Intelligent channel management method and system based on multi-source data fusion

Abstract

The invention relates to the field of channel management, in particular to an intelligent channel management method and system based on multi-source data fusion. The invention discloses an intelligent navigation channel management system based on multi-source data fusion. The system comprises a combined ship identification module, a positioning frequency updating module, a collision risk assessment module and an intervention management module. According to the invention, the system can sense multi-dimensional environmental parameters such as illumination intensity, visibility, AIS signal quality, radar echo signal-to-noise ratio and the like in real time by constructing a combined ship identification model containing AIS, radar, vision and infrared, and dynamically adjust the fusion weight of each sensor to compensate for the deficiency of a single sensor; according to the method, the robustness of ship dynamic monitoring is improved, high-precision tracking can be kept in severe weather, the data redundancy contradiction of a traditional multi-sensor system is avoided, and therefore continuous and stable target state input is provided for follow-up risk analysis.

Description

Technical Field

[0001] The present invention relates to the field of waterway management, and in particular to a waterway intelligent management method and system based on multi-source data fusion. Background Art

[0002] With the rapid development of the global shipping industry, the density of waterway traffic continues to grow, and the navigation environment is becoming increasingly complex. Faced with safety risks such as ship collisions and groundings, traditional waterway management methods often rely on a single sensor for dynamic monitoring of ships, which has significant limitations. For example, a single data source is susceptible to environmental interference, resulting in reduced positioning accuracy or even target loss. Fixed-frequency data acquisition modes are difficult to adapt to fluctuations in ship density or changes in severe weather conditions, and may result in information update delays or waste of resources. At the same time, existing collision risk assessment models are mostly based on static rule thresholds, which do not fully integrate the real-time impact of environmental interference factors such as water flow and wind speed on ship control, resulting in insufficient accuracy and timeliness of risk warnings.

[0003] In this context, existing technologies expose three core problems: First, the multi-sensor coordination capability is insufficient. Different sensors have significant performance differences in specific scenarios, but traditional solutions lack a dynamic weight allocation mechanism and cannot adapt to environmental changes; second, the data collection strategy is rigid. Existing systems usually use a fixed sampling frequency and cannot dynamically optimize resource allocation. For example, high-frequency monitoring is urgently needed when visibility is low and ships are dense, while high energy consumption still needs to be maintained during low-risk periods; third, the risk modeling dimension is single. Traditional methods rely heavily on the distance and speed between two ships when calculating collision risks, ignoring the comprehensive interference of hydrological conditions and meteorological factors on ship trajectories, resulting in deviations between risk assessment results and actual navigation threats.

[0004] To overcome the above-mentioned defects, the present invention proposes a waterway intelligent management method and system based on multi-source data fusion, which dynamically optimizes information collection and risk modeling strategies and improves the intelligence level of waterway safety management. Summary of the Invention

[0005] By constructing a joint ship identification model that includes AIS, radar, vision and infrared, the present invention enables the system to perceive multi-dimensional environmental parameters such as light intensity, visibility, AIS signal quality and radar echo signal-to-noise ratio in real time, and dynamically adjust the fusion weight of each sensor to compensate for the shortcomings of a single sensor. This improves the robustness of dynamic ship monitoring, enabling high-precision tracking in adverse weather while avoiding the data redundancy and contradictions of traditional multi-sensor systems, thereby providing continuous and stable target state input for subsequent risk analysis.

[0006] A waterway intelligent management method based on multi-source data fusion, comprising:

[0007] A joint ship identification model is constructed that incorporates AIS, radar, visual sensors, and infrared thermal imaging. At any given positioning time point, current data accuracy influencing indicators, including light intensity, visibility, AIS signal reception strength, and radar echo signal-to-noise ratio, are collected. Based on these data accuracy influencing indicators, the weights of each data source in the joint ship identification model are adjusted. The joint ship identification model and the adjusted weights of each data source are then applied to obtain vessel information within the target management channel, including the position, speed, and draft of each vessel within the target management channel.

[0008] Set the frequency update cycle. At the beginning of any frequency update cycle, obtain the current channel information of the target managed channel, including the number of ships, water velocity, visibility and wind speed, and adjust the collection frequency of the joint ship identification model based on the channel information;

[0009] At the beginning of any frequency update period, for any ship, a collision risk score is calculated based on the distance and relative speed between the ship and the nearest ship, and the collision risk score is corrected by water velocity, visibility and wind speed;

[0010] Based on the collision risk scores of all ships, key concern ships are screened out and intervention management measures are implemented on all key concern ships.

[0011] Preferably, the weight of each data source in the joint ship identification model is adjusted based on the data accuracy impact index. The specific operation is as follows:

[0012] Set the maximum light intensity, maximum visibility, ideal peak value of AIS signal reception intensity, and ideal peak value of radar echo signal-to-noise ratio respectively. Use these values ​​to normalize the accuracy influencing indicators of the currently collected data to obtain the light intensity score, visibility score, AIS signal score, and radar echo score.

[0013] Set basic weight values ​​for AIS, radar, visual sensor, and infrared thermal imaging respectively, indicating the weight of each data source in its corresponding optimal working environment;

[0014] For AIS, the AIS signal score is directly used as the AIS weight adjustment coefficient. The product of the AIS basic weight value and the AIS weight adjustment coefficient is calculated to obtain the AIS weight value.

[0015] For radar, the radar echo score is directly used as the radar weight adjustment coefficient. The product of the radar's basic weight value and the radar weight adjustment coefficient is calculated to obtain the radar weight value.

[0016] For the visual sensor, the average of the light intensity score and the visibility score is used as the visual weight adjustment coefficient. The product of the basic weight value of the visual sensor and the visual weight adjustment coefficient is calculated to obtain the visual weight value.

[0017] For infrared thermal imaging, the average of the light intensity score and the visibility score is calculated, and the difference between the average and 1 is used as the infrared weight adjustment coefficient of the infrared thermal imaging. The product of the basic weight value of the infrared thermal imaging and the infrared weight adjustment coefficient is calculated to obtain the infrared weight value.

[0018] Preferably, a joint ship identification model and the adjusted weights of each data source are applied to obtain the ship information in the target management channel. The ship information includes the position, speed and draft of each ship in the target management channel, wherein the ship position adopts the weighted average method, and the real-time coordinates of each data source are superimposed according to the weight. The real-time coordinates include the GPS positioning of AIS, the polar coordinate conversion of radar, the pixel inversion position of vision and the heat source positioning of infrared; the speed is obtained by the weighted fusion of the SOG speed of AIS, the radar Doppler speed measurement, and the speed tracking by the visual optical flow method; the draft uses the visual water mark line recognition as the main weight data, and the radar-assisted ranging reverse draft as the secondary weight. When the vision fails due to low light, it automatically switches to the infrared enhanced image.

[0019] Preferably, the acquisition frequency of the joint ship identification model is adjusted based on the channel information, and the specific operations are as follows:

[0020] Set the minimum and maximum collection frequencies for indicators affecting data accuracy, and set the maximum number of ships, maximum water velocity, maximum visibility, and maximum wind speed respectively. Use the maximum number of ships, maximum water velocity, maximum visibility, and maximum wind speed to normalize each indicator in the currently collected channel information, and use the normalized results of each indicator in the channel information to obtain the frequency control coefficient.

[0021] The frequency control coefficient is linearly mapped between the minimum acquisition frequency and the maximum acquisition frequency to obtain the adjusted acquisition frequency. Subsequently, within the current frequency update cycle, the adjusted acquisition frequency is applied to collect indicators affecting data accuracy until the next frequency update cycle begins.

[0022] Preferably, the collision risk score of the vessel is calculated and corrected by the water velocity, visibility and wind speed, as follows:

[0023] For any ship, based on the distance and relative speed between the ship and the nearest ship; using the formula Calculate the basic collision risk score of the ship, where k is the preset attenuation coefficient, with a default value of 0.01, which can be adjusted by professionals according to actual conditions; d is the distance in meters; d0 is equal to 1 meter, which is used to digitize the distance; Δv is the relative speed, v max speed limits for waterways;

[0024] Calculate the average of the normalized results of the current water velocity, visibility, and wind speed to obtain the correction factor, and then apply the correction factor to correct the basic collision risk score. The specific formula is as follows:

[0025] R′=R(1+C)

[0026] Where C represents the correction coefficient; R′ is the corrected collision risk score of the current ship, and the collision risk score ranges from 0 to 2.

[0027] Preferably, focus on the ships to be screened out. The specific operations are as follows:

[0028] Set a risk threshold and mark ships with collision risk scores higher than the risk threshold as ships of special concern at the beginning of any frequency update cycle;

[0029] The value of the risk threshold is determined by the particle swarm optimization algorithm, which specifically includes the following steps:

[0030] Step 1: Randomly generate a particle swarm within the range of 0-2. The position of each particle represents the value of a risk threshold, and the maximum number of iterations is set. For any particle, initialize the particle's velocity and use the current position as the individual optimal position of the particle. Define the objective function with the optimization goals of minimizing the consumption of monitoring resources and minimizing the incidence of ship collision accidents, and calculate the objective function value of the particle. Traverse all particles and take the position corresponding to the particle with the highest objective function value as the global optimal position.

[0031] Step 2: For any particle, update the speed and position of the particle and recalculate the objective function value; if the updated objective function value is lower than the objective function value of the particle's individual best position, update the particle's individual best position to the current position; traverse all particles, and if there is any particle whose individual best position has a fitness value lower than the global best position, update the global best position to the particle's individual best position;

[0032] Step 3: Repeat step 2 until the maximum number of iterations is reached. The global optimal position obtained is the optimal value of the risk threshold.

[0033] A waterway intelligent management system based on multi-source data fusion, including:

[0034] The joint ship identification module includes a weight adjustment unit and a joint ship identification unit. The weight adjustment unit is used to collect the current data accuracy impact index at any positioning time point and adjust the weights of each data source in the joint ship identification model. The joint ship identification unit is used to construct the joint ship identification model and then apply the joint ship identification model and the adjusted weights of each data source to obtain ship information in the target management channel.

[0035] The positioning frequency update module is used to set the frequency update cycle. At the beginning of any frequency update cycle, the current channel information of the target management channel is obtained and the acquisition frequency of the joint ship identification model is adjusted based on the channel information.

[0036] A collision risk assessment module is used to calculate a collision risk score for any vessel at the beginning of any frequency update period based on the distance and relative speed between the vessel and the nearest vessel, and to modify the collision risk score based on water velocity, visibility and wind speed;

[0037] The intervention management module is used to screen out key ships based on the collision risk scores of all ships and implement intervention management measures on all key ships.

[0038] The present invention has the following advantages:

[0039] 1. By constructing a joint ship identification model that includes AIS, radar, vision, and infrared, the present invention enables the system to perceive multi-dimensional environmental parameters such as light intensity, visibility, AIS signal quality, and radar echo signal-to-noise ratio in real time, and dynamically adjust the fusion weights of each sensor to compensate for the shortcomings of a single sensor. This improves the robustness of dynamic ship monitoring, enabling high-precision tracking in adverse weather while avoiding the data redundancy and contradictions of traditional multi-sensor systems, thereby providing continuous and stable target state input for subsequent risk analysis.

[0040] 2. This invention dynamically corrects the collision risk threshold by quantifying the impact of interference factors such as water resistance and wind pressure deviation on the ship's trajectory. At the same time, it innovatively links channel density, meteorological conditions, and data processing requirements to autonomously adjust the data collection frequency and computing resource allocation strategy. This adaptive control mechanism not only effectively reduces resource consumption in low-risk scenarios, but also can quickly switch to enhanced monitoring mode when ships are densely packed or visibility drops sharply, achieving an optimal balance of performance across the entire chain from data collection to risk warning, breaking through the performance bottleneck of traditional fixed architectures. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 This is a flow chart of the waterway intelligent management method based on multi-source data fusion adopted in an embodiment of the present invention.

[0042] Figure 2 This is a structural diagram of the waterway intelligent management system based on multi-source data fusion adopted in an embodiment of the present invention. DETAILED DESCRIPTION

[0043] In order to enable those skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0044] Example 1, a waterway intelligent management method based on multi-source data fusion, such as Figure 1 Shown, including:

[0045] A joint ship identification model is constructed that incorporates AIS, radar, visual sensors, and infrared thermal imaging. At any given positioning time point, current data accuracy influencing indicators, including light intensity, visibility, AIS signal reception strength, and radar echo signal-to-noise ratio, are collected. Based on these data accuracy influencing indicators, the weights of each data source in the joint ship identification model are adjusted. The joint ship identification model and the adjusted weights of each data source are then applied to obtain vessel information within the target management channel, including the position, speed, and draft of each vessel within the target management channel.

[0046] Set the frequency update cycle. At the beginning of any frequency update cycle, obtain the current channel information of the target managed channel, including the number of ships, water velocity, visibility and wind speed. Based on the channel information, adjust the collection frequency of the joint ship identification model (i.e., adjust the time interval between positioning time points).

[0047] At the beginning of any frequency update period, for any ship, a collision risk score is calculated based on the distance and relative speed between the ship and the nearest ship, and the collision risk score is corrected by water velocity, visibility and wind speed;

[0048] Based on the collision risk scores of all ships, key focus ships are screened out, and intervention management measures are implemented for all key focus ships; intervention management measures may include multi-dimensional control measures such as active heading and speed guidance and compulsory risk avoidance coordination: voice / text instructions such as deceleration and steering are sent through VHF automatic broadcast or AIS module; electronic chart warning layer is triggered and recommended routes for traffic separation are pushed; high-risk targets that fail to respond in time are automatically subject to remote speed limit control or navigation control notices are issued through the maritime supervision platform; at the same time, a ship-shore collaborative collision avoidance channel is constructed, and high-precision trajectory prediction and route game algorithm are combined to issue coordinated avoidance routes to related ships, and the collision avoidance execution effect is monitored in real time; in addition, the manual intervention review mechanism is triggered to automatically generate a risk disposal recommendation report containing the ship's historical violation records and real-time navigation parameters to the duty center, realizing the integrated response of intelligent and manual decision-making.

[0049] The weights of each data source in the joint ship identification model are adjusted based on the data accuracy impact index. The specific operations are as follows:

[0050] The maximum light intensity, maximum visibility, ideal peak value of AIS signal reception intensity, and ideal peak value of radar echo signal-to-noise ratio are set respectively. These values ​​are then used to normalize the accuracy influencing indicators of the currently collected data to obtain the light intensity score, visibility score, AIS signal score, and radar echo score. The specific steps of normalization are: each actual collected data value is divided by its corresponding ideal peak reference value; the result of this division operation is a value between 0 and 1. Through this operation, the original raw data with huge differences in dimensions and numerical ranges is converted into a unified, dimensionless score, which facilitates subsequent comparison, weighting, or comprehensive evaluation.

[0051] Set basic weight values ​​for AIS, radar, visual sensor, and infrared thermal imaging respectively, indicating the weight of each data source in its corresponding optimal working environment;

[0052] For AIS, the AIS signal score is directly used as the AIS weight adjustment coefficient. The product of the AIS basic weight value and the AIS weight adjustment coefficient is calculated to obtain the AIS weight value.

[0053] For radar, the radar echo score is directly used as the radar weight adjustment coefficient. The product of the radar's basic weight value and the radar weight adjustment coefficient is calculated to obtain the radar weight value.

[0054] For the visual sensor, the average of the light intensity score and the visibility score is used as the visual weight adjustment coefficient. The product of the basic weight value of the visual sensor and the visual weight adjustment coefficient is calculated to obtain the visual weight value.

[0055] For infrared thermal imaging, the average of the light intensity score and the visibility score is calculated, and the difference between the average and 1 is used as the infrared weight adjustment coefficient of the infrared thermal imaging. The product of the basic weight value of the infrared thermal imaging and the infrared weight adjustment coefficient is calculated to obtain the infrared weight value.

[0056] The joint ship identification model and the adjusted weights of each data source are used to obtain the ship information in the target management channel. The ship information includes the position, speed and draft of each ship in the target management channel. Among them, the ship position adopts the weighted average method, and the real-time coordinates of each data source are superimposed according to the weight, including AIS GPS positioning, radar polar coordinate conversion, visual pixel inversion position and infrared heat source positioning, to dynamically compensate for the signal drift or blind spot error of a single data source; the speed is achieved through the weighted fusion of AIS SOG speed, radar Doppler speed measurement, and visual optical flow method tracking speed; the draft uses visual water mark recognition as the main weight data, and radar-assisted ranging reverse draft as the secondary weight. When the vision fails due to low light, it automatically switches to infrared enhanced image and radar data for cross-verification; the weight ratio is updated in real time throughout the process to ensure that stable and reliable ship dynamic information can still be output in scenarios such as severe weather and signal interference, providing core data support for intelligent channel scheduling and safety warning.

[0057] Adjust the acquisition frequency of the joint ship identification model based on the channel information. The specific operations are as follows:

[0058] Set the minimum and maximum collection frequencies for indicators that affect data accuracy. Set the maximum number of ships, maximum water velocity, maximum visibility, and maximum wind speed respectively. Apply the maximum number of ships, maximum water velocity, maximum visibility, and maximum wind speed to normalize each indicator in the currently collected channel information. Use the normalized results of each indicator in the channel information to obtain the frequency control coefficient. The specific formula is as follows:

[0059]

[0060] Among them, h is the frequency control coefficient, N, U, W, and V represent the current number of ships, water velocity, visibility, and wind speed respectively; N max 、U max 、W max 、V max represents the maximum number of ships, the highest water velocity, the highest visibility, and the highest wind speed, respectively; ω1, ω2, ω3, and ω4 are the influence weights of the number of ships, water velocity, visibility, and wind speed, respectively, and the sum of ω1, ω2, ω3, and ω4 is 1;

[0061] The frequency control coefficient is linearly mapped between the minimum acquisition frequency and the maximum acquisition frequency to obtain the adjusted acquisition frequency. Subsequently, within the current frequency update cycle, the adjusted acquisition frequency is applied to collect indicators affecting data accuracy until the next frequency update cycle begins.

[0062] Calculate the collision risk score of the vessel and adjust it by water velocity, visibility and wind speed as follows:

[0063] For any ship, based on the distance and relative speed between the ship and the nearest ship; using the formula Calculate the basic collision risk score of the ship, where k is the preset attenuation coefficient, the default value is 0.01, which can be adjusted by professionals according to actual conditions; D is the distance in meters; D0 is equal to 1 meter, which is used to digitize the distance; Δv is the relative speed, v max speed limits for waterways;

[0064] Calculate the average of the normalized results of the current water velocity, visibility, and wind speed to obtain the correction factor, and then apply the correction factor to correct the basic collision risk score. The specific formula is as follows:

[0065] R′=R(1+C)

[0066] Where C represents the correction coefficient; R′ is the corrected collision risk score of the current ship, and the collision risk score ranges from 0 to 2;

[0067] The design of the collision risk calculation formula is based on the core principles of distance attenuation effect and speed threat superposition. It quantifies the urgency of ship spacing through an exponential function (the closer the distance, the steeper the risk), and combines the ratio of relative speed to the channel speed limit to reflect the intensity of the dynamic threat. It further introduces environmental correction factors (such as water flow deviation, visibility shortening reaction time, and wind-induced drift) to convert the interference of natural conditions on ship control into a risk coefficient multiplier effect. Ultimately, a multi-dimensional coupled assessment model is formed, which not only reflects static spatial relationships but also dynamically responds to complex environmental changes, realizing a scientific mapping from physical parameters to risk probabilities.

[0068] Filter out ships of key concern. The specific operations are as follows:

[0069] Set a risk threshold and mark ships with collision risk scores higher than the risk threshold as ships of special concern at the beginning of any frequency update cycle;

[0070] The value of the risk threshold is determined by the particle swarm optimization algorithm, which specifically includes the following steps:

[0071] Step 1: Randomly generate a particle swarm within the range of 0-2. The position of each particle represents the value of a risk threshold, and the maximum number of iterations is set. For any particle, initialize the particle's velocity and use the current position as the individual optimal position of the particle. Define the objective function with the optimization goals of minimizing the consumption of monitoring resources and minimizing the incidence of ship collision accidents, and calculate the objective function value of the particle. Traverse all particles and take the position corresponding to the particle with the highest objective function value as the global optimal position.

[0072] Step 2: For any particle, update the speed and position of the particle and recalculate the objective function value; if the updated objective function value is lower than the objective function value of the particle's individual best position, update the particle's individual best position to the current position; traverse all particles, and if there is any particle whose individual best position has a fitness value lower than the global best position, update the global best position to the particle's individual best position;

[0073] Step 3: Repeat step 2 until the maximum number of iterations is reached. The global optimal position obtained is the optimal value of the risk threshold.

[0074] Example 2, a waterway intelligent management system based on multi-source data fusion, such as Figure 2 Shown, including:

[0075] The joint ship identification module includes a weight adjustment unit and a joint ship identification unit. The weight adjustment unit is used to collect the current data accuracy influencing indicators, including light intensity, visibility, AIS signal reception strength, and radar echo signal-to-noise ratio, at any positioning time point. The weights of each data source in the joint ship identification model are adjusted based on the data accuracy influencing indicators. The joint ship identification unit is used to construct a joint ship identification model that includes AIS, radar, visual sensors, and infrared thermal imaging. The joint ship identification model and the adjusted weights of each data source are then applied to obtain ship information in the target management channel. The ship information includes the position, speed, and draft of each ship in the target management channel.

[0076] The positioning frequency update module is used to set the frequency update cycle. At the beginning of any frequency update cycle, the module obtains the current channel information of the target management channel, including the number of ships, water velocity, visibility and wind speed, and adjusts the acquisition frequency of the joint ship identification model based on the channel information.

[0077] A collision risk assessment module is used to calculate a collision risk score for any vessel at the beginning of any frequency update period based on the distance and relative speed between the vessel and the nearest vessel, and to modify the collision risk score based on water velocity, visibility and wind speed;

[0078] The intervention management module is used to screen out key ships based on the collision risk scores of all ships and implement intervention management measures on all key ships.

[0079] It should be understood that those skilled in the art may make improvements or modifications based on the above description, and all such improvements and modifications shall fall within the scope of protection of the appended claims. Any portion of this specification not described in detail is prior art known to those skilled in the art.

Claims

1. A waterway intelligent management method based on multi-source data fusion, characterized in that: include: A joint ship identification model is constructed that incorporates AIS, radar, visual sensors, and infrared thermal imaging. At any given positioning time point, current data accuracy influencing indicators, including light intensity, visibility, AIS signal reception strength, and radar echo signal-to-noise ratio, are collected. Based on these data accuracy influencing indicators, the weights of each data source in the joint ship identification model are adjusted. The joint ship identification model and the adjusted weights of each data source are then applied to obtain vessel information within the target management channel, including the position, speed, and draft of each vessel within the target management channel. Set the frequency update cycle. At the beginning of any frequency update cycle, obtain the current channel information of the target managed channel, including the number of ships, water velocity, visibility and wind speed, and adjust the collection frequency of the joint ship identification model based on the channel information; At the beginning of any frequency update period, for any ship, a collision risk score is calculated based on the distance and relative speed between the ship and the nearest ship, and the collision risk score is corrected by water velocity, visibility and wind speed; Based on the collision risk scores of all ships, key concern ships are screened out and intervention management measures are implemented on all key concern ships.

2. The method for intelligent waterway management based on multi-source data fusion according to claim 1, characterized in that: The weights of each data source in the joint ship identification model are adjusted based on the data accuracy impact index. The specific operations are as follows: Set the maximum light intensity, maximum visibility, ideal peak value of AIS signal reception intensity, and ideal peak value of radar echo signal-to-noise ratio respectively. Use these values ​​to normalize the accuracy influencing indicators of the currently collected data to obtain the light intensity score, visibility score, AIS signal score, and radar echo score. Set basic weight values ​​for AIS, radar, visual sensor, and infrared thermal imaging respectively, indicating the weight of each data source in its corresponding optimal working environment; For AIS, the AIS signal score is directly used as the AIS weight adjustment coefficient. The product of the AIS basic weight value and the AIS weight adjustment coefficient is calculated to obtain the AIS weight value. For radar, the radar echo score is directly used as the radar weight adjustment coefficient. The product of the radar's basic weight value and the radar weight adjustment coefficient is calculated to obtain the radar weight value. For the visual sensor, the average of the light intensity score and the visibility score is used as the visual weight adjustment coefficient. The product of the basic weight value of the visual sensor and the visual weight adjustment coefficient is calculated to obtain the visual weight value. For infrared thermal imaging, the average of the light intensity score and the visibility score is calculated, and the difference between the average and 1 is used as the infrared weight adjustment coefficient of the infrared thermal imaging. The product of the basic weight value of the infrared thermal imaging and the infrared weight adjustment coefficient is calculated to obtain the infrared weight value.

3. The waterway intelligent management method based on multi-source data fusion according to claim 2 is characterized in that: The joint ship identification model and the adjusted weights of each data source are applied to obtain the ship information in the target management channel. The ship information includes the position, speed and draft of each ship in the target management channel. Among them, the ship position adopts the weighted average method, and the real-time coordinates of each data source are superimposed according to the weight. The real-time coordinates include GPS positioning of AIS, polar coordinate conversion of radar, pixel inversion position of vision and heat source positioning of infrared; the speed is obtained through the weighted fusion of SOG speed of AIS, radar Doppler speed measurement, and speed tracking by visual optical flow method; the draft uses visual water mark recognition as the main weight data, and radar-assisted ranging reverse draft as the secondary weight. When the vision fails due to low light, it automatically switches to infrared enhanced image.

4. The method for intelligent waterway management based on multi-source data fusion according to claim 3 is characterized in that: Adjust the acquisition frequency of the joint ship identification model based on the channel information. The specific operations are as follows: Set the minimum and maximum collection frequencies for indicators affecting data accuracy, and set the maximum number of ships, maximum water velocity, maximum visibility, and maximum wind speed respectively. Use the maximum number of ships, maximum water velocity, maximum visibility, and maximum wind speed to normalize each indicator in the currently collected channel information, and use the normalized results of each indicator in the channel information to obtain the frequency control coefficient. The frequency control coefficient is linearly mapped between the minimum acquisition frequency and the maximum acquisition frequency to obtain the adjusted acquisition frequency. Subsequently, within the current frequency update cycle, the adjusted acquisition frequency is applied to collect indicators affecting data accuracy until the next frequency update cycle begins.

5. The method for intelligent waterway management based on multi-source data fusion according to claim 4 is characterized in that: Calculate the collision risk score of the vessel and adjust it by water velocity, visibility and wind speed as follows: For any ship, based on the distance and relative speed between the ship and the nearest ship; using the formula Calculate the basic collision risk score of the ship, where k is the preset attenuation coefficient, which is adjusted by professionals according to actual conditions; D is the distance in meters; D0 is equal to 1 meter, which is used to digitize the distance; Δv is the relative speed, v max speed limits for waterways; Calculate the average of the normalized results of the current water velocity, visibility, and wind speed to obtain the correction factor, and then apply the correction factor to correct the basic collision risk score. The specific formula is as follows: R′=R(1+C) Where C represents the correction coefficient; R′ is the corrected collision risk score of the current ship, and the collision risk score ranges from 0 to 2.

6. The method for intelligent waterway management based on multi-source data fusion according to claim 5, characterized in that: Filter out ships of key concern. The specific operations are as follows: Set a risk threshold and mark ships with collision risk scores higher than the risk threshold as ships of special concern at the beginning of any frequency update cycle; The value of the risk threshold is determined by the particle swarm optimization algorithm, which specifically includes the following steps: Step 1: Randomly generate a particle swarm within the range of 0-2. The position of each particle represents the value of a risk threshold, and the maximum number of iterations is set. For any particle, initialize the particle's velocity and use the current position as the individual optimal position of the particle. Define the objective function with the optimization goals of minimizing the consumption of monitoring resources and minimizing the incidence of ship collision accidents, and calculate the objective function value of the particle. Traverse all particles and take the position corresponding to the particle with the highest objective function value as the global optimal position. Step 2: For any particle, update the speed and position of the particle and recalculate the objective function value; if the updated objective function value is lower than the objective function value of the particle's individual best position, update the particle's individual best position to the current position; traverse all particles, and if there is any particle whose individual best position has a fitness value lower than the global best position, update the global best position to the particle's individual best position; Step 3: Repeat step 2 until the maximum number of iterations is reached. The global optimal position obtained is the optimal value of the risk threshold.

7. A waterway intelligent management system based on multi-source data fusion, characterized in that: The system is applied to a waterway intelligent management method based on multi-source data fusion as described in any one of claims 1 to 6 above, comprising: The joint ship identification module includes a weight adjustment unit and a joint ship identification unit. The weight adjustment unit is used to collect the current data accuracy impact index at any positioning time point and adjust the weights of each data source in the joint ship identification model. The joint ship identification unit is used to construct the joint ship identification model and then apply the joint ship identification model and the adjusted weights of each data source to obtain ship information in the target management channel. The positioning frequency update module is used to set the frequency update cycle. At the beginning of any frequency update cycle, the current channel information of the target management channel is obtained and the acquisition frequency of the joint ship identification model is adjusted based on the channel information. A collision risk assessment module is used to calculate a collision risk score for any vessel at the beginning of any frequency update period based on the distance and relative speed between the vessel and the nearest vessel, and to modify the collision risk score based on water velocity, visibility and wind speed; The intervention management module is used to screen out key ships based on the collision risk scores of all ships and implement intervention management measures on all key ships.

Images