A ship sewage well monitoring system and method

Abstract

The present invention provides a ship sewage well monitoring system and method, comprising the following steps: S1: acquiring real-time monitoring video of a ship target area including a sewage well; S2: inputting time-series image frames of the real-time monitoring video into a trained sewage well leakage monitoring model to output the seepage area and the seepage situation, realizing leakage detection of the ship sewage well through a specially designed sewage well detection model, and realizing automatic, real-time monitoring and alarm of the ship sewage well through a reasonable hardware arrangement, without the need for crew members to frequently enter a dangerous sewage well environment, thereby ensuring the safety of the crew members.

Description

Technical Field

[0001] The present invention relates to the technical field of ship equipment monitoring, and in particular to a ship sewage well monitoring system and method. Background Art

[0002] In the long-term operation of ships, the management of sewage wells is related to ship safety and environmental protection. If the sewage in the sewage well cannot be effectively managed, long-term accumulation may cause the sewage well to be overloaded, increase the uneven distribution of local weight of the ship, and affect the overall center of gravity stability of the ship. In addition, the sewage of some ships may contain flammable and explosive substances, such as oily sewage. If the sewage well is not managed properly, these substances will accumulate in the sewage well and easily cause fire or explosion when encountering a fire source or high temperature environment.

[0003] Traditional sewage well monitoring relies on regular inspections by crew members, which has many problems: ① The location of sewage wells is often relatively hidden and the environment is complex. They are often hidden deep in the cabin. Crew inspections are time-consuming and pose safety risks such as hypoxia and harmful gases. They are also inefficient; ② Minor changes or initial abnormalities, such as slow leakage and small water level fluctuations, are difficult to detect in time, which may affect the normal operation of the ship.

[0004] Of course, the prior art also has a ship bilge water system BILGE SYSTEM for managing the accumulation of water in the bilge of a ship. A ship bilge water alarm system is also provided in the ship bilge water system BILGE SYSTEM. The ship bilge water alarm system monitors the operating status of the ship bilge water system and the bilge water level, and issues an alarm when an abnormal situation occurs. However, most of the current ship bilge water alarm systems use a contact type high-level alarm for alarming, such as a water level alarm device for a tugboat bilge sewage tank proposed in CN107655544A, and another example is a water level alarm device for a tugboat bilge sewage tank proposed in CN117508553A. A bilge water alarm device and a method for using the same are disclosed. The implementation method is to trigger an alarm by checking whether a water level sensor triggers an alarm through contact action. However, this method also has many problems: ① The method of relying on a water level sensor is limited by the accuracy of the sensor. Some sensors have poor reliability in harsh environments. When the ship is in special working conditions such as severe shaking and vibration, the sensor signal is easily interfered with, resulting in false alarms or missed alarms; ② It can only determine whether the bilge water has reached the set water level, and cannot accurately locate the seepage area. It is difficult to detect slow seepage, initial seepage, etc. For example, when a slow seepage is caused by a minor local damage to the ship, as long as the water level does not reach the threshold, there will be no alarm. When the water level rises and triggers the alarm, the seepage may have continued for a long time. ③ It is impossible to intuitively view the specific seepage and alarm conditions of each ship.

[0005] In other words, the current ship bilge water system has many problems in monitoring sewage well seepage, such as poor timeliness and poor accuracy; at the same time, considering the particularity of the ship's sewage well, the existing technology lacks integrated equipment designed for the special environment of the ship's sewage well in terms of hardware equipment; in terms of installation method, the complex cabin structure and network wiring, lighting and other on-site conditions of the ship are not fully considered, resulting in installation difficulties and affecting the monitoring effect; at the algorithm level, there is a lack of effective algorithms specifically for accurate analysis and fault warning of ship sewage well monitoring data. These defects have led to many obstacles in the current sewage well monitoring in ship scenarios. Summary of the invention

[0006] The purpose of the present invention is to provide a ship sewage well monitoring system and method, which realizes leakage detection of ship sewage wells through a specially designed sewage well detection model, and realizes automatic and real-time monitoring and alarm of ship sewage wells with a reasonable supporting hardware arrangement, without the need for crew members to frequently enter the dangerous sewage well environment, thereby ensuring the safety of crew members' lives.

[0007] To achieve the above objectives, the present technical solution provides a method for monitoring a ship sewage well, comprising the following steps:

[0008] S1: Obtain real-time surveillance video of the ship's target area including the sewage well;

[0009] S2: Input the time-series image frames of the real-time monitoring video into the trained sewage well leakage monitoring model to output the seepage area and seepage situation, wherein the sewage well leakage monitoring model includes a plurality of modules connected in sequence, wherein the modules include a backbone network for extracting image frame features, an aggregation module for fusing multi-scale features, a time-series module for extracting time-series features, and a decoding module for generating the final monitoring results, wherein the decoding module optimizes features in combination with the attention mechanism, wherein the backbone network uses depthwise separable convolution to extract seepage features, the aggregation module fuses seepage features of different scales to obtain fused features, the time-series module extracts and enhances the fused features of each image frame by analyzing the time-series differences of different image frames and using multi-branch convolution to obtain enhanced features, the decoding module optimizes the enhanced features to obtain optimized features and outputs the seepage area and seepage situation.

[0010] On the other hand, the present invention provides a ship sewage well monitoring system, comprising:

[0011] Management and monitoring platform;

[0012] The edge devices are connected to the management and monitoring platform through the gateway. Each edge device is connected to the AI ​​edge computing camera, and each edge device is loaded with the trained sewage well leakage monitoring model.

[0013] The AI ​​edge computing camera obtains real-time monitoring video of the ship's target area including the sewage well and transmits it to the edge device. The sewage well leakage monitoring model in the edge device analyzes the real-time monitoring video output to obtain the seepage area and seepage situation. The sewage well leakage monitoring model uses a lightweight backbone network to extract the seepage features in each image frame, and uses an aggregation module to integrate seepage features of different scales to obtain the fusion features of each image frame. The timing module is used to enhance the expression of the fusion features of each image frame based on the fusion features of different image frames to obtain enhanced features. The attention mechanism is integrated into the decoding module to predict the seepage area and seepage situation based on the enhanced features.

[0014] Compared with the prior art, this technical solution has the following characteristics and beneficial effects:

[0015] 1. Realize automatic monitoring: This solution uses AI edge computing cameras to obtain real-time monitoring videos of the target area of ​​the ship, and inputs them into the trained sewage well leakage monitoring model for analysis. It can automatically and in real time monitor the sewage well seepage area and seepage conditions, without the need for crew members to frequently enter dangerous environments, thus ensuring the safety of crew members' lives. It can also accurately locate the seepage area and promptly detect various seepage conditions.

[0016] 2. Achieve high-precision monitoring: This solution constructs a sewage well leakage monitoring model specifically suitable for ship sewage well monitoring. It uses a lightweight backbone network to extract seepage features, the aggregation module integrates features of different scales, the timing module enhances feature expression, and the decoding module incorporates the attention mechanism. The model performance is improved through multi-dimensional optimization, such as model lightweighting to reduce computational complexity, multi-scale aggregation to accurately extract features, and attention mechanism to focus on key areas, etc., to enhance detection accuracy and better adapt to the complex and changeable ship sewage well environment.

[0017] 3. Realize reasonable hardware layout: This plan proposes a reasonable hardware layout plan. Angle iron is fixedly installed in the ship cavity. AI edge computing cameras and floodlights can be detachably installed through L-shaped brackets. A waterproof junction box is installed in the ship cavity to connect the network and power lines to realize network and power supply. The characteristics of cameras and floodlights are optimized, such as modular design of cameras and functions such as automatic elimination of image jitter. The floodlights have many excellent characteristics (such as intelligent adjustment, corrosion resistance and waterproofness, etc.), which improve the stability and reliability of the system and adapt to the special environment of the ship.

[0018] 4. Improve timeliness: The seepage area, seepage situation and real-time video stream of this solution can be transmitted to the management and monitoring platform for centralized monitoring and local storage, which is convenient for staff to respond, analyze and process, and give treatment suggestions according to the situation; the management and monitoring platform can display information such as camera number, location, operating status, etc. When an alarm is triggered, the alarm device will be linked and the alarm information will be recorded, which is convenient for retrospective analysis and improves practicality and functionality.

[0019] In summary, the present invention comprehensively solves the problems of poor timeliness, low accuracy, weak adaptability, unreasonable hardware layout and insufficient system stability in existing ship sewage well monitoring solutions, and provides a more efficient, accurate, reliable and adaptable solution for ship sewage well monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a flow chart of the ship sewage well monitoring method of this scheme.

[0021] Figure 2 It is a comparison chart of the seepage area detected by the sewage well leakage monitoring model of this scheme and the seepage area detected by the traditional algorithm.

[0022] Figure 3 It is a schematic diagram of the layout of the ship sewage well monitoring system of this scheme.

[0023] Figure 4 This is a schematic diagram of the layout of the AI ​​edge computing camera of this solution.

[0024] Figure 5 It is a schematic diagram of the layout structure of the AI ​​edge computing camera of this solution.

[0025] Figure 6 It is a schematic diagram of a specific example of the operation of the ship sewage well monitoring system of this scheme.

[0026] Figure 7 It is a schematic diagram of the electronic hardware device for implementing the ship sewage well monitoring method of this scheme.

[0027] In the picture: 1-AI edge computing camera, 2-floodlight, 3-installation angle iron, 4-waterproof junction box, 5-first L-shaped bracket, 6-second L-shaped bracket. DETAILED DESCRIPTION

[0028] The following will be combined with the accompanying 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 belong to the scope of protection of the present invention.

[0029] Those skilled in the art should understand that, in the disclosure of the present invention, the terms "longitudinal", "lateral", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating the orientation or position relationship, are based on the orientation or position relationship shown in the drawings, which are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operate in a specific orientation. Therefore, the above terms should not be understood as limiting the present invention.

[0030] Embodiment 1

[0031] In view of the many drawbacks of traditional ship sewage well monitoring methods when dealing with ship sewage well monitoring: for example, they are extremely sensitive to light changes and noise, and have strong parameter dependence, making it difficult to effectively deal with noise, false changes, and changes in complex scenes. This scheme constructs a sewage well leakage monitoring model specifically suitable for ship sewage well monitoring, and realizes automatic and comprehensive monitoring of ship sewage wells based on the sewage well leakage monitoring model.

[0032] like Figure 1 As shown, this scheme provides a method for monitoring a ship sewage well, comprising the following steps:

[0033] S1: Obtain real-time surveillance video of the ship's target area including the sewage well;

[0034] S2: Input the time-series image frames of the real-time monitoring video into the trained sewage well leakage monitoring model to output the seepage area and seepage situation, wherein the sewage well leakage monitoring model includes a plurality of modules connected in sequence, wherein the modules include a backbone network for extracting image frame features, an aggregation module for fusing multi-scale features, a time-series module for extracting time-series features, and a decoding module for generating the final monitoring results, wherein the decoding module optimizes features in combination with the attention mechanism, wherein the backbone network uses depthwise separable convolution to extract seepage features, the aggregation module fuses seepage features of different scales to obtain fused features, the time-series module extracts and enhances the fused features of each image frame by analyzing the time-series differences of different image frames and using multi-branch convolution to obtain enhanced features, the decoding module optimizes the enhanced features to obtain optimized features and outputs the seepage area and seepage situation.

[0035] The sewage well leakage monitoring model provided in this solution is specially designed for ship sewage wells. It uses features extracted from a small number of FLOPs network, multi-scale feature aggregation, time series feature fusion and supervised attention mechanism to accurately identify the seepage area and seepage situation of the leakage.

[0036] Specifically, the sewage well leakage monitoring model provided by the present scheme includes a lightweight backbone network, an aggregation module, a timing module and a decoding module connected in sequence, and an attention mechanism is integrated in the decoding module, wherein the backbone network adopts a deep separable convolution to extract the seepage features of the image frame, which has the advantage of ensuring a low amount of calculation while efficiently extracting high-level features in the image frame; wherein the aggregation module fuses the seepage features of adjacent layers output by the lightweight backbone network to obtain a fused feature, which has the advantage of significantly improving the expression ability of the fused features in the image frame; wherein the timing module calculates the timing difference of the fused features of the image frames of adjacent time sequences, and uses different convolution branches to gradually extract the timing difference, performs a fusion operation after each convolution branch completes the convolution operation, and after the fusion, gradually enhances the feature expression through addition and activation functions to obtain an enhanced feature, which has the advantage of extracting the timing change information between multiple image frames of the real-time monitoring video, and performing deep fusion in a scaled manner; wherein the attention mechanism is introduced in the decoding module to obtain an optimized feature by aggregating the multi-level enhanced features layer by layer from top to bottom and optimizing the feature weights, and outputs the seepage area and seepage situation based on the optimized feature.

[0037] In an embodiment of the present solution, the backbone network is selected as a lightweight backbone network, which refers to a type of network structure designed in the field of computer vision and deep learning to effectively process image or video data in a resource-constrained environment.

[0038] In some embodiments, the backbone network in the sewage well leakage monitoring model adopts a lightweight backbone network MobileNetV2, and uses the deep separable convolution structure of MobileNetV2 as the backbone network to extract high-level features of the image. The deep separable convolution first performs deep convolution to process the image of each input channel separately, and then realizes information fusion between channels through point-by-point convolution. While ensuring a low amount of computation, it can effectively extract water seepage features related to sewage well monitoring in the image frame, such as the structure of the sewage well, the liquid state and other features, providing a basis for subsequent leakage detection.

[0039] In some embodiments, the aggregation module in the sewage well leakage monitoring model extracts seepage features of different scales through a convolution layer, processes the seepage features of each scale using a corresponding convolution block, and aggregates the processed seepage features of adjacent levels to obtain fused features. The fused features exist in the form of fused feature maps and are passed to subsequent layers.

[0040] In some embodiments, the timing module in the sewage well leakage monitoring model calculates the timing differences between the fusion features of the image frames of the time series input, uses 5 different convolution branches to gradually extract the timing features and then performs a fusion operation to fuse the information in the timing differences, and after the fusion, the feature expression is gradually enhanced by addition and activation functions to obtain enhanced features, thereby highlighting the dynamic change features such as leakage in the sewage well, so that the sewage well leakage monitoring model can more accurately identify the changing trend of seepage and leakage in the time dimension.

[0041] In some embodiments, the decoding module in the sewage well leakage monitoring model incorporates an attention mechanism, which can automatically focus on the seepage area containing leakage and suppress the interference of background noise. By aggregating multi-level features layer by layer from top to bottom and optimizing feature weights, the sewage well leakage monitoring model pays more attention to key areas related to leakage when processing images, improves the sewage well leakage monitoring model's ability to identify leakage areas, reduces misjudgments and missed judgments, and enhances the accuracy of overall detection.

[0042] Regarding the training of the sewage well leakage monitoring model: the video stream marked with water seepage area and water seepage situation is used as the training sample set for training. The trained sewage well leakage monitoring model can analyze the input video stream or image to output the water seepage area and water seepage situation. Figure 2 As shown, Figure 2 The left picture shows the seepage area monitored by the sewage well leakage monitoring model of this scheme. Figure 2 The right picture in the figure shows the recognition result of the traditional monitoring algorithm.

[0043] In order to ensure the adaptability of the sewage well leakage monitoring model to various scenarios, a diverse training sample set was constructed to train the sewage well leakage monitoring model. The training sample set comprehensively covers different seepage speeds, different seepage locations, and seepage areas with different background interferences, so as to ensure that the trained sewage well leakage monitoring model can cope with the recognition of various scenarios. In addition, the training sample set can also be data enhanced to simulate random changes in the actual environment, such as incorporating factors such as lighting changes and motion blur, so that the sewage well leakage monitoring model can still maintain good detection performance and adaptability when facing the complex and changeable actual ship sewage well environment, thereby providing solid and reliable algorithm support for the accurate monitoring of ship sewage wells.

[0044] In some embodiments, each image frame of the real-time monitoring video is processed into a digital signal and then preprocessed, wherein the preprocessing means includes one or any combination of operations of normalizing the magnitude data of the digital signal, enhancing and scaling the image frame.

[0045] In some embodiments, the preprocessing technical means includes normalizing the magnitude data of the digital signal. The formula for normalizing the magnitude data of the preprocessed digital signal is as follows:

[0046] ;

[0047] Where y represents each digital signal, Indicates the minimum value of a digital signal. Indicates the maximum value of the digital signal. Represents a standardized digital signal.

[0048] In some embodiments, the pre-processing techniques also include operations of enhancing and scaling the image frames to enhance noise interference and highlight water seepage features.

[0049] It should be noted that the present invention has carried out multi-dimensional optimization and adaptability improvement on the sewage well leakage monitoring model used to monitor sewage well leakage. In terms of time-consuming optimization, the sewage well leakage monitoring model innovatively adopts the means of model lightweighting, successfully converts floating-point calculations into fixed-point calculations, greatly reduces the calculation complexity, reduces memory usage, and maintains the relative stability of model accuracy; in terms of feature extraction and seepage detection accuracy, with the help of the aggregation module, the local and global features of the seepage area of ​​the sewage well can be accurately extracted, providing the sewage well leakage monitoring model with richer and more accurate information input, greatly enhancing the accuracy of detection; at the same time, the attention mechanism is used to enable the sewage well leakage monitoring model to focus on potential seepage areas, effectively avoid the interference of irrelevant features, and further improve the accuracy of detection.

[0050] In some embodiments, the solution additionally includes step S3: when the water seepage reaches a set threshold or the water seepage area covers the target area, the alarm mechanism is triggered. In other words, if serious water seepage occurs or the water seepage area covers the target area, the alarm mechanism is triggered to promptly notify the ship's staff. Of course, the method of notifying the ship's staff in this solution includes but is not limited to sound, light, etc.

[0051] In some embodiments, the seepage area, seepage situation and corresponding real-time video stream are transmitted to the ship management and monitoring platform for centralized monitoring and stored locally, so that the staff of the management and monitoring platform can respond after receiving the relevant information, and it can also facilitate subsequent analysis and processing.

[0052] In some embodiments, the present solution can also provide suggested treatment solutions based on the analysis of the water seepage situation and the water seepage area, such as repairing damaged parts of sewage wells, adjusting ship navigation plans, etc., to help managers solve technical problems efficiently.

[0053] In some embodiments, an AI edge computing camera is used to obtain real-time monitoring video of the target area of ​​the ship, including the sewage well, and the trained sewage well leakage monitoring model is loaded into the edge computing node of the edge device. At this time, the real-time monitoring video obtained by the AI ​​edge computing camera is sent to the corresponding edge computing node for calculation. The advantage of this is that the AI ​​edge computing camera directly processes the acquired real-time monitoring video locally (edge ​​side), without the need to transmit a large amount of video data to the ship management monitoring platform for centralized calculation, which greatly reduces the delay of data transmission and can monitor and analyze the status of the sewage well in near real time. For example, once a leak occurs, the system can respond quickly and issue an alarm in time, allowing the crew to take measures at the first time to prevent the leak from further deteriorating and ensure the safety of the ship. In addition, network resources are relatively limited in the ship's operating environment. Transmitting all video data to the management center for calculation will take up a lot of network bandwidth. In edge computing mode, only the processed results (such as whether there is a leak, the severity of the leak, and other key information) and necessary real-time video streams will be transmitted to the ship management center, which greatly reduces the amount of data transmitted over the network, reduces the burden on the ship's network, and ensures that the network stability of other key systems of the ship (such as navigation, communication, etc.) is not affected.

[0054] The ship sewage well monitoring method proposed in this scheme can be applied to various daily operation scenarios, whether it is the oily sewage generated in the cargo hold when the cargo ship transports various types of goods, or the bilge water generated during the operation of the engine room of the passenger ship, the ship sewage well monitoring method can accurately monitor the sewage well in real time. Moreover, in view of the unique cabin layout and sewage generation characteristics of different types of ships, the ship sewage well monitoring method can flexibly adjust the monitoring parameters and strategies: for cargo ships, due to the wide variety of cargoes and the possible flammable and explosive risks, the ship sewage well monitoring method focuses on ensuring the explosion-proof safety of electrical equipment when monitoring the sewage well, and optimizes the data collection frequency according to the fluctuation law of sewage generation during cargo loading and unloading. For passenger ships with dense personnel: the ship sewage well monitoring method pays more attention to the potential impact of the sewage well status on passenger comfort and safety. Once it is detected that sewage may soak the cabin area or affect the normal operation of electromechanical equipment, an early warning is immediately triggered for timely processing to prevent the impact on passenger experience and navigation safety.

[0055] In emergency scenarios, such as when a ship encounters a collision or other accident, the sewage well, as a key part of the drainage system, the ship sewage well monitoring method will give priority to ensuring the smooth monitoring of sewage wells related to the drainage channel, increase the monitoring frequency and data transmission speed, and provide real-time feedback to the ship control center on the sewage well drainage status and possible blockage risks. In addition, the ship sewage well monitoring method will intelligently adjust the key monitoring areas according to the damage and tilt angle of the ship to fully ensure the stability of the ship. Whether it is routine navigation in calm waters, difficult progress in severe sea conditions, or in emergencies in the face of sudden accidents, the ship sewage well monitoring system can rely on its excellent adaptability to reliably complete the monitoring task of the sewage well and escort the safe and stable operation of the ship.

[0056] Embodiment 2

[0057] Based on the same idea, refer to Figure 3 , the present application also proposes a ship sewage well monitoring system, comprising:

[0058] Management and monitoring platform;

[0059] The edge devices are connected to the management and monitoring platform through the gateway. Each edge device is connected to the AI ​​edge computing camera, and each edge device is loaded with the trained sewage well leakage monitoring model.

[0060] The AI ​​edge computing camera obtains real-time monitoring video of the ship's target area including the sewage well and transmits it to the edge device. The sewage well leakage monitoring model in the edge device analyzes the real-time monitoring video output to obtain the seepage area and seepage situation. The sewage well leakage monitoring model uses a lightweight backbone network to extract the seepage features in each image frame, and uses an aggregation module to integrate seepage features of different scales to obtain the fusion features of each image frame. The timing module is used to enhance the expression of the fusion features of each image frame based on the fusion features of different image frames to obtain enhanced features. The attention mechanism is integrated into the decoding module to predict the seepage area and seepage situation based on the enhanced features.

[0061] In some specific embodiments, the number and location of each AI edge computing camera are displayed on the management and monitoring platform, and the real-time monitoring video acquired by the AI ​​edge computing camera is transmitted to the management and monitoring platform for unified viewing and management. In other words, users can easily view the real-time monitoring screen of the AI ​​edge computing cameras at all points and the real-time operating status of the AI ​​edge computing cameras. The operating status is subdivided into offline, normal, warning, and dangerous, so that users can see the overall situation at a glance. In addition, the management and monitoring platform also displays the cabin distribution details of the entire ship, and at the page level of each cabin, you can intuitively see the point screen and operating status of all AI edge computing cameras in the current cabin, providing a strong basis for accurate positioning problems.

[0062] In some embodiments, when the water seepage at a certain point of the corresponding AI edge computing camera reaches a set threshold or the water seepage area covers the target area, an alarm mechanism is triggered on the management and monitoring platform. Correspondingly, in a specific embodiment, an alarm button is set on the management and monitoring platform, and the alarm button is linked to the alarm device of the management personnel. Once the alarm mechanism is triggered, the alarm device is immediately linked to sound an alarm, ensuring that relevant personnel can know and deal with abnormal situations in a timely manner, which comprehensively improves the practicality and functionality of the ship sewage well monitoring system.

[0063] Of course, whenever the alarm mechanism is triggered, the current alarm mechanism is recorded and stored on the management and monitoring platform. At this time, the management and monitoring platform can see the real-time alarm text of all points, including the alarm records of all camera points on that day and the historical alarm records of all points, which is convenient for tracing back and analyzing the alarm situation.

[0064] The schematic diagram of the layout of the AI ​​edge computing camera of the ship sewage well monitoring system is as follows Figure 4 As shown, in this specific embodiment, an angle iron is fixedly installed in the ship cavity where the target area of ​​the ship is located, an AI edge computing camera is detachably installed on the installation angle iron through a first L-shaped bracket, and a floodlight is detachably installed on the installation angle iron through a second L-shaped bracket. This arrangement structure is stable and works in coordination, laying a solid hardware foundation for the stable operation of the entire ship sewage well monitoring system.

[0065] In some embodiments, the floodlight has the characteristics of large beam angle and high lumen. When the ambient light is insufficient, it can provide sufficient fill light for the AI ​​edge computing camera, ensuring the clarity and accuracy of the pictures taken by the AI ​​edge computing camera.

[0066] In some embodiments, the floodlight uses a high-brightness and low-power LED lamp, which is installed near the AI ​​edge computing camera. The installation position and angle of the floodlight are optimized to avoid glare caused by direct light to the AI ​​edge computing camera and reduce the impact of shadows on the image.

[0067] In other embodiments, the floodlight has an intelligent adjustment function, and cooperates with the light sensor to automatically adjust the light intensity according to the feedback of the light sensor under different lighting conditions, such as daytime or nighttime, and different cabin lighting environments, and achieves precise dimming by controlling the LED drive current to ensure that the image captured by the AI ​​edge computing camera always remains clear. Alternatively, the floodlight works in conjunction with the entire monitoring system, automatically adjusts when monitoring is started or the image quality is affected by light, and its status information is fed back to the system, and the system can alarm in case of failure.

[0068] In some embodiments, the outer shell of the floodlight is sealed with corrosion-resistant and waterproof materials, and the interior is filled with thermally conductive insulating materials to adapt to the harsh environment of the sewage well, which is humid and has corrosive gases.

[0069] In addition, since ships face complex and changeable electromagnetic environments, shaking and vibration, and different climatic conditions, in order to improve the quality of the image frames of the real-time monitoring video, the AI ​​edge computing camera of this solution has the following characteristics: 1. Modular design: The mainboard of the AI ​​edge computing camera is fixed, and different types of cameras can be replaced according to needs, such as 1080P, low light, night vision and other types. 2. The AI ​​edge computing camera has the function of automatically eliminating picture jitter and dynamic mode analysis, which can improve the accuracy and reliability of monitoring. In some embodiments, since the sewage well of the ship is in a dark and closed cabin, in order to ensure the accuracy of image recognition, it is often necessary to turn on the lighting for fill light.

[0070] In some embodiments, the AI ​​edge computing camera is equipped with a buzzer alarm function so that an alarm can be issued in time when an abnormal situation is detected.

[0071] In some embodiments, a waterproof junction box is installed in the ship cavity, and the network-electricity integrated line in the waterproof junction box is connected to the corresponding AI edge computing camera and floodlight to realize the connection between the network and the power. This solution leads the network-electricity integrated line from the ship's central control room to supply power to each lower cabin. The network-electricity integrated line adopts an armored design and its internal network transmission line is additionally equipped with a shielding layer. This design can better adapt to the special use environment of the cabin and effectively reduce the interference and influence of external factors.

[0072] like Figure 4 As shown, the AI ​​edge computing camera is set at the position corresponding to the sewage well to obtain the target area of ​​the ship including the sewage well.

[0073] like Figure 5 As shown, the AI ​​edge computing camera and the floodlight are placed on both sides of the mounting angle iron. Different connection positions are provided on the mounting angle iron. One side of the first L-shaped bracket is fixed on the connection position, and the AI ​​edge computing camera is installed on the other side of the first L-shaped bracket; similarly, one side of the second L-shaped bracket is fixed on the connection position, and the floodlight is installed on the other side of the second L-shaped bracket.

[0074] In a specific embodiment, the AI ​​edge computing camera is fixed to a high-toughness first L-shaped bracket by means of stainless steel hexagon socket bolts, spring washers, and flat washers, wherein the first L-shaped bracket is a stainless steel L-shaped round hole bracket; the floodlight is installed on the second L-shaped bracket by means of stainless steel hexagon socket bolts, spring washers, flat washers, and anti-loosening nuts, wherein the second L-shaped bracket is a stainless steel L-shaped long waist hole bracket, and the mounting angle iron, the first L-shaped bracket, and the second L-shaped bracket are precisely assembled by means of stainless steel hexagon socket bolts, spring washers, flat washers, and anti-loosening nuts.

[0075] In addition, in order to ensure the stability and reliability of the system during long-term operation of the ship, in the heat dissipation design of the hardware equipment, key components such as the AI ​​edge computing camera adopt a combination of efficient passive heat dissipation and active heat dissipation. Preferably, in a closed cabin environment, the shell of the AI ​​edge computing camera is made of metal with excellent thermal conductivity to increase the contact area with the air, which is conducive to the natural dissipation of heat. At the same time, it is equipped with a micro silent fan inside, which automatically adjusts the speed according to the real-time monitoring data of the temperature sensor. When the temperature rises to the set threshold, the fan accelerates to discharge the heat in time, ensuring that the equipment operates within the normal operating temperature range and avoiding performance degradation or failure due to overheating.

[0076] In some embodiments, when a network interruption occurs, the edge device immediately starts the local data cache function to continue to collect and temporarily store real-time monitoring videos and try to reconnect to the network. Once the network is restored, the cached data is automatically uploaded to the management and monitoring platform to ensure data integrity. If an abnormality occurs in the AI ​​edge computing camera, the management and monitoring platform will quickly switch to the backup AI edge computing camera or start a fault warning, and adjust the monitoring strategy at the same time, using the surrounding related sensor data for auxiliary analysis, and try to maintain the basic monitoring function of the sewage well. In addition, the system regularly performs self-diagnosis and optimization, checks the integrity of the software code, the connection status of the hardware equipment, etc., and promptly discovers and repairs potential problems. These measures greatly improve the practicality and reliability of the system in the ship's operating environment, and ensure that the sewage well monitoring work is carried out continuously and stably.

[0077] like Figure 6 As shown, Figure 6A specific example of the operation of the ship's sewage well monitoring system is provided. The status of each AI edge computing camera can be checked on the management and monitoring platform. If an AI edge computing camera has an abnormality, it will be reported and processed. If all AI edge computing cameras are normal, the sewage well monitoring information will be sent to the edge device. The edge device will monitor the sewage wells at the corresponding points and store each monitoring process. If a situation that meets the alarm mechanism occurs, it will be reported to the management and monitoring platform in a timely manner.

[0078] Embodiment 3

[0079] This embodiment also provides an electronic device, referring to Figure 7 , including a memory 404 and a processor 402, wherein the memory 404 stores a computer program, and the processor 402 is configured to run the computer program to execute the steps in any of the above-mentioned ship sewage well monitoring method embodiments.

[0080] Specifically, the processor 402 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application.

[0081] The memory 404 may include a large-capacity memory 404 for data or instructions. By way of example and not limitation, the memory 404 may be used to store or cache various data files that need to be processed and / or used for communication, as well as possible computer program instructions executed by the processor 402.

[0082] The processor 402 implements any one of the ship sewage well monitoring methods in the above embodiments by reading and executing computer program instructions stored in the memory 404 .

[0083] Optionally, the electronic device may further include a transmission device 406 and an input / output device 408 , wherein the transmission device 406 is connected to the processor 402 , and the input / output device 408 is connected to the processor 402 .

[0084] The transmission device 406 may be used to receive or send data via a network. Specific examples of the above network may include a wired or wireless network provided by a communication provider of the electronic device.

[0085] The input and output device 408 is used to input or output information. In this embodiment, the input information may be real-time monitoring video, etc., and the output information may be water seepage area and water seepage situation, etc.

[0086] Optionally, in this embodiment, the processor 402 may be configured to perform the following steps through a computer program:

[0087] S1: Obtain real-time surveillance video of the ship's target area including the sewage well;

[0088] S2: The time-series image frames of the real-time monitoring video are input into the trained sewage well leakage monitoring model to output the seepage area and seepage situation. The sewage well leakage monitoring model uses a lightweight backbone network to extract the seepage features in each image frame, and uses an aggregation module to integrate the seepage features of different scales to obtain the fusion features of each image frame. The time-series module is used to enhance the expression of the fusion features of each image frame based on the fusion features of different image frames to obtain enhanced features. The attention mechanism is integrated into the decoding module to predict the seepage area and seepage situation based on the enhanced features.

[0089] It should be noted that the specific examples in this embodiment can refer to the examples described in the above embodiments and optional implementation modes, and this embodiment will not be described in detail here.

[0090] In general, various embodiments may be implemented in hardware or dedicated circuits, software, logic, or any combination thereof. Some aspects of the invention may be implemented in hardware, while other aspects may be implemented in firmware or software that may be executed by a controller, microprocessor, or other computing device, but the invention is not limited thereto. Although various aspects of the invention may be shown and described as block diagrams, flow charts, or using some other graphical representation, it should be understood that, as non-limiting examples, the boxes, devices, systems, techniques, or methods described herein may be implemented in hardware, software, firmware, dedicated circuits or logic, general-purpose hardware or controllers or other computing devices, or some combination thereof.

[0091] Embodiments of the present invention can be implemented by computer software, which is executable by a data processor of a mobile device, such as in a processor entity, or implemented by hardware, or implemented by a combination of software and hardware. Computer software or programs (also referred to as program products) including software routines, applets and / or macros can be stored in any device readable data storage medium, and they include program instructions for performing specific tasks. Computer program products can include one or more computer executable components configured to perform embodiments when the program is running. One or more computer executable components can be at least one software code or a part thereof. In addition, at this point, it should be noted that any box of the logic flow in the figure can represent a program step, or interconnected logic circuits, boxes and functions, or a combination of program steps and logic circuits, boxes and functions. Software can be stored in physical media such as memory chips or storage blocks implemented in processors, magnetic media such as hard disks or floppy disks, and optical media such as, for example, DVDs and data variants thereof, CDs. Physical media are non-transient media.

[0092] Those skilled in the art should understand that the technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0093] The above embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A method for monitoring a ship sewage well, characterized in that: The following steps are involved: S1: Obtain real-time surveillance video of the ship's target area including the sewage well; S2: Input the time-series image frames of the real-time monitoring video into the trained sewage well leakage monitoring model to output the seepage area and seepage situation, wherein the sewage well leakage monitoring model includes a plurality of modules connected in sequence, the modules include a backbone network for extracting image frame features, an aggregation module for fusing multi-scale features, a time-series module for extracting time-series features, and a decoding module for generating the final monitoring results, the decoding module optimizes features in combination with the attention mechanism, wherein the backbone network uses a depth-separable convolution to extract seepage features, the aggregation module fuses seepage features of different scales to obtain fused features, the time-series module extracts and enhances the fused features of each image frame by analyzing the time-series differences of different image frames and using multi-branch convolution to obtain enhanced features, the decoding module optimizes the enhanced features to obtain optimized features and outputs the seepage area and seepage situation; The aggregation module fuses the water seepage features of adjacent layers output by the backbone network to obtain fused features; The timing module calculates the timing difference of the fusion features of adjacent time-series image frames, and uses different convolution branches to gradually extract the timing difference. After each convolution branch completes the convolution operation, a fusion operation is performed. After the fusion, the feature expression is gradually enhanced by addition and activation functions to obtain enhanced features. The attention mechanism is introduced into the decoding module to obtain optimized features by aggregating multi-level enhanced features layer by layer from top to bottom and optimizing the feature weights. The seepage area and seepage situation are output based on the optimized features.

2. The ship sewage well monitoring method according to claim 1, characterized in that: The method comprises step S3: when the water seepage reaches a set threshold or the water seepage area covers the target area, triggering an alarm mechanism.

3. The ship sewage well monitoring method according to claim 1, characterized in that: AI edge computing cameras are used to obtain real-time monitoring videos of target areas of the ship, including sewage wells, and the trained sewage well leakage monitoring model is loaded into the edge computing node of the edge device.

4. The ship sewage well monitoring method according to claim 1, characterized in that: Each image frame of the real-time monitoring video is processed into a digital signal and then preprocessed, wherein the preprocessing means includes one or any combination of operations of normalizing the magnitude data of the digital signal, enhancing and scaling the image frame.

5. A ship sewage well monitoring system, characterized in that: include: Management and monitoring platform; The edge devices are connected to the management and monitoring platform through the gateway. Each edge device is connected to the AI ​​edge computing camera, and each edge device is loaded with the trained sewage well leakage monitoring model. The AI ​​edge computing camera obtains real-time monitoring video of the target area of ​​the ship including the sewage well and transmits it to the edge device. The sewage well leakage monitoring model in the edge device analyzes the real-time monitoring video output of the seepage area and the seepage situation. The sewage well leakage monitoring model uses the backbone network to extract the seepage features in each image frame, and uses the aggregation module to integrate the seepage features of different scales to obtain the fusion features of each image frame. The time series module is used to enhance the expression of the fusion features of each image frame based on the fusion features of different image frames to obtain enhanced features. The attention mechanism is integrated into the decoding module to predict the seepage area and seepage situation based on the enhanced features. The aggregation module fuses the seepage features of adjacent levels output by the backbone network to obtain fusion features. The timing module calculates the timing difference of the fusion features of adjacent time-series image frames, and uses different convolution branches to gradually extract the timing difference. After each convolution branch completes the convolution operation, a fusion operation is performed. After the fusion, the feature expression is gradually enhanced by addition and activation functions to obtain enhanced features. The attention mechanism is introduced into the decoding module to obtain optimized features by aggregating multi-level enhanced features layer by layer from top to bottom and optimizing the feature weights. The seepage area and seepage situation are output based on the optimized features.

6. The ship sewage well monitoring system according to claim 5, characterized in that: include: When the water seepage at a certain point corresponding to the AI ​​edge computing camera reaches the set threshold or the water seepage area covers the target area, an alarm mechanism is triggered on the management and monitoring platform.

7. The ship sewage well monitoring system according to claim 5, characterized in that: An angle iron is fixedly installed in the ship cavity where the target area of ​​the ship is located, an AI edge computing camera is detachably installed on the installation angle iron through a first L-shaped bracket, and a floodlight is detachably installed on the installation angle iron through a second L-shaped bracket.

8. The ship sewage well monitoring system according to claim 5, characterized in that: A waterproof junction box is installed in the ship cavity, and the network-electricity integrated wires in the waterproof junction box are connected to the corresponding AI edge computing camera and floodlight to realize the connection between network and power.

9. The ship sewage well monitoring system according to claim 5, characterized in that: When a network interruption occurs, the edge device immediately starts the local data caching function to continue collecting and temporarily storing real-time surveillance videos and attempts to reconnect to the network. Once the network is restored, the cached data is automatically uploaded to the management and monitoring platform. If an abnormality occurs with the AI ​​edge computing camera, the management and monitoring platform quickly switches to the backup AI edge computing camera or initiates a fault warning.

Images