A ship information integrated management platform system for shipping enterprises

Abstract

This invention relates to the field of intelligent shipping management technology, specifically to an integrated management platform system for personnel and vessels in shipping enterprises. The system includes modules for data acquisition and labeling, feature analysis, feature fusion, risk assessment, and strategy generation. The data acquisition and labeling module collects original human-machine operation records and vessel equipment status records from shipping operations, and generates a unified set of operation records through format normalization and correlation annotation. The feature analysis module extracts personnel behavior feature codes, equipment status evolution trajectories, and environmental event correlation networks, and generates an integrated dynamic feature map of personnel and vessels through interactive evolution via a feature fusion device. The risk assessment module constructs a network based on the map to dynamically determine risk levels and form a hierarchical risk list. The strategy generation module generates adapted management strategies and converts them into a set of executable instructions. This system achieves the technical effects of collaborative association of personnel and vessel data, accurate dynamic risk assessment, and direct implementation of management strategies.

Description

Technical Field

[0001] This invention relates to the field of intelligent shipping management technology, and in particular to an integrated management platform system for personnel and ship information for shipping companies. Background Technology

[0002] The current operational management of shipping companies employs a separate data collection and processing model, collecting human and machine operation records, ship equipment status records, and environmental event records separately. These raw records are stored independently with inconsistent formats, and only basic feature processing is performed on a single data type. Personnel identification features are extracted only once, ship equipment status is only detected with simple anomalies, and environmental events are only statistically analyzed independently. No correlation or labeling relationship has been established between the three types of data (personnel, equipment, and environment). Risk assessment relies on static level determination based on single-dimensional data, management strategies are formulated based on human experience, and strategy content is presented only in text form, unable to be directly converted into executable management instructions. Risk assessment results only form single risk items.

[0003] Separate data processing models cannot achieve collaborative correlation of personnel, ship equipment, and environmental event data; simplistic feature extraction methods cannot form a dynamic expression of human-ship collaborative features; static risk assessment methods cannot adapt to the dynamic changes in shipping operations; manually formulated management strategies lack adaptability to target shipping operations, and there is a disconnect between strategy content and actual execution. This invention requires format normalization and associative annotation of the original record sequence to form a unified format set of operation records. After feature extraction from multiple data types, a dynamic feature map integrating human and ship elements is generated through interactive evolution using a feature fusion device. Based on this feature map, an operation risk assessment network needs to be constructed to achieve dynamic risk level determination and form a hierarchical risk list. An adaptive management strategy is generated through a management strategy generation engine, and the strategy is converted into a set of executable management instructions. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of existing technologies and propose an integrated management platform system for personnel and ship information for shipping companies.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: an integrated management platform system for personnel and vessel information for shipping enterprises, comprising: The data acquisition and labeling module collects original human-machine operation records and ship equipment status records in shipping operation scenarios to form a raw record sequence. It performs format normalization and association labeling on the raw record sequence to generate a unified format operation record set. The operation record set includes personnel identification tags, multi-dimensional status time series of ship equipment, and environmental event time series. The feature analysis module performs multi-round identity feature extraction on the personnel identification tags to obtain personnel behavior feature codes, performs abnormal pattern recognition on the multi-dimensional state time series of the ship equipment to obtain the equipment state evolution trajectory, and performs event correlation analysis on the environmental event time series to obtain the environmental event correlation network. The feature fusion module inputs the personnel behavior feature code, the equipment state evolution trajectory and the environmental event association network into the feature fusion device for interactive evolution, generating a dynamic feature map of integrated human and ship; The risk assessment module, based on the integrated human-vessel dynamic feature map, constructs an operational risk assessment network to determine the dynamic risk level and form a hierarchical risk list; The strategy generation module generates adaptive management strategies for target shipping operations based on the hierarchical risk list through the management strategy generation engine, and converts the adaptive management strategies into a set of executable management instructions.

[0006] As a further aspect of the present invention, multiple rounds of identity feature extraction are performed on the personnel identification tag to obtain a personnel behavior feature code, including: The personnel identification tags are matched with a preset personnel qualification database to output a preliminary set of qualification attributes; Historical operational behavior tracing and habit pattern mining are performed on the preliminary qualification attribute set to extract personnel operation pattern clusters; The current task scenario is embedded into the personnel operation mode cluster by a scenario encoder to generate scenario operation feature vectors; Feature compression technology is used to extract and encode the core features of the scenario-based operation feature vector to generate a distinctive personnel behavior feature code.

[0007] As a further aspect of the present invention, anomaly pattern recognition is performed on the multi-dimensional state time series of the ship equipment to obtain the equipment state evolution trajectory, including: Multiple monitoring dimensions are defined on the multi-dimensional state time series of the ship equipment. The original equipment state data is slide-segmented under different monitoring dimensions to form multi-angle state data segments. The stationarity test and deviation detection are performed on the equipment status data in each multi-angle status data segment to separate the time series of normal baseline component, abnormal fluctuation component and noise component. By performing spatiotemporal alignment and trend synthesis on the normal baseline component, abnormal fluctuation component, and noise component corresponding to different monitoring dimensions, the device state evolution trajectory reflecting the comprehensive changes in the device state across the three dimensions of normal, abnormal, and disturbance is constructed.

[0008] As a further aspect of the present invention, the personnel behavior feature code, the equipment state evolution trajectory, and the environmental event association network are interactively evolved into a feature fusion device to generate a human-ship integrated dynamic feature map, including: An interactive evolution layer is established in the feature fusion device. The interactive evolution layer receives the personnel behavior feature code, the equipment state evolution trajectory and the environmental event association network as the initial node state. In the interactive evolution layer, a bidirectional influence link is established between feature nodes. The bidirectional influence link simulates the state coupling and causal transmission between the personnel behavior feature code, the equipment state evolution trajectory and the environmental event association network according to the preset causal logic rules. After a preset evolution cycle, the states of each feature node in the interactive evolution layer complete the interaction and reach stability. At this time, the stable node states after the interaction and their bidirectional influence link relationships are output in a graph form, forming the integrated human-ship dynamic feature graph.

[0009] As a further aspect of the present invention, based on the integrated human-vessel dynamic feature map, an operational risk assessment network is constructed to determine the dynamic risk level and form a hierarchical risk list, including: The tiered risk list includes work unit identifiers and quantified risk levels; The feature nodes in the integrated human-ship dynamic feature map are used as the input layer nodes of the operation risk assessment network; A multi-layer risk perception neuron is constructed, and the features of the input layer nodes are non-linearly mapped and fused through the multi-layer risk perception neuron, and then passed to the output layer layer by layer. In the output layer, each job unit identifier corresponds to a risk assessment neuron, which converts the received fusion information into a continuous quantitative risk level value. All work unit identifiers and their corresponding quantitative risk level values ​​are automatically divided and sorted according to the level range of the quantitative risk level values ​​to generate the hierarchical risk list.

[0010] As a further aspect of the present invention, based on the hierarchical risk list, an adaptive management strategy for the target shipping operation is generated through a management strategy generation engine, including: The adaptability management strategy includes suggested sequences of personnel operations and ship equipment maintenance plans; The hierarchical risk list, along with the resource and constraint list obtained in real time from the shipping operation site, is input into the management strategy generation engine. In the management strategy generation engine, an optimal model of resource constraints and risk control objectives is established. The optimal model uses the quantified risk level in the hierarchical risk list as the quantitative expression of the risk control objectives, and the available resources and operational requirements in the resource and constraint list as the boundary conditions of resource constraints. Solving the optimal model yields a work unit management scheme that minimizes overall operational risk and the corresponding ideal resource configuration under the resource constraint boundary conditions. The operational unit management scheme and the corresponding ideal resource configuration are adjusted for operability according to the actual shipping operation process and specifications to generate the personnel operation suggestion sequence and the ship equipment maintenance plan, which together constitute the adaptability management strategy.

[0011] As a further aspect of the present invention, the operation unit management scheme and the corresponding ideal resource configuration are adjusted for operability according to actual shipping operation processes and specifications, including: Identify the standard operating procedures and maritime regulations followed by the target shipping operations; The expected intervention method for each work unit in the work unit management scheme is analyzed, and the expected intervention method is derived from the equipment state evolution trajectory and the personnel behavior feature code; The expected intervention methods will be verified for compliance and feasibility against the standard operating procedures and maritime management regulations. After the verification is passed, the final executable resource configuration scheme and operation sequence are fine-tuned according to the size of the ideal resource configuration and the availability of manpower and equipment on site, thus completing the operability adjustment.

[0012] As a further aspect of the present invention, the adaptability management strategy is converted into a set of executable management instructions, including: The personnel operation suggestion sequence and ship equipment maintenance plan in the adaptability management strategy are analyzed to identify the personnel instruction items that need to be issued, the ship equipment parameter items that need to be configured, and their corresponding operation types. Consult the instruction and parameter protocol manual that matches the target personnel terminal type and the target ship equipment controller model, and map the identified operation type to specific personnel instruction codes and ship equipment control parameters; According to the logic and time sequence implied in the personnel operation suggestion sequence and the ship equipment maintenance plan, the specific personnel instruction codes and ship equipment control parameters are arranged into an instruction sequence with execution conditions and triggering timing, forming the set of executable management instructions.

[0013] As a further aspect of the present invention, according to the logic and temporal sequence implied in the personnel operation suggestion sequence and the ship equipment maintenance plan, the specific personnel instruction codes and ship equipment control parameters are arranged into an instruction sequence with execution conditions and triggering timing, including: By analyzing the suggested sequence of personnel operations and the ship equipment maintenance plan, a logical chain for human-machine collaborative execution is determined. This logical chain indicates that the setting of certain ship equipment parameters is a prerequisite for relevant personnel to perform specific operations. Based on the aforementioned logic chain, preconditions and effective times are marked for each personnel instruction code and each ship equipment control parameter; Based on the aforementioned preconditions and effective time, a time-series logic model is used to schedule and sort the specific personnel instruction codes and ship equipment control parameters, generating a linear execution queue that is conflict-free in both time and logic, which serves as the instruction sequence.

[0014] As a further aspect of the present invention, the construction of the management strategy generation engine includes the following steps: Acquire a pre-defined risk control knowledge base, which includes historical operational accident cases, industry safety management standards, and expert risk handling experience; Based on the aforementioned historical operational accident cases, industry safety management standards, and expert risk handling experience, a model of the correspondence between risk factors and intervention measures was extracted. Establish a rule-based reasoning engine, which is used to receive the work unit identifiers and quantified risk levels in the hierarchical risk list, and infer a set of candidate intervention measures based on the correspondence model; An optimization model solver is established, which receives the set of candidate intervention measures and the current operational resource constraints, and calculates the optimal intervention plan with the ratio of operational resource consumption to risk reduction as the optimization objective. The rule reasoning engine and the optimization model solver are logically cascaded to construct the management strategy generation engine. The management strategy generation engine generates the adaptive management strategy based on the hierarchical risk list and resource constraint list through a process of reasoning followed by optimization.

[0015] Compared with the prior art, the advantages and positive effects of the present invention are as follows: The process involves collecting raw human-machine operation records and ship equipment status records from shipping operations to form raw record sequences. These sequences are then formatted and annotated to generate a unified set of operation records containing personnel identification tags, multi-dimensional ship equipment status time series, and environmental event time series. Multiple rounds of identity feature extraction are performed on the personnel identification tags to obtain personnel behavior feature codes. Anomaly pattern recognition is performed on the multi-dimensional ship equipment status time series to obtain equipment status evolution trajectories. Event correlation analysis is performed on the environmental event time series to obtain an environmental event correlation network. The personnel behavior feature codes, equipment status evolution trajectories, and environmental event correlation networks are input into a feature fusion device for interactive evolution, generating a dynamic feature map integrating human and ship elements. The personnel behavior feature codes fully represent the personalized characteristics of personnel operations; the equipment status evolution trajectories continuously reflect the state changes of ship equipment; the environmental event correlation network intuitively reflects the inherent correlation attributes of various environmental events; and the interactive evolution of the feature fusion device breaks down the independent processing boundaries of personnel, equipment, and environmental data, achieving dynamic coupling and correlated presentation of multi-dimensional features.

[0016] An operational risk assessment network is constructed based on a dynamic feature map integrating human and vessel characteristics. This network performs dynamic risk level determination and generates a hierarchical risk list. Based on this list, a management strategy generation engine generates adaptive management strategies for the target shipping operations, which are then converted into a set of executable management instructions. The operational risk assessment network enables real-time risk assessment based on the dynamic feature map. The hierarchical risk list distinguishes the specific attributes and differences between different risk levels. The management strategy generation engine generates corresponding adaptive strategies tailored to the target shipping operation scenario. The set of executable management instructions can be directly applied to the shipping operation management process, eliminating barriers between management strategies and actual execution. Attached Figure Description

[0017] Figure 1 This is a sequence diagram of the integrated personnel and vessel information management platform system for shipping enterprises described in this invention; Figure 2 A flowchart for obtaining the device state evolution trajectory for abnormal pattern recognition; Figure 3 A trajectory diagram of the evolution of ship equipment status; Figure 4 A comparative analysis chart of the effectiveness of intervention measures in shipping operations; Figure 5 This is a diagram for monitoring and analyzing the execution of management commands and system load. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0019] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0020] See Figure 1 This invention provides an integrated management platform system for personnel and vessel information in shipping enterprises. Its core lies in achieving deep integration, analysis, and intelligent decision-making of personnel, vessel, and environmental information during shipping operations through a modular architecture. The system includes a data acquisition and labeling module, a feature analysis module, a feature fusion module, a risk assessment module, and a strategy generation module. The data acquisition and labeling module is responsible for collecting original human-machine operation records and vessel equipment status records from the shipping operation site, forming a raw record sequence. This sequence is then formatted and associated with the records, generating a unified set of operation records containing personnel identification tags, multi-dimensional vessel equipment status time series, and environmental event time series. The feature analysis module receives this set of operation records, performs multi-round identity feature extraction on the personnel identification tags to obtain personnel behavior feature codes, performs anomaly pattern recognition on the multi-dimensional vessel equipment status time series to obtain equipment status evolution trajectories, and performs event correlation analysis on the environmental event time series to obtain an environmental event correlation network. The feature fusion module inputs the obtained personnel behavior feature codes, equipment state evolution trajectories, and environmental event correlation networks into the feature fusion device for interactive evolution, ultimately generating a dynamic, integrated human-ship dynamic feature map that comprehensively reflects the interaction state of humans, ships, and the environment. Based on this integrated human-ship dynamic feature map, the risk assessment module constructs an operational risk assessment network to dynamically determine risk levels and outputs a hierarchical risk list sorted by risk level. The strategy generation module then uses this hierarchical risk list and its internal management strategy generation engine to generate an adaptive management strategy for the current target shipping operation. This strategy is further converted into a set of executable management instructions that can be directly issued to personnel terminals or ship equipment controllers, thus completing a closed loop from perception, analysis, assessment to decision execution.

[0021] In one embodiment of the present invention, multiple rounds of identity feature extraction are performed on personnel identification tags to obtain personnel behavior feature codes. The process is as follows: The personnel identification tags are matched with a preset personnel qualification database to output a preliminary qualification attribute set corresponding to the personnel. This set includes basic information such as certificates, positions, and historical training records. The system traces historical operational behaviors and mines habitual patterns for the personnel's preliminary qualification attribute set. By analyzing their past operation logs, a personnel operation pattern cluster reflecting their personal operation style, proficiency, and common behavioral patterns is extracted. The personnel operation pattern cluster is embedded into the current task scenario using a scenario-based encoder. The scenario-based encoder combines the current task type, ship status, and environmental conditions to transform general operation patterns into scenario-based operation feature vectors strongly correlated with the current scenario. Feature compression technology is used to extract and encode the core features of the scenario-based operation feature vectors, removing redundant information and retaining the most distinctive features to generate a compact and uniquely identifiable personnel behavior feature code.

[0022] See Figure 2 Anomaly pattern recognition is performed on the multi-dimensional state time series of ship equipment to obtain the equipment state evolution trajectory. The process is as follows: Multiple monitoring dimensions are defined on the multi-dimensional state time series of ship equipment, such as mechanical vibration, temperature, pressure, and power consumption. The original equipment state data is segmented by sliding windows under different monitoring dimensions to form a series of multi-angle state data segments that are continuous or overlapping in time. The stationarity test and deviation detection are performed on the equipment state data in each multi-angle state data segment. Through statistical analysis models, the time series of the normal baseline component representing the steady-state operation characteristics of the equipment, the abnormal fluctuation component representing the potential fault or performance degradation trend, and the noise component caused by random disturbances are separated. The normal baseline component, abnormal fluctuation component, and noise component corresponding to different monitoring dimensions are spatiotemporally aligned and trend synthesized, that is, aligned according to timestamps. Taking into account the physical coupling relationship between each dimension, an equipment state evolution trajectory that can comprehensively reflect the changing trends and mutual influences of the equipment state at the three levels of normal, abnormal, and random disturbances is constructed.

[0023] In practical implementation, the process of extracting multiple rounds of identity features from personnel identification tags to obtain personnel behavior feature codes is achieved through the following method: The system matches the collected personnel identification tags with a preset personnel qualification database, outputting a preliminary set of qualification attributes associated with the personnel. This preliminary set of qualification attributes includes the personnel's seafarer competency certificate number, certificate level, historical training records, and a list of permitted operating equipment types. The system then traces historical operational behaviors and mines habitual patterns from the obtained preliminary qualification attribute set. It retrieves the personnel's operational records from historical operations and extracts personnel operation pattern clusters representing their personal operational rhythm, concentrated error intervals, and high-frequency operation combinations through time-series pattern analysis. In practical implementation, a scenario-based encoder embeds the mined personnel operation pattern clusters into the current task scenario. The scenario-based encoder receives current operational task instructions, real-time ship cargo status, and waterway weather information, and weights and maps the general personnel operation pattern clusters to the current scenario parameter space, generating scenario-based operation feature vectors. Feature compression technology is used to extract and encode the core features of the scenario-based operation feature vector. Principal component analysis is used to reduce the dimensionality of the high-dimensional vector. Feature components with variance contribution rates exceeding a preset threshold are retained and binary encoded to finally generate highly recognizable personnel behavior feature codes.

[0024] In some embodiments, the process of performing anomaly pattern recognition on the multi-dimensional state time series of ship equipment to obtain the equipment state evolution trajectory follows these steps: Multiple monitoring dimensions, including mechanical vibration, bearing temperature, fuel pressure, and cooling water flow, are defined on the multi-dimensional state time series of ship equipment. Under different monitoring dimensions, the original equipment state data is segmented using a sliding window of fixed time length, forming multiple multi-angle state data segments that are temporally continuous and partially overlapping. For the equipment state data within each multi-angle state data segment, stationarity testing and deviation detection are performed. The stationarity test uses the unit root test method to determine the trend characteristics of the data segment, while the deviation detection identifies outliers by calculating the standardized distance between data points and the moving average. This separates the normal baseline component time series representing steady-state operation of the equipment, the abnormal fluctuation component time series characterizing potential faults, and the random noise component time series. It is understandable that the normal baseline component, abnormal fluctuation component, and noise component corresponding to different monitoring dimensions are spatiotemporally aligned and trend synthesized separately. Spatiotemporal alignment is performed by interpolating and aligning the components of each dimension based on a unified timestamp sequence. Trend synthesis combines the components of multiple dimensions into a comprehensive index through a weighted fusion function, which is expressed as follows:

[0025] Where: characters This represents the device state evolution trajectory value synthesized at time t, character... This represents the value of the aligned normal baseline component at time t, character... This represents the value of the aligned anomalous fluctuation component at time t, character... This represents the value of the aligned noise component at time t, character ,character and characters These represent the preset weighting coefficients for each component. After synthesis, a trajectory reflecting the overall state evolution of the equipment across three dimensions—normal, abnormal, and random disturbances—is constructed.

[0026] Optionally, during feature extraction, the refinement of personnel operation pattern clusters relies on frequent pattern mining algorithms for continuous action sequences in historical operation records. It can be understood that the embedding process of the contextual encoder utilizes an attention mechanism, enabling the encoder to dynamically adjust the weights of different parts of the personnel's historical operation patterns based on the key elements of the current task. In some embodiments, deviation detection for multi-angle state data segments can employ a dynamic threshold algorithm, where the threshold is adaptively adjusted based on the historical statistical characteristics of the data segment itself. Optionally, the weight coefficients in the weighted fusion function... , and The values ​​assigned are based on the importance of different monitoring dimensions for predicting the overall health status of the equipment. They are predefined by domain experts or obtained through training with historical fault data. The synthesized equipment status evolution trajectory is a continuous time series, used to intuitively display the evolution trend of the equipment status.

[0027] In one embodiment of the present invention, personnel behavior feature codes, equipment state evolution trajectories, and environmental event association networks are input into a feature fusion device for interactive evolution to generate a dynamic feature map integrating humans and ships. The process is as follows: An interactive evolution layer is established in the feature fusion device. This layer defines the core elements in the received personnel behavior feature codes, equipment state evolution trajectories, and environmental event association networks as initial feature nodes and assigns them initial states. In the interactive evolution layer, according to preset causal logic rules, bidirectional influence links are established between personnel behavior feature nodes, equipment state evolution trajectory nodes, and environmental event association network nodes. These links simulate complex causal transmission and state coupling relationships, such as how personnel operations affect equipment states, how equipment anomalies require personnel intervention, and how environmental events simultaneously affect both people and equipment. After a preset evolution cycle, the state of each feature node in the interactive evolution layer continuously interacts and updates under the drive of the bidirectional influence link, eventually reaching a stable state. At this point, the stable node state after interaction and the established bidirectional influence link relationship between them are output in the form of a graph. This graph is the human-ship integrated dynamic feature graph, which dynamically depicts the real-time relationship of key elements in a specific operation scenario.

[0028] In specific implementation, the process of inputting personnel behavior feature codes, equipment state evolution trajectories, and environmental event correlation networks into a feature fusion device for interactive evolution to generate a dynamic feature map integrating human and vessel is achieved in the following way: An interactive evolution layer is established within the feature fusion device. This layer receives personnel behavior feature codes, equipment state evolution trajectories, and environmental event correlation networks from upstream modules as initial node states. Personnel behavior feature codes are mapped to nodes representing operator behavior tendencies, equipment state evolution trajectories are mapped to nodes representing equipment health trends, and key events in the environmental event correlation network are extracted and mapped to nodes representing external environmental factors. In some embodiments, the initial values ​​of node states can be obtained through normalization processing. For example, the binary encoding of the personnel behavior feature codes is converted to decimal values ​​and divided by the maximum possible value; the synthesized value of the equipment state evolution trajectory is normalized to a minimum and maximum value; and the states of environmental event nodes are predefined and assigned values ​​according to the event type and severity. In practical implementation, bidirectional influence links are established between feature nodes in the interactive evolution layer. The construction of these bidirectional influence links strictly adheres to preset causal logic rules. These rules include, for example, "If the abnormal vibration fluctuation component of the main equipment continues to increase, the attention level of the on-duty engineer needs to be increased," and "If the visibility environmental event node status deteriorates, the expected value of the ship speed node needs to be reduced." The bidirectional influence links simulate the state coupling and causal transmission between personnel behavior feature code nodes, equipment state evolution trajectory nodes, and environmental event-related network nodes based on these rules. It can be understood that the calculation process of state coupling and causal transmission can be implemented through an iterative message passing mechanism. Within each calculation step, a node updates its own state based on information from the connected links, and this state update can be expressed as the formula:

[0029] Where: characters This represents the state vector of node i after the kth evolution cycle, and the character... This represents the state vector of node i in the previous cycle, and the character... Represents the set of all neighboring nodes connected to node i via bidirectional influence links, character This represents the attribute weight of the bidirectional influence link from node j to node i, represented by the character. It is a message passing function used to calculate incoming information based on link attributes and neighbor node states, characters It is a state fusion function used to combine the node's own historical state with incoming information to generate a new state. After a preset evolution cycle, the states of each feature node in the interaction evolution layer complete multiple rounds of interaction driven by the bidirectional influence link. When the change in the state vector of all nodes within a continuous cycle is less than a preset threshold, the node state is considered to have reached stability. At this time, the interaction evolution layer outputs the stable node state vectors after interaction and the bidirectional influence link relationships between nodes in the form of an adjacency matrix and a graph node attribute list. The resulting graph structure is the human-ship integrated dynamic feature graph.

[0030] In some embodiments, pre-defined causal logic rules can be structurally extracted from historical accident reports and expert experience and stored as a rule base format of "IF condition THEN influence intensity and direction". Optionally, an information transfer function... It can be designed as a weight for link attributes State of neighboring nodes The product operation. This can be understood as the state fusion function. A network structure with gated recurrent units can be used to preserve the historical dependency characteristics of node state evolution. After a preset number of evolution cycles or when the state stability condition is reached, the interactive evolution process terminates. At this point, the stable node state vector and link relationships together constitute a dynamic feature map that describes the overall situation of the human-ship-environment system at a specific moment.

[0031] In one embodiment of the present invention, a work risk assessment network is constructed based on the integrated human-ship dynamic feature map to determine the dynamic risk level and form a hierarchical risk list. The process is as follows: The resulting hierarchical risk list includes work unit identifiers and their corresponding quantified risk levels. When constructing the work risk assessment network, each feature node in the integrated human-ship dynamic feature map is used as the input layer node of the network. Multi-layer risk perception neurons are constructed in the network. The features of the input layer nodes are nonlinearly mapped and fused with information between layers through these multi-layer risk perception neurons, so that the features are abstracted and transmitted layer by layer until the output layer. In the output layer, a risk assessment neuron is set for each work unit identifier that needs to be assessed. This neuron converts and maps the fused information received from the previous layer into a continuous quantified risk level value. After the system obtains all work unit identifiers and their corresponding quantified risk level values, it automatically divides them according to the predefined level range of the quantified risk level values, and sorts them according to the value in the same level, generating a hierarchical risk list with a clear structure and clear priority.

[0032] In specific implementation, the process of constructing an operational risk assessment network based on the integrated human-ship dynamic feature map to determine dynamic risk levels and form a hierarchical risk list is achieved in the following way. The resulting hierarchical risk list must include the identification of the operational unit being assessed and the quantified risk level value corresponding to that operational unit. When constructing the operational risk assessment network, each feature node in the integrated human-ship dynamic feature map is defined as a node in the input layer of the network, and the values ​​of each dimension of the node state vector in the integrated human-ship dynamic feature map are used as the initial input features of the corresponding input layer node. In some embodiments, the multi-layer risk perception neurons constructed in the operational risk assessment network are responsible for abstracting the input features layer by layer. The first layer of risk perception neurons receives the features of all input layer nodes, completes the first nonlinear mapping through weighted summation and bias transformation with a nonlinear activation function, and passes the output result as information to the next layer of risk perception neurons. In specific implementation, information fusion occurs between risk perception neurons in adjacent layers. Each node of the next layer of risk perception neurons receives the outputs from multiple nodes in the previous layer and fuses this information. The fusion process can be expressed as the formula:

[0033] Where: characters This represents the output value of the m-th risk-aware neuron in the l-th layer of the network, where the character... This represents the output value of the nth risk-aware neuron in the (l-1)th layer of the network, where the character... This represents the connection weight between the nth neuron in layer (l-1) and the mth neuron in layer l. This represents the bias term of the m-th neuron in the l-th layer, character. This represents a preset nonlinear activation function, such as the ReLU function. Features are nonlinearly mapped and fused layer by layer through multiple layers of risk-aware neurons, ultimately being passed to the output layer. In the output layer, a dedicated risk assessment neuron is configured for each job unit identifier that needs to be evaluated. The risk assessment neuron transforms and maps the fused information received from the previous layer of risk-aware neurons into a continuous quantified risk level value through a linear transformation function or another specific activation function. The quantified risk level value can be understood as a real number between 0 and 1, with a higher value indicating a higher risk level for the job unit assessed by the system.

[0034] In practice, after obtaining all work unit identifiers and their corresponding quantitative risk level values, the system automatically categorizes them according to preset risk level ranges. For example, values ​​in the range [0, 0.3) are defined as "low risk," [0.3, 0.7) as "medium risk," and [0.7, 1.0] as "high risk." Within the same risk level range, the system sorts the values ​​in descending order to indicate the relative level of risk. Optionally, the tiered risk list is ultimately output as a structured data list, as shown in Table 1.

[0035] Table 1: Layered Risk List

[0036] In some embodiments, the mapping function of the output layer risk assessment neurons can employ the Sigmoid function to ensure that the output value falls within the range of 0 to 1. The number of layers and the number of neurons in each layer of the multi-layer risk perception neurons can be pre-configured based on the complexity of the human-ship integrated dynamic feature map and the accuracy requirements of risk assessment. It is understood that work unit identifiers are typically associated with specific job responsibilities, equipment locations, and operational tasks; for example, "Main Engine Room Inspection_Shift A" identifies the work location, task, and responsible shift. The final generated hierarchical risk list, grouped according to risk level and sorted numerically within each group, provides a clear basis for the priority allocation of risk management resources.

[0037] See Figure 3 This is a trajectory diagram of the evolution of ship equipment status, visually presenting the three types of status components of the equipment over time. The normal baseline represents the stable operating state of the equipment without anomalies, with values ​​consistently between 0.2 and 0.25, exhibiting minimal fluctuations and reflecting the basic health level of the equipment. The evolution trajectory of abnormal fluctuations, indicating equipment failure or abnormal operating conditions, shows a significant deterioration trend, rising from an initial 0.3 to 0.88, serving as a core indicator for risk warning. The noise component, representing random disturbances caused by environmental interference or measurement errors, maintains extremely low values ​​between 0.04 and 0.08, effectively distinguishing it from genuine abnormal signals. By comparing the difference between abnormal fluctuations and the normal baseline, the starting point of equipment status deterioration can be accurately located. The continuously rising trend of abnormal fluctuations provides a quantitative basis for early maintenance intervention and preventing equipment failure. The separation of the noise component ensures the accuracy of anomaly detection, avoiding misjudging environmental interference as equipment failure.

[0038] In one embodiment of the present invention, an adaptive management strategy for the target shipping operation is generated through a management strategy generation engine based on a hierarchical risk list. This adaptive management strategy includes a suggested sequence of personnel operations and a ship equipment maintenance plan. During generation, the hierarchical risk list and a resource and constraint list obtained in real time from the shipping operation site are input into the management strategy generation engine. Within the management strategy generation engine, a preset risk control knowledge base is first acquired. This knowledge base contains historical operational accident cases, industry safety management standards, and expert risk handling experience. Based on this knowledge, a correspondence model between risk factors and effective intervention measures is extracted. Subsequently, the rule reasoning engine in the engine receives the operational unit identifier and quantified risk level from the hierarchical risk list and infers a set of candidate intervention measures for each risk point based on the aforementioned correspondence model. Next, the optimization model solver is activated. It receives the set of candidate intervention measures and the current operational resource constraint list, and establishes an optimal model with the operational resource consumption and risk reduction ratio as the optimization objective. This model uses the quantified risk level in the hierarchical risk list as the quantitative expression of the risk control objective and the available resources and operational requirements in the resource and constraint list as the boundary conditions of the resource constraints. The optimization model solver solves this optimal model to obtain a work unit management scheme and corresponding ideal resource configuration that minimizes overall operational risk while satisfying resource constraints. Subsequently, the obtained scheme is adjusted for operability by identifying the standard operating procedures and maritime regulations followed by the target shipping operations. The expected intervention method for each work unit in the work unit management scheme is analyzed, derived from equipment state evolution trajectories and personnel behavior characteristic codes. The expected intervention method is then verified for compliance and feasibility with the standard operating procedures and maritime regulations. After successful verification, based on the ideal resource configuration and the real-time availability of on-site manpower and equipment, the resource configuration scheme and operation sequence are fine-tuned to form the final confirmed executable resource configuration scheme and operation sequence. This generates a personnel operation suggestion sequence and a ship equipment maintenance plan, which together constitute the adaptive management strategy. The rule reasoning engine and the optimization model solver are logically cascaded in the management strategy generation engine, realizing a process of reasoning before optimization.

[0039] In practical implementation, an adaptive management strategy for the target shipping operation is generated through a management strategy generation engine based on the hierarchical risk list. This adaptive management strategy includes suggested sequences for personnel operations and ship equipment maintenance plans. The construction of the management strategy generation engine includes the following steps: First, a pre-set risk control knowledge base is acquired. This knowledge base stores textual and data analysis reports of historical operational accident cases, industry safety management regulations issued by the International Maritime Organization and classification societies, and expert risk handling experience recorded in rule form. Second, based on historical operational accident cases, industry safety management regulations, and expert risk handling experience, a correspondence model between risk factors and intervention measures is extracted. This model associates "abnormal main engine temperature" with intervention measures such as "increasing inspection frequency" and "reducing load operation." Third, a rule reasoning engine is established. This engine receives the operational unit identifier and quantified risk level from the hierarchical risk list, matches and reasons based on the correspondence model, and outputs a set of candidate intervention measures for each risk point. This set of candidate intervention measures may include multiple options such as "suggesting a senior engineer to conduct a review" and "suggesting the activation of the auxiliary cooling system." Establish an optimization model solver. The optimization model solver receives a set of candidate intervention measures and a list of resources and constraints obtained in real time from the shipping operation site. The list of resources and constraints lists the currently available manpower, equipment, time, and operational procedure limitations, as shown in Table 2.

[0040] Table 2: List of Task Resources and Constraints

[0041] The optimization model solver constructs an optimal model with the ratio of operational resource consumption to risk reduction as the optimization objective. The optimal model uses the quantified risk level in the hierarchical risk list as the quantitative expression of the risk control objective, and the available resources and operational requirements in the resource and constraint list as the boundary conditions for resource constraints. The objective function of the optimal model can be expressed as:

[0042] Where: characters Indicates the target comprehensive value, character This represents the resource cost (such as manpower hours, equipment depreciation) consumed in implementing a candidate intervention. This represents the expected reduction in risk (calculated based on a quantified risk level) after implementing the candidate intervention. and characters These are the weighting coefficients for resource consumption and risk reduction, respectively. Solving this optimal model yields the optimal work unit management scheme and corresponding ideal resource configuration that maximizes the objective function while satisfying resource constraints. In implementation, the obtained work unit management scheme and corresponding ideal resource configuration are adjusted for operability according to actual shipping operation procedures and regulations. The standard operating procedures and maritime management regulations followed by the target shipping operation are identified, such as the company's "Main Engine Operation Manual" and relevant chapters of the "International Convention for the Safety of Life at Sea". The expected intervention methods for each work unit in the work unit management scheme are analyzed. These expected intervention methods are derived from the equipment state evolution trajectory and personnel behavior characteristic codes. For example, for the "Main Engine High Temperature Risk" work unit, the intervention "requiring the activation of backup cooling" is derived from the equipment state evolution trajectory, and the intervention "requiring operation by personnel with experience values ​​exceeding the threshold" is derived from relevant personnel behavior characteristic codes. The expected intervention methods are then verified for compliance and feasibility against the standard operating procedures and maritime management regulations. The startup procedure of the backup cooling pump is verified to conform to the manual, and the qualifications of the assigned personnel are checked to ensure they meet the requirements of the convention. After successful verification, fine-tuning is performed based on the ideal resource configuration and the availability of on-site manpower and equipment. For example, if the ideal configuration requires two engineers, but only one is available immediately, the fine-tuning sequence would be "one person performs the key steps first, and the other joins after completing their current task." This yields the final confirmed executable resource configuration plan and operation sequence, completing the operability adjustment. It can be understood that the rule reasoning engine and the optimization model solver are logically cascaded within the management strategy generation engine. The rule reasoning engine first performs knowledge-driven measure reasoning, and the optimization model solver then performs optimization calculations under resource constraints. Finally, the operable executable resource configuration plan and operation sequence, after operability adjustments, are structured and output as personnel operation suggestion sequences and a ship equipment maintenance plan, respectively. Together, these constitute the final adaptive management strategy.

[0043] In some embodiments, the rule-based reasoning engine can employ a generative rule system or an ontology-based inference engine to implement the correspondence model. The candidate intervention set can include attribute fields such as measure description, expected resource consumption estimate, and expected risk reduction estimate. The resource and constraint list is dynamically updated, reflecting real-time changes in manpower scheduling, equipment status, and operational progress at the shipping operation site. Optionally, the optimization model solver can employ linear programming, integer programming, or heuristic algorithms to solve the aforementioned optimal model. It is understood that the compliance and feasibility verification process can automatically compare data using a digitized procedural knowledge base, generating warnings and prompting adjustments for conflicting intervention methods. In some embodiments, the fine-tuning process for operability adjustments can involve the system providing several alternative solutions, which are then finalized by management personnel. The final generated adaptability management strategy clarifies the specific instruction framework of "who," "when," "what," and "how to configure the equipment."

[0044] See Figure 4 This is a comparative analysis chart of the effectiveness of shipping operation intervention measures, which intuitively presents the performance of different intervention measures on three core indicators. Risk reduction effectiveness: Activating backup equipment reduces risk the most (50%), making it the most effective risk control method. Replacing personnel reduces risk the least (25%), suitable only for low-priority risk scenarios. Resource consumption level: Increasing inspection frequency consumes the least resources (15), making it the most cost-effective intervention. Activating backup equipment consumes the most resources (40), requiring a risk-benefit balance. Comprehensive optimization score: Increasing inspection frequency scores the highest (85), achieving considerable risk reduction with low resource consumption. Replacing personnel scores the lowest (70), with the worst cost-effectiveness. Prioritizing the activation of backup equipment can quickly halve the risk, suitable for dealing with emergency high-risk situations. Increasing inspection frequency or adjusting routes is recommended to effectively reduce risk while controlling costs.

[0045] In one embodiment of the present invention, the adaptation management strategy is converted into a set of executable management instructions, the process of which is as follows: The system parses the personnel operation suggestion sequence and ship equipment maintenance plan in the adaptation management strategy, identifying all personnel instruction entries that need to be issued to specific personnel, equipment parameter entries that need to be configured for specific ship equipment, and their corresponding operation types. The system consults the instruction and parameter protocol manual that perfectly matches the target personnel terminal type and the target ship equipment controller model, mapping the identified abstract operation types to specific personnel instruction codes and ship equipment control parameters. The system analyzes the personnel operation suggestion sequence and ship equipment maintenance plan to determine the logical chain of human-machine collaborative execution, which indicates that the setting of certain ship equipment parameters is a prerequisite for relevant personnel to perform specific operations. Based on the determined logical chain, each personnel instruction code and each ship equipment control parameter is labeled with the preconditions that need to be met for execution and the expected effective time. Based on the preconditions and effective time, a time-series logic model is used to uniformly schedule and sort all specific personnel instruction codes and ship equipment control parameters, eliminating time and logic conflicts, and generating a linear execution queue that can be triggered sequentially or according to conditions. This queue is the final set of executable management instructions.

[0046] In practical implementation, the process of converting the adaptability management strategy into a set of executable management instructions is achieved through the following methods: The system parses the personnel operation suggestion sequence and ship equipment maintenance plan in the adaptability management strategy to identify the personnel instruction items to be issued, the ship equipment parameter items to be configured, and their corresponding operation types. Personnel instruction items may include descriptions such as "check the main engine lubricating oil pressure" and "check the status of the standby pump valve," while ship equipment parameter items may include descriptions such as "set the main engine speed to 85 RPM" and "start the No. 1 fuel circulation pump." The corresponding operation types are categorized into "confirm," "set," and "start." The system consults the instruction and parameter protocol manual that matches the target personnel terminal type and the target ship equipment controller model. For example, it consults the display instruction protocol that matches "Marine Tablet PC Type-A" and the Modbus communication protocol that matches "Main Engine Remote Control System Model-Z." The identified "confirm" operation type is mapped to the specific personnel instruction code "CMD_CONFIRM_001," and the "set speed" operation type is mapped to the specific ship equipment control parameter "Write_Register(40101,85)." In practical implementation, the analysis of personnel operation suggestion sequences and ship equipment maintenance plans determines the logical chain of human-machine collaborative execution. This logical chain indicates that the setting of certain ship equipment parameters is a prerequisite for relevant personnel to perform specific operations. For example, the logical chain indicates that "ship equipment control parameter 'start No. 1 fuel circulation pump'" can only be safely executed after "personnel instruction 'confirm fuel supply valve is open'". Based on the determined logical chain, each personnel instruction code and each ship equipment control parameter is labeled with preconditions and effective times. The precondition field records the status of other instructions or parameters that must be met to trigger the instruction, and the effective time field records the specific time point or relative time interval at which the instruction is scheduled to be executed. It can be understood that, based on preconditions and effective times, a temporal logic model is used to schedule and sort specific personnel instruction codes and ship equipment control parameters. The temporal logic model handles the dependencies and time constraints between instructions, generating a linear execution queue that is conflict-free in both time and logic. The scheduling process can be expressed as finding an instruction sequence that satisfies all constraints. Its objective function is:

[0047] Where: characters This represents the k-th instruction (personnel instruction code or ship equipment control parameter) in the sequence, represented by the character... Indication of instructions The actual time point at which the character is scheduled for execution. Indication of instructions The expected effective time is determined by adjusting the instruction order and actual execution time in the sequential logic model. This ensures that the actual execution time of all instructions is as close as possible to their expected effective time, while also ensuring that for any instruction... All instructions corresponding to its preconditions have been scheduled and completed before it. The linear execution queue obtained by solving this model is the final set of executable management instructions.

[0048] In some embodiments, the instruction and parameter protocol manual is stored in the form of a database table, containing fields such as "operation type," "target device model," "control instruction code / parameter address," and "data format." When marking the effective time of instructions, an absolute time format is used for operations with strict time window requirements, while a relative time format relative to the job start time can be used for sequential operations. Optionally, the solution of the timing logic model can use a topology sorting-based algorithm to handle logical dependencies, combined with heuristic rules to adjust and meet time window constraints. The generated linear execution queue is an ordered list, where each entry contains the instruction code, target object, execution parameters, triggering conditions, and planned time. It can be understood that the analysis of the logical chain can be achieved by establishing a directed graph between instructions and parameters, where nodes represent instructions and edges represent dependencies that "must be executed after...". In some embodiments, preconditions can be expressed as Boolean logical expressions, such as "Pump_Status==RUNNING&&Valve_Position==OPEN." It can be understood that the final scheduling order needs to ensure that there are no circular dependencies and that the triggering conditions of all instructions are satisfied at their planned execution times. The resulting set of executable management instructions can be directly sent to various execution terminals and controllers in the ship's network.

[0049] See Figure 5 This is a monitoring and analysis chart of management instruction execution and system load, which intuitively reflects the dynamic relationship between the number of instructions executed, resource utilization, and system load. The number of instructions executed remains at an extremely low level of 0-3, with minimal fluctuations, indicating that the instruction execution queue is well-organized and there are no large-scale concurrent execution scenarios. The peak occurs at the 40th second, with the remaining time between 0-2 instructions, demonstrating the stability of instruction timing scheduling. Resource utilization fluctuates dramatically, ranging from 52% to 84%, with peaks occurring at the 15th and 30th seconds. The high resource utilization periods are highly synchronized with the system load peak, reflecting the direct correlation between instruction execution and resource scheduling. The system load ranges from 55% to 85%, with the peak occurring at the 35th second (85%), showing a lag of approximately 5 seconds compared to the resource utilization peak. The overall load level is high, indicating that the system is operating under continuous high load during instruction execution.

[0050] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A unified management platform system for personnel and vessel information for shipping companies, characterized in that, include: The data acquisition and labeling module collects original human-machine operation records and ship equipment status records in shipping operation scenarios to form a raw record sequence. It performs format normalization and association labeling on the raw record sequence to generate a unified format operation record set. The operation record set includes personnel identification tags, multi-dimensional status time series of ship equipment, and environmental event time series. The feature analysis module performs multi-round identity feature extraction on the personnel identification tags to obtain personnel behavior feature codes, performs abnormal pattern recognition on the multi-dimensional state time series of the ship equipment to obtain the equipment state evolution trajectory, and performs event correlation analysis on the environmental event time series to obtain the environmental event correlation network. The feature fusion module inputs the personnel behavior feature code, the equipment state evolution trajectory and the environmental event association network into the feature fusion device for interactive evolution, generating a dynamic feature map of integrated human and ship; The risk assessment module, based on the integrated human-vessel dynamic feature map, constructs an operational risk assessment network to determine the dynamic risk level and form a hierarchical risk list; The strategy generation module generates adaptive management strategies for target shipping operations based on the hierarchical risk list through the management strategy generation engine, and converts the adaptive management strategies into a set of executable management instructions.

2. The integrated management platform system for personnel and vessel information of shipping enterprises according to claim 1, characterized in that, Multiple rounds of identity feature extraction are performed on the personnel identification tags to obtain personnel behavior feature codes, including: The personnel identification tags are matched with a preset personnel qualification database to output a preliminary set of qualification attributes; Historical operational behavior tracing and habit pattern mining are performed on the preliminary qualification attribute set to extract personnel operation pattern clusters; The current task scenario is embedded into the personnel operation mode cluster by a scenario encoder to generate scenario operation feature vectors; Feature compression technology is used to extract and encode the core features of the scenario-based operation feature vector to generate a distinctive personnel behavior feature code.

3. The integrated management platform system for personnel and vessel information of shipping enterprises according to claim 2, characterized in that, Anomaly pattern recognition is performed on the multi-dimensional state time series of the ship's equipment to obtain the equipment state evolution trajectory, including: Multiple monitoring dimensions are defined on the multi-dimensional state time series of the ship equipment. The original equipment state data is slide-segmented under different monitoring dimensions to form multi-angle state data segments. The stationarity test and deviation detection are performed on the equipment status data in each multi-angle status data segment to separate the time series of normal baseline component, abnormal fluctuation component and noise component. By performing spatiotemporal alignment and trend synthesis on the normal baseline component, abnormal fluctuation component, and noise component corresponding to different monitoring dimensions, the device state evolution trajectory reflecting the comprehensive changes in the device state across the three dimensions of normal, abnormal, and disturbance is constructed.

4. The integrated management platform system for personnel and vessel information of shipping enterprises according to claim 3, characterized in that, The personnel behavior feature codes, the equipment state evolution trajectory, and the environmental event association network are input into a feature fusion device for interactive evolution to generate a human-ship integrated dynamic feature map, including: An interactive evolution layer is established in the feature fusion device. The interactive evolution layer receives the personnel behavior feature code, the equipment state evolution trajectory and the environmental event association network as the initial node state. In the interactive evolution layer, a bidirectional influence link is established between feature nodes. The bidirectional influence link simulates the state coupling and causal transmission between the personnel behavior feature code, the equipment state evolution trajectory and the environmental event association network according to the preset causal logic rules. After a preset evolution cycle, the states of each feature node in the interactive evolution layer complete the interaction and reach stability. At this time, the stable node states after the interaction and their bidirectional influence link relationships are output in a graph form, forming the integrated human-ship dynamic feature graph.

5. The integrated management platform system for personnel and vessel information of shipping enterprises according to claim 4, characterized in that, Based on the aforementioned integrated human-vessel dynamic feature map, an operational risk assessment network is constructed to determine dynamic risk levels and generate a hierarchical risk list, including: The tiered risk list includes work unit identifiers and quantified risk levels; The feature nodes in the integrated human-ship dynamic feature map are used as the input layer nodes of the operation risk assessment network; A multi-layer risk perception neuron is constructed, and the features of the input layer nodes are non-linearly mapped and fused through the multi-layer risk perception neuron, and then passed to the output layer layer by layer. In the output layer, each job unit identifier corresponds to a risk assessment neuron, which converts the received fusion information into a continuous quantitative risk level value. All work unit identifiers and their corresponding quantitative risk level values ​​are automatically divided and sorted according to the level range of the quantitative risk level values ​​to generate the hierarchical risk list.

6. The integrated management platform system for personnel and vessel information of shipping enterprises according to claim 5, characterized in that, Based on the aforementioned tiered risk list, an adaptive management strategy for the target shipping operations is generated through a management strategy generation engine, including: The adaptability management strategy includes suggested sequences of personnel operations and ship equipment maintenance plans; The hierarchical risk list, along with the resource and constraint list obtained in real time from the shipping operation site, is input into the management strategy generation engine. In the management strategy generation engine, an optimal model of resource constraints and risk control objectives is established. The optimal model uses the quantified risk level in the hierarchical risk list as the quantitative expression of the risk control objectives, and the available resources and operational requirements in the resource and constraint list as the boundary conditions of resource constraints. Solving the optimal model yields a work unit management scheme that minimizes overall operational risk and the corresponding ideal resource configuration under the resource constraint boundary conditions. The operational unit management scheme and the corresponding ideal resource configuration are adjusted for operability according to the actual shipping operation process and specifications to generate the personnel operation suggestion sequence and the ship equipment maintenance plan, which together constitute the adaptability management strategy.

7. A unified management platform system for personnel and vessel information for shipping enterprises according to claim 6, characterized in that, The aforementioned work unit management scheme and corresponding ideal resource configuration are adjusted for operability according to actual shipping operation processes and standards, including: Identify the standard operating procedures and maritime regulations followed by the target shipping operations; The expected intervention method for each work unit in the work unit management scheme is analyzed, and the expected intervention method is derived from the equipment state evolution trajectory and the personnel behavior feature code; The expected intervention methods will be verified for compliance and feasibility against the standard operating procedures and maritime management regulations. After the verification is passed, the final executable resource configuration scheme and operation sequence are fine-tuned based on the size of the ideal resource configuration and the availability of on-site manpower and equipment, thus completing the operability adjustment.

8. The integrated management platform system for personnel and vessel information of shipping enterprises according to claim 7, characterized in that, The adaptation management strategy is converted into a set of executable management instructions, including: The personnel operation suggestion sequence and ship equipment maintenance plan in the adaptability management strategy are analyzed to identify the personnel instruction items that need to be issued, the ship equipment parameter items that need to be configured, and their corresponding operation types. Consult the instruction and parameter protocol manual that matches the target personnel terminal type and the target ship equipment controller model, and map the identified operation type to specific personnel instruction codes and ship equipment control parameters; According to the logic and time sequence implied in the personnel operation suggestion sequence and the ship equipment maintenance plan, the specific personnel instruction codes and ship equipment control parameters are arranged into an instruction sequence with execution conditions and triggering timing, forming the set of executable management instructions.

9. A unified management platform system for personnel and vessel information for shipping enterprises according to claim 8, characterized in that, Based on the logic and temporal sequence implied in the aforementioned personnel operation suggestion sequence and ship equipment maintenance plan, the specific personnel instruction codes and ship equipment control parameters are arranged into an instruction sequence with execution conditions and triggering timing, including: By analyzing the suggested sequence of personnel operations and the ship equipment maintenance plan, a logical chain for human-machine collaborative execution is determined. This logical chain indicates that the setting of certain ship equipment parameters is a prerequisite for relevant personnel to perform specific operations. Based on the aforementioned logic chain, preconditions and effective times are marked for each personnel instruction code and each ship equipment control parameter; Based on the aforementioned preconditions and effective time, a time-series logic model is used to schedule and sort the specific personnel instruction codes and ship equipment control parameters, generating a linear execution queue that is conflict-free in time and logic, which serves as the instruction sequence.

10. A unified management platform system for personnel and vessel information for shipping enterprises according to claim 9, characterized in that, The construction of the management strategy generation engine includes the following steps: Acquire a pre-defined risk control knowledge base, which includes historical operational accident cases, industry safety management standards, and expert risk handling experience; Based on the aforementioned historical operational accident cases, industry safety management standards, and expert risk handling experience, a model of the correspondence between risk factors and intervention measures was extracted. A rule-based reasoning engine is established, which is used to receive the work unit identifier and quantified risk level in the hierarchical risk list, and infer a set of candidate intervention measures based on the correspondence model. An optimization model solver is established, which receives the set of candidate intervention measures and the current operational resource constraints, and calculates the optimal intervention plan with the ratio of operational resource consumption to risk reduction as the optimization objective. The rule reasoning engine and the optimization model solver are logically cascaded to construct the management strategy generation engine. The management strategy generation engine generates the adaptive management strategy based on the hierarchical risk list and resource constraint list through a process of reasoning followed by optimization.

Images