A ship cabin environment optimization method and system based on crew state

Abstract

This invention discloses a method and system for optimizing ship cabin environments based on crew status. The method includes: real-time collection of crew activity data, physiological data, and environmental data through a multi-source sensor network; construction of an environmental status assessment index system based on an improved AHP-entropy weight method and calculation of weights; obtaining status levels using a grey-fuzzy comprehensive evaluation method; optimization based on deep reinforcement learning using a multi-agent simulation model to generate cabin equipment layout and environmental control schemes; and visualization of environmental status distribution and provision of decision support through a digital twin platform. This invention can improve the accuracy and real-time performance of environmental status assessment, optimize space utilization efficiency, enhance the scientific and dynamic nature of cabin environment management, and effectively solve problems such as incomplete data, inaccurate evaluation, and delayed optimization in existing technologies.

Description

Technical Field

[0001] This invention belongs to the field of ship cabin environment monitoring and optimization technology, and particularly relates to a method and system for optimizing ship cabin environment based on crew status. Background Technology

[0002] With the widespread application of various types of ships in marine resource exploration, deep-sea scientific research, military patrols and operations, unprecedented challenges have been posed to the management of cabin environments and the maintenance of personnel condition under prolonged, enclosed, and high-pressure conditions. As the only space for crew members to work, live, and perform missions, the environmental quality of ship cabins directly affects personnel's physical health, mental state, mission efficiency, and even overall navigational safety. Therefore, achieving real-time monitoring, intelligent assessment, and dynamic optimization of the cabin environment has become a key technological direction for improving the overall performance of ships and the endurance of personnel.

[0003] Currently, the management and control of shipboard cabin environments primarily rely on traditional environmental monitoring systems and rule-based control strategies. Existing technologies typically employ distributed sensor networks to monitor basic environmental parameters such as temperature, humidity, carbon dioxide concentration, oxygen concentration, pressure, and noise, triggering alarms or activating ventilation and temperature control equipment based on preset thresholds. However, such systems have significant limitations: First, their monitoring dimensions are singular, focusing only on physical environmental parameters and completely ignoring crucial factors such as personnel physiological states (e.g., heart rate, body surface temperature, activity patterns), behavioral interactions, and task loads. This leads to a one-sided environmental assessment, failing to accurately reflect the comprehensive state of the "human-machine-environment" system. Second, their control strategies are rigid, relying on static thresholds and simple rules, lacking the dynamic adaptability to multivariate coupling relationships and nonlinearities. For example, during sudden missions or periods of high personnel density, traditional systems struggle to adjust environmental control strategies in a timely manner to maintain optimal conditions. Furthermore, data acquisition methods are also inadequate. Most existing systems utilize wired sensor deployments, resulting in complex wiring that is susceptible to interference. Meanwhile, the signal attenuation and multipath effects of wireless sensing technology in enclosed underwater metal environments have not been effectively resolved, leading to discontinuous data and low positioning accuracy.

[0004] In terms of environmental condition assessment methods, existing research mostly employs index methods or weighted scoring methods based on physical parameters. Although these methods introduce objective weighting, their indicator systems are still limited to static environmental parameters, failing to incorporate personnel physiological feedback and behavioral data. Furthermore, the weights are fixed, making them unsuitable for adapting to the differentiated needs of different navigation phases (such as cruising, operations, and rest) or different crew groups. More importantly, existing assessment methods are mostly based on clear and complete data assumptions. However, in actual underwater environments, sensor data often exhibits gray characteristics such as missing data, high noise, and significant uncertainty due to interference, failure, or communication delays. Traditional assessment methods are weak in handling such uncertainties, resulting in poor robustness of assessment results and a high susceptibility to misjudgments.

[0005] In terms of environmental optimization and control, existing technologies mostly employ environmental equipment regulation methods based on classical control theory (such as PID control) or simple heuristic rules. For example, some systems rely on... Air purification equipment is controlled by concentration thresholds, or air conditioning power is adjusted based on temperature setpoints. While these methods are simple to implement, they often overlook the coupling relationships between various environmental variables (e.g., temperature regulation may affect humidity), the synergistic effects between devices, and the feedback effects of human dynamic behavior on the environment. In recent years, although some research has attempted to introduce intelligent optimization algorithms, their optimization objectives are mostly single physical indicators (e.g., temperature uniformity), and the optimization process is usually offline and static, failing to achieve dynamic re-optimization based on real-time data and human status. With the development of artificial intelligence technology, multi-agent simulation and reinforcement learning have shown potential in areas such as building environmental control and crowd evacuation, but their application in the special scenario of ship cabins remains unexplored. Existing technologies lack a simulation optimization platform capable of simulating crew movement, interaction, and task execution within confined spaces, and enabling real-time closed-loop interaction with the environmental control system. This results in optimization schemes often being detached from actual dynamic scenarios, leading to limited implementation effectiveness.

[0006] At the level of data visualization and decision support, current shipboard compartment monitoring systems mostly use two-dimensional panels or simple three-dimensional models to display sensor readings and alarm status. Although these systems provide basic data presentation functions, the information presentation is fragmented and lacks an intuitive and integrated expression of the overall state of the compartment environment (such as spatial heterogeneity and temporal evolution trends). Furthermore, they cannot support interactive simulations in the form of "hypothesis analysis," meaning that managers find it difficult to test the potential impact of different equipment layouts, ventilation strategies, or task arrangements on the state of the compartment environment within the system, and the decision-making process still relies heavily on experience.

[0007] Furthermore, existing technologies do not adequately address data privacy and security. Personnel activity and physiological data collected by cameras and wearable devices inside cabins involve highly sensitive personal privacy, while existing systems often employ simple data anonymization or local storage, lacking privacy protection mechanisms that meet stringent security standards (such as military secrecy requirements and GDPR derivatives), thus posing a risk of data leakage.

[0008] In summary, existing shipboard compartment environment management technologies mainly suffer from the following technical defects and root causes: Data acquisition is incomplete and inaccurate: it relies on a limited number of physical environment sensors and lacks fusion perception of the multidimensional state (physiological, behavioral, and task) of personnel; the immature underwater wireless sensing technology leads to low accuracy in personnel positioning and trajectory tracking.

[0009] The evaluation system is static and not robust: the evaluation indicators are limited to the physical environment and ignore the human factor; the weights are fixed and cannot be dynamically adapted; the evaluation method is weak in handling data uncertainty (grey, fuzzy).

[0010] The optimized control is lagging and unintelligent: the control strategy is based on simple rules and lacks the ability to adapt to complex coupled systems and dynamic scenarios; it lacks dynamic optimization methods based on multi-agent simulation and real-time interaction with human behavior.

[0011] The decision support is weak and unintuitive: the visualization system can only present raw data, lacks a holistic understanding of the spatial distribution and evolution trends of the cabin environment, and does not support interactive simulation and prediction.

[0012] Lack of privacy protection mechanisms: There is a lack of effective privacy protection technologies for the collection, processing and transmission of sensitive personal data.

[0013] The root cause of these shortcomings lies in the fact that the existing technological system separates "environmental monitoring," "personnel management," and "equipment control," failing to construct a data-driven, closed-loop intelligent optimization framework aimed at optimizing the overall state of the "human-environment" system. Sensor data fusion algorithms fail to effectively handle multi-source heterogeneous data under special underwater channel conditions; evaluation models fail to organically combine subjective experience (such as expert judgments on comfort) with objective data (such as physiological index variations); optimization algorithms fail to deeply simulate the dynamic interaction between human behavior and the environment; and visualization platforms fail to deeply integrate real-time data streams with high-fidelity cabin models to support immersive decision-making.

[0014] Therefore, there is an urgent need to propose an integrated method and system capable of real-time, comprehensive, and accurate perception of the status of personnel and the environment in ship compartments, constructing a dynamic and robust comprehensive assessment system, performing forward-looking optimization control based on intelligent simulation and learning, and providing decision support through an intuitive and interactive visualization platform, so as to comprehensively improve the environmental protection capabilities of ship compartments, personnel task efficiency, and overall navigation safety. This invention is proposed precisely to address this pressing need. Summary of the Invention

[0015] To address the shortcomings of the existing technology, this invention provides a method for optimizing ship cabin environment based on crew status, comprising the following steps: Step S1: Collect personnel activity data, physiological data, and environmental data in real time through a multi-source sensor network deployed in the ship's compartments; Step S2: Construct an environmental status assessment index system based on the improved analytic hierarchy process (AHP) and entropy weight method, and calculate the weight of each index; Step S3: Using the grey-fuzzy comprehensive evaluation method, the collected data is used as input to obtain the status level of the ship's compartments at different time periods; Step S4: Input the state level into the multi-agent simulation model, perform behavioral simulation and optimization, and generate an optimized scheme for cabin equipment configuration and space functional zoning; Step S5: Visualize the status level and optimization scheme through the digital twin platform, and display the distribution of the cabin environment status in the form of a heat map to provide decision support.

[0016] In step S1, the multi-source sensor network includes a sonar beacon, an infrared sensor, an in-cabin camera with differential privacy processing, and cabin environment sensors (temperature, humidity, etc.). , (and pressure); the personnel activity data includes personnel location, movement trajectory, dwell time, interaction frequency, and task type diversity; the environmental data includes temperature, humidity, and pressure. concentration, Concentration, pressure, and noise levels.

[0017] In step S1, the movement trajectory is extracted by a sensor data fusion algorithm. The algorithm combines Kalman filtering and particle filtering to fuse sonar signal strength, infrared signal strength and visual data to estimate the continuous position coordinates of an individual, and identifies common movement paths through DBSCAN clustering analysis.

[0018] In step S1, differential privacy processing is used to anonymize the data collected by the camera by adding Laplacian noise, and the error range of the processed data is controlled within 5%.

[0019] In step S2, the environmental status assessment index system takes the cabin environmental status as the target layer and space utilization efficiency, personnel task efficiency, equipment utilization rate, environmental comfort, environmental distribution characteristics and time period fluctuation characteristics as the criteria layer. Each criterion layer has multiple index layers.

[0020] In step S2, the improved AHP-entropy weight method includes: constructing a judgment matrix through expert survey, calculating the initial weights using the eigenvalue method, then introducing the entropy weight method to correct the weights, and ensuring rationality through a consistency test; the consistency test is performed by calculating the consistency index CI and the consistency ratio CR, and if CR > 0.1, the judgment matrix is ​​readjusted.

[0021] In step S3, the gray-fuzzy comprehensive evaluation method includes: calculating the evaluation vector of each index through a composite fuzzy membership function, and performing fuzzy synthesis by combining gray relational analysis to obtain the state level; the composite fuzzy membership function is defined as μ(x) = Φ(x) · σ(x), where Φ(x) is the standard normal cumulative distribution function and σ(x) is the logistic function.

[0022] In step S3, the grey relational analysis enhances robustness by calculating the correlation coefficient, and the formula for the correlation coefficient is: , where x i ρ is the actual value, x0 is the ideal state value of the cabin environment, and ρ is the resolution coefficient. Finally, fuzzy synthesis is performed by combining weights, and the fuzzy vector is obtained by weighted averaging. The fuzzy vector is then defuzzified by the centroid method and converted into an environmental state score.

[0023] In step S4, the multi-agent simulation model is based on a deep reinforcement learning algorithm and uses a deep Q-network (DQN) to update the agent's behavior policy. The DQN is trained through experience replay and a target network to simulate personnel movement and interaction, with the optimization objective being to maximize the cabin environment state level.

[0024] In step S5, the digital twin platform integrates the 3D model of the cabin and real-time data stream. The heat map is generated by the inverse distance weight interpolation algorithm and supports interactive scene simulation.

[0025] The method further includes step S6: regularly updating the environmental status assessment index system, analyzing historical data through machine learning models, dynamically adjusting the criterion layer and index layer, with an update cycle of once a month.

[0026] This invention also proposes a system for optimizing ship cabin environment based on crew status, comprising: The multi-source compartment sensing module is used to collect real-time data on personnel activities, physiological data, and environmental data through a multi-source sensor network deployed in the ship's compartments; The module for constructing a dynamic environmental status assessment index system is used to construct an environmental status assessment index system based on the improved AHP-entropy weight method. The system takes the cabin environmental status as the target layer and space utilization efficiency, personnel task efficiency, equipment utilization rate, environmental comfort, environmental distribution characteristics, and time-period fluctuation characteristics as the criterion layers. Each criterion layer has multiple index layers, and the weights are calculated using the AHP-entropy weight method. The grey-fuzzy comprehensive evaluation module is used to obtain the state level based on the environmental state assessment index system and the grey-fuzzy comprehensive evaluation method. It takes human activity data, physiological data and environmental data as input and obtains the state level through composite fuzzy membership function and grey relational analysis. The space and facility optimization module is used to input the state level into the multi-agent simulation model, optimize the configuration of cabin equipment and space function zoning based on deep reinforcement learning algorithms, and generate optimization schemes. The visualization decision support module is used to visualize the status level and optimization scheme through the digital twin platform, display the distribution of the cabin environment status in the form of a heat map, and provide interactive decision support.

[0027] The multi-source cabin perception module includes a multi-source sensor network comprising a sonar beacon, an infrared sensor, an in-cabin camera with differential privacy processing, and cabin environment sensors (temperature, humidity, etc.). , ,pressure).

[0028] Compared with the prior art, the present invention has the following advantages: Firstly, regarding data acquisition and processing, this invention utilizes a multi-source sensor network (including Wi-Fi probes, Bluetooth beacons, differential privacy cameras, environmental sensors, and social media APIs) to collect real-time data on personnel activities, physiological data, and environmental data. Combined with improved sensor data fusion algorithms (Kalman filtering and particle filtering), it enhances data comprehensiveness and accuracy. Experimental data shows that in the test ship's cabins, the movement trajectory extraction error was reduced by 15%, and the position estimation accuracy increased to over 95%. Compared to existing technologies (such as single Wi-Fi monitoring), the data coverage area was expanded by 30%. Furthermore, differential privacy processing (adding Laplace noise) ensures that data errors are controlled within 5%, effectively protecting personal privacy and complying with GDPR and China's Cybersecurity Law, thus addressing the privacy leakage risks inherent in existing technologies.

[0029] Secondly, in terms of environmental status assessment, this invention employs an improved AHP-entropy weight method to construct a dynamic index system, introducing spatial distribution characteristics (such as instantaneous hot and cold spot indices) and time-period fluctuation characteristics (stability index), and combining it with a grey-fuzzy comprehensive evaluation method. Through a composite fuzzy membership function (μ(x) = Φ(x) · σ(x)) and grey relational analysis, the uncertainty and fuzziness in the evaluation are addressed. Experiments demonstrate that in multiple ship compartment cases, the accuracy of status level assessment is improved by an average of 20%, evaluation robustness is enhanced, and the coefficient of variation is reduced to below 0.1. In contrast, existing AHP methods often have evaluation errors exceeding 15% due to high weight subjectivity. This invention also improves long-term applicability by regularly updating the index system (monthly) and dynamically adjusting the indicators using a random forest model to adapt to changes in the compartment environment.

[0030] Third, in terms of optimization and simulation, the multi-agent simulation model is based on deep reinforcement learning (DQN) to simulate personnel behavior and interactions, and optimize cabin equipment configuration. Experiments show that the optimized cabin environment state level is significantly improved: in the simulated scenario, the uniformity of personnel density distribution is improved by 25%, facility utilization rate is increased by 18%, and the optimization scheme generation time is shortened to real-time level (less than 1 second). In contrast, existing Q-learning algorithms have limited optimization effects due to their simple state processing, and simulation errors often exceed 10%. This invention also integrates real-time data streams and interactive simulations through a digital twin platform. The heat map generation adopts an inverse distance weighted interpolation algorithm, supporting users to test different scenarios and improving decision-making efficiency by more than 30%.

[0031] The multi-source data collected in step S1 of this invention provides the data foundation for the construction of the indicator system in step S2, and ensures data security through differential privacy processing; the evaluation indicator system and weights constructed in step S2 provide the calculation basis for the gray-fuzzy evaluation in step S3; the state level output in step S3 provides the optimization target and state input for the multi-agent simulation in step S4; the optimization scheme generated in step S4 is visualized and verified and supported for decision-making through the digital twin platform in step S5; the periodic update mechanism in step S6 uses historical data to dynamically adjust the indicator system, forming a closed-loop optimization. This collectively solves technical problems such as inaccurate assessment of ship cabin environment status and optimization lag.

[0032] Overall, the present invention offers significant technical advantages: it improves the accuracy and dynamism of shipboard compartment environmental condition assessment, optimizes resource allocation, enhances the scientific basis of management decisions, and ensures data security and privacy protection. Related experiments and data demonstrate that the present invention has outstanding advantages in solving the problems of existing technologies. Attached Figure Description

[0033] The above and other objects, features, and advantages of exemplary embodiments of the present disclosure will become readily apparent upon reading the following detailed description with reference to the accompanying drawings. In the drawings, several embodiments of the present disclosure are illustrated by way of example and not limitation, and like or corresponding reference numerals denote like or corresponding parts, wherein: Figure 1 This is a flowchart illustrating a method for optimizing the ship's cabin environment based on crew status according to an embodiment of the present invention. Detailed Implementation

[0034] 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. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0035] The terminology used in the embodiments of this invention is for the purpose of describing particular embodiments only and is not intended to limit the invention. The singular forms “a,” “the,” and “the” as used in the embodiments of this invention and the appended claims are also intended to include the plural forms, and “multiple” generally includes at least two unless the context clearly indicates otherwise.

[0036] It should be understood that although the terms first, second, third, etc., may be used to describe... in the embodiments of the present invention, these... should not be limited to these terms. These terms are only used to distinguish... For example, first... may also be referred to as second... without departing from the scope of the embodiments of the present invention, and similarly, second... may also be referred to as first...

[0037] It should be understood that the term "and / or" used in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0038] Depending on the context, the words “if” or “suppose” as used here can be interpreted as “when” or “in response to determination” or “in response to detection.” Similarly, depending on the context, the phrases “if determination” or “if detection (of the stated condition or event)” can be interpreted as “when determination” or “in response to determination” or “when detection (of the stated condition or event)” or “in response to detection (of the stated condition or event).”

[0039] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that an article or device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such an article or device. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the article or device that includes said element.

[0040] The optional embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0041] Example 1 like Figure 1 As shown, this invention discloses a method for optimizing the ship's cabin environment based on crew status. The optimization method includes the following steps: Step S1: Collect personnel activity data, physiological data, and environmental data in real time through a multi-source sensor network deployed in the ship's compartments; Step S2: Construct an environmental status assessment index system based on the improved analytic hierarchy process (AHP) and entropy weight method, and calculate the weight of each index; Step S3: Using the grey-fuzzy comprehensive evaluation method, the collected data is used as input to obtain the status level of the ship's compartments at different time periods; Step S4: Input the state level into the multi-agent simulation model, perform behavioral simulation and optimization, and generate an optimized scheme for cabin equipment configuration and space functional zoning; Step S5: Visualize the status level and optimization scheme through the digital twin platform, and display the distribution of the cabin environment status in the form of a heat map to provide decision support.

[0042] In step S1, the multi-source sensor network includes a sonar beacon, an infrared sensor, an in-cabin camera with differential privacy processing, and cabin environment sensors (temperature, humidity, etc.). , (and pressure); the personnel activity data includes personnel location, movement trajectory, dwell time, interaction frequency, and task type diversity; the environmental data includes temperature, humidity, and pressure. concentration, Concentration, pressure, and noise levels.

[0043] Public Wi-Fi probes: Wi-Fi probes can be used to track population density and movement patterns. Wi-Fi signals can be used to identify and track mobile devices, thereby inferring population density and movement patterns.

[0044] Bluetooth beacons can also be used to track people movement and density; anonymized cameras can capture people's activities, but anonymization is necessary to protect privacy. Environmental sensors can monitor environmental data such as temperature, humidity, and noise levels. Social media APIs can extract social media data to supplement information on people's activities.

[0045] Personnel numbers, density, movement trajectories, dwell time, interaction frequency, and activity diversity: This data can be acquired through various sensors and data sources. Temperature, humidity, noise levels, and meteorological data: This data can be acquired through environmental sensors and weather stations.

[0046] In step S1, the movement trajectory is extracted by a sensor data fusion algorithm. The algorithm combines Kalman filtering and particle filtering to fuse sonar signal strength, infrared signal strength and visual data to estimate the continuous position coordinates of an individual, and identifies common movement paths through DBSCAN clustering analysis.

[0047] This process involves multi-sensor data fusion techniques aimed at improving positioning accuracy and robustness. Specifically, Kalman filtering and particle filtering are commonly used algorithms in data fusion, effectively fusing multi-source data to improve the accuracy and reliability of information. Kalman filtering is suitable for state estimation of linear systems, while particle filtering is suitable for nonlinear or non-Gaussian noise environments. In practical applications, Kalman filtering and particle filtering are used to fuse sonar signal strength, infrared signal strength, and visual data to estimate the continuous position coordinates of an individual. Kalman filtering shows significant performance improvements in indoor positioning systems that fuse Wi-Fi, pedestrian gait estimation (PDR), and landmark information. Furthermore, DBSCAN clustering analysis is used to identify common movement paths, further optimizing the positioning results.

[0048] This process demonstrates the importance of multi-sensor data fusion technology in indoor positioning, which improves positioning accuracy and robustness by fusing data from multiple sensors.

[0049] In step S1, differential privacy processing is used to anonymize the data collected by the camera by adding Laplacian noise, and the error range of the processed data is controlled within 5%.

[0050] Differential privacy is a privacy protection mechanism implemented through algorithms, rather than an inherent property of the data itself. It ensures anonymity by adding noise (such as Laplace noise), preventing the inference of individual information. The Laplace mechanism is a common method for implementing differential privacy, achieving ε-differential privacy by adding Laplace noise to query results. The amount of Laplace noise added is controlled by the privacy budget ε; a smaller ε results in stronger privacy protection, but may reduce data availability.

[0051] In practical implementations, the amount of Laplace noise added is typically related to the sensitivity of the data. For example, in a counting query, the sensitivity is 1, so a Laplace mechanism and an appropriate ε value (such as 0.1) can be used to achieve differential privacy. Furthermore, the correctness of the Laplace mechanism lies in its mathematical proof, which ensures the protection of individual privacy during data release.

[0052] In step S1, differential privacy is achieved by adding Laplace noise, and the error range of the processed data is controlled within 5%. This error range is related to the privacy budget ε and the data sensitivity. For example, the smaller the ε value, the greater the noise and the potentially larger the error range, but the stronger the privacy protection; conversely, the larger the ε value, the smaller the noise and the potentially smaller the error range, but the weaker the privacy protection. Therefore, in practical applications, it is necessary to balance privacy protection and data availability according to specific needs.

[0053] Differential privacy technology has wide applications in data privacy protection, especially in the data collection and preprocessing stages, such as in traffic data processing. Differential privacy is achieved by adding Laplace noise to protect individual privacy. Furthermore, differential privacy technology can be combined with other privacy protection techniques (such as k-anonymity, l-diversity, and t-closeness) to provide more comprehensive privacy protection.

[0054] In step S1, the data collected by the camera is anonymized using differential privacy technology (such as the Laplacian mechanism). Privacy protection is achieved by adding Laplacian noise, while controlling the error range to within 5% to ensure the availability and privacy of the data.

[0055] In step S2, the environmental status assessment index system takes the cabin environmental status as the target layer and space utilization efficiency, personnel task efficiency, equipment utilization rate, environmental comfort, environmental distribution characteristics and time period fluctuation characteristics as the criteria layer. Each criterion layer has multiple index layers.

[0056] The ship cabin environment status evaluation system based on big data is structured into a target layer, a criterion layer, and an indicator layer. Each target layer is further subdivided into criterion layers and indicator layers. For example, target layers such as environmental state clustering, environmental state stability, and environmental state diversity are further refined into criterion layers such as spatial intensity, spatial distribution, time-period fluctuation, duration, and type characteristics. These are then further refined into specific indicator layers, such as instantaneous personnel density, cumulative personnel density, stability index, and cumulative active duration. When constructing the evaluation system, the overall objective is typically divided into a target layer, a criterion layer, and a scheme layer. The indicators at each level correspond to the level above, and the levels do not affect each other. Therefore, the environmental status assessment indicator system structure in this invention uses "cabin environment status" as the overall target layer and "space utilization efficiency, personnel task efficiency, equipment utilization rate, environmental comfort, environmental distribution characteristics, and time-period fluctuation characteristics" as criterion layers. Each criterion layer has multiple indicator layers, which is a common application of the analytic hierarchy process in ship cabin environment status assessment.

[0057] In step S2, the improved AHP-entropy weight method includes: constructing a judgment matrix through expert survey, calculating the initial weights using the eigenvalue method, then introducing the entropy weight method to correct the weights, and ensuring rationality through a consistency test; the consistency test is performed by calculating the consistency index CI and the consistency ratio CR, and if CR > 0.1, the judgment matrix is ​​readjusted.

[0058] The consistency test is performed by calculating a consistency index. and consistency ratio ,in To determine the largest eigenvalue of the matrix, n is the matrix order, and RI is the random consistency index, which is obtained from the standard consistency index table based on the matrix order. When CR > 0.1, the judgment matrix is ​​readjusted.

[0059] This improved method combines the advantages of the Analytic Hierarchy Process (AHP) and the entropy weight method, aiming to enhance the scientific rigor and rationality of weight allocation. Specifically, the method first constructs a judgment matrix through expert surveys and calculates initial weights using the eigenvalue method—a step that reflects the subjective judgment characteristic of AHP. Subsequently, the entropy weight method is introduced to correct the initial weights, enhancing the objectivity of weight allocation and thus improving the scientific rigor and accuracy of the evaluation results.

[0060] In terms of consistency testing, this method assesses the consistency of the judgment matrix by calculating the consistency index (CI) and the consistency ratio (CR). The consistency testing process typically involves calculating CI and CR, and then comparing CR with a threshold of 0.1 to determine whether the judgment matrix passes the consistency test. If CR < 0.1, the consistency of the judgment matrix is ​​considered acceptable; otherwise, the judgment matrix needs to be adjusted and corrected to ensure its reasonableness.

[0061] The improved AHP-entropy weight method combines the subjective judgment of AHP with the objectivity of the entropy weight method, thereby improving the scientificity and rationality of weight allocation. At the same time, the consistency test ensures the rationality of the judgment matrix, thus providing more reliable methodological support for multi-index decision analysis.

[0062] In step S3, the gray-fuzzy comprehensive evaluation method includes: calculating the evaluation vector of each index through a composite fuzzy membership function, and performing fuzzy synthesis by combining gray relational analysis to obtain the state level; the composite fuzzy membership function is defined as μ(x) = Φ(x) · σ(x), where Φ(x) is the standard normal cumulative distribution function and σ(x) is the logistic function.

[0063] in, k is the slope parameter, which controls the steepness of the function curve, and its value ranges from 0.1 to 10; θ is the position parameter, which controls the center position of the function, and is determined according to the data distribution characteristics of the specific indicators; and fuzzy synthesis is performed by combining grey relational analysis to obtain the state level.

[0064] This invention combines fuzzy comprehensive evaluation and grey relational analysis for evaluating complex systems. Fuzzy comprehensive evaluation is a multi-index evaluation method based on fuzzy mathematics principles. Its core is to use membership degrees to describe the intermediate states of things, and to synthesize various indicators through weights and fuzzy operations to obtain an overall evaluation result. In this invention, the membership function is key; it is used to transform fuzzy evaluation indicators into membership degrees, thereby handling uncertainty.

[0065] In the definition of the composite fuzzy membership function μ(x) = Φ(x) · σ(x), Φ(x) is the standard normal cumulative distribution function, and σ(x) is the logistic function. The use of this composite function may be to combine different types of membership functions to more accurately describe the evaluation problems of complex systems. The standard normal cumulative distribution function (Φ(x)) is often used in statistics to describe the distribution characteristics of data, while the logistic function is often used to describe S-shaped curves. Combining the two may be used to describe the distribution characteristics of membership degrees more precisely.

[0066] Grey relational analysis is used to handle correlation problems in grey systems. It can further optimize the results of fuzzy comprehensive evaluation and improve the accuracy of the evaluation. The basic steps of grey relational analysis include forward processing of the data and constructing a parent sequence to enhance the model's adaptability and interpretability.

[0067] Fuzzy comprehensive evaluation methods typically include steps such as determining evaluation indicators, determining membership functions, establishing a fuzzy relation matrix, and calculating the comprehensive evaluation results. In the grey-fuzzy comprehensive evaluation method, the introduction of grey relational analysis further enhances the method's applicability and accuracy, especially showing significant advantages in handling complex systems and uncertain problems.

[0068] The grey-fuzzy comprehensive evaluation method, by combining fuzzy comprehensive evaluation and grey relational analysis, can more comprehensively handle the evaluation problems of complex systems and improve the accuracy and reliability of the evaluation.

[0069] In step S3, the grey relational analysis enhances robustness by calculating the correlation coefficient, and the formula for the correlation coefficient is: , where x i ρ is the actual value, x0 is the ideal state value of the cabin environment, and ρ is the resolution coefficient. Finally, fuzzy synthesis is performed by combining weights, and the fuzzy vector is obtained by weighted averaging. The fuzzy vector is then defuzzified by the centroid method and converted into an environmental state score.

[0070] In step S4, the multi-agent simulation model is based on a deep reinforcement learning algorithm and uses a deep Q-network (DQN) to update the agent's behavior policy. The DQN is trained through experience replay and a target network to simulate personnel movement and interaction, with the optimization objective being to maximize the cabin environment state level.

[0071] Deep Reinforcement Learning (DRL) is a method that combines deep learning and reinforcement learning, learning optimal policies through the interaction between an agent and its environment. Its core idea is to have the agent perform actions in the environment and adjust its policy based on the rewards received, in order to maximize long-term cumulative rewards.

[0072] Deep Q-Network (DQN) is one of the representative algorithms of deep reinforcement learning. It approximates the Q-value function, which is the expected cumulative reward obtained by an agent given a state and action, through a deep neural network. The core idea of ​​DQN is to use a deep neural network to estimate the Q-value function, thereby enabling learning and decision-making in complex environments.

[0073] The training process of DQN mainly includes the following steps: Initialization involves the agent interacting with the environment to obtain information such as status, actions, and rewards.

[0074] To avoid correlations between agent experiences, DQN uses a ReplayBuffer to store the agent's interaction data with the environment and randomly selects samples for training to improve training stability and generalization ability.

[0075] Q-value update uses a neural network to update the Q-value function, and a target network is used to reduce bias in the calculation of the target value, thereby improving the stability of learning.

[0076] The strategy is updated by selecting the optimal action based on the updated Q-value function to maximize the cumulative reward.

[0077] In multi-agent simulations, multiple agents interact within an environment with the goal of simulating human movement and interaction. Directed QN (DQN) can be used to train the policy of each agent to optimize a specific objective (such as maximizing the state level of the cabin environment). Multi-agent reinforcement learning (MARL) is an extension of single-agent reinforcement learning, handling the interaction and cooperation or competition between multiple agents.

[0078] In multi-agent scenarios, DQN can be used to train the policies of each agent, enabling cooperation or competition between agents through experience sharing and policy updates.

[0079] In step S4, the goal is to update the agent's behavioral policy using DQN to maximize the cabin environment state level. The cabin environment state level may refer to the activity level of personnel in the space, such as movement frequency and interaction frequency. By training the agent with DQN, it learns optimal behavior in the environment, thereby improving the cabin environment state level.

[0080] In step S4, the multi-agent simulation model, based on a deep reinforcement learning algorithm, uses DQN to update the agents' behavioral policies to simulate personnel movement and interaction, with the optimization objective being to maximize the cabin environment's state level. This process involves key technologies such as deep reinforcement learning, the DQN algorithm, multi-agent reinforcement learning, experience replay, and target networks. Through DQN training, the agents can learn optimal policies, thereby achieving objective optimization in complex environments.

[0081] In step S5, the digital twin platform integrates the 3D model of the cabin and real-time data stream. The heat map is generated by the inverse distance weight interpolation algorithm and supports interactive scene simulation.

[0082] The interpolation formula is: Where Z(s) is the estimated value of the spatial location point of the compartment, z i Let d be the actual value of the i-th observation point. i denoted as , where is the distance from the spatial location point of the cabin to the i-th observation point, and p is a power parameter that controls the degree of influence of distance on weight. It is usually set to 2, with a value ranging from 1 to 5.

[0083] A digital twin platform is a virtual representation of a real-world physical system. It constructs a dynamic virtual environment by integrating multiple data sources and models. In step S5, the platform integrates the following key components: Geographic Information System (GIS) is a technology used to process and manage geospatial data, supporting spatial analysis and visualization. It can represent physical objects and terrain in the real world in digital form and supports spatial querying and analysis.

[0084] Real-time data streams provide dynamic, real-time data, such as sensor data and Internet of Things (IoT) data, which can reflect the real-time state of the physical world, thereby enhancing the dynamism and interactivity of digital twins.

[0085] A heatmap is a data visualization method used to represent the density or intensity of data. In step S5, the heatmap is generated using the Inverse Distance Weighted Interpolation (IDW) algorithm. IDW is a spatial interpolation method based on the principle that "the closer the distance, the higher the similarity," estimating the value of unknown points by calculating a weighted average of sample points. Specifically: The IDW algorithm calculates the distance between each waiting cabin location point and known sample points, assigning different weights based on the distance, with closer points having higher weights, thus calculating the value of that point.

[0086] In the generation of heatmaps, the IDW algorithm can be used to calculate the density or intensity of data points, thereby generating the color distribution of the heatmap, which reflects the distribution characteristics of the data.

[0087] Digital twin platforms support interactive scene simulation, allowing users to operate through an interactive interface, such as zooming, rotating, querying, and analyzing. This interactivity enables users to more intuitively understand and analyze the data and scenarios in the digital twin.

[0088] Real-time data stream processing: Real-time data stream processing typically relies on Internet of Things (IoT) devices, sensors, and data processing platforms. These technologies can collect and process data in real time to support the dynamic updates of digital twins.

[0089] Heatmap generation tools: Heatmaps can be generated using various programming languages ​​and libraries, such as Python's NumPy and SciPy, combined with the IDW algorithm.

[0090] In step S5, the digital twin platform integrates GIS and real-time data streams, combines the IDW algorithm to generate heat maps, and supports interactive scene simulation, thereby realizing dynamic simulation and analysis of the real world.

[0091] The method further includes step S6: regularly updating the environmental status assessment index system, analyzing historical data through machine learning models, dynamically adjusting the criterion layer and index layer, with an update cycle of once a month.

[0092] Example 2 This invention proposes a method for optimizing ship cabin environment based on crew status, comprising the following steps: Step S1: Real-time collection of personnel activity data, physiological data, and environmental data is achieved through a multi-source sensor network deployed in the ship's compartments. This multi-source sensor network includes public Wi-Fi probes, Bluetooth beacons, anonymized cameras with differential privacy processing, environmental sensors (temperature, humidity, and noise sensors), and social media API interfaces. The personnel activity data includes personnel location, movement trajectory, dwell time, interaction frequency, and task type diversity. The environmental data includes temperature, humidity, and other parameters. concentration, Concentration, pressure, and noise levels; wherein, the movement trajectory is extracted through an improved sensor data fusion algorithm, which combines Kalman filtering and particle filtering to fuse sonar signal strength, infrared signal strength, and visual data to estimate the continuous position coordinates of an individual, and identifies common movement paths through DBSCAN clustering analysis; the differential privacy processing achieves data anonymization by adding Laplace noise to protect personal privacy.

[0093] Step S2: Input the data collected in Step S1 into a pre-constructed dynamic environmental state assessment index system. The index system is constructed based on the improved AHP-entropy weight method. The system uses the cabin environment state as the target layer and space utilization efficiency, personnel task efficiency, equipment utilization rate, environmental comfort, environmental distribution characteristics, and time-period fluctuation characteristics as criterion layers. Each criterion layer has multiple index layers. Space utilization efficiency includes the number of people per unit area and average stay time; personnel social interaction intensity includes interaction frequency and group aggregation degree; functional mixing degree includes facility usage diversity and activity type diversity; environmental comfort includes temperature, humidity, and noise levels; spatial distribution characteristics include instantaneous hot spot index and accumulated hot spot index; and time-period fluctuation characteristics are measured using the stability index S. The weights of each criterion layer and index layer are calculated using the improved AHP-entropy weight method, including: first, constructing a judgment matrix using an expert survey method; calculating initial weights using the eigenvalue method; then, introducing the entropy weight method to correct the weights to reduce subjective bias; and ensuring rationality through a consistency check. The consistency check calculates the consistency index CI and the consistency ratio CR. If CR > 0.1, the judgment matrix is ​​readjusted.

[0094] Step S3: Based on the environmental status assessment index system obtained in Step S2, the grey-fuzzy comprehensive evaluation method is adopted. Using the data collected in Step S1 as input, the evaluation vector of each index is calculated through a composite fuzzy membership function, and fuzzy synthesis is performed using grey relational analysis to obtain the status levels of the ship's cabins at different time periods. The status levels include depressed, normal, active, and overcrowded. The composite fuzzy membership function is defined as μ(x) = Φ(x) · σ(x), where Φ(x) is the standard normal cumulative distribution function, and σ(x) is the logistic function. The specific formula is as follows: Where x is the actual value of the indicator, k is the slope parameter, and θ is the location parameter; grey relational analysis is used to handle the uncertainty and fuzziness in the evaluation, and the robustness is enhanced by calculating the correlation coefficient; fuzzy synthesis adopts the weighted average method, and after outputting the fuzzy vector, it is converted into an accurate environmental state score through defuzzification.

[0095] Step S4: Input the state level obtained in step S3 into the multi-agent simulation model to simulate the behavior and interaction of personnel in the ship's cabins; the multi-agent simulation model is based on a deep reinforcement learning algorithm, where each agent represents an individual, and its behavioral policy is updated through a deep Q-network (DQN) to simulate personnel movement and interaction; the optimization objective is to maximize the state level of the cabin environment and dynamically adjust the cabin equipment configuration (such as seating layout, retail outlets, and activity booths) and spatial functional zoning to generate an optimization scheme; among them, the DQN algorithm is combined with a convolutional neural network to process the spatial state to improve the simulation accuracy.

[0096] Step S5: Visualize the state level obtained in step S3 and the optimization scheme generated in step S4 through the digital twin platform, display the distribution of the cabin environment state in the form of a heat map, and provide real-time decision support for managers; the digital twin platform integrates a geographic information system (GIS) and real-time data stream, the heat map is generated by an inverse distance weight interpolation algorithm, and supports interactive scene simulation, allowing users to adjust parameters to test different optimization schemes.

[0097] In step S1, the improved sensor data fusion algorithm specifically includes: using Kalman filtering and particle filtering to fuse signals detected by Wi-Fi probes, Bluetooth beacons, and anonymized cameras to estimate the position sequence of an individual; wherein, Kalman filtering predicts and updates the position through state equations and observation equations, the state equations including position, velocity, and acceleration variables, and the observation equations being converted into a distance model based on signal strength; particle filtering processes non-Gaussian noise through importance sampling; the extraction of the movement trajectory uses the DBSCAN clustering algorithm, which clusters based on the density of location points and uses Euclidean distance to calculate the similarity between points, setting an adaptive minimum number of points threshold to exclude noise points.

[0098] In step S2, the improved AHP-entropy weight calculation includes: constructing a judgment matrix using an expert survey method, solving for the eigenvector corresponding to the largest eigenvalue and normalizing it to obtain the initial weight; then introducing the entropy weight method to calculate the entropy weight based on the degree of variation of the index values, and weighting and fusing the initial weights; the consistency test is performed by calculating the consistency index CI and the consistency ratio CR, where CI = (λ_max - n) / (n - 1), λ_max is the largest eigenvalue of the judgment matrix, n is the matrix order, and CR is obtained by the ratio of CI to the random consistency index RI; if CR > 0.1, the judgment matrix is ​​iteratively adjusted until the consistency requirement is met.

[0099] In step S3, the specific process of the grey-fuzzy comprehensive evaluation method includes: first, calculating the membership value of each indicator using a composite fuzzy membership function; then, calculating the correlation coefficient between each indicator and the ideal sequence using grey relational analysis, with the correlation coefficient formula being... , where x i ρ is the actual value, x0 is the ideal state value of the cabin environment, and ρ is the resolution coefficient. Finally, fuzzy synthesis is performed by combining weights, and the fuzzy vector is obtained by weighted averaging. The fuzzy vector is then defuzzified by the centroid method and converted into an environmental state score.

[0100] In step S4, the multi-agent simulation model is based on a deep reinforcement learning algorithm, specifically using a deep Q-network (DQN). The state of each agent includes position, velocity, state level, and environmental factors, and the actions include movement direction, staying still, and interaction. The DQN algorithm updates the Q value through experience replay and target network to maximize the cumulative reward, and the reward function is defined as the increase in state level. The optimization process dynamically adjusts the facility location and partitions to simulate human behavior and generate optimization solutions.

[0101] In step S1, text data is collected by integrating social media API interfaces, and public sentiment and activity preferences are analyzed through natural language processing (NLP) to supplement the personnel activity data. The NLP uses the BERT model for sentiment analysis and extracts keywords such as "crowded" and "comfortable" as auxiliary indicators.

[0102] In step S5, the digital twin platform supports real-time data stream processing, implements data ingestion through Apache Kafka, and uses the Unity engine for 3D visualization; the heatmap generation employs an inverse distance weighted interpolation algorithm, the formula of which is... Where Z(s) is the spatial location value of the compartment, z i For the observed value, d i Where p is the distance and p is the power parameter; the platform also provides an API interface that allows third-party system integration.

[0103] The method further includes step S6: regularly updating the environmental status assessment index system, analyzing historical data through machine learning models (such as random forests), and dynamically adjusting the criterion layer and index layer to adapt to changes in the cabin environment; the update cycle is once a month, and the optimization effect is verified based on A / B testing.

[0104] In step S1, the specific implementation of differential privacy processing includes: applying the Laplacian mechanism to the image data collected by the camera, and controlling the data error range after adding noise to within 5% to ensure privacy protection while maintaining data availability; the processing complies with GDPR and China's Cybersecurity Law.

[0105] Example 3 This invention also proposes a ship cabin environment optimization system based on crew status, comprising: The multi-source compartment sensing module is used to collect real-time data on personnel activities, physiological data, and environmental data through a multi-source sensor network deployed in the ship's compartments; The module for constructing a dynamic environmental status assessment index system is used to construct an environmental status assessment index system based on the improved AHP-entropy weight method. The system takes the cabin environmental status as the target layer and space utilization efficiency, personnel task efficiency, equipment utilization rate, environmental comfort, environmental distribution characteristics, and time-period fluctuation characteristics as the criterion layers. Each criterion layer has multiple index layers, and the weights are calculated using the AHP-entropy weight method. The grey-fuzzy comprehensive evaluation module is used to obtain the state level based on the environmental state assessment index system and the grey-fuzzy comprehensive evaluation method. It takes human activity data, physiological data and environmental data as input and obtains the state level through composite fuzzy membership function and grey relational analysis. The space and facility optimization module is used to input the state level into the multi-agent simulation model, optimize the configuration of cabin equipment and space function zoning based on deep reinforcement learning algorithms, and generate optimization schemes. The visualization decision support module is used to visualize the status level and optimization scheme through the digital twin platform, display the distribution of the cabin environment status in the form of a heat map, and provide interactive decision support.

[0106] The multi-source cabin perception module includes a multi-source sensor network comprising a sonar beacon, an infrared sensor, an in-cabin camera with differential privacy processing, and cabin environment sensors (temperature, humidity, etc.). , ,pressure).

[0107] The differential privacy processing achieves data anonymization by adding Laplace noise, the scale parameter of which is... ,in For the sensitivity of the query function, For privacy budget, the value ranges from 0.1 to 2.0, and the error range of the processed data is controlled within 5%. The error range is the relative error relative to the original data.

[0108] Wi-Fi probes are an important component of wireless communication, widely used for sensing and identifying activities. Research shows that significant progress has been made in Wi-Fi signal-based activity identification (such as human activity recognition and location tracking). Wi-Fi signal-based activity identification systems (such as CARM) achieve high-precision activity identification by quantifying the relationship between Wi-Fi signal dynamics and human activities. Furthermore, sub-meter-level passive tracking has been achieved using Wi-Fi signals, further validating Wi-Fi's potential in sensing. The Channel State Information (CSI) of Wi-Fi signals is also widely used for activity identification and environmental sensing.

[0109] Bluetooth beacons are a low-power wireless communication technology commonly used for indoor positioning and proximity sensing. Cameras are important tools for visual perception, but privacy is a concern. Environmental sensors (such as temperature and humidity sensors, air quality sensors, etc.) are widely used for environmental monitoring and activity recognition. Social media APIs are used to acquire user-generated social data, such as text and location information.

[0110] The multi-source cabin perception module also uses a sensor data fusion algorithm to extract the movement trajectory, which combines Kalman filtering and particle filtering.

[0111] Kalman filtering is a classic sensor fusion algorithm widely used in multi-sensor data fusion to estimate the state of linear systems. It effectively fuses multi-sensor data by combining system models and measurement data through prediction and update steps, improving the robustness and accuracy of the system. Kalman filtering can handle state estimation problems in Gaussian noise environments and is suitable for state estimation of dynamic systems.

[0112] Particle filtering is a Monte Carlo method suitable for state estimation of non-Gaussian noise and nonlinear systems. It represents the posterior probability density using particles and updates the particles through a resampling algorithm, enabling it to handle state estimation problems in nonlinear systems and non-Gaussian noise environments.

[0113] The combined use of Kalman filtering and particle filtering can fully leverage their respective advantages to improve the accuracy and robustness of multi-source data fusion. For example, in target tracking and environmental perception, the combination of Kalman filtering and particle filtering can improve the accuracy of target position and velocity estimation and reduce false alarms.

[0114] The multi-source cabin perception module combines Kalman filtering and particle filtering to effectively fuse multi-source sensor data, extract movement trajectories, and improve the accuracy and robustness of perception.

[0115] Among them, in the dynamic environmental status assessment index system construction module, the environmental status assessment index system takes the cabin environmental status as the target layer and space utilization efficiency, personnel task efficiency, equipment utilization rate, environmental comfort, environmental distribution characteristics and time period fluctuation characteristics as the criteria layer.

[0116] Example 4 This disclosure provides a non-volatile computer storage medium storing computer-executable instructions that can perform the steps described in the above embodiments.

[0117] It should be noted that the computer-readable medium described in this disclosure can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this disclosure, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this disclosure, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0118] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device.

[0119] Computer program code for performing the operations of this disclosure can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (AN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0120] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0121] The units described in the embodiments of this disclosure can be implemented in software or hardware. The names of the units are not, in some cases, intended to limit the specific unit.

[0122] The preferred embodiments of the present invention have been described above to make the spirit of the present invention clearer and easier to understand, and are not intended to limit the present invention. All modifications, substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope summarized by the appended claims.

Claims

1. A method for optimizing ship cabin environment based on crew status, characterized in that, Includes the following steps: Step S1: Collect personnel activity data, physiological data, and environmental data in real time through a multi-source sensor network deployed in the ship's compartments; Step S2: Construct an environmental status assessment index system based on the improved analytic hierarchy process (AHP) and entropy weight method, and calculate the weight of each index. Step S3: Using the grey-fuzzy comprehensive evaluation method, the collected data is used as input to obtain the status level of the cabin at different time periods; Step S4: Input the state level into the multi-agent simulation model, perform behavioral simulation and optimization, and generate an optimized scheme for cabin equipment configuration and space functional zoning. Step S5: Visualize the status level and optimization scheme through the digital twin platform, and display the distribution of the cabin environment status in the form of a heat map to provide decision support.

2. The method as described in claim 1, characterized in that, In step S1, the multi-source sensor network includes sonar beacons, infrared sensors, differentially privacy-processed in-cabin cameras, and cabin environment sensors; the personnel activity data includes personnel location, movement trajectory, dwell time, interaction frequency, and task type diversity; the environmental data includes temperature, humidity, and other data. concentration, Concentration, pressure, and noise levels.

3. The method as described in claim 1 or 2, characterized in that, In step S1, the movement trajectory is extracted by a sensor data fusion algorithm. The algorithm combines Kalman filtering and particle filtering to fuse sonar signal strength, infrared signal strength and visual data to estimate the continuous position coordinates of an individual, and identifies common movement paths through DBSCAN cluster analysis.

4. The method as described in claim 1 or 2, characterized in that, In step S1, differential privacy processing is also used to anonymize the data collected by the camera by adding Laplacian noise, and the error range of the processed data is controlled within 5%.

5. The method as described in claim 1, characterized in that, In step S2, the environmental status assessment index system takes the cabin environmental status as the target layer and space utilization efficiency, personnel task efficiency, equipment utilization rate, environmental comfort, environmental distribution characteristics and time period fluctuation characteristics as the criteria layer. Each criterion layer has multiple index layers.

6. The method as described in claim 1 or 5, characterized in that, In step S2, the improved AHP-entropy weight method includes: constructing a judgment matrix through expert survey, calculating the initial weights using the eigenvalue method, then introducing the entropy weight method to correct the weights, and ensuring rationality through a consistency test; the consistency test is performed by calculating the consistency index CI and the consistency ratio CR, and if CR>0.1, the judgment matrix is ​​readjusted.

7. The method as described in claim 1, characterized in that, In step S3, the gray-fuzzy comprehensive evaluation method includes: calculating the evaluation vector of each index through a composite fuzzy membership function, and performing fuzzy synthesis by combining gray relational analysis to obtain the state level.

8. The method as described in claim 1, characterized in that, In step S4, the multi-agent simulation model is based on a deep reinforcement learning algorithm and uses a deep Q-network (DQN) to update the agent's behavior policy. The DQN is trained through experience replay and a target network to simulate personnel movement and interaction, with the optimization objective being to maximize the cabin environment state level.

9. The method as described in claim 1, characterized in that, In step S5, the digital twin platform integrates the 3D model of the cabin and real-time data stream. The heat map is generated by the inverse distance weight interpolation algorithm and supports interactive scene simulation.

10. A ship cabin environment optimization system based on crew status, comprising: The multi-source compartment sensing module is used to collect real-time data on personnel activities, physiological data, and environmental data through a multi-source sensor network deployed in the ship's compartments; The module for constructing a dynamic environmental status assessment index system is used to construct an environmental status assessment index system based on the improved AHP-entropy weight method. The system takes the cabin environmental status as the target layer and space utilization efficiency, personnel task efficiency, equipment utilization rate, environmental comfort, environmental distribution characteristics, and time-period fluctuation characteristics as the criterion layers. Each criterion layer has multiple index layers, and the weights are calculated using the AHP-entropy weight method. The grey-fuzzy comprehensive evaluation module is used to obtain the state level based on the environmental state assessment index system and the grey-fuzzy comprehensive evaluation method. It takes human activity data, physiological data and environmental data as input and obtains the state level through composite fuzzy membership function and grey relational analysis. The space and facility optimization module is used to input the state level into the multi-agent simulation model, optimize the configuration of cabin equipment and space function zoning based on deep reinforcement learning algorithms, and generate optimization schemes. The visualization decision support module is used to visualize the status level and optimization scheme through the digital twin platform, display the distribution of the cabin environment status in the form of a heat map, and provide interactive decision support. The multi-source cabin perception module includes a multi-source sensor network comprising a sonar beacon, an infrared sensor, an in-cabin camera with differential privacy processing, and a cabin environment sensor.

Images