Intelligent management and control system and method for few-person marine engine room

Abstract

According to the intelligent management and control system and method for the few-person marine engine room, situation information such as current system operation, operation and maintenance personnel and equipment states of the engine room is collected based on technologies such as intelligent image collection, environment state collection, image processing and artificial intelligence recognition; intelligent recognition, processing and calculation functions such as face recognition, human body operation posture recognition, personnel trajectory tracking, cabin foreign matter recognition, temperature monitoring and alarming, important equipment states and cabin three-dimensional fusion scenes are integrated and fused, and intelligence and visualization of safety management and control in a cabin are achieved. And support is provided for safe operation of the cabin and staff reduction and efficiency improvement of the system.

Description

Technical Field

[0001] The present invention belongs to the technical field of ship intelligent management, and specifically relates to a system and method for intelligent management and control of a ship engine room with few personnel. Background Art

[0002] On ships, the engine room, as the cabin where the ship's main propulsion engine and auxiliary engine are located, has always been one of the most complex and important cabins on the ship. At the same time, the engine room has many main engines, auxiliary engines, supporting lubricating oil, water, steam and other systems, and the working environment is relatively harsh, requiring more personnel and high inspection frequency. At present, the intelligent technology of large ship engine rooms is developing rapidly, but the main research is still on the operation status monitoring of the main propulsion-related equipment and systems in the engine room, the main propulsion equipment fault diagnosis and health management technology research, and there is a lack of relevant research on the management of operating personnel in the engine room and the overall operation status of the engine room. At the same time, the engine room also has harsh environments such as high temperature, high humidity, high noise, high temperature steam leakage, and lubricating oil leakage. The use of advanced intelligent technology to improve the efficiency of engine room control and reduce the demand for engine room operation and maintenance personnel is the development trend of ship engine room technology. Summary of the invention

[0003] In order to solve the technical problems in the prior art, the present invention provides a less-manned ship engine room intelligent control system, including an infrared imaging information acquisition terminal, a scene image information acquisition terminal, a personnel image information acquisition terminal, an intelligent control cabinet, a cabin scene foreign body recognition controller module, a personnel control controller module, a cabin temperature recognition and alarm controller module, a cabin three-dimensional fusion scene controller module, a cabin intelligent control console, an information acquisition terminal signal cable, and a network cable;

[0004] Based on intelligent image acquisition, environmental status acquisition, image processing and artificial intelligence recognition technology, the system collects information on the current system operation, operation and maintenance personnel, equipment status, etc. in the cabin, and integrates and fuses face recognition, human work posture recognition, personnel trajectory tracking, cabin foreign object recognition, temperature monitoring and alarm, important equipment status and cabin three-dimensional fusion scene intelligent recognition, processing and calculation functions to realize the intelligence and visualization of safety management and control in the cabin.

[0005] Furthermore, the infrared imaging information acquisition terminal is mainly used to detect infrared radiation from the surface of equipment such as pipelines and pressure vessels in the cabin by using infrared thermal imaging technology, and convert the temperature distribution image of the target object into a video image through photoelectric conversion and signal processing. The acquisition terminal can penetrate harsh environments such as smoke and steam and has a high degree of detection.

[0006] The scene image information acquisition terminal is mainly used to collect cabin scene image information, and has the ability to perform global image monitoring of important areas in the cabin; it can flexibly select key details in the global picture, select a target in the panoramic picture at a single point or by frame, and drive the ball camera to monitor the target in detail;

[0007] The personnel image information acquisition terminal mainly uses industrial-grade image recognition camera units to collect video image information of the target object and its surroundings, encodes and compresses it into digital video signals, and then transmits it for analysis and processing.

[0008] Furthermore, the intelligent management and control cabinet is mainly used for distributed processing, calculation, and analysis of cabin temperature fields, equipment images, and personnel images, three-dimensional scene visualization, and outputs data to the cabin intelligent management and control console.

[0009] Furthermore, the cabin scene foreign object recognition controller module is used to process the cabin image information collected by the scene image information collection terminal, perform image detection, recognition and comparison, and identify the type of foreign objects in the scene and their location by comparing and analyzing with the reference scene picture.

[0010] Furthermore, the personnel management and control controller module is used to process the cabin personnel's facial, action, and position images collected by the personnel image information collection terminal, perform image detection, recognition, and comparison, and identify personnel identity, movement path, and working posture.

[0011] Furthermore, the cabin temperature recognition and alarm controller module is used to process the temperature information and temperature distribution image collected by the infrared imaging information collection terminal of personnel 1, identify the abnormal temperature parts, and issue early warnings for high-temperature steam leakage and high-temperature water leakage.

[0012] Furthermore, the cabin three-dimensional fusion scene controller module is used to visualize the alarms of specific equipment for personnel, foreign objects, and infrared detection in the cabin three-dimensional scene by means of three-dimensional visualization using the information processed by the cabin scene foreign object recognition controller module, the personnel management and control controller module, and the cabin temperature recognition and alarm controller module.

[0013] Furthermore, the cabin intelligent control console is used for cabin foreign object identification, personnel control, temperature identification, three-dimensional scene human-machine interface display, and information management.

[0014] The above-mentioned intelligent control system method for a less-manned ship engine room includes:

[0015] The information collection layer collects the temperature and image information of the ship's engine room through the system information collection terminal, and provides information resources for subsequent controller modules;

[0016] The image recognition, calculation, and information processing layer, through the training of the image recognition model, performs image recognition, calculation, and processing on the front-end collected images in series or parallel mode, and sends the results to the display, alarm, and management levels through the event mechanism or directly sends the pictures and videos of the recognition results in the form of files;

[0017] Display, alarm, and management layers: Based on the various recognition and calculation results pushed by the image recognition, calculation, and information processing layers, foreign object recognition, personnel control, temperature recognition, human-machine interface display of three-dimensional scenes, and information management are performed in the cabin;

[0018] The iterative optimization layer collects new data samples from the cabin, processes data, optimizes models, and optimizes parameter training, thereby optimizing the performance of the entire intelligent management and control system.

[0019] Furthermore, in the cabin image recognition model method, the personnel, foreign objects and temperature field in the cabin are identified through images, and the convolutional neural network is used to train and extract features from the personnel, foreign objects and temperature data sets to increase the network's robustness to image feature loss and obstacles. Then the matching degree between adjacent image frames is calculated, and a tracking code is assigned to each object.

[0020] Compared with the prior art, the present invention has the following beneficial effects:

[0021] The present invention is based on intelligent image acquisition, environmental status acquisition, image processing and artificial intelligence recognition technologies to collect information on the current system operation, operation and maintenance personnel, equipment status and other situation information in the cabin, and integrates and fuses intelligent recognition, processing and calculation functions such as face recognition, human work posture recognition, personnel trajectory tracking, cabin foreign object recognition, temperature monitoring and alarm, important equipment status and cabin three-dimensional fusion scene, to achieve intelligent and visualized safety management and control in the cabin, and provide support for achieving safe operation of the cabin and reducing staff and increasing efficiency of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 This is the composition and principle of the intelligent control system for ship engine rooms with few people in the present invention;

[0023] Figure 2 This is a method for intelligent control system of ship engine room with few personnel in the present invention;

[0024] Figure 3 It is a cabin target recognition and tracking method model constructed by the present invention. DETAILED DESCRIPTION

[0025] In order to enable those skilled in the art to better understand the technical solution of the present invention, the present invention is further described below in conjunction with the accompanying drawings and specific implementation methods.

[0026] The present application proposes a system and method for intelligent control of ship engine rooms with few personnel. The system collects the current situation information of the engine room through information collection equipment, integrates and fuses functions such as face recognition, human posture recognition and personnel trajectory tracking, cabin scene foreign object recognition, cabin temperature monitoring and alarm, and three-dimensional fusion scene virtual and real fusion display, so as to realize the intelligent and visual control of the engine room safety.

[0027] Basic working principle: Based on intelligent image acquisition, environmental status acquisition, image processing and artificial intelligence recognition technologies, the system collects information on the current system operation, operation and maintenance personnel, equipment status and other situation information in the cabin, integrates and fuses intelligent recognition, processing and computing functions such as face recognition, human posture recognition, personnel trajectory tracking, cabin foreign body recognition, temperature monitoring and alarm, important equipment status and cabin three-dimensional fusion scene, and realizes intelligent and visualized safety management and control in the cabin. It provides support for the safe operation of the cabin and the reduction of system staff and increase of efficiency.

[0028] A less-manned ship engine room intelligent control system mainly consists of: 1 infrared imaging information acquisition terminal, 2 scene image information acquisition terminal, 3 personnel image information acquisition terminal, 4 intelligent control cabinet, 5 cabin scene foreign body recognition controller module, 6 personnel control controller module, 7 cabin temperature recognition and alarm controller module, 8 cabin three-dimensional fusion scene controller module, 9 cabin intelligent control console, 10 information acquisition terminal signal cable, 11 network cable. Figure 1 shown.

[0029] Among them, 1 infrared imaging information acquisition terminal is mainly used to detect infrared radiation from the surface of equipment such as pipelines and pressure vessels in the cabin by infrared thermal imaging technology, and convert the temperature distribution image of the target object into a video image through photoelectric conversion and signal processing. The acquisition terminal can penetrate harsh environments such as smoke and steam and has a high degree of detection.

[0030] 2 The scene image information acquisition terminal is mainly used to collect cabin scene image information. It has the ability to monitor the image of important areas in the cabin globally; it can flexibly select key details in the global picture, select a target in the panoramic picture by single point or frame, and drive the ball camera to monitor the target in detail.

[0031] 3. The personnel image information acquisition terminal mainly uses industrial-grade image recognition camera units to collect video image information of the target object and its surroundings, encodes and compresses it into digital video signals, and then transmits, analyzes and processes them.

[0032] 4 Intelligent management and control cabinets are mainly used for distributed processing, calculation, analysis of cabin temperature field, equipment images, and personnel images, three-dimensional scene visualization processing, and output data to 9 cabin intelligent management and control consoles.

[0033] 5 The cabin scene foreign object recognition controller module is mainly used to process the cabin image information collected by the 2 scene image information acquisition terminal, perform image detection, recognition and comparison, and identify the type and location of foreign objects in the scene by comparing and analyzing with the reference scene picture.

[0034] 6 The personnel management and control controller module is mainly used to process the facial, action and position images of cabin personnel collected by the personnel image information collection terminal, perform image detection, recognition and comparison, and identify personnel identity, movement path and working posture.

[0035] 7 The cabin temperature recognition and alarm controller module is mainly used to process the temperature information and temperature distribution images collected by the infrared imaging information acquisition terminal of personnel 1, identify abnormal temperature locations, and issue early warnings for high-temperature steam leakage, high-temperature water leakage, etc.

[0036] The cabin three-dimensional fusion scene controller module 8 is mainly used to visualize the information processed by the cabin scene foreign object recognition controller module 5, the personnel management and control controller module 6, and the cabin temperature recognition and alarm controller module 7 in the cabin three-dimensional scene by means of three-dimensional visualization, such as personnel, foreign objects, and infrared detection specific equipment alarms.

[0037] 9The intelligent cabin control console is used for cabin foreign object identification, personnel control, temperature identification, three-dimensional scene human-machine interface display, and information management.

[0038] Method of intelligent control system for less-manned ship engine room:

[0039] How to use the intelligent control system of the ship engine room with few people Figure 2 The method is divided into information collection layer, image recognition, calculation, information processing layer, display, alarm, management layer, and iterative optimization layer.

[0040] 1) The information collection layer collects the temperature and image information of the ship's engine room through the system information collection terminal, and provides information resources for subsequent controller modules.

[0041] 2) The image recognition, calculation, and information processing layer mainly trains the image recognition model, performs image recognition, calculation, and processing on the front-end collected images in series or parallel mode, and sends the results to the display, alarm, and management levels through an event mechanism or directly sends the pictures and videos of the recognition results in the form of files.

[0042] 3) The display, alarm, and management layers mainly perform cabin foreign object recognition, personnel control, temperature recognition, three-dimensional scene human-machine interface display, and information management based on the various recognition and calculation results pushed by the image recognition, calculation, and information processing layers.

[0043] 4) The iterative optimization layer collects new data samples from the cabin, processes data, optimizes models, and optimizes parameter training, thereby optimizing the performance of the entire intelligent management and control system.

[0044] In the cabin image recognition model method, the personnel, foreign objects and temperature field in the cabin are identified through images, and the convolutional neural network is used to train and extract features from the personnel, foreign objects and temperature data sets to increase the network's robustness to image feature loss and obstacles. Then the matching degree between adjacent image frames is calculated, and a tracking code is assigned to each object.

[0045] A model of cabin target recognition and tracking method combining deep learning-based target detection algorithm and multi-target tracking algorithm is constructed to realize personnel, foreign objects and temperature point recognition, personnel trajectory tracking and reproduction, and track multiple targets in continuous video images of single and multiple information collection terminal cameras. When a cabin target is detected, the tracking coding trajectory is continuously recorded and the coordinate information is sent to the cabin 3D fusion scene to realize the real-time position and trajectory display of the target.

[0046] Due to the complex environment of the cabin, with many pipes and equipment, the moving personnel detection target may be blocked by pipes and equipment. Based on the traditional multi-target tracking algorithm, the constructed recognition and tracking method model adds target appearance feature information to reduce the tracking code switching phenomenon when the target is blocked by pipes and equipment. At the same time, the cascade matching method is introduced to assign a tracker to each detector to prevent the increase of Kalman filter prediction uncertainty due to long-term occlusion. The constructed cabin target recognition and tracking method model is shown in Figure 3.

[0047] In the cabin target recognition and tracking method model, firstly, all the targets detected in the first frame of the video sequence are assigned tracking coding information and initialized as independent new trajectories. At the same time, the Kalman filter and trajectory state judger are initialized, and the principle of state determination is: when the trajectory is successfully matched for multiple consecutive frames, it is a confirmed state, otherwise it is considered an unconfirmed state. Update the Kalman filter to determine whether the trajectory is in a confirmed state. When the trajectory is in an unconfirmed state, the target trajectory is cascade matched with the detection execution, and the matching result is output: when the target trajectory is matched with the detection, Kalman prediction is executed to predict the position of the new trajectory; when the target detection is not matched with the trajectory, it is initialized as a new trajectory and a new target is assigned. Secondly, when the trajectory is in a confirmed state, the target trajectory is cascade matched with the new detection, and the cascade matching also responds to the output result. Finally, the loop is executed until the end of the video frame.

[0048] The model of the cabin target recognition and tracking method adds a state confirmation mechanism for the trajectory in the loop process. When the trajectory is matched to the detected trajectory for multiple consecutive frames, it is converted to the confirmation state, which is conducive to reducing the object coding switching phenomenon caused by misjudgment. At the same time, in the trajectory deletion mechanism, the existing trajectory is deleted when the trajectory has not been matched for multiple consecutive frames, which has a high advantage in dealing with target occlusion. Finally, based on the identified coordinate position information and object coding information of the operation and maintenance personnel, the trajectory and real-time position of the operation and maintenance personnel are reproduced in the three-dimensional fusion scene.

[0049] The main functions of a less-manned ship engine room intelligent management and control system:

[0050] First, the cabin foreign object recognition function: It can identify and monitor foreign objects such as tools left behind by personnel and equipment parts dropped by vibration in the collection screen of the 2-scene image information collection terminal in the cabin. Including: 1) Real-time preview of cabin video images, supporting multiple layouts and multiple screens of the 2-scene image information collection terminal camera monitoring screen display in the cabin; 2) Cabin benchmark recognition screen configuration, supporting the monitoring screen or the foreign object recognition benchmark screen configuration, filtering configuration, image preprocessing configuration, confidence configuration, etc. in the specific area of ​​the screen, and performing cabin foreign object recognition and monitoring based on the benchmark recognition screen configuration; 3) Cabin foreign object recognition, by comparing with the benchmark screen, intelligently identifying foreign objects in the monitored screen, and outputting the location and contour information of the foreign objects according to the synchronous coordinate system, and sending it to the cabin three-dimensional scene, the location and type of foreign objects are reproduced and prompted in the cabin three-dimensional scene.

[0051] Second, the function of identifying and controlling people entering the cabin: mainly realizing functions such as cabin inspection, face recognition of operation and maintenance personnel, personnel trajectory tracking, human body specific posture recognition, and historical trajectory statistical analysis. Including: 1) Face recognition, through image recognition technology, the facial images of inspection and operation and maintenance personnel appearing in the surveillance screen are recognized, the identity information of the personnel is identified, and it is determined whether they are authorized to enter. When unauthorized personnel are detected to enter, an alarm will be issued at the cabin intelligent control console; 2) Personnel trajectory tracking, mainly realizes the real-time tracking of personnel between multiple lenses in the surveillance screen, and stores the real-time position information, and sends it to the three-dimensional fusion scene at the same time, and reproduces the position of personnel in the three-dimensional scene of the cabin; 3) Human posture recognition, mainly realizes the recognition of specific postures of operation and maintenance personnel in the cabin, such as abnormal postures such as falling, falling, being squeezed, and smoking in violation of regulations, and sends it to the three-dimensional fusion scene at the same time, and reproduces the posture of personnel in the three-dimensional scene of the cabin and alarms; 4) Historical trajectory statistical analysis, according to actual needs, statistical analysis of the historical trajectories of inspection and operation and maintenance personnel is carried out, such as historical activity trajectories of personnel, duration of special positions, intrusions into dangerous areas, etc., for management personnel decision-making, accident tracing analysis, etc.

[0052] Third, the cabin temperature field recognition and alarm function: the temperature information is collected by 1 infrared imaging information collection terminal to realize the high and low temperature values ​​and position display of the cabin infrared thermal imaging monitoring screen, and the temperature display is implemented at any position in the cabin. At the same time, the temperature field monitoring and alarm are carried out for specific equipment or pipeline areas where there are leaks of steam, high-temperature water, lubricating oil, fuel, etc.

[0053] Fourth, the cabin 3D fusion scene display function: by constructing a model of the cabin and equipment in the cabin, and then integrating and rendering the airport 3D scene according to the information collected by the information collection terminal and the calculation results of the controller, the cabin 3D fusion scene status display, information display, alarm, etc. are realized. Including: 1) Cabin 3D scene status display, realize the display of the entire cabin 3D fusion scene environment status, which can be displayed from multiple angles such as roaming, global perspective, local perspective, positioning to equipment, pipelines, etc.; personnel position, trajectory and foreign object reproduction, inspection in the cabin 3D fusion scene, operation and maintenance personnel in the cabin position, activity trajectory, working posture reproduction, cabin foreign object identification results reproduction; 2) Alarm display, in the cabin 3D fusion scene, according to the temperature recognition, foreign object recognition, personnel working posture recognition results, the corresponding abnormal position in the cabin 3D fusion scene is displayed, and the alarm events of high-temperature steam, water, oil leakage, personnel falling, falling, being squeezed, and asking for help are identified. Alarm events such as personnel entering dangerous areas, missing tools, and equipment parts falling off.

[0054] (1) Arrangement of the intelligent control system acquisition terminal: 1 infrared imaging information acquisition terminal, 2 scene image information acquisition terminal, 3 personnel image information acquisition terminal are arranged at appropriate locations in the ship's engine room. During the arrangement process, considering the problem that the acquisition terminal may be blocked, the acquisition terminal can be set at different locations to ensure the coverage of the engine room by the acquisition terminal.

[0055] (2) 4 intelligent control cabinets are installed in the important operation cabins of the ship. All information collection terminal lines are connected to the 4 intelligent control cabinets. The signals are respectively distributed to the 5 cabin scene foreign object recognition controller module, the 6 personnel control controller module, the 7 cabin temperature recognition and alarm controller module, and the 8 cabin three-dimensional fusion scene display controller module for calculation, analysis and processing.

[0056] (3) Install the 9-cabin intelligent control console and connect each device with signal cables and network cables.

[0057] (4) System wiring connection is completed; interface and function debugging is carried out and debugging is completed.

[0058] (5) Usage process: The information collection, identification, calculation and analysis of the engine room are all automated processes. 1 Infrared imaging information collection terminal, 2 Scene image information collection terminal, 3 Personnel image information collection terminal, etc. continuously collect information such as personnel, equipment, temperature, etc. in the engine room, and transmit it to 4 Intelligent control cabinet through signal lines, and distribute it to 5 Engine room scene foreign body identification controller module, 6 Personnel control controller module, 7 Engine room temperature identification and alarm controller module, 8 Engine room three-dimensional fusion scene display controller module for calculation, analysis and processing; The results are transmitted to 9 Operator status management and decision terminal through the network, and the management personnel can display, alarm and manage the engine room through 9 Engine room intelligent control console human-machine interface. Realize the monitoring of the current personnel status information of the ship's engine room, scene foreign body identification, temperature monitoring and three-dimensional fusion scene display, etc., and realize the intelligent and visual control of personnel and status in the cabin.

[0059] In summary, the above are only preferred embodiments of the present invention, and the protection scope of the present invention is not limited thereto. All equivalent changes and modifications made according to the patent scope of the present invention and the contents of the specification are within the scope covered by the patent of the present invention.

Claims

1. An intelligent control system for ship engine rooms with few personnel, characterized by: Including infrared imaging information collection terminal, scene image information collection terminal, personnel image information collection terminal, intelligent control cabinet, cabin scene foreign body recognition controller module, personnel control controller module, cabin temperature recognition and alarm controller module, cabin three-dimensional fusion scene controller module, cabin intelligent control console, information collection terminal signal cable, network cable; Based on intelligent image acquisition, environmental status acquisition, image processing and artificial intelligence recognition technology, the system collects information on the current system operation, operation and maintenance personnel, equipment status, etc. in the cabin, and integrates and fuses face recognition, human work posture recognition, personnel trajectory tracking, cabin foreign object recognition, temperature monitoring and alarm, important equipment status and cabin three-dimensional fusion scene intelligent recognition, processing and calculation functions to realize the intelligence and visualization of safety management and control in the cabin.

2. The system according to claim 1, characterized in that: The infrared imaging information acquisition terminal is mainly used to detect infrared radiation from the surface of equipment such as pipelines and pressure vessels in the cabin using infrared thermal imaging technology, and convert the temperature distribution image of the target object into a video image through photoelectric conversion and signal processing. The acquisition terminal can penetrate harsh environments such as smoke and steam and has a high degree of detection. The scene image information acquisition terminal is mainly used to collect cabin scene image information, and has the ability to perform global image monitoring of important areas in the cabin; it can flexibly select key details in the global picture, select a target in the panoramic picture at a single point or by frame, and drive the ball camera to monitor the target in detail; The personnel image information acquisition terminal mainly uses industrial-grade image recognition camera units to collect video image information of the target object and its surroundings, encodes and compresses it into digital video signals, and then transmits it for analysis and processing.

3. The system according to claim 1, characterized in that: The intelligent management and control cabinet is mainly used for distributed processing, calculation, and analysis of cabin temperature fields, equipment images, and personnel images, as well as three-dimensional scene visualization processing, and outputs data to the cabin intelligent management and control console.

4. The system according to claim 1, characterized in that: The cabin scene foreign object recognition controller module is used to process the cabin image information collected by the scene image information acquisition terminal, perform image detection, recognition and comparison, and identify the type and location of foreign objects in the scene by comparing and analyzing with the reference scene image.

5. The system according to claim 1, characterized in that: The personnel management and control controller module is used to process the facial, action and position images of cabin personnel collected by the personnel image information collection terminal, perform image detection, recognition and comparison, and identify personnel identity, movement path and working posture.

6. The system according to claim 1, characterized in that: The cabin temperature recognition and alarm controller module is used to process the temperature information and temperature distribution images collected by the infrared imaging information acquisition terminal of personnel 1, identify abnormal temperature locations, and issue early warnings for high-temperature steam leakage and high-temperature water leakage.

7. The system of claim 1, wherein: The cabin three-dimensional fusion scene controller module is used to visualize the alarms of specific equipment for personnel, foreign objects, and infrared detection in the cabin three-dimensional scene by means of three-dimensional visualization, using the information processed by the cabin scene foreign object recognition controller module, personnel management and control controller module, and cabin temperature recognition and alarm controller module.

8. The system of claim 1, wherein: The intelligent cabin control console is used for cabin foreign object identification, personnel control, temperature identification, three-dimensional scene human-machine interface display, and information management.

9. The method for intelligent control of a ship engine room with few personnel as claimed in any one of claims 1 to 8, characterized in that: include: The information collection layer collects the temperature and image information of the ship's engine room through the system information collection terminal, and provides information resources for subsequent controller modules; The image recognition, calculation, and information processing layer, through the training of the image recognition model, performs image recognition, calculation, and processing on the front-end collected images in series or parallel mode, and sends the results to the display, alarm, and management levels through the event mechanism or directly sends the pictures and videos of the recognition results in the form of files; Display, alarm, and management layers: Based on the various recognition and calculation results pushed by the image recognition, calculation, and information processing layers, foreign object recognition, personnel control, temperature recognition, human-machine interface display of three-dimensional scenes, and information management are performed in the cabin; The iterative optimization layer collects new data samples from the cabin, processes data, optimizes models, and optimizes parameter training, thereby optimizing the performance of the entire intelligent management and control system.

10. The method according to claim 9, characterized in that: In the cabin image recognition model method, people, foreign objects and temperature fields in the cabin are identified through images, and a convolutional neural network is used to train and extract features from the people, foreign objects and temperature data sets to increase the network's robustness to image feature loss and obstacles. The matching degree between adjacent image frames is then calculated and a tracking code is assigned to each object.

Images