Ship safety management method and system based on visual identification technology

Abstract

The invention discloses a ship safety management method and system based on a visual identification technology, and relates to the technical field of ship management, the system is in communication connection with a perception interaction layer, an intelligent decision-making layer and an execution supervision layer in sequence, and the perception interaction layer deploys a high-definition camera in a ship key area; human face and environment data are collected in real time and are in butt joint with a sailor database; the intelligent decision-making layer calculates a feature hash value by using a convolutional neural network and a hash algorithm based on the management request to determine a security level; the execution supervision layer executes a corresponding supervision strategy according to the security level, and the system also comprises a data management layer which is responsible for storing sailor information, the number of ship approved passengers and historical supervision data, and carrying out data interaction with the AIS ship trajectory system and the electronic chart system; according to the method, through data acquisition, request receiving, hash value calculation, safety level determination and strategy execution, accurate dynamic supervision of ship safety is realized, and the efficiency and accuracy of ship safety supervision are effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of ship management, and more particularly to a ship safety management method and system based on visual recognition technology. Background Art

[0002] In the shipping industry, the Maritime Safety Administration needs to manage ships, but existing technologies are inefficient and have loopholes in ship safety management. Traditional management relies on manual inspections and spot checks, which are inefficient and insufficiently covered, making it difficult to achieve full-area supervision and prone to safety hazards. Especially during non-working hours and in remote waters, supervision becomes even more difficult. Coastal fixed-operation ships are limited in their crew's range of activities, and traditional supervision methods are difficult to meet their dynamic supervision needs. In addition, the shipping market is highly competitive, and ship operating costs are rising. Some ships are understaffed to reduce costs, and existing management methods have limited monitoring and early warning capabilities for such problems and cannot prevent maritime safety accidents from the source.

[0003] To meet these challenges, a ship safety management system based on visual recognition technology has emerged. The system integrates artificial intelligence, big data analysis and facial recognition technology, aiming to achieve real-time monitoring and intelligent assessment of key areas of the ship. By deploying high-definition cameras in key areas of the ship, the system can collect facial data and ship environmental data in real time, and synchronize them with the crew database. The intelligent decision-making layer calculates the feature hash value and determines the target safety level based on management requests by combining convolutional neural networks with hash algorithms. The executive supervision layer executes corresponding supervision strategies according to the target safety level to achieve precise regulation of the ship's safety status.

[0004] Therefore, the existing ship safety management has the problems of low efficiency, insufficient coverage and safety hazards. The intelligent ship safety management system realizes precise dynamic supervision, effectively improves management efficiency and reduces accident risks. Summary of the Invention

[0005] In order to overcome the problems of low efficiency, insufficient coverage and safety hazards in existing ship safety management, the present invention discloses a ship safety management method and system based on visual recognition technology, which can effectively solve the above technical problems.

[0006] In order to solve the above technical problems, the technical solutions of the present invention are as follows:

[0007] A ship safety management system based on visual recognition technology includes a perception interaction layer, an intelligent decision-making layer, and an execution and supervision layer that are sequentially connected in communication;

[0008] The perception interaction layer is used to deploy high-definition cameras in the ship's bridge, engine room, and passenger cabin areas to collect real-time facial data and ship environment data, and synchronize with the crew database, which contains crew certification information;

[0009] The intelligent decision layer is configured to determine, based on the management request, a target security feature chain from a plurality of built-in security feature chain rings, calculate a feature hash value of the management request, and determine a target security level corresponding to the feature hash value from the target security feature chain ring;

[0010] The execution supervision layer is configured to respond to the target security level and execute the supervision policy mapped to the target security level;

[0011] The security feature chain is composed of multiple feature hash values; the execution supervision layer includes multiple supervision policy agents, each of which is associated with multiple policy execution addresses; the supervision policy agent is used to determine the supervision policy mapped to the target security level from the multiple policy execution addresses; and in response to the target security level, execute the supervision policy to adjust the ship's safety status;

[0012] It also includes a data management layer; the data management layer includes a ship safety database; the ship safety database is used to store crew information, the ship's approved passenger capacity and historical supervision data, and to interact with the AIS ship tracking system and the electronic chart system.

[0013] The perception interaction layer is also used to classify the collected facial data and ship environment data according to preset security classification rules, generate corresponding security description files and cache them; the preset security classification rules include personnel qualification rules, ship manning rules and navigation safety rules.

[0014] Preferably, the intelligent decision-making layer includes a global security decision chain ring; the global security decision chain ring is composed of a security risk analysis engine and a plurality of local security decision chain ring agents connected; the local security decision chain ring agents are associated one by one with the local security decision chain rings;

[0015] The security risk analysis engine is used to respond to an input management request and forward the management request to the multiple local security decision chain ring agents; the local security decision chain ring agent is used to send the management request to the associated local security decision chain ring.

[0016] Preferably, the intelligent decision layer is specifically configured to: determine a target local security feature chain ring from a plurality of built-in local security feature chain rings based on the security category carried in the management request;

[0017] A convolutional neural network and a hash algorithm are used to calculate the characteristic hash value corresponding to the management request, and it is determined whether the target local security characteristic chain link has a hash value identical to the characteristic hash value; if so, the target security level is determined to be the hash value identical to the characteristic hash value; if not, the target security level is determined to be the hash value with the smallest difference from the characteristic hash value;

[0018] The management request includes one or more of a crew identity verification request, a ship manning verification request, and a navigation safety supervision request.

[0019] Preferably, the intelligent decision layer is further configured to construct a plurality of updated local security feature chain rings using the updated feature hash values ​​respectively calculated for the plurality of management requests, and determine the updated local security feature chain ring agent corresponding to each of the updated local security feature chain rings;

[0020] The data management layer is also used to construct an updated global security feature chain ring using all the updated local security feature chain ring agents.

[0021] Preferably, the perception interaction layer is further configured to: classify the plurality of management requests according to the personnel qualification rules, the ship manning rules or the navigation safety rules, and generate a security description file corresponding to each of the management requests;

[0022] The management request associated with the security description file is cached.

[0023] Preferably, the intelligent decision layer is further configured to: calculate the update feature hash value corresponding to each of the management requests; classify the multiple update feature hash values ​​into multiple security clusters according to the security features contained in each of the security description files;

[0024] Using the updated feature hash value in each of the security clusters, respectively construct an updated local security feature chain ring; wherein each of the updated feature hash values ​​is mapped to an updated supervision policy in the execution supervision layer;

[0025] Determine, from each of the update local security feature chain rings, an update supervision policy associated with an update feature hash value having a small-world network characteristic as an update local security feature chain ring agent;

[0026] The small-world network feature is an update feature hash value that appears the most times in the multiple update local security feature chain rings.

[0027] Preferably, the data management layer is further configured to: select, according to the feature hash value of the updated local security feature chain ring agent, a target agent corresponding to the feature hash value having the small-world network feature as the first updated security risk analysis engine;

[0028] All of the updated local security feature chain ring agents and the first updated security risk analysis engine are used to construct an updated global security feature chain ring; or all of the updated local security feature chain ring agents and the preset second updated security risk analysis engine are used to construct an updated global security feature chain ring.

[0029] Preferably, the intelligent decision-making layer is also used to: when the management request carries new security supervision requirements, calculate the new feature hash value corresponding to the new security supervision requirements; store the new security supervision requirements in the new supervision policy mapped to the new feature hash value in the execution supervision layer.

[0030] Preferably, a ship safety management method based on visual recognition technology comprises:

[0031] High-definition cameras deployed in key areas of the ship collect facial data and ship environment data to the perception interaction layer; the perception interaction layer receives input management requests; and the perception interaction layer broadcasts the management requests to the intelligent decision-making layer;

[0032] Determining, by the intelligent decision layer based on the management request, a target security feature chain ring from a plurality of built-in security feature chain rings; the security feature chain ring is composed of a plurality of feature hash values;

[0033] Calculate the characteristic hash value of the management request by the intelligent decision layer;

[0034] Determine the target security level corresponding to the feature hash value from the target security feature chain by the intelligent decision layer;

[0035] executing a supervisory policy mapped to the target security level by executing a supervisory layer in response to the target security level;

[0036] The step of executing the regulatory policy mapped to the target security level in response to the target security level by the execution regulatory layer includes: determining, by the regulatory policy agent, the regulatory policy mapped to the target security level from the plurality of policy execution addresses;

[0037] In response to the target safety level, executing the supervision strategy to adjust the safety status of the ship;

[0038] It also includes classifying multiple management requests according to preset security classification rules through the perception interaction layer, generating corresponding security description files and caching them; the preset security classification rules include personnel qualification rules, ship manning rules and navigation safety rules.

[0039] A computer-readable storage medium stores a computer program, wherein the computer program implements the security management method described above when executed by a processor.

[0040] Compared with the prior art, the present invention has the following beneficial effects: it solves the problems of low efficiency, insufficient coverage, and potential safety hazards in existing ship safety management. The system includes a perception interaction layer, an intelligent decision-making layer, an execution and supervision layer, and a data management layer. The perception interaction layer deploys high-definition cameras in key areas of the ship to collect real-time facial data and ship environment data, and simultaneously connects to the crew database to achieve real-time monitoring of crew identity and ship environment. The intelligent decision-making layer calculates feature hash values ​​based on management requests using a combination of convolutional neural networks and hash algorithms to quickly and accurately determine the target safety level. The execution and supervision layer implements corresponding supervision policies based on the target safety level and promptly adjusts the ship's safety status. The data management layer is responsible for storing crew information, the ship's approved passenger capacity, and historical supervision data, and interacts with the AIS ship tracking system and electronic nautical chart system to achieve data sharing and integration. The system realizes accurate dynamic monitoring and intelligent assessment of ship safety, improving the efficiency and accuracy of ship safety supervision. In addition, the present invention uses a multi-source data linkage mechanism to connect multiple system data such as the crew database, AIS ship tracking, and electronic nautical charts to construct a three-dimensional ship-personnel-behavior supervision model, which can accurately identify evasive supervision behaviors and effectively prevent the occurrence of safety accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the implementation methods of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the implementation methods or the description of the prior art. Obviously, the drawings described below are merely exemplary. For ordinary technicians in this field, other implementation drawings can be derived based on the provided drawings without any creative work.

[0042] Figure 1 This is a system structure diagram of the present invention;

[0043] Figure 2 This is a flow chart of the safety management method of the present invention;

[0044] Figure 3 This is a flow chart for updating the safety chain of the present invention. DETAILED DESCRIPTION

[0045] The accompanying drawings are for illustrative purposes only and are not to be construed as limiting this patent;

[0046] In order to better illustrate this embodiment, some parts in the drawings may be omitted, enlarged, or reduced, and do not represent the actual product size;

[0047] It is understandable to those skilled in the art that some well-known structures and descriptions thereof may be omitted in the drawings.

[0048] The technical solution of the present invention is further described below with reference to the accompanying drawings and embodiments.

[0049] Example 1

[0050] A ship safety management system based on visual recognition technology includes a perception interaction layer, an intelligent decision-making layer, and an execution and supervision layer that are sequentially connected in communication;

[0051] The perception interaction layer is used to deploy high-definition cameras in the ship's bridge, engine room, and passenger cabin areas to collect real-time facial data and ship environment data, and synchronize with the crew database, which contains crew certification information;

[0052] The intelligent decision layer is configured to determine, based on the management request, a target security feature chain from a plurality of built-in security feature chain rings, calculate a feature hash value of the management request, and determine a target security level corresponding to the feature hash value from the target security feature chain ring;

[0053] The execution supervision layer is configured to respond to the target security level and execute the supervision policy mapped to the target security level;

[0054] The security feature chain is composed of multiple feature hash values; the execution supervision layer includes multiple supervision policy agents, each of which is associated with multiple policy execution addresses; the supervision policy agent is used to determine the supervision policy mapped to the target security level from the multiple policy execution addresses; and in response to the target security level, execute the supervision policy to adjust the ship's safety status;

[0055] It also includes a data management layer; the data management layer includes a ship safety database; the ship safety database is used to store crew information, the ship's approved passenger capacity and historical supervision data, and to interact with the AIS ship tracking system and the electronic chart system.

[0056] The perception interaction layer is also used to classify the collected facial data and ship environment data according to preset security classification rules, generate corresponding security description files and cache them; the preset security classification rules include personnel qualification rules, ship manning rules and navigation safety rules.

[0057] The intelligent decision-making layer includes a global security decision chain; the global security decision chain is composed of a security risk analysis engine and a plurality of local security decision chain agents; the local security decision chain agents are associated with the local security decision chains one by one;

[0058] The security risk analysis engine is used to respond to an input management request and forward the management request to the multiple local security decision chain ring agents; the local security decision chain ring agent is used to send the management request to the associated local security decision chain ring.

[0059] The intelligent decision layer is specifically configured to: determine a target local security feature chain ring from a plurality of built-in local security feature chain rings based on the security category carried by the management request;

[0060] A convolutional neural network and a hash algorithm are used to calculate the characteristic hash value corresponding to the management request, and it is determined whether the target local security characteristic chain link has a hash value identical to the characteristic hash value; if so, the target security level is determined to be the hash value identical to the characteristic hash value; if not, the target security level is determined to be the hash value with the smallest difference from the characteristic hash value;

[0061] The management request includes one or more of a crew identity verification request, a ship manning verification request, and a navigation safety supervision request.

[0062] The intelligent decision layer is further configured to construct multiple updated local security feature chain rings using the updated feature hash values ​​respectively calculated for the multiple management requests, and determine the updated local security feature chain ring agent corresponding to each of the updated local security feature chain rings;

[0063] The data management layer is also used to construct an updated global security feature chain ring using all the updated local security feature chain ring agents.

[0064] The perception interaction layer is further configured to: classify the plurality of management requests according to the personnel qualification rules, the ship manning rules, or the navigation safety rules, and generate a security description file corresponding to each of the management requests;

[0065] The management request associated with the security description file is cached.

[0066] The intelligent decision layer is further configured to: calculate the update feature hash value corresponding to each of the management requests; classify the multiple update feature hash values ​​into multiple security clusters according to the security features contained in each of the security description files;

[0067] Using the updated feature hash value in each of the security clusters, respectively construct an updated local security feature chain ring; wherein each of the updated feature hash values ​​is mapped to an updated supervision policy in the execution supervision layer;

[0068] Determine, from each of the update local security feature chain rings, an update supervision policy associated with an update feature hash value having a small-world network characteristic as an update local security feature chain ring agent;

[0069] The small-world network feature is an update feature hash value that appears the most times in the multiple update local security feature chain rings.

[0070] The data management layer is further configured to: select, according to the feature hash value of the updated local security feature chain ring agent, a target agent corresponding to the feature hash value having the small-world network feature as a first updated security risk analysis engine;

[0071] All of the updated local security feature chain ring agents and the first updated security risk analysis engine are used to construct an updated global security feature chain ring; or all of the updated local security feature chain ring agents and the preset second updated security risk analysis engine are used to construct an updated global security feature chain ring.

[0072] The intelligent decision-making layer is also used to: when the management request carries a new security supervision requirement, calculate the new feature hash value corresponding to the new security supervision requirement; store the new security supervision requirement in the new supervision policy mapped to the new feature hash value in the execution supervision layer.

[0073] See also Figure 1-3 High-definition cameras are deployed in key areas such as the ship's bridge, engine room and passenger cabin. The cameras on the bridge are used to collect real-time facial images and operational behavior data of the crew, such as the operation of the ship's navigation equipment and communication equipment. The cabin cameras are mainly used to monitor the crew's operation and inspection of the ship's power equipment and electrical equipment, and at the same time collect environmental data on the equipment's operating status, such as temperature and humidity. The cameras in the passenger cabin are used to collect passengers' facial data, passenger activities, and luggage placement. These high-definition cameras have high resolution, wide dynamic range, and low-light sensing capabilities, which can ensure clear image data under various lighting conditions in different areas of the ship, such as the strong light environment of the bridge and the dim environment of the engine room, and transmit video streams to the storage device of the perception interaction layer in real time at a frequency of 30 frames per second.

[0074] The perception interaction layer maintains real-time synchronization with the crew database through a secure data interface. The crew database stores detailed information of each crew member, including name, ID number, type and validity period of crew competency certificate, professional skills training records, etc.

[0075] When the camera collects facial data, the system immediately compares this data with the crew database. For example, on the bridge, when the camera captures the facial image of a crew member operating navigation equipment, it uses a facial recognition algorithm to match it with the crew member's facial information in the database, quickly determining the crew member's identity and further verifying whether he or she has the qualification certificate to operate the corresponding equipment, such as a senior ship driver certificate.

[0076] The perception interaction layer classifies and processes the collected data according to preset safety classification rules. The preset safety classification rules include personnel qualification rules, ship manning rules and navigation safety rules.

[0077] Personnel qualification rules are used to determine whether crew members are qualified to work in specific positions. For example, based on the operating behavior and facial recognition results of the crew members on the bridge, the system verifies whether they hold a valid seafarer certificate and whether the level of the certificate matches the equipment they are operating.

[0078] The ship manning rules are used to check whether the number of crew members in each position on the ship meets the minimum safety manning standards. The system generates a ship manning situation report by counting the number of crew members in each area of ​​the ship and their job qualifications.

[0079] Navigation safety rules are used to assess whether a ship's navigation status is safe. For example, by analyzing the distribution of people in the passenger cabin, the placement of luggage, and the ship's navigation speed, route and other data, it is determined whether there are safety hazards such as overcrowding, cargo blocking the passage, or excessive navigation speed.

[0080] Based on the classification results, corresponding safety description files are generated. These files record detailed information about the data classification in a structured data format, including classification categories, such as compliance of personnel qualifications, adequacy of ship manning or navigation safety risks, data collection time, collection location, and identity information of relevant crew members or passengers. These files are temporarily stored in the local cache of the perception interaction layer for subsequent analysis and processing by the intelligent decision-making layer.

[0081] The intelligent decision-making layer has multiple built-in security feature chains. These security feature chains are constructed based on historical data, regulatory requirements and expert experience of ship safety management. Each chain contains multiple feature hash values. These feature hash values ​​represent specific security features or risk patterns. For example, for crew identity verification requests, the intelligent decision-making layer locates the local security feature chain related to personnel qualification verification from many local security feature chains. This chain contains feature hash values ​​of various crew qualification compliance situations, such as valid certificates, expired certificates, unlicensed operations and other feature representations of different situations.

[0082] When the perception interaction layer sends a crew identity verification request and related data, the intelligent decision-making layer first uses a convolutional neural network to extract features from the collected crew facial images and operational behavior data. After being trained on a large amount of labeled data, the convolutional neural network can automatically learn key features of facial images and operational behaviors, such as key point features of the face and sequence features of operational actions. It then uses a hash algorithm to convert these features into a unique feature hash value. The system then searches the target local security feature chain to see if there is a hash value identical to the feature hash value. If so, it directly determines that the hash value identical to the feature hash value is the feature corresponding to the target security level. If the feature hash value is identical to the valid certificate operation feature hash value in the chain, it indicates that the crew qualification is compliant and the corresponding security level is safe. If there is no identical hash value, the system finds the hash value with the smallest difference as the corresponding feature for the target security level. For example, if the calculated feature hash value has the smallest difference with the certificate expiration feature hash value, it is determined that there is a certain risk in the crew qualification and the corresponding security level is low risk, indicating that the validity period of the crew certificate needs to be paid attention to.

[0083] The safety risk analysis engine in the global safety decision chain receives various input management requests, such as ship manning verification requests and navigation safety supervision requests. When receiving a ship manning verification request, the safety risk analysis engine forwards the request to multiple local safety decision chain agents based on the type and urgency of the request. For example, emergency verification requests for ships about to sail have a higher priority.

[0084] Each local safety decision chain ring agent is associated with a specific local safety decision chain ring. For example, there is a local safety decision chain ring agent specifically responsible for processing the decision chain ring related to ship manning. After receiving the request, the agent sends it to the associated local safety decision chain ring, and uses the feature hash value in the chain ring to match and analyze the data features related to the request to determine the safety level of the ship's manning situation.

[0085] The execution supervision layer contains multiple supervision policy agents, each of which is associated with multiple policy execution addresses. These policy execution addresses can be various execution devices or systems on the ship, such as alarm devices, ship operation authority control systems, communication equipment, etc. For example, in the case of insufficient ship manning, there is a supervision policy agent specifically responsible for handling such safety risks. The policy execution addresses associated with this agent include the ship's sound and light alarms, the bridge operation authority restriction system, and the communication interface for sending alarm information to the port maritime management department.

[0086] After the intelligent decision-making layer determines the target safety level, the supervisory layer responds and executes the corresponding supervisory strategy. For example, if a ship is judged to be undermanned and at high risk, the supervisory strategy agent extracts the corresponding supervisory strategy from the associated strategy execution address.

[0087] The system immediately triggers the sound and light alarm to alert the crew and ship management personnel of the insufficient manning situation. At the same time, the supervisory authorities restrict the ship's departure operation authority through the ship operation authority restriction system to ensure that the ship cannot sail until the manning problem is resolved.

[0088] In addition, the system sends detailed alarm information to the port maritime management department through the communication interface, including the ship's name, location, specific circumstances of the missing crew, and screenshots of relevant evidence, such as images of an unmanned bridge and images of insufficient personnel in key engine room positions, so that the management department can take further measures in a timely manner, such as dispatching law enforcement personnel to board the ship for inspection or requiring the ship operator to immediately dispatch qualified crew members to board the ship.

[0089] The ship safety database is a core component of the data management layer. It stores detailed information about crew members, including personal information, certificate updates, the ship's approved passenger capacity, and historical regulatory data, such as past manning inspection results, violation records, and handling results.

[0090] The database also maintains real-time data interaction with the AIS ship tracking system and electronic nautical chart system. The AIS system can obtain the ship's real-time position information, speed, heading and other trajectory data. Combined with the electronic nautical chart system, it can accurately determine whether the ship is sailing within the specified operating area and channel, providing important location and channel information for navigation safety supervision. For example, when a ship enters a specific traffic control area, the data management layer can obtain the ship's position information in a timely manner and compare it with the navigation regulations of the area, providing accurate data support for the ship's navigation status to the intelligent decision-making layer.

[0091] After the intelligent decision-making layer uses the new management request to calculate the updated feature hash value, the data management layer constructs an updated local safety feature chain and an updated global safety feature chain based on it. For example, when a ship completes a successful manning adjustment, the system incorporates the new manning data and the corresponding feature hash value into the safety feature chain system. The data management layer constructs a new global safety feature chain based on these updated local safety feature chain agents, optimizes the algorithm model, and improves the system's recognition accuracy and adaptability to future ship manning situations.

[0092] Specific implementation: One day, a harbor vessel was about to carry out a material transportation mission within the port. Before the ship started, the high-definition camera in the perception interaction layer began to work, collecting facial data of the crew on the bridge and image information of passengers in the passenger cabin, and synchronizing this data with the crew database, while generating a corresponding security description file.

[0093] The system broadcasts requests for ship manning verification and navigation safety supervision to the intelligent decision-making layer. Based on the request category, the intelligent decision-making layer locates the corresponding manning verification and navigation safety feature chain from the local safety feature chain, calculates the feature hash value and matches it with the hash value in the chain.

[0094] After analysis, it was found that the qualifications of the bridge crew met the requirements, but the number of passengers in the passenger cabin was close to the approved passenger capacity limit, and there was a certain risk of overcrowding. The system determined that the target safety level was medium risk.

[0095] After the executive supervisory layer receives the safety level, the supervisory policy agent extracts the corresponding supervisory policy from the policy execution address, such as issuing an alarm to remind the crew to pay attention to the safety of passenger distribution, restricting the ship from adding additional passengers, and pushing the relevant situation to the port maritime command center. At the same time, the data management layer updates the verification and supervision data to the ship safety database, providing experience reference for subsequent supervision of similar scenarios and optimizing the system's algorithm model for monitoring passenger cabin personnel density.

[0096] Example 2

[0097] A ship safety management method based on visual recognition technology, the method comprising:

[0098] High-definition cameras deployed in key areas of the ship collect facial data and ship environment data to the perception interaction layer; the perception interaction layer receives input management requests; and the perception interaction layer broadcasts the management requests to the intelligent decision-making layer;

[0099] Determining, by the intelligent decision layer based on the management request, a target security feature chain ring from a plurality of built-in security feature chain rings; the security feature chain ring is composed of a plurality of feature hash values;

[0100] Calculate the characteristic hash value of the management request by the intelligent decision layer;

[0101] Determine the target security level corresponding to the feature hash value from the target security feature chain by the intelligent decision layer;

[0102] executing a supervisory policy mapped to the target security level by executing a supervisory layer in response to the target security level;

[0103] The step of executing the regulatory policy mapped to the target security level in response to the target security level by the execution regulatory layer includes: determining, by the regulatory policy agent, the regulatory policy mapped to the target security level from the plurality of policy execution addresses;

[0104] In response to the target safety level, executing the supervision strategy to adjust the safety status of the ship;

[0105] It also includes classifying multiple management requests according to preset security classification rules through the perception interaction layer, generating corresponding security description files and caching them; the preset security classification rules include personnel qualification rules, ship manning rules and navigation safety rules.

[0106] A computer-readable storage medium stores a computer program, wherein the computer program implements the security management method described above when executed by a processor.

[0107] See also Figure 1-3 High-definition cameras deployed in key areas of the ship, such as the bridge, engine room, and passenger cabin, work continuously, collecting real-time facial data (including crew members and passengers) and ship environment data, such as equipment operating status, cargo stacking, and personnel activity trajectories. For example, in the engine room, the camera can capture the crew's operation of various types of machinery and determine whether there are any illegal operations; in the passenger cabin, it can monitor whether passengers are wearing life-saving equipment correctly.

[0108] The camera on the bridge collects facial images and operational behavior data of the crew every 5 seconds, and transmits this data in the form of a video stream to the storage device of the perception interaction layer. At the same time, the camera also collects the ship's navigation status data, such as speed and heading, which are transmitted to the perception interaction layer through the ship's data bus.

[0109] The perception interaction layer receives management requests from ship staff, such as the captain, watchkeeper, or the ship's automatic detection system. For example, the captain enters a crew identity verification request through the ship management system interface. The request includes the personal information of the new crew member (name, ID number, etc.) and the area of ​​the ship where they are located, such as the bridge.

[0110] When the ship's automatic detection system finds that the passenger cabin's occupant density is close to the upper limit of the approved passenger capacity, it automatically sends a navigation safety supervision request to the perception interaction layer. The request contains information such as the current number of people in the passenger cabin, the approved passenger capacity, and the current location of the ship.

[0111] The perception interaction layer classifies the collected data and management requests according to preset security classification rules (personnel qualification rules, ship manning rules, and navigation safety rules). For crew identity verification requests, it uses facial recognition technology to compare with the crew database in accordance with personnel qualification rules to determine whether the crew has the corresponding qualification certificate.

[0112] The system compares the facial image of the new crew member with the image in the crew database and verifies their certificate information. If the crew member holds a valid crew competency certificate and the certificate level matches the requirements of the bridge position, their qualifications are deemed to be compliant; otherwise, it is determined that there is a qualification risk.

[0113] In response to navigation safety supervision requests, the system analyzes the distribution of people in the passenger cabin based on navigation safety rules. If the distribution of people is uneven, it may affect the stability of the ship, and the system will determine that there is a navigation safety risk.

[0114] Based on the classification results, a detailed security description file is generated to record information such as the management request category, data characteristics, preliminary analysis results, etc., and is temporarily stored in the perception interaction layer cache. For example, the security description file corresponding to the crew identity verification request records the facial feature data of the new crew member, the matching results with the database (certificate type, validity period, etc.), the ship area information and the current timestamp, etc. The security description file corresponding to the navigation safety supervision request records the current number of people in the passenger cabin, the approved passenger capacity, the distribution of people, the ship location and the possible safety risk level, etc.

[0115] The perception interaction layer broadcasts the management request to the intelligent decision-making layer. The intelligent decision-making layer determines the target security feature chain from multiple built-in security feature chain links based on the security category in the management request.

[0116] For crew identity verification requests, the intelligent decision-making layer locates the local security feature chain link related to personnel qualifications from the local security feature chain link, and uses a combination of convolutional neural network and hash algorithm to calculate the feature hash value of the management request.

[0117] The convolutional neural network performs feature extraction on the collected crew facial images and certificate information, extracting key facial feature points and certificate information features, such as certificate number, issuing agency, etc., and then the hash algorithm converts these features into a unique feature hash value.

[0118] The hash value that is identical to or has the smallest difference with the feature hash value is retrieved in the target security feature chain to determine the corresponding target security level. Assuming that the feature hash value of a new crew member is exactly the same as a hash value in the chain, it indicates that the crew member's qualifications fully meet the requirements, and the target security level is determined to be safe.

[0119] If the characteristic hash value calculated by another ship's navigation safety supervision request is very different from a hash value in the chain link, it indicates that there is a slight uneven distribution of people in the passenger cabin, and the target safety level is determined to be low risk, prompting the crew to guide the passengers to redistribute to ensure the safety of the ship's navigation.

[0120] The executive supervisor responds to the target safety level, and the supervisory policy agent determines the corresponding supervisory policy from multiple policy execution addresses. For safe crew identity verification results, the supervisory policy may include allowing the crew to operate on the bridge and recording their boarding operation compliance information in the ship management system; for low-risk navigation safety issues, the supervisory policy includes issuing voice warnings to remind the crew to guide passengers to redistribute, displaying warning information on the ship's control panel, and recording relevant situations in the ship's navigation log.

[0121] Execute regulatory strategies to adjust the ship's safety status. For example, after the system issues a warning for passenger redistribution, the crew promptly guides passengers to adjust their positions, restoring the ship to a safe navigation state and reducing safety risks.

[0122] The data management layer updates the ship safety database in real time, storing information such as collected data, management request processing results, and execution supervision status. It also interacts with the AIS ship tracking system and electronic nautical chart system to obtain data such as the latest position of the ship and the surrounding sea environment, enriching data resources. For example, when a ship completes a navigation mission, the data management layer updates all safety inspection data, supervision execution status, and ship trajectory data during the navigation process to the ship safety database, providing data support for subsequent data analysis and system optimization.

[0123] The intelligent decision-making layer uses the new feature hash value to build and update the local safety feature chain, and the data management layer builds and updates the global safety feature chain based on this, optimizing the system's ability to identify and judge ship safety features and improving the accuracy of subsequent supervision.

[0124] When a management request carries new safety supervision requirements, such as supervision of the operation of ship anti-pollution equipment, the intelligent decision-making layer calculates the new feature hash value corresponding to the new requirement, and stores the new supervision requirement in the new supervision policy mapped to the new feature hash value in the execution supervision layer; the execution supervision layer monitors and manages the operation of ship anti-pollution equipment based on the new supervision policy, such as regularly checking the equipment operating status, recording emission data, etc., to achieve effective control of new safety risks.

[0125] The same or similar reference numerals correspond to the same or similar components;

[0126] The terms used in the drawings to describe positional relationships are for illustrative purposes only and should not be construed as limiting this patent;

[0127] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not limitations on the implementation methods of the present invention. For ordinary technicians in the field, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to enumerate all the implementation methods here. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the claims of the present invention.

Claims

1. A ship safety management system based on visual recognition technology, characterized in that: It includes a perception interaction layer, an intelligent decision-making layer, and an execution and supervision layer that are sequentially connected; The perception interaction layer is used to deploy high-definition cameras in the ship's bridge, engine room, and passenger cabin areas to collect real-time facial data and ship environment data, and synchronize with the crew database, which contains crew certification information; The intelligent decision layer is configured to determine, based on the management request, a target security feature chain from a plurality of built-in security feature chain rings, calculate a feature hash value of the management request, and determine a target security level corresponding to the feature hash value from the target security feature chain ring; The execution supervision layer is configured to respond to the target security level and execute the supervision policy mapped to the target security level; The security feature chain is composed of multiple feature hash values; the execution supervision layer includes multiple supervision policy agents, each of which is associated with multiple policy execution addresses; the supervision policy agent is used to determine the supervision policy mapped to the target security level from the multiple policy execution addresses; and in response to the target security level, execute the supervision policy to adjust the ship's safety status; It also includes a data management layer; the data management layer includes a ship safety database; the ship safety database is used to store crew information, the ship's approved passenger capacity and historical supervision data, and to interact with the AIS ship tracking system and the electronic chart system. The perception interaction layer is also used to classify the collected facial data and ship environment data according to preset security classification rules, generate corresponding security description files and cache them; the preset security classification rules include personnel qualification rules, ship manning rules and navigation safety rules.

2. The ship safety management system based on visual recognition technology according to claim 1 is characterized in that: The intelligent decision-making layer includes a global security decision-making chain; The global security decision chain is composed of a security risk analysis engine and a plurality of local security decision chain agents; the local security decision chain agents are associated with the local security decision chains one by one; The security risk analysis engine is configured to respond to an input management request and forward the management request to the plurality of local security decision chain agents; The local security decision chain ring agent is used to send the management request to the associated local security decision chain ring.

3. The ship safety management system based on visual recognition technology according to claim 1 is characterized in that: The intelligent decision layer is specifically configured to: determine a target local security feature chain ring from a plurality of built-in local security feature chain rings based on the security category carried by the management request; A convolutional neural network and a hash algorithm are used to calculate the characteristic hash value corresponding to the management request, and it is determined whether the target local security characteristic chain link has a hash value identical to the characteristic hash value; if so, the target security level is determined to be the hash value identical to the characteristic hash value; if not, the target security level is determined to be the hash value with the smallest difference from the characteristic hash value; The management request includes one or more of a crew identity verification request, a ship manning verification request, and a navigation safety supervision request.

4. The ship safety management system based on visual recognition technology according to claim 1 is characterized in that: The intelligent decision layer is further configured to construct multiple updated local security feature chain rings using the updated feature hash values ​​respectively calculated for the multiple management requests, and determine the updated local security feature chain ring agent corresponding to each of the updated local security feature chain rings; The data management layer is also used to construct an updated global security feature chain ring using all the updated local security feature chain ring agents.

5. The ship safety management system based on visual recognition technology according to claim 4 is characterized in that: The perception interaction layer is further configured to: classify the plurality of management requests according to the personnel qualification rules, the ship manning rules, or the navigation safety rules, and generate a security description file corresponding to each of the management requests; The management request associated with the security description file is cached.

6. The ship safety management system based on visual recognition technology according to claim 5 is characterized in that: The intelligent decision layer is further configured to: calculate the update feature hash value corresponding to each of the management requests; classify the multiple update feature hash values ​​into multiple security clusters according to the security features contained in each of the security description files; Using the updated feature hash value in each of the security clusters, respectively construct an updated local security feature chain ring; wherein each of the updated feature hash values ​​is mapped to an updated supervision policy in the execution supervision layer; Determine, from each of the update local security feature chain rings, an update supervision policy associated with an update feature hash value having a small-world network characteristic as an update local security feature chain ring agent; The small-world network feature is an update feature hash value that appears the most times in the multiple update local security feature chain rings.

7. The ship safety management system based on visual recognition technology according to claim 6 is characterized in that: The data management layer is further configured to: select, according to the feature hash value of the updated local security feature chain ring agent, a target agent corresponding to the feature hash value having the small-world network feature as a first updated security risk analysis engine; Using all of the updated local security feature chain ring agents and the first updated security risk analysis engine to construct an updated global security feature chain ring; Alternatively, all of the update local security feature chain ring agents and the preset second update security risk analysis engine are used to construct an update global security feature chain ring.

8. The ship safety management system based on visual recognition technology according to claim 1 is characterized in that: The intelligent decision-making layer is also used to: when the management request carries a new security supervision requirement, calculate the new feature hash value corresponding to the new security supervision requirement; store the new security supervision requirement in the new supervision policy mapped to the new feature hash value in the execution supervision layer.

9. A ship safety management method based on visual recognition technology, characterized in that: Applied to the security management system according to any one of claims 1 to 8, the method comprises: High-definition cameras deployed in key areas of the ship collect facial data and ship environment data to the perception interaction layer; the perception interaction layer receives input management requests; and the perception interaction layer broadcasts the management requests to the intelligent decision-making layer; Determining, by the intelligent decision layer based on the management request, a target security feature chain ring from a plurality of built-in security feature chain rings; the security feature chain ring is composed of a plurality of feature hash values; Calculate the characteristic hash value of the management request by the intelligent decision layer; Determine the target security level corresponding to the feature hash value from the target security feature chain by the intelligent decision layer; executing a supervisory policy mapped to the target security level by executing a supervisory layer in response to the target security level; The step of executing the regulatory policy mapped to the target security level in response to the target security level by the execution regulatory layer includes: determining, by the regulatory policy agent, the regulatory policy mapped to the target security level from the plurality of policy execution addresses; In response to the target safety level, executing the supervision strategy to adjust the safety status of the ship; It also includes classifying multiple management requests according to preset security classification rules through the perception interaction layer, generating corresponding security description files and caching them; the preset security classification rules include personnel qualification rules, ship manning rules and navigation safety rules.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the security management method according to claim 9 is implemented.

Images