risk indicators are identified by comparing real-time data of work environment elements with the work standard layer of a hierarchical knowledge base. An attention mechanism is used to assign dual-dimensional weights to these risk indicators, and risk values ​​are calculated based on this weighted average. The dual-dimensional weights include historical data attention weights and real-time data attention weights for the work environment elements. Based on these dual-dimensional weights, entity relationships are established between the risk indicators and their risk values, forming a safety risk analysis and evaluation set.

[0075] Step S103: Construct a risk knowledge graph for analyzing and evaluating the safety of the work environment based on the safety risk analysis and evaluation set and the initial knowledge graph, and conduct safety analysis and evaluation of the work environment based on the risk knowledge graph.

[0076] Specifically, each risk indicator in the safety risk analysis and evaluation set is matched with the entity nodes in the indicator data layer to establish the association between the risk indicator and the entity nodes in its respective level. The semantic similarity between the risk indicator and the entity nodes in the operation standard layer, impact analysis layer, and accident case layer is calculated. If the similarity exceeds a preset threshold, a cross-level association is established to generate a risk knowledge graph for evaluating the safety of the work environment.

[0077] The work environment safety analysis and evaluation method provided in this embodiment constructs a multi-level knowledge base, integrates multi-source data, compares pre-acquired work environment indicator data with the hierarchical knowledge base to accurately identify risk indicators, and calculates the risk value of the risk indicators. Finally, it forms a risk knowledge graph that can integrate multiple aspects of information, which can be effectively applied to the safety evaluation of work environments in different industries, improves the accuracy and universality of safety evaluation, and solves the problems of rigid knowledge graph architecture and inaccurate risk indicator judgment in work environment scenarios.

[0078] This embodiment provides a method for analyzing and evaluating the safety of the work environment, which can be used with the aforementioned computer equipment. Figure 2 This is a flowchart of the work environment safety analysis and evaluation method according to an embodiment of the present invention, such as... Figure 2 As shown, the process includes the following steps:

[0079] Step S201: Establish a hierarchical knowledge base for the elements of the work environment, and establish entity relationships between its levels to generate an initial knowledge graph.

[0080] Specifically, step S201 includes:

[0081] Step S2011: Obtain indicator data, work standards, influencing factors on work environment elements, and work accident cases.

[0082] Specifically, the operational environment indicator data includes real-time data and historical data of operational environment elements. The operational standards include the name of the regulations / standards, their effective date, and the corresponding safety range. The safety range refers to the permissible fluctuation range of the monitoring indicators specified in the regulations / standards, determined by the threshold requirements in the regulations, and denoted as the range [L]. i H i Each rule node embeds the regulation name, safety scope, etc. Operational influencing factors include the impact of operational environment indicators on workers, the impact on production, and the factors affecting operational environment indicators. Operational accident cases include the cause of the accident, the characteristics of the operational environment indicators, and improvement measures.

[0083] Step S2012: Establish an indicator data layer based on indicator data of work environment elements, establish a work standard layer based on work standards of work environment elements, establish an impact analysis layer based on factors affecting work environment elements, and establish an accident case layer based on work accident cases.

[0084] Specifically, the indicator data layer is used to store real-time data and historical data of the work environment elements; the work standard layer is used to store the names of regulations and standards, their effective dates, and corresponding safety ranges; the impact analysis layer is used to store the impact of work environment indicator data on workers, on production, and the factors that affect work environment indicator data; and the accident case layer is used to store the causes of related work accidents, the characteristics of work environment indicator data, and improvement measures. Among these, the entity extraction technique is used to extract the causes of accidents, the characteristics of work environment indicator data, and improvement measures from accident cases.

[0085] Step S2013: Filter the data and entity relationship network in the indicator data layer, operation standard layer, impact analysis layer and accident case layer, store the data in a relational database, store the entity relationship network in a graph database, and generate a hierarchical knowledge base based on the stored relational database and the stored graph database.

[0086] Specifically, relational databases store data in tabular form, possessing a strict data structure and relational model, making them suitable for storing highly structured data with clear row and column relationships. In this work environment safety analysis and evaluation method, the data in the indicator data layer, work standard layer, impact analysis layer, and accident case layer—such as real-time and historical data of work environment elements, regulatory and standard names, effective dates, and the impact of work environment indicators on workers—all have clear field attributes and fixed formats. Storing this data in a relational database leverages its powerful transaction processing capabilities and data integrity constraints to ensure data consistency and accuracy. For example, when updating work standard layer data, transaction operations can ensure the synchronous updating of relevant fields such as effective dates and safety scope, avoiding data inconsistency issues.

[0087] Graph databases store data using a graph structure, excelling at handling complex relationship networks and efficiently representing and querying the relationships between entities. The entity relationship network in workplace safety analysis and evaluation, including relationships between entities within each level and relationships between entities across levels, is highly complex and dynamic. Storing it in a graph database facilitates relationship-based queries and analysis. For example, when querying the relationship between a workplace indicator and related accident cases, a graph database can quickly traverse nodes and edges to obtain the relationship information without the complex multi-table join operations required by relational databases, greatly improving the efficiency of relationship queries. Graph databases store (entity, relationship, entity) triples.

[0088] A hierarchical knowledge base is generated based on the stored relational database and the stored graph database.

[0089] Step S2014: Calculate the semantic similarity of cross-level entities in the hierarchical knowledge base using a pre-trained model, and associate entities with semantic similarity exceeding a preset threshold across levels to generate an initial knowledge graph.

[0090] Specifically, the hierarchical knowledge base encompasses multi-source heterogeneous data, including indicator data, operational standards, impact analysis, and accident case studies. Before calculating semantic similarity, this data needs to be cleaned and standardized. For example, the units and formats of real-time monitoring data in the indicator data layer are standardized; key entities, such as standard names, applicable scope, and specific values, are extracted from the text data in the operational standards layer; and the unstructured text in the accident case study layer undergoes preprocessing such as word segmentation and part-of-speech tagging to transform it into a structured form that can be processed by a computer.

[0091] The pre-processed data is input into a pre-trained model (such as BERT, GPT series, and other natural language processing models). Leveraging the model's powerful feature extraction capabilities, each entity is mapped to a high-dimensional semantic vector. These vectors contain the entity's semantic information, contextual information, and potential relationships with other entities. For example, the entity "hydrogen sulfide concentration" not only contains the literal meaning of the indicator but also implies semantic features such as its relevant attributes and impacts in the field of workplace safety.

[0092] Based on the generated entity semantic vectors, appropriate similarity metrics (such as cosine similarity, Euclidean distance, etc.) are used to calculate the semantic similarity between entities at different levels. Taking cosine similarity as an example, the cosine value of the angle between two vectors is calculated to measure their similarity in the semantic space. The closer the value is to 1, the more similar the two entities are semantically.

[0093] The system performs cross-level traversal of entities in the hierarchical knowledge base. For each entity from the indicator data layer, semantic similarity is calculated sequentially with all entities in the operation standards layer, impact analysis layer, and accident case layer. For example, for the entity "dust concentration exceeds the standard" in the indicator data layer, its semantic similarity is calculated with entities such as dust concentration-related standard clauses in the operation standards layer, dust hazard descriptions in the impact analysis layer, and dust explosion accident cases in the accident case layer, resulting in a series of similarity scores.

[0094] A preset threshold (e.g., 0.7) is set to filter entity pairs with strong semantic association. The calculated semantic similarity score is compared with the preset threshold. Only entity pairs with similarity scores exceeding the threshold are considered to have sufficient semantic association, thus establishing a cross-level association. For example, if the semantic similarity between "excessive dust concentration" and a certain regulatory standard clause is 0.8, which is greater than the preset threshold of 0.7, then the two are determined to be related.

[0095] Based on the characteristics of the entity's hierarchical level, different types are defined for the established cross-level relationships. For example, the relationship between the indicator data layer and the operational standards layer might be "compliant" or "violation"; the relationship between the indicator data layer and the impact analysis layer might be "may lead to" or "cause"; and the relationship between the indicator data layer and the accident case layer might be "similar to" or "related to". Simultaneously, relevant attributes are added to the relationships, such as the confidence level of the association (which can be converted based on semantic similarity scores) and the basis for the association.

[0096] Entities and their relationships, after being filtered and defined, are stored in a graph database as nodes and edges, constructing an initial knowledge graph. Each entity serves as a node, and the relationships between entities serve as edges, thus forming a structured and semantic knowledge graph, enabling the systematic organization and efficient management of knowledge related to workplace safety.

[0097] For example, taking a chemical plant as an example, the indicator data layer stores real-time data and historical data of operational environment elements such as light intensity, temperature, and concentration of toxic and harmful gases. The operation standard layer parses relevant laws and standards to obtain the names, effective dates, and safety scope of the laws and standards. The impact analysis layer stores the impact of operational environment indicators on workers, on production, and the factors affecting the operational environment indicators. The accident case layer is used to extract accident causes, operational environment indicator data characteristics, and improvement measures. After the hierarchical knowledge base is established, the data of each level is stored in a relational database, the entity relationship network is stored in a graph database, and the semantic similarity of cross-level entities is calculated through a pre-trained model. Entities with similarity exceeding a threshold are associated across levels to generate an initial knowledge graph.

[0098] For example, taking a chemical plant as an example, an entity node "Illumination Monitoring Point A01" is created in the indicator data layer to store real-time illuminance data and historical illuminance data for the past 30 days. In the operation standard layer, by parsing HG / T20586 "Technical Regulations for Lighting Design of Chemical Enterprises", the illuminance threshold [40, 300] of the chemical plant is extracted, and a rule node "Rule_Light_lx" is created. In the impact analysis layer, a node "Hydrogen sulfide concentration health risk" and its attribute "may cause respiratory diseases and nervous system damage to workers, affecting work efficiency" are created. In the accident case layer, a hydrogen sulfide poisoning accident report from a certain location in August 2021 is extracted, and nodes such as "Incident_2021-08", "H2S_Conc_High", "H2S_15mg m / m³", and "Arrange for dedicated personnel to regularly inspect, calibrate, and maintain the gas detector" are obtained, and a "has_risk_relation" relationship is established with the monitoring point. The semantic alignment engine calculates the semantic similarity between different nodes. If the threshold is met, a "comply_with" relationship is established to generate an initial knowledge graph across levels.

[0099] Step S202: The work environment index data includes real-time data of work environment elements. The real-time data of work environment elements is obtained in the following way: a sensor network is deployed in a distributed manner in the work environment monitoring scenario, and real-time data of work environment elements are collected from multiple sensors in the sensor network. The real-time data of work environment elements includes timestamps, sensor measurement values, and measurement locations. Based on a preset reference time start point or the initial acquisition time of the first sensor, the timestamps of the other sensors are offset and compensated to generate real-time data of work environment elements under a unified time reference. If there are missing values ​​in the real-time data of work environment elements, supplementary values ​​are calculated based on the data points before and after the missing location through bidirectional linear interpolation to generate complete real-time data of work environment elements.

[0100] For example, in a chemical plant, real-time data on environmental factors such as illuminance, temperature, and hydrogen sulfide concentration are collected through a sensor network. Using the initial acquisition time of the first sensor as a reference, the timestamps of the remaining sensors are offset to generate a data sequence with a unified time reference. When hydrogen sulfide sensor data is lost between 14:05 and 14:07, a supplementary value is calculated using bidirectional linear interpolation. For example, the weight of the data point to the left of the missing location is 0.7, and the weight of the data point to the right is 0.3. In this way, the missing value is estimated based on information from adjacent data points, and after supplementing the missing value, the hydrogen sulfide concentration data is maintained at 12.0 mg / m³. This embodiment achieves accurate temporal alignment and fusion of operational data and supplements missing values ​​in the real-time data of environmental factors using bidirectional linear interpolation. This allows the operational environment data to more accurately reflect the actual situation of the operational environment, effectively solving the problems of incomplete data and inconsistent time, and improving data quality and usability.

[0101] Step S203 involves comparing pre-acquired operational environment indicator data with the hierarchical knowledge base to identify risk indicators, calculating the risk values ​​of these indicators, and constructing a safety risk analysis and evaluation set based on the risk indicators and their corresponding risk values. For details, please refer to [link to relevant documentation]. Figure 1 Step S102 of the illustrated embodiment will not be described again here.

[0102] Step S204: Construct a risk knowledge graph for analyzing and evaluating workplace safety based on the safety risk analysis and evaluation set and the initial knowledge graph, and perform safety analysis and evaluation of the workplace environment based on the risk knowledge graph. For details, please refer to [link to details]. Figure 1 Step S103 of the illustrated embodiment will not be described again here.

[0103] The operational environment safety analysis and evaluation method provided in this embodiment systematically classifies and structures heterogeneous data from multiple sources by clearly defining the functions and storage content of different levels. It calculates the semantic similarity between entities at each level using a pre-trained model, associating entities with similarity greater than a similarity threshold. This optimizes the logical hierarchy of knowledge representation, significantly improving the accuracy and response speed of data retrieval, enabling a more comprehensive and accurate safety evaluation of the operational environment, and effectively solving the problem of weak semantic connections between levels. The interconnected levels form a comprehensive and systematic hierarchical knowledge base, providing rich and accurate data support for operational environment safety analysis and evaluation, and improving data utilization.

[0104] This embodiment provides a method for analyzing and evaluating the safety of the work environment, which can be used with the aforementioned computer equipment. Figure 3 This is a flowchart of the work environment safety analysis and evaluation method according to an embodiment of the present invention, such as... Figure 3 As shown, the process includes the following steps:

[0105] Step S301: Establish a hierarchical knowledge base for the elements of the work environment, and establish entity relationships between its levels to generate an initial knowledge graph. For details, please refer to [link to relevant documentation]. Figure 2 Step S201 of the illustrated embodiment will not be described again here.

[0106] Step S302: Compare the pre-acquired indicator data of the work environment elements with the hierarchical knowledge base of the work environment elements to identify risk indicators, calculate the risk values ​​of the risk indicators, and construct a safety risk analysis and evaluation set based on the risk indicators and their corresponding risk values.

[0107] Specifically, the operational standard layer includes the safety scope; the operational environment indicator data also includes historical data on operational environment elements; the above step S302 includes:

[0108] Step S3021: Compare the real-time data of the work environment elements with the safety range of the work standard layer, and identify the real-time data of the work environment elements that are not within the safety range as risk indicators.

[0109] Specifically, in the risk indicator identification phase, the system first conducts a detailed comparison between the real-time data of the work environment elements and the safety range of the work standard layer. The real-time data of the work environment elements is collected in real time by the sensor network, covering various environmental parameters such as gas concentration, temperature and humidity, and light intensity. Each data point contains a precise timestamp, measurement value, and measurement location information.

[0110] The operational standard layer stores safety ranges defined by national regulations and industry standards. For example, the permissible concentration standard for short-term exposure to hydrogen sulfide is 0-10.0 mg / m³. The system uses a preset comparison algorithm to check the measured value of each parameter in the real-time data one by one. If a parameter (such as a measured hydrogen sulfide concentration of 12.0 mg / m³) exceeds the corresponding safety range, that real-time data point will be initially marked as abnormal.

[0111] For flagged abnormal real-time data, the system further traces its historical environmental data. If similar anomalies or abnormal fluctuation trends frequently occur in historical data, the parameter is identified as a risk indicator. For example, if hydrogen sulfide concentration repeatedly approaches or exceeds the standard value recently, and historical data shows that the concentration tends to rise under certain operating conditions, then hydrogen sulfide concentration will be formally identified as a risk indicator and included in the subsequent assessment process.

[0112] Step S3022: Based on historical data of work environment elements, an attention mechanism is used to assign two-dimensional weights to risk indicators, and a weighted calculation method is used to calculate the risk value corresponding to the risk indicator based on the two-dimensional weights.

[0113] In some optional implementations, the two-dimensional weighting includes historical data attention weighting of the work environment elements and real-time data attention weighting of the work environment elements; step S3022 above includes:

[0114] Step a1: Calculate the historical data attention weights of the work environment elements based on the pre-trained gated recurrent unit model.

[0115] Specifically, the gated recurrent unit model is abbreviated as GRU model. GRU (Gate Recurrent Unit) is an improved recurrent neural network (RNN) that effectively captures long sequence dependencies through a gating mechanism, solving the gradient vanishing problem of traditional RNNs.

[0116] Historical data on the operational environment elements of the target risk indicators are collected. This data is typically in time series format, containing measurements of the indicator over a past period. For example, for the risk indicator "hydrogen sulfide concentration," concentration measurements at each point in time over the past few hours, days, or even months are obtained. This historical data is then organized into a sequence and input into a pre-trained GRU model. During training, the model learns from a large amount of historical data, mastering the potential patterns and regularities of various operational environment indicators. When processing input data, the hidden layers of the GRU continuously update their state, combining information from previous time steps with the input of the current time step, gradually extracting key features from the data. The model outputs a value representing the attention weight of the historical data for the operational environment element. This weight reflects the importance of the historical data to the current risk assessment; a higher weight indicates a stronger correlation between the information contained in the historical data and the current risk situation. For example, if a work area has a history of frequently experiencing accidents caused by abnormally high hydrogen sulfide concentrations, then historical data will be given relatively high weight when assessing the current risk of hydrogen sulfide concentration in that area, indicating that historical experience has important reference value for current risk assessment.

[0117] The formula for calculating the attention weight of historical data for work environment elements is as follows:

[0118] α hist,i = softma x( Attn ( h i {(t)} , h i {(t-k)} ,…, h i {(t-1)} )),in, denoted as the current time feature, k is the historical window length, and Attn(·) is the additive attention function.

[0119] Step a2: Calculate the degree of anomaly and rate of change of risk indicators, calculate the real-time emergency score based on the degree of anomaly and rate of change of risk indicators, and calculate the real-time data attention weight of the work environment elements based on the real-time emergency score.

[0120] Specifically, the formula for calculating the degree of abnormality of risk indicators, that is, the degree of deviation between the current measured value and the normal range or standard value, is as follows: , Indicates the normal range. Indicates the safe range.

[0121] Taking "hydrogen sulfide concentration" as an example, if its short-term exposure allowable concentration standard is 10.0 mg / m³, while the current real-time measurement value is 12.0 mg / m³, then the degree of abnormality can be calculated by (12.0-10.0) / 10.0, resulting in an abnormality level of 0.2, which indicates that the current concentration exceeds the standard value by 20%.

[0122] Simultaneously, the rate of change of risk indicators is calculated by analyzing changes in measured values ​​at adjacent time points. The calculation formula is: .

[0123] For example, comparing the hydrogen sulfide concentration at the current moment with that at the previous moment, if the concentration rises rapidly in a short period of time, it indicates a large rate of change and an increased potential risk.

[0124] Based on the calculated degree of anomaly and rate of change, a real-time emergency score is further calculated. A weighted summation method can be used to combine the degree of anomaly and rate of change according to a certain weight ratio to obtain a value that reflects the current level of risk urgency. The emergency score calculation formula is as follows: ,in, For the Sigmoid function, z 1, z 2 represents the industry weight coefficient, determined by a large number of accident cases in the accident case layer based on logistic regression analysis; this is the default value. z 1 = 0.6 z 2 = 0.4.

[0125] For example, setting the weight of the degree of abnormality to 0.6 and the weight of the rate of change to 0.4, using the formula: A specific score is calculated. Finally, the real-time data attention weight of the work environment elements is calculated based on the real-time emergency score. Generally, the higher the real-time emergency score, the greater the real-time data attention weight of the corresponding work environment elements. A preset mapping rule can be used to map the real-time emergency score to a value within the range [0, 1] as the real-time data attention weight of the work environment elements. For example, when the real-time emergency score is 0.8, according to the mapping rule, it is converted to a real-time data attention weight of 0.9 for the work environment elements, indicating that the current real-time data has extremely high importance in risk assessment.

[0126] Step a3: Calculate the comprehensive attention weight of the risk index based on the historical data attention weight of the work environment elements and the real-time data attention weight of the work environment elements.

[0127] Specifically, after obtaining the historical data attention weights and real-time data attention weights of the work environment elements, the two are merged to calculate the comprehensive attention weight of the risk indicator. A weighted average method is typically used, assigning different weight coefficients to the historical and real-time weights based on actual needs. For example, if it is believed that real-time data is more critical for risk assessment in the current work environment, the historical data attention weight coefficient for the work environment elements can be set to 0.4, and the real-time data attention weight coefficient for the work environment elements can be set to 0.6. The comprehensive attention weight is then calculated using the formula: Comprehensive Attention Weight = Historical Data Attention Weight of Work Environment Elements × 0.4 + Real-time Data Attention Weight of Work Environment Elements × 0.6.

[0128] Step a4: Calculate the risk value corresponding to the risk indicator using a weighted calculation method based on the comprehensive attention weight of the risk indicator.

[0129] Specifically, relevant characteristics of risk indicators (such as the degree of anomaly and the degree of matching with historical risk patterns) are combined with comprehensive attention weights, and the final risk value is obtained through weighted summation and other calculation methods. The risk value calculation formula is as follows: , This is an adjustment coefficient used to balance the influence of historical data and real-time data of work environment elements. Historical data attention weights for work environment elements. Attention weights for real-time data of operational environment elements.

[0130] For example, by setting the anomaly severity weight to 0.5, the matching degree with historical risk patterns weight to 0.3, and the overall attention weight to 0.2, the specific risk value can be calculated using the formula: Risk Value = Anomaly Severity × 0.5 + Matching Degree with Historical Risk Patterns × 0.3 + Overall Attention Weight × 0.2. This risk value comprehensively reflects the impact of historical data experience and real-time data conditions on current risk indicators, providing a quantitative basis for workplace safety analysis and evaluation. This facilitates safety managers in more accurately judging risk levels and taking corresponding preventative measures.

[0131] Step S3023: Establish entity associations between risk indicators and corresponding risk values, construct a safety risk analysis and evaluation set based on entity associations, and store the safety risk analysis and evaluation set in the indicator data layer.

[0132] Specifically, after obtaining the risk indicators and their corresponding risk values, the system establishes an entity association between the two. Based on the nodes and edges of the graph database, the risk indicators (such as exceeding the hydrogen sulfide concentration limit) and risk values ​​(such as the calculated 0.85) are treated as different nodes and connected by specific types of edges (such as "has a risk value") to clarify their correspondence. At the same time, attributes are added to the associated edges, such as details of weight allocation and calculation basis, to enhance the interpretability of the association.

[0133] Based on the aforementioned entity relationships, the system constructs a safety risk analysis and evaluation set. The evaluation set is presented in structured data format, including detailed information such as risk indicator name, measured value, measurement location, timestamp, risk value, and dual-dimensional weight values. Finally, the constructed safety risk analysis and evaluation set is stored in the indicator data layer. For example, in a chemical plant, comparing the operating environment data [50 lx, 25℃, 12.0 mg / m³] with the safe range [0, 10.0 mg / m³] of the operating standard layer reveals that the short-term exposure allowable concentration of hydrogen sulfide is outside the safe range and is identified as a risk indicator.

[0134] A pre-trained GRU model is invoked, inputting historical data of work environment factors related to the short-term permissible concentration of hydrogen sulfide for nearly one hour, and the attention weights of the historical data for the work environment factors are calculated. Simultaneously, based on the degree of abnormality and rate of change of the indicators, a real-time emergency score E1=0.785 was calculated, yielding the real-time data attention weights for the operational environment elements. And calculate the comprehensive attention weight of the risk index. It is 0.888.

[0135] Establish entity relationships between risk indicator nodes and their risk value nodes, combine them into a safety risk analysis and evaluation set, and store it in the indicator data layer. Establish risk indicator W. i and its risk value w i The entity relationships between them are combined into a security risk analysis and evaluation set, W={(W1,w1),(W2,w2),…,(W n ,w n )}, where i is the i-th risk indicator and is stored in the indicator data layer.

[0136] The technical solution of this embodiment can more accurately identify risk indicators of the working environment, effectively solving the problems of difficulty in dynamically allocating weights and accurately identifying risk indicators of the working environment, and improving the accuracy and reliability of safety assessment.

[0137] Step S303: Construct a risk knowledge graph for analyzing and evaluating the safety of the work environment based on the safety risk analysis and evaluation set and the initial knowledge graph, and conduct safety analysis and evaluation of the work environment based on the risk knowledge graph.

[0138] Specifically, step S303 includes:

[0139] Step S3031: Semantically match each risk indicator in the security risk analysis and evaluation set with the entity nodes in the indicator data layer, and establish the association between the risk indicator and the entity nodes in its respective layer.

[0140] Specifically, when matching risk indicators in the safety risk analysis and evaluation set with entity nodes in the indicator data layer, the system first parses the text description of the risk indicators to extract key information. For example, for the risk indicator "hydrogen sulfide concentration exceeds the standard," "hydrogen sulfide concentration" is extracted as the core keyword. Next, in the indicator data layer, matching entity nodes are found through keyword search combined with semantic understanding. The indicator data layer stores a large amount of historical and real-time operational environment indicator data. The system uses pre-trained language models (such as the BERT model, Bidirectional Encoder Representation from Transformers) to calculate the semantic similarity between the risk indicators and the text of each entity node. For example, semantic similarity is calculated between "hydrogen sulfide concentration exceeds the standard" and entity nodes such as "hydrogen sulfide concentration monitoring data" and "historical trend of hydrogen sulfide concentration" recorded in the indicator data layer.

[0141] When the semantic similarity between an entity node and a risk indicator exceeds a preset threshold (e.g., 0.7), the system determines that the two are successfully matched and establishes an association. This association includes not only a simple correspondence but also records metadata such as the confidence level and association time. For example, it clearly establishes a strong association between the risk indicator "exceeding the standard hydrogen sulfide concentration" and the entity node "hydrogen sulfide concentration monitoring data" within a specific time period in the indicator data layer, providing a data traceability path for subsequent risk analysis.

[0142] Step S3032: Adjust the visualization attributes of the risk indicator according to the risk value. The visualization attributes include color and size.

[0143] Specifically, to more intuitively display the level of risk, the system adjusts the visualization attributes of risk indicators based on the risk value. For the color attribute, a gradient mapping method is used, assigning different colors to different risk value ranges. For example, the risk value range of 0-0.3 is mapped to green, indicating low risk; the range of 0.3-0.7 is mapped to yellow, indicating medium risk; and the range of 0.7-1 is mapped to red, indicating high risk. For instance, if the risk value of the "hydrogen sulfide concentration exceeding the standard" risk indicator is 0.85, the system sets its visualization node color to red to warn of the severity of the risk.

[0144] In terms of size attribute adjustment, linear or non-linear scaling is also applied based on the risk value. When the risk value is set to 0, the node size is the minimum baseline value (e.g., a diameter of 10 pixels); as the risk value increases, the node size increases accordingly. For example, when the risk value is 1, the node diameter is enlarged to 30 pixels. Through size changes, safety managers can quickly identify high-risk indicators from the visualization interface and focus on key risk points.

[0145] Step S3033: Calculate the semantic similarity between the risk indicators and the entity nodes in the operation standard layer, impact analysis layer, and accident case layer based on the pre-trained model, and establish cross-level association relationships based on the relationship between semantic similarity and preset thresholds.

[0146] In some optional implementations, the two-dimensional weighting includes historical data attention weighting and real-time data attention weighting of the work environment elements; step S3033 above includes:

[0147] Step b1: Calculate the semantic similarity between risk indicators and entities in the operation standard layer using a pre-trained model to establish compliance or violation relationships.

[0148] Specifically, a pre-trained model is used to calculate the semantic similarity between risk indicators and entity nodes in the work standard layer. The work standard layer stores various safety regulations, industry standards, and other textual content. For each risk indicator, the system extracts relevant clauses, limits, and other information from the work standard layer as entity nodes.

[0149] Taking "excessive hydrogen sulfide concentration" as an example, the system will retrieve relevant standard clauses from the operational standard layer, such as the provisions on short-term exposure limits for hydrogen sulfide in the "Occupational Exposure Limits for Hazardous Factors in the Workplace". The system calculates the semantic similarity between this risk indicator and these standard clauses using a pre-trained model. If the similarity exceeds a preset threshold (e.g., 0.6), a "violation" or "non-compliance" association is established, and the specific violated standard content and limit value are marked, clearly presenting the deviation between the risk indicator and regulatory requirements.

[0150] Step b2: Based on the semantic matching of risk indicators, the potential consequences description in the impact analysis layer is used to form a risk propagation path and construct the causal relationship between risk and consequences.

[0151] Specifically, the impact analysis layer includes analyses of the potential consequences and impact paths of operational environment indicators. The system semantically matches risk indicators with entity nodes in the impact analysis layer to uncover potential consequences that risks may cause. For example, for the risk indicator of "excessive hydrogen sulfide concentration," semantic matching identifies entity nodes in the impact analysis layer that describe "may cause personnel poisoning" or "may corrode equipment." When semantic similarity meets the criteria, causal relationships such as "may cause" and "may trigger" are established, forming a propagation path from the risk indicator to potential consequences. Simultaneously, attributes such as weights or confidence levels are added to these relationships to reflect the likelihood of the risk causing corresponding consequences, helping safety managers comprehensively understand the scope and severity of the risk's impact.

[0152] Step b3 involves performing multi-dimensional matching between the current risk indicators and historical accident cases at the accident case layer. When the matching degree exceeds a preset threshold, a similarity relationship is established and historical handling measures are associated to form risk prediction and handling suggestions. Multi-dimensional matching includes indicator type similarity matching, numerical range similarity matching, and time pattern similarity matching.

[0153] Specifically, the accident case layer stores detailed information on historical operational environment accidents. The system performs multi-dimensional matching between current risk indicators and historical accident cases in the accident case layer. Besides semantic similarity calculation, it also considers dimensions such as indicator type, numerical range, and time pattern. For example, for the risk indicator "hydrogen sulfide concentration exceeding the standard," it not only compares its semantic similarity with descriptions of hydrogen sulfide accidents in historical cases but also compares the numerical range and trend of hydrogen sulfide concentration at the time of the accident. When the multi-dimensional matching degree exceeds a preset threshold (e.g., 0.7), a "similar to" association is established, and information such as the handling measures and lessons learned from historical accident cases is linked. Through this association, a reference basis is provided for responding to current risks, enabling the reuse of historical experience and improving the effectiveness of risk management.

[0154] Step S3034: Construct a risk knowledge graph for analyzing and evaluating workplace safety based on cross-level relationships.

[0155] Specifically, after establishing the aforementioned cross-level relationships, the system constructs a risk knowledge graph in graph form based on these relationships. Risk indicators, operational standards entities, impact analysis entities, and accident case entities are used as nodes, and the established relationships are used as edges, forming a structured and semantic knowledge network. A schematic diagram of the constructed risk knowledge graph is shown below. Figure 6 As shown.

[0156] Each node and edge is accompanied by rich attribute information, such as the node's name, type, risk value, and visualization attributes, as well as the edge's association type, weight, and confidence level. In this way, the risk knowledge graph not only presents the basic information of risk indicators but also comprehensively demonstrates their complex relationships with regulations, standards, potential impacts, and historical cases. It provides an intuitive, comprehensive, and in-depth analytical framework for workplace safety analysis and evaluation, facilitating risk assessment, prevention measure development, and decision support for safety managers.

[0157] For example, taking the monitoring of hydrogen sulfide concentration in a chemical plant as an example, the implementation process of S400 is specifically explained. The risk index W1 in the safety risk analysis and evaluation set W is matched with the entity node "hydrogen sulfide monitoring point C03" in the index data layer to establish a "has_risk" association relationship, and the visualization attributes of the node are adjusted according to the risk value w1=0.888: the node color changes from black to red, and the diameter is expanded from 20px to 28px. After calculating the semantic similarity between the risk indicator W1 and cross-level nodes based on the BERT pre-trained model, for the operation standard layer, the regulatory text of the node "Risk Indicator W1" and the node "Rule_H2S_Conc" is semantically matched, generating a similarity score of 0.92 (threshold 0.85), and establishing a "violates_standard" association. In the impact analysis layer, the nodes "Hydrogen sulfide concentration health risk" and "Factors affecting hydrogen sulfide concentration" are associated, and "triggers_health_impact" and "requires_intervention" relationships are constructed respectively. In the accident case layer, the semantic similarity between the risk of the node "Risk Indicator W1" and the accident cause "H2S_Conc_High" of the historical accident node "Incident_2021-08" reaches 0.89, and a "similar_to_incident" relationship is established. After semantic association was completed, triples such as (hydrogen sulfide monitoring point C03, has_risk, risk indicator W1), (risk indicator W1, violents_standard, Rule_H2S_Conc), and (risk indicator W1, similar_to_incident, Incident_2021-08) were added to the graph database. In the final risk knowledge graph, the nodes "risk indicator W1" and "hydrogen sulfide monitoring point C03" are highlighted with large red icons. This provides decision support for enterprise safety production management.

[0158] The work environment safety analysis and evaluation method provided in this embodiment associates real-time risk indicators from the safety risk analysis and evaluation set with different types of data in the initial knowledge graph, such as the indicator data layer, work standard layer, impact analysis layer, and accident case layer. It not only associates with historical monitoring data in the indicator data layer but also with occupational exposure limits in the work standard layer, poisoning hazards in the impact analysis layer, and similar accident cases in the accident case layer, forming a complete knowledge chain. By establishing semantic relationships between entities, the originally fragmented safety data is transformed into structured knowledge. Presented in a graph structure, it clearly displays the source, impact, and countermeasures of risk indicators, facilitating understanding and analysis. The visualization attributes of risk indicators are adjusted according to the risk value; this intuitive visualization allows safety managers to quickly locate high-risk areas and key risk indicators and take timely measures. Using a pre-trained model to calculate semantic similarity and establish cross-level relationships can uncover implicit connections between risk indicators and other knowledge entities. As one or more specific application embodiments of this invention, combined with... Figures 4 to 6 The method for analyzing and evaluating workplace safety provided by this invention will be further described in detail below:

[0159] like Figure 4 As shown in the figure, this embodiment provides a method for analyzing and evaluating the safety of the work environment, the method including:

[0160] S100: Establish a hierarchical knowledge base for the elements of the work environment, and establish entity relationships between its levels to generate an initial knowledge graph.

[0161] The S200 sensor network collects real-time data on environmental elements and performs timestamp alignment and missing value supplementation.

[0162] S300 compares the real-time data of the work environment elements with the work standard layer of the hierarchical knowledge base, identifies risk indicators, assigns dual-dimensional weights to the risk indicators using an attention mechanism and calculates risk values, and combines the risk indicators and their risk values ​​into a safety risk analysis and evaluation set.

[0163] S400, together with the safety risk analysis and evaluation set and the initial knowledge graph, forms a risk knowledge graph for analyzing and evaluating the safety of the work environment.

[0164] Furthermore, such as Figure 5 As shown, this embodiment provides a knowledge graph-based method for workplace safety analysis and evaluation, including:

[0165] Step S1: Establish a hierarchical knowledge base D for the work environment, and establish entity relationships between its levels to generate an initial knowledge graph. The hierarchical knowledge base D includes an indicator data layer, a work standard layer, an impact analysis layer, and an accident case layer.

[0166] Specifically, a hierarchical knowledge base D is established, comprising: an indicator data layer, including but not limited to storing real-time data and historical data of operational environment elements corresponding to operational environment indicators such as lighting, temperature, humidity, noise, vibration, dust, radiation, concentration of toxic and harmful substances, and ventilation volume; and an operational standard layer, which extracts the safety range from standard specifications using structured parsing technology. The safety range refers to the allowable fluctuation range of monitoring indicators specified in regulations and standards, determined by threshold requirements in regulatory clauses, and denoted as the interval [L]. i H i Each rule node embeds the regulation name, safety scope, etc.; the impact analysis layer stores the impact of work environment indicators on workers, the impact on production, and the factors affecting work environment indicators; the accident case layer stores a large number of accident cases, and extracts accident causes, work environment indicator data characteristics, and improvement measures from the accident cases through entity extraction technology.

[0167] When generating the initial knowledge graph based on the hierarchical knowledge base D, a hybrid storage architecture and semantic alignment algorithm are used to achieve multi-source data fusion. A relational database stores hierarchical metadata and index information, while a graph database stores (entity, relation, entity) triples. The semantic alignment engine calculates text similarity using a pre-trained model, identifies semantically consistent nodes at different levels, and establishes cross-level relationships.

[0168] Step S2: Collect real-time data U of the working environment elements from multiple sensors in the sensor network, and perform timestamp alignment and missing value supplementation.

[0169] Specifically, the process of collecting real-time data U of operational environment elements from multiple sensors in a sensor network and performing timestamp alignment and missing value supplementation includes: collecting real-time data of operational environment elements through a distributed sensor network deployed in the operational environment, wherein sensor nodes employ a redundancy check protocol to ensure data transmission integrity; and the real-time data of the operational environment elements includes timestamps, sensor measurements, and measurement locations. Using a preset reference time start point or the initial acquisition time of the first sensor as a benchmark, the timestamps of the remaining sensors are offset to generate a data sequence under a unified time benchmark. If missing values ​​exist in the real-time data of the operational environment elements, supplementary values ​​are calculated based on the left and right data points at the missing location using bidirectional linear interpolation to generate a complete data sequence. For example, the weight of the data point to the left of the missing location is 0.7, and the weight of the data point to the right is 0.3.

[0170] Step S3: Compare the real-time data U of the work environment elements with the work standard layer of the hierarchical knowledge base of the work environment elements, identify the risk indicators Wi, use the attention mechanism to assign weights to the risk indicators and calculate the risk value wi, and combine the risk indicators and their risk values ​​into a safety risk analysis and evaluation set W={(W1,w1),(W2,w2),…,(Wn,wn)}, where i is the i-th risk indicator.

[0171] Specifically, let the real-time data of the working environment elements be an n-dimensional vector U=[ u 1, u 2, ..., u n The safety range corresponding to the standard operating procedure level is [L]. i H i The risk indicator is: An attention mechanism is used to assign two-dimensional weights to risk indicators and the risk value is obtained based on the weighted calculation. The two-dimensional weights include the historical data attention weights of the work environment elements and the real-time data attention weights of the work environment elements.

[0172] First, based on deep learning models (LSTM, GRU, etc.), the temporal features of the indicators are extracted, and the attention weights of historical data for the work environment elements are calculated: = softma x( Attn ( h i {(t)} , h i {(t-k)} ,…, h i {(t-1)} )),in, Let k be the historical window length, and Attn(·) be the additive attention function, representing the current time-to-time feature. Secondly, based on the degree of anomaly in the risk indicators... and rate of change Calculate the real-time emergency score: ,in, For the Sigmoid function, z 1, z 2 represents the industry weight coefficient, determined by a large number of accident cases in the accident case layer based on logistic regression analysis; this is the default value. z 1 = 0.6 z 2=0.4, thus obtaining the real-time data attention weights of the work environment elements: Based on the dual-dimensional weights of the risk indicators, the risk value of the risk indicators is calculated using a weighted average. The formula for calculating the risk value is as follows: , This is an adjustment factor used to balance the influence of historical and real-time data on work environment factors. A risk indicator W is established. i and its risk value w i The entity relationships between them are combined into a security risk analysis and evaluation set, W={(W1,w1),(W2,w2),…,(W n ,w n )}, where i is the i-th risk indicator, and it is stored in the indicator data layer. Historical data attention weights for work environment elements. Attention weights for real-time data of operational environment elements. This represents the real-time emergency score of the j-th risk indicator. This represents the total number of risk indicators currently identified.

[0173] Step S4: Associate the safety risk analysis and evaluation set with the initial knowledge graph to form a risk knowledge graph for analyzing and evaluating the safety of the work environment.

[0174] Specifically, associating the safety risk analysis and evaluation set with the initial knowledge graph to form a risk knowledge graph for evaluating workplace safety includes: matching each risk indicator in the safety risk analysis and evaluation set with the entity nodes of the indicator data layer, establishing the association between the risk indicator and the entity nodes of its respective level, and adjusting the visualization attributes of the risk indicator, including color and size, according to the risk value; and calculating the semantic similarity between the risk indicator and the entity nodes in the work standard layer, impact analysis layer, and accident case layer based on the pre-trained model. If the similarity exceeds a preset threshold, a cross-level association is established.

[0175] The following uses a hazardous chemical production workshop in a chemical plant as an example to illustrate the implementation process of this invention. This area is equipped with a light intensity sensor to monitor the area's illuminance (Area_lx); a temperature sensor to monitor the area's temperature (Area_temp); and a hydrogen sulfide sensor to monitor the hydrogen sulfide concentration (H2S_Conc), as shown below. Figure 5 As shown, the specific implementation steps are as follows:

[0176] S1: Establish a hierarchical knowledge base D for specific work environment elements, and establish entity relationships between its levels to generate an initial knowledge graph. The hierarchical knowledge base D includes an indicator data layer, a work standard layer, an accident case layer, and an impact analysis layer.

[0177] Specifically, the indicator data layer creates entity nodes for "illuminance monitoring point A01", "temperature monitoring point B02", and "hydrogen sulfide monitoring point C03". The monitoring data of these nodes are stored and associated with the corresponding relational databases. In the operational standards layer, HG / T 20586 "Technical Regulations for Lighting Design of Chemical Enterprises" is parsed to extract the illuminance threshold [40, 300] for chemical plant buildings, creating nodes "Rule_Light_lx" and "[40, 300]" for "HG / T 20586 Technical Regulations for Lighting Design of Chemical Enterprises". Similarly, GB50016-2014 "Design Code for Industrial Enterprises" is parsed to create nodes "Rule_Area_Temp" and "[20℃, 28℃]" for "GB 50016-2014 Design Code for Industrial Enterprises". GBZ 2.1-2019 "Occupational Exposure Limits for Hazardous Factors in the Workplace Part 1: Chemical Hazardous Factors" is parsed to extract the short-term permissible concentration of hydrogen sulfide as 10.0 mg. / m³, create the node “Rule_H2S_Conc”, “GBZ 2.1-2019 Occupational Exposure Limits for Hazardous Factors in the Workplace Part 1: Chemical Hazardous Factors”, “10.0mg / m³”; Impact Analysis Layer, based on the detailed analysis of the work environment report, create the node “Hydrogen Sulfide Concentration Health Risk” and its attributes “May cause respiratory diseases and nervous system damage to workers, affecting work efficiency”; the node “Hydrogen Sulfide Concentration Production Impact” and its attributes “Causes equipment corrosion, shortens equipment life, increases maintenance costs”; the node “Factors Affecting Hydrogen Sulfide Concentration” and its attributes “Inefficient ventilation system, failure to repair gas leak sources in a timely manner, unreasonable workshop layout”; Accident Case Layer, extract hydrogen sulfide concentration data from a certain location in August 2021. The report of the toxic incident, along with associated environmental characteristics (illuminance 50 lx, temperature 30℃, peak hydrogen sulfide concentration 15 mg / m³), establishes the following nodes: event node "Incident_2021-08", accident cause node "H2S_Conc_High", operational environmental indicator data characteristics "Light_50 lx", "Temp_30℃", "H2S_15 mg / m³", and improvement measure node "Arrange for a dedicated person to regularly inspect, calibrate, and maintain the gas detector". Simultaneously, a "has_risk_relation" relationship is established with "illuminance monitoring point A01", "temperature monitoring point B02", and "hydrogen sulfide monitoring point C03". These nodes are then linked to the accident influencing node "H2S_Conc_High" with a "has_influence_on" relationship. The semantic alignment engine based on the BERT model establishes triples for cross-level associations, such as the "comply_with" relationship between nodes "illuminance monitoring point A01" and "Rule_Light_lx". Similarly, it establishes relational triples for other work environment indicators. The dataset corresponding to the knowledge graph is shown in Table 1 below:

[0178] Table 1 shows the datasets corresponding to the knowledge graphs.

[0179]

[0180] S2: Collect real-time data U of operational environment elements from multiple sensors in the sensor network, and perform timestamp alignment and missing value supplementation. The steps include: collecting real-time data U of operational environment elements from multiple sensors in the sensor network, including illuminance at monitoring points. u 1=50 lx Monitoring point temperature 25℃, hydrogen sulfide concentration u =12.0 mg / m³ (600 sampling points within 10 minutes, data is for example only); Based on the initial acquisition time of the first sensor, the timestamps of the remaining sensors are offset to generate a data sequence under a unified time reference; When the hydrogen sulfide sensor data is lost between 14:05 and 14:07, the supplementary value is calculated by bidirectional linear interpolation, for example, the weight is 0.7 when the distance is 6m. In this way, the missing value is estimated based on the information of adjacent data points, and after the missing value is filled, the hydrogen sulfide concentration data is kept at 12.0 mg / m³.

[0181] S3: Compare the real-time data U of the work environment elements with the work standard layer of the hierarchical knowledge base to identify risk indicators W. i An attention mechanism is used to assign weights to risk indicators and calculate the risk value w. i The risk indicators and their risk values ​​are combined into a safety risk analysis and evaluation set W = {(W1, w1), (W2, w2), ..., (W...}. n ,w n )}, where i is the i-th risk indicator. Includes:

[0182] Comparing the operational environment data [50 lx, 25℃, 12.0 mg / m³] with the safety range of the standard operational layer [0, 10.0 mg / m³], it was found that the short-term exposure allowable concentration of hydrogen sulfide was outside the safe range, and this was identified as risk indicator W1. An attention mechanism was used to assign two-dimensional weights to this risk indicator. A pre-trained GRU model was called, and historical data of the operational environment elements for the past hour of short-term exposure allowable concentration of hydrogen sulfide were input. The attention weight α of the historical data of the operational environment elements was calculated. hist =0.72, and also based on the degree of abnormality in the short-term exposure concentration of hydrogen sulfide. With rate of change Substitute into the Sigmoid function to calculate the real-time emergency score. E 1 = 0.785, thus obtaining the real-time attention weight. Assuming the adjustment coefficient =0.2, and the calculated risk value w1 is 0.888. The risk indicators and their risk values ​​are combined into a safety risk analysis and evaluation set W={(W1,w1)}.

[0183] S4: Associate the safety risk analysis and evaluation set W with the initial knowledge graph to form a risk knowledge graph for analyzing and evaluating workplace safety, including:

[0184] The risk indicator W1 in the safety risk analysis and evaluation set W is matched with the entity node "hydrogen sulfide monitoring point C03" in the indicator data layer to establish a "has_risk" association. Based on the risk value w1=0.888, the visualization attributes of this node are adjusted: the node color changes from black to red, and the diameter expands from 20px to 28px. After calculating the semantic similarity between the risk indicator W1 and cross-level nodes using a BERT pre-trained model, for the operational standard layer, the regulatory text of the node "risk indicator W1" is semantically matched with that of the node "Rule_H2S_Conc", generating a similarity score of 0.92 (threshold 0.85), establishing a "violates_standard" association. In the accident case layer, the semantic similarity between the risk of the node "risk indicator W1" and the accident cause "H2S_Conc_High" of the historical accident node "Incident_2021-08" reaches 0.89, establishing a "similar_to_incident" relationship. After semantic association is completed, triples such as (hydrogen sulfide monitoring point C03, has_risk, risk indicator W1), (risk indicator W1, violates_standard, Rule_H2S_Conc), and (risk indicator W1, similar_to_incident, Incident_2021-08) are added to the graph database, ultimately forming a risk knowledge graph, as shown below. Figure 6 As shown in the image, the nodes "Risk Indicator W1" and "Hydrogen Sulfide Monitoring Point C03" are highlighted with large red icons, providing decision support for enterprise safety production management.

[0185] By establishing multi-dimensional classification paths and semantic association networks, hierarchical structures not only optimize the logical hierarchy of knowledge representation but also significantly improve the accuracy and response speed of data retrieval. It is worth noting that although different industries exhibit significant differences in process standards and operating procedures, the work environment, as a common core element of safe production, remains a key area for risk management in all industries. Therefore, constructing a work environment safety analysis and evaluation system based on hierarchical knowledge graphs has significant technological innovation value and practical guiding significance.

[0186] This embodiment also provides a work environment safety analysis and evaluation device, which is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can be a combination of software and / or hardware that performs a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0187] This embodiment provides a device for analyzing and evaluating the safety of the work environment, such as... Figure 7 As shown, it includes:

[0188] Initialization module 701 is used to establish a hierarchical knowledge base for the elements of the work environment, establish entity relationships between its levels, and generate an initial knowledge graph.

[0189] Attention module 702 is used to compare the indicator data of pre-acquired work environment elements with the hierarchical knowledge base to identify risk indicators, calculate the risk value of the risk indicators, and construct a safety risk analysis and evaluation set based on the risk indicators and their corresponding risk values.

[0190] The safety evaluation module 703 is used to construct a risk knowledge graph for analyzing and evaluating the safety of the work environment based on the safety risk analysis and evaluation set and the initial knowledge graph, and to perform safety analysis and evaluation of the work environment based on the risk knowledge graph.

[0191] In some optional implementations, the initialization module 701 includes:

[0192] The data acquisition unit is used to acquire indicator data of work environment elements, work standards, factors affecting work environment elements, and work accident cases.

[0193] The hierarchical establishment unit is used to establish an indicator data layer based on indicator data of work environment elements, an operation standard layer based on operation standards of work environment elements, an impact analysis layer based on factors affecting work environment elements, and an accident case layer based on operation accident cases.

[0194] The hierarchical knowledge base generation unit is used to filter data and entity relationship networks in the indicator data layer, operation standard layer, impact analysis layer, and accident case layer, and store the data in a relational database and the entity relationship network in a graph database. The hierarchical knowledge base is generated based on the stored relational database and the stored graph database.

[0195] The initial knowledge graph construction unit is used to calculate the semantic similarity of entities across levels in the hierarchical knowledge base through a pre-trained model, and to associate entities with semantic similarity exceeding a preset threshold across levels to generate the initial knowledge graph.

[0196] In some optional implementations, the indicator data of the work environment elements include real-time data of the work environment elements, and the work environment safety analysis and evaluation device further includes:

[0197] The real-time data acquisition module for operational environment elements is used to deploy a distributed sensor network in the operational environment monitoring scenario and collect real-time data of operational environment elements from multiple sensors in the sensor network. The real-time data of operational environment elements includes timestamps, sensor measurements, and measurement locations. Based on a preset reference time start point or the initial acquisition time of the first sensor, the timestamps of the remaining sensors are offset and compensated to generate real-time data of operational environment elements under a unified time reference. If there are missing values ​​in the real-time data of operational environment elements, the missing values ​​are calculated by bidirectional linear interpolation based on the data points before and after the missing location to generate complete real-time data of operational environment elements.

[0198] In some optional implementations, the work standard layer includes a safety range; the work environment indicator data also includes historical data on work environment elements; the attention module 702 includes:

[0199] The risk indicator identification unit is used to compare the real-time data of the work environment elements with the safety range of the work standard layer, and to identify the real-time data and historical environmental data of the work environment elements that are not within the safety range as risk indicators.

[0200] The attention unit is used to assign two-dimensional weights to risk indicators based on historical data of work environment elements using an attention mechanism, and to calculate the risk value corresponding to the risk indicator using a weighted calculation method based on the two-dimensional weights.

[0201] The entity association establishment and safety risk analysis and evaluation set construction unit is used to establish entity associations between risk indicators and their corresponding risk values, and to construct a safety risk analysis and evaluation set based on these entity associations. This set is then stored in the indicator data layer. In some optional implementations, the dual-dimensional weighting includes historical data attention weights and real-time data attention weights for operational environment elements; the attention unit includes:

[0202] The historical data attention weight calculation subunit for work environment elements is used to calculate the historical data attention weights of work environment elements based on a pre-trained gated recurrent unit model.

[0203] The real-time data attention weight calculation subunit for work environment elements is used to calculate the degree of anomaly and rate of change of risk indicators, calculate the real-time emergency score based on the degree of anomaly and rate of change of risk indicators, and calculate the real-time data attention weight of work environment elements based on the real-time emergency score.

[0204] The comprehensive attention weight calculation subunit is used to calculate the comprehensive attention weight of risk indicators based on the historical data attention weight of work environment elements and the real-time data attention weight of work environment elements.

[0205] The risk value calculation subunit is used to calculate the risk value corresponding to the risk indicator based on the comprehensive attention weight of the risk indicator using a weighted calculation method.

[0206] In some alternative implementations, the safety evaluation module 703 includes:

[0207] The association establishment unit is used to semantically match each risk indicator in the security risk analysis and evaluation set with the entity nodes in the indicator data layer, and to establish the association relationship between the risk indicator and the entity node in its respective layer.

[0208] The visualization unit is used to adjust the visualization attributes of risk indicators according to the risk value. The visualization attributes include color and size.

[0209] The cross-level association establishment unit is used to calculate the semantic similarity between risk indicators and entity nodes in the operation standard layer, impact analysis layer, and accident case layer based on the pre-trained model, and to establish cross-level associations based on the relationship between semantic similarity and preset thresholds.

[0210] The risk knowledge graph construction unit is used to build a risk knowledge graph for analyzing and evaluating workplace safety based on cross-level relationships.

[0211] In some optional implementations, the cross-level association establishment unit includes:

[0212] The first relation establishment subunit is used to calculate the semantic similarity between risk indicators and entities in the operation standard layer through a pre-trained model, and to establish compliance or violation relations.

[0213] The second relationship establishment subunit is used to describe the potential consequences in the impact analysis layer based on the semantic matching of risk indicators, form a risk propagation path, and construct the causal relationship between risk and consequences.

[0214] The third relationship establishment subunit is used to match the current risk indicators with historical accident cases in the accident case layer in multiple dimensions. When the matching degree exceeds the preset threshold, a similarity relationship is established and historical handling measures are associated to form risk prediction and handling suggestions. The multi-dimensional matching includes indicator type similarity matching, numerical range similarity matching and time pattern similarity matching.

[0215] Further functional descriptions of the above modules and units are the same as those in the corresponding embodiments described above, and will not be repeated here.

[0216] The work environment safety analysis and evaluation device in this embodiment is presented in the form of functional units. Here, a unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0217] This invention also provides a computer device having the above-described features. Figure 7 The device shown is for analyzing and evaluating the safety of the work environment.

[0218] Please see Figure 8 , Figure 8 This is a schematic diagram of the structure of a computer device provided in an optional embodiment of the present invention, such as... Figure 8 As shown, the computer device includes one or more processors 10, memory 20, and interfaces for connecting the components, including high-speed interfaces and low-speed interfaces. The components communicate with each other via different buses and can be mounted on a common motherboard or otherwise installed as needed. The processors can process instructions executed within the computer device, including instructions stored in or on memory to display graphical information of a GUI on external input / output devices (such as display devices coupled to the interfaces). In some alternative implementations, multiple processors and / or multiple buses can be used with multiple memories and multiple memory modules, if desired. Similarly, multiple computer devices can be connected, each providing some of the necessary operations (e.g., as a server array, a group of blade servers, or a multiprocessor system). Figure 8 Take a processor 10 as an example.

[0219] Processor 10 may be a central processing unit, a network processor, or a combination thereof. Processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The programmable logic device may be a complex programmable logic device (CAMP), a field-programmable gate array (FPGA), a general-purpose array logic (GDA), or any combination thereof.

[0220] The memory 20 stores instructions executable by at least one processor 10 to cause the at least one processor 10 to perform the method shown in the above embodiments.

[0221] The memory 20 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the computer device. Furthermore, the memory 20 may include high-speed random access memory and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some alternative embodiments, the memory 20 may optionally include memory remotely located relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0222] The memory 20 may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as flash memory, hard disk or solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0223] The computer device also includes an input device 30 and an output device 40. The processor 10, memory 20, input device 30, and output device 40 can be connected via a bus or other means. Figure 8 Taking the example of a connection between China and Israel via a bus.

[0224] Input device 30 can receive input numerical or character information, and generate key signal inputs related to user settings and function control of the computer device, such as a touchscreen, keypad, mouse, trackpad, touchpad, joystick, one or more mouse buttons, trackball, joystick, etc. Output device 40 may include display devices, auxiliary lighting devices (e.g., LEDs), and haptic feedback devices (e.g., vibration motors). The aforementioned display devices include, but are not limited to, liquid crystal displays, light-emitting diodes, displays, and plasma displays. In some alternative embodiments, the display device may be a touchscreen.

[0225] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code, which, when accessed and executed by the computer, processor, or hardware, implements the methods shown in the above embodiments.

[0226] A portion of this invention can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can invoke or provide the methods and / or technical solutions according to the invention through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions, or the computer compiling the instructions and then executing the corresponding compiled program, or the computer reading and executing the instructions, or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.

[0227] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A method for safety analysis and evaluation of industrial production operation environment, characterized in that, The method includes: Establish a hierarchical knowledge base for operational environment elements, and establish entity relationships between its levels to generate an initial knowledge graph; this includes: Obtain indicator data, work standards, influencing factors, and case studies of work accidents related to the work environment. An indicator data layer is established based on the indicator data of the work environment elements; an operation standard layer is established based on the operation standards of the work environment elements; an impact analysis layer is established based on the factors affecting the work environment elements; and an accident case layer is established based on the operation accident cases. Risk indicators are identified by comparing the pre-acquired indicator data of the work environment elements with the hierarchical knowledge base of the work environment elements. At the same time, the risk values ​​of the risk indicators are calculated, and a safety risk analysis and evaluation set is constructed based on the risk indicators and their corresponding risk values. The indicator data of the work environment elements include real-time data and historical data of the work environment elements; the work standard layer includes the safety range; the real-time data acquisition method of the work environment elements includes: collecting real-time data of work environment elements such as air, lighting, temperature, humidity, noise, vibration, dust, radiation, and toxic and harmful substances at monitoring points through a sensor network, using the initial acquisition time of the first sensor as the reference, offset compensation is performed on the timestamps of the other sensors to generate real-time data of work environment elements under a unified time reference; The process involves comparing pre-acquired indicator data of operational environment elements with risk indicators identified by the hierarchical knowledge base, simultaneously calculating the risk values ​​of the risk indicators of the operational environment elements, and constructing a safety risk analysis and evaluation set based on the risk indicators and corresponding risk values, including: By comparing the real-time data of the work environment elements with the safety range of the work standard layer, the real-time data of the work environment elements that are not within the safety range are identified as risk indicators. Based on historical data of work environment elements, an attention mechanism is used to assign two-dimensional weights to the risk indicators, and a weighted calculation method is used to calculate the risk value corresponding to the risk indicator based on the two-dimensional weights. Establish entity associations between the risk indicators and their corresponding risk values, and construct a security risk analysis and evaluation set based on the entity associations, storing the security risk analysis and evaluation set in the indicator data layer; The dual-dimensional weighting includes historical data attention weighting and real-time data attention weighting of work environment elements. The historical data based on work environment elements employs an attention mechanism to assign dual-dimensional weights to the risk indicators, and calculates the risk value corresponding to the risk indicators using a weighted calculation method based on these dual-dimensional weights, including: The attention weights of historical data for work environment elements are calculated based on a pre-trained gated recurrent unit model. Calculate the degree of anomaly and rate of change of risk indicators, calculate a real-time emergency score based on the degree of anomaly and rate of change of risk indicators, and calculate the real-time data attention weight of the work environment elements based on the real-time emergency score. The comprehensive attention weight of the risk index is calculated based on the historical data attention weight and the real-time data attention weight of the work environment elements. Based on the comprehensive attention weight of risk indicators, the risk value corresponding to the risk indicators is calculated using a weighted calculation method. Based on the aforementioned safety risk analysis and evaluation set and the initial knowledge graph, a risk knowledge graph is constructed for analyzing and evaluating the safety of the work environment, and the work environment is analyzed and evaluated based on the risk knowledge graph.

2. The method according to claim 1, characterized in that, The process of establishing a hierarchical knowledge base for operational environment elements, creating entity relationships between its levels, and generating an initial knowledge graph also includes: The data and entity relationship network in the indicator data layer, operation standard layer, impact analysis layer and accident case layer are filtered, and the data is stored in a relational database and the entity relationship network is stored in a graph database. A hierarchical knowledge base is generated based on the stored relational database and the stored graph database. The semantic similarity of entities across levels in the hierarchical knowledge base is calculated using a pre-trained model, and entities with semantic similarity exceeding a preset threshold are associated across levels to generate an initial knowledge graph.

3. The method according to claim 1, characterized in that, The methods for acquiring real-time data on the operational environment elements also include: If there are missing values ​​in the real-time data of the work environment elements, supplementary values ​​are calculated by bidirectional linear interpolation based on the data points before and after the missing location to generate complete real-time data of the work environment elements.

4. The method according to claim 2, characterized in that, The construction of a risk knowledge graph for evaluating workplace safety based on the safety risk analysis and evaluation set and the initial knowledge graph includes: Semantically match each risk indicator in the security risk analysis and evaluation set with the entity nodes in the indicator data layer, and establish the association between the risk indicator and the entity nodes in its respective layer. The visualization attributes of risk indicators are adjusted according to the magnitude of the risk value, and the visualization attributes include color and size; Based on the pre-trained model, the semantic similarity between risk indicators and entity nodes in the operation standard layer, impact analysis layer, and accident case layer is calculated, and cross-level association is established based on the relationship between semantic similarity and preset thresholds. Based on the aforementioned cross-level relationships, a risk knowledge graph for evaluating workplace safety is constructed.

5. The method according to claim 4, characterized in that, The step of calculating the semantic similarity between risk indicators based on the pre-trained model and entity nodes in the operation standard layer, impact analysis layer, and accident case layer, and establishing cross-level associations based on the relationship between semantic similarity and preset thresholds, includes: The semantic similarity between risk indicators and entities in the operation standard layer is calculated using a pre-trained model to establish compliance or violation relationships. Based on the semantic matching of risk indicators, the potential consequences described in the impact analysis layer are formed to form a risk propagation path and construct a causal relationship between risk and consequences. The current risk indicators are matched with historical accident cases in the accident case layer in multiple dimensions. When the matching degree exceeds a preset threshold, a similarity relationship is established and historical handling measures are associated to form risk prediction and handling suggestions. The multi-dimensional matching includes indicator type similarity matching, numerical range similarity matching and time pattern similarity matching.

6. An industrial production operation environment safety analysis and evaluation device, characterized in that, The device includes: The initialization module is used to establish a hierarchical knowledge base of work environment elements, establish entity relationships between its levels, and generate an initial knowledge graph; establishing a hierarchical knowledge base of work environment elements, establishing entity relationships between its levels, and generating an initial knowledge graph includes: Obtain indicator data, work standards, influencing factors, and case studies of work accidents related to the work environment. An indicator data layer is established based on the indicator data of the work environment elements; an operation standard layer is established based on the operation standards of the work environment elements; an impact analysis layer is established based on the factors affecting the work environment elements; and an accident case layer is established based on the operation accident cases. The attention module is used to compare the indicator data of the pre-acquired work environment elements with the hierarchical knowledge base to identify risk indicators, calculate the risk value of the risk indicators, and construct a safety risk analysis and evaluation set based on the risk indicators and their corresponding risk values. The index data of the work environment elements include real-time data and historical data of the work environment elements; the work standard layer includes the safety range; real-time data of work environment elements such as air, lighting, temperature, humidity, noise, vibration, dust, radiation, and toxic and harmful substances are collected at monitoring points through a sensor network; the timestamps of the remaining sensors are offset and compensated based on the initial acquisition time of the first sensor to generate real-time data of work environment elements under a unified time reference. The process involves comparing pre-acquired indicator data of operational environment elements with risk indicators identified by the hierarchical knowledge base, simultaneously calculating the risk values ​​of the risk indicators of the operational environment elements, and constructing a safety risk analysis and evaluation set based on the risk indicators and corresponding risk values, including: By comparing the real-time data of the work environment elements with the safety range of the work standard layer, the real-time data of the work environment elements that are not within the safety range are identified as risk indicators. Based on historical data of work environment elements, an attention mechanism is used to assign two-dimensional weights to the risk indicators, and a weighted calculation method is used to calculate the risk value corresponding to the risk indicator based on the two-dimensional weights. Establish entity associations between the risk indicators and their corresponding risk values, and construct a security risk analysis and evaluation set based on the entity associations, storing the security risk analysis and evaluation set in the indicator data layer; The dual-dimensional weighting includes historical data attention weighting and real-time data attention weighting of work environment elements. The historical data based on work environment elements employs an attention mechanism to assign dual-dimensional weights to the risk indicators, and calculates the risk value corresponding to the risk indicators using a weighted calculation method based on these dual-dimensional weights, including: The attention weights of historical data for work environment elements are calculated based on a pre-trained gated recurrent unit model. Calculate the degree of anomaly and rate of change of risk indicators, calculate a real-time emergency score based on the degree of anomaly and rate of change of risk indicators, and calculate the real-time data attention weight of the work environment elements based on the real-time emergency score. The comprehensive attention weight of the risk index is calculated based on the historical data attention weight and the real-time data attention weight of the work environment elements. Based on the comprehensive attention weight of risk indicators, the risk value corresponding to the risk indicators is calculated using a weighted calculation method. The safety evaluation module is used to construct a risk knowledge graph for evaluating the safety of the work environment based on the safety risk analysis and evaluation set and the initial knowledge graph, and to perform safety analysis and evaluation of the work environment based on the risk knowledge graph.

7. A computer device, characterized in that, include: The system includes a memory and a processor, which are interconnected and communicate with each other. The memory stores computer instructions, and the processor executes the computer instructions to perform the industrial production operation environment safety analysis and evaluation method according to any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to execute the industrial production operation environment safety analysis and evaluation method according to any one of claims 1 to 5.

Images