An intelligent safety supervision system for ships

Abstract

The present invention discloses an intelligent safety supervision system applied to ships, and relates to the field of ship safety technology. The image acquisition module acquires high-definition images in real time; the automatic identification module obtains ship dynamic data and ship static data; the ship server performs feature recognition on the ship dynamic data and ship static data to obtain data processing results; sends ship dynamic data, ship static data and data processing results; receives alarm indication information; the alarm module outputs an alarm; the ship client displays the data processing results; determines whether to send alarm indication information according to the data processing results; the communication module receives and sends ship dynamic data, ship static data and data processing results; the shore supervision system includes: the ship safety supervision big data analysis platform performs secondary feature recognition on the ship dynamic data and ship static data to obtain secondary data processing results. The present invention realizes online real-time supervision and early warning of ships, and improves supervision efficiency.

Description

Technical Field

[0001] The present invention relates to the technical fields of the Internet, artificial intelligence and safe operation of ships, and in particular to an intelligent safety supervision system applied to ships. Background Art

[0002] Ship supervision is an important part of the management and safety of the ship industry. There are two main ways to deal with the surveillance video deployed on ships:

[0003] 1) Shipping companies require the captain to monitor the operation of each position of the ship in real time and stop any violation immediately. The captain is also required to check the surveillance video of the previous day every day and fill in the "Daily Surveillance Video Inspection Record of the Ship" for future reference.

[0004] 2) The shore-based safety management personnel of the shipping company remotely receives the monitoring video data of the ships at sea through the satellite network, and manually searches and screens the irregularities and unsafe situations in the data.

[0005] It can be found that the current processing of surveillance video data is completely dependent on personnel's analysis and judgment of various behaviors and states, which will consume a lot of manpower, resulting in high operating costs and low efficiency, and there are serious safety hazards.

[0006] In recent years, remote supervision has been carried out with the help of video surveillance systems and Internet systems, which can achieve full-process recording and improve the coverage of supervision and inspection. However, there are also problems such as the inability to accurately supervise at close range and from multiple angles, the inability to conduct online real-time intelligent analysis, and insufficient early warning. The safety supervision system still needs further innovation. Summary of the invention

[0007] The purpose of the embodiment of the present invention is to provide an intelligent safety supervision system applied to ships, so as to realize online real-time supervision and early warning of ships, thereby improving supervision efficiency.

[0008] To achieve the above purpose, the embodiment of the present invention provides the following solution:

[0009] An intelligent safety supervision system applied to ships, comprising: a ship-side supervision system and a shore-side supervision system;

[0010] The ship-side monitoring system comprises:

[0011] Image acquisition module, used to acquire high-definition images in real time;

[0012] An automatic identification module, used to obtain ship dynamic data and ship static data at least based on the high-definition image through a deep learning algorithm of a convolutional neural network;

[0013] The ship server is connected to the automatic identification module and is used to:

[0014] Performing feature recognition on the ship dynamic data and the ship static data to obtain a data processing result;

[0015] Sending the ship dynamic data, the ship static data and the data processing result;

[0016] Receiving alarm indication information;

[0017] an alarm module, connected to the ship server, and configured to output an alarm according to the alarm indication information;

[0018] The ship client is connected to the ship server and is used for:

[0019] Display the data processing result;

[0020] Determining whether to send the alarm indication information according to the data processing result;

[0021] If the characteristic value in the data processing result is greater than or equal to a first preset threshold, sending the alarm indication information;

[0022] If the characteristic value in the data processing result is less than a first preset threshold, sending normal operation information;

[0023] A communication module, connected to the ship server, for receiving and sending the ship dynamic data, the ship static data and the data processing result;

[0024] The shore supervision system includes:

[0025] Ship safety supervision big data analysis platform, used for:

[0026] A convolutional neural network deep learning algorithm is used to perform secondary feature recognition on the ship dynamic data and the ship static data to obtain a secondary data processing result.

[0027] Optionally, the ship-side monitoring system further includes:

[0028] The fire monitoring module is connected to the ship server and is used to:

[0029] Acquire the alarm indication information, smoke data, water pressure data and water level data of the ship; and send the smoke data, water pressure data and water level data of the ship;

[0030] Firefighting is carried out according to the alarm indication information.

[0031] Optionally, the shore monitoring system further includes:

[0032] A land server, connected to the communication module, is used to:

[0033] Receiving the ship dynamic data, the ship static data and the data processing result;

[0034] Classify and store the ship dynamic data, the ship static data and the data processing results;

[0035] The land client is connected to the ship safety supervision big data analysis platform and is used to:

[0036] Display the ship dynamic data, the ship static data, the data processing result and the secondary data processing result;

[0037] Determining whether to send the alarm indication information according to the secondary data processing result;

[0038] If the characteristic value in the secondary data processing result is greater than or equal to the first preset threshold, sending the alarm indication information;

[0039] If the characteristic value in the secondary data processing result is less than the first preset threshold, normal operation information is sent.

[0040] Optionally,

[0041] The ship dynamic data includes: at least one of ship fire data, staff operation data, staff on-duty data, ship position data, ship speed data, ship heading data, smoke data, water pressure data and water level data;

[0042] The ship static data includes: at least one of the ship name, call sign and ship draft.

[0043] Optionally, the ship-side monitoring system further includes:

[0044] The master clock is connected to the ship server and is used to provide a unified time reference for the slave clocks on the ship and the ship server.

[0045] Optionally, the ship server includes:

[0046] The feature fusion classification unit is used to classify the fusion features under the influence of different factors in the ship dynamic data and the ship static data according to the character feature fusion method to obtain feature fusion classification; the different factors include: color, illumination or pitch angle; the feature fusion classification function is:

[0047]

[0048] Among them, the number of characteristic components under the influence of different factors is U, f jis the fusion feature of the jth person under the influence of different factors, η i is the feature component before fusion, a i is the weight of the fused feature component.

[0049] Optionally, the ship safety supervision big data analysis platform includes:

[0050] The prediction unit is used to use the state prediction equation to predict the current value by the value of the final state to obtain the prior state estimate value.

[0051] An updating unit is used to update the prior state estimate using a Kalman gain coefficient equation and a state update equation Perform optimization updates to obtain the posterior state estimate The posterior state estimate is the secondary data processing result.

[0052] Optionally, the state prediction equation is:

[0053]

[0054]

[0055] in, is the prior estimated covariance at time k, P k-1 is the prior estimated covariance at the K-1 moment; A k is the state transfer matrix; T represents transpose; Q is the system noise covariance matrix; represents the prior state estimate at time k, The optimal estimate at time k-1 predicts the state at time k; u k is the state control vector; B k is the control variable matrix.

[0056] Optionally, the Kalman gain coefficient equation is:

[0057]

[0058] Among them, K k is the Kalman gain, R is the covariance matrix of the measurement noise;

[0059] The state update equation is:

[0060]

[0061]

[0062] in, is the posterior state estimate at time k; Pk is the posterior estimated covariance at time k; z k is the measurement vector; H k is the transformation matrix from state vector to measurement vector; I is the identity matrix.

[0063] In the embodiment of the present invention, high-definition images are collected in real time at close range and from multiple angles by an image acquisition module, and feature recognition is performed on a large number of high-definition images by an automatic recognition module, thereby realizing intelligent and automated supervision and monitoring, changing the traditional operating mechanism that completely relies on personnel to analyze and judge various behaviors and states, and saving a lot of manpower.

[0064] The alarm module outputs an alarm to maximize the reminder effect. High-definition images can record the normal behavior of staff, form personnel attendance and work records, and facilitate data analysis for ship managers. It realizes online real-time supervision of ships, improves early warning capabilities, and improves supervision efficiency.

[0065] The communication module is connected to the ship server. The communication module receives and sends ship dynamic data, ship static data and data processing results, classifies data transmission and data reception, ensures the stable operation of the high-speed data return function in a weak network environment, and ensures the continuity and stability of data return during the ship's navigation.

[0066] The ship safety supervision big data analysis platform realizes secondary identification of features and improves the accuracy of feature identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0068] Figure 1 A schematic diagram of the structure of an intelligent safety supervision system applied to ships provided by an embodiment of the present invention;

[0069] Figure 2 A detailed structural diagram of an intelligent safety supervision system applied to ships provided by an embodiment of the present invention;

[0070] Figure 3 A schematic diagram of a process flow of an intelligent safety supervision system applied to ships provided in an embodiment of the present invention.

[0071] Explanation of symbols:

[0072] Image acquisition module-1, automatic identification module-2, ship server-3, ship client-4, fire monitoring module-5, master clock-6, alarm module-7, communication module-8, ship safety supervision big data analysis platform-9. DETAILED DESCRIPTION

[0073] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0074] The purpose of the present invention is to provide an intelligent safety supervision system for ships, so as to solve the existing problem that the current processing of monitoring video data completely relies on personnel's analysis and judgment of various behaviors and states, which will consume a lot of manpower, resulting in high operating costs and low efficiency, serious safety hazards, and cannot be accurately supervised at close range and from multiple angles, and cannot be online for real-time intelligent analysis and insufficient early warning.

[0075] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0076] Figure 1 An exemplary structure of the above-mentioned intelligent safety supervision system applied to ships is shown, including: a ship-side supervision system and a shore-side supervision system. Each part is introduced in detail below.

[0077] The ship-side monitoring system at least includes: an image acquisition module 1, an automatic identification module 2, a ship server 3, a ship client 4, a fire monitoring module 5, a master clock 6 and an alarm module 7.

[0078] The image acquisition module 1 is used to acquire high-definition images in real time.

[0079] In one example, the image acquisition module 1 may be a high-definition explosion-proof camera. The high-definition explosion-proof camera uploads high-definition images within a region in real time. The region includes the range within which the high-definition explosion-proof camera rotates to record.

[0080] The automatic identification module 2 is used to obtain ship dynamic data and ship static data based on at least high-definition images through a deep learning algorithm of a convolutional neural network.

[0081] The ship dynamic data includes: at least one of ship fire data, staff on-duty data, ship position data, ship speed data, ship heading data, smoke data, water pressure data and water level data.

[0082] The static data of the ship includes at least one of the ship name, call sign and ship draft.

[0083] In one example, the automatic identification module 2 receives high-definition images and can at least obtain the data of staff on duty. The automatic identification module 2 cooperates with the Beidou satellite navigation system and / or GPS to send out ship fire data, staff on duty data, ship position data, ship speed data, ship heading data, smoke data, water pressure data, water level data and other ship dynamic data and ship static data such as ship name, call sign and ship draft depth.

[0084] The ship server 3 is connected to the automatic identification module 2, and the ship server 3 is used to:

[0085] Perform feature recognition on ship dynamic data and ship static data to obtain data processing results;

[0086] Send ship dynamic data, ship static data and data processing results;

[0087] Receive alarm indication information.

[0088] In one example, after receiving the ship dynamic data and the ship static data, the ship server 3 performs feature recognition on them, for example, comparing the ship fire data with the temperature threshold to obtain normal temperature data or abnormal temperature data; comparing the ship position data with the preset heading to obtain the normal position or abnormal position; comparing the ship speed data with the preset speed to obtain the normal speed or abnormal speed; comparing the ship heading data with the preset heading to obtain the normal heading or abnormal heading; comparing the smoke data with the preset smoke concentration to obtain normal smoke data or abnormal smoke data; comparing the water pressure data with the preset water pressure to obtain normal water pressure or abnormal water pressure; comparing the water level data with the preset water level to obtain the normal water level or abnormal water level; comparing the ship draft with the preset depth to obtain the normal depth or abnormal depth; performing face recognition on the staff on-the-job data to obtain normal on-the-job or off-the-job data.

[0089] The data processing results include at least: normal temperature data or abnormal temperature data, normal position or abnormal position, normal speed or abnormal speed, normal heading or abnormal heading, normal smoke data or abnormal smoke data, normal water pressure or abnormal water pressure, normal water level or abnormal water level, normal depth or abnormal depth, normal on-duty data or off-duty data.

[0090] Feature recognition includes big data storage analysis and behavior recognition. Big data storage analysis includes massive data models and natural language processing. Big data storage analysis has real-time data processing, high-speed data transmission, search, data analysis and other functions. Behavior recognition is specifically image recognition technology. Collect images of people who are making phone calls, sleeping, not wearing helmets, not wearing work clothes, fighting, etc., and extract their features such as making phone calls, sleeping, not wearing helmets, not wearing work clothes, fighting, etc. through the deep learning algorithm of convolutional neural network, and then match the features of making phone calls, sleeping, not wearing helmets, not wearing work clothes, fighting, etc. with the biological features in the database to obtain data processing results.

[0091] The ship server 3 at least includes: a feature fusion classification unit.

[0092] The feature fusion classification unit is used to classify the fusion features under the influence of different factors in the ship dynamic data and the ship static data according to the character feature fusion method to obtain feature fusion classification; the different factors include: color, illumination or pitch angle; the feature fusion classification function is:

[0093]

[0094] Among them, the number of characteristic components under the influence of different factors is U, f j is the fusion feature of the jth person under the influence of different factors, η i is the feature component before fusion, a i is the weight of the fused feature component.

[0095] In one example, fusion features influenced by different factors in on-the-job data of staff members are graded according to a person feature fusion method to obtain a feature fusion grade.

[0096] The ship server 3 can also perform data compression and network segmentation on the high-definition image and then display the high-definition image on the display screen in real time. The network segmentation can be segmented at least according to time.

[0097] The feature fusion grading unit can be specifically composed of a high-performance parallel computing server and an extremely fast graphics processor to achieve automatic recognition of staff behavior and high-definition image encoding.

[0098] In an embodiment of the present invention, in terms of personnel supervision, a technology based on face recognition and edge differential adaptive dynamic tracking is adopted. Face recognition adopts the deep learning method of convolutional neural network, which accurately identifies effective features through a large amount of training, can be optimized for complex scenes such as cabin passages, and uses the high-speed computing performance of the graphics processor to improve the speed and accuracy of face recognition. When employees enter the working area, the feature fusion grading unit performs "unconscious" personnel clocking in and tracking and positioning for the employees. During the working period, the behavior of employees is recorded, such as whether they wear a safety helmet, whether they are absent from work, whether they fight, whether they sleep, etc. During non-working hours, personnel leaving the work area perform "unconscious" personnel clocking out.

[0099] The identification of fighting within the area is to detect and capture the human body posture based on the key points of the human body to detect whether there is a fighting behavior; the identification of wearing of safety helmets is to designate cameras and monitor the presence of people without safety helmets within the monitoring range; the identification of illegal use of mobile phones in the workplace is to designate cameras and monitor the presence of people making phone calls within the monitoring range; the identification of sleeping behavior in the cab is suspected sleeping behavior (motionless, dazed) lasting for 5 minutes; the identification of absence in the cab is to record the entry and exit cards of all staff in the cab. If no new staff member punches in after 5 minutes, an absence warning will be issued.

[0100] Warning behaviors also include statistics of people getting on and off the ship, and warnings for on-duty personnel being disturbed. Statistics of people getting on and off the ship are used to identify unauthorized people boarding the ship, and to count all non-crew members leaving the ship when leaving the port; warnings for on-duty personnel being disturbed are used for behaviors such as non-duty personnel chatting on the bridge, or drivers doing work unrelated to their duty.

[0101] Non-warning behaviors can be realized by using face recognition technology to make statistics and records of personnel attendance, regular meetings, and exercises. Attendance, that is, automatically recording the working hours of staff in the designated workplace, whether the captain and the chief engineer stick to their posts under the specified conditions, and whether the duty personnel take over on time, is used to count the working hours of relevant staff in the engine room and the bridge; regular meetings, that is, selecting relevant staff to attend meetings according to company requirements, is used to ensure that relevant staff participate on time; exercises, that is, conducting life-saving and fire drills according to company requirements, ensuring that relevant staff participate and wear work clothes and life jackets.

[0102] Attendance records are warning pictures and text descriptions with timestamps. The text descriptions include: person name, time of occurrence, area and warning behavior, and support the export of warning pictures and text descriptions.

[0103] The ship server 3 can also classify and store the data processing results for viewing, for example, by time, or by normal data and abnormal data. The classified data is then converted into data that matches the transmission in the satellite broadband dynamic return technology for transmission.

[0104] The alarm module 7 is connected to the ship server 3, and the alarm module 7 is used to output an alarm according to the alarm indication information.

[0105] In one example, the alarm module 7 may specifically be a broadcast general alarm system. When face recognition is performed, a dangerous situation or a warning behavior occurs, the broadcast general alarm system at least outputs an audio file or a video file for reminders or warnings. Dangerous situations refer to situations such as fires within the area. Warning behaviors include fighting among staff within the area, not wearing safety helmets, sleeping in the work area, playing with mobile phones, and leaving the post. Normal behaviors include attendance, regular meetings, exercises, etc.

[0106] The ship client 4 is connected to the ship server 3, and the ship client 4 is used for:

[0107] Display data processing results;

[0108] Determine whether to send alarm indication information according to the data processing result;

[0109] If the characteristic value in the data processing result is greater than or equal to the first preset threshold, an alarm indication message is sent;

[0110] If the characteristic value in the data processing result is less than the first preset threshold, normal operation information is sent.

[0111] In one example, the ship client 4 may be a display terminal, such as a display, a computer, etc. The display displays the data processing result. The computer determines whether to send the alarm indication information according to the data processing result.

[0112] The ship client 4 can also be a mobile terminal, such as a mobile phone, a tablet computer, etc. The mobile terminal is used to synchronize the recognition results of the staff's behavior. The staff can view the results of attendance, regular meetings, exercises, etc. through the mobile terminal. For example, face recognition is performed on the staff entering the cab, and voice feedback is provided during face recognition. The recognition screen and text description with a timestamp will be synchronized to the mobile terminal.

[0113] When the ship client 4 is applied to an electronic device in the form of software, in one example, the client can be an independent software (such as an APP deployed on a terminal), or an operating system or operating system-level program of the terminal. In addition, it can also be a mini program (Mini Program) in WeChat or an H5 (html5) web page, etc. It should be noted that a "mini program" is an application that can be used without downloading and installing.

[0114] The communication module 8 is connected to the ship server 3, and the communication module 8 is used to receive and send ship dynamic data, ship static data and data processing results.

[0115] In one example, the communication module 8 may be a satellite broadband dynamic backhaul technology of the Beidou satellite navigation system or GPS. In addition, when the ship is sailing in the offshore, it may also receive and send ship dynamic data, ship static data and data processing results through a wireless data network (5G).

[0116] Satellite broadband dynamic return technology separates data transmission and data reception without affecting each other. When transmitting data back, the system automatically cuts and compresses the data, and transmits data back in segments of 3MB each time. The data transmission rate is 0.5MB / s. After all segments are transmitted, decompression and segment merging are performed. In addition, during the data transmission process, the system continuously records the progress of the transmission. If the line is disconnected or other abnormalities occur, the system automatically queries the data transmission status and resumes the transmission from the breakpoint. The high-speed data transmission function is guaranteed to operate stably in a weak network environment, ensuring the continuity and stability of data transmission during the ship's navigation.

[0117] The shore monitoring system includes:

[0118] The Ship Safety Supervision Big Data Analysis Platform 9 is used for:

[0119] The deep learning algorithm of convolutional neural network is used to perform secondary feature recognition on the ship dynamic data and ship static data to obtain the secondary data processing results.

[0120] In an example, the secondary feature recognition process of the ship safety supervision big data analysis platform 9 refers to the process of feature recognition of ship dynamic data and ship static data by the above-mentioned ship server 3, which will not be repeated here.

[0121] The Ship Safety Supervision Big Data Analysis Platform 9 adopts a deep learning kernel with gradient adaptive optimization. By accumulating square gradients, it makes the process smoother, thereby speeding up the training process. At the same time, it adds a decay coefficient to the accumulated result to control the accumulation of historical information, change the learning rate, and solve the problem that the subsequent learning rate is very small and the weights cannot be effectively updated. It improves the generalization ability of the training model, forms effective features for accurately identifying head posture, occlusion, lighting conditions, and facial expressions, and realizes more accurate recognition through secondary feature recognition, filtering out erroneous information in feature recognition.

[0122] The ship safety supervision big data analysis platform 9 at least includes: a prediction unit and an update unit.

[0123] The prediction unit is used to use the state prediction equation to predict the current value by the value of the final state to obtain the prior state estimate.

[0124] The state prediction equation is:

[0125]

[0126]

[0127] in, is the prior estimated covariance at time k, P k-1 is the prior estimated covariance at the K-1 moment; A k is the state transfer matrix; T represents transpose; Q is the system noise covariance matrix; represents the prior state estimate at time k, The optimal estimate at time k-1 predicts the state at time k; u k is the state control vector; B k is the control variable matrix.

[0128] An updating unit is used to update the prior state estimate using a Kalman gain coefficient equation and a state update equation Perform optimization updates to obtain the posterior state estimate The posterior state estimate is the secondary data processing result.

[0129] The Kalman gain coefficient equation is:

[0130]

[0131] Among them, K k is the Kalman gain, and R is the covariance matrix of the measurement noise.

[0132] The state update equation is:

[0133]

[0134]

[0135] in, is the posterior state estimate at time k; P k is the posterior estimated covariance at time k; z k is the measurement vector; H k is the transformation matrix from state vector to measurement vector; I is the identity matrix.

[0136] Because there may be a lot of occlusion problems in the image, the ship safety supervision big data analysis platform 9 adopts the target tracking technology based on Kalman filtering and related filtering algorithms. The ship safety supervision big data analysis platform 9 can be specifically a Kalman filter. The Kalman filter includes a prediction unit and an update unit to obtain a more accurate prediction value. By repeating the above Kalman gain coefficient equation and state update equation, the optimal value of the current prediction value can be calculated, and the dynamic trajectory of the multi-target position of the staff is predicted. It adapts to various operating conditions in the cabin, deck, and stern, greatly improving the success rate of dynamic violation identification, and providing a solid foundation for the intelligent navigation safety system.

[0137] To sum up, in the embodiments of the present invention, high-definition images are collected in real time at close range and from multiple angles through the image acquisition module, and feature recognition is performed on a large number of high-definition images using the automatic recognition module, thereby realizing intelligent and automated supervision and monitoring, changing the traditional operating mechanism that completely relies on personnel to analyze and judge various behaviors and states, and saving a lot of manpower.

[0138] The alarm module outputs an alarm to maximize the reminder effect. High-definition images can record the normal behavior of staff, form personnel attendance and work records, and facilitate data analysis for ship managers. It realizes online real-time supervision of ships, improves early warning capabilities, and improves supervision efficiency.

[0139] The communication module is connected to the ship server. The communication module receives and sends ship dynamic data, ship static data and data processing results, classifies data transmission and data reception, ensures the stable operation of the high-speed data return function in a weak network environment, and ensures the continuity and stability of data return during the ship's navigation.

[0140] The ship safety supervision big data analysis platform realizes secondary identification of features and improves the accuracy of feature identification.

[0141] In other embodiments of the present invention, the ship-side monitoring system further includes: a fire monitoring module 5.

[0142] See also Figure 2The fire monitoring module 5 is connected to the ship server 3, and the fire monitoring module 5 is used to:

[0143] Obtain alarm indication information, smoke data, water pressure data and water level data of the ship; and send the smoke data, water pressure data and water level data of the ship;

[0144] Carry out fire fighting according to the alarm instructions.

[0145] In one example, the fire monitoring module 5 includes at least an automatic fire alarm sensor, a fire water tank level sensor, a fire pipe network pressure sensor, a fire power supply voltage sensor, and an automatic fire extinguishing equipment controller.

[0146] The automatic fire alarm sensor has the function of sound and light alarm. When a fire occurs, it will promptly send out sound and light alarm and upload the alarm indication information to the fire monitoring module 5.

[0147] Fire water tank level sensor, fire pipe network pressure sensor: used to monitor the fire water tank level and fire pipe network pressure in real time, and upload the data to the fire monitoring module 5.

[0148] Fire power supply voltage sensor: monitors the power supply voltage in real time and uploads the voltage data to the fire monitoring module 5.

[0149] Automatic fire extinguishing equipment controller: includes water spray fire extinguishing system, fine water mist fire extinguishing system, foam spray fire extinguishing system, oil drainage and nitrogen injection fire extinguishing system, gas fire extinguishing system, etc., with automatic, manual, alarm indication information remote control and emergency mechanical operation functions.

[0150] In other embodiments of the present invention, the shore monitoring system further includes:

[0151] The land server is connected to the communication module 8, and the land server is used to:

[0152] Receive ship dynamic data, ship static data and data processing results;

[0153] The ship dynamic data, ship static data and data processing results are classified and stored.

[0154] In one example, the land server includes data synchronization and application services.

[0155] Data synchronization is used to receive the ship dynamic data, ship static data and data processing results transmitted in real time by satellite broadband dynamic return technology, classify and store the ship dynamic data, ship static data and data processing results, and support data query and backtracking. Please refer to the above for the classification method.

[0156] The application service is used to establish a big data analysis platform for ship safety supervision9 to achieve situational awareness and global awareness of ship safety supervision.

[0157] The land client is connected to the ship safety supervision big data analysis platform 9, and the land client is used to:

[0158] Display ship dynamic data, ship static data, data processing results and secondary data processing results;

[0159] Determining whether to send the alarm indication information according to the secondary data processing result;

[0160] If the characteristic value in the secondary data processing result is greater than or equal to the first preset threshold, sending an alarm indication message;

[0161] If the characteristic value in the secondary data processing result is less than the first preset threshold, normal operation information is sent.

[0162] The number of times the ship safety supervision big data analysis platform 9 sends alarm indication information is compared with the number of times the ship client 4 sends alarm indication information to obtain the overlap rate; if the overlap rate is greater than or equal to the second preset threshold, the output alarm is accurate; if the overlap rate is less than the second preset threshold, the output alarm is inaccurate.

[0163] In one example, the land client is used to manage and display in real time the alarm indication information of all ships in the fleet, enabling the alarm indication information to be viewed and automatically exported by ship, scene, type, and time period.

[0164] In other embodiments of the present invention, the ship-side monitoring system further includes:

[0165] The master clock 6 is connected to the ship server 3, and the master clock 6 is used to provide a unified time reference for the slave clocks and the ship server on the ship.

[0166] See also Figure 3 , the image acquisition module 1 acquires high-definition images in real time (monitoring video import), the automatic identification module 2 obtains ship dynamic data and ship static data (streaming media analysis), the ship server 3 performs feature recognition on the ship dynamic data and ship static data to obtain data processing results (big data storage analysis, behavior recognition), and sends ship dynamic data, ship static data and data processing results (early warning behavior). If an alarm indication message is received, the alarm module 7 outputs an alarm. If not, OCR records are performed, and the ship dynamic data, ship static data and data processing results are stored (data storage). The ship client 4 displays the data processing results, and determines whether to send an alarm indication message based on the data processing results. The communication module 8 receives and sends ship dynamic data, ship static data and data processing results (data storage).

[0167] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0168] This article uses specific examples to illustrate the principles and implementation methods of the embodiments of the present invention. The description of the above embodiments is only used to help understand the methods and core ideas of the embodiments of the present invention. At the same time, for those skilled in the art, according to the ideas of the embodiments of the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the embodiments of the present invention.

Claims

1. An intelligent safety supervision system applied to ships, characterized in that: include: Ship-side supervision system and shore-side supervision system; The ship-side monitoring system comprises: Image acquisition module, used to acquire high-definition images in real time; An automatic identification module, used to obtain ship dynamic data and ship static data at least based on the high-definition image through a deep learning algorithm of a convolutional neural network; The ship server is connected to the automatic identification module and is used to: Performing feature recognition on the ship dynamic data and the ship static data to obtain a data processing result; The feature recognition includes big data storage analysis and behavior recognition; The data processing result at least includes: normal temperature data or abnormal temperature data, normal position or abnormal position, normal speed or abnormal speed, normal heading or abnormal heading, normal smoke data or abnormal smoke data, normal water pressure or abnormal water pressure, normal water level or abnormal water level, normal depth or abnormal depth, normal on-duty or off-duty data; Sending the ship dynamic data, the ship static data and the data processing result; Receiving alarm indication information; an alarm module, connected to the ship server, and configured to output an alarm according to the alarm indication information; The ship client is connected to the ship server and is used for: Display the data processing result; Determining whether to send the alarm indication information according to the data processing result; If the characteristic value in the data processing result is greater than or equal to a first preset threshold, sending the alarm indication information; If the characteristic value in the data processing result is less than a first preset threshold, sending normal operation information; A communication module, connected to the ship server, for receiving and sending the ship dynamic data, the ship static data and the data processing result; The shore supervision system includes: Ship safety supervision big data analysis platform, used for: Using a deep learning algorithm of a convolutional neural network to perform secondary feature recognition on the ship dynamic data and the ship static data, and obtain a secondary data processing result; The land client is connected to the ship safety supervision big data analysis platform and is used to: Display the ship dynamic data, the ship static data, the data processing result and the secondary data processing result; Determining whether to send the alarm indication information according to the secondary data processing result; If the characteristic value in the secondary data processing result is greater than or equal to the first preset threshold, sending the alarm indication information; If the characteristic value in the secondary data processing result is less than the first preset threshold, sending normal operation information; The number of times the ship safety supervision big data analysis platform sends the alarm indication information is compared with the number of times the ship client sends the alarm indication information to obtain the overlap rate; if the overlap rate is greater than or equal to the second preset threshold, the output alarm is accurate; if the overlap rate is less than the second preset threshold, the output alarm is inaccurate.

2. The intelligent safety supervision system for ships according to claim 1 is characterized in that: The ship-side monitoring system also includes: The fire monitoring module is connected to the ship server and is used to: Acquire the alarm indication information, smoke data, water pressure data and water level data of the ship; and send the smoke data, water pressure data and water level data of the ship; Firefighting is carried out according to the alarm indication information.

3. The intelligent safety supervision system for ships according to claim 1 is characterized in that: The shore supervision system also includes: A land server, connected to the communication module, is used to: Receiving the ship dynamic data, the ship static data and the data processing result; The ship dynamic data, the ship static data and the data processing results are classified and stored.

4. The intelligent safety supervision system for ships according to claim 2 is characterized in that: The ship dynamic data includes: at least one of ship fire data, staff on-duty data, ship position data, ship speed data, ship heading data, smoke data, water pressure data and water level data; The ship static data includes: at least one of the ship name, call sign and ship draft.

5. The intelligent safety supervision system for ships according to claim 1 is characterized in that: The ship-side monitoring system also includes: The master clock is connected to the ship server and is used to provide a unified time reference for the slave clocks on the ship and the ship server.

6. The intelligent safety supervision system for ships according to claim 1, characterized in that: The ship server comprises: The feature fusion classification unit is used to classify the fusion features under the influence of different factors in the ship dynamic data and the ship static data according to the character feature fusion method to obtain feature fusion classification; the different factors include: color, illumination or pitch angle; the feature fusion classification function is: Among them, the number of characteristic components under the influence of different factors is U, f j is the fusion feature of the jth person under the influence of different factors, η i is the feature component before fusion, a i is the weight of the fused feature component.

7. The intelligent safety supervision system for ships according to claim 1 is characterized in that: The ship safety supervision big data analysis platform includes: The prediction unit is used to use the state prediction equation to predict the current value by the value of the final state to obtain the prior state estimate value. An updating unit is used to update the prior state estimate using a Kalman gain coefficient equation and a state update equation Perform optimization updates to obtain the posterior state estimate The posterior state estimate is the secondary data processing result.

8. The intelligent safety supervision system for ships according to claim 7 is characterized in that: The state prediction equation is: in, is the prior estimated covariance at time k, P k-1 is the prior estimated covariance at the K-1 moment; A k is the state transfer matrix; T represents transpose; Q is the system noise covariance matrix; represents the prior state estimate at time k, The optimal estimate at time k-1 predicts the state at time k; u k is the state control vector; B k is the control variable matrix.

9. The intelligent safety supervision system for ships according to claim 8, characterized in that: The Kalman gain coefficient equation is: Among them, K k is the Kalman gain, R is the covariance matrix of the measurement noise; The state update equation is: in, is the posterior state estimate at time k; P k is the posterior estimated covariance at time k; z k is the measurement vector; H k is the transformation matrix from state vector to measurement vector; I is the identity matrix.

Images