Channel electronic fence calculation method and system based on unmanned aerial vehicle

Abstract

The invention discloses a channel electronic fence calculation method and system based on an unmanned aerial vehicle. The method comprises the following steps: setting a channel basic fence according to geographic features of a channel; setting a lowest fence standard according to ship attributes; fusing the navigation channel basic fence and the lowest fence through a label priority mechanism to obtain a navigation electronic fence of the ship; acquiring channel monitoring data through an unmanned aerial vehicle group to obtain a real-time monitoring label; and dynamically adjusting the navigation electronic fence based on a preset rule engine and the real-time monitoring tag. According to the invention, dynamic adjustment of fence parameters is realized through linkage of the real-time monitoring label and the rule engine in cooperation with the unmanned aerial vehicle group; customizing a core protection area according to different types, loading states and maneuvering performance of cargo ships, passenger ships and the like; by constructing a multi-dimensional label system, multi-source data is converted into labels with unified semantics, and deep data fusion is achieved; and a machine learning model is introduced to dynamically calibrate the label, and a label system and a calibration rule are optimized.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicles (UAVs), and in particular to a method and system for calculating a waterway electronic fence based on an UAV. Background Art

[0002] Waterways are crucial routes for shipping, carrying over 90% of global cargo. These routes are complex and subject to natural risks (floods, fog, etc.), human interference (overloading, illegal navigation), and damaged facilities.

[0003] Existing technologies primarily rely on AIS systems and electronic nautical charts for vessel positioning, shore-based radar and video surveillance for channel monitoring, and fixed electronic fences to demarcate hazardous areas. Some solutions employ drones to patrol pre-set routes and assist in identifying obstacles and navigational anomalies.

[0004] However, existing technologies have significant flaws: First, electronic fences are static and cannot adapt to dynamic scenarios such as construction and moving obstacles. Second, they lack differentiated ship type controls, resulting in a crude handling of the safety requirements of cargo and passenger ships. Third, data from AIS, radar, and drones is not integrated, resulting in weak risk prediction capabilities. Fourth, there is a lack of a unified labeling system, resulting in inconsistent data semantics and a low level of intelligence. These factors have resulted in an accident warning accuracy rate of less than 60% and delayed emergency response. There is an urgent need for electronic fence management technologies that enable dynamic adaptation and intelligent decision-making. Summary of the Invention

[0005] The embodiments of the present invention provide a method and system for calculating electronic fences for waterways based on drones, which are used to address the problems in the existing technology of static electronic fences, inability to adapt to dynamic scenarios such as construction and obstacle drift, lack of differentiated ship type management and control, rough processing of safety requirements for cargo ships, passenger ships, etc., unintegrated data such as AIS, radar, and drones, weak risk prediction capabilities, lack of a unified labeling system, inconsistent data semantics, and low intelligence level.

[0006] In one aspect, an embodiment of the present invention provides a method for calculating a waterway electronic fence based on a drone, comprising:

[0007] Setting up basic waterway fences based on the geographical characteristics of the waterway, including waterway type, risk level, and topography;

[0008] Setting minimum fence standards based on vessel attributes, including vessel type, load status, and maneuverability;

[0009] The navigation electronic fence of the ship is obtained by fusing the basic fence of the waterway and the lowest fence through a tag priority mechanism;

[0010] The drone group collects waterway monitoring data to obtain real-time monitoring tags. The monitoring data includes obstacles, navigation mark status, meteorological and hydrological parameters;

[0011] The navigation electronic fence is dynamically adjusted based on a preset rule engine and the real-time monitoring tag.

[0012] In a possible implementation, setting the waterway basic fence according to the geographical features of the waterway includes:

[0013] Labeling the geographical features with waterway type labels, risk level labels, and navigation density labels;

[0014] Calculate the basic protection area width according to the preset expansion coefficient of the waterway type label, risk level label and navigation density label;

[0015] Priority protection is implemented for areas with high-risk labels, wherein the high-risk labels include narrow channel labels and high-risk bend labels of the channel type labels.

[0016] In a possible implementation, setting the minimum fence standard according to the ship attributes includes:

[0017] Establishing a mapping relationship between a ship type tag and the minimum fence standard, wherein the ship type tag includes a cargo ship tag, a passenger ship tag, and a special operation ship tag;

[0018] The minimum fence standards are revised based on the load status label and the maneuverability label, wherein the load status label includes a heavy load label and a light load label.

[0019] In a possible implementation, the fusing of the waterway basic fence and the lowest fence through a tag priority mechanism to obtain the navigation electronic fence of the ship includes:

[0020] Calculate the boundary value between the basic fence of the waterway and the minimum fence standard;

[0021] Obtaining a final fence boundary based on the boundary value;

[0022] Processing conflicting parameters based on the tag priority mechanism;

[0023] The tag priority mechanism includes high-risk tags taking precedence over regular tags.

[0024] In a possible implementation, the dynamically adjusting the navigation electronic fence based on a preset rule engine and the real-time monitoring tag includes:

[0025] triggering a fence parameter adjustment rule based on the real-time monitoring tag, wherein the real-time monitoring tag includes an obstacle type tag and a navigation mark offset tag;

[0026] Expanding the adjustment range through a multi-source tag linkage mechanism, which includes a combined response mechanism of weather tags and ship type tags;

[0027] Generate a fence adjustment instruction through the rule engine;

[0028] The rule engine sends the fence adjustment instruction to the ship and the drone group.

[0029] In a possible implementation, the dynamically adjusting the navigation electronic fence based on a preset rule engine and the real-time monitoring tag further includes:

[0030] The accuracy of the real-time monitoring tags is ensured through a cross-validation mechanism of multi-source data, including AIS data, electronic chart data, and drone monitoring data;

[0031] Establishing a dynamic update mechanism for the real-time monitoring tag, the dynamic update mechanism including event-triggered update and periodic calibration;

[0032] Blockchain technology is used to achieve label traceability of the real-time monitoring label.

[0033] In a possible implementation, the dynamically adjusting the navigation electronic fence based on a preset rule engine and the real-time monitoring tag further includes:

[0034] Training a label calibration model, wherein the input of the label calibration model includes basic label data, real-time perception data, and historical behavior data;

[0035] identifying an error pattern of the real-time monitoring tag based on the tag calibration model;

[0036] generating correction suggestions based on the error pattern of the real-time monitoring tag;

[0037] Verify and apply the correction suggestions through human-machine collaboration mechanism.

[0038] On the other hand, an embodiment of the present invention provides a waterway electronic fence system based on a drone, comprising:

[0039] A waterway labeling module is used to set waterway basic fences based on the geographical characteristics of the waterway, including waterway type, risk level, and topography;

[0040] A ship type tag module is used to set minimum fence standards based on ship attributes, including ship type, load status, and maneuverability;

[0041] A fence calculation module is used to fuse the basic fence of the waterway and the lowest fence through a tag priority mechanism to obtain the navigation electronic fence of the ship;

[0042] The monitoring tag module is used to collect waterway monitoring data through the drone group to obtain real-time monitoring tags. The monitoring data includes obstacles, navigation mark status, meteorological and hydrological parameters;

[0043] A rule engine module, configured to dynamically adjust the navigation electronic fence based on a preset rule engine and the real-time monitoring tag;

[0044] The command execution module is used to send adjustment commands to the ship terminal and the UAV group

[0045] In one possible implementation, the rule engine module includes:

[0046] a tag matching unit, configured to identify an adjustment rule corresponding to the real-time monitoring tag;

[0047] Priority determination unit, used to handle rule priorities when multiple tags conflict;

[0048] A parameter calculation unit is used to calculate the correction value of the fence parameter according to the adjustment rule.

[0049] In one possible implementation, a label quality control module is also included, which is used to verify label accuracy, including multi-source data cross-comparison and expert knowledge base verification; manage the label life cycle, including full process records of label creation, update, and expiration; and generate a label quality assessment report, including indicators such as label accuracy and update timeliness.

[0050] The drone-based waterway electronic fence calculation method and system of the present invention have the following advantages:

[0051] (1) Through the linkage between real-time monitoring tags and rule engines, and in conjunction with drone groups, dynamic adjustment of fence parameters can be achieved.

[0052] (2) Establish a mapping relationship between ship type labels and fence parameters, and customize core protection areas and monitoring operation areas according to different types of cargo ships, passenger ships, etc., as well as their load status and maneuverability.

[0053] (3) By building a multi-dimensional labeling system, multi-source data such as AIS, radar, and drones are converted into labels with unified semantics to achieve deep data integration.

[0054] (4) Introduce machine learning models to dynamically calibrate labels, use historical data to mine potential patterns, and optimize the labeling system and calibration rules. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0056] Figure 1 A flowchart of a method for calculating electronic fences for a waterway based on a drone is provided in an embodiment of the present application. DETAILED DESCRIPTION

[0057] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0058] Figure 1 A flowchart of a method for calculating electronic fences for waterways based on drones is provided in an embodiment of the present invention. The method includes: setting a basic waterway fence based on the geographical features of the waterway, wherein the geographical features include the waterway type, risk level, and topography;

[0059] Setting minimum fence standards based on vessel attributes, including vessel type, load status, and maneuverability;

[0060] The navigation electronic fence of the ship is obtained by fusing the basic fence of the waterway and the lowest fence through a tag priority mechanism;

[0061] The drone group collects waterway monitoring data to obtain real-time monitoring tags. The monitoring data includes obstacles, navigation mark status, meteorological and hydrological parameters;

[0062] Dynamically adjust the navigation electronic fence based on a preset rule engine and the real-time monitoring tag;

[0063] The setting of the waterway basic fence according to the geographical features of the waterway includes:

[0064] Labeling the geographical features with waterway type labels, risk level labels, and navigation density labels;

[0065] Calculate the basic protection area width according to the preset expansion coefficient of the waterway type label, risk level label and navigation density label;

[0066] Prioritize protection for areas with high-risk labels, including narrow channel labels and high-risk bend labels of the channel type labels;

[0067] The minimum fence standards set according to the ship attributes include:

[0068] Establishing a mapping relationship between a ship type tag and the minimum fence standard, wherein the ship type tag includes a cargo ship tag, a passenger ship tag, and a special operation ship tag;

[0069] amending the minimum fencing standards based on a load status label and a maneuverability label, wherein the load status label comprises a heavy load label and a light load label;

[0070] The method of fusing the waterway basic fence and the minimum fence to obtain the navigation electronic fence of the ship through the tag priority mechanism includes:

[0071] Calculate the boundary value between the basic fence of the waterway and the minimum fence standard;

[0072] Obtaining a final fence boundary based on the boundary value;

[0073] Processing conflicting parameters based on the tag priority mechanism;

[0074] The label priority mechanism includes high-risk labels taking precedence over regular labels;

[0075] The dynamically adjusting the navigation electronic fence based on the preset rule engine and the real-time monitoring tag includes:

[0076] triggering a fence parameter adjustment rule based on the real-time monitoring tag, wherein the real-time monitoring tag includes an obstacle type tag and a navigation mark offset tag;

[0077] Expanding the adjustment range through a multi-source tag linkage mechanism, which includes a combined response mechanism of weather tags and ship type tags;

[0078] Generate a fence adjustment instruction through the rule engine;

[0079] The rule engine sends the fence adjustment instruction to the ship and the drone group;

[0080] After dynamically adjusting the navigation electronic fence based on the preset rule engine and the real-time monitoring tag, the method further includes:

[0081] The accuracy of the real-time monitoring tags is ensured through a cross-validation mechanism of multi-source data, including AIS data, electronic chart data, and drone monitoring data;

[0082] Establishing a dynamic update mechanism for the real-time monitoring tag, the dynamic update mechanism including event-triggered update and periodic calibration;

[0083] Using blockchain technology to achieve label traceability of the real-time monitoring label;

[0084] After dynamically adjusting the navigation electronic fence based on the preset rule engine and the real-time monitoring tag, the method further includes:

[0085] Training a label calibration model, wherein the input of the label calibration model includes basic label data, real-time perception data, and historical behavior data;

[0086] identifying an error pattern of the real-time monitoring tag based on the tag calibration model;

[0087] generating correction suggestions based on the error pattern of the real-time monitoring tags;

[0088] Verify and apply the correction suggestions through human-machine collaboration mechanism.

[0089] For example, in this application, a ship carries multiple drone groups, and realizes dynamic and differentiated management of waterway electronic fences through real-time drone detection, a multi-dimensional tagging system, and a rule engine;

[0090] The first step is to build the basic fence of the waterway. First, data such as waterway type (such as main channel, narrow channel), risk level (high / medium / low), topography (curve, shoal) are collected and marked with channel type labels (such as narrow_channel), risk level labels (such as high_risk), and traffic density labels (such as busy_traffic). Then, the width of the basic protection area is calculated using the preset expansion coefficient of the label. For example, the narrow channel label triggers a 1.2-1.5 times width expansion, and the high-risk curve label adds an additional 20% width to complete the dynamic width calculation. Then, priority protection is implemented for areas marked with high-risk labels (such as narrow channels and high-risk curves), the core protection area is expanded, and the monitoring frequency is increased.

[0091] The second step is to customize the minimum fence standard. First, a mapping relationship is established between the ship type label (cargo ship, passenger ship, special operation ship) and the basic fence parameters. For example, the default radius of the core protection zone for cargo ships is 80 meters, and for passenger ships it is 60 meters. Then, the fence parameters are corrected according to the load status label (heavy load / light load) and the maneuverability label (good / poor). For example, the radius of the core zone for heavily loaded cargo ships is increased by 30%, and the length of the monitoring zone for ships with poor maneuverability is extended by 20%.

[0092] Next is the fusion generation of navigation electronic fences. First, the boundary parameters of the basic fence of the channel and the minimum fence of the ship type (such as the core area radius and the monitoring area length) are extracted respectively to complete the boundary value calculation. Then, the label priority mechanism is used to handle conflicts. High-risk labels (such as high_risk) take precedence over regular labels. The stricter boundary value of the two is taken as the final fence parameter (for example, the core area of ​​a cargo ship in a narrow channel takes the basic fence of 100 meters instead of the default 80 meters of the ship type) to complete the priority fusion.

[0093] The second step is real-time monitoring and label generation. First, drones equipped with visible light / infrared cameras, lidar, and hydrological sensors collect data on obstacles (such as sunken ships and floating objects), navigation beacon status (offset / damage), meteorological and hydrological data (wind speed, water level), etc. in real time. The monitoring data is then converted into real-time labels (such as obstacle_type:sunken_ship, buoy_status:offset) and transmitted back to the system via 5G / Beidou to complete the label conversion.

[0094] The lowest level is dynamic adjustment and instruction issuance. The rule engine matches the preset adjustment rules according to the real-time tags. For example, when obstacle_type:sunken_ship is detected, the radius of the core area is expanded by 1.5 times. Combined with the meteorological tag (such as weather:typhoon) and the ship type tag (such as vessel_type:passenger_ship), the adjustment range is expanded (such as increasing the passenger ship warning buffer zone by 50% during a typhoon). Then, the accuracy of the tag is cross-verified through AIS data and electronic nautical charts, and the blockchain is used to record the tag change log. The machine learning model is used to identify tag errors and generate correction suggestions (such as correcting the incorrectly labeled "light load" to "heavy load").

[0095] The above solution requires data structure definition first, that is, defining the label and fence parameters. The code example is as follows:

[0096] fromdataclassesimportdataclass

[0097] fromenumimportEnum

[0098] import numpy asnp

[0099] #---------------------------Tag Definition---------------------------

[0100] classChannelType(Enum):

[0101] MAIN="main_channel"#Main channel

[0102] NARROW="narrow_channel"#Narrow channel

[0103] HIGH_CURVE="high_curve"#High-risk curve

[0104] classRiskLevel(Enum):

[0105] HIGH="high_risk"

[0106] MEDIUM="medium_risk"

[0107] LOW="low_risk"

[0108] classVesselType(Enum):

[0109] CARGO="cargo_ship"#Cargo ship

[0110] PASSENGER="passenger_ship"#Passenger ship

[0111] SPECIAL="special_ship"#Special operation ship

[0112] classLoadStatus(Enum):

[0113] HEAVY="heavy_load"#Heavy load

[0114] LIGHT="light_load"#Light load

[0115] classManeuverability(Enum):

[0116] POOR="poor_maneuver"#Poor maneuverability

[0117] GOOD="good_maneuver"#Good maneuverability

[0118] #Real-time monitoring tags (example)

[0119] @dataclass

[0120] classMonitorTag:

[0121] tag_type:str#Tag type (such as "obstacle", "buoy_offset")

[0122] value:float#Tag value (e.g. obstacle distance 30m, beacon offset 5m)

[0123] timestamp:str#timestamp

[0124] #---------------------------Fence Parameter Class---------------------------@dataclass

[0125] classFenceParams:

[0126] core_radius: float#Core protection zone radius (meters)

[0127] monitor_length:float#Monitoring area length (meters)

[0128] warning_buffer: float#Warning buffer width (meters)

[0129] Then, the geo-driven waterway foundation fence is constructed. The code example is as follows:

[0130] classChannelFenceBuilder:

[0131] #Preset expansion factor (can be adjusted according to actual scenario)

[0132] _CHANNEL_EXT_COEFF={

[0133] ChannelType.MAIN:1.0,

[0134] ChannelType.NARROW:1.3,#Narrow channel expansion 30%

[0135] ChannelType.HIGH_CURVE:1.2#High-risk curve expansion 20%

[0136] }

[0137] _RISK_EXT_COEFF={

[0138] RiskLevel.HIGH:1.2,#High risk additional expansion 20%

[0139] RiskLevel.MEDIUM:1.1,

[0140] RiskLevel.LOW:1.0

[0141] }

[0142] defbuild_basic_fence(self,channel_type:ChannelType,risk_level:RiskLevel,base_width:float)->FenceParams:

[0143] """

[0144] Calculate waterway foundation fence parameters based on geographical features

[0145] :paramchannel_type: Channel type label

[0146] :paramrisk_level: Risk level label

[0147] :parambase_width: Channel base width (meters)

[0148] :return: Basic fence parameters

[0149] """

[0150] #Calculate the core protection zone radius (1 / 2 of the base width * channel type expansion coefficient)

[0151] core_radius=(base_width / 2)*self._CHANNEL_EXT_COEFF[channel_type]

[0152] #Risk level correction core area radius

[0153] core_radius*=self._RISK_EXT_COEFF[risk_level]

[0154] #The length of the monitoring area is 5 times the radius of the core area (experience value)

[0155] monitor_length=core_radius*5

[0156] #The warning buffer zone is 1.5 times the radius of the core area

[0157] warning_buffer=core_radius*1.5

[0158] returnFenceParams(

[0159] core_radius=round(core_radius,2),

[0160] monitor_length=round(monitor_length,2),

[0161] warning_buffer=round(warning_buffer,2) )

[0163] #Example call: A narrow and dangerous bend in the Yangtze River (basic width 90 meters)

[0164] builder=ChannelFenceBuilder()

[0165] basic_fence=builder.build_basic_fence(

[0166] channel_type=ChannelType.NARROW,

[0167] risk_level=RiskLevel.HIGH,

[0168] base_width=90)print(f"Basic fence parameters: {basic_fence}")#Output: core_radius=70.2, monitor_length=351.0, warning_buffer=105.3

[0169] Then, the minimum fence standard is set based on the ship type attributes. The code example is as follows:

[0170] classVesselFenceBuilder:

[0171] #Default minimum fence parameters for ship types (can be adjusted according to ship specifications)

[0172] _VESSEL_BASE_PARAMS={

[0173] VesselType.CARGO:FenceParams(core_radius=50,monitor_length=250,warning_buffer=75),

[0174] VesselType.PASSENGER:FenceParams(core_radius=40,monitor_length=200,warning_buffer=60),

[0175] VesselType.SPECIAL:FenceParams(core_radius=60,monitor_length=300,warning_buffer=90)

[0176] }

[0177] #Load status correction factor

[0178] _LOAD_COEFF={LoadStatus.HEAVY:1.3,LoadStatus.LIGHT:1.0}

[0179] #Maneuverability correction factor

[0180] _MANEUVER_COEFF={Maneuverability.POOR:1.2,Maneuverability.GOOD:1.0}

[0181] defbuild_min_fence(self,vessel_type:VesselType,load_status:LoadStatus,maneuver:Maneuverability)->FenceParams:

[0182] """

[0183] Calculate minimum fence standards based on vessel attributes

[0184] :paramvessel_type: Vessel type tag

[0185] :paramload_status: Load status label

[0186] :parammaneuver:Manipulation performance tag

[0187] :return: Minimum fence parameter

[0188] """

[0189] base_params=self._VESSEL_BASE_PARAMS[vessel_type]

[0190] #Load correction core area radius

[0191] core_radius=base_params.core_radius*self._LOAD_COEFF[load_status]

[0192] #Maneuverability correction monitoring area length

[0193] monitor_length=base_params.monitor_length*self._MANEUVER_COEFF[maneuver]

[0194] #The buffer zone is maintained with the core zone.

[0195] warning_buffer=base_params.warning_buffer*self._LOAD_COEFF[load_status]

[0196] returnFenceParams(

[0197] core_radius=round(core_radius,2),

[0198] monitor_length=round(monitor_length,2),

[0199] warning_buffer=round(warning_buffer,2) )

[0201] #Example call: Heavy-loaded, poorly maneuverable cargo ship

[0202] vessel_builder=VesselFenceBuilder()

[0203] min_fence=vessel_builder.build_min_fence(

[0204] vessel_type=VesselType.CARGO,

[0205] load_status=LoadStatus.HEAVY,

[0206] maneuver = Maneuverability.POOR) print(f"Minimum fence parameter: {min_fence}") # Output: core_radius = 65.0, monitor_length = 300.0, warning_buffer = 97.5

[0207] Then, navigation electronic fences are fused and generated based on the tag priority mechanism. The code example is as follows:

[0208] classFenceFusionEngine:

[0209] @staticmethod

[0210] deffuse_fences(basic_fence:FenceParams,min_fence:FenceParams,risk_level:RiskLevel)->FenceParams:

[0211] """

[0212] Fusion of basic fence and minimum fence, with high-risk tags prioritized

[0213] :parambasic_fence: Waterway basic fence parameters

[0214] :parammin_fence: Minimum fence parameters for ship types

[0215] :paramrisk_level: Waterway risk level (used for priority determination)

[0216] :return: Final navigation electronic fence parameters

[0217] """

[0218] #High-risk scenarios take stricter parameters (basic fences take priority)

[0219] ifrisk_level==RiskLevel.HIGH:

[0220] core_radius=max(basic_fence.core_radius,min_fence.core_radius)

[0221] monitor_length=max(basic_fence.monitor_length,min_fence.monitor_length)

[0222] warning_buffer=max(basic_fence.warning_buffer,min_fence.warning_buffer)

[0223] else:

[0224] #Minimum standards for ship types in low-risk scenarios (avoiding excessive restrictions)

[0225] core_radius=min(basic_fence.core_radius,min_fence.core_radius)

[0226] monitor_length=min(basic_fence.monitor_length,min_fence.monitor_length)warning_buffer=min(basic_fence.warning_buffer,min_fence.warning_buffer)returnFenceParams(

[0227] core_radius=round(core_radius,2),

[0228] monitor_length=round(monitor_length,2),

[0229] warning_buffer=round(warning_buffer,2) )

[0231] #Example call: High-risk waterway fusion base fence (core = 70.2) and cargo ship minimum fence (core = 65.0)

[0232] fusion_engine=FenceFusionEngine()

[0233] final_fence=fusion_engine.fuse_fences(

[0234] basic_fence=basic_fence,

[0235] min_fence=min_fence,

[0236] risk_level = RiskLevel.HIGH)print(f"Final fence parameters: {final_fence}")#Output: core_radius = 70.2, monitor_length = 351.0, warning_buffer = 105.3

[0237] Then, real-time monitoring and label generation are performed based on the drone data. The code example is as follows:

[0238] classDroneMonitor:

[0239] def__init__(self):

[0240] self.drone_id="DRONE-001" #Drone ID

[0241] defcapture_monitor_tags(self,obstacle_distance:float,buoy_offset:float,wind_speed:float)->list[MonitorTag]:

[0242] """

[0243] Simulate drone data collection and generate real-time monitoring labels

[0244] :paramobstacle_distance: obstacle distance (meters)

[0245] :parambuoy_offset: Navigation beacon offset (meters)

[0246] :paramwind_speed: Wind speed (m / s)

[0247] :return: Real-time monitoring tag list

[0248] """

[0249] tags=[]

[0250] #Obstacle label (distance <50m is high risk)

[0251] ifobstacle_distance<50:

[0252] tags.append(MonitorTag(tag_type="obstacle_high_risk",value="obstacle_distanc e,timestamp="2025-06-01T09:15:00"))

[0253] # Navigation mark offset label (offset > 2m needs to be adjusted)

[0254] ifbuoy_offset>2:

[0255] tags.append(MonitorTag(tag_type="buoy_offset",value="buoy_offset,timestamp="2025-06-01T09:15:00"))

[0256] #High wind tag (wind speed > 10m / s affects navigation)

[0257] ifwind_speed>10:

[0258] tags.append(MonitorTag(tag_type="high_wind", value="wind_speed, timestamp="2025-06-01T09:15:00"))

[0259] returntags

[0260] #Example call: The drone detects an obstacle distance of 30m, a navigation mark offset of 3m, and a wind speed of 12m / s

[0261] drone=DroneMonitor()

[0262] monitor_tags=drone.capture_monitor_tags(

[0263] obstacle_distance=30,

[0264] buoy_offset=3,

[0265] wind_speed=12)print("Real-time monitoring tags:",[tag.tag_typefortaginmonitor_tags])#Output: ['obstacle_high_risk','buoy_offset','high_wind']

[0266] Finally, the rule engine drives dynamic adjustments and instruction issuance. The code example is as follows;

[0267] classRuleEngine:

[0268] def__init__(self):

[0269] #Preset adjustment rules (label type->parameter correction coefficient)

[0270] self.rules = {

[0271] "obstacle_high_risk":{"core_radius":1.5,"monitor_length":1.2},#When there is an obstacle, the core area will expand by 50%

[0272] "buoy_offset":{"core_radius":1.3,"warning_buffer":1.1},#The core area expands by 30% when the beacon is offset

[0273] "high_wind":{"monitor_length":1.4,"speed_limit":0.8}#In high winds, the monitoring area is extended by 40% and the speed limit is 20%.

[0274] }

[0275] defadjust_fence(self,current_fence:FenceParams,tags:list[MonitorTag])->tuple[FenceParams,dict]:

[0276] """

[0277] Dynamically adjust fence parameters and generate instructions based on real-time tags

[0278] :paramcurrent_fence: Current fence parameters

[0279] :paramtags: Real-time monitoring tag list

[0280] :return:Adjusted fence parameters and command content

[0281] """

[0282] adjusted_params=current_fence.__dict__.copy()

[0283] commands = {"speed_limit": 1.0} # No speed limit by default (1.0 is the original speed)

[0284] fortagintags:

[0285] iftag.tag_typeinself.rules:

[0286] # Apply rule modification parameters

[0287] rule=self.rules[tag.tag_type]

[0288] forparam,coeffinrule.items():

[0289] ifparam=="speed_limit":

[0290] commands["speed_limit"]*=coeff#Speed ​​limit coefficient superposition

[0291] else:

[0292] adjusted_params[param]*=coeff

[0293] #Round to 2 decimal places

[0294] adjusted_fence = FenceParams(

[0295] core_radius=round(adjusted_params["core_radius"],2),

[0296] monitor_length=round(adjusted_params["monitor_length"],2),

[0297] warning_buffer=round(adjusted_params["warning_buffer"],2) )

[0299] returnadjusted_fence,commands

[0300] #Example call: Adjust fence based on real-time tags

[0301] rule_engine = RuleEngine()

[0302] adjusted_fence, commands = rule_engine.adjust_fence(final_fence, monitor_tags) print(f"Adjusted fence: {adjusted_fence}") # Output: core_radius = 157.95(70.2*1.5*1.3), monitor_length = 589.68(351*1.2*1.4) ... print(f"Command: Speed ​​limit {round(commands['speed_limit']*100)}%") # Output: Speed ​​limit 80%

[0303] To identify and correct label errors through machine learning, data is first prepared to construct a training dataset. The input data includes: basic label data (such as waterway type, risk level, and traffic density (structured labels); real-time perception data (such as obstacle coordinates, navigation mark offsets, and meteorological parameters monitored by drones (numeric data); and historical behavior data (such as the ship's actual navigation trajectory (whether it deviates from the fence) and accident records (whether a collision or grounding has occurred). The data's true value is then set, and the "true label" (such as actual obstacle distance and true navigation mark status) is verified through manual verification or AIS / electronic nautical charts.

[0304] Then, we perform model training, which is error pattern recognition. We select the random forest regression model, input basic labels, real-time perception data, and historical behavior data, and predict the label error (perception value - true value). The code example is as follows:

[0305] importpandasaspdfromsklearn.ensembleimportRandomForestRegressorfromskle arn.model_selectionimporttrain_test_splitfromsklearn.metricsimportmean_absolute_error

[0306] #Load training data (assuming it has been preprocessed into a DataFrame)

[0307] data=pd.read_csv("label_error_dataset.csv")

[0308] X=data[["channel_type","obstacle_distance","deviation_flag"]]#Input features

[0309] y = data["error"] #label error (perception - reality)

[0310] # Divide the training set and test set

[0311] X_train,X_test,y_train,y_test=train_test_split(X,y,test_size=0.2)

[0312] #Train random forest model

[0313] model=RandomForestRegressor(n_estimators=100)

[0314] model.fit(X_train,y_train)

[0315] #Evaluate the model

[0316] y_pred = model.predict(X_test) print(f"Error prediction MAE: {mean_absolute_error(y_test,y_pred):.2f}m") # Output: Error prediction mean absolute error ≤ 2m

[0317] Error identification and correction suggestion generation are performed through online inference. This involves acquiring current label data in real time (e.g., a drone reports "obstacle distance 30m"), inputting the model's prediction error (e.g., the model outputs "error = 5m"), and then performing a correction. If the absolute value of the error exceeds a threshold (e.g., 3m), it is marked as a "high error label." The correction suggestion is: true value = perceived value - prediction error (e.g., 30m - 5m = 25m).

[0318] After correction, human-computer collaborative verification is carried out, and then the manually verified correction results (true values) are added to the training set. The model is retrained regularly (such as once a week) to achieve iterative optimization of the model.

[0319] The embodiment of the present invention further provides a waterway electronic fence system based on a drone, comprising:

[0320] A waterway labeling module is used to set waterway basic fences based on the geographical characteristics of the waterway, including waterway type, risk level, and topography;

[0321] A ship type tag module is used to set minimum fence standards based on ship attributes, including ship type, load status, and maneuverability;

[0322] A fence calculation module is used to fuse the basic fence of the waterway and the lowest fence through a tag priority mechanism to obtain the navigation electronic fence of the ship;

[0323] The monitoring tag module is used to collect waterway monitoring data through the drone group to obtain real-time monitoring tags. The monitoring data includes obstacles, navigation mark status, meteorological and hydrological parameters;

[0324] A rule engine module, configured to dynamically adjust the navigation electronic fence based on a preset rule engine and the real-time monitoring tag;

[0325] The command execution module is used to send adjustment commands to the ship terminal and drone group.

[0326] For example, the system consists of six core modules and a quality control module, realizing closed-loop management from data collection to instruction execution.

[0327] Channel label module: parses channel geographic data and generates basic fence labels (such as channel_type and risk_level).

[0328] Vessel type label module: reads AIS / ship registration data, generates vessel type, load capacity, and maneuverability labels, and matches minimum fence standards.

[0329] Fence calculation module: Integrates basic fences and ship type tags to generate navigation electronic fences through a priority mechanism (e.g., the core area radius of a cargo ship in a high-risk curve = max (basic 120 meters, ship type 100 meters) = 120 meters).

[0330] Monitoring label module: dispatch drones to monitor according to preset routes, and generate obstacles, navigation marks, and weather labels (such as hydrology: high_water) in real time.

[0331] Rule engine module: includes a label matching unit (identifying the buoy_status:offset corresponding navigation mark correction rule), a priority determination unit (handling multi-label conflicts), and a parameter calculation unit (calculating the fence expansion range).

[0332] Command execution module: Sends speed limit commands to ships via VHF / 5G and sends route adjustment commands to the drone swarm.

[0333] Label quality control module: multi-source data cross-validation (such as comparison of AIS ship type and registration data), label lifecycle management (recording creation / update / expiration time), and quality report generation (such as label accuracy ≥ 95%).

[0334] In a possible embodiment, the rule engine module includes:

[0335] a tag matching unit, configured to identify an adjustment rule corresponding to the real-time monitoring tag;

[0336] Priority determination unit, used to handle rule priorities when multiple tags conflict;

[0337] A parameter calculation unit is used to calculate the correction value of the fence parameter according to the adjustment rule.

[0338] In a possible embodiment, a label quality control module is further included, which is used to verify label accuracy, including multi-source data cross-comparison and expert knowledge base verification; manage the label life cycle, including full process records of label creation, update, and expiration; and generate a label quality assessment report, including indicators such as label accuracy and update timeliness.

[0339] In a possible embodiment, the present application is applied based on a complex waterway of the Yangtze River:

[0340] (1) Scene background

[0341] The Yichang to Jingzhou section of the Yangtze River (K120+300-K130+500) is a narrow curve (90 meters wide and 35° curvature) with a high navigation density (an average of 50 cargo ships / 10 passenger ships per day), and has recently seen frequent shoaling accidents.

[0342] (2) Technical solution implementation steps

[0343] 1. Construction of waterway basic fence:

[0344] Geographic feature annotation: channel type label narrow_channel, risk level label high_risk, traffic density label busy_traffic.

[0345] Width calculation: Basic width = 90 meters × 1.2 (narrow extension) × 1.2 (high-risk correction) = 129.6 meters, 64.8 meters on each side of the center line.

[0346] Priority protection: set up a core protection area (center line ± 30 meters), a monitoring area (300 meters in front × 64.8 meters), and an early warning buffer zone (60 meters outside).

[0347] 2. Minimum fence standard customization:

[0348] Ship type tag: A heavy-load cargo ship (load capacity 85% of the approved tonnage, length 110 meters, poor maneuverability) corresponds to vessel_type:cargo_ship, load_status:heavy_load, maneuverability:poor.

[0349] Parameter correction: core area radius = 80 meters × 1.3 (heavy load) = 104 meters, monitoring area length = 300 meters × 1.2 (maneuvering correction) = 360 meters, warning buffer zone = 120 meters.

[0350] 3. Navigation electronic fence integration:

[0351] Boundary value comparison: basic fence core area 60 meters vs. ship type minimum 104 meters → take 104 meters; monitoring area foundation 300 meters × 64.8 meters vs. ship type 360 ​​meters × 80 meters → take 360 ​​meters × 64.8 meters (limited by channel width).

[0352] Priority processing: High-risk tags are given priority, and the final fence is a core area of ​​104 meters, a monitoring area of ​​360 meters × 64.8 meters, and a buffer zone of 120 meters.

[0353] 4. Real-time monitoring and dynamic adjustment:

[0354] Drone monitoring: The lidar detected a shoal at K125+200 (water depth 3.2 meters, cargo ship draft 3.5 meters), generating the labels obstacle_type:shoal and distance_to_vessel:80 meters.

[0355] Rule triggering: The rule engine matches the shoal processing rules, the core area radius is expanded to 120 meters (80 meters × 1.5), the monitoring area is extended to 500 meters, and the weather tag (current wind speed is level 10) is linked to limit the ship speed to ≤8 knots.

[0356] Command execution: The ship terminal receives the speed limit command, and the drone swarm adjusts its route to focus on scanning 200 meters around the shallows.

[0357] 5. Label quality control:

[0358] Cross-validation: AIS load data (85%) is consistent with vessel registration data (rated tonnage), confirming that the load_status:heavy_load tag is accurate.

[0359] Dynamic update: After the shoal is cleared, the obstacle_type:shoal tag is deleted based on an event trigger, and a periodic calibration is performed the next day to confirm that the channel has returned to normal.

[0360] Blockchain traceability: Query the tag change log to confirm the shoal tag creation time (2025-06-01 09:00), adjustment range (core area + 16 meters) and operator (automatically triggered by the system).

[0361] (3) Implementation Effect

[0362] Dynamic Adaptation: It takes 12 seconds from detecting the shoal to adjusting the fence, which is 90% more responsive than traditional fixed fences.

[0363] Differentiated management and control: The core area radius of heavy-loaded cargo ships is 30% larger than that of conventional cargo ships, and the risk of running aground is reduced by 65%.

[0364] Data fusion: Cross-validation of multi-source data has increased labeling accuracy to 98%, and the accuracy of accident warnings has increased from 58% to 93%.

[0365] Quality assurance: Blockchain records the entire life cycle of labels, reducing manual verification workload by 70% and ensuring 99% timely label updates.

[0366] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including all changes and modifications that fall within the scope of the present invention and the preferred embodiments.

[0367] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. A method for calculating electronic fences for waterways based on drones, characterized in that: include: Setting up basic waterway fences based on the geographical characteristics of the waterway, including waterway type, risk level, and topography; Setting minimum fence standards based on vessel attributes, including vessel type, load status, and maneuverability; The navigation electronic fence of the ship is obtained by fusing the basic fence of the waterway and the lowest fence through a tag priority mechanism; The drone group collects waterway monitoring data to obtain real-time monitoring tags. The monitoring data includes obstacles, navigation mark status, meteorological and hydrological parameters; The navigation electronic fence is dynamically adjusted based on a preset rule engine and the real-time monitoring tag.

2. A method for calculating electronic fences for waterways based on drones according to claim 1, characterized in that: The setting of the waterway basic fence according to the geographical features of the waterway includes: Labeling the geographical features with waterway type labels, risk level labels, and navigation density labels; Calculate the basic protection area width according to the preset expansion coefficient of the waterway type label, risk level label and navigation density label; Priority protection is implemented for areas with high-risk labels, wherein the high-risk labels include narrow channel labels and high-risk bend labels of the channel type labels.

3. The method for calculating electronic fences for waterways based on drones according to claim 1, characterized in that: The minimum fence standards set according to the ship attributes include: Establishing a mapping relationship between a ship type tag and the minimum fence standard, wherein the ship type tag includes a cargo ship tag, a passenger ship tag, and a special operation ship tag; The minimum fence standards are revised based on the load status label and the maneuverability label, wherein the load status label includes a heavy load label and a light load label.

4. The method for calculating electronic fences for waterways based on drones according to claim 1, characterized in that: The method of fusing the waterway basic fence and the minimum fence to obtain the navigation electronic fence of the ship through the tag priority mechanism includes: Calculate the boundary value between the basic fence of the waterway and the minimum fence standard; Obtaining a final fence boundary based on the boundary value; Processing conflicting parameters based on the tag priority mechanism; The tag priority mechanism includes high-risk tags taking precedence over regular tags.

5. The method for calculating electronic fences for waterways based on drones according to claim 1, characterized in that: The dynamically adjusting the navigation electronic fence based on the preset rule engine and the real-time monitoring tag includes: triggering a fence parameter adjustment rule based on the real-time monitoring tag, wherein the real-time monitoring tag includes an obstacle type tag and a navigation mark offset tag; Expanding the adjustment range through a multi-source tag linkage mechanism, which includes a combined response mechanism of weather tags and ship type tags; Generate a fence adjustment instruction through the rule engine; The rule engine sends the fence adjustment instruction to the ship and the drone group.

6. The method for calculating electronic fences for waterways based on drones according to claim 1, characterized in that: After dynamically adjusting the navigation electronic fence based on the preset rule engine and the real-time monitoring tag, the method further includes: The accuracy of the real-time monitoring tags is ensured through a cross-validation mechanism of multi-source data, including AIS data, electronic chart data, and drone monitoring data; Establishing a dynamic update mechanism for the real-time monitoring tag, the dynamic update mechanism including event-triggered update and periodic calibration; Blockchain technology is used to achieve label traceability of the real-time monitoring label.

7. The method for calculating electronic fences for waterways based on drones according to claim 1, characterized in that: After dynamically adjusting the navigation electronic fence based on the preset rule engine and the real-time monitoring tag, the method further includes: Training a label calibration model, wherein the input of the label calibration model includes basic label data, real-time perception data, and historical behavior data; identifying an error pattern of the real-time monitoring tag based on the tag calibration model; generating correction suggestions based on the error pattern of the real-time monitoring tag; Verify and apply the correction suggestions through human-machine collaboration mechanism.

8. A waterway electronic fence system based on drones, characterized in that: include: A waterway labeling module is used to set waterway basic fences based on the geographical characteristics of the waterway, including waterway type, risk level, and topography; A ship type tag module is used to set minimum fence standards based on ship attributes, including ship type, load status, and maneuverability; A fence calculation module is used to fuse the basic fence of the waterway and the lowest fence through a tag priority mechanism to obtain the navigation electronic fence of the ship; The monitoring tag module is used to collect waterway monitoring data through the drone group to obtain real-time monitoring tags. The monitoring data includes obstacles, navigation mark status, meteorological and hydrological parameters; A rule engine module, configured to dynamically adjust the navigation electronic fence based on a preset rule engine and the real-time monitoring tag; The command execution module is used to send adjustment commands to the ship terminal and drone group.

9. The UAV-based waterway electronic fence system according to claim 8, characterized in that: The rule engine module includes: a tag matching unit, configured to identify an adjustment rule corresponding to the real-time monitoring tag; Priority determination unit, used to handle rule priorities when multiple tags conflict; A parameter calculation unit is used to calculate the correction value of the fence parameter according to the adjustment rule.

10. The UAV-based waterway electronic fence system according to claim 8, characterized in that: It also includes a label quality control module, which is used to verify label accuracy, including multi-source data cross-comparison and expert knowledge base verification; manage the label life cycle, including full process records of label creation, update, and expiration; and generate label quality assessment reports, including indicators such as label accuracy and update timeliness.

Images