Port tractor early warning system based on multi-source fusion and control method applying same

Abstract

The invention provides a port tractor early warning system based on multi-source fusion, and the system comprises a vehicle-mounted subsystem and a cloud server. The vehicle-mounted subsystem comprises a Beidou satellite positioning unit, a UWB ultra wide band positioning unit, an RTK real-time dynamic difference unit, a time synchronization module, a data acquisition device, a DMS camera, an HOD camera, a cockpit acousto-optic device, a steering wheel vibration module and a control device. The control device is in physical medium / communication link connection with the Beidou satellite positioning unit, the UWB positioning unit, the RTK real-time dynamic difference unit, the time synchronization module, the data acquisition device, the cockpit acousto-optic device and the steering wheel vibration module. The invention also provides a control method applying the early warning system, and the method comprises the steps: receiving the fusion positioning data in real time, generating the real-time speed of the trailer, and generating a visual recognition obstacle distance through an AI visual analysis model; and automatically triggering risk level judgment. According to the invention, the tractor in-port positioning precision and safety early warning comprehensiveness are improved.

Description

Technical Field

[0001] This invention relates to the field of automatic driving and safety control technology for port tractors, specifically to a port tractor early warning system based on multi-source fusion and a control method using the same. Background Technology

[0002] Current port tractor safety management relies on solutions such as single BeiDou positioning early warning, "manual inspection + video surveillance," and general commercial vehicle early warning systems. These solutions have technical shortcomings and limitations in scenario adaptation, for example:

[0003] Disadvantage 1: Insufficient vehicle positioning coverage and accuracy; single Beidou positioning is prone to failure in areas with dense metal or signal obstruction, such as container yards and under bridges, resulting in positioning blind spots; and the error of ordinary positioning technology is mostly between 1 and 3 meters, which cannot meet the "centimeter-level" boundary crossing warning requirements in scenarios such as shoreline boundaries and narrow passages, and is prone to misjudgment or missed judgment, making it difficult to accurately determine whether port vehicles are approaching dangerous areas.

[0004] Disadvantage 2: Vehicle warning response is delayed and has only one dimension; the "manual inspection + video monitoring" solution relies on 24-hour shifts of manpower, which is affected by fatigue and limited perspective, resulting in a high rate of missed detection. It takes more than 30 seconds from the discovery of a risk to the reminder to the driver, which cannot prevent sudden accidents; at the same time, most existing warning systems only monitor the vehicle's location and do not link the driver's status (such as fatigue or distraction) or surrounding environmental data, resulting in a single dimension of warning and an inability to comprehensively prevent and control risks.

[0005] Disadvantage 3: Poor vehicle system compatibility and scenario adaptation; the general commercial vehicle warning system follows the logic of road commercial vehicles, does not support UWB indoor positioning, and cannot cover areas with blocked signals in the port; the electronic fence parameters have not been optimized for the characteristics of "low speed heavy load and shoreline operation" of tractor vehicles, which easily triggers warnings falsely; and it cannot be linked with the port dispatch system, and the warning information is only pushed to the cockpit, so the management center cannot monitor the overall risk in real time.

[0006] Disadvantage 4: Weak data collaboration and traceability capabilities; existing solutions mostly lack a unified data processing and cloud collaboration mechanism, making it impossible to collect vehicle driving data (speed, steering, braking) and driver status data in real time and form correlation analysis. After an accident, it is difficult to trace the vehicle trajectory and the cause of the risk trigger, which is not conducive to subsequent safety management optimization.

[0007] In summary, existing technologies cannot meet the requirements of "smart ports" for high-precision, all-weather, blind-spot-free, and multi-dimensional safety control of port tractors. There is an urgent need for an integrated safety early warning system that integrates multi-source positioning, dynamic early warning, intelligent intervention, and multi-vehicle adaptation. Summary of the Invention

[0008] In view of this, it is necessary to address the shortcomings and deficiencies of existing technologies by proposing a port tractor early warning system based on multi-source fusion and a control method for its application. This would simultaneously improve the blind spot positioning accuracy, the comprehensiveness of driving boundary safety warnings, and the adaptability to multiple driving scenarios of port tractors while driving in the port, thereby improving the efficiency of port safety management and scheduling.

[0009] To achieve the above objectives, the present invention adopts the following technical solution:

[0010] This invention proposes a port tractor early warning system based on multi-source fusion, comprising an onboard subsystem and a cloud server. The onboard subsystem interacts with the cloud server for data exchange. The onboard subsystem is installed on the port tractor. The cloud server is used for data storage, remote monitoring, statistical analysis, and emergency response. The onboard subsystem includes a BeiDou satellite positioning unit, a UWB ultra-wideband positioning unit, an RTK real-time dynamic differential unit, a time synchronization module, a data acquisition device, a DMS camera, a HOD camera, cockpit audio-visual equipment, a steering wheel vibration module, and a control device. The control device has a vehicle braking linkage interface and is physically / communicationally connected to the BeiDou satellite positioning unit, UWB ultra-wideband positioning unit, RTK real-time dynamic differential unit, time synchronization module, data acquisition device, cockpit audio-visual equipment, and steering wheel vibration module. The data acquisition device is also communicatively connected to the DMS camera and the HOD camera.

[0011] The BeiDou satellite positioning unit is used to obtain the real-time location information of the port tractor.

[0012] UWB (Ultra-Wideband) positioning units are used in ports for fine-grained positioning using wireless carrier communication, thereby reducing the risk of positioning failures caused by satellite signal blockage or weak signals for port tractors.

[0013] The RTK real-time dynamic differential unit is used to receive differential signals sent by the port base station, thereby facilitating real-time correction of BeiDou satellite positioning errors;

[0014] The data acquisition device is used to collect the first data packet of the on-board management system of the port tractor in real time. The first data packet includes operating status data, environmental perception data and / or positioning data, and transmits the collected data synchronously to the control device and the cloud server.

[0015] The DMS camera is used to monitor the head and head organs of the tractor driver in real time and to identify abnormal situations of the driver through AI algorithms.

[0016] The HOD camera is used to monitor whether the driver's hands are on the steering wheel, and also to adapt to the frequent turning operation rules of port tractors to identify abnormal situations of the driver's hands operating the steering wheel.

[0017] The time synchronization module is used to unify the time reference of BeiDou satellite positioning units, UWB ultra-wideband positioning units, data acquisition devices, DMS cameras, and HOD cameras;

[0018] The steering wheel vibration module is used to provide tactile warnings through steering wheel vibration;

[0019] The control device is used to receive and integrate positioning data from the Beidou satellite positioning unit, UWB ultra-wideband positioning unit, RTK real-time dynamic differential unit, vehicle status data from the data acquisition device, driver status data from the DMS camera, and driver status data from the HOD camera, based on the unified time reference of the time synchronization module and the first data packet, and generate hierarchical control commands according to the preset risk judgment logic.

[0020] The cockpit audio-visual equipment is used to issue graded warning signals to the driver when the control device detects abnormal risks in traffic conditions, vehicle movement, and driver driving.

[0021] Furthermore, the control device includes a data processing unit and an application processing unit;

[0022] The data processing unit is used for collaborative fusion of positioning data, vehicle data acquisition, and driver status data acquisition; the operational status data includes the speed, steering, and braking status data of the port tractor; the positioning data includes BeiDou satellite positioning data, UWB data, and RTK data;

[0023] The application processing unit is used for data processing and decision-making, hierarchical early warning and intervention processing, and also for risk analysis and early warning execution through a multi-source fusion dynamic risk algorithm model; the multi-source fusion dynamic risk algorithm model is used to construct a dynamic risk assessment model based on multi-source fusion data, and the dynamic risk assessment model is used to achieve three-dimensional risk classification about distance, speed and driver status.

[0024] Furthermore, the multi-source fusion dynamic risk algorithm model includes a first algorithm, a second algorithm, and a third algorithm that are executed sequentially or operated collaboratively; the first algorithm, the second algorithm, and the third algorithm correspond to the risk perception stage, the decision output stage, and the control execution stage of the port tractor safety early warning process, respectively.

[0025] Furthermore, the first algorithm is a multi-source positioning data fusion algorithm used to overcome the accuracy deviation defects of a single positioning method in complex port scenarios; the second algorithm is a dynamic risk level assessment algorithm used to realize three-dimensional risk classification and provide decision-making basis for matching warning intensity and braking level; the third algorithm is a graded braking adaptive control algorithm used to output adaptive braking commands based on dynamic risk values ​​and real-time vehicle status.

[0026] Furthermore, the port tractor warning system also includes several user terminals that interact with the cloud server, such as cockpit controllers, mobile apps, or management center terminals.

[0027] This invention further proposes a control method for a port tractor early warning system based on multi-source fusion as described in any of the preceding claims, comprising the following steps:

[0028] Step 1: Receive fused positioning data from BeiDou satellite positioning unit, UWB ultra-wideband positioning unit and RTK real-time dynamic differential unit in real time, and generate the trailer's real-time speed and obstacle distances for visual recognition through AI visual analysis model.

[0029] Step two: Based on the preset spatial coordinates of the electronic fence used to define the shoreline boundary and hazardous work area, the risk level determination is automatically triggered.

[0030] Furthermore, in step two, the process of triggering the risk level determination includes:

[0031] Staged braking execution and dynamic pressure regulation;

[0032] Dynamic correction and precision compensation are performed during the braking process;

[0033] The locking and emergency linkage mechanism is executed according to the state after braking; the locking and emergency linkage mechanism is released when the mechanism for manual release and system reset is activated.

[0034] This invention further proposes a port tractor early warning system based on multi-source fusion, including a port tractor's onboard management system, an onboard execution subsystem controlled by the port tractor's onboard management system, a positioning subsystem, a data acquisition and processing unit, an application processing unit, and a port cloud service layer. The port cloud service layer is used for data storage, remote monitoring, statistical analysis, and emergency response. The port tractor early warning system also includes a time synchronization unit, a DMS camera, and a HOD camera. The positioning subsystem, DMS camera, and HOD camera are installed on the port tractor. The positioning subsystem includes a Beidou satellite positioning unit, a UWB ultra-wideband positioning unit, and an RTK real-time dynamic differential unit. The onboard execution subsystem includes a vehicle braking device, a cockpit audio-visual equipment, and a steering wheel vibration module. The data acquisition and processing unit is connected to the Beidou satellite positioning unit, UWB ultra-wideband positioning unit, RTK real-time dynamic differential unit, time synchronization unit, DMS camera, HOD camera, port tractor's onboard management system, and port cloud service layer via physical media / communication links.

[0035] The BeiDou satellite positioning unit is used to obtain the real-time location information of the port tractor.

[0036] UWB ultra-wideband positioning units are used in ports for fine positioning using wireless carrier communication, thereby reducing the risk of positioning failures caused by satellite signal blockage or weak signals for port tractors.

[0037] The RTK real-time dynamic differential unit is used to receive differential signals sent by the port base station, thereby facilitating real-time correction of BeiDou satellite positioning errors;

[0038] The DMS camera is used to monitor the head and head organs of the tractor driver in real time and to identify abnormal situations of the driver through AI algorithms.

[0039] The HOD camera is used to monitor whether the driver's hands are on the steering wheel, and also to adapt to the frequent turning operation rules of port tractors to identify abnormal situations of the driver's hands operating the steering wheel.

[0040] The time synchronization unit is used to unify the time reference of the BeiDou satellite positioning unit, UWB ultra-wideband positioning unit, data acquisition and processing unit, DMS camera, and HOD camera.

[0041] The steering wheel vibration module is used to provide tactile warnings through steering wheel vibration;

[0042] The data acquisition and processing unit is used to collect driver status data and the operating status data, environmental perception data and / or positioning data of the port tractor's on-board management system in real time, and synchronously transmit the relevant collected data to the application processing unit and the port cloud service layer; the operating status data includes the port tractor's speed, steering and braking status data; the positioning data includes Beidou satellite positioning data, UWB data and RTK data;

[0043] The application processing unit is used for data processing and decision-making, hierarchical early warning and intervention processing, and generates hierarchical control instructions based on the preset risk judgment logic. That is, risk analysis and early warning execution are realized through a multi-source fusion dynamic risk algorithm model and the vehicle management system.

[0044] The cockpit audio-visual equipment is used to issue graded warning signals to the driver when the application processing unit detects abnormal risks in traffic conditions, vehicle movement, and driver driving.

[0045] Furthermore, the port tractor early warning system also includes several user terminals that interact with the port's cloud service layer. These user terminals can be cockpit controllers, mobile apps, or management center terminals.

[0046] This invention further proposes a control method for a port tractor early warning system based on multi-source fusion as described above, comprising the following steps:

[0047] Step 1: Receive fused positioning data from BeiDou satellite positioning unit, UWB ultra-wideband positioning unit and RTK real-time dynamic differential unit in real time, and generate the trailer's real-time speed and obstacle distances for visual recognition through AI visual analysis model.

[0048] Step two: Based on the preset spatial coordinates of the electronic fence used to define the shoreline boundary and hazardous work area, the risk level determination is automatically triggered.

[0049] The beneficial effects of this invention are as follows:

[0050] This invention improves the blind spot positioning accuracy of port tractors when driving in ports, the comprehensiveness of safety warnings at driving boundaries, and the adaptability to multiple driving scenarios. It significantly reduces the risk of misjudgment and missed judgment when port tractors perform positioning warnings, greatly shortens the warning response time, and also realizes the versatility of multi-vehicle and multi-scenario application or modification in ports, thereby improving the efficiency of port safety management and scheduling. Attached Figure Description

[0051] Figure 1 This is a schematic diagram of the port tractor early warning system based on multi-source fusion according to Embodiment 1 of the present invention;

[0052] Figure 2 This is a schematic diagram of the port tractor early warning system based on multi-source fusion in Embodiment 2 of the present invention;

[0053] Figure 3 This is a flowchart illustrating the control method of the port tractor early warning system applied to multi-source fusion according to the present invention. Detailed Implementation

[0054] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be further described clearly and completely below in conjunction with the embodiments of this invention. It should be noted that the described embodiments are merely some embodiments of this invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0055] As used in this specification and the following claims, the words “a,” “an,” and “the” have the meaning of plural reference unless the context clearly indicates otherwise. Furthermore, as used in the description herein, unless the context clearly indicates otherwise, “in” has the meaning of both “in…” and “on…”.

[0056] The terms “first,” “second,” “third,” and “fourth” are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, the use of “first,” “second,” “third,” and “fourth” to designate a feature may explicitly or implicitly include one or more of that feature.

[0057] The following is a detailed description of embodiments of the invention depicted in the accompanying drawings. The embodiments are detailed in order to clearly convey the invention. However, the amount of detail provided is not intended to limit the contemplative variations of the embodiments; rather, it is intended to cover all modifications, equivalents, and substitutions falling within the spirit and scope of the invention as defined by the appended claims.

[0058] In the following description, numerous specific details are set forth in order to provide a thorough understanding of embodiments of the invention. It will be apparent to those skilled in the art that embodiments of the invention may be practiced without some of these specific details.

[0059] Example 1

[0060] like Figure 1 As shown:

[0061] This embodiment proposes a port tractor early warning system based on multi-source fusion, including an on-board subsystem and a cloud server. The on-board subsystem interacts with the cloud server. The on-board subsystem is installed on the port tractor. The cloud server is used for data storage, remote monitoring, statistical analysis, and emergency response. The on-board subsystem includes a Beidou satellite positioning unit, a UWB ultra-wideband positioning unit, an RTK real-time dynamic differential unit, a time synchronization module, a data acquisition device, a DMS camera, a HOD camera, a cockpit audio-visual equipment, a steering wheel vibration module, and a control device. The control device has a vehicle braking linkage interface and is physically / communicationally connected to the Beidou satellite positioning unit, the UWB ultra-wideband positioning unit, the RTK real-time dynamic differential unit, the time synchronization module, the data acquisition device, the cockpit audio-visual equipment, and the steering wheel vibration module. The data acquisition device is also communicatively connected to the DMS camera and the HOD camera.

[0062] The BeiDou satellite positioning unit is used to obtain the real-time location information of the port tractor.

[0063] UWB (Ultra-Wideband) positioning units are used in ports for fine-grained positioning using wireless carrier communication, thereby reducing the risk of positioning failures caused by satellite signal blockage or weak signals for port tractors.

[0064] The RTK real-time dynamic differential unit is used to receive differential signals sent by the port base station, thereby facilitating real-time correction of BeiDou satellite positioning errors;

[0065] The data acquisition device is used to collect the first data packet of the on-board management system of the port tractor in real time. The first data packet includes operating status data, environmental perception data and / or positioning data, and transmits the collected data synchronously to the control device and the cloud server.

[0066] The DMS camera is used to monitor the head and head organs of the tractor driver in real time and to identify abnormal situations of the driver through AI algorithms.

[0067] The HOD camera is used to monitor whether the driver's hands are on the steering wheel, and also to adapt to the frequent turning operation rules of port tractors to identify abnormal situations of the driver's hands operating the steering wheel.

[0068] The time synchronization module is used to unify the time reference of BeiDou satellite positioning units, UWB ultra-wideband positioning units, data acquisition devices, DMS cameras, and HOD cameras;

[0069] The steering wheel vibration module is used to provide tactile warnings by vibrating the steering wheel when the driver does not respond in time after the cockpit audio-visual equipment provides a warning.

[0070] The control device is used to receive and integrate positioning data from the Beidou satellite positioning unit, UWB ultra-wideband positioning unit, RTK real-time dynamic differential unit, vehicle status data from the data acquisition device, driver status data from the DMS camera, and driver status data from the HOD camera, based on the unified time reference of the time synchronization module and the first data packet, and generate hierarchical control commands according to the preset risk judgment logic.

[0071] The cockpit audio-visual equipment is used to issue graded warning signals to the driver when the control device detects abnormal risks in traffic conditions, vehicle movement, and driver driving.

[0072] Specifically, the DMS camera focuses on monitoring the driver's head and organ status, and can accurately identify abnormal behaviors such as driver fatigue, distraction, and yawning through AI algorithms. Compared with cameras that monitor both the driver's head and hands, the dedicated DMS camera can concentrate computing power and algorithm resources on facial feature analysis, is not affected by hand monitoring, and improves the accuracy of head status judgment. The dedicated HOD camera specifically monitors hands, and can accurately detect whether hands are on the steering wheel and abnormal hand operation of the steering wheel. It is better adapted to the frequent turning scenarios of port tractors, accurately identifies hand movements, and is not affected by head monitoring-related factors. The HOD camera, through visual image recognition, can directly observe whether there are hands on the steering wheel, as well as the specific position and posture of the hands, and is not affected by the amount of hand pressure, so as to obtain hand information more intuitively and accurately.

[0073] Preferably, the control device, as the core control component of the system, is used to receive and integrate positioning data from Beidou, UWB, and RTK, vehicle status data from the data acquisition device, and driver status data from the DMS / HOD camera. Based on preset risk judgment logic (such as distance to the boundary, driver status, and vehicle speed), it generates tiered control commands. When braking is deemed necessary, it sends precise commands (such as the pressure value for light braking and the triggering timing for emergency braking) to the tractor braking system via the CAN bus. Simultaneously, it activates the cockpit audio-visual equipment and steering wheel vibration module to enhance warnings, and synchronously uploads braking process data (such as braking time, deceleration, and braking distance) to the port management center. After braking is completed, the system can be reset based on driver operation (such as manually releasing the brakes or confirming the risk has been eliminated). This adapts to the control characteristics of port tractors, which require "diverse risks and rapid response," and serves as the core hub connecting "risk perception - warning - braking execution."

[0074] Furthermore, the control device includes a data processing unit and an application processing unit;

[0075] The data processing unit is used for collaborative fusion of positioning data, vehicle data acquisition, and driver status data acquisition; the operational status data includes the speed, steering, and braking status data of the port tractor; the positioning data includes BeiDou satellite positioning data, UWB data, and RTK data;

[0076] The application processing unit is used for data processing and decision-making, hierarchical early warning and intervention processing, and also for risk analysis and early warning execution through a multi-source fusion dynamic risk algorithm model. The multi-source fusion dynamic risk algorithm model is used to construct a dynamic risk assessment model based on multi-source fusion data. This dynamic risk assessment model is used to achieve three-dimensional risk classification of distance, speed and driver status, thereby providing a decision basis for matching the early warning intensity with the braking level, thus avoiding the abnormal risk of excessive or delayed early warning.

[0077] Ideally, the multi-source fusion dynamic risk algorithm model includes a first algorithm, a second algorithm, and a third algorithm that are executed sequentially or operated collaboratively; the first algorithm, the second algorithm, and the third algorithm correspond to the risk perception stage, the decision output stage, and the control execution stage of the port tractor safety early warning process, respectively.

[0078] Further optimized, the first algorithm is a multi-source positioning data fusion algorithm to overcome the accuracy deviation defects of single positioning methods in complex port scenarios; the second algorithm is a dynamic risk level assessment algorithm to realize three-dimensional risk classification and provide decision-making basis for matching warning intensity and braking level; the third algorithm is a graded braking adaptive control algorithm to output adaptive braking commands based on dynamic risk values ​​and real-time vehicle status.

[0079] Ideally, the port tractor warning system also includes several user terminals that interact with the cloud server, such as cockpit controllers, mobile apps, or management center terminals.

[0080] like Figure 3 As shown, this embodiment proposes a control method for a port tractor early warning system based on multi-source fusion as described in any of the above technical solutions, including the following steps:

[0081] Step 1: Receive fused positioning data from BeiDou satellite positioning unit, UWB ultra-wideband positioning unit and RTK real-time dynamic differential unit in real time, and generate the trailer's real-time speed and obstacle distances visually recognized through AI visual analysis model; then proceed to Step 2.

[0082] Step two: Based on the preset spatial coordinates of the electronic fence used to define the shoreline boundary and hazardous work area, the risk level determination is automatically triggered.

[0083] Specifically, in step one, the real-time speed of the trailer is obtained from the vehicle's CAN bus, and the result is acquired and parsed by the control device;

[0084] In a further optimized manner, the process of triggering the risk level determination in step two includes:

[0085] S201, graded braking execution and dynamic pressure regulation;

[0086] S202, Dynamic correction and precision compensation are performed during the braking process;

[0087] S203, the locking and emergency linkage mechanism is executed according to the state after braking; the locking and emergency linkage mechanism is released when the mechanism for manual release and system reset is activated.

[0088] Example 2

[0089] like Figure 2 As shown:

[0090] This embodiment proposes a port tractor early warning system based on multi-source fusion, including a port tractor's onboard management system, an onboard execution subsystem controlled by the port tractor's onboard management system, a positioning subsystem, a data acquisition and processing unit, an application processing unit, and a port cloud service layer. The port cloud service layer is used for data storage, remote monitoring, statistical analysis, and emergency response. The port tractor early warning system also includes a time synchronization unit, a DMS camera, and a HOD camera. The positioning subsystem, DMS camera, and HOD camera are installed on the port tractor. The positioning subsystem includes a Beidou satellite positioning unit, a UWB ultra-wideband positioning unit, and an RTK real-time dynamic differential unit. The onboard execution subsystem includes a vehicle braking device, a cockpit audio-visual equipment, and a steering wheel vibration module. The data acquisition and processing unit is connected to the Beidou satellite positioning unit, UWB ultra-wideband positioning unit, RTK real-time dynamic differential unit, time synchronization unit, DMS camera, HOD camera, port tractor's onboard management system, and port cloud service layer via physical media / communication links.

[0091] The BeiDou satellite positioning unit is used to obtain the real-time location information of the port tractor.

[0092] UWB ultra-wideband positioning units are used in ports for fine positioning using wireless carrier communication, thereby reducing the risk of positioning failures caused by satellite signal blockage or weak signals for port tractors.

[0093] The RTK real-time dynamic differential unit is used to receive differential signals sent by the port base station, thereby facilitating real-time correction of BeiDou satellite positioning errors;

[0094] The DMS camera is used to monitor the head and head organs of the tractor driver in real time and to identify abnormal situations of the driver through AI algorithms.

[0095] The HOD camera is used to monitor whether the driver's hands are on the steering wheel, and also to adapt to the frequent turning operation rules of port tractors to identify abnormal situations of the driver's hands operating the steering wheel.

[0096] The time synchronization unit is used to unify the time reference of the BeiDou satellite positioning unit, UWB ultra-wideband positioning unit, data acquisition and processing unit, DMS camera, and HOD camera.

[0097] The steering wheel vibration module is used to provide tactile warnings by vibrating the steering wheel when the driver does not respond in time after the cockpit audio-visual equipment provides a warning.

[0098] The data acquisition and processing unit is used to collect driver status data and the operating status data, environmental perception data and / or positioning data of the port tractor's on-board management system in real time, and synchronously transmit the relevant collected data to the application processing unit and the port cloud service layer; the operating status data includes the port tractor's speed, steering and braking status data; the positioning data includes Beidou satellite positioning data, UWB data and RTK data;

[0099] The application processing unit is used for data processing and decision-making, hierarchical early warning and intervention processing, and generates hierarchical control instructions based on the preset risk judgment logic. That is, risk analysis and early warning execution are realized through a multi-source fusion dynamic risk algorithm model and the vehicle management system.

[0100] The cockpit audio-visual equipment is used to issue graded warning signals to the driver when the application processing unit detects abnormal risks in traffic conditions, vehicle movement, and driver driving.

[0101] Ideally, the port tractor warning system also includes several user terminals that interact with the port's cloud service layer. These user terminals can be cockpit controllers, mobile apps, or management center terminals.

[0102] like Figure 3 As shown, this embodiment proposes a control method for a port tractor early warning system based on multi-source fusion as described in any of the above technical solutions, including the following steps:

[0103] Step 1: Receive fused positioning data from BeiDou satellite positioning unit, UWB ultra-wideband positioning unit and RTK real-time dynamic differential unit in real time, and generate the trailer's real-time speed and obstacle distances visually recognized through AI visual analysis model; then proceed to Step 2.

[0104] Step two: Based on the preset spatial coordinates of the electronic fence used to define the shoreline boundary and hazardous work area, the risk level determination is automatically triggered.

[0105] As an optional implementation technical solution, in step one, the precise location (centimeter-level accuracy) of the port tractor is obtained in real time through BeiDou + UWB + RTK positioning technology. Combined with the trailer motion parameters (such as displacement change rate) collected by the relevant on-board sensors of the port tractor, the real-time driving speed of the port tractor is calculated. At the same time, the positioning data of the port tractor is combined with GIS geofences (such as coastline boundaries and hazardous operation area ranges) to help verify whether the speed data of the port tractor matches the safe driving requirements of the current area (such as whether speeding is allowed in hazardous areas).

[0106] As an optional specific implementation technical solution, in step one, the high-definition camera integrated into the port tractor captures real-time images of the trailer operation scene (such as the dock front and shoreline operation area), and uses AI vision algorithms to identify targets such as pedestrians and fixed obstacles (such as shoreline facilities and equipment), and calculates the relative distance between the obstacles and the trailer, providing visual data support for subsequent early warning judgment.

[0107] In a further optimized manner, the process of triggering the risk level determination in step two includes:

[0108] Staged braking execution and dynamic pressure regulation;

[0109] Dynamic correction and precision compensation are performed during the braking process;

[0110] The locking and emergency linkage mechanism is executed according to the state after braking; the locking and emergency linkage mechanism is released when the mechanism for manual release and system reset is activated.

[0111] As an alternative implementation, the interaction between the port tractor's onboard management system and the data acquisition and processing layer is as follows:

[0112] The interaction between the onboard management system (core of which is the vehicle CAN bus control system) and the data acquisition and processing layer of the port tractor is manifested as "real-time bidirectional data flow + hardware status monitoring". Its control principle is as follows:

[0113] Principle 1, Data Acquisition and Processing Layer → Onboard Management System of Port Tractor: Raw Data and Status Signal Input:

[0114] The data acquisition and processing layer packages multi-source positioning data (latitude and longitude and altitude fused from BeiDou + UWB + RTK, with an update frequency of 10Hz), environmental perception data (obstacle distance and road friction coefficient, with a sampling frequency of 5Hz), and driver status data (fatigue coefficient from the DMS camera and hand status signal from the HOD camera) into standardized data frames (format conforming to the J1939 protocol) via sensor interfaces (such as RS485 and Ethernet), and transmits them to the onboard management system of the port tractor through the CAN bus.

[0115] Meanwhile, the data acquisition and processing layer monitors its own hardware status in real time (such as UWB module signal strength and camera obstruction status). When an anomaly occurs (such as signal loss or hardware failure), it sends a "data invalid" flag (1-byte status code) to the onboard management system of the port tractor, triggering the onboard management system of the port tractor to start a backup data acquisition mode (such as using Beidou positioning only).

[0116] Principle 2, Port tractor's onboard management system → Data acquisition and processing layer: Vehicle status feedback and control commands:

[0117] The onboard management system of the port tractor transmits real-time vehicle operating parameters to the data acquisition and processing layer via the CAN bus: engine speed (0-2500rpm), current gear (forward / reverse / neutral), braking system pressure (0-12MPa), load information (0-30 tons), vehicle speed (0-30km / h), etc., for the data acquisition and processing layer to perform data calibration (such as correcting the multipath error of UWB positioning according to the load).

[0118] When the onboard management system of the port tractor detects a vehicle malfunction (such as insufficient brake fluid), it will send a "system degradation" command to the data acquisition and processing layer to limit the sampling frequency of high-power sensors (such as high-definition cameras) (from 30fps to 10fps) and prioritize the operation of the core positioning module.

[0119] The interactive features of Principle 2: With hardware interfaces as the link, it focuses on "raw data transmission and status monitoring", with data latency ≤100ms, ensuring the real-time nature of vehicle dynamics and environmental perception, and providing a full-dimensional data foundation of "vehicle-environment-person" for subsequent risk analysis.

[0120] Specifically, the interaction between the onboard management system and the application processing unit of the port tractor is optimized as follows:

[0121] The application processing unit (with a multi-source fusion dynamic risk algorithm at its core) is the "decision-making center" of the port tractor's onboard management system. The interaction between the two is manifested as a closed-loop control of "decision instruction issuance + execution result feedback," as detailed below:

[0122] Principle 3, Application Processing Unit → Onboard Management System of Port Tractor: Hierarchical Control Command Output:

[0123] Based on the fused data from the data acquisition and processing layer, the application processing unit calculates the risk level (Level 1 / Level 2 / Level 3) using a dynamic risk algorithm and generates corresponding control commands.

[0124] The specific level one risk involved is: sending a "warning command" (including audio-visual equipment frequency and steering wheel vibration intensity parameters) to the on-board management system of the port tractor, triggering the yellow indicator light in the cockpit to flash (frequency 2Hz) + low-frequency voice prompt (volume 80dB);

[0125] The specific secondary risk involved is: sending a "light braking + enhanced warning command", which includes the braking pressure value (30%-50% of the rated pressure) and braking duration (2-5s), and simultaneously linking the steering wheel vibration module (frequency 5Hz);

[0126] The three-level risks involved are as follows: sending an "emergency braking + emergency linkage command", which includes the maximum braking pressure (80%-100% of the rated pressure), ABS / EBD activation signal, and simultaneously triggering hazard warning lights, horn (110dB) and V2X warning of surrounding vehicles;

[0127] Principle 4, Onboard Management System of Port Tractor → Application Processing Unit: Execution Status and Anomaly Feedback:

[0128] After receiving the instruction, the on-board management system of the port tractor provides real-time feedback on the execution progress to the application processing unit, such as the current value of braking pressure (e.g., 3.2MPa / target 5MPa), deceleration (e.g., 0.8m / s²), and the working status of the warning equipment (e.g., "normal sound and light" or "vibration module failure").

[0129] When an execution anomaly occurs (such as the brake pressure not reaching the target value or the steering wheel vibration module failing), the on-board management system of the port tractor immediately sends a "command execution failure" signal (including a fault code), and the application processing unit triggers a backup plan (such as increasing the brake pressure to 120% of the rated value or extending the duration of the audible and visual warnings).

[0130] Its interactive features include: algorithm-driven decision-making, emphasizing closed-loop control of "command-execution-feedback," with a command response delay of ≤50ms to ensure rapid intervention in risky scenarios. The application processing unit can dynamically adjust commands based on feedback from the port tractor's onboard management system (e.g., increasing output when braking pressure is insufficient), demonstrating "adaptive control" characteristics.

[0131] Example 3

[0132] Example 3 is an optimized design of Example 1 or Example 2;

[0133] During the risk level determination process in step two, boundary safety warning control and vehicle braking control are also executed. The boundary safety warning control and vehicle braking control processes include the following steps:

[0134] S2.1, the process of performing risk threshold determination and braking level matching is as follows: The data acquisition unit of Example 2 (or the control device of Example 1) receives real-time Beidou + UWB + RTK fusion positioning data (positioning accuracy ≤10cm), obstacle distance recognized by AI vision (error ≤0.5m), and real-time speed of the trailer (sampling frequency ≥10Hz). The application processing unit of Example 2 (or the control device of Example 1) combines the spatial coordinates of the preset electronic fence (such as the coastline boundary, dangerous work area) to automatically trigger risk level determination; when the distance between the trailer and the boundary is >5m and there is no sudden obstacle, it is determined as "Level 1 warning" and braking is not initiated; when the distance is 2-5m or there is a stationary obstacle, it is determined as "Level 2 warning" and light braking is initiated; when the distance is <2m, there is a moving obstacle (such as pedestrians, other vehicles) or the driver does not respond to the audible and visual alarm, it is determined as "Level 3 warning" and emergency braking is initiated.

[0135] S2.2, the process of prioritizing braking commands is as follows: The application processing unit of Embodiment 2 (or the control device of Embodiment 1) has built-in command priority logic. When both the driver's manual braking signal and the system's automatic braking signal are received simultaneously, manual braking is executed first (response delay ≤ 0.1s). If it is detected that the driver is not operating due to fatigue or distraction (identified by the control device of Embodiment 1, the DMS driver monitoring software subunit of the application processing unit of Embodiment 2, a DMS driver monitoring device that communicates with the control device of Embodiment 1 / the application processing unit of Embodiment 2, a DMS driver monitoring unit that interacts with the application processing unit, or the application processing unit, such as eye closure duration > 0.5s, no steering wheel operation > 3s), the priority of the system braking command is automatically increased, and the system is forced to intervene in the control. At the same time, the port management center cloud platform (i.e., the port cloud service layer) is linked to upload data such as the braking trigger cause, current vehicle position, and speed synchronously (4G / 5G transmission delay ≤ 1s) to ensure that the management end monitors the braking process in real time. Specifically, the DMS camera is installed in the cab of the port tractor.

[0136] S2.3, performs graded braking and dynamic pressure regulation;

[0137] S2.4, Perform dynamic correction and accuracy compensation during braking, specifically: During braking, the RTK real-time dynamic differential unit receives the differential signal from the base station in real time (update frequency ≥ 1Hz) to correct vehicle positioning deviations (such as positioning drift caused by ground bumps or metal obstructions). If the actual braking trajectory of the vehicle deviates from the preset safe trajectory by more than 0.3m, the control device of Embodiment 1 or the application processing unit of Embodiment 2 immediately adjusts the brake caliper pressure (adjustment accuracy ≤ 0.1MPa) to correct the driving direction and prevent the vehicle from skidding or deviating from the braking path. At the same time, the UWB ultra-wideband positioning unit continuously scans the surrounding environment. If the position of an obstacle changes (such as a pedestrian suddenly moving), the control device of Embodiment 1 or the application processing unit of Embodiment 2 recalculates the braking distance within 0.2s and dynamically adjusts the braking intensity to avoid over-braking or under-braking.

[0138] S2.5, Execute post-braking state lock and emergency linkage, specifically: When the vehicle brakes to a speed ≤5km / h (Level 1 and Level 2 warnings) or comes to a complete stop (Level 3 warning), the control device of Example 1 or the application processing unit of Example 2 automatically triggers state lock: Under Level 2 warning, maintain a light braking pressure (20%-30% of rated pressure) until the driver operates the accelerator or manually unlocks the vehicle; Under Level 3 warning, maintain the braking pressure at 100%, simultaneously activate the vehicle's hazard warning lights and horn (volume ≥110dB), and send a "forward braking warning" signal to other port vehicles within a 500m range (via V2X). Vehicle-to-vehicle communication technology (with a response delay of ≤0.5s) alerts surrounding vehicles to give way. Furthermore, upon receiving the braking completion signal, the management center of the cloud server or a separately configured backend server communicating with the vehicle subsystem automatically generates a "Braking Event Report," containing data such as braking trigger time, location, cause, braking distance, and vehicle status, for subsequent operational analysis and system optimization. Specifically, for example, in the case of a Level 2 warning light braking: the control device of Example 1 or the application processing unit of Example 2 sends a command to the vehicle braking system via the CAN bus, controlling the brake master cylinder to output 30%-50% of the rated pressure, maintaining the vehicle deceleration at 0.5-1m / s², while simultaneously triggering a voice prompt in the driver's cab (e.g., "Approaching the boundary, please slow down"), and displaying a braking progress bar on the instrument panel, allowing for manual adjustment by the driver. For example, in the case of a Level 3 warning emergency braking: after the command is triggered, the brake master cylinder pressure instantly increases to 80%-100% of the rated pressure, simultaneously activating the ABS anti-lock braking system and EBD. Electronic brake force distribution ensures that the wheels do not lock up and that the braking force is evenly distributed (the front and rear wheel braking force distribution ratio is adjusted in real time according to the vehicle load, with a load error of ≤5%), achieving rapid deceleration of 1.5-2.5m / s², while automatically shutting off the vehicle throttle and cutting off power output to avoid conflict between braking and power.

[0139] S2.6, the system reset mechanism of the control device in Embodiment 1 or the application processing unit in Embodiment 2 is executed manually. Specifically, after the driver confirms that the risk has been eliminated, he can trigger the "brake release" command through the steering wheel shortcut key or the vehicle terminal touch screen. Then, the port tractor releases the braking pressure and restores power output within 0.1s. If the three-level emergency braking is triggered three times in a row, the vehicle system will automatically lock the "automatic braking function". The driver needs to contact the management center for manual review (such as confirming the driver's status and troubleshooting the vehicle). After the review is approved, the system will be unlocked remotely through the cloud to prevent frequent braking caused by accidental triggering or vehicle failure, thereby ensuring the continuity of port tractor operation.

[0140] Specifically, preferably, the Beidou satellite positioning unit is used to acquire the real-time location of port tractors in open outdoor scenarios (such as main roads in storage yards and the shoreline at the front of wharves), providing meter-level to centimeter-level positioning data, accurately capturing the vehicle's driving trajectory along the operating route, the distance from the electronic fence of the coastline, and the real-time driving speed, providing basic location information for judging whether the vehicle deviates from the safety boundary and whether there is a risk of crossing the boundary, adapting to the characteristics of the port's open-air areas with no obstructions and a wide operating range.

[0141] Specifically, the UWB ultra-wideband positioning unit is used for high-precision positioning of port tractors in scenarios with obstructed or weak satellite signals (such as densely stacked container areas, under-bridge operation channels, and indoor maintenance workshops). It achieves positioning accuracy of 10-30 centimeters through short-range ranging technology, making up for the positioning blind spots of Beidou satellite positioning in obstructed areas, and the positioning deficiencies of vehicles in complex scenarios such as turning in container areas and close-range berthing. It can still accurately identify the distance to surrounding containers and fixed facilities, avoiding collision accidents caused by positioning failure, and adapting to the needs of port indoor and outdoor alternating operations.

[0142] Specifically, preferably, the RTK real-time dynamic differential unit is used to receive differential signals sent by the port base station, correct errors in BeiDou satellite positioning (such as ionospheric delay and multipath interference) in real time, and stabilize the positioning accuracy of the tractor at the centimeter level. Especially in scenarios where the tractor is heavily loaded (fully loaded with containers) or operating at low speed along the shoreline, which require extremely high position accuracy, it ensures that the distance calculation between the vehicle and the safety boundary (such as the shoreline edge and the isolation zone of the work area) is without deviation, and provides accurate position data support for subsequent graded early warning and braking control.

[0143] Specifically, preferably, the data acquisition device described in Embodiment 1 is used to collect in real time the first data packet of the port tractor's onboard management system, including operating status data (such as engine speed, braking system pressure, load information, and current gear), environmental perception data (such as distance to surrounding obstacles and weather visibility), and positioning data (BeiDou + UWB + RTK fusion data). This data is then synchronously transmitted to the control device and the port cloud service layer via wireless networks such as 4G / 5G. During peak port operations, data from multiple tractors can be collected simultaneously, adapting to multi-vehicle collaborative operation scenarios in the port, and providing comprehensive data input for the system to analyze whether vehicles have risks such as "decreased braking performance under heavy load" or "speeding".

[0144] Specifically, preferably, the DMS camera is used to monitor the driver's status in real time (such as facial expressions, eye movements, and head posture). Through AI algorithms, it identifies whether the driver is fatigued (such as eye closure time > 0.5s, excessive yawning frequency), distracted (such as looking down at a mobile phone, gaze deviating from the driving direction > 3s), or violates regulations (such as not wearing a seat belt). In this way, in scenarios of long-term continuous operation in the port (such as nighttime transfer), abnormal driver status can be detected in time, avoiding vehicle overtaking or collisions due to human error, and providing the system with a basis for determining whether forced braking is necessary.

[0145] Specifically, preferably, the HOD camera is used to monitor whether the driver's hands are on the steering wheel (Handon Dial), adapting to the frequent turning characteristics of port tractors (such as container area repositioning and docking), and identifying whether the driver has risks such as "taking both hands off the steering wheel to operate other equipment (such as walkie-talkies, vehicle terminals)" or "using one hand to operate the steering wheel for a long time, resulting in untimely turning"; when it is detected that both hands are off the steering wheel and the vehicle is close to the safety boundary, a risk warning is triggered, providing operational behavior basis for whether the system should activate braking.

[0146] Specifically, preferably, the time synchronization module of Embodiment 1 (time synchronization unit of Embodiment 2) is used to unify the time base of all components such as the Beidou positioning unit, UWB positioning unit, data acquisition device / data acquisition and processing unit, and high-definition camera (synchronization accuracy ≤ 1ms), avoiding problems such as "time misalignment between positioning data and video footage" and "mismatch between speed data and distance data" caused by time asynchrony of various components; when the port tractor is traveling at high speed (such as cross-yard transfer) or in case of sudden danger (such as pedestrians entering the work area), it ensures that the control device of Embodiment 1 or the application processing unit of Embodiment 2 can accurately associate the "vehicle position-time-surrounding environment" data, providing accurate time dimension basis for risk assessment, braking command triggering and subsequent accident tracing.

[0147] Preferably, the cockpit audio-visual equipment is used to issue a graded warning signal to the driver when the control device of Embodiment 1 or the application processing unit of Embodiment 2 detects a risk (such as the vehicle approaching the boundary, driver fatigue, or an obstacle intrusion). For Level 1 risk (5-10m from the boundary), a yellow indicator light flashes and a low-frequency voice prompt (such as "Please be aware of approaching the safety boundary") is triggered. For Level 2 risk (2-5m from the boundary), a red indicator light stays on and a high-frequency voice warning (such as "You are about to cross the boundary, please slow down immediately") is triggered. For Level 3 risk (<2m from the boundary or no response from the driver), a synchronized audio-visual alarm (volume ≥110dB) is triggered. This is adapted to noisy port operating environments (such as mechanical noise and ship horns), ensuring the driver can perceive the risk in a timely manner and reserving time for manual intervention by the control device of Embodiment 1 or the application processing unit of Embodiment 2 to determine whether automatic braking needs to be initiated.

[0148] Specifically, preferably, the steering wheel vibration module is used to provide tactile warnings by vibrating the steering wheel (the vibration frequency increases with the risk level) when the driver does not respond in time after the warning from the cockpit audio-visual equipment. This is to adapt to the situation where port drivers may be "insensitive to audio-visual warnings" due to long-term operations, and to further enhance the warning effect. For example, when the vehicle approaches the shoreline boundary and the driver does not notice the audio-visual alarm, the steering wheel triggers high-frequency vibration to remind the driver to slow down. If there is still no response, the system will activate automatic braking.

[0149] Specifically, preferably, the multi-source fusion dynamic risk algorithm is manifested as a port tractor multi-source data fusion dynamic risk classification early warning and control algorithm during execution. This algorithm is the core of the system to realize the whole process closed loop of "data processing-risk judgment-early warning execution-vehicle control". By integrating Beidou + UWB + RTK positioning data, vehicle operation data and driver status data, a dynamic risk assessment model is constructed, and accurate classification early warning instructions and braking control signals are output.

[0150] Specifically, preferably, the multi-source fusion dynamic risk algorithm includes three key subordinate algorithms, corresponding to the three stages of "risk perception - decision output - control execution". Each subordinate algorithm is deeply integrated with the port tractor driving scenario (such as shoreline operations, container area turning, heavy-load driving) and vehicle control process (graded braking, emergency intervention). The three key subordinate algorithms are as follows:

[0151] (I) Lower-level Algorithm 1: Multi-source positioning data fusion algorithm, the detailed description of which is as follows:

[0152] Explanation 1.1. Function of the multi-source positioning data fusion algorithm: It solves the accuracy deviation problem of single positioning technology in complex port scenarios, realizes the seamless fusion of Beidou (outdoor), UWB (obstructed area), and RTK (dynamic correction) data, and outputs centimeter-level, blind-spot-free real-time vehicle location and motion trajectory data, providing a precise spatial benchmark for risk assessment.

[0153] Explanation 1.2. Scenarios and vehicle control integration methods related to multi-source positioning data fusion algorithms; Shoreline operation scenarios related to multi-source positioning data fusion algorithms: When the tractor is traveling along the front edge of the wharf (outdoor open area), the algorithm prioritizes the use of "BeiDou + RTK" data, receives differential signals from the port base station through RTK, corrects ionospheric delay errors, stabilizes the positioning accuracy at ≤3cm, and accurately calculates the real-time distance between the vehicle and the shoreline boundary (e.g., 4.2m from the boundary), avoiding "misjudgment of boundary crossing" or "missed judgment risk" caused by positioning drift;

[0154] Explanation 1.3. Container-dense area scenario related to multi-source positioning data fusion algorithm: When the tractor enters the container area (satellite signal is blocked by containers), the algorithm automatically switches to "UWB+RTK" fusion mode. UWB captures the relative position of the vehicle and surrounding containers through short-range pulse signals (accuracy ≤10cm), and RTK synchronously corrects the cumulative error of UWB to ensure that when the vehicle turns in the container area (e.g., turning radius 5m), it can accurately identify the distance to adjacent containers (e.g., distance 0.8m), providing positional basis for light braking commands (to avoid collision);

[0155] Explanation 1.4. Data switching logic related to the multi-source positioning data fusion algorithm: The algorithm has a built-in "signal strength threshold trigger" mechanism. When the BeiDou signal strength is <-120dBm, UWB data is automatically enabled; when the BeiDou signal strength recovers to >-110dBm, it seamlessly switches back to BeiDou + RTK mode. The switching process has a delay of ≤0.2s and no data interruption, ensuring that the positioning data is not lost and the vehicle control commands (such as braking) remain effective when the vehicle is operating continuously in "outdoor-box area-outdoor".

[0156] (II) Subordinate Algorithm 2: Dynamic Risk Level Assessment Algorithm, the detailed description of which is as follows:

[0157] Explanation 2.1. The algorithm function of the dynamic risk level assessment algorithm: Based on multi-source fusion data (location, speed, driver status, environmental obstacles), a dynamic risk assessment model is constructed to realize the three-dimensional risk classification of "distance-speed-driver status", providing a decision basis for the matching of "warning intensity-braking level" and avoiding "over-warning" or "warning lag";

[0158] Explanation 2.2. Scenarios related to the dynamic risk level assessment algorithm and their integration with vehicle control:

[0159] Explanation 2.3. Heavy-load shoreline driving scenario related to the dynamic risk level assessment algorithm: A tractor-trailer fully loaded with containers (load capacity 20 tons) travels along the shoreline (speed 15km / h). The algorithm input parameters are "distance from the shoreline 3m (BeiDou + RTK data), speed 15km / h, driver fatigue-free (DMS camera), hands on the steering wheel (HOD camera)". The risk value is calculated through the risk assessment model (risk value = distance weight × 0.4 + speed weight × 0.3 + driver status weight × 0.3), resulting in a risk value of 0.6 (level 2 risk), triggering "red audible and visual warning + low-frequency vibration of the steering wheel". If the driver does not decelerate within 5 seconds (speed still ≥ 12km / h), a "light braking command" (braking pressure 30% of rated pressure) is output to prevent the heavy-load vehicle from crossing the boundary due to excessive inertia.

[0160] Explanation 2.4. Pedestrian intrusion scenario related to the dynamic risk level assessment algorithm: The tractor is operating in the yard (speed 8km / h). The AI ​​visual system detects a pedestrian intruding 5m away (the obstacle type "moving target" is uploaded simultaneously). The algorithm input parameters are "distance from pedestrian 5m, speed 8km / h, driver distracted (DMS detects looking down at mobile phone)". The calculated risk value is 0.85 (level 3 risk). This directly triggers "synchronous sound and light alarm + high-frequency vibration of steering wheel", and simultaneously outputs "emergency braking command" (brake pressure 80% of rated pressure) and cuts off the throttle to ensure that the vehicle stops within 2 seconds (braking distance ≤ 3m) to avoid collision with the pedestrian.

[0161] Note 2.5. The core assessment formula related to the dynamic risk level assessment algorithm is:

[0162] Dynamic risk value (R) = α×(1 / D) + β×V + γ×S (1)

[0163] The symbols in Formula (1) have the following meanings: R: dynamic risk value, ranging from 0 to 1 (R < 0.3 is level 1 risk, 0.3 ≤ R < 0.7 is level 2 risk, and R ≥ 0.7 is level 3 risk); D: real-time distance between the vehicle and the risk source (boundary / obstacle) (unit: m), D > 0; V: real-time speed of the vehicle (unit: km / h), V ≥ 0; S: driver status coefficient (normal driver S = 0, fatigue / distraction S = 0.5, violation operation S = 1); α, β, γ: weight coefficients (dynamically adjusted according to the port scenario, α + β + γ = 1, default α = 0.4, β = 0.3, γ = 0.3, shoreline scenario can be adjusted α = 0.5, β = 0.2, γ = 0.3);

[0164] The function of formula (1) is:

[0165] Function 1: Introducing the "1 / D" nonlinear relationship to reflect the actual scenario that "the closer the distance, the greater the risk (e.g., when D=2m, 1 / D=0.5, and when D=1m, 1 / D=1, the risk growth rate doubles).

[0166] Function 2: Combined with the driver's state coefficient (S), it avoids misjudgments caused by "only looking at distance / speed and ignoring human factors" (e.g., with D=3m and V=10km / h, when the driver is distracted, R=0.4×(1 / 3)+0.3×10+0.3×0.5≈0.73, triggering a level 3 risk; when the driver is normal, R≈0.58, triggering a level 2 risk).

[0167] Function 3: The weighting coefficient can be dynamically adjusted to adapt to the risk focus of different port operation scenarios (shoreline, container area, storage yard).

[0168] (III) Lower-level algorithm 3: Graded braking adaptive control algorithm, the detailed description of which is as follows:

[0169] Explanation 3.1. Function of the graded braking adaptive control algorithm: Based on the dynamic risk value (R) and the real-time vehicle status (load, braking system pressure, road conditions), it outputs adaptive braking commands (pressure, duration, whether to link ABS / EBD) to avoid "under-braking" (inability to stop in time) or "over-braking" (causing vehicle sideslip / cargo deviation), ensuring a safe and smooth braking process;

[0170] Explanation 3.2. Integration of Scenarios with Vehicle Control Related to the Graded Braking Adaptive Control Algorithm: Heavy-load, slippery road surface scenario related to the graded braking adaptive control algorithm: A fully loaded tractor (load weight 25 tons) travels along the shoreline in rainy weather (road surface friction coefficient μ=0.4), triggering a level 3 risk (R=0.8). The algorithm input parameters are "risk value 0.8, load weight 25 tons, road surface μ=0.4, current braking pressure 0MPa". The adaptive control model calculates the "optimal braking pressure" and "ABS activation threshold", and outputs the following instructions:

[0171] Command 1: Braking pressure is increased in two stages (the first stage increases to 50% of rated pressure within 0.5 seconds, and the second stage increases to 70% of rated pressure within 1 second) to avoid wheel lock-up caused by sudden high pressure;

[0172] Command 2: Link the ABS system. When a sudden drop in wheel speed is detected (drop > 20% / 0.1s), automatically reduce the braking pressure of the corresponding wheel (drop ≤ 10% of rated pressure) to prevent skidding.

[0173] Command 3: Link the EBD system to adjust the braking pressure ratio of the front and rear wheels (60% for the front wheels and 40% for the rear wheels) to adapt to the increased load on the rear wheels under heavy loads, ensuring that the braking distance is ≤5m (≤4m on dry roads) and that the cargo does not deviate.

[0174] Explanation 3.3. Light-load box truck braking scenario related to the graded braking adaptive control algorithm: The tractor unit is unloaded (load 5 tons) turning in the box truck area (speed 5 km / h), triggering a level 2 risk (R=0.5), and the algorithm outputs a "light braking command":

[0175] Instruction 4: Maintain braking pressure at 20% of rated pressure and control deceleration at 0.5 m / s² to avoid loss of steering control due to excessive braking of unloaded vehicles;

[0176] Instruction 5: Braking duration should be synchronized with turning progress (e.g., turning duration 3s, braking duration 2s) to ensure that the vehicle speed drops to 2km / h after turning, so as not to affect work efficiency and avoid collision with the equipment in the container area.

[0177] Explanation 3.3. The core control formula of the graded braking adaptive control algorithm is:

[0178] Optimal braking pressure (P) = P0×(R×K1 + W×K2 + (1-μ)×K3) (2)

[0179] The symbols in Formula (2) have the following meanings: P: Optimal braking pressure (unit: MPa), P≤P0 (rated pressure of vehicle braking system, default 10MPa); P0: Rated pressure of braking system (MPa), preset according to the tractor model (e.g., P0=12MPa for Sinotruk trailer); R: Dynamic risk value (0-1); W: Vehicle relative load coefficient (W=0.3 for unloaded, W=0.6 for half-loaded, W=1.0 for fully loaded); μ: Road friction coefficient (μ=0.8 for dry road, μ=0.4 for wet road, μ=0.2 for icy and snowy road); K1, K2, K3: Adjustment coefficients (K1+K2+K3=1, default K1=0.5, K2=0.3, K3=0.2).

[0180] The function of formula (2) is as follows: Function 1: It integrates the three factors of "risk value, load and road condition" to solve the problem of "unable to stop under heavy load and too sudden braking under light load" in traditional fixed pressure braking (for example, when fully loaded W=1.0 and wet road surface μ=0.4, P=10×(0.8×0.5+1.0×0.3+(1-0.4)×0.2)=10×0.72=7.2MPa, ensuring effective braking; when unloaded W=0.3 and dry road surface μ=0.8, P=10×(0.5×0.5+0.3×0.3+(1-0.8)×0.2)=10×0.38=3.8MPa, avoiding excessive braking); Function 2: Based on the vehicle model, P0 is preset to adapt to the differences in braking systems of different mainstream brands of trailers in the port, and has strong universality.

[0181] The advantages of formula (2) are:

[0182] Strong scene adaptability: For multiple scenarios in ports, including "outdoor-obstructed area-indoor", "heavy load-light load" and "dry-wet-ice and snow" road conditions, through dynamic weight adjustment and parameter adaptation, it avoids misjudgment of a single algorithm in complex scenarios. Compared with traditional fixed threshold early warning algorithms, the scene adaptability rate is improved by more than 60%.

[0183] High control precision: The multi-source positioning fusion algorithm achieves a positioning accuracy of ≤10cm, the dynamic risk assessment algorithm achieves a risk value error of ≤0.05, and the graded braking adaptive algorithm achieves a braking pressure control accuracy of ≤0.1MPa, ensuring that the braking distance error of the vehicle is ≤0.3m in high-risk scenarios such as shorelines and container areas. Compared with traditional early warning systems, the control precision is improved by 40%.

[0184] Balancing safety and efficiency: Through the "three-level risk - graded control" logic, level one risk only provides a warning without braking (to avoid affecting operational efficiency), level two risk involves mild braking (balancing safety and efficiency), and level three risk involves emergency braking (prioritizing safety). Compared to the traditional system that "emerges as soon as there is a risk," the port tractor's operational efficiency is increased by 15%, while the accident rate is reduced by 80%.

[0185] High compatibility: The corresponding algorithms of the control device in Example 1 or the application processing unit in Example 2 support the interface with the CAN bus of different brands of tractor vehicles, existing port reference stations, and cloud management platforms. No modification to existing equipment is required, reducing port deployment costs. Compared with customized algorithms, compatibility is improved by 70%.

[0186] Specifically, the optimized user terminal application process is as follows:

[0187] Phase 1 (which is completed prior to step one of the control method), user terminal pre-configuration and permission binding (initialization phase):

[0188] When deploying the border security early warning system, the port management center user terminal (PC management platform), the on-site safety officer terminal (mobile APP), and the driver's vehicle terminal (vehicle touch screen) need to complete the initial binding with the port cloud (the cloud server in Example 1 or the cloud service layer in Example 2):

[0189] Regarding the application process of the PC-based management platform: The port administrator logs into the cloud service layer through an account to complete the "electronic fence parameter configuration" (such as shoreline boundary coordinates, hazardous operation area range, and warning distance threshold) and "user permission division" (such as the administrator can modify warning parameters, while the safety officer can only view warning information). The configuration parameters are then synchronized to the port cloud database. The port cloud automatically associates with the corresponding port's RTK base station signal source to ensure that the positioning data calibration benchmark is consistent.

[0190] Regarding the application process of the mobile APP (safety officer terminal): The safety officer registers an account through the port's intranet, and after the port verifies his / her identity on the cloud, he / she binds the work area he / she is responsible for (such as shoreline section A, container stacking area B) and authorizes the permissions of "early warning information reception" and "on-site handling feedback". At the same time, he / she downloads the offline electronic fence map of the area to ensure that core early warning data can still be received when there is no network.

[0191] Regarding the application process of the vehicle-mounted terminal (driver terminal): The vehicle-mounted touch screen of each tractor unit is bound to the cloud through the device SN code. The cloud automatically sends the "braking control parameter threshold" corresponding to the tractor unit model (such as the range of light braking pressure when fully loaded, and the emergency braking response delay), and synchronizes the basic positioning data of the port operation route to complete the communication link test between the terminal and the port cloud (ensuring that the 4G / 5G transmission delay is ≤1s).

[0192] Phase 2 (which takes precedence over step two of the control method), real-time data push from the port cloud (preparation before early warning triggering):

[0193] During normal operation of the port tractor, the port cloud continuously receives the "positioning-vehicle status" collaborative data (such as real-time vehicle coordinates, speed, and driver status) uploaded by the data acquisition and processing unit of Embodiment 2 (or the data acquisition device of Embodiment 1), and pushes the data to each user terminal in real time.

[0194] Push a “Global Operation Status Map” to the PC management platform: dynamically display the location trajectory of all tractor vehicles (based on Beidou + UWB + RTK fusion positioning), electronic fence boundaries, and triggered early warning events (distinguished by different color icons: yellow level 1, orange level 2, and red level 3). It also allows administrators to view the historical trajectory (last 24 hours) and braking records of individual vehicles, providing data support for global dispatch.

[0195] Pushing "Local Early Warning Prompt" to the vehicle terminal: When the towing vehicle approaches the electronic fence warning distance (such as 10m from the shoreline boundary, without triggering a formal warning), the cloud service layer pushes the "approaching the safety boundary" prompt information (including the current distance and recommended driving speed) to the vehicle touch screen, and simultaneously synchronizes it to the driver's voice broadcast system, to guide the driver to operate in a standardized manner in advance and reduce the probability of triggering the warning;

[0196] Push "Regional Operation Dynamics" to the mobile APP (safety officer terminal): The cloud filters the real-time status data of tractor vehicles in the operation area bound to the safety officer (such as whether they have entered a high-risk area or whether they are speeding) and updates the vehicle distribution heat map in the area every 30 seconds. When the number of tractor vehicles in the area exceeds the threshold (such as ≥5 vehicles operating at the same time in the shoreline operation area), push "Regional Congestion Alert" to assist the safety officer in on-site traffic management.

[0197] Phase 3 (which is performed after step two of the control method is completed), multi-terminal collaborative interaction after the early warning event is triggered (core early warning phase):

[0198] When the tractor triggers a boundary safety warning (e.g., if the distance to the shoreline boundary is less than 2m, a Level 3 warning is triggered), the cloud service layer initiates a multi-terminal collaborative interaction process to support the warning response.

[0199] The cloud immediately pushes an "emergency warning command" to the vehicle terminal: a red warning pop-up window appears on the vehicle touch screen (displaying "About to cross the boundary, emergency braking is about to start"), simultaneously playing a high-frequency voice warning (volume ≥100dB), and displaying the current real-time distance to the boundary and suggested actions (such as "Slow down immediately, and cancel the warning after confirming safety"); if the driver does not manually intervene within 3 seconds, the vehicle terminal receives an "automatic braking trigger command" from the cloud, and links with the port tractor's vehicle management system to execute emergency braking, while simultaneously feeding back the "braking execution status" (such as braking pressure and deceleration) to the cloud.

[0200] Pushing "On-site handling tasks" to the mobile APP (safety officer terminal): Based on the real-time location of the warning vehicle, the cloud pushes a "Level 3 warning handling notification" to the APP of the safety officer responsible for the area. The notification includes the warning vehicle number, current location (with navigation link, supporting one-click navigation to the scene), warning reason (such as "risk of shoreline crossing, emergency braking has been triggered"), and provides three operation buttons: "On-site confirmation", "risk clearance" and "request support". After the safety officer arrives at the scene, he can upload photos of the scene through the APP (such as confirming that there are no obstacles and the vehicle has stopped) and report the handling results to the cloud.

[0201] Push "Warning Event Details and Linkage Control Options" to the PC management platform: The management platform will pop up a warning event pop-up window, displaying complete data of the warning vehicle (location coordinates, speed, load, driver status), warning trigger time, and current handling progress (e.g., "Emergency braking has been triggered, safety officer has departed"). Administrators can perform "remote intervention" operations through the platform (e.g., sending a "extend braking time" command to the vehicle terminal, or pushing a "warning ahead, slow down and avoid" prompt to other tractor vehicles within 500m). At the same time, a "warning event log" (including a complete timeline of data collection, cloud push, and terminal response) will be automatically generated and stored in the cloud database.

[0202] Example 4

[0203] Example 4 is an optimized design of Example 1;

[0204] The overall system architecture of this embodiment is as follows: The system adopts a four-layer architecture of "hardware layer - data layer - application layer - cloud layer". Each layer coordinates with the internal bus through 4G / 5G network. The architecture composition and relationship are as follows:

[0205] 1. Hardware Layer: Includes multi-source positioning hardware (BeiDou-3 positioning unit, UWB ultra-wideband positioning unit, RTK real-time dynamic differential unit, time synchronization module), data acquisition device, early warning and intervention hardware (cockpit audio-visual equipment, steering wheel vibration module, vehicle braking linkage interface), and installation and adaptation hardware (lightweight roof bracket, IP68 protective shell), providing the foundation for system data acquisition and execution;

[0206] 2. Data Layer: Responsible for data reception, fusion and preprocessing, including positioning data fusion (BeiDou + UWB + RTK data collaboration), vehicle data acquisition (speed, steering, braking status), driver status data acquisition (fatigue, distraction identification), forming a three-dimensional data matrix of "location-vehicle-driver", with a data update frequency ≥10Hz;

[0207] 3. Application Layer: Includes data processing and decision-making modules, and hierarchical early warning and intervention modules. It realizes risk analysis and early warning execution through core algorithms and is the "decision center" of the system.

[0208] 4. Cloud layer: namely the port cloud service layer, which is responsible for data storage, remote monitoring, statistical analysis and emergency response, and supports data synchronization and interaction across multiple terminals (cockpit controller, mobile APP, management center terminal).

[0209] Example 5

[0210] Example 5 illustrates a specific application of Example 1, Example 2, or Example 3 in the field of computer and electronic information technology.

[0211] This embodiment proposes a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the control method as described in any one of Embodiments 1, 2, or 3.

[0212] This embodiment proposes an electronic device, including a memory, a processor, and a computer program stored in the memory and executable by the processor. When the processor executes the computer program, it implements the steps of the control method as described in any one of the technical solutions in Embodiment 1, Embodiment 2, or Embodiment 3.

[0213] The memory involved in this embodiment can be volatile memory or non-volatile memory, or it can include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous linked dynamic random access memory (SLDRAM), and direct rambus RAM (DRRAM).

[0214] This invention also provides an electronic device that may include a processor, a communications interface, a memory, and a communication bus, wherein the processor, communications interface, and memory communicate with each other via the communication bus. The processor can invoke logical instructions stored in the memory to execute relevant methods.

[0215] Embodiments of the present invention may also provide another electronic device, which may include one or more processors, system control logic connected to at least one of the processors, system memory connected to the system control logic, memory connected to the system control logic, and network interface connected to the system control logic.

[0216] It should be understood that the client of this invention can be an electronic device running a third-party application. The electronic device can be a terminal device or a server. Specifically, the terminal device can include any electronic device such as a mobile phone, computer, virtual reality (VR) device, tablet computer, augmented reality (AR) device, laptop computer, etc., and this application does not limit this.

[0217] Embodiments of the present invention include various steps, which will be described below. These steps may be performed by hardware components or may be contained in machine-executable instructions, which may be used by a general-purpose or special-purpose processor programmed with the instructions to perform these steps. Alternatively, the steps may be performed by a combination of hardware, software, and firmware and / or by a human operator. The processor involved in the embodiments of this application may be a chip. For example, it may be a field-programmable gate array (FPGA), an application-specific integrated circuit (ASIC), a system-on-chip (SoC), a central processing unit (CPU), a network processor (NP), a digital signal processor (DSP), a microcontroller unit (MCU), a programmable logic device (PLD), or other integrated chips.

[0218] This invention can provide a computer program product that may include a machine-readable storage medium on which instructions are tangibly implemented, which can be used to program a computer (or other electronic device) to perform processing. The terms "machine-readable storage medium" or "computer-readable storage medium" include, but are not limited to, fixed (hardware) drives, magnetic tape, floppy disks, optical discs, optical disc read-only memory (CD-ROM) and magneto-optical discs, semiconductor memories such as ROMs, PROMs, random access memories (RAM), programmable read-only memories (PROMs), erasable PROMs (EPROMs), eMMC, electrically erasable PROMs (EEPROMs), SSDs, SDs, flash memory, magnetic or optical cards, or other types of media / machine-readable media suitable for storing electronic instructions (e.g., computer programming code, such as software or firmware). Machine-readable media may include non-transitory media in which data can be stored and does not include carrier waves and / or transient electronic signals propagated via wireless or wired connections. Examples of non-transitory media may include, but are not limited to, magnetic disks or magnetic tapes, optical storage media such as compact discs (CDs) or digital universal discs (DVDs), flash memory, memory, or memory devices. Computer program products may include code and / or machine-executable instructions, which may represent any combination of procedures, functions, subroutines, programs, routines, subroutines, modules, software packages, classes, or instructions, data structures, or program statements. Code segments may be coupled to other code segments or hardware circuitry by passing and / or receiving information, data, variables, parameters, or memory contents. Information, variables, parameters, data, etc., may be passed, forwarded, or transmitted by any suitable means, including memory sharing, message passing, token passing, network transmission, etc.

[0219] The system, device, and storage medium in this invention are based on multiple aspects of the same inventive concept as the method in the foregoing embodiments. The implementation process of the method has been described in detail above, so those skilled in the art can clearly understand the structure and implementation process of the system, device, and storage medium in this embodiment based on the foregoing description. For the sake of brevity, it will not be described again here.

[0220] The present invention can also provide some of the systems depicted in the figures in various configurations. In some embodiments, the system can be configured as a distributed system, wherein one or more components of the system are distributed across one or more networks of a cloud computing system.

[0221] Compared with existing technologies, this invention has significant advantages in terms of safety, economy, efficiency, and ecology. Specific effects and test data are as follows:

[0222] Solved the positioning blind spot and accuracy problems: Through the fusion of Beidou, UWB and RTK multi-source positioning, seamless positioning is achieved in open outdoor areas of the port (front of the wharf, long-distance transportation channels), signal-blocked areas (container yard, under the bridge), and transitional scenarios, eliminating positioning blind spots (100% blind spot elimination rate) and improving positioning accuracy to ≤3cm outdoors and ≤30cm indoors, meeting the "centimeter-level" boundary crossing warning requirements in scenarios such as shoreline boundaries, and avoiding misjudgment and missed judgment;

[0223] It solves the problems of delayed early warning and limited dimensions: It constructs a full-process mechanism of "location - data fusion - risk decision - graded early warning - intelligent intervention", collects vehicle location, driving data and driver status data in real time (update frequency ≥10Hz), realizes graded early warning of low, medium and high risks based on risk level model, and automatically triggers emergency deceleration within 5 seconds in high-risk scenarios, which significantly shortens the early warning response time. At the same time, it forms a "location + status" dual protection to improve the comprehensiveness of early warning.

[0224] The system compatibility and scenario adaptation issues have been resolved: it is compatible with the universal CAN bus interface (supporting protocols such as SAEJ1939 and ISO 15765) and the modular cockpit controller, and is compatible with mainstream Chinese brand tractor vehicles without the need to modify the original circuit; it supports adjusting the braking intervention force according to the load capacity of the port tractor (50 tons / 80 tons) and optimizing the data acquisition frequency according to the operation frequency (10 / 20 times per day), achieving customized adaptation for multiple vehicle models and multiple scenarios;

[0225] Improved data collaboration and management efficiency: Real-time data collaboration with the port cloud service layer is achieved through wireless networks such as 4G / 5G, supporting remote viewing of real-time vehicle trajectories and historical alarm records (storage period ≥ 1 year), automatically generating statistical reports (daily / weekly / monthly reports) such as equipment failure rate and boundary risk frequency, reducing manual inspection costs by 50% and improving port safety management and scheduling efficiency;

[0226] Eliminates positioning blind spots: Through the fusion of BeiDou + UWB + RTK, it covers all operational scenarios in ports, including outdoor areas, storage yards, and under bridges. After a pilot test at a major coastal port, the positioning blind spot elimination rate reached 100%, solving the problem of traditional single BeiDou positioning not being able to "see everything".

[0227] Improved early warning accuracy: outdoor positioning accuracy ≤3cm, indoor positioning accuracy ≤30cm, combined with multi-dimensional risk decision-making, shoreline boundary crossing early warning accuracy reached 99.5%, with no missed cases; port tractor driver status recognition accuracy ≥98%, avoiding accidents caused by fatigue and distraction;

[0228] Significantly reduced the accident rate: During the 6-month pilot application of 30 tractor units in Tianjin Port, there were 2 port tractor collision accidents (with equipment / pedestrians), a 60% decrease compared to before the pilot (5 incidents / 6 months); the risk of shoreline crossing incidents dropped from 8 incidents / 6 months before the pilot to 0 incidents, and the risk of falling into the sea from the shoreline approached zero.

[0229] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A port tractor early warning system based on multi-source fusion, comprising an on-board subsystem and a cloud server, wherein the on-board subsystem interacts with the cloud server, the on-board subsystem is installed on the port tractor, and the cloud server is used for data storage, remote monitoring, statistical analysis, and emergency response processing, characterized in that, The vehicle-mounted subsystem includes a BeiDou satellite positioning unit, a UWB ultra-wideband positioning unit, an RTK real-time dynamic differential unit, a time synchronization module, a data acquisition device, a DMS camera, a HOD camera, cockpit audio-visual equipment, a steering wheel vibration module, and a control device. The control device is equipped with a vehicle braking linkage interface and is physically / communicationally connected to the BeiDou satellite positioning unit, the UWB ultra-wideband positioning unit, the RTK real-time dynamic differential unit, the time synchronization module, the data acquisition device, the cockpit audio-visual equipment, and the steering wheel vibration module. The data acquisition device is also communicatively connected to the DMS camera and the HOD camera. The BeiDou satellite positioning unit is used to obtain the real-time location information of the port tractor. UWB (Ultra-Wideband) positioning units are used in ports for fine-grained positioning using wireless carrier communication, thereby reducing the risk of positioning failures caused by satellite signal blockage or weak signals for port tractors. The RTK real-time dynamic differential unit is used to receive differential signals sent by the port base station, thereby facilitating real-time correction of BeiDou satellite positioning errors; The data acquisition device is used to collect the first data packet of the on-board management system of the port tractor in real time. The first data packet includes operating status data, environmental perception data and / or positioning data, and transmits the collected data synchronously to the control device and the cloud server. The DMS camera is used to monitor the head and head organs of the tractor driver in real time and to identify abnormal situations of the driver through AI algorithms. The HOD camera is used to monitor whether the driver's hands are on the steering wheel, and also to adapt to the frequent turning operation rules of port tractors to identify abnormal situations of the driver's hands operating the steering wheel. The time synchronization module is used to unify the time reference of BeiDou satellite positioning units, UWB ultra-wideband positioning units, data acquisition devices, DMS cameras, and HOD cameras; The steering wheel vibration module is used to provide tactile warnings through steering wheel vibration; The control device is used to receive and integrate positioning data from the Beidou satellite positioning unit, UWB ultra-wideband positioning unit, RTK real-time dynamic differential unit, vehicle status data from the data acquisition device, driver status data from the DMS camera, and driver status data from the HOD camera, based on the unified time reference of the time synchronization module and the first data packet, and generate hierarchical control commands according to the preset risk judgment logic. The cockpit audio-visual equipment is used to issue graded warning signals to the driver when the control device detects abnormal risks in traffic conditions, vehicle movement, and driver driving.

2. The port tractor early warning system based on multi-source fusion according to claim 1, characterized in that, The control device includes a data processing unit and an application processing unit; The data processing unit is used for collaborative fusion of positioning data, vehicle data acquisition, and driver status data acquisition; the operational status data includes the speed, steering, and braking status data of the port tractor; the positioning data includes BeiDou satellite positioning data, UWB data, and RTK data; The application processing unit is used for data processing and decision-making, hierarchical early warning and intervention processing, and also for risk analysis and early warning execution through a multi-source fusion dynamic risk algorithm model; the multi-source fusion dynamic risk algorithm model is used to construct a dynamic risk assessment model based on multi-source fusion data, and the dynamic risk assessment model is used to achieve three-dimensional risk classification about distance, speed and driver status.

3. The port tractor early warning system based on multi-source fusion according to claim 2, characterized in that, The multi-source fusion dynamic risk algorithm model includes a first algorithm, a second algorithm, and a third algorithm that are executed sequentially or operated collaboratively; the first algorithm, the second algorithm, and the third algorithm correspond to the risk perception stage, the decision output stage, and the control execution stage of the port tractor safety early warning process, respectively.

4. The port tractor early warning system based on multi-source fusion according to claim 3, characterized in that, The first algorithm is a multi-source positioning data fusion algorithm used to overcome the accuracy deviation defects of single positioning methods in complex port scenarios; the second algorithm is a dynamic risk level assessment algorithm used to realize three-dimensional risk classification and provide decision-making basis for matching early warning intensity and braking level. The third algorithm is a graded adaptive braking control algorithm used to output adaptive braking commands based on dynamic risk values ​​and real-time vehicle status.

5. The port tractor early warning system based on multi-source fusion according to claim 1, characterized in that, The port tractor warning system also includes several user terminals that interact with the cloud server. These user terminals can be cockpit controllers, mobile apps, or management center terminals.

6. A control method for a port tractor early warning system based on multi-source fusion as described in any one of claims 1-5, characterized in that, Includes the following steps: Step 1: Receive fused positioning data from BeiDou satellite positioning unit, UWB ultra-wideband positioning unit and RTK real-time dynamic differential unit in real time, and generate the trailer's real-time speed and obstacle distances for visual recognition through AI visual analysis model. Step two: Based on the preset spatial coordinates of the electronic fence used to define the shoreline boundary and hazardous work area, the risk level determination is automatically triggered.

7. The control method according to claim 6, characterized in that, In step two, the process of triggering the risk level determination includes: Staged braking execution and dynamic pressure regulation; Dynamic correction and precision compensation are performed during the braking process; The locking and emergency linkage mechanism is executed according to the state after braking; the locking and emergency linkage mechanism is released when the mechanism for manual release and system reset is activated.

8. A port tractor early warning system based on multi-source fusion, comprising a port tractor's onboard management system, an onboard execution subsystem controlled by the port tractor's onboard management system, a positioning subsystem, a data acquisition and processing unit, an application processing unit, and a port cloud service layer, wherein the port cloud service layer is used for data storage, remote monitoring, statistical analysis, and emergency response processing, characterized in that, The port tractor early warning system also includes a time synchronization unit, a DMS camera, and a HOD camera. The positioning subsystem, DMS camera, and HOD camera are installed on the port tractor. The positioning subsystem includes a Beidou satellite positioning unit, a UWB ultra-wideband positioning unit, and an RTK real-time dynamic differential unit. The vehicle-mounted execution subsystem includes a vehicle braking device, a cockpit audio-visual equipment, and a steering wheel vibration module. The data acquisition and processing unit is connected to the Beidou satellite positioning unit, UWB ultra-wideband positioning unit, RTK real-time dynamic differential unit, time synchronization unit, DMS camera, HOD camera, the port tractor's vehicle management system, and the port cloud service layer via physical media / communication links. The BeiDou satellite positioning unit is used to obtain the real-time location information of the port tractor. UWB ultra-wideband positioning units are used in ports for fine positioning using wireless carrier communication, thereby reducing the risk of positioning failures caused by satellite signal blockage or weak signals for port tractors. The RTK real-time dynamic differential unit is used to receive differential signals sent by the port base station, thereby facilitating real-time correction of BeiDou satellite positioning errors; The DMS camera is used to monitor the head and head organs of the tractor driver in real time and to identify abnormal situations of the driver through AI algorithms. The HOD camera is used to monitor whether the driver's hands are on the steering wheel, and also to adapt to the frequent turning operation rules of port tractors to identify abnormal situations of the driver's hands operating the steering wheel. The time synchronization unit is used to unify the time reference of the BeiDou satellite positioning unit, UWB ultra-wideband positioning unit, data acquisition and processing unit, DMS camera, and HOD camera. The steering wheel vibration module is used to provide tactile warnings through steering wheel vibration; The data acquisition and processing unit is used to collect driver status data and the operating status data, environmental perception data and / or positioning data of the port tractor's on-board management system in real time, and synchronously transmit the relevant collected data to the application processing unit and the port cloud service layer; the operating status data includes the port tractor's speed, steering and braking status data; the positioning data includes Beidou satellite positioning data, UWB data and RTK data; The application processing unit is used for data processing and decision-making, hierarchical early warning and intervention processing, and generates hierarchical control instructions based on the preset risk judgment logic. That is, risk analysis and early warning execution are realized through a multi-source fusion dynamic risk algorithm model and the vehicle management system. The cockpit audio-visual equipment is used to issue graded warning signals to the driver when the application processing unit detects abnormal risks in traffic conditions, vehicle movement, and driver driving.

9. The port tractor early warning system based on multi-source fusion according to claim 8, characterized in that, The port tractor warning system also includes several user terminals that interact with the port's cloud service layer. These user terminals can be cockpit controllers, mobile apps, or management center terminals.

10. A control method for a port tractor early warning system based on multi-source fusion, as described in 8 or 9, characterized in that, Includes the following steps: Step 1: Receive fused positioning data from BeiDou satellite positioning unit, UWB ultra-wideband positioning unit and RTK real-time dynamic differential unit in real time, and generate the trailer's real-time speed and obstacle distances for visual recognition through AI visual analysis model. Step two: Based on the preset spatial coordinates of the electronic fence used to define the shoreline boundary and hazardous work area, the risk level determination is automatically triggered.

Images