Water LED formation display method and system based on group intelligence collaboration

Abstract

The present application discloses a method and system for water LED formation display based on group intelligence collaboration, the method includes: the arrangement method includes: obtaining the real-time position information of the unmanned ship; generating formation pattern data according to the task requirements and real-time position information; scanning the site to obtain scene information data, performing path planning and design according to the scene information data and the formation pattern data, and obtaining the unmanned ship path arrangement set; the group intelligence collaboration method includes: collecting the environmental data, motion data and obstacle sensor data of the unmanned ship during the movement process; performing natural group simulation according to the unmanned ship path arrangement set and motion data, and obtaining natural control data. This solution ensures that the unmanned ship always maintains a stable and efficient performance effect in the continuous formation display, and effectively solves the problem of insufficient natural coordination of the unmanned ship in the water LED formation display.

Description

Technical Field

[0001] The present application relates to the technical field of unmanned control systems, and in particular to a method and system for displaying an LED formation on water based on group intelligence collaboration. Background Art

[0002] With the rapid development of unmanned boat technology, water LED formation display has gradually become an emerging form of artistic expression. However, in the current application of water LED formation display, despite the continuous advancement of technology, unmanned boats often face the problem of insufficient natural coordination when performing formation tasks in complex water environments. This defect stems from the influence of various external factors, such as dynamic environmental conditions such as water flow, wind speed and waves. These factors not only interfere with the movement trajectory of the unmanned boat, but also lead to an imbalance in the interaction between the formation members, thereby weakening the overall coordination of the display effect.

[0003] This unnatural coordination is manifested in the deviation of the position of the unmanned boats in the formation, uneven speed and delayed action, which ultimately reduces the visual impact felt by the audience and greatly reduces the performance effect. Especially in large-scale events or important celebrations, the audience has higher expectations for visual performance. Once the formation is out of control or chaotic, it will inevitably affect their viewing experience and the overall atmosphere of the event. In addition, frequent light flashes and color changes, if not properly controlled, will inevitably cause visual fatigue to the audience, further weakening the appeal and artistic value of the display. Summary of the invention

[0004] In order to solve the above problems, an embodiment of the present invention provides a method for displaying a water LED formation based on group intelligence collaboration, the method including an arrangement method, a group intelligence collaboration method and a lighting adaptation method;

[0005] The arrangement method comprises:

[0006] Obtain the real-time location information of the unmanned ship; generate formation pattern data according to the task requirements and the real-time location information; scan the site to obtain scene information data, perform path planning and design according to the scene information data and the formation pattern data, and obtain the unmanned ship path arrangement set;

[0007] The group intelligence collaboration method comprises:

[0008] Collect environmental data, motion data and obstacle sensor data of the unmanned boat during its movement; perform natural group simulation based on the unmanned boat path arrangement set and motion data to obtain natural control data; perform reset control based on the natural control data, environmental data and obstacle sensor data to obtain reset adjustment data, and the unmanned boat performs posture control based on the reset adjustment data; the method for obtaining the reset adjustment data includes:

[0009] Input the natural control data, environmental data and obstacle sensing data into the pre-built reset control model, and output the reset adjustment data;

[0010] The training method of the reset control model includes:

[0011] The natural control data, environmental data and obstacle sensor data are used as inputs of the reset control model. The reset control model uses the reset adjustment data corresponding to the prediction of each set of natural control data, environmental data and obstacle sensor data as output, uses the actual reset adjustment data corresponding to each set of natural control data, environmental data and obstacle sensor data as the prediction target, and uses minimizing the sum of the first prediction accuracies of all predicted reset adjustment data as the training target; the reset control model is trained until the sum of the first prediction accuracies reaches convergence and the training is stopped; the reset control model is a long short-term memory network model;

[0012] The lighting adaptation method comprises:

[0013] Collect background data, input the formation pattern data, background data and timestamp into the pre-built visual fatigue prediction model, and output the lighting control data; adjust the lights of the unmanned ship according to the lighting control data.

[0014] Furthermore, the method for acquiring the formation pattern data includes algorithm generation and picture input;

[0015] The algorithm generation method comprises:

[0016] Set necessary image generation parameters;

[0017] Create a two-dimensional grid array where each cell corresponds to a point on the complex plane;

[0018] Perform complex mapping, iterative calculation and escape determination on each pixel;

[0019] Complex mapping: Map the coordinates of each pixel point to a point in the complex plane. The calculation method includes:

[0020] ;

[0021] In the formula, represents a point on the complex plane; is the horizontal coordinate of the pixel; The vertical coordinate of the pixel; is the corresponding minimum in the complex plane value; is the corresponding maximum value in the complex plane value; The corresponding minimum in the complex plane value; is the corresponding maximum value in the complex plane value; is the width of the image; is the height of the image; is an imaginary unit;

[0022] The calculation methods for iterative calculation include:

[0023] ;

[0024] In the formula, For the The complex value after iterations; is the number of iterations;

[0025] Escape Detection: Calculating Complex Values If the modulus is greater than 2, then Escaped, record the number of iterations of this point, store the number of iterations of each pixel in the image array, generate the image, map the image array to the generated image, and assign a color to each pixel based on the number of iterations.

[0026] Furthermore, the method for obtaining the naturalness control data includes:

[0027] Allocate initial natural parameters of the unmanned ship based on the unmanned ship path arrangement set, the natural parameters include position vector, velocity vector and field of view;

[0028] The natural swarm simulation algorithm is run once at each time step to update the position vector and velocity vector of the unmanned ship and obtain natural control data.

[0029] Furthermore, the natural population simulation algorithm includes:

[0030] Taking the unmanned ship as a single object, all unmanned ships whose distance is smaller than the field of view of the unmanned ship are marked as neighbors;

[0031] Calculate the behavior forces of all unmanned ships based on the motion data of the neighbors. The motion data includes velocity vector, acceleration vector and position vector. The behavior forces include separation force, alignment force and aggregation force.

[0032] Separation force: Calculate the distance between all neighbors and the current unmanned ship, and compare the distance with the preset minimum safety distance. If the distance is greater than or equal to the preset minimum safety distance, there is no need to calculate the separation force. If the distance is less than the preset minimum safety distance, it is marked as avoidance and the separation force vector is calculated at the same time. :

[0033] ;

[0034] In the formula, is the current position vector of the unmanned ship, is the position vector of the neighbor, For normalization, is the current distance between the unmanned ship and its neighbor;

[0035] Alignment force: Accumulate the current speed vectors of all neighbors and calculate the average speed vector of all neighbors to obtain the average speed vector of all neighbors. Calculate the difference between the current speed vector of the unmanned ship and the average speed vector to obtain the alignment force.

[0036] Aggregation power: Accumulate the current position vectors of all neighbors and calculate the average position vector of all neighbors to obtain the average position vector of all neighbors. Calculate the difference between the current position vector of the unmanned ship and the average position vector to obtain the aggregation power.

[0037] Furthermore, the background data includes light intensity, background complexity and viewing distance;

[0038] The method for obtaining the background complexity includes:

[0039] Input the environment photo into the pre-built visual processing model and output the background complexity;

[0040] Methods for training visual processing models include:

[0041] The environmental photos are used as the input of the visual processing model. The visual processing model takes the predicted background complexity corresponding to each group of environmental photos as the output, takes the actual background complexity corresponding to each group of environmental photos as the prediction target, and takes minimizing the sum of the second prediction accuracies of all predicted background complexities as the training target; the visual processing model is trained until the sum of the second prediction accuracies reaches convergence and the training is stopped; the visual processing model is a convolutional neural network model.

[0042] Furthermore, the training method of the visual fatigue prediction model includes:

[0043] The formation pattern data, background data and timestamp are used as inputs of the visual fatigue prediction model. The visual fatigue prediction model takes the lighting control data corresponding to each set of formation pattern data, background data and timestamp predictions as outputs, and takes the actual lighting control data corresponding to each set of formation pattern data, background data and timestamp as prediction targets. The training target is to minimize the sum of the third prediction accuracies of all predicted lighting control data. The visual fatigue prediction model is trained until the sum of the third prediction accuracies converges. The training is stopped. The visual fatigue prediction model is a decision tree model.

[0044] The water LED formation display system based on group intelligence collaboration includes a scheduling module, a group intelligence collaboration module and a lighting adaptation module;

[0045] The arrangement module includes a positioning and navigation unit, a formation arrangement unit and a site path planning unit; the positioning and navigation unit is used to obtain the real-time position information of the unmanned ship; the formation arrangement unit is used to generate formation pattern data according to the task requirements and the real-time position information; the site path planning unit is used to scan the site, obtain the scene information data, perform path planning and design according to the scene information data and the formation pattern data, and obtain the unmanned ship path arrangement set;

[0046] The group intelligence collaboration module includes a dynamic data collection unit, a natural simulation unit and a reset control unit; the dynamic data collection unit is used to collect environmental data, motion data and obstacle sensor data of the unmanned ship during movement; the natural simulation unit is used to perform natural group simulation according to the unmanned ship path arrangement set and motion data to obtain natural control data; the reset control unit is used to perform reset control according to the natural control data, environmental data and obstacle sensor data to obtain reset adjustment data, and the unmanned ship performs posture control according to the reset adjustment data; the method for obtaining the reset adjustment data includes:

[0047] Input the natural control data, environmental data and obstacle sensing data into the pre-built reset control model, and output the reset adjustment data;

[0048] The training method of the reset control model includes:

[0049] The natural control data, environmental data and obstacle sensor data are used as inputs of the reset control model. The reset control model uses the reset adjustment data corresponding to the prediction of each set of natural control data, environmental data and obstacle sensor data as output, uses the actual reset adjustment data corresponding to each set of natural control data, environmental data and obstacle sensor data as the prediction target, and uses minimizing the sum of the first prediction accuracies of all predicted reset adjustment data as the training target; the reset control model is trained until the sum of the first prediction accuracies reaches convergence and the training is stopped; the reset control model is a long short-term memory network model;

[0050] The lighting adaptation module includes a visual fatigue prediction unit and a lighting control unit; the visual fatigue prediction unit is used to collect background data, input formation pattern data, background data and timestamp into a pre-built visual fatigue prediction model, and output lighting control data; the lighting control unit is used to control the unmanned ship lighting according to the lighting control data.

[0051] The technical effects and advantages of the water LED formation display system based on group intelligence collaboration provided by the present invention are as follows:

[0052] This solution integrates machine learning and intelligent control technologies. Through models such as long short-term memory networks and convolutional neural networks, it improves the response speed and decision-making accuracy of unmanned ships in complex environments. It can identify and adapt to environmental changes in real time, ensuring that unmanned ships always maintain stable and efficient performance effects in continuous formation displays, and effectively solves the problem of insufficient natural coordination of unmanned ships in water LED formation displays.

[0053] This solution uses dynamic data collection and natural group simulation algorithms to obtain the position information and environmental data of unmanned ships in real time, thereby realizing real-time adaptive control. By accurately calculating the relative positions and speeds between formation members, it ensures that the movement of unmanned ships in complex water environments is more coordinated and natural, thereby improving the overall display effect. Combining environmental dynamic monitoring and obstacle sensing data, a reset control strategy is adopted to enable unmanned ships to quickly adjust their course and speed when facing external interference such as wind and waves, maintain the stability and consistency of the formation, and effectively reduce the impact of environmental factors on the performance of the unmanned ship formation, ensuring the reliability and smoothness of the formation. By establishing a visual fatigue prediction model, combined with environmental changes and audience feedback, an adaptive lighting control strategy is generated to achieve dynamic adjustment of lighting brightness, color and flashing mode. This strategy not only improves the coordination of lighting performance, but also effectively reduces the audience's visual fatigue, making the display more attractive and artistic. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 This is a connection diagram of the water LED formation display system based on group intelligence collaboration in Example 1;

[0055] Figure 2 This is a flow chart of the natural population simulation method in Example 1;

[0056] Figure 3 This is a flow chart of the method for displaying an LED formation on water based on group intelligence collaboration in Example 2. DETAILED DESCRIPTION

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

[0058] Embodiment 1:

[0059] See also Figure 1 As shown, the water LED formation display system based on group intelligence collaboration described in this embodiment includes an arrangement module, a group intelligence collaboration module and a lighting adaptation module;

[0060] The arrangement module includes a positioning and navigation unit, a formation arrangement unit and a site path planning unit; the positioning and navigation unit is used to obtain the real-time position information of the unmanned ship; the formation arrangement unit is used to generate formation pattern data according to the task requirements and the real-time position information; the site path planning unit is used to scan the site, obtain the scene information data, perform path planning and design according to the scene information data and the formation pattern data, and obtain the unmanned ship path arrangement set;

[0061] The group intelligence collaboration module includes a dynamic data collection unit, a natural simulation unit and a reset control unit; the dynamic data collection unit is used to collect the environmental data, motion data and obstacle sensor data of the unmanned ship during the movement; the natural simulation unit is used to simulate the natural group according to the unmanned ship path arrangement set and motion data to obtain natural control data; the reset control unit is used to perform reset control according to the natural control data, environmental data and obstacle sensor data to obtain reset adjustment data, and the unmanned ship performs posture control according to the reset adjustment data;

[0062] The lighting adaptation module includes a visual fatigue prediction unit and a lighting control unit; the visual fatigue prediction unit is used to collect background data, input the formation pattern data, background data and timestamp into a pre-built visual fatigue prediction model, and output lighting control data; the lighting control unit is used to control the lights of the unmanned ship according to the lighting control data.

[0063] The real-time position information of the unmanned ship includes the real-time position vector, velocity vector and direction information of the unmanned ship, which are obtained by navigation software (such as GPS and Beidou positioning system) and inertial navigation unit (IMU).

[0064] Task requirements include user editing and formation types. User editing allows users to select (image input) or customize formation patterns (algorithm generation) through the user interface. Customized formation patterns require setting necessary image generation parameters (such as complex plane range, iteration depth, and color mapping).

[0065] The formation pattern data is the data required for unmanned ship imaging, such as images and coloring points forming the images, and the acquisition method includes algorithm generation and picture input;

[0066] Algorithm generation methods include:

[0067] Set necessary image generation parameters, such as plane range, image resolution, iteration depth, and color mapping;

[0068] The complex plane range includes the range of the X axis and the range of the Y axis. For example, the range of the X axis is [−2, 1] and the range of the Y axis is [−1.5, 1.5]. These values ​​will affect the overall position and scale of the generated pattern.

[0069] Image Resolution: Determines the size of the generated image, for example setting the width and height to 800 pixels, which will determine the level of detail in the image.

[0070] Iteration depth: Set the maximum number of iterations, for example 100. This is the condition for the algorithm to determine whether a point "escapes". The deeper the iteration depth, the richer the pattern details.

[0071] Color Mapping: Select a color mapping scheme, where the user can choose different color gradients, such as cool or warm tones, and determine how to map colors based on the number of iterations, for example: points with fewer iterations use bright colors, and points with more iterations use dark colors.

[0072] Create a two-dimensional grid array where each cell corresponds to a point on the complex plane;

[0073] Perform complex mapping, iterative calculation and escape determination on each pixel;

[0074] Complex mapping: Map the coordinates of each pixel point to a point in the complex plane. The calculation method includes:

[0075] ;

[0076] In the formula, is a complex number, representing a point on the complex plane; The horizontal coordinate (column index) of the pixel shows the position of the current pixel in the image, usually ranging from 0 to ; The vertical coordinate (row index) of the pixel shows the position of the current pixel in the image, usually ranging from 0 to ; is the corresponding minimum in the complex plane Value, sets the left border of the complex plane image; is the corresponding maximum value in the complex plane Value, sets the right boundary of the complex plane image; The corresponding minimum in the complex plane Value, value sets the lower boundary of the complex plane image; is the corresponding maximum value in the complex plane Value, sets the upper edge of the complex plane image; is the width of the image (in pixels), which is the number of horizontal pixels of the image; is the height of the image (in pixels), which is the number of vertical pixels in the image; Is an imaginary unit.

[0077] The calculation methods for iterative calculation include:

[0078] ;

[0079] In the formula, For the The complex value after iterations; is the number of iterations.

[0080] Escape Detection: Calculating Complex Values If the modulus is greater than 2, then Escaped, record the number of iterations of this point, store the number of iterations of each pixel in the image array, generate the image, map the image array to the generated image, and assign a color to each pixel based on the number of iterations.

[0081] The images generated by the algorithm will not be as rigid as those input directly by users. They can generate diverse and complex patterns. At the same time, because they are generated by the algorithm, the data will be stored directly. There is no need to identify and split them like picture input. They can be directly matched with the unmanned ship. However, the patterns generated by the algorithm have a certain screening time cost. If the purpose is to conduct a standardized unmanned ship demonstration, direct picture input can also be used.

[0082] Scene information data includes water flow, wind speed and sensor data, and sensor data includes lidar, camera and ultrasonic sensor;

[0083] The path planning design uses a fast random tree algorithm to find the path and obtain the unmanned ship path arrangement set. The path in the unmanned ship path arrangement set here is the initial path and needs to be fine-tuned in the subsequent process.

[0084] Environmental data includes water flow speed, water flow direction, water depth, wind speed, wind direction, wave height and frequency; motion data includes the speed and acceleration of the unmanned boat; obstacle sensing data includes fixed obstacle data information and mobile obstacle data information. Fixed obstacles include shore structures (piers, bridges, buoys, etc.), other ships and shore plants, etc.; mobile obstacles include other water surface vehicles (ships, boats, yachts, etc.) or any moving objects (such as swimmers, floats, etc.), which can be measured using sensors including lidar, cameras and ultrasonic sensors.

[0085] It should be noted that the scene information data is based on the site as the scope, and the data of the entire site is measured, but the drones in the site are individuals and need to be fine-tuned. The environmental data and obstacle sensing data are the basic data for subsequent fine-tuning.

[0086] Methods for obtaining natural control data include:

[0087] The initial natural parameters of the unmanned ship are allocated based on the unmanned ship path arrangement set. The natural parameters include position vector, velocity vector and field of view, where the field of view is the maximum range of the unmanned ship to perceive the surrounding environment and nearby individuals.

[0088] The natural swarm simulation algorithm is run once at each time step to update the position vector and velocity vector of the unmanned ship and obtain natural control data. The time step is a preset value; the natural control data is a data set obtained from each operation of the natural swarm simulation algorithm.

[0089] like Figure 2 As shown, the natural population simulation algorithm includes:

[0090] Taking the unmanned ship as a single object, all unmanned ships within the field of view of the unmanned ship are determined, that is, all unmanned ships whose distance is less than the field of view of the unmanned ship are marked as neighbors.

[0091] Calculate the behavior forces of all unmanned ships based on the motion data of the neighbors. The motion data includes velocity vector, acceleration vector and position vector. The behavior forces include separation force, alignment force and aggregation force.

[0092] Separation force: Calculate the distance between all neighbors and the current unmanned ship (can be measured using both lidar and ultrasonic sensors), and compare the distance with the preset minimum safety distance. If the distance is greater than or equal to the preset minimum safety distance, no separation force is required, that is, no separation force needs to be calculated. If the distance is less than the preset minimum safety distance, it is marked as avoidance, and the separation force vector is calculated at the same time. :

[0093] ;

[0094] In the formula, is the current position vector of the unmanned ship, is the position vector of the neighbor, For normalization, is the current distance between the unmanned ship and its neighbor;

[0095] Exemplary:

[0096] The current position vector of the unmanned ship is (5, 5), and the position vector of a neighbor nearby is (4, 4);

[0097] Calculating distance ≈1.41, assuming it is less than the preset minimum safety distance, calculate the separation force at this time:

[0098] = ≈ ; = ≈ ≈ ;

[0099] It should be noted that the calculation formula for the separation force vector is a vector calculation, the position vector is the position coordinate, which is a two-dimensional vector calculation, and the separation force points in the direction away from the neighbor and increases as the neighbor approaches; the three forces are all vector calculations, that is, two-dimensional calculations. Since they are in the horizontal plane, when the change in water depth, that is, the change in water wave height, is not considered, it is only necessary to calculate the two-dimensional plane.

[0100] Alignment force: Accumulate the current speed vectors of all neighbors and calculate the average speed vector of all neighbors to obtain the average speed vector of all neighbors. Calculate the difference between the current speed vector of the unmanned ship and the average speed vector to obtain the alignment force. This force points in the direction of the average speed, so that the speed of the unmanned ship gradually changes to keep consistent with the neighbors.

[0101] Aggregation force: Accumulate the current position vectors of all neighbors and calculate the average position vector of all neighbors to obtain the average position vector of all neighbors. Calculate the difference between the current position vector of the unmanned ship and the average position vector to obtain the aggregation force, which drives the unmanned ship to move closer to the average position.

[0102] The natural swarm simulation algorithm enables multiple unmanned ships to move in a coordinated manner in the water. Through interaction mechanisms and behavior adjustments, the unmanned ships can adapt to various changes in a dynamic environment, making them exhibit the naturalness of group behavior.

[0103] The method for obtaining reset adjustment data includes:

[0104] Input the natural control data, environmental data and obstacle sensing data into the pre-built reset control model, and output the reset adjustment data;

[0105] The training method of the reset control model includes:

[0106] The natural control data, environmental data and obstacle sensor data are taken as the input of the reset control model. The reset control model takes the reset adjustment data corresponding to the prediction of each group of natural control data, environmental data and obstacle sensor data as the output, and takes the actual reset adjustment data corresponding to each group of natural control data, environmental data and obstacle sensor data as the prediction target. The training goal is to minimize the sum of the first prediction accuracies of all predicted reset adjustment data.

[0107] Among them, the calculation formula for the first prediction accuracy is: ,in, is the number of each set of natural control data, environmental data and obstacle sensing data, is the first prediction accuracy, For the Predicted reset adjustment data corresponding to natural control data, environmental data and obstacle sensing data, For the The actual reset adjustment data corresponding to the natural control data, environmental data and obstacle sensing data are grouped; the reset control model is trained until the sum of the first prediction accuracy reaches convergence and the training is stopped; the reset control model is a long short-term memory network model.

[0108] While ensuring that the unmanned ship formation changes its pattern smoothly and naturally, the reset adjustment data needs to be adjusted in three dimensions in combination with the influence of environmental factors (such as water waves and wind). By acquiring natural control data, environmental data and obstacle sensor data in real time and using appropriate machine learning algorithms (such as long short-term memory network models) for training, it can achieve rapid response and effective compensation to environmental changes, ensuring that the unmanned ship formation performance is efficient and stable.

[0109] Background data includes light intensity, background complexity and viewing distance;

[0110] Methods for obtaining background complexity include:

[0111] Input the environmental photos into the pre-built visual processing model to output the background complexity; the environmental photos are images taken by the camera of a drone or unmanned ship.

[0112] Methods for training visual processing models include:

[0113] The environmental photos are used as the input of the visual processing model. The visual processing model predicts the corresponding background complexity for each group of environmental photos as output, takes the actual background complexity corresponding to each group of environmental photos as the prediction target, and takes minimizing the sum of the second prediction accuracies of all predicted background complexities as the training target.

[0114] Among them, the calculation formula for the second prediction accuracy is: ,in, is the number of each group of environmental photos, is the second prediction accuracy, For the The predicted background complexity corresponding to the environment photo, is the actual background complexity corresponding to the jth group of environmental photos; the visual processing model is trained until the sum of the second prediction accuracies reaches convergence and the training is stopped; the visual processing model is a convolutional neural network model.

[0115] The camera on the unmanned boat is used to capture the surrounding environment, and the convolutional neural network (CNN) model is used to extract image features. Indicators such as the entropy, edge strength, and number of objects in the background image are generalized into background complexity.

[0116] It should be noted that the background complexity here is a relatively vague generalization result and is only used as one of the indicators for predicting visual fatigue. The posture adjustment time of the unmanned ship formation during the performance is limited. Too long a time will destroy the coordination of the unmanned ship formation. Therefore, it is not suitable for too precise data measurement.

[0117] The training method of the visual fatigue prediction model includes:

[0118] The formation pattern data, background data and timestamp are used as the input of the visual fatigue prediction model. The visual fatigue prediction model takes the lighting control data corresponding to each set of formation pattern data, background data and timestamp as its output, and the actual lighting control data corresponding to each set of formation pattern data, background data and timestamp as its prediction target. Minimizing the sum of the third prediction accuracies of all predicted lighting control data is used as the training target.

[0119] Among them, the calculation formula for the third prediction accuracy is: ,in, The number of each set of formation pattern data, background data and timestamp, is the third prediction accuracy, For the Formation pattern data, background data and predicted lighting control data corresponding to timestamps, For the The formation pattern data, background data and actual lighting control data corresponding to the timestamp are collected; the visual fatigue prediction model is trained until the sum of the third prediction accuracies reaches convergence; the visual fatigue prediction model is a decision tree model.

[0120] Since visual fatigue is based on the natural coordination of the unmanned ship formation, in the actual environment, some input data may be missing, such as formation status or background changes. The decision tree can handle missing values ​​better, and there is no need to strictly preprocess the data, saving data processing time. This feature is suitable for real-time operation of unmanned ships in dynamic environments, and helps to maintain natural coordination and stability of visual effects.

[0121] Lighting control data includes specific parameters that control lighting performance, such as brightness, flashing mode and duration.

[0122] Embodiment 2:

[0123] like Figure 3As shown, based on the same inventive concept as the water LED formation display system based on group intelligence collaboration in the aforementioned embodiment, this application provides a water LED formation display method based on group intelligence collaboration. The method and system embodiment in the embodiment of this application are based on the same inventive concept. Among them, the method includes an arrangement method, a group intelligence collaboration method and a lighting adaptation method;

[0124] Arrangement method:

[0125] Obtain the real-time location information of the unmanned ship; generate formation pattern data according to the task requirements and the real-time location information; scan the site to obtain scene information data, perform path planning and design according to the scene information data and the formation pattern data, and obtain the unmanned ship path arrangement set;

[0126] Crowd Intelligence Collaboration Method:

[0127] Collect environmental data, motion data and obstacle sensor data of the unmanned boat during its movement; perform natural group simulation based on the unmanned boat path arrangement set and motion data to obtain natural control data; perform reset control based on the natural control data, environmental data and obstacle sensor data to obtain reset adjustment data, and the unmanned boat performs posture control based on the reset adjustment data; the method for obtaining the reset adjustment data includes:

[0128] Input the natural control data, environmental data and obstacle sensing data into the pre-built reset control model, and output the reset adjustment data;

[0129] The training method of the reset control model includes:

[0130] The natural control data, environmental data and obstacle sensor data are used as inputs of the reset control model. The reset control model uses the reset adjustment data corresponding to the prediction of each set of natural control data, environmental data and obstacle sensor data as output, uses the actual reset adjustment data corresponding to each set of natural control data, environmental data and obstacle sensor data as the prediction target, and uses minimizing the sum of the first prediction accuracies of all predicted reset adjustment data as the training target; the reset control model is trained until the sum of the first prediction accuracies reaches convergence and the training is stopped; the reset control model is a long short-term memory network model;

[0131] Lighting Adaptation Method:

[0132] Collect background data, input the formation pattern data, background data and timestamp into the pre-built visual fatigue prediction model, and output the lighting control data; adjust the lights of the unmanned ship according to the lighting control data.

[0133] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

[0134] What has been described above is only a preferred specific implementation manner of the embodiments of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can make equivalent substitutions or changes according to the technical scheme and concept of the present application within the technical scope disclosed in the present application, which should be covered by the protection scope of the present application.

Claims

1. The method for displaying LED formations on water based on group intelligence collaboration is characterized by: The methods include choreography method, group intelligence collaboration method and lighting adaptive method; The arrangement method comprises: Obtain the real-time location information of the unmanned ship; generate formation pattern data according to the task requirements and the real-time location information; scan the site to obtain scene information data, perform path planning and design according to the scene information data and the formation pattern data, and obtain the unmanned ship path arrangement set; The group intelligence collaboration method comprises: Collect environmental data, motion data and obstacle sensor data of the unmanned ship during its movement; perform natural group simulation based on the unmanned ship path arrangement set and motion data to obtain natural control data; perform reset control based on the natural control data, environmental data and obstacle sensor data to obtain reset adjustment data, and the unmanned ship performs posture control based on the reset adjustment data; The method for obtaining reset adjustment data includes: Input the natural control data, environmental data and obstacle sensing data into the pre-built reset control model, and output the reset adjustment data; The training method of the reset control model includes: The natural control data, environmental data and obstacle sensor data are used as inputs of the reset control model. The reset control model uses the reset adjustment data corresponding to the prediction of each set of natural control data, environmental data and obstacle sensor data as output, uses the actual reset adjustment data corresponding to each set of natural control data, environmental data and obstacle sensor data as the prediction target, and uses minimizing the sum of the first prediction accuracies of all predicted reset adjustment data as the training target; the reset control model is trained until the sum of the first prediction accuracies reaches convergence and the training is stopped; the reset control model is a long short-term memory network model; The lighting adaptation method comprises: Collect background data, input the formation pattern data, background data and timestamp into the pre-built visual fatigue prediction model, and output the lighting control data; adjust the lights of the unmanned ship according to the lighting control data.

2. The method according to claim 1, characterized in that The method for acquiring the formation pattern data includes algorithm generation and picture input; The algorithm generation method comprises: Set necessary image generation parameters; Create a two-dimensional grid array where each cell corresponds to a point on the complex plane; Perform complex mapping, iterative calculation and escape determination on each pixel; Complex mapping: Map the coordinates of each pixel point to a point in the complex plane. The calculation method includes: ； In the formula, represents a point on the complex plane; is the horizontal coordinate of the pixel; The vertical coordinate of the pixel; is the corresponding minimum in the complex plane value; is the corresponding maximum value in the complex plane value; The corresponding minimum in the complex plane value; is the corresponding maximum value in the complex plane value; is the width of the image; is the height of the image; is an imaginary unit; The calculation methods for iterative calculation include: ； In the formula, For the The complex value after iterations; is the number of iterations; Escape Detection: Calculating Complex Values If the modulus is greater than 2, then Escaped, record the number of iterations of this point, store the number of iterations of each pixel in the image array, generate the image, map the image array to the generated image, and assign a color to each pixel based on the number of iterations.

3. The method according to claim 1, characterized in that The method for obtaining the naturalness control data includes: Allocate initial natural parameters of the unmanned ship based on the unmanned ship path arrangement set, the natural parameters include position vector, velocity vector and field of view; The natural swarm simulation algorithm is run once at each time step to update the position vector and velocity vector of the unmanned ship and obtain natural control data.

4. The method according to claim 3, characterized in that The natural population simulation algorithm includes: Taking the unmanned ship as a single object, all unmanned ships whose distance is less than the field of view of the unmanned ship are marked as neighbors; based on the motion data of the neighbors, the behavioral forces of all unmanned ships are calculated. The motion data includes velocity vector, acceleration vector and position vector. The behavioral forces include separation force, alignment force and aggregation force. Separation force: Calculate the distance between all neighbors and the current unmanned ship, and compare the distance with the preset minimum safety distance. If the distance is greater than or equal to the preset minimum safety distance, there is no need to calculate the separation force. If the distance is less than the preset minimum safety distance, it is marked as avoidance and the separation force vector is calculated at the same time. : ； In the formula, is the current position vector of the unmanned ship, is the position vector of the neighbor, For normalization, is the current distance between the unmanned ship and its neighbor; Alignment force: Accumulate the current speed vectors of all neighbors and calculate the average speed vector of all neighbors to obtain the average speed vector of all neighbors. Calculate the difference between the current speed vector of the unmanned ship and the average speed vector to obtain the alignment force. Aggregation power: Accumulate the current position vectors of all neighbors and calculate the average position vector of all neighbors to obtain the average position vector of all neighbors. Calculate the difference between the current position vector of the unmanned ship and the average position vector to obtain the aggregation power.

5. The method according to claim 1, characterized in that The background data includes light intensity, background complexity and viewing distance; The method for obtaining the background complexity includes: Input the environment photo into the pre-built visual processing model and output the background complexity; The training method of the visual processing model includes: The environmental photos are used as the input of the visual processing model. The visual processing model uses the predicted background complexity corresponding to each group of environmental photos as the output, the actual background complexity corresponding to each group of environmental photos as the prediction target, and the sum of the second prediction accuracies of all predicted background complexities as the training target; the visual processing model is trained until the sum of the second prediction accuracies reaches convergence and the training is stopped; the visual processing model is a convolutional neural network model.

6. The method according to claim 1, characterized in that The training method of the visual fatigue prediction model includes: The formation pattern data, background data and timestamp are used as inputs of the visual fatigue prediction model. The visual fatigue prediction model takes the lighting control data corresponding to each set of formation pattern data, background data and timestamp predictions as outputs, takes the actual lighting control data corresponding to each set of formation pattern data, background data and timestamps as prediction targets, and takes minimizing the sum of the third prediction accuracies of all predicted lighting control data as the training target; the visual fatigue prediction model is trained until the sum of the third prediction accuracies converges; the training is stopped; the visual fatigue prediction model is a decision tree model.

7. The water LED formation display system based on group intelligence collaboration is characterized by: The system includes an orchestration module, a group intelligence collaboration module, and a lighting adaptation module; The arrangement module includes a positioning and navigation unit, a formation arrangement unit and a site path planning unit; the positioning and navigation unit is used to obtain the real-time position information of the unmanned ship; The formation arrangement unit is used to generate formation pattern data according to task requirements and real-time position information; The site path planning unit is used to scan the site, obtain scene information data, perform path planning and design according to the scene information data and formation pattern data, and obtain the unmanned ship path arrangement set; The group intelligence collaboration module includes a dynamic data collection unit, a natural simulation unit and a reset control unit; The dynamic data collection unit is used to collect environmental data, motion data and obstacle sensing data of the unmanned ship during its movement; The natural simulation unit is used to perform natural group simulation according to the unmanned ship path arrangement set and motion data to obtain natural control data; the reset control unit is used to perform reset control according to the natural control data, environmental data and obstacle sensing data to obtain reset adjustment data, and the unmanned ship performs posture control according to the reset adjustment data; The method for obtaining reset adjustment data includes: Input the natural control data, environmental data and obstacle sensing data into the pre-built reset control model, and output the reset adjustment data; The training method of the reset control model includes: The natural control data, environmental data and obstacle sensor data are used as inputs of the reset control model. The reset control model uses the reset adjustment data corresponding to the prediction of each set of natural control data, environmental data and obstacle sensor data as output, uses the actual reset adjustment data corresponding to each set of natural control data, environmental data and obstacle sensor data as the prediction target, and uses minimizing the sum of the first prediction accuracies of all predicted reset adjustment data as the training target; the reset control model is trained until the sum of the first prediction accuracies reaches convergence and the training is stopped; the reset control model is a long short-term memory network model; The lighting adaptation module includes a visual fatigue prediction unit and a lighting control unit; the visual fatigue prediction unit is used to collect background data, input formation pattern data, background data and timestamp into a pre-built visual fatigue prediction model, and output lighting control data; the lighting control unit is used to control the unmanned ship lighting according to the lighting control data.

Images