Ship task execution efficiency evaluation method and device and electronic equipment

Abstract

The invention discloses a ship task execution efficiency evaluation method and device and electronic equipment, and belongs to the technical field of ship control. The method comprises the following steps: constructing a multi-role network diagram system according to an interaction relationship and a dependency relationship among a plurality of task roles of a ship in task execution, the task roles including crew, equipment and environment, the multi-role network diagram system comprises a man-machine system efficiency evaluation model, a human-ring system efficiency evaluation model and a man-machine-ring collaborative system efficiency evaluation model; analyzing the multi-role network graph system, and identifying an efficiency performance portrait and a key node of each task role in cooperative work; obtaining the operation performance, psychological load and scale data of the sailors according to the performance expression portrait and the key nodes; and according to the job performance, the psychological load and the scale data, obtaining an efficiency evaluation result of the collaborative efficiency of ship task execution. The method can comprehensively and accurately evaluate the overall work efficiency in the ship task execution process.

Description

Technical Field

[0001] This application belongs to the field of photovoltaic power generation technology, and in particular relates to a method, device and electronic equipment for evaluating the performance of ship missions. Background Technology

[0002] As modern ship systems become increasingly complex and intelligent, the execution of ship missions is highly dependent on the crew's operation of equipment, the equipment's adaptation to the environment, and the environment's feedback on the crew's status. Therefore, it is necessary to conduct performance evaluations of ship mission execution.

[0003] Traditional assessment methods often focus on a single dimension, such as equipment performance or crew operation, lacking a comprehensive consideration of the synergistic effects of crew, equipment, and environment. This makes it difficult to capture the impact of dynamic changes and key moments, resulting in incomplete assessment results.

[0004] Therefore, a comprehensive and accurate evaluation method for ship mission performance is needed. Summary of the Invention

[0005] This application aims to address at least one of the technical problems existing in the prior art. To this end, this application proposes a method, apparatus, and electronic equipment for evaluating the performance of ship missions, which can comprehensively and accurately evaluate the overall operational efficiency during the execution of ship missions.

[0006] Firstly, this application provides a method for evaluating the effectiveness of ship mission execution, the method comprising: Based on the interaction and dependency relationships between multiple mission roles of a ship during mission execution, a multi-role network graph system is constructed. The mission roles include crew, equipment, and environment. The multi-role network graph system includes a human-machine system effectiveness evaluation model, a human-environment system effectiveness evaluation model, and a human-machine-environment collaborative system effectiveness evaluation model. The multi-role network graph system is analyzed to identify the performance profile and key nodes of each task role in collaborative work; Based on the performance profile and the key points, the crew's work performance, psychological workload, and scale data are obtained. Based on the operational performance, psychological load, and scale data, the effectiveness evaluation results of the collaborative effectiveness of the ship's mission execution are obtained.

[0007] According to one embodiment of this application, the step of constructing a multi-role network graph system based on the interaction and dependency relationships between multiple mission roles of a ship during mission execution includes: The heterogeneous nodes and feature vectors of the task roles are determined. The heterogeneous nodes include crew heterogeneous nodes, equipment heterogeneous nodes and environmental heterogeneous nodes. The crew heterogeneous nodes carry state feature vectors of operational skills and psychological state. The equipment heterogeneous nodes carry attribute feature vectors of performance parameters and fault history. The environmental heterogeneous nodes include environmental parameter feature vectors of sea state and weather. Based on the interaction and dependency relationships between the heterogeneous nodes, a multi-layer adjacency matrix is ​​constructed; Based on the heterogeneous nodes, the state feature vectors, and the multi-layer adjacency matrix, the multi-role network graph system is constructed. The ontology knowledge graph of the ship is converted into triples, and the entity embedding vector is determined in the triples using the TransE algorithm. The entity embedding vector is then concatenated with the feature vector in the multi-layer adjacency matrix to obtain an enhanced node feature representation. In each time window, the temporal changes of the enhanced node feature representation are captured to obtain the temporal-aware node state representation and the interaction events of each time window; Based on the multi-layer adjacency matrix of the first time window and the interaction events of the second time window, the changes in the interaction relationships between the heterogeneous nodes are detected, the multi-layer adjacency matrix is ​​updated, and an incremental adjacency matrix is ​​obtained. The multi-role network graph system is updated based on the time-aware node state representation, the enhanced node feature representation, and the incremental adjacency matrix.

[0008] According to one embodiment of this application, a multi-layer adjacency matrix is ​​constructed based on the interaction and dependency relationships between the heterogeneous nodes, including: Based on the aforementioned dependencies, the topological skeleton of multiple semantic adjacency layers is determined; Based on the topological skeleton, according to the interaction events in the interaction relationship and the feature vectors of the heterogeneous nodes, the dynamic weight of each edge in the semantic adjacency layer is calculated through a real-time state function to generate the corresponding weighted adjacency submatrix. Multiple weighted adjacency sub-matrices are stacked into a three-dimensional adjacency tensor. The importance weight of each semantic layer in the three-dimensional adjacency tensor at the current task stage is learned through a cross-layer graph attention mechanism. Task-adaptive adjacency fusion is then performed to obtain the multi-layer adjacency matrix.

[0009] According to one embodiment of this application, determining the topological skeleton of multiple semantic adjacency layers based on the dependency relationship includes: In the dependency relationship, the key dependency paths of each link in the mission process of the ship are determined, and the human-machine-loop closed-loop dependency chain between the mission roles is extracted. The machine-loop closed-loop dependency chain includes the crew operating equipment main chain, the equipment response environment main chain, and the environmental feedback crew main chain. The machine-loop closed-loop dependency chain is topologically decomposed by dependency strength analysis to identify the semantic adjacency layer, which includes a human-computer interaction layer, a human-loop interaction layer, and a machine-loop interaction layer. Using a dependency propagation algorithm, a topological skeleton based on a directed acyclic graph is constructed for each semantic adjacency layer, wherein the human-computer interaction layer uses operation instructions as edge directions, the human-environment interaction layer uses environmental parameter thresholds as edge directions, and the machine-environment interaction layer uses device performance constraints as edge directions. By performing dependency closure detection, redundant edges are pruned and critical edges are strengthened on the topological skeleton.

[0010] According to one embodiment of this application, the step of calculating the dynamic weight of each edge in the semantic adjacency layer based on the topological skeleton, the interaction events in the interaction relationship, and the feature vectors of the heterogeneous nodes, through a real-time state function, and generating a corresponding weighted adjacency sub-matrix, includes: Based on the edges in the topological skeleton, context aggregation of micro-state, meso-state and macro-state is performed to construct a multi-granularity spatiotemporal state context vector. The micro-state includes the instantaneous attributes of the node, the meso-state includes the behavioral statistical features within the recent time window, and the macro-state includes the semantic label of the current task stage. The semantic adjacency layer is configured with a domain knowledge guidance function module as a real-time state function. In the human-computer interaction layer, the operation matching degree weight between crew members and equipment is calculated by configuring a skill-complexity matching network. In the machine-environment interaction layer, a simplified physical model is embedded to calculate the impact weight of the environment on equipment performance. In the human-environment interaction layer, the disturbance weight of the environment on personnel state is calculated by configuring a piecewise function based on human factors engineering standards. The causality test module verifies the consistency of the edge weights. If the edge weights are found to violate the ship's operating mechanism, an exponential decay penalty is applied to the edge weights to obtain the mechanism calibration weights. Based on the data integrity and interaction event compliance of the feature vectors of the heterogeneous nodes, a confidence factor is calculated. The data integrity is determined by evaluating the online status of sensors and the data missing rate, and the interaction event compliance is determined by rule matching based on the operation procedure knowledge graph. The confidence factor is multiplied by the mechanism calibration weight to obtain the dynamic weight of each edge; The dynamic weights are filled into the adjacent positions corresponding to the topological skeleton, and the unconnected positions are set to zero, generating the weighted adjacency submatrix that corresponds one-to-one with the semantic adjacency layer.

[0011] According to one embodiment of this application, the analysis of the multi-role network graph system to identify the performance profile and key nodes of each task role in collaborative work includes: Based on time-aware node state representation, enhanced node feature vectors, and incremental adjacency matrix, a dynamic heterogeneous time sequence is constructed. A multi-layer heterogeneous graph attention network is used to jointly encode the state evolution of each task role in the dynamic heterogeneous time sequence within multiple time windows to generate a collaborative context embedding vector. The collaborative context embedding vector is input into the human-machine system effectiveness evaluation model, human-environment system effectiveness evaluation model, and human-machine-environment collaborative system effectiveness evaluation model in the role network graph system, respectively, to perform multi-dimensional effectiveness decoupling and obtain dynamic weights and risk factors. The human-machine system effectiveness evaluation model is used to quantify operational effectiveness, the human-environment system effectiveness evaluation model is used to quantify adaptive effectiveness, and the human-machine-environment collaborative system effectiveness evaluation model is used to quantify collaborative effectiveness and robustness effectiveness. The task flow path edges are weighted by the operational efficiency, the information propagation probability is modulated by the collaborative efficiency, the human vulnerability index and environmental sensitivity score are calculated by the adaptive efficiency, and the redundant collaborative paths are dynamically activated and the node failure recovery rate is adjusted by the robust efficiency, thus constructing a dynamic graph attribute with efficiency awareness. Using a dynamic multi-level centrality fusion algorithm, based on the attributes of the dynamic graph, task flow betweenness centrality, information propagation influence, human vulnerability index, and environmental sensitivity scores are calculated on the performance-aware dynamic graph to generate a dynamic criticality score for each task role. Based on the operational efficiency, adaptive efficiency, collaborative efficiency, and robustness efficiency, a performance profile for each task role is determined. Based on the dynamic criticality score, key nodes in collaborative work are identified.

[0012] According to one embodiment of this application, obtaining the crew member's work performance, psychological workload, and scale data based on the performance profile and the key nodes includes: Based on the quantified values ​​of operational effectiveness, adaptive effectiveness, cooperative effectiveness, and robust effectiveness in the aforementioned performance profile, these values ​​are back-mapped to the corresponding original observation dimensions: Based on the dynamic criticality scores of the key nodes and the task stage context, a collaborative attribution graph is constructed. Based on the dynamic criticality scores of heterogeneous crew nodes in the performance perception dynamic graph, and combined with the weighted fusion results of operational performance and collaborative performance, the operational performance is generated. By coupling analysis of environmental sensitivity scores and psychological state feature vectors in the human-environment interaction layer, the psychological load of the crew members under each sea state condition is obtained. The psychological load includes three sub-indicators: cognitive load, emotional stress, and physiological fatigue. A structured questionnaire was constructed using a five-level Likert scale. The scale data was obtained by comparing the operating procedures in the equipment technical manual with actual interaction events. The scale data included job satisfaction, task suitability, and team trust.

[0013] According to one embodiment of this application, obtaining the effectiveness evaluation result of the collaborative effectiveness of the ship's mission execution based on the operational performance, the psychological load, and the scale data includes: The job performance is mapped to a behavioral dimension representation, the psychological load is mapped to a physiological dimension representation, and the scale data is mapped to a task dimension representation. The behavioral dimension representation, the physiological dimension representation, and the task dimension representation are then fused together with temporal alignment and semantic complementarity to generate a collaborative state embedding vector. Based on the cooperative state embedding vector, calculate the multi-granularity cooperative performance index; Based on the semantic tags of the current task stage, the multi-granularity collaborative performance indicators are adaptively weighted and fused to obtain the performance evaluation results.

[0014] Secondly, this application provides a ship mission performance evaluation device, the device comprising: The first processing module is used to construct a multi-role network graph system based on the interaction and dependency relationships between multiple task roles of the ship in mission execution. The task roles include crew, equipment and environment. The multi-role network graph system includes a human-machine system effectiveness evaluation model, a human-environment system effectiveness evaluation model and a human-machine-environment collaborative system effectiveness evaluation model. The second processing module is used to analyze the multi-role network graph system and identify the performance profile and key nodes of each task role in collaborative work. The third processing module is used to obtain the crew's work performance, psychological load, and scale data based on the performance profile and the key nodes. The fourth processing module is used to obtain the effectiveness evaluation result of the collaborative effectiveness of the ship's mission execution based on the operational performance, the psychological load, and the scale data.

[0015] Thirdly, this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the ship mission performance evaluation method as described in the first aspect above.

[0016] Fourthly, this application provides a non-transitory computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the ship mission performance evaluation method as described in the first aspect above.

[0017] Fifthly, this application provides a chip including a processor and a communication interface, the communication interface being coupled to the processor, the processor being used to run programs or instructions to implement the ship mission execution performance evaluation method as described in the first aspect.

[0018] In a sixth aspect, this application provides a computer program product, including a computer program that, when executed by a processor, implements the ship mission performance evaluation method as described in the first aspect above.

[0019] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application.

[0020] The method, apparatus, and electronic equipment for evaluating the performance of ship missions provided in this application have the following advantages over the prior art: (1) By analyzing the interaction and dependency relationships between crew, equipment and environment, a multi-role network diagram system is constructed to analyze the performance profile and key nodes. By integrating operational performance, psychological load and scale data, the collaborative effectiveness between roles is quantified and analyzed, dynamically adapting to task changes. This allows for a comprehensive and accurate assessment of the overall work efficiency of ship missions during multi-task and multi-role collaboration.

[0021] (2) By constructing a dynamic multi-layer network graph system that integrates the heterogeneous roles of crew, equipment and environment, and combining ontology knowledge graph enhancement and real-time interactive perception, a comprehensive evaluation of the ship’s human-machine-environment collaborative efficiency with high precision, interpretability and second-level updates is achieved. Compared with traditional static or single-dimensional methods, it improves the evaluation accuracy, multi-source data fusion capability, key bottleneck attribution capability and dynamic adaptability to complex sea conditions and mission changes. Attached Figure Description

[0022] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which: Figure 1This is a flowchart illustrating the ship mission performance evaluation method provided in the embodiments of this application; Figure 2 This is a schematic diagram of the structure of the multi-role network graph system provided in the embodiments of this application; Figure 3 This is a schematic diagram of the structure of the ship mission performance evaluation device provided in the embodiments of this application; Figure 4 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0023] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0024] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0025] The following description, in conjunction with the accompanying drawings, details the ship mission execution performance evaluation method, ship mission execution performance evaluation device, electronic equipment, and readable storage medium provided in this application through specific embodiments and application scenarios.

[0026] Among them, the ship mission execution effectiveness evaluation method can be applied to the terminal, specifically by the hardware or software in the terminal.

[0027] The terminal includes, but is not limited to, portable communication devices such as mobile phones or tablets with touch-sensitive surfaces (e.g., touchscreen displays and / or touchpads). It should also be understood that, in some embodiments, the terminal may not be a portable communication device, but rather a desktop computer with touch-sensitive surfaces (e.g., touchscreen displays and / or touchpads).

[0028] The following embodiments describe a terminal including a display and a touch-sensitive surface. However, it should be understood that the terminal may include one or more other physical user interface devices such as a physical keyboard, mouse, and joystick.

[0029] The ship mission execution performance evaluation method provided in this application embodiment can be implemented by an electronic device or a functional module or entity in an electronic device that can realize the ship mission execution performance evaluation method. The electronic devices mentioned in this application embodiment include, but are not limited to, mobile phones, tablets, computers, cameras and wearable devices. The ship mission execution performance evaluation method provided in this application embodiment will be described below using an electronic device as the implementation subject.

[0030] like Figure 1 As shown, the ship's mission performance evaluation method includes: Step 110: Based on the interaction and dependency relationships between multiple mission roles of the ship during mission execution, construct a multi-role network graph system. The mission roles include crew, equipment, and environment. The multi-role network graph system includes a human-machine system performance evaluation model, a human-environment system performance evaluation model, and a human-machine-environment collaborative system performance evaluation model.

[0031] A mission role is an entity that performs a specific function during the execution of a ship's mission.

[0032] In terms of mission roles, crew members are personnel who perform various operational, command, and support tasks on ships. Their operational skills, psychological qualities, and work efficiency directly affect the progress of various tasks on the ship. Equipment refers to the various hardware facilities and systems on the ship, such as power systems, weapon systems, communication systems, and navigation systems. The environment refers to environmental factors that affect the work performance of crew members and the effectiveness of equipment, including the sea state and weather conditions where the ship is located, such as the size of the waves, temperature, and humidity, as well as the electromagnetic environment, such as the interference of surrounding electromagnetic signals, and the battlefield environment, such as the degree of enemy firepower threat and the battlefield situation.

[0033] Interactions are actual operational, communication, or physical impact events that occur between task roles, such as a crew member pressing a button on equipment or high waves causing increased equipment vibration, which can be captured in real time through sensor logs, operation records, or video analytics.

[0034] Dependency is a logical sequence relationship defined by operating procedures or physical constraints in a task flow.

[0035] In step 110, a thorough analysis of the mission to be performed by the ship is conducted, identifying all involved crew members, equipment, and environmental factors as mission roles. For example, for an anti-submarine warfare mission, the participating crew members are identified as sonar operators, anti-submarine weapon operators, and command and decision-making personnel; the equipment includes sonar systems, anti-submarine torpedo launching devices, and the ship's power supply equipment; and the environmental factors include the hydrographic conditions of the sea area and the possible activities of enemy submarines in the surrounding area.

[0036] By examining ship mission execution procedures, relevant technical documents, and operational doctrines, the interaction methods and interdependencies among mission roles during mission completion are determined. For example, the sonar operator needs to transmit the location information of detected enemy submarines to the anti-submarine weapon operator, representing an information exchange relationship between the two; while the normal launch of anti-submarine weapons depends on the stable operation of the power supply equipment, representing an interdependence between the equipment.

[0037] Using the interaction and dependency relationships between mission roles as connecting lines, and various mission roles as nodes, a multi-role network graph system is constructed. This system can intuitively show the cooperation and mutual influence between various roles, as well as their respective roles and positions in completing the overall mission of the ship.

[0038] like Figure 2 As shown, in the multi-role network diagram system, for the human-machine system, the human-environment system, and the multi-person-machine-environment collaborative system, the control variables of each system are clearly defined, such as ambient temperature, humidity, noise level, lighting intensity, and personnel workload, to ensure the controllability and reliability of each variable during the experiment.

[0039] Human-machine system effectiveness assessment models consider the interaction between crew and equipment, including crew proficiency in operating equipment, the speed and accuracy of equipment response to crew commands, and the impact of equipment operating status on crew work. For example, a model can be established to measure the time required for crew to detect a target and complete firing when operating a naval gun system, as well as the hit rate, to assess the effectiveness of human-machine collaborative combat.

[0040] The human-environment system performance evaluation model is used to model the work performance of crew members under different environmental conditions, such as high temperature, high humidity, high salinity marine environment, or complex electromagnetic interference environment, for example, the work efficiency and psychological state changes of crew members under adverse sea conditions.

[0041] The human-machine-environment collaborative system performance evaluation model comprehensively considers the interaction and collaborative work among crew, equipment, and environment to measure the overall effectiveness of a ship in completing its missions. For example, in a ship's daily patrol missions, the model analyzes the impact of crew watch status, equipment operational support, and the surrounding marine environment on patrol effectiveness, constructing a model that comprehensively considers these three factors. The model parameters and evaluation methods are determined through data collection and analysis of actual patrol missions.

[0042] Step 120: Analyze the multi-role network graph system to identify the performance profile and key nodes of each task role in collaborative work.

[0043] The performance profile is a multi-dimensional performance profile of a task role in the current task phase, based on the quantitative values ​​of operational performance, adaptive performance, collaborative performance, and robust performance.

[0044] Key nodes are those task roles or specific interactions between them that have a significant impact on the collaborative work efficiency of ships in a multi-role network graph system, and can be identified through a dynamic multi-level centrality fusion algorithm.

[0045] In step 120, computer simulation software is used to simulate the performance evaluation model in the constructed multi-role network graph system, simulating the ship's operation under different mission scenarios. Through the reactions and interactions of each mission role, their performance profiles are analyzed. Node centrality analysis of the multi-role network graph system identifies which nodes are in critical positions, have numerous connections with other nodes, and significantly impact information transmission and task flow. Path analysis identifies important nodes for completing tasks and the critical paths formed by these connections, thus recognizing the performance profile and key nodes of each mission role in collaborative work. For example, by calculating the degree centrality of each node in the multi-role network graph, a crew member in a key position is identified as having a high degree centrality, indicating that they play a crucial bridging role in information communication and task collaboration, and are likely a key node.

[0046] Step 130: Based on the performance profile and the key nodes, obtain the crew's work performance, psychological workload, and scale data; Operational performance is used to characterize the work results and performance of crew members in the process of performing tasks, including efficiency, accuracy, number of tasks completed, operational accuracy rate, and consistency indicators of equipment status.

[0047] Psychological load is used to characterize the degree of psychological stress and mental burden experienced by crew members when performing tasks, including psychological tension and anxiety caused by factors such as work intensity, task difficulty, and changes in the surrounding environment, as well as cognitive load index and emotional arousal based on physiological signal inversion.

[0048] The scale data is used to characterize subjective feelings such as job satisfaction and confidence in the task, psychological states such as emotional state and fatigue level, and work performance such as operational proficiency. The data is obtained by quantifying these aspects. It includes scores from the post-task subjective evaluation scale and the standardized human factors engineering questionnaire, which are used to help assess the crew's work situation and psychological state, and reflect the ship's effectiveness.

[0049] In step 130, the quantitative values ​​of the operational effectiveness and collaborative effectiveness of the crew member node are extracted from the performance profile. The dynamic criticality score of the crew member is obtained from the critical node analysis. The current mission phase context is obtained to dynamically adjust the weights of collaborative effectiveness, operational effectiveness, and dynamic criticality score, and to calculate the crew member's operational performance. In highly collaborative missions such as joint search and rescue, the weights of collaborative effectiveness and criticality score will increase. In high-precision operational missions such as weapon aiming, the weight of operational effectiveness will increase.

[0050] In the human-environment interaction layer, the environmental sensitivity score of the crew member node is determined to quantify the potential impact of the environment on the crew member.

[0051] The psychological state feature vector of the crew is determined by the self-crew heterogeneous node, including real-time data obtained through physiological sensors and eye trackers, and the initial indicators of cognitive load, emotional stress and physical fatigue are derived.

[0052] By coupling environmental sensitivity scores, environmental parameters, and psychological state feature vectors, and relating them to the psychological load of crew members, an environment-psychology coupling function is constructed.

[0053] The cognitive load is obtained by calling the environment-psychology coupling function and multiplying the task complexity by a gain coefficient modulated by the environmental sensitivity score and the current environmental parameters. The emotional stress is obtained by calculating unpredictable and sudden stressful events in the environment and combining them with the emotional stability extracted from the crew's psychological state feature vector. The physiological fatigue is obtained by calculating the task duration, environmental physical intensity, and environmental sensitivity score. The psychological load of the crew is obtained based on the cognitive load, emotional stress, and physiological fatigue.

[0054] Step 140: Based on the operational performance, the psychological load, and the scale data, obtain the effectiveness evaluation results of the collaborative effectiveness of the ship's mission execution.

[0055] Synergy effectiveness is the overall effectiveness generated by the cooperation and coordination of various mission roles on a ship. It reflects the effect and capability of the ship's various elements working together and is a key indicator for measuring whether a ship can efficiently complete complex tasks.

[0056] In step 140, based on the characteristics of operational performance, psychological load, and scale data, and their respective importance in the collaborative effectiveness assessment, a comprehensive assessment model based on weighted summation is determined. The collected operational performance data of the crew, the processed psychological load data, and the scale data are used in the comprehensive assessment model to calculate the ship's collaborative effectiveness quantitative score, which serves as the effectiveness assessment result of collaborative effectiveness.

[0057] Furthermore, the calculated collaborative effectiveness results are analyzed in depth to identify key factors and weaknesses affecting ship collaborative effectiveness. For example, if the collaborative effectiveness score is low, analysis of various data sources may reveal that it is due to poor operational performance or excessive psychological burden on certain crew members. These analytical results are then fed back to ship management and relevant personnel to enable targeted improvement measures, such as strengthening crew training, improving the working environment, and optimizing task allocation, thereby enhancing the overall collaborative effectiveness of the ship.

[0058] According to the ship mission execution performance evaluation method provided in the embodiments of this application, by analyzing the interaction and dependency relationships between crew, equipment and environment, a multi-role network diagram system is constructed, the performance profile and key nodes are analyzed, and the collaborative performance between each role is quantified and analyzed by integrating operational performance, psychological load and scale data, dynamically adapting to mission changes, and comprehensively and accurately evaluating the overall work efficiency in the ship mission execution process for multi-task and multi-role collaboration.

[0059] In some embodiments, constructing a multi-role network graph system based on the interaction and dependency relationships between multiple mission roles of a ship during mission execution includes: The heterogeneous nodes and feature vectors of the task roles are determined. The heterogeneous nodes include crew heterogeneous nodes, equipment heterogeneous nodes and environmental heterogeneous nodes. The crew heterogeneous nodes carry state feature vectors of operational skills and psychological state. The equipment heterogeneous nodes carry attribute feature vectors of performance parameters and fault history. The environmental heterogeneous nodes include environmental parameter feature vectors of sea state and weather. Based on the interaction and dependency relationships between the heterogeneous nodes, a multi-layer adjacency matrix is ​​constructed; Based on the heterogeneous nodes, the state feature vectors, and the multi-layer adjacency matrix, the multi-role network graph system is constructed. The ontology knowledge graph of the ship is converted into triples, and the entity embedding vector is determined in the triples using the TransE algorithm. The entity embedding vector is then concatenated with the feature vector in the multi-layer adjacency matrix to obtain an enhanced node feature representation. In each time window, the temporal changes of the enhanced node feature representation are captured to obtain the temporal-aware node state representation and the interaction events of each time window; Based on the multi-layer adjacency matrix of the first time window and the interaction events of the second time window, the changes in the interaction relationships between the heterogeneous nodes are detected, the multi-layer adjacency matrix is ​​updated, and an incremental adjacency matrix is ​​obtained. The multi-role network graph system is updated based on the time-aware node state representation, the enhanced node feature representation, and the incremental adjacency matrix.

[0060] It is understandable that heterogeneous nodes include crew heterogeneous nodes, equipment heterogeneous nodes, and environment heterogeneous nodes, each carrying its own state feature vector, attribute feature vector, or environmental parameter feature vector. The feature vector is a numerical vector used to represent the node's attributes. For example, the crew feature vector includes operational skills and psychological state, the equipment feature vector includes performance parameters and fault history, and the environmental feature vector includes sea state and weather.

[0061] A multi-level adjacency matrix is ​​a stack of multiple adjacency matrices, each matrix corresponding to a semantic layer, representing the connections of nodes under different relationships.

[0062] An ontology knowledge graph is a structured knowledge representation of a ship, containing entities, attributes, and relationships in the ship domain, such as the structured knowledge of equipment specifications and operating procedures in an equipment technical manual.

[0063] A triple is the basic unit of a knowledge graph, in the form of (head entity, relation, tail entity), for example, (device A, operation method, specification B).

[0064] The TransE algorithm is a knowledge graph embedding algorithm that learns low-dimensional vector representations of entities and relations by minimizing the distance between the head entity plus relation and the tail entity. The entity embedding vector is a low-dimensional numerical representation obtained by the TransE algorithm, capturing semantic information.

[0065] Enhanced node feature representation is a new feature vector obtained by concatenating the original feature vector with the entity embedding vector, which is used to capture temporal changes and enrich the semantic information of nodes.

[0066] Time-aware node state representation is a node feature representation that takes into account historical states, reflecting time-series dependencies and evolutionary patterns; interactive events are specific interactions that occur between nodes, such as crew members operating equipment or environmental alarms.

[0067] The first and second time windows are two adjacent time windows; the incremental adjacency matrix is ​​an adjacency matrix that only contains the changed parts, and is used to efficiently update the network structure.

[0068] In actual execution, task roles are classified into heterogeneous nodes. The feature vector of each heterogeneous node is obtained in real time through sensors, logs or evaluation tools. Crew heterogeneous nodes carry state feature vectors such as operational skills and psychological state, equipment heterogeneous nodes carry attribute feature vectors such as performance parameters and fault history, and environmental heterogeneous nodes carry environmental parameter feature vectors such as sea state and weather.

[0069] By capturing crew operation equipment events, equipment response to environmental events, and environmental impact on personnel events through real-time event streams, interaction relationships are obtained. Dependencies are extracted from a structured operating procedure knowledge base, and three semantic adjacency layers are defined, each corresponding to an interaction type, including human-computer interaction layer, human-environment interaction layer, and machine-environment interaction layer. Each layer corresponds to an adjacency submatrix, and the elements in the matrix represent the connection strength or probability between nodes. The three submatrices are stacked into a three-dimensional tensor to obtain a multi-layer adjacency matrix.

[0070] In some embodiments, a multi-layer adjacency matrix is ​​constructed based on the interaction and dependency relationships between the heterogeneous nodes, including: Based on the aforementioned dependencies, the topological skeleton of multiple semantic adjacency layers is determined; Based on the topological skeleton, according to the interaction events in the interaction relationship and the feature vectors of the heterogeneous nodes, the dynamic weight of each edge in the semantic adjacency layer is calculated through a real-time state function to generate the corresponding weighted adjacency submatrix. Multiple weighted adjacency sub-matrices are stacked into a three-dimensional adjacency tensor. The importance weight of each semantic layer in the three-dimensional adjacency tensor at the current task stage is learned through a cross-layer graph attention mechanism. Task-adaptive adjacency fusion is then performed to obtain the multi-layer adjacency matrix.

[0071] Understandably, the topological skeleton is a static directed graph structure for each semantic adjacency layer, defining the allowed connections between heterogeneous nodes in each semantic layer, without weights.

[0072] The real-time state function is a computable mapping function. The input is the current interaction event and the node state, and the output is the dynamic weight of the edge, which reflects the effectiveness or strength of the current interaction.

[0073] The semantic adjacency layer is based on different layers of the topological skeleton, including the human-computer interaction layer, the human-environment interaction layer, and the machine-environment interaction layer.

[0074] The weighted adjacency submatrix is ​​a matrix formed by assigning weight values ​​to each edge based on the topological skeleton, with unconnected positions set to 0.

[0075] A three-dimensional adjacency tensor is a tensor formed by stacking three weighted adjacency submatrices along the third dimension, preserving semantic hierarchical information.

[0076] The cross-layer graph attention mechanism is a learnable fusion module that automatically calculates the importance weights of each semantic layer based on the semantics of the current task stage, thereby achieving task-adaptive fusion.

[0077] Task-adaptive adjacency fusion compresses a three-dimensional tensor into a single adjacency matrix with weights dynamically adjusted according to the task, allowing the graph structure to focus on the current key interaction type.

[0078] In actual implementation, human-machine dependency rules, human-environment dependency rules, and machine-environment dependency rules are extracted from the ship structured operation procedure knowledge base; for each dependency rule, a corresponding Boolean adjacency matrix is ​​constructed as the topological skeleton of the three layers of human-machine, human-environment, and machine-environment; for each semantic layer, all existing edges in its topological skeleton are traversed and dynamic weights are calculated respectively. In the human-computer interaction layer, the real-time state function is:

[0079] in, In the human-computer interaction layer, from the crew Pointing device Dynamic edge weights; crew members Status includes: skill level Current psychological load ;equipment The states include: operational complexity Current fault warning signs Interaction events are the frequency of operations within the past 10 seconds. ; For the Sigmoid function, This is the first layer weight matrix of the pre-trained MLP, used to calculate the matching degree between skill and complexity.

[0080] In the human-environment interaction layer, the real-time state function is:

[0081] in, In the human-environment interaction layer, from the environment node Pointing device Dynamic edge weights; environment nodes Parameters include: wave height The unit is meters (m) and wind speed. The unit is m / s; crew Conditions include: history of seasickness Current Heart Rate Variability (HRV) value.

[0082] In the machine-ring interaction layer, the real-time state function is:

[0083] in, In the machine-environment interaction layer, from the environment node Pointing to the crew Dynamic edge weights; environmental parameters include wind speed. and equipment environmental sensitivity coefficient .

[0084] For each semantic adjacency layer Generate a weighted adjacency submatrix:

[0085] in, It is the first l Nodes in a semantic layer i To the node j The connection weights, including , and ; It is the first l The topological skeleton of a semantic layer Middle node i and j There are potential connections within this semantic layer.

[0086] Stacking the three weighted adjacency sub-matrices along the third dimension yields a three-dimensional adjacency tensor. :

[0087] in, The first l A weighted adjacency submatrix of semantic layers.

[0088] For each semantic layer adjacency Calculate its global importance score:

[0089] in, It is a learnable query vector used to calculate the importance score of each semantic layer; It is a learnable weight matrix in the cross-layer attention module, used to fuse interaction intensity and task semantics; For the weighted adjacency submatrix The arithmetic mean of all elements is used to reflect the overall interaction strength. It is the semantic embedding vector of the current task stage, which is obtained by mapping the task label through a pre-trained embedding layer.

[0090] Calculate the first l Normalized attention weights for each semantic layer:

[0091] in, It is the first l The unnormalized importance score of each semantic layer.

[0092] The weighted fusion yields the final multi-layer adjacency matrix. :

[0093] In this embodiment, by constructing a multi-semantic topological skeleton based on dependencies, integrating real-time interactive events and node states to dynamically calculate edge weights, and combining a task-phase adaptive cross-layer graph attention mechanism, accurate modeling and dynamic representation of the complex human-machine-environment interaction relationship of the ship are achieved. Compared with traditional static or single-layer graph methods, this improves the graph structure's ability to perceive the task context and the rationality of weight allocation, thereby enhancing the system's real-time performance, adaptability, and interpretability.

[0094] The set of heterogeneous nodes, the set of feature vectors, and the multi-level adjacency matrix are packaged into a dynamic heterogeneous graph object, which is denoted as the initial multi-role network graph system. The spaCy+ rule template is used as an information extraction tool to extract structured knowledge from the ship ontology knowledge graph and convert it into a triple form (head entity-relationship-tail entity).

[0095] The TransE algorithm is used to learn the embedding vector of each entity (heterogeneous node) in the triplet. The entity embedding vector is concatenated with the feature vector of the embedded node to obtain the enhanced node feature representation, thereby incorporating domain knowledge.

[0096] Within each time window, the enhancement features of heterogeneous nodes are monitored and enhanced. The evolutionary patterns are captured using a Long Short-Term Memory (LSTM) time series model to obtain a time series-aware node state representation. At the same time, the interaction events within the window are recorded through an event list.

[0097] Compare the interaction events in the first time window and the second time window to detect changes in the interaction relationship. If an event does not appear in the first time window but appears in the second time window and continues to appear in the next two time windows, it is determined to be a new interaction. If the event does not appear in three consecutive windows, it is determined to be a lost interaction. Based on whether the interaction event is a new interaction or a lost interaction, perform incremental local updates on the multi-level adjacency matrix, retain the changed parts, and obtain the incremental adjacency matrix.

[0098] Based on the time-aware node state representation, enhanced node feature representation, and incremental adjacency matrix, the node state and edge relationships of the multi-role network graph system are updated to ensure that the system dynamically adapts to changes.

[0099] In this embodiment, by constructing a dynamic multi-layer network graph system that integrates heterogeneous roles of crew, equipment and environment, and combining ontology knowledge graph enhancement with real-time interactive perception, a comprehensive assessment of the ship's human-machine-environment collaborative efficiency with high precision, interpretability and second-level updates is achieved. Compared with traditional static or single-dimensional methods, this improves the accuracy of assessment, multi-source data fusion capability, key bottleneck attribution capability and dynamic adaptability to complex sea conditions and mission changes.

[0100] In some embodiments, determining the topological skeleton of multiple semantic adjacency layers based on the dependencies includes: In the dependency relationship, the key dependency paths of each link in the mission process of the ship are determined, and the human-machine-loop closed-loop dependency chain between the mission roles is extracted. The machine-loop closed-loop dependency chain includes the crew operating equipment main chain, the equipment response environment main chain, and the environmental feedback crew main chain. The machine-loop closed-loop dependency chain is topologically decomposed by dependency strength analysis to identify the semantic adjacency layer, which includes a human-computer interaction layer, a human-loop interaction layer, and a machine-loop interaction layer. Using a dependency propagation algorithm, a topological skeleton based on a directed acyclic graph is constructed for each semantic adjacency layer, wherein the human-computer interaction layer uses operation instructions as edge directions, the human-environment interaction layer uses environmental parameter thresholds as edge directions, and the machine-environment interaction layer uses device performance constraints as edge directions. By performing dependency closure detection, redundant edges are pruned and critical edges are strengthened on the topological skeleton.

[0101] It is understandable that the human-machine-environment closed-loop dependency chain is a feedback loop structure formed during task execution, including the crew operating equipment main chain, the equipment responding to the environment main chain, and the environment feeding back to the crew main chain.

[0102] The crew-operated equipment main chain describes the process from crew issuing instructions to equipment execution; the equipment response environment main chain describes the process from environmental disturbances to changes in equipment performance; and the environmental feedback crew main chain describes the process from environmental changes to crew perception / state changes.

[0103] Dependency strength analysis quantifies the importance of each dependency in a closed-loop chain, calculating strength values ​​based on indicators such as operation frequency, task criticality, and security level.

[0104] Topological decomposition splits closed-loop dependency chains into three independent semantic subgraphs based on interaction type: human-machine, human-loop, and machine-loop.

[0105] Dependency transitive algorithms are based on path propagation rules in directed graphs, ensuring that dependencies are acyclic and non-contradictory in the graph. For example, topological sorting can be used to handle the order of operations.

[0106] In the topological skeleton of the directed acyclic graph, the static structure of each semantic layer is a directed acyclic graph, which ensures that the task logic has no circular dependencies and conforms to the timing constraints of the ship operation procedures.

[0107] At the human-machine layer, the edges point from the crew to the equipment, describing the flow of operating instructions; at the human-environment layer, the edges point from the environment to the crew, describing the flow of environmental influences; at the machine-environment layer, the edges point from the environment to the equipment, describing the flow of environmental disturbances.

[0108] Dependency closure detection is used to check whether there are redundant paths in the graph. For example, if A→B→C and A→C exist simultaneously, low-intensity redundant edges are identified and pruned through path coverage analysis, while high-intensity critical edges are weighted or their structure is preserved.

[0109] In actual execution, key task chains are extracted from the ship mission process knowledge graph. The graph stores operation steps and their dependencies in the form of triples. The process nodes are traversed to identify triple sequences that form closed loops, and their paths are combined into human-machine loop closed-loop dependency chains and stored in the closed-loop chain set. For each dependent edge in the closed-loop chain, calculate the dependency strength and cluster them according to edge type. Assign "crew → equipment" to the human-computer interaction layer, "environment → crew" to the human-environment interaction layer, and "environment → equipment" to the machine-environment interaction layer, and output three edge sets. For the edge set of each semantic layer, a directed graph is constructed, and the Kahn algorithm is used for topological sorting to detect the existence of cycles. If a cycle exists, high-strength edges are retained and low-strength edges are deleted to break the cycle, ensuring that each semantic layer is a directed acyclic graph, which serves as the topological skeleton of that layer.

[0110] For each semantic layer, perform path closure analysis and use the Floyd-Warshall algorithm to calculate the reachability matrix of all node pairs. If there is a direct edge and a path with a length ≥ 2, it is determined to be a redundant edge. If the strength of a direct edge is less than the preset strength threshold, the edge is deleted; if the edge strength ranks in the top 10%, it is assigned a lower limit of the basic weight in the subsequent dynamic weight calculation to obtain the optimized topological skeleton.

[0111] In this embodiment, by extracting the human-machine loop closed-loop dependency chain from the ship's mission flow, and combining dependency strength analysis and closure detection, a three-layer directed acyclic graph topological skeleton with acyclicity, redundancy, and semantic clarity is constructed. This not only strictly follows the logical constraints of the operating procedures, but also significantly improves the simplification and task relevance of the graph structure through key edge strengthening and redundancy pruning. This provides a high-fidelity, low-noise structural foundation for subsequent dynamic weight calculation and collaborative performance evaluation, improves the accuracy of key interaction path identification and graph reasoning efficiency, and allows each semantic layer to independently represent specific interaction patterns and form a complete performance transfer network through dependency relationships.

[0112] In some embodiments, the step of calculating the dynamic weights of edges in each semantic adjacency layer based on the topological skeleton, according to the interaction events in the interaction relationship and the feature vectors of the heterogeneous nodes, and generating a corresponding weighted adjacency sub-matrix, includes: Based on the edges in the topological skeleton, context aggregation of micro-state, meso-state and macro-state is performed to construct a multi-granularity spatiotemporal state context vector. The micro-state includes the instantaneous attributes of the node, the meso-state includes the behavioral statistical features within the recent time window, and the macro-state includes the semantic label of the current task stage. The semantic adjacency layer is configured with a domain knowledge guidance function module as a real-time state function. In the human-computer interaction layer, the operation matching degree weight between crew members and equipment is calculated by configuring a skill-complexity matching network. In the machine-environment interaction layer, a simplified physical model is embedded to calculate the impact weight of the environment on equipment performance. In the human-environment interaction layer, the disturbance weight of the environment on personnel state is calculated by configuring a piecewise function based on human factors engineering standards. The causality test module verifies the consistency of the edge weights. If the edge weights are found to violate the ship's operating mechanism, an exponential decay penalty is applied to the edge weights to obtain the mechanism calibration weights. Based on the data integrity and interaction event compliance of the feature vectors of the heterogeneous nodes, a confidence factor is calculated. The data integrity is determined by evaluating the online status of sensors and the data missing rate, and the interaction event compliance is determined by rule matching based on the operation procedure knowledge graph. The confidence factor is multiplied by the mechanism calibration weight to obtain the dynamic weight of each edge; The dynamic weights are filled into the adjacent positions corresponding to the topological skeleton, and the unconnected positions are set to zero, generating the weighted adjacency submatrix that corresponds one-to-one with the semantic adjacency layer.

[0113] It is understandable that the microstate is the instantaneous attribute value of a heterogeneous node at the current moment, such as the current heart rate of the crew, the current temperature of the equipment, and the current wave height of the environment, which comes from real-time sensors or state monitoring systems.

[0114] The meso-level state refers to the statistical characteristics of the behavior of heterogeneous nodes within a certain time window, such as the frequency of crew operations, the number of equipment failure alarms, and the standard deviation of environmental parameter fluctuations, reflecting short-term trends.

[0115] Macro-state is the semantic label of the current mission phase of a ship, such as combat, cruising, emergency damage control, and port replenishment. It comes from the mission management system and is used to provide contextual semantics.

[0116] Multi-granularity spatiotemporal state context vector integrates micro, meso, and macro state information into a unified vector, which serves as the context input for edge weight calculation.

[0117] The domain knowledge guidance function module is a dedicated weight calculation model designed for different semantic layers. It integrates engineering rules and data-driven methods. For the human-machine layer, it uses a skill-complexity matching network built based on multilayer perceptron (MLP); for the machine-environment layer, it uses a simplified physical model; and for the human-environment layer, it uses a piecewise function based on human factors standards.

[0118] The causality verification module is based on a rule base of ship operation mechanisms. For example, fatigued personnel are prohibited from operating high-risk equipment. It is used to verify the rationality of the calculated edge weights, and penalties are imposed if violations are found.

[0119] Mechanism calibration weights are edge weights corrected by causal testing to ensure compliance with physical and human factors logic.

[0120] The confidence factor is a coefficient that measures the credibility of edge weights, including data integrity and interaction event compliance. Data integrity is affected by factors such as whether the sensor is online and whether the data missing rate is below the threshold. Interaction event compliance is determined by matching knowledge graph rules to determine whether the current interaction conforms to the operating procedures.

[0121] The weighted adjacency submatrix is ​​formed by filling dynamic weights into the corresponding positions on the topological skeleton and setting unconnected edges to zero, thus creating a weighted graph matrix.

[0122] In actual execution, for each existing edge in the topological skeleton, the state information of the nodes at both ends is collected. For the micro-state, the sensor or state monitoring values ​​at the current moment are read in real time, such as the crew's heart rate, psychological stress score, equipment temperature, vibration value, and environmental wave height and wind speed; for the meso-state, the behavioral characteristics within the past 15-second time window are statistically analyzed, such as the number of times the crew operated the equipment, the average response time, the frequency of the equipment triggering alarms, and the amplitude and standard deviation of environmental parameter changes; for the macro-state, the system obtains the current mission stage label from the ship's mission management system and uses this label as semantic context, concatenating the above three types of state information into a unified context vector to obtain a multi-granularity spatiotemporal state context vector.

[0123] For the edges in the human-computer interaction layer, a pre-trained multilayer perceptron model is called as a skill-complexity matching network. The input is the crew's operational skill level and the equipment's operational complexity level, and the output is the operational matching degree between 0 and 1, which is used as the initial weight.

[0124] For edges in the machine-ring interaction layer, a simplified physical model is embedded. For example, based on the current wind speed and the equipment's wind load sensitivity coefficient, the degree of influence of the environment on the equipment performance is calculated using a preset exponential decay formula, and used as the initial weight.

[0125] For the edges in the human-environment interaction layer, a piecewise function based on human factors engineering standards is adopted. For example, when the wave height exceeds 2.5 meters, the environmental disturbance weight is linearly increased at different slopes depending on whether the crew has a history of seasickness; if the threshold is not exceeded, the weight is zero.

[0126] The ship operation mechanism rule base contains several safety and operational logic rules. For example, crew members with a psychological stress score higher than 0.8 are not allowed to operate weapon equipment, and equipment is considered unusable when the temperature exceeds 80 degrees Celsius.

[0127] The interaction scenario corresponding to the currently calculated edge weight is matched with the rule base. If an edge is found to violate any rule, it is determined to be inconsistent in mechanism, and an exponential decay penalty is applied to the edge weight, such as multiplying it by 0.1 or 0.01, to obtain the calibrated mechanism calibration weight; if no rule is violated, the weight remains unchanged.

[0128] First, data integrity is assessed by checking if the sensors involved in edge computing are online and calculating the percentage of missing data in the past 10 seconds. If the missing rate exceeds 30%, the data integrity score is low. Second, the compliance of the interaction event is assessed by converting the current interaction event into a structured triple and querying the ship's operating procedures knowledge graph to see if a corresponding authorization relationship exists. If it exists, the compliance score is 1; otherwise, it is 0. Finally, the confidence factor is equal to the product of the data integrity score and the compliance score, and its value ranges from 0 to 1.

[0129] The confidence factor is multiplied by the mechanism calibration weight to obtain the final dynamic weight of the edge. The dynamic weight reflects the physical rationality of the interaction, data reliability, and procedural compliance.

[0130] For each of the human-computer interaction layer, human-loop interaction layer, and machine-loop interaction layer, a zero matrix of the same size as the total number of nodes is initialized. All edges in the topological skeleton of each layer are traversed. If a position is 1 in the topological skeleton, there is a dependency relationship, and the corresponding final dynamic weight is filled into that position; if it is 0, there is no dependency relationship, and it is kept as 0. Three weighted adjacency sub-matrices are generated that correspond one-to-one with the semantic adjacency layer, representing the dynamic intensity of the three types of human-computer, human-loop, and machine-loop interactions at the current time.

[0131] In this embodiment, by integrating microscopic instantaneous states, mesoscopic behavioral statistics, and macroscopic task semantics, a multi-granularity context-aware edge weight calculation mechanism is constructed. Combined with domain knowledge guidance functions, mechanism causal verification, and confidence assessment, a dynamic quantification of the intensity of human-machine, human-environment, and machine-environment interactions with high fidelity, interpretability, and compliance is achieved.

[0132] In some embodiments, the analysis of the multi-role network graph system to identify the performance profile and key nodes of each task role in collaborative work includes: Based on time-aware node state representation, enhanced node feature vectors, and incremental adjacency matrix, a dynamic heterogeneous time sequence is constructed. A multi-layer heterogeneous graph attention network is used to jointly encode the state evolution of each task role in the dynamic heterogeneous time sequence within multiple time windows to generate a collaborative context embedding vector. The collaborative context embedding vector is input into the human-machine system effectiveness evaluation model, human-environment system effectiveness evaluation model, and human-machine-environment collaborative system effectiveness evaluation model in the role network graph system, respectively, to perform multi-dimensional effectiveness decoupling and obtain dynamic weights and risk factors. The human-machine system effectiveness evaluation model is used to quantify operational effectiveness, the human-environment system effectiveness evaluation model is used to quantify adaptive effectiveness, and the human-machine-environment collaborative system effectiveness evaluation model is used to quantify collaborative effectiveness and robustness effectiveness. The task flow path edges are weighted by the operational efficiency, the information propagation probability is modulated by the collaborative efficiency, the human vulnerability index and environmental sensitivity score are calculated by the adaptive efficiency, and the redundant collaborative paths are dynamically activated and the node failure recovery rate is adjusted by the robust efficiency, thus constructing a dynamic graph attribute with efficiency awareness. Using a dynamic multi-level centrality fusion algorithm, based on the attributes of the dynamic graph, task flow betweenness centrality, information propagation influence, human vulnerability index, and environmental sensitivity scores are calculated on the performance-aware dynamic graph to generate a dynamic criticality score for each task role. Based on the operational efficiency, adaptive efficiency, collaborative efficiency, and robustness efficiency, a performance profile for each task role is determined. Based on the dynamic criticality score, key nodes in collaborative work are identified.

[0133] It is understandable that a dynamic heterogeneous time series graph sequence is a graph data sequence composed of multiple consecutive time windows. Each graph contains heterogeneous nodes, enhanced features, and an adjacency structure with incremental updates, reflecting the temporal evolution of the system state.

[0134] Multimodal heterogeneous graph attention network (MHGAT) can distinguish different node types and edge semantics, aggregate neighbor information through attention mechanism, and support state propagation across time windows.

[0135] The collaborative context embedding vector is a high-dimensional vector of each task role obtained after MHGAT encoding. It integrates its own state, neighbor interactions and historical evolution information, and serves as a unified input for performance evaluation.

[0136] Multidimensional performance decoupling is used to decompose the overall collaborative capability into four independent dimensions, including operational performance (human-machine), adaptive performance (human-environment), collaborative performance, and robust performance (human-machine-environment), which are calculated by dedicated models.

[0137] The dynamic graph attribute of performance awareness is used to represent the back-injection of performance evaluation results into the graph structure, dynamically adjusting attributes such as edge weights, propagation probabilities, and node vulnerability, so that the graph itself has performance awareness.

[0138] The dynamic multi-level centrality fusion algorithm is used to comprehensively consider multiple network centrality indicators such as task flow betweenness, information influence, and vulnerability, and adaptively weights them according to the current task stage to calculate the criticality of nodes.

[0139] The performance profile is a four-dimensional quantitative description of each task role in the current collaborative task, including operational performance, adaptive performance, collaborative performance, and robust performance, with values ​​ranging from 0 to 1.

[0140] The task flow betweenness centrality is weighted based on the operational efficiency output by the human-machine system efficiency evaluation model, which is used to weight the task path edges; the human vulnerability index is generated by the inverse mapping of the adaptive efficiency output by the human-environment system efficiency evaluation model; and the information propagation influence dynamically adjusts the information diffusion rate between nodes based on the collaborative efficiency output by the human-machine-environment collaborative system efficiency evaluation model.

[0141] Critical nodes are task roles that rank highly in the dynamic criticality score. Their abnormal status or failure will significantly affect the overall task execution and require close monitoring or intervention.

[0142] In actual execution, the multi-role network graphs from multiple consecutive time windows are organized into a temporal graph sequence. Each graph contains enhanced feature vectors of nodes, incremental adjacency matrices, and node type labels. The sequence is processed by a multilayer heterogeneous graph attention network (MHGAT), which calculates attention weights for each node type within each time window and aggregates neighbor information. Then, the state is passed across time windows through gated recurrent units, ultimately outputting a collaborative context embedding vector that integrates the history and current context for each task role.

[0143] The collaborative context embedding vector of each task role is fed into three pre-trained deep evaluation models. The human-machine system effectiveness evaluation model is a fully connected neural network that outputs an operational effectiveness value between 0 and 1, reflecting the crew's accuracy and responsiveness in operating the equipment. The human-environment system effectiveness evaluation model is a neural network that combines the intensity of environmental disturbances with the physiological state of the personnel to output adaptive effectiveness, characterizing the personnel's tolerance to harsh sea conditions or high-load environments. The human-machine-environment collaborative system effectiveness evaluation model is a graph-aware network that outputs collaborative effectiveness and robustness effectiveness.

[0144] Among them, the operational efficiency is calculated based on the accuracy of operation command execution, response latency and consistency of equipment status feedback; the collaborative efficiency is calculated based on the integrity of information transmission, frequency of cross-role interaction and smoothness of task chain connection, reflecting the level of information sharing and team cooperation; the adaptive efficiency is calculated based on the state stability under environmental disturbances, task switching recovery speed and psychological load tolerance; and the robust efficiency is calculated based on the functional maintenance capability and redundant path activation efficiency under abnormal events, reflecting the maintenance capability when some nodes fail. In addition, each model also outputs corresponding dynamic weights and risk factors such as the probability of operational errors and the risk of seasickness.

[0145] The operational efficiency, adaptive efficiency, collaborative efficiency, and robust efficiency are back-mapped into a graph structure. Operational efficiency is used to weight the edges on the task flow path, collaborative efficiency is used to modulate the information propagation probability, adaptive efficiency is used to calculate the human vulnerability index and environmental sensitivity score, and robust efficiency is used to dynamically activate preset redundant collaborative paths and adjust their recovery rate when nodes fail. The operational attributes of the graph are dynamically adjusted to construct performance-aware dynamic graph attributes. On the updated performance perception map, the number of critical task paths on which each heterogeneous node is located is counted, and the task flow betweenness centrality is calculated. Based on the independent cascade model, information propagation is simulated, and its scope of influence is calculated to obtain the information propagation influence. Adaptive performance is used to calculate the human vulnerability index and environmental sensitivity score, reflecting the degree to which the node is affected by environmental disturbances.

[0146] Subsequently, the system adaptively allocates the weights of each indicator according to the current mission phase (e.g., the combat phase focuses more on mission flow, while the cruise phase focuses more on human factors), and then sums the weights to obtain a dynamic criticality score for each mission role.

[0147] The four types of performance values ​​for each task role are combined into a four-dimensional vector to serve as a performance profile. All task roles are sorted from high to low according to their dynamic criticality scores. The top 10% or those with scores exceeding the threshold are selected and marked as key nodes, and pushed to the command and decision-making system for key monitoring or resource support.

[0148] In this embodiment, by constructing a dynamic heterogeneous time sequence graph and introducing a multi-dimensional performance decoupling mechanism, the fine quantification of the ship's mission role operation, adaptation, cooperation and robustness is realized. The performance results are then injected back into the graph structure to form a dynamic attribute of "performance perception". Combined with a multi-layer centrality fusion algorithm for mission adaptation, the key nodes that have a decisive impact on the success or failure of the mission are accurately identified.

[0149] In some embodiments, obtaining the crew member's operational performance, psychological workload, and scale data based on the performance profile and the key nodes includes: Based on the quantified values ​​of operational effectiveness, adaptive effectiveness, cooperative effectiveness, and robust effectiveness in the aforementioned performance profile, these values ​​are back-mapped to the corresponding original observation dimensions: Based on the dynamic criticality scores of the key nodes and the task stage context, a collaborative attribution graph is constructed. Based on the dynamic criticality scores of heterogeneous crew nodes in the performance perception dynamic graph, and combined with the weighted fusion results of operational performance and collaborative performance, the operational performance is generated. By coupling analysis of environmental sensitivity scores and psychological state feature vectors in the human-environment interaction layer, the psychological load of the crew members under each sea state condition is obtained. The psychological load includes three sub-indicators: cognitive load, emotional stress, and physiological fatigue. A structured questionnaire was constructed using a five-level Likert scale. The scale data was obtained by comparing the operating procedures in the equipment technical manual with actual interaction events. The scale data included job satisfaction, task suitability, and team trust.

[0150] Understandably, the reverse mapping to the original observation dimension is to map the operational effectiveness, adaptive effectiveness, collaborative effectiveness, and robust effectiveness in the effectiveness performance profile back to the original observation dimension.

[0151] Collaborative attribution mapping is a graphical tool used to visualize the relationship between dynamic criticality scores of key nodes and their task phase context, helping to analyze the factors influencing each role's overall effectiveness at different task phases.

[0152] Operational performance is a comprehensive evaluation of the effectiveness and efficiency with which crew members complete tasks. It is based on a weighted fusion of operational effectiveness and collaborative effectiveness, and takes into account the dynamic criticality scores of crew members in the effectiveness perception dynamic map. Quantitative indicators of operational performance include task completion rate, operational accuracy rate, and timeliness of anomaly response.

[0153] Psychological load is assessed in three aspects: cognitive load, emotional stress, and physical fatigue. It is derived through coupling analysis of environmental sensitivity scores and psychological state feature vectors in the human-environment interaction layer, reflecting the degree of psychological burden on crew members under specific sea conditions.

[0154] The five-point Likert scale is a survey questionnaire design method that provides five response options to measure the degree of a respondent's opinion or feeling, ranging from completely disagreeing to completely agreeing.

[0155] In practice, the quantitative values ​​of operational efficacy, adaptive efficacy, collaborative efficacy, and robust efficacy in the performance profile are analyzed and converted into specific raw observation indicators according to predefined mapping rules. For example, low operational efficacy is associated with a decrease in operational accuracy or abnormal response delay in job performance; low adaptive efficacy is associated with a high cognitive load index or fluctuations in emotional arousal in psychological load; low collaborative efficacy is associated with subjective ratings in scale data; and low robust efficacy is associated with behavioral records of a sudden drop in job performance without effective recovery in abnormal events.

[0156] For the identified key nodes, the system uses their dynamic criticality scores and current task stage context information to construct a collaborative attribution graph, which is used to show how the importance of key nodes changes over time and how it affects the overall system performance.

[0157] By combining the dynamic criticality scores, operational effectiveness, and collaborative effectiveness of heterogeneous crew nodes in the performance perception dynamic graph, a weighted average is calculated as the crew's operational performance. The weights can be adjusted according to the specific task requirements.

[0158] By analyzing the environmental sensitivity score and the state feature vector of psychological state in the human-environment interaction layer, the psychological load of the crew members is calculated by calling a model that includes three sub-indicators: cognitive load, emotional stress, and physiological fatigue.

[0159] A structured questionnaire based on a five-point Likert scale was designed and used to collect data on job satisfaction, task suitability, and team trust by comparing the operating procedures in the equipment technical manual with actual interaction events.

[0160] In the collaborative attribution mapping, if a crew member is identified as a high-risk critical node, characterized by high criticality and low efficiency, then the efficiency bottleneck is marked in their attribution path; if multiple non-critical nodes simultaneously experience a decline in collaborative efficiency, it is attributed to the failure of the team's collaborative mechanism, rather than an individual problem. In addition, the reliability of the data was verified by using Cronbach's α coefficient. The time-series variation curve of work performance and the fluctuation characteristics of psychological load were cross-validated. Combined with the long-term work attitude indicators reflected in the scale data, a comprehensive crew performance database including immediate performance and continuous status was constructed.

[0161] In this embodiment, by back-mapping the quantitative values ​​in the performance profile to the original observation dimension and combining them with the dynamic characteristics of key nodes, a collaborative attribution map was successfully constructed, which effectively improved the understanding and evaluation accuracy of crew member work performance. At the same time, by deeply analyzing the psychological state and environmental sensitivity in human-environment interaction, a refined assessment of crew member psychological load was achieved.

[0162] In some embodiments, obtaining the effectiveness assessment result of the collaborative effectiveness of the ship's mission execution based on the operational performance, the psychological load, and the scale data includes: The job performance is mapped to a behavioral dimension representation, the psychological load is mapped to a physiological dimension representation, and the scale data is mapped to a task dimension representation. The behavioral dimension representation, the physiological dimension representation, and the task dimension representation are then fused together with temporal alignment and semantic complementarity to generate a collaborative state embedding vector. Based on the cooperative state embedding vector, calculate the multi-granularity cooperative performance index; Based on the semantic tags of the current task stage, the multi-granularity collaborative performance indicators are adaptively weighted and fused to obtain the performance evaluation results.

[0163] Understandably, behavioral dimension representation transforms crew members' task completion rate, operational accuracy, response latency, and other operational performance metrics into structured, computable numerical vectors to characterize their overt behavioral performance.

[0164] Physiological dimension representation refers to the transformation of psychological loads such as cognitive load, emotional stress, and physiological fatigue into quantitative vectors based on physiological and psychological states, reflecting the impact of crew members' internal state on their collaborative capabilities.

[0165] Task dimension representation encodes data from scales such as job satisfaction, task suitability, and team trust into vectors that reflect subjective task experience and team cognition, embodying the willingness to collaborate and cognitive consistency within the task context.

[0166] Temporal alignment and semantic complementarity fusion aligns behavioral dimension representations, physiological dimension representations, and task dimension representations to the same time window in time, and integrates their complementary information through splicing and projection to generate a unified co-state embedding vector.

[0167] Multi-granularity synergistic effectiveness index refers to multiple synergistic effectiveness sub-indicators calculated from different granularities such as individuals, groups, and the whole system, as well as from different perspectives such as rate, stability, resilience, and consistency.

[0168] Adaptive weighted fusion dynamically adjusts the weights of each collaborative performance indicator based on the semantic labels of the current task stage, so that the final evaluation result is more in line with the current task objective.

[0169] The performance evaluation results are a comprehensive score and interpretable report of the ship's human-machine-environment collaborative performance, which are used for command decision-making, human factor intervention, or mission optimization.

[0170] In actual execution, for job performance, task completion, operational accuracy, response speed, and error rate are extracted and normalized to form a behavioral dimension representation vector; for psychological load, cognitive load, emotional stress, and physiological fatigue are quantified into 0-1 values ​​to form a physiological dimension representation vector; for scale data, job satisfaction, task suitability, and team trust in the five-level Likert scale are converted into continuous 0-1 values ​​to form a task dimension representation vector.

[0171] Using a 1-second time window, the three types of representations are ensured to come from the same time segment. The three representation vectors are concatenated into a joint vector and input into a pre-trained lightweight multilayer perceptron (MLP) for semantic fusion. The MLP identifies the complementary relationship between dimensions and outputs a fixed-dimensional cooperative state embedding vector as a unified representation of the ship's cooperative state at the current moment.

[0172] The predefined performance calculation module is invoked to extract multiple indicators from the collaborative state embedding vector. At the individual level, the comprehensive performance score of each crew member is calculated; at the group level, the embedding vectors of members in the same position or task group are averaged to calculate the group collaboration consistency index; at the system level, the whole graph embedding is aggregated through graph neural network to calculate indicators such as overall task execution efficiency, system stability, and robustness. The predefined performance calculation module outputs 5–8 interpretable multi-granularity collaborative performance indicators.

[0173] The system reads the current mission phase label from the ship mission management system and queries the preset weight configuration table. It increases the mission layer weight during the emergency response phase, the team layer weight during the cruise monitoring phase, and the individual layer weight during the complex operation phase. The system then sums the weights of the multi-granularity indicators according to these weights to obtain the final collaborative effectiveness evaluation result between 0 and 1, and generates a structured evaluation report that includes key shortcomings.

[0174] In this embodiment, by mapping operational performance, psychological load, and scale data to three dimensions of behavior, physiology, and task respectively, and performing temporal alignment and semantic complementarity fusion, a high-fidelity collaborative state embedding vector is generated. Then, multi-granularity collaborative effectiveness indicators are calculated, and adaptive weighted fusion is achieved by combining task stage semantics. Finally, accurate, interpretable, and task-oriented collaborative effectiveness evaluation results of ship mission execution are output.

[0175] The ship mission execution performance evaluation method provided in this application can be implemented by a ship mission execution performance evaluation device. This application uses the ship mission execution performance evaluation device executing the method as an example to illustrate the ship mission execution performance evaluation device provided in this application.

[0176] This application also provides a device for evaluating the performance of ship missions.

[0177] like Figure 3 As shown, the ship's mission performance evaluation device includes: The first processing module 310 is used to construct a multi-role network graph system based on the interaction and dependency relationships between multiple task roles of the ship in the execution of the mission. The task roles include crew, equipment and environment. The multi-role network graph system includes a human-machine system performance evaluation model, a human-environment system performance evaluation model and a human-machine-environment collaborative system performance evaluation model. The second processing module 320 is used to analyze the multi-role network graph system and identify the performance profile and key nodes of each task role in collaborative work. The third processing module 330 is used to obtain the crew's work performance, psychological load, and scale data based on the performance profile and the key nodes. The fourth processing module 340 is used to obtain the effectiveness evaluation result of the collaborative effectiveness of the ship's mission execution based on the operational performance, the psychological load, and the scale data.

[0178] According to the ship mission execution performance evaluation device provided in the embodiments of this application, by analyzing the interaction and dependency relationships between crew, equipment and environment, a multi-role network diagram system is constructed, the performance profile and key nodes are analyzed, and the collaborative performance between each role is quantified and analyzed by integrating operational performance, psychological load and scale data, dynamically adapting to mission changes, and comprehensively and accurately evaluating the overall work efficiency in the ship mission execution process for multi-task and multi-role collaboration.

[0179] The ship mission performance evaluation device in this application embodiment can be an electronic device or a component of an electronic device, such as an integrated circuit or a chip. The electronic device can be a terminal or other devices besides a terminal.

[0180] The ship mission performance evaluation device in this application embodiment can be a device with an operating system. This operating system can be Android, iOS, or other possible operating systems; this application embodiment does not specifically limit it.

[0181] The ship mission execution performance evaluation device provided in this application embodiment can realize the various processes implemented in the ship mission execution performance evaluation method embodiment as described above. To avoid repetition, it will not be described again here.

[0182] In some embodiments, such as Figure 4 As shown, this application embodiment also provides an electronic device 400, including a processor 401, a memory 402, and a computer program stored on the memory 402 and executable on the processor 401. When the program is executed by the processor 401, it implements the various processes of the above-described ship mission performance evaluation method embodiment and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0183] It should be noted that the electronic devices in the embodiments of this application include the mobile electronic devices and non-mobile electronic devices described above.

[0184] This application also provides a non-transitory computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the various processes of the above-described ship mission performance evaluation method embodiment and achieves the same technical effect. To avoid repetition, it will not be described again here.

[0185] The processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0186] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described ship mission performance evaluation method.

[0187] The processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0188] This application embodiment also provides a chip, which includes a processor and a communication interface. The communication interface and the processor are coupled. The processor is used to run programs or instructions to implement the various processes of the above-described ship mission performance evaluation method embodiment and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0189] It should be understood that the chip mentioned in the embodiments of this application may also be referred to as a system-on-a-chip, system chip, chip system, or system-on-a-chip, etc.

[0190] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element. Furthermore, it should be noted that the scope of the methods and apparatuses in the embodiments of this application is not limited to performing functions in the order shown or discussed, but may also include performing functions substantially simultaneously or in the reverse order, depending on the functions involved. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.

[0191] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a computer software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, or network device, etc.) to execute the ship mission performance evaluation method of the various embodiments of this application.

[0192] In the description of this application, "first feature" and "second feature" may include one or more of the features.

[0193] In the description of this application, "multiple" means two or more.

[0194] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.

[0195] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0196] Although embodiments of this application have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of this application, the scope of which is defined by the claims and their equivalents.

Claims

1. A method for evaluating the performance of ship missions, characterized in that, include: Based on the interaction and dependency relationships between multiple mission roles of a ship during mission execution, a multi-role network graph system is constructed. The mission roles include crew, equipment, and environment. The multi-role network graph system includes a human-machine system effectiveness evaluation model, a human-environment system effectiveness evaluation model, and a human-machine-environment collaborative system effectiveness evaluation model. The multi-role network graph system is analyzed to identify the performance profile and key nodes of each task role in collaborative work; Based on the performance profile and the key points, the crew's work performance, psychological workload, and scale data are obtained. Based on the operational performance, psychological load, and scale data, the effectiveness evaluation results of the collaborative effectiveness of the ship's mission execution are obtained.

2. The method for evaluating the effectiveness of ship mission execution according to claim 1, characterized in that, The system constructs a multi-role network graph system based on the interaction and dependency relationships between multiple mission roles of a ship during mission execution, including: The heterogeneous nodes and feature vectors of the task roles are determined. The heterogeneous nodes include crew heterogeneous nodes, equipment heterogeneous nodes and environmental heterogeneous nodes. The crew heterogeneous nodes carry state feature vectors of operational skills and psychological state. The equipment heterogeneous nodes carry attribute feature vectors of performance parameters and fault history. The environmental heterogeneous nodes include environmental parameter feature vectors of sea state and weather. Based on the interaction and dependency relationships between the heterogeneous nodes, a multi-layer adjacency matrix is ​​constructed; Based on the heterogeneous nodes, the state feature vectors, and the multi-layer adjacency matrix, the multi-role network graph system is constructed. The ontology knowledge graph of the ship is converted into triples, and the entity embedding vector is determined in the triples using the TransE algorithm. The entity embedding vector is then concatenated with the feature vector in the multi-layer adjacency matrix to obtain an enhanced node feature representation. In each time window, the temporal changes of the enhanced node feature representation are captured to obtain the temporal-aware node state representation and the interaction events of each time window; Based on the multi-layer adjacency matrix of the first time window and the interaction events of the second time window, the changes in the interaction relationships between the heterogeneous nodes are detected, the multi-layer adjacency matrix is ​​updated, and an incremental adjacency matrix is ​​obtained. The multi-role network graph system is updated based on the time-aware node state representation, the enhanced node feature representation, and the incremental adjacency matrix.

3. The method for evaluating the effectiveness of ship mission execution according to claim 2, characterized in that, Based on the interaction and dependency relationships between the heterogeneous nodes, a multi-level adjacency matrix is ​​constructed, including: Based on the aforementioned dependencies, the topological skeleton of multiple semantic adjacency layers is determined; Based on the topological skeleton, according to the interaction events in the interaction relationship and the feature vectors of the heterogeneous nodes, the dynamic weight of each edge in the semantic adjacency layer is calculated through a real-time state function to generate the corresponding weighted adjacency submatrix. Multiple weighted adjacency sub-matrices are stacked into a three-dimensional adjacency tensor. The importance weight of each semantic layer in the three-dimensional adjacency tensor at the current task stage is learned through a cross-layer graph attention mechanism. Task-adaptive adjacency fusion is then performed to obtain the multi-layer adjacency matrix.

4. The method for evaluating the effectiveness of ship mission execution according to claim 3, characterized in that, The process of determining the topological skeleton of multiple semantic adjacency layers based on the dependencies includes: In the dependency relationship, the key dependency paths of each link in the mission process of the ship are determined, and the human-machine-loop closed-loop dependency chain between the mission roles is extracted. The machine-loop closed-loop dependency chain includes the crew operating equipment main chain, the equipment response environment main chain, and the environmental feedback crew main chain. The machine-loop closed-loop dependency chain is topologically decomposed by dependency strength analysis to identify the semantic adjacency layer, which includes a human-computer interaction layer, a human-loop interaction layer, and a machine-loop interaction layer. Using a dependency propagation algorithm, a topological skeleton based on a directed acyclic graph is constructed for each semantic adjacency layer, wherein the human-computer interaction layer uses operation instructions as edge directions, the human-environment interaction layer uses environmental parameter thresholds as edge directions, and the machine-environment interaction layer uses device performance constraints as edge directions. By performing dependency closure detection, redundant edges are pruned and critical edges are strengthened on the topological skeleton.

5. The method for evaluating the effectiveness of ship mission execution according to claim 3, characterized in that, Based on the topological skeleton, and according to the interaction events in the interaction relationship and the feature vectors of the heterogeneous nodes, the dynamic weights of the edges in each semantic adjacency layer are calculated through a real-time state function to generate the corresponding weighted adjacency sub-matrix, including: Based on the edges in the topological skeleton, context aggregation of micro-state, meso-state and macro-state is performed to construct a multi-granularity spatiotemporal state context vector. The micro-state includes the instantaneous attributes of the node, the meso-state includes the behavioral statistical features within the recent time window, and the macro-state includes the semantic label of the current task stage. The semantic adjacency layer is configured with a domain knowledge guidance function module as a real-time state function. In the human-computer interaction layer, the operation matching degree weight between crew members and equipment is calculated by configuring a skill-complexity matching network. In the machine-environment interaction layer, a simplified physical model is embedded to calculate the impact weight of the environment on equipment performance. In the human-environment interaction layer, the disturbance weight of the environment on personnel state is calculated by configuring a piecewise function based on human factors engineering standards. The causality test module verifies the consistency of the edge weights. If the edge weights are found to violate the ship's operating mechanism, an exponential decay penalty is applied to the edge weights to obtain the mechanism calibration weights. Based on the data integrity and interaction event compliance of the feature vectors of the heterogeneous nodes, a confidence factor is calculated. The data integrity is determined by evaluating the online status of sensors and the data missing rate, and the interaction event compliance is determined by rule matching based on the operation procedure knowledge graph. The confidence factor is multiplied by the mechanism calibration weight to obtain the dynamic weight of each edge; The dynamic weights are filled into the adjacent positions corresponding to the topological skeleton, and the unconnected positions are set to zero, generating the weighted adjacency submatrix that corresponds one-to-one with the semantic adjacency layer.

6. The method for evaluating the effectiveness of ship mission execution according to any one of claims 1-5, characterized in that, The analysis of the multi-role network graph system identifies the performance profile and key nodes of each task role in collaborative work, including: Based on time-aware node state representation, enhanced node feature vectors, and incremental adjacency matrix, a dynamic heterogeneous time sequence is constructed. A multi-layer heterogeneous graph attention network is used to jointly encode the state evolution of each task role in the dynamic heterogeneous time sequence within multiple time windows to generate a collaborative context embedding vector. The collaborative context embedding vector is input into the human-machine system effectiveness evaluation model, human-environment system effectiveness evaluation model, and human-machine-environment collaborative system effectiveness evaluation model in the role network graph system, respectively, to perform multi-dimensional effectiveness decoupling and obtain dynamic weights and risk factors. The human-machine system effectiveness evaluation model is used to quantify operational effectiveness, the human-environment system effectiveness evaluation model is used to quantify adaptive effectiveness, and the human-machine-environment collaborative system effectiveness evaluation model is used to quantify collaborative effectiveness and robustness effectiveness. The task flow path edges are weighted by the operational efficiency, the information propagation probability is modulated by the collaborative efficiency, the human vulnerability index and environmental sensitivity score are calculated by the adaptive efficiency, and the redundant collaborative paths are dynamically activated and the node failure recovery rate is adjusted by the robust efficiency, thus constructing a dynamic graph attribute with efficiency awareness. Using a dynamic multi-level centrality fusion algorithm, based on the attributes of the dynamic graph, task flow betweenness centrality, information propagation influence, human vulnerability index, and environmental sensitivity scores are calculated on the performance-aware dynamic graph to generate a dynamic criticality score for each task role. Based on the operational efficiency, adaptive efficiency, collaborative efficiency, and robustness efficiency, a performance profile for each task role is determined. Based on the dynamic criticality score, key nodes in collaborative work are identified.

7. The method for evaluating the effectiveness of ship mission execution according to any one of claims 1-5, characterized in that, The process of obtaining the crew's operational performance, psychological workload, and scale data based on the performance profile and key nodes includes: Based on the quantified values ​​of operational effectiveness, adaptive effectiveness, cooperative effectiveness, and robust effectiveness in the aforementioned performance profile, these values ​​are back-mapped to the corresponding original observation dimensions: Based on the dynamic criticality scores of the key nodes and the task stage context, a collaborative attribution graph is constructed. Based on the dynamic criticality scores of heterogeneous crew nodes in the performance perception dynamic graph, and combined with the weighted fusion results of operational performance and collaborative performance, the operational performance is generated. By coupling analysis of environmental sensitivity scores and psychological state feature vectors in the human-environment interaction layer, the psychological load of the crew members under each sea state condition is obtained. The psychological load includes three sub-indicators: cognitive load, emotional stress, and physiological fatigue. A structured questionnaire was constructed using a five-level Likert scale. The scale data was obtained by comparing the operating procedures in the equipment technical manual with actual interaction events. The scale data included job satisfaction, task suitability, and team trust.

8. The method for evaluating the effectiveness of ship mission execution according to any one of claims 1-5, characterized in that, The effectiveness evaluation result of the collaborative effectiveness of the ship's mission execution, obtained based on the operational performance, the psychological load, and the scale data, includes: The job performance is mapped to a behavioral dimension representation, the psychological load is mapped to a physiological dimension representation, and the scale data is mapped to a task dimension representation. The behavioral dimension representation, the physiological dimension representation, and the task dimension representation are then fused together with temporal alignment and semantic complementarity to generate a collaborative state embedding vector. Based on the cooperative state embedding vector, calculate the multi-granularity cooperative performance index; Based on the semantic tags of the current task stage, the multi-granularity collaborative performance indicators are adaptively weighted and fused to obtain the performance evaluation results.

9. A device for evaluating the performance of ship missions, characterized in that, include: The first processing module is used to construct a multi-role network graph system based on the interaction and dependency relationships between multiple task roles of the ship in mission execution. The task roles include crew, equipment and environment. The multi-role network graph system includes a human-machine system effectiveness evaluation model, a human-environment system effectiveness evaluation model and a human-machine-environment collaborative system effectiveness evaluation model. The second processing module is used to analyze the multi-role network graph system and identify the performance profile and key nodes of each task role in collaborative work. The third processing module is used to obtain the crew's work performance, psychological load, and scale data based on the performance profile and the key nodes. The fourth processing module is used to obtain the effectiveness evaluation result of the ship's collaborative effectiveness based on the operational performance, the psychological load, and the scale data.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the ship mission performance evaluation method as described in any one of claims 1-8.

Images