Industrial internet of things-based safety training data analysis method, system and device

Abstract

This application discloses a method, system, and device for safety training data analysis based on the Industrial Internet of Things (IIoT), relating to the technical field of IIoT. The method includes: performing cluster analysis on the training group based on personality test data to obtain clusters and personality profiles; performing temporal pattern transformation on operation sequence data to obtain operation temporal data; performing semantic extraction on scenario protocol temporal data to obtain scenario semantic representation; performing empirical mode decomposition on the operation temporal data and the physiological temporal data to obtain operation intrinsic mode function components and physiological intrinsic mode function components; performing semantic fusion on the operation intrinsic mode function components and the physiological intrinsic mode function components based on the scenario semantic representation to obtain cluster semantic representation; and performing pattern parsing on the cluster semantic representation based on the personality profile to obtain group analysis results. This application has the effect of improving the accuracy of safety performance assessment of trainees.

Description

Technical Field

[0001] This application relates to the technical field of industrial Internet of Things (IIoT), and in particular to methods, systems and devices for analyzing safety training data based on IIoT. Background Technology

[0002] In industrial production, safe operation of personnel is crucial for ensuring production continuity and personnel safety. Traditional safety training systems typically rely on standardized courses, written assessments, and regular practical drills. While these methods have accumulated rich experience, their inherent limitations are becoming increasingly apparent with the increasing complexity of industrial environments and the widespread adoption of IoT data collection capabilities. They are no longer sufficient to meet the urgent needs of modern industry for precise, efficient, and personalized safety training. Current training data analysis suffers from the following main shortcomings: First, the training model is rigid and lacks individual adaptability. Existing training systems mostly adopt a "one-size-fits-all" approach, failing to effectively identify and differentiate the differences in trainees' cognitive styles, risk-taking tendencies, stress responses, and other inherent personality and psychological traits. This leads to a disconnect between training content, pace, and assessment standards and individuals' actual learning abilities and behavioral tendencies, making it difficult to accurately evaluate and optimize training effectiveness, and leaving the direction for personalized improvement unclear.

[0003] Secondly, assessments are limited in scope and fail to capture the real-world pressure scenarios. Traditional assessments heavily rely on final operational results or simple written test scores, failing to replicate the complex scenarios in real-world work where multiple pressures are coupled, including time constraints, environmental distractions, and high cognitive load. Under such real-world pressure, the sequence of an individual's operational behaviors, decision-making logic, and accompanying physiological stress responses are crucial for assessing their safety, reliability, and psychological resilience. However, current technologies lack the systematic collection and correlation analysis of such multi-dimensional, time-series data.

[0004] Finally, data value mining is superficial and lacks in-depth insights. Although modern industrial IoT systems can collect massive amounts of operation logs, equipment status, and some environmental data through sensor networks, forming a so-called "data-rich" environment, the analytical methods for safety training remain lagging behind. Existing methods mostly remain at the level of data recording, or only perform simple statistics and visualization, failing to deeply integrate and jointly analyze discrete operational events, continuous physiological signals, and dynamically changing scene stress semantics. This results in the inability to extract interpretable patterns from the data, such as the behavioral instability patterns of different trait groups under specific stress, physiological warning thresholds, and common characteristics of skill weaknesses, making training optimization lack a reliable data-driven basis.

[0005] In summary, existing industrial safety training systems and supporting technologies have significant shortcomings in personalized adaptation, multimodal behavioral and physiological assessment under real-world stress scenarios, and semantic-level fusion and intelligent parsing of multi-source heterogeneous data. Overcoming these deficiencies and constructing an intelligent analysis system that can deeply integrate personnel characteristics, behavioral responses, physiological states, and scenario stress has become a critical technical issue that urgently needs to be addressed in improving the effectiveness of industrial safety training and achieving the transformation from "generalized training" to "precise empowerment." Summary of the Invention

[0006] To improve the accuracy of safety performance assessments of trainees, this application provides a method, system, and equipment for safety training data analysis based on the Industrial Internet of Things.

[0007] Firstly, this application provides a method for analyzing safety training data based on the Industrial Internet of Things, employing the following technical solution: A safety training data analysis method based on the Industrial Internet of Things (IIoT) is applied to an IIoT system, which includes a management platform, a sensor network platform, and an object platform connected in sequence. The method is executed by the management platform and includes: Obtain personality test data of the training group, and perform cluster analysis on the training group based on the personality test data to obtain at least one cluster and personality profiles of each cluster. For each cluster, under configurable stress scenarios, the operation sequence data, physiological time series data, and corresponding scenario protocol time series data of the corresponding trainees are obtained. The operation sequence data is transformed into a time series pattern to obtain operation time series data. The scenario protocol time series data is semantically extracted to obtain a scenario semantic representation. The stress scenario includes at least one of time stress, environmental stress, and cognitive stress. Empirical mode decomposition is performed on the operational time series data and the physiological time series data to obtain operational intrinsic mode function components and physiological intrinsic mode function components; Based on the scene semantic representation, the operational intrinsic mode function components and the physiological intrinsic mode function components are semantically fused to obtain a cluster semantic representation; Based on the personality profile, the cluster semantic representation is parsed to obtain the group analysis results. The analysis results are used to represent the behavioral reliability, physiological stress patterns, and corresponding training weaknesses of trainees with different personality types under the stress scenario.

[0008] By adopting the above technical solution, personality test data of the training group is first obtained, and cluster analysis is performed on the training group based on the personality test data to obtain at least one cluster and personality profiles of each cluster. Then, for each cluster, under configurable stress scenarios, the corresponding trainees' operational sequence data, physiological time-series data, and corresponding scenario protocol time-series data are obtained. The operational sequence data is transformed into a time-series pattern to obtain operational time-series data, and semantic extraction is performed on the scenario protocol time-series data to obtain a scenario semantic representation. The stress scenario includes at least one of time stress, environmental stress, and cognitive stress. Then, empirical mode decomposition is performed on the operational time-series data and physiological time-series data to obtain operational intrinsic mode function components and physiological intrinsic mode function components. Then, based on the scenario semantic representation, semantic fusion is performed on the operational intrinsic mode function components and physiological intrinsic mode function components to obtain a cluster semantic representation. Finally, based on the personality profiles, pattern parsing is performed on the cluster semantic representation to obtain the group analysis results. The analysis results are used to represent the behavioral reliability, physiological stress patterns, and corresponding training weaknesses of trainees with different personality types under stress scenarios. This invention first achieves personalized group segmentation based on personality clustering, breaking through the traditional "one-size-fits-all" analysis model. Second, by simultaneously collecting operational and physiological data under configurable stress scenarios, it restores the complexity of real-world operations and achieves multi-dimensional process evaluation. Third, by utilizing experience modality decomposition and semantic fusion technology, it deeply correlates behavior, physiology, and scenario stress, solving the problem of superficial data value mining. Finally, through pattern parsing, it accurately identifies the behavioral patterns, stress response modes, and weaknesses of different personality groups under stress. This method as a whole realizes the transformation from experience-driven to data-driven, and from generalized training to precise empowerment. It not only improves the accuracy of personnel safety performance evaluation but also provides a reliable basis for developing more targeted and effective safety training programs.

[0009] Optionally, the step of performing cluster analysis on the training group based on the personality test data to obtain at least one cluster and a personality profile for each cluster includes: The personality test data is processed into structured data to obtain structured data; Based on a personality psychology model, the structured data is feature-mapped and constructed to obtain a personality feature vector oriented towards industrial safety. The personality psychology model is trained by supervised learning using personality test sample data and corresponding safety personality dimension labels as training data. The personality psychology model is used to map the input personality test sample data into a personality feature representation associated with industrial safety behavior. Based on the personality trait vector, the training group is clustered to obtain at least one cluster. For each cluster, a profile of industrial safety is generated based on the cluster's feature centers and distribution characteristics.

[0010] By adopting the above technical solution, in order to achieve cluster analysis, the personality test data is first processed into structured data. Then, based on the personality psychology model, the structured data is feature-mapped and constructed to obtain a personality feature vector oriented towards industrial safety. The personality psychology model is trained by supervised learning using the personality test sample data and the corresponding safety personality dimension labels as training data. The personality psychology model is used to map the input personality test sample data into personality feature representations associated with industrial safety behavior. Then, based on the personality feature vector, the training group is clustered to obtain at least one cluster. For each cluster, a personality profile oriented towards industrial safety is generated based on the feature center and distribution characteristics of the cluster.

[0011] Optionally, the step of performing time-series pattern conversion on the operation sequence data to obtain operation time-series data includes: Parse the discrete operation events in the operation sequence data, wherein the discrete operation events include at least the operation object, the action type, and the action parameters; Based on a predefined operation semantic encoding table, each discrete operation event is mapped to a numerical code with security semantics; Based on the timestamps of the discrete operation events, the training time axis is divided into intervals to obtain at least one continuous time sequence window; For each time series window, the numerical codes corresponding to all operation events within the time series window are weighted, aggregated, and smoothed to obtain the window feature vector that represents the corresponding time series window. The window feature vectors of each of the continuous time-series windows are concatenated in time sequence to form operation time-series data.

[0012] By adopting the above technical solution, in order to obtain operation time-series data, discrete operation events in the operation sequence data are parsed. The discrete operation events include at least the operation object, action type, and action parameters. Then, based on a predefined operation semantic encoding table, each discrete operation event is mapped to a numerical code with security semantics. Then, based on the timestamps of the discrete operation events, the training time axis is divided into intervals to obtain at least one continuous time-series window. Then, for each time-series window, the numerical codes corresponding to all operation events within the time-series window are weighted, aggregated, and smoothed to obtain a window feature vector representing the corresponding time-series window. Finally, the window feature vectors of each continuous time-series window are concatenated in time sequence to form operation time-series data.

[0013] Optionally, the step of semantic extraction from the scene protocol time-series data to obtain a scene semantic representation includes: The scenario protocol timing data is analyzed to identify key pressure events, which include pressure type, pressure trigger time, pressure intensity parameters, and pressure duration. Based on all key stress events, a stress impact model is constructed, wherein the stress impact model is used to represent the stress semantic vector formed by the superposition of the key stress events at any time on the training timeline; Based on the stress impact model, the dynamic scene semantic vector corresponding to each sampling moment on the training time axis is determined. The stress impact model is obtained by supervised learning training using historical key stress event data and corresponding stress impact labels as training data, and is used to map the input historical key stress events into stress semantic vectors. The dynamic scene semantic vector at each sampling time is matrixed to generate a scene semantic representation.

[0014] By adopting the above technical solution, in order to obtain the scene semantic representation, the scene protocol time series data is parsed to identify key stress events. The key stress events include stress type, stress trigger time, stress intensity parameters, and stress duration. Then, based on all key stress events, a stress impact model is constructed. The stress impact model is trained by supervised learning using historical key stress event data and corresponding stress impact labels as training data. It is used to map the input historical key stress events into stress semantic vectors. Then, based on the stress impact model, the dynamic scene semantic vectors corresponding to each sampling moment on the training time axis are determined. Finally, the dynamic scene semantic vectors at each sampling moment are matrixed to generate the scene semantic representation.

[0015] Optionally, the step of semantically fusing the operational intrinsic mode function components and the physiological intrinsic mode function components based on the scene semantic representation to obtain a cluster semantic representation includes: Based on the scene semantic representation, the operational intrinsic mode function components are subjected to semantic transformation under stress scene constraints to obtain the scene operational semantic representation. Based on the scene semantic representation, the physiological intrinsic mode function components are subjected to semantic transformation under stress scene constraints to obtain the scene physiological semantic representation. For each cluster, the scene operation semantic representation and the scene physiological semantic representation of the cluster are aggregated across modalities to obtain the behavioral physiological semantic representation; Principal component analysis is performed on the behavioral physiological semantic representations of the same cluster to obtain the cluster semantic representation.

[0016] By adopting the above technical solution, in order to obtain the cluster semantic representation, firstly, based on the scene semantic representation, the semantic transformation of the operational intrinsic mode function components under stress scene constraints is performed to obtain the scene operational semantic representation. Then, based on the scene semantic representation, the semantic transformation of the physiological intrinsic mode function components under stress scene constraints is performed to obtain the scene physiological semantic representation. Then, for each cluster, the scene operational semantic representation and the scene physiological semantic representation of the cluster are aggregated across modes to obtain the behavioral physiological semantic representation. Finally, principal component analysis is performed on the behavioral physiological semantic representations of the same cluster to obtain the cluster semantic representation.

[0017] Optionally, the step of performing a semantic transformation on the operational intrinsic mode function components under stress scenario constraints based on the scene semantic representation to obtain the scene operational semantic representation includes: An operational mode matrix is ​​constructed based on each of the operational intrinsic mode function components, wherein the rows of the operational mode matrix represent the mode order, the columns represent the sampling time, and the corresponding matrix element values ​​represent the amplitude; The scene semantic representation is reconstructed to obtain a scene matrix, wherein the rows of the scene matrix represent the stress semantic dimension, the columns represent the sampling time, and the corresponding matrix element values ​​represent the intensity value of the stress semantic dimension at the sampling time. Based on the operation mode matrix, a query vector sequence is generated, and based on the scene matrix, a key vector sequence and a value vector sequence are generated. Calculate the attention scores of the query vector sequence and the key vector sequence to obtain the temporal attention weight matrix; Based on the temporal attention weight matrix, the value vector sequence is weighted and aggregated to obtain a weighted modal representation; The weighted modal representations are aggregated across time to generate scene operation semantic representations.

[0018] By adopting the above technical solution, in order to obtain the scene operation semantic representation, an operation mode matrix is ​​constructed based on the eigenmode function components of each operation. In the operation mode matrix, the rows represent the mode order, the columns represent the sampling time, and the corresponding matrix element values ​​represent the amplitude. Then, the scene semantic representation is reconstructed to obtain the scene matrix. In the scene matrix, the rows represent the stress semantic dimension, the columns represent the sampling time, and the corresponding matrix element values ​​represent the intensity value of the stress semantic dimension at the sampling time. Then, based on the operation mode matrix, a query vector sequence is generated, and based on the scene matrix, a key vector sequence and a value vector sequence are generated. Then, the attention score of the query vector sequence and the key vector sequence is calculated to obtain the temporal attention weight matrix. Then, based on the temporal attention weight matrix, the value vector sequence is weighted and aggregated to obtain the weighted mode representation. Finally, the weighted mode representation is aggregated across time to generate the scene operation semantic representation.

[0019] Optionally, the step of performing cross-modal aggregation of the scene operation semantic representation and the scene physiological semantic representation of the cluster to obtain the behavioral physiological semantic representation includes: The scene operation semantic representation and the scene physiological semantic representation are projected onto the same shared semantic space to obtain the scene operation projection vector and the scene physiological projection vector. Based on the scene operation projection vector and the scene physiological projection vector, a cross-modal association graph is constructed, wherein the nodes are composed of the scene operation projection vector and the scene physiological projection vector, and the edge weights between nodes are determined by the mutual information metric. Graph convolution is performed on the cross-modal association graph to fuse operational and physiological modality information within the shared semantic space and update node features; The updated node features are aggregated to obtain a multimodal feature vector; The multimodal feature vectors are compressed to obtain a behavioral physiological semantic representation.

[0020] By adopting the above technical solution, in order to obtain the behavioral physiological semantic representation, the scene operation semantic representation and the scene physiological semantic representation are projected onto the same shared semantic space to obtain scene operation projection vectors and scene physiological projection vectors. Then, based on the scene operation projection vectors and scene physiological projection vectors, a cross-modal association graph is constructed, where each node is composed of scene operation projection vectors and scene physiological projection vectors, and the edge weights between nodes are determined by mutual information metrics. Then, graph convolution operation is performed on the cross-modal association graph to fuse the information of operation and physiological modalities in the shared semantic space and update the node features. Then, the updated node features are aggregated to obtain multimodal feature vectors. Finally, the multimodal feature vectors are compressed to obtain the behavioral physiological semantic representation.

[0021] Optionally, the step of performing pattern parsing on the cluster semantic representation based on the personality profile to obtain the group analysis results includes: The personality profile and the cluster semantic representation are correlated and modeled to obtain a mapping model for representing personality and behavioral physiological patterns; Based on the mapping model, the typical behavioral and physiological patterns represented by the cluster semantic representation are analyzed, and the personality profile is traced according to the analysis results to obtain key personality dimensions. Based on the key personality dimensions and the typical behavioral and physiological patterns, a group analysis result is generated for this cluster.

[0022] By adopting the above technical solution, in order to obtain the group analysis results, the personality profile and the cluster semantic representation are correlated and modeled to obtain a mapping model for representing personality and behavioral physiological patterns. Then, based on the mapping model, the typical behavioral physiological patterns represented by the cluster semantic representation are analyzed, and the personality profile is traced according to the analysis results to obtain key personality dimensions. Then, based on the key personality dimensions and typical behavioral physiological patterns, the group analysis results for the cluster are generated.

[0023] Secondly, this application also provides a safety training data analysis system based on the Industrial Internet of Things, which adopts the following technical solution: A safety training data analysis system based on the Industrial Internet of Things (IIoT) includes a management platform, a sensor network platform, and an object platform that are sequentially connected via communication. The management platform is configured with: The clustering module is used to acquire personality test data of the training group and perform cluster analysis on the training group based on the personality test data to obtain at least one cluster and personality profiles of each cluster. The data processing module is used to acquire, for each cluster, the corresponding trainees' operation sequence data, physiological time series data and corresponding scenario protocol time series data under configurable stress scenarios, and to perform time series pattern conversion on the operation sequence data to obtain operation time series data, and to perform semantic extraction on the scenario protocol time series data to obtain scenario semantic representation, wherein the stress scenario includes at least one of time stress, environmental stress and cognitive stress. The decomposition module is used to perform empirical mode decomposition on the operational time series data and the physiological time series data to obtain operational intrinsic mode function components and physiological intrinsic mode function components. The cluster semantic representation generation module is used to perform semantic fusion on the operational intrinsic mode function components and the physiological intrinsic mode function components based on the scene semantic representation to obtain the cluster semantic representation; The analysis result generation module is used to perform pattern parsing on the cluster semantic representation based on the personality profile to obtain group analysis results. The analysis results are used to represent the behavioral reliability, physiological stress patterns, and corresponding training weaknesses of trainees with different personality types under the stress scenario.

[0024] Thirdly, this application also provides a computer device, which adopts the following technical solution: A computer device includes a memory and a processor, the memory storing a computer program executable on the processor, the processor executing the computer program to implement the method described in the first aspect.

[0025] In summary, this application includes at least the following beneficial technical effects: First, personality test data of the training group is acquired, and cluster analysis is performed on the training group based on the personality test data to obtain at least one cluster and personality profiles of each cluster. Then, for each cluster, under configurable stress scenarios, the corresponding trainees' operational sequence data, physiological time-series data, and corresponding scenario protocol time-series data are acquired. The operational sequence data is then transformed into a time-series pattern to obtain operational time-series data, and semantic extraction is performed on the scenario protocol time-series data to obtain a scenario semantic representation. The stress scenario includes at least one of time stress, environmental stress, and cognitive stress. Then, empirical mode decomposition is performed on the operational time-series data and physiological time-series data to obtain operational intrinsic mode function components and physiological intrinsic mode function components. Then, based on the scenario semantic representation, semantic fusion is performed on the operational intrinsic mode function components and physiological intrinsic mode function components to obtain a cluster semantic representation. Finally, based on the personality profiles, pattern parsing is performed on the cluster semantic representation to obtain the group analysis results. The analysis results are used to represent the behavioral reliability, physiological stress patterns, and corresponding training weaknesses of trainees with different personality types under stress scenarios. This invention first achieves personalized group segmentation based on personality clustering, breaking through the traditional "one-size-fits-all" analysis model. Second, by simultaneously collecting operational and physiological data under configurable stress scenarios, it restores the complexity of real-world operations and achieves multi-dimensional process evaluation. Third, by utilizing experience modality decomposition and semantic fusion technology, it deeply correlates behavior, physiology, and scenario stress, solving the problem of superficial data value mining. Finally, through pattern parsing, it accurately identifies the behavioral patterns, stress response modes, and weaknesses of different personality groups under stress. This method as a whole realizes the transformation from experience-driven to data-driven, and from generalized training to precise empowerment. It not only improves the accuracy of personnel safety performance evaluation but also provides a reliable basis for developing more targeted and effective safety training programs. Attached Figure Description

[0026] Figure 1 This is a schematic diagram of the overall process of an embodiment of this application.

[0027] Figure 2 This is a structural diagram of one application scenario of the system in this application embodiment.

[0028] Figure 3 This is a structural diagram of another application scenario of the system according to an embodiment of this application.

[0029] Figure 4 This is a structural block diagram of the computer device described in this application. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0031] This application discloses a method for analyzing safety training data based on the Industrial Internet of Things.

[0032] Reference Figure 1 A safety training data analysis method based on the Industrial Internet of Things (IIoT) is applied to an IIoT system, which includes a management platform, a sensor network platform, and an object platform connected in sequence. The method is executed by the management platform and includes: Step S11: Obtain personality test data of the training group, and perform cluster analysis on the training group based on the personality test data to obtain at least one cluster and personality profiles of each cluster.

[0033] It should be noted that in step S11, the management platform first collects personality test data obtained by trainees through standardized psychological tests or questionnaires. This data usually covers dimensions such as an individual's cognitive style, risk preference, and stress response tendency. Subsequently, the platform uses clustering algorithms (such as K-means, hierarchical clustering, etc.) to group all trainees, and groups people with similar personality traits into the same cluster. Each cluster ultimately forms a representative personality profile, which abstractly describes the common characteristics and typical patterns of the group in terms of personality dimensions.

[0034] Step S12: For each cluster, under configurable stress scenarios, acquire the corresponding trainees' operation sequence data, physiological time series data, and corresponding scenario protocol time series data, and perform time series pattern transformation on the operation sequence data to obtain operation time series data, and perform semantic extraction on the scenario protocol time series data to obtain scenario semantic representation.

[0035] Among them, stressful scenarios include at least one of time stress, environmental stress, and cognitive stress.

[0036] It should be noted that in step S12, for each cluster, the system collects multi-source data from trainees under simulated or real configurable stress scenarios. These stress scenarios can be configured by the system and include one or more combinations of time pressure (e.g., time-limited operations), environmental pressure (e.g., noise, light interference), and cognitive pressure (e.g., multitasking, information overload). The collected data mainly includes: operation sequence data, recording the operation steps and order performed by personnel on industrial equipment or simulation interfaces; physiological time-series data, acquiring real-time physiological signals such as heart rate, skin conductance, and electroencephalogram (EEG) through wearable sensors; and scenario protocol time-series data, i.e., the time series of scenario events, state changes, and rule triggers recorded in system logs or communication protocols. Subsequently, the operation sequence data undergoes time-series pattern transformation, converting it into operation time-series data with a unified timestamp reflecting the rhythm and intervals of operations. Simultaneously, semantic extraction is performed on the scenario protocol time-series data, converting it into a structured semantic representation that describes the scenario state, stress sources, and task requirements by parsing event types, parameters, and context.

[0037] Step S13: Perform empirical mode decomposition on the operational time series data and physiological time series data to obtain the operational intrinsic mode function components and physiological intrinsic mode function components.

[0038] It should be noted that in step S13, to further analyze the intrinsic patterns and fluctuation characteristics in operational behavior and physiological response, this step employs the Empirical Mode Decomposition (EMD) method to adaptively decompose the operational time-series data and physiological time-series data respectively. EMD can decompose the non-stationary, nonlinear original signal layer by layer into a series of intrinsic mode function (IMF) components with frequencies ranging from high to low, and a residual trend term. Through this process, the operational time-series data is decomposed into multiple operational IMF components, and the physiological time-series data is decomposed into multiple physiological IMF components. These components respectively reveal subtle changes in operational behavior at different time scales (such as operational fluency and hesitation characteristics) and hierarchical information such as stress response and fatigue accumulation implied in the physiological signals.

[0039] Step S14: Based on the scene semantic representation, perform semantic fusion on the operational intrinsic mode function components and the physiological intrinsic mode function components to obtain the cluster semantic representation.

[0040] It should be noted that in step S14, the system, guided by the scene semantic representation extracted in step S12, aligns and integrates the features related to scene events and stress states in the operational IMF component and the physiological IMF component. For example, when a specific stress event occurs, the abnormal operational patterns reflected in the operational IMF within the corresponding time window and the features representing increased stress levels in the physiological IMF are extracted and bound to the event description and stress level in the scene semantics. Through methods such as feature weighting, association rule mining, or deep learning encoding, a multi-dimensional "cluster semantic representation" that comprehensively reflects the "behavior-physiology-scene" interaction relationship of the cluster under stress scenarios is finally generated. This representation is a holistic encoding of the group's response pattern under specific stress conditions.

[0041] Step S15: Based on the personality profile, perform pattern parsing on the cluster semantic representation to obtain the group analysis results.

[0042] The analysis results are used to represent the behavioral reliability, physiological stress patterns, and corresponding training weaknesses of trainees with different personality types under stress scenarios.

[0043] It should be noted that in step S15, the management platform, combined with the personality profiles of each cluster obtained in step S11, performs in-depth analysis of the cluster semantic representation generated in step S14. Through pattern recognition, comparative analysis, or the establishment of a mapping model between personality traits and semantic patterns, the system extracts the commonalities and differences exhibited by trainees of different personality types under stress scenarios. The analysis results specifically include: behavioral reliability: such as the frequency and type of operational errors and procedural violations by different personality groups under stress; physiological stress patterns: such as the typical characteristics of physiological arousal levels and recovery speeds of each group when coping with different types of stress; training weaknesses: identifying the safety awareness blind spots, skill deficiencies, or psychological resilience bottlenecks most likely to occur in specific personality groups under specific stress scenarios. These group-based analysis results can be directly used to guide the improvement of customized safety training content, the design of targeted drills, and the formulation of personalized intervention measures, thereby improving the overall effectiveness of personnel safety training in the industrial IoT environment.

[0044] In the above implementation, firstly, personality test data of the training group is acquired, and cluster analysis is performed on the training group based on the personality test data to obtain at least one cluster and personality profiles of each cluster. Then, for each cluster, under configurable stress scenarios, the corresponding trainees' operational sequence data, physiological time-series data, and corresponding scenario protocol time-series data are acquired. The operational sequence data is then transformed into a time-series pattern to obtain operational time-series data, and semantic extraction is performed on the scenario protocol time-series data to obtain a scenario semantic representation. The stress scenario includes at least one of time stress, environmental stress, and cognitive stress. Then, empirical mode decomposition is performed on the operational time-series data and physiological time-series data to obtain operational intrinsic mode function components and physiological intrinsic mode function components. Then, based on the scenario semantic representation, semantic fusion is performed on the operational intrinsic mode function components and physiological intrinsic mode function components to obtain a cluster semantic representation. Finally, based on the personality profiles, pattern parsing is performed on the cluster semantic representation to obtain the group analysis results. The analysis results are used to represent the behavioral reliability, physiological stress patterns, and corresponding training weaknesses of trainees with different personality types under stress scenarios. This invention first achieves personalized group segmentation based on personality clustering, breaking through the traditional "one-size-fits-all" analysis model. Second, by simultaneously collecting operational and physiological data under configurable stress scenarios, it restores the complexity of real-world operations and achieves multi-dimensional process evaluation. Third, by utilizing experience modality decomposition and semantic fusion technology, it deeply correlates behavior, physiology, and scenario stress, solving the problem of superficial data value mining. Finally, through pattern parsing, it accurately identifies the behavioral patterns, stress response modes, and weaknesses of different personality groups under stress. This method as a whole realizes the transformation from experience-driven to data-driven, and from generalized training to precise empowerment. It not only improves the accuracy of personnel safety performance evaluation but also provides a reliable basis for developing more targeted and effective safety training programs.

[0045] As a further implementation of the method, the step of performing cluster analysis on the training group based on personality test data to obtain at least one cluster and a personality profile for each cluster includes: Step S21: Perform structured processing on the personality test data to obtain structured data.

[0046] Step S22: Based on the personality psychology model, feature mapping and construction are performed on the structured data to obtain a personality feature vector oriented towards industrial safety. The personality psychology model is trained by supervised learning using personality test sample data and corresponding safety personality dimension labels as training data. The personality psychology model is used to map the input personality test sample data into personality feature representations associated with industrial safety behavior.

[0047] Step S23: Based on the personality trait vector, cluster the training group to obtain at least one cluster.

[0048] Step S24: For each cluster, generate a profile for industrial safety based on the cluster’s feature center and distribution characteristics.

[0049] In the above implementation, in order to achieve cluster analysis, the personality test data is first processed into structured data. Then, based on the personality psychology model, the structured data is feature-mapped and constructed to obtain a personality feature vector oriented towards industrial safety. The personality psychology model is trained by supervised learning using the personality test sample data and the corresponding safety personality dimension labels as training data. The personality psychology model is used to map the input personality test sample data into a personality feature representation associated with industrial safety behavior. Then, based on the personality feature vector, the training group is clustered to obtain at least one cluster. Then, for each cluster, a personality profile oriented towards industrial safety is generated based on the feature center and distribution characteristics of the cluster.

[0050] As a further implementation of the method, the step of performing time-series pattern transformation on the operation sequence data to obtain operation time-series data includes: Step S31: parse the discrete operation events in the operation sequence data, wherein the discrete operation events include at least the operation object, the action type, and the action parameters.

[0051] Step S32: Based on a predefined operation semantic encoding table, map each discrete operation event to a numerical code with security semantics.

[0052] Step S33: Based on the timestamps of discrete operation events, divide the training time axis into intervals to obtain at least one continuous time window.

[0053] Step S34: For each time series window, perform weighted aggregation and smoothing on the numerical codes corresponding to all operation events within the time series window to obtain the window feature vector used to represent the corresponding time series window.

[0054] Step S35: The window feature vectors of each consecutive time series window are concatenated according to the time sequence to form operation time series data.

[0055] In the above implementation, in order to obtain operation time-series data, discrete operation events in the operation sequence data are parsed. Each discrete operation event includes at least an operation object, an action type, and action parameters. Then, based on a predefined operation semantic encoding table, each discrete operation event is mapped to a numerical code with security semantics. Then, based on the timestamps of the discrete operation events, the training time axis is divided into intervals to obtain at least one continuous time-series window. Then, for each time-series window, the numerical codes corresponding to all operation events within the time-series window are weighted, aggregated, and smoothed to obtain a window feature vector representing the corresponding time-series window. Finally, the window feature vectors of each continuous time-series window are concatenated in time sequence to form operation time-series data.

[0056] As a further implementation of the method, the step of semantic extraction of scene protocol time-series data to obtain scene semantic representation includes: Step S41: Analyze the scenario protocol timing data and identify key pressure events, including pressure type, pressure trigger time, pressure intensity parameters and pressure duration.

[0057] Step S42: Based on all key stress events, construct a stress impact model. The stress impact model is used to represent the stress semantic vector formed by the superposition of key stress events at any time on the training timeline. The stress impact model is trained by supervised learning using historical key stress event data and corresponding stress impact labels as training data, and is used to map the input historical key stress events into stress semantic vectors.

[0058] Step S43: Based on the pressure influence model, determine the dynamic scene semantic vector corresponding to each sampling moment on the training time axis.

[0059] Step S44: Matrix the dynamic scene semantic vectors at each sampling time to generate a scene semantic representation.

[0060] In the above implementation, in order to obtain the scene semantic representation, the scene protocol time series data is parsed to identify key stress events. The key stress events include stress type, stress trigger time, stress intensity parameters, and stress duration. Then, based on all key stress events, a stress impact model is constructed. The stress impact model is trained using historical key stress event data and corresponding stress impact labels as training data through supervised learning. It is used to map the input historical key stress events into stress semantic vectors. Then, based on the stress impact model, the dynamic scene semantic vectors corresponding to each sampling moment on the training time axis are determined. Finally, the dynamic scene semantic vectors at each sampling moment are matrixed to generate the scene semantic representation.

[0061] As a further implementation of the method, the step of semantically fusing operational intrinsic mode function components and physiological intrinsic mode function components based on scene semantic representation to obtain cluster semantic representation includes: Step S51: Based on the scene semantic representation, perform semantic transformation on the operational intrinsic mode function components under stress scene constraints to obtain the scene operational semantic representation.

[0062] Step S52: Based on the scene semantic representation, perform semantic transformation on the physiological intrinsic mode function components under stress scene constraints to obtain the scene physiological semantic representation.

[0063] It should be noted that step S52 can be performed in the same way as step S51.

[0064] Step S53: For each cluster, perform cross-modal aggregation of the scene operation semantic representation and the scene physiological semantic representation of the cluster to obtain the behavior physiological semantic representation.

[0065] Step S54: Perform principal component analysis on the behavioral and physiological semantic representations of the same cluster to obtain the cluster semantic representation.

[0066] In the above implementation, in order to obtain the cluster semantic representation, the semantic transformation of the operational intrinsic mode function components under stress scenario constraints is first performed based on the scene semantic representation to obtain the scene operational semantic representation. Then, the semantic transformation of the physiological intrinsic mode function components under stress scenario constraints is performed based on the scene semantic representation to obtain the scene physiological semantic representation. Then, for each cluster, the scene operational semantic representation and the scene physiological semantic representation of the cluster are aggregated across modes to obtain the behavioral physiological semantic representation. Finally, principal component analysis is performed on the behavioral physiological semantic representations of the same cluster to obtain the cluster semantic representation.

[0067] As a further implementation of the method, the step of performing semantic transformation on the operational intrinsic mode function components under stress scene constraints based on scene semantic representation to obtain the scene operational semantic representation includes: Step S61: Construct an operational mode matrix based on each operational eigenmode function component, where the rows of the operational mode matrix represent the mode order, the columns represent the sampling time, and the corresponding matrix element values ​​represent the amplitude.

[0068] Step S62: Reconstruct the scene semantic representation to obtain a scene matrix, where the rows of the scene matrix represent the stress semantic dimension, the columns represent the sampling time, and the corresponding matrix element values ​​represent the intensity values ​​of the stress semantic dimension at the sampling time.

[0069] Step S63: Generate a query vector sequence based on the operation mode matrix, and generate a key vector sequence and a value vector sequence based on the scene matrix.

[0070] Step S64: Calculate the attention scores of the query vector sequence and the key vector sequence to obtain the temporal attention weight matrix.

[0071] Step S65: Based on the temporal attention weight matrix, the value vector sequence is weighted and aggregated to obtain a weighted modal representation.

[0072] Step S66: Perform cross-time aggregation on the weighted modal representation to generate a scene operation semantic representation.

[0073] In the above implementation, to obtain the scene operation semantic representation, an operation mode matrix is ​​constructed based on each operation intrinsic mode function component. The rows of the operation mode matrix represent the mode order, the columns represent the sampling time, and the corresponding matrix element values ​​represent the amplitude. Then, the scene semantic representation is reconstructed to obtain a scene matrix. The rows of the scene matrix represent the stress semantic dimension, the columns represent the sampling time, and the corresponding matrix element values ​​represent the intensity value of the stress semantic dimension at the sampling time. Based on the operation mode matrix, a query vector sequence is generated, and based on the scene matrix, a key vector sequence and a value vector sequence are generated. Then, the attention score between the query vector sequence and the key vector sequence is calculated to obtain a temporal attention weight matrix. Then, based on the temporal attention weight matrix, the value vector sequence is weighted and aggregated to obtain a weighted mode representation. Finally, the weighted mode representation is aggregated across time to generate the scene operation semantic representation.

[0074] As a further implementation of the method, the step of performing cross-modal aggregation of the scene operation semantic representation and scene physiological semantic representation of the cluster to obtain the behavioral physiological semantic representation includes: Step S71: Project the scene operation semantic representation and the scene physiological semantic representation onto the same shared semantic space to obtain the scene operation projection vector and the scene physiological projection vector.

[0075] Step S72: Construct a cross-modal association graph based on scene operation projection vector and scene physiological projection vector, wherein nodes are composed of scene operation projection vector and scene physiological projection vector, and the edge weights between nodes are determined by mutual information metric.

[0076] Step S73: Perform graph convolution operation on the cross-modal association graph to fuse information of operational and physiological modalities in the shared semantic space and update node features.

[0077] Step S74: Aggregate the updated node features to obtain a multimodal feature vector.

[0078] Step S75: Compress the multimodal feature vectors to obtain a behavioral physiological semantic representation.

[0079] In the above implementation, in order to obtain the behavioral physiological semantic representation, the scene operation semantic representation and the scene physiological semantic representation are projected onto the same shared semantic space to obtain scene operation projection vectors and scene physiological projection vectors. Then, based on the scene operation projection vectors and scene physiological projection vectors, a cross-modal association graph is constructed, wherein each node is composed of scene operation projection vectors and scene physiological projection vectors, and the edge weights between nodes are determined by mutual information metrics. Then, graph convolution operation is performed on the cross-modal association graph to fuse the information of operation and physiological modalities in the shared semantic space and update the node features. Then, the updated node features are aggregated to obtain multimodal feature vectors. Finally, the multimodal feature vectors are compressed to obtain the behavioral physiological semantic representation.

[0080] As a further implementation of the method, the step of performing pattern parsing on the cluster semantic representation based on personality profiling to obtain the group analysis results includes: Step S81: Perform correlation modeling on personality profile and cluster semantic representation to obtain a mapping model for representing personality and behavioral physiological patterns.

[0081] Step S81: Based on the mapping model, analyze the typical behavioral and physiological patterns represented by the cluster semantic representation, and trace the personality profile according to the analysis results to obtain the key personality dimensions.

[0082] Step S81: Based on key personality dimensions and typical behavioral and physiological patterns, generate group analysis results for this cluster.

[0083] In the above implementation, in order to obtain the group analysis results, the personality profile and the cluster semantic representation are correlated and modeled to obtain a mapping model for representing personality and behavioral physiological patterns. Then, based on the mapping model, the typical behavioral physiological patterns represented by the cluster semantic representation are analyzed, and the personality profile is traced according to the analysis results to obtain key personality dimensions. Then, based on the key personality dimensions and typical behavioral physiological patterns, the group analysis results for the cluster are generated.

[0084] This application also discloses a safety training data analysis system based on the Industrial Internet of Things.

[0085] refer to Figure 2 The safety training data analysis system based on the Industrial Internet of Things includes a management platform, a sensor network platform, and an object platform that are connected in sequence. The management platform is configured with: The clustering module is used to obtain personality test data of the training group and perform cluster analysis on the training group based on the personality test data to obtain at least one cluster and personality profiles of each cluster. The data processing module is used to acquire the operation sequence data, physiological time series data and corresponding scenario protocol time series data of the corresponding trainees for each cluster under configurable stress scenarios, and to perform time series pattern transformation on the operation sequence data to obtain operation time series data, and to perform semantic extraction on the scenario protocol time series data to obtain scenario semantic representation. The stress scenario includes at least one of time stress, environmental stress and cognitive stress. The decomposition module is used to perform empirical mode decomposition on operational time series data and physiological time series data to obtain operational intrinsic mode function components and physiological intrinsic mode function components; The cluster semantic representation generation module is used to perform semantic fusion on operational intrinsic mode function components and physiological intrinsic mode function components based on scene semantic representation to obtain cluster semantic representation; The analysis results generation module is used to perform pattern parsing on the cluster semantic representation based on personality profiles to obtain group analysis results. The analysis results are used to represent the behavioral reliability, physiological stress patterns and corresponding training weaknesses of trainees with different personality types under stress scenarios.

[0086] The overall framework of another application scenario of the safety training data analysis system based on the Industrial Internet of Things in this application is as follows: Figure 3 As shown, the system can include a user platform, a service platform, a management platform, a sensor network platform, and an object platform that interact sequentially, forming a five-platform architecture based on the Industrial Internet of Things (IIoT). The management platform is configured with a clustering module, a data processing module, a decomposition module, a cluster semantic representation generation module, and an analysis result generation module. The service platform includes a main service database, n service sub-platforms, and n service sub-databases. Each service sub-platform can communicate with its corresponding service sub-database, and each service sub-database can communicate with the main service database. The sensor network platform includes a main sensor database and n sensor network sub-platforms. Each sensor network sub-platform has its own sensor sub-database, and each sensor network sub-platform can communicate with the main sensor database.

[0087] Specifically, in another application scenario mentioned above, the safety training data analysis system based on the Industrial Internet of Things includes a management platform. The management platform is configured to: acquire personality test data of the training group, and perform cluster analysis on the training group based on the personality test data to obtain at least one cluster and personality profiles for each cluster; for each cluster, under configurable stress scenarios, acquire the corresponding trainees' operation sequence data, physiological time-series data, and corresponding scenario protocol time-series data, and perform time-series pattern transformation on the operation sequence data to obtain operation time-series data, and perform semantic extraction on the scenario protocol time-series data to obtain scenario language. The semantic representation is as follows: Stress scenarios include at least one of time stress, environmental stress, and cognitive stress; empirical mode decomposition is performed on operational and physiological time-series data to obtain operational intrinsic mode function (IMM) components and physiological intrinsic mode function (PIM) components; based on scenario semantic representation, semantic fusion is performed on the operational and physiological IMM components to obtain cluster semantic representation; based on personality profiles, pattern parsing is performed on the cluster semantic representation to obtain group analysis results. These results are used to represent the behavioral reliability, physiological stress patterns, and corresponding training weaknesses of trainees with different personality types under stress scenarios.

[0088] By leveraging the interaction between the various functional platforms of the industrial IoT-based safety training data analysis system, which is based on the aforementioned three or five platforms, a complete closed-loop information operation logic is established, ensuring the orderly operation of sensing and control information and realizing intelligent equipment management.

[0089] The safety training data analysis system based on the Industrial Internet of Things of the present invention can implement any of the methods in the safety training data analysis method based on the Industrial Internet of Things, and the specific working process of the safety training data analysis system based on the Industrial Internet of Things of the present invention can refer to the corresponding process in the above-mentioned safety training data analysis method based on the Industrial Internet of Things.

[0090] This application also discloses a computer device.

[0091] refer to Figure 4 A computer device includes a memory and a processor, wherein the memory stores a computer program that can run on the processor, and the processor executes the computer program to implement any of the above-described methods for security training data analysis based on the Industrial Internet of Things.

[0092] This application also discloses a computer-readable storage medium.

[0093] A computer-readable storage medium storing a computer program capable of being loaded by a processor and executing any of the above-described methods for security training data analysis based on the Industrial Internet of Things.

[0094] The computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, apparatus, or device; the program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0095] The above are all preferred embodiments of this application and are not intended to limit the scope of protection of this application. Any feature disclosed in this specification (including the abstract and drawings) may be replaced by other equivalent or similar features unless specifically stated otherwise. That is, unless specifically stated otherwise, each feature is only one example of a series of equivalent or similar features.

Claims

1. A method for analyzing safety training data based on the Industrial Internet of Things, characterized in that, Applied to an industrial Internet of Things (IIoT) system, the IIoT system includes a management platform, a sensor network platform, and an object platform that are sequentially and communicatively connected. The method is executed by the management platform and includes: Obtain personality test data of the training group, and perform cluster analysis on the training group based on the personality test data to obtain at least one cluster and personality profiles of each cluster. For each cluster, under configurable stress scenarios, the operation sequence data, physiological time series data, and corresponding scenario protocol time series data of the corresponding trainees are obtained. The operation sequence data is transformed into a time series pattern to obtain operation time series data. The scenario protocol time series data is semantically extracted to obtain a scenario semantic representation. The stress scenario includes at least one of time stress, environmental stress, and cognitive stress. Empirical mode decomposition is performed on the operational time series data and the physiological time series data to obtain operational intrinsic mode function components and physiological intrinsic mode function components; Based on the scene semantic representation, the operational intrinsic mode function components and the physiological intrinsic mode function components are semantically fused to obtain a cluster semantic representation; Based on the personality profile, the cluster semantic representation is parsed to obtain the group analysis results. The analysis results are used to represent the behavioral reliability, physiological stress patterns, and corresponding training weaknesses of trainees with different personality types under the stress scenario.

2. The safety training data analysis method based on the Industrial Internet of Things according to claim 1, characterized in that, The step of performing cluster analysis on the training group based on the personality test data to obtain at least one cluster and personality profiles for each cluster includes: The personality test data is processed into structured data to obtain structured data; Based on a personality psychology model, the structured data is feature-mapped and constructed to obtain a personality feature vector oriented towards industrial safety. The personality psychology model is trained by supervised learning using personality test sample data and corresponding safety personality dimension labels as training data. The personality psychology model is used to map the input personality test sample data into a personality feature representation associated with industrial safety behavior. Based on the personality trait vector, the training group is clustered to obtain at least one cluster. For each cluster, a profile of industrial safety is generated based on the cluster's feature centers and distribution characteristics.

3. The safety training data analysis method based on the Industrial Internet of Things according to claim 1, characterized in that, The step of performing time-series pattern conversion on the operation sequence data to obtain operation time-series data includes: Parse the discrete operation events in the operation sequence data, wherein the discrete operation events include at least the operation object, the action type, and the action parameters; Based on a predefined operation semantic encoding table, each discrete operation event is mapped to a numerical code with security semantics; Based on the timestamps of the discrete operation events, the training time axis is divided into intervals to obtain at least one continuous time sequence window; For each time series window, the numerical codes corresponding to all operation events within the time series window are weighted, aggregated, and smoothed to obtain the window feature vector that represents the corresponding time series window. The window feature vectors of each of the continuous time-series windows are concatenated in time sequence to form operation time-series data.

4. The safety training data analysis method based on the Industrial Internet of Things according to claim 1, characterized in that, The step of extracting semantics from the scene protocol time-series data to obtain a scene semantic representation includes: The scenario protocol timing data is analyzed to identify key pressure events, which include pressure type, pressure trigger time, pressure intensity parameters, and pressure duration. Based on all key stress events, a stress impact model is constructed. The stress impact model is trained by supervised learning using historical key stress event data and corresponding stress impact labels as training data. It is used to map the input historical key stress events into stress semantic vectors. Based on the pressure influence model, determine the dynamic scene semantic vector corresponding to each sampling moment on the training time axis; The dynamic scene semantic vector at each sampling time is matrixed to generate a scene semantic representation.

5. The safety training data analysis method based on the Industrial Internet of Things according to claim 4, characterized in that, The step of semantically fusing the operational intrinsic mode function components and the physiological intrinsic mode function components based on the scene semantic representation to obtain a cluster semantic representation includes: Based on the scene semantic representation, the operational intrinsic mode function components are subjected to semantic transformation under stress scene constraints to obtain the scene operational semantic representation. Based on the scene semantic representation, the physiological intrinsic mode function components are subjected to semantic transformation under stress scene constraints to obtain the scene physiological semantic representation. For each cluster, the scene operation semantic representation and the scene physiological semantic representation of the cluster are aggregated across modalities to obtain the behavioral physiological semantic representation; Principal component analysis is performed on the behavioral physiological semantic representations of the same cluster to obtain the cluster semantic representation.

6. The safety training data analysis method based on the Industrial Internet of Things according to claim 5, characterized in that, The step of performing semantic transformation on the operational intrinsic mode function components under stress scene constraints based on the scene semantic representation to obtain the scene operational semantic representation includes: An operational mode matrix is ​​constructed based on each of the operational intrinsic mode function components, wherein the rows of the operational mode matrix represent the mode order, the columns represent the sampling time, and the corresponding matrix element values ​​represent the amplitude; The scene semantic representation is reconstructed to obtain a scene matrix, wherein the rows of the scene matrix represent the stress semantic dimension, the columns represent the sampling time, and the corresponding matrix element values ​​represent the intensity value of the stress semantic dimension at the sampling time. Based on the operation mode matrix, a query vector sequence is generated, and based on the scene matrix, a key vector sequence and a value vector sequence are generated. Calculate the attention scores of the query vector sequence and the key vector sequence to obtain the temporal attention weight matrix; Based on the temporal attention weight matrix, the value vector sequence is weighted and aggregated to obtain a weighted modal representation; The weighted modal representations are aggregated across time to generate scene operation semantic representations.

7. The safety training data analysis method based on the Industrial Internet of Things according to claim 5, characterized in that, The step of performing cross-modal aggregation of the scene operation semantic representation and the scene physiological semantic representation of the cluster to obtain the behavioral physiological semantic representation includes: The scene operation semantic representation and the scene physiological semantic representation are projected onto the same shared semantic space to obtain the scene operation projection vector and the scene physiological projection vector. Based on the scene operation projection vector and the scene physiological projection vector, a cross-modal association graph is constructed, wherein the nodes are composed of the scene operation projection vector and the scene physiological projection vector, and the edge weights between nodes are determined by the mutual information metric. Graph convolution is performed on the cross-modal association graph to fuse operational and physiological modality information within the shared semantic space and update node features; The updated node features are aggregated to obtain a multimodal feature vector; The multimodal feature vectors are compressed to obtain a behavioral physiological semantic representation.

8. The method for analyzing safety training data based on the Industrial Internet of Things according to any one of claims 1-7, characterized in that, The step of performing pattern parsing on the cluster semantic representation based on the personality profile to obtain the group analysis results includes: The personality profile and the cluster semantic representation are correlated and modeled to obtain a mapping model for representing personality and behavioral physiological patterns; Based on the mapping model, the typical behavioral and physiological patterns represented by the cluster semantic representation are analyzed, and the personality profile is traced according to the analysis results to obtain key personality dimensions. Based on the key personality dimensions and the typical behavioral and physiological patterns, a group analysis result is generated for this cluster.

9. A safety training data analysis system based on the Industrial Internet of Things, characterized in that, It includes a management platform, a sensor network platform, and an object platform that are connected in sequence. The management platform is configured with: The clustering module is used to acquire personality test data of the training group and perform cluster analysis on the training group based on the personality test data to obtain at least one cluster and personality profiles of each cluster. The data processing module is used to acquire, for each cluster, the corresponding trainees' operation sequence data, physiological time series data and corresponding scenario protocol time series data under configurable stress scenarios, and to perform time series pattern conversion on the operation sequence data to obtain operation time series data, and to perform semantic extraction on the scenario protocol time series data to obtain scenario semantic representation, wherein the stress scenario includes at least one of time stress, environmental stress and cognitive stress. The decomposition module is used to perform empirical mode decomposition on the operational time series data and the physiological time series data to obtain operational intrinsic mode function components and physiological intrinsic mode function components. The cluster semantic representation generation module is used to perform semantic fusion on the operational intrinsic mode function components and the physiological intrinsic mode function components based on the scene semantic representation to obtain the cluster semantic representation; The analysis result generation module is used to perform pattern parsing on the cluster semantic representation based on the personality profile to obtain group analysis results. The analysis results are used to represent the behavioral reliability, physiological stress patterns, and corresponding training weaknesses of trainees with different personality types under the stress scenario.

10. A computer device, characterized in that, The method includes a memory and a processor, wherein the memory stores a computer program that can run on the processor, and the processor executes the computer program to implement the method of any one of claims 1 to 8.

Images