Marine net cage unmanned ship intelligent inspection method based on multi-source data and environment perception

Abstract

The invention provides an ocean net cage unmanned ship intelligent inspection method based on multi-source data and environment perception, which comprises the following steps: collecting and preprocessing multi-source ocean environment data of an ocean net cage group area, and simultaneously obtaining net cage distribution information, net cage inspection task requirements and an initial state of each unmanned ship; predicting the number of unmanned ships by using an improved random forest model; adopting an improved genetic algorithm to generate a plurality of optimal inspection paths; intelligently distributing a plurality of optimal routing inspection paths to each unmanned ship, and performing routing inspection on each unmanned ship according to the distributed path; multi-source marine environment data collected by a fixed monitoring station deployed in a net cage group area and a sensor of the unmanned ship can be comprehensively utilized, unmanned ship inspection path pre-planning, intelligent task allocation and dynamic adjustment of the inspection process are achieved, and therefore the inspection efficiency is improved, inspection safety is guaranteed, resource allocation is optimized, and the cost is reduced. And the intelligent level of mariculture management is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned ship path planning and marine aquaculture management, and more specifically, to an intelligent inspection method of an unmanned marine cage ship based on multi-source data and environmental perception. Background Art

[0002] In the management of marine cage aquaculture, regular inspections are essential to ensure aquaculture safety and to promptly detect and deal with potential problems. Traditional inspections mainly rely on manually driven boats, which are inefficient, costly, and greatly affected by human factors and weather conditions.

[0003] Although some technologies that use unmanned ships for inspection have emerged, the existing technologies still have the following shortcomings: 1) Insufficient pre-planning capabilities: Existing technologies lack the ability to reasonably predict the number of unmanned boats and plan their routes based on the environmental data of the cage group area before the unmanned boats set off; 2) Insufficient environmental adaptability: Existing technologies cannot fully utilize the real-time environmental data of the cage group area for optimization before inspection, and lack dynamic adaptability to environmental changes during the inspection process; 3) Low efficiency of resource allocation: Existing technologies cannot perform optimal resource allocation before inspection based on the overall environmental conditions of the cage group and the status of the unmanned boat itself; 4) Insufficient decision support: Existing technologies lack intuitive data display and intelligent analysis methods, making it difficult for managers to quickly understand environmental conditions and inspection results, affecting decision-making efficiency. Summary of the invention

[0004] In order to overcome the defects of the above-mentioned prior art, such as rigid unmanned ship path planning, poor environmental adaptability and unreasonable resource allocation, the present invention provides an intelligent inspection method for marine cage unmanned ships based on multi-source data and environmental perception. The method can comprehensively utilize the multi-source marine environmental data collected by fixed monitoring stations deployed in the cage group area and the unmanned ship's own sensors to realize the pre-planning of unmanned ship inspection paths, intelligent task allocation and dynamic adjustment of the inspection process, thereby improving inspection efficiency, ensuring inspection safety, optimizing resource allocation, and enhancing the intelligence level of marine aquaculture management.

[0005] In order to solve the above technical problems, the technical solution of the present invention is as follows: An intelligent inspection method for marine cage unmanned boats based on multi-source data and environmental perception includes the following steps: S1: Collect and pre-process multi-source marine environmental data in the marine cage group area, and obtain cage distribution information, cage inspection task requirements, and the initial status of each unmanned vessel; S2: Based on the pre-processed multi-source ocean environment data, the number of unmanned ships is predicted using an improved random forest model; the improved random forest model introduces an adaptive weight adjustment mechanism based on environmental perception, and the prediction weight of each decision tree is dynamically adjusted according to the real-time multi-source ocean environment data; S3: Based on the pre-processed multi-source marine environment data, cage distribution information, cage inspection task requirements and the initial status of each unmanned ship, an improved genetic algorithm is used to generate several optimal inspection paths; the improved genetic algorithm adaptively adjusts the optimization target and the genetic operator according to the real-time multi-source marine environment data and the status of the unmanned ship; S4: intelligently assigning a number of optimal inspection paths to each unmanned ship, and each unmanned ship performs inspections in the marine cage group area according to the optimal inspection path assigned to it.

[0006] Compared with the prior art, the technical solution of the present invention has the following beneficial effects: The present invention provides an intelligent inspection method for marine cage unmanned ships based on multi-source data and environmental perception. First, multi-source marine environmental data of a marine cage group area is collected and preprocessed, and cage distribution information, cage inspection task requirements, and the initial state of each unmanned ship are obtained at the same time; then, based on the preprocessed multi-source marine environmental data, an improved random forest model is used to predict the number of unmanned ships; then, based on the preprocessed multi-source marine environmental data, cage distribution information, cage inspection task requirements, and the initial state of each unmanned ship, an improved genetic algorithm is used to generate a plurality of optimal inspection paths; finally, a plurality of optimal inspection paths are intelligently allocated to each unmanned ship, and each unmanned ship performs inspection in the marine cage group area according to the optimal inspection path allocated to it; The present invention collects multi-source environmental data, uses an improved random forest model to predict the number of unmanned boats, uses an improved genetic algorithm to predict the optimal inspection path, and uses an intelligent allocation algorithm to allocate the predicted path to the most suitable unmanned boat for execution, which has the following beneficial effects: 1) Improve inspection efficiency: The present invention comprehensively considers the overall environmental factors and cage requirements of the cage group area, optimizes the inspection path of the unmanned boat in advance, reduces invalid navigation, and significantly improves the inspection efficiency; for example, in a cage farming scenario in a certain island area, more than 50% of the inspection time can be saved compared with manual inspection; 2) Enhance inspection safety: The present invention takes environmental data into consideration during pre-path planning and inspection, especially focusing on key factors such as wind speed, to prevent the unmanned ship from operating in harsh environments and improve inspection safety; 3) Optimize resource allocation: According to the overall environmental conditions of the cage group area and the status of the unmanned boats themselves, the present invention makes reasonable predictions on the number of unmanned boats and routes planning before inspection, so as to achieve reasonable allocation and efficient utilization of resources and reduce operating costs; 4) Improve management decision-making level: The present invention can further introduce data visualization technology to monitor and display multi-source environmental data and the inspection status of unmanned ships in real time, thereby providing managers with intuitive and comprehensive information to assist them in making scientific decisions and responding to potential problems in a timely manner; 5) Strong environmental adaptability: The present invention can utilize the data collected by fixed monitoring stations and the unmanned ship's own sensors to achieve pre-planning before inspection and dynamic adjustment during the inspection process, better adapt to the ever-changing marine environment, ensure the smooth progress of inspection tasks, and improve the robustness of inspection task execution. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] Figure 1 This is a flow chart of an intelligent inspection method for marine cage unmanned boats based on multi-source data and environmental perception provided in Example 1.

[0008] Figure 2 This is a flow chart of an intelligent inspection method for marine cage unmanned boats based on multi-source data and environmental perception provided in Example 2.

[0009] Figure 3 This is a layout diagram of the actual application scenario provided in Example 2.

[0010] Figure 4 This is a flow chart for predicting the number of unmanned ships provided in Example 2.

[0011] Figure 5 A flow chart for generating the optimal inspection path provided in Example 2. DETAILED DESCRIPTION

[0012] The accompanying drawings are only used for illustrative purposes and are not to be construed as limiting the present invention; In order to better illustrate the present embodiment, some parts in the drawings may be omitted, enlarged or reduced, and do not represent the size of the actual product; It is understandable to those skilled in the art that some well-known structures and their descriptions may be omitted in the drawings.

[0013] The technical solution of the present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0014] Example 1 like Figure 1 As shown, this embodiment provides an intelligent inspection method for marine cage unmanned boats based on multi-source data and environmental perception, comprising the following steps: S1: Collect and pre-process multi-source marine environmental data in the marine cage group area, and obtain cage distribution information, cage inspection task requirements, and the initial status of each unmanned vessel; S2: Based on the pre-processed multi-source ocean environment data, the number of unmanned ships is predicted using an improved random forest model; the improved random forest model introduces an adaptive weight adjustment mechanism based on environmental perception, and the prediction weight of each decision tree is dynamically adjusted according to the real-time multi-source ocean environment data; S3: Based on the pre-processed multi-source marine environment data, cage distribution information, cage inspection task requirements and the initial status of each unmanned ship, an improved genetic algorithm is used to generate several optimal inspection paths; the improved genetic algorithm adaptively adjusts the optimization target and the genetic operator according to the real-time multi-source marine environment data and the status of the unmanned ship; S4: intelligently assigning a number of optimal inspection paths to each unmanned ship, and each unmanned ship performs inspections in the marine cage group area according to the optimal inspection path assigned to it.

[0015] In the specific implementation process, multi-source marine environmental data of the marine cage group area is first collected and pre-processed, and the cage distribution information, cage inspection task requirements and the initial status of each unmanned ship are obtained; Then, based on the preprocessed multi-source ocean environment data, the improved random forest model is used to predict the number of unmanned ships. Then, based on the pre-processed multi-source marine environment data, cage distribution information, cage inspection task requirements, and the initial status of each unmanned vessel, an improved genetic algorithm was used to generate several optimal inspection paths; Finally, several optimal inspection paths are intelligently allocated to each unmanned ship, and each unmanned ship conducts inspections in the marine cage group area according to the optimal inspection path allocated to it; In addition, in this embodiment, during the unmanned boat inspection process, the operating status of the unmanned boat is monitored in real time, including location information, remaining power, sensor data, etc., and the monitoring data is fed back to the central control screen; at the same time, the real-time multi-source marine environment data, the operating status of the unmanned boat, the inspection path, the cage distribution information and the inspection results are visualized on the central control screen to achieve unified monitoring of information; During the inspection process, the real-time environmental data and status collected by the unmanned ship are continuously monitored to determine whether the inspection plan needs to be adjusted; if the environmental conditions change significantly (such as wind speed exceeds the set threshold) or the state of the unmanned ship is abnormal, the steps in the above-mentioned unmanned ship intelligent inspection method are re-executed based on the real-time data collected by the unmanned ship itself, and the inspection path is re-evaluated and adjusted; This method can comprehensively utilize the multi-source marine environmental data collected by fixed monitoring stations deployed in the cage group area and the unmanned ship's own sensors to achieve pre-planning of the unmanned ship's inspection path, intelligent task allocation, and dynamic adjustment of the inspection process, thereby improving inspection efficiency, ensuring inspection safety, optimizing resource allocation, and enhancing the scientific level of marine aquaculture management.

[0016] Example 2 This embodiment provides an intelligent inspection method for marine cage unmanned boats based on multi-source data and environmental perception, comprising the following steps: S1: Collect and pre-process multi-source marine environmental data in the marine cage group area, and obtain cage distribution information, cage inspection task requirements, and the initial status of each unmanned vessel; S2: Based on the pre-processed multi-source ocean environment data, the number of unmanned ships is predicted using an improved random forest model; the improved random forest model introduces an adaptive weight adjustment mechanism based on environmental perception, and the prediction weight of each decision tree is dynamically adjusted according to the real-time multi-source ocean environment data; S3: Based on the pre-processed multi-source marine environment data, cage distribution information, cage inspection task requirements and the initial status of each unmanned ship, an improved genetic algorithm is used to generate several optimal inspection paths; the improved genetic algorithm adaptively adjusts the optimization target and the genetic operator according to the real-time multi-source marine environment data and the status of the unmanned ship; S4: intelligently assigning a number of optimal inspection paths to each unmanned ship, and each unmanned ship performs inspections in the marine cage group area according to the optimal inspection path assigned to it.

[0017] In the specific implementation process, Figure 2 Shown is a flow chart of the method in this embodiment; Step S1: Collect and pre-process multi-source marine environmental data in the marine cage group area, and simultaneously obtain cage distribution information, cage inspection task requirements, and the initial status of each unmanned vessel; Specifically, a plurality of fixed monitoring stations (buoys equipped with sensors) deployed in the marine cage group area and sensors pre-installed on each unmanned ship are used to collect multi-source marine environmental data of the marine cage group area; the multi-source marine environmental data at least include: wind speed, wind direction, current speed, current direction, wave height, wave period, wave speed, current speed, current direction and environmental meteorological data; the initial state of each unmanned ship is collected by using sensors pre-installed on each unmanned ship; The specific application scenarios of this embodiment are as follows Figure 3As shown, a group of aquaculture cages are distributed in the sea area 15 kilometers south of the shore on one side of an island. The cages are arranged in 3 rows and 10 columns, with rows in the east-west direction and columns in the north-south direction. The spacing between the cage outlines is 20 meters, and the diameter of each cage is 10 meters. There is a management center on the shore, and there is an unmanned boat dock in the sea area next to the management center. The unmanned boat is required to depart from the dock to complete the cage inspection and return to the dock; this embodiment adopts the following fixed monitoring station layout scheme (a total of 8): F1: outside the upper left corner; F2: outside the upper right corner; F3: outside the lower left corner; F4: outside the lower right corner; F5: outside the center of the northern boundary; F6: outside the center of the southern boundary; F7: northeast direction, far away, indicating upwind or upstream direction; F8: southwest direction, far away, indicating downwind or downstream direction; Among them, the specific distances of F7 and F8 are relative; the directions of upwind upstream and downwind downstream are assumed according to the general conditions of the environment; F7 and F8 monitoring stations are of great significance in the entire cage monitoring system. Their location and function enable them to capture key information that other monitoring stations may not be able to obtain. They can capture macro-environmental changes that affect the entire cage group, not just local information around the cages; long-distance monitoring data is helpful for long-term trend analysis and assessment of the long-term impact of environmental changes on cage aquaculture, as well as the long-term impact of cage aquaculture activities on the environment; F7 and F8 monitoring stations are like sentinels, guarding the upstream and downstream of the cage group respectively, and can provide broader and more forward-looking environmental information, which is crucial for ensuring the sustainability of cage aquaculture and assessing its environmental impact; together with the nearby monitoring stations, they form a more complete monitoring network, providing managers with a more comprehensive and in-depth understanding of the environment; This layout scheme can better monitor the environmental information when the unmanned boat patrols the sea area outside the cages and capture the macro-environmental changes that affect the entire cage group; In this embodiment, sensors are configured on 8 fixed monitoring stations. The fixed monitoring stations use DTS-WSD series wind speed and direction sensors (accuracy: wind speed ±0.3m / s, wind direction ±3°), acoustic Doppler current profilers (ADCP, such as NortekSignature series, accuracy: ±0.5% of measured value ±0.5 cm / s), pressure sensors (for wave height measurement, such as Keller 33X series, accuracy: ±0.05% FS), and temperature and humidity sensors (accuracy: temperature ±0.2°C, humidity ±2%RH) to measure data. According to actual needs, it can be set to collect data every 10 minutes and transmit the data to the management center through the 4G / 5G network. At the same time, unmanned ships are also equipped with a variety of sensors. They can use sensors similar to those used in fixed monitoring stations, but may require smaller, low-power versions, such as integrated weather stations (similar in accuracy to the above), GPS modules (such as the U-blox NEO series, with a positioning accuracy of <2.5m), current sensors (used to monitor their own power, with an accuracy of ±1%), and may include cameras (for image recognition, such as cage damage detection) and water quality sensors (such as dissolved oxygen and pH sensors); unmanned ships collect data more frequently, such as collecting location information once a second and environmental data once a minute; data is also transmitted to the management center in real time through 4G / 5G networks; After completing data collection, the data needs to be preprocessed to ensure data quality and consistency, providing a reliable foundation for subsequent model construction; In this embodiment, the preprocessing includes: a) Outlier detection and processing: Due to factors such as sensor errors and environmental interference, the collected data may contain outliers. Based on the 3σ principle, outlier detection is performed on all collected data. For example, assuming that the mean of wind speed data collected in the past week is 6 m / s and the standard deviation is 1 m / s, the upper limit of the threshold is 9 m / s and the lower limit is 3 m / s. If a fixed monitoring station collects a wind speed of 11 m / s, it is marked as an outlier. Delete the detected outliers, or replace them with the average value of several adjacent normal data; b) Missing value interpolation: Due to sensor failure, communication interruption and other reasons, there may be missing data; use linear interpolation to interpolate the missing data. Assuming that the data changes linearly, use the two valid data points before and after the missing value for linear interpolation. For example, if the current speed collected by the unmanned ship at 10:00 is 0.4m / s, and the current speed collected at 10:02 is 0.6m / s, then the missing value at 10:01 can be obtained by linear interpolation: 0.4+(0.6-0.4)×(10:01-10:00) / (10:02-10:00)=0.5m / s; c) Data standardization: Data collected by different sensors have different dimensions and numerical ranges. In order to eliminate the impact of dimensions and improve the convergence speed and accuracy of the model, data standardization is required. All data that have undergone outlier detection and processing, as well as missing value interpolation, are Z-score standardized to scale the data to a distribution with a mean of 0 and a standard deviation of 1. For example, assuming that the mean of the historical wind speed data collected by a fixed monitoring station is 5 m / s and the standard deviation is 2 m / s, and the current wind speed collected is 9 m / s, then the standardized value is (9-5) / 2=2. d) Data formatting: converting all data that have undergone data standardization into data of a pre-set unified type and format; For example, assume that the ocean environment data from different sensors contains the following two timestamp formats and a field representing the sensor ID: the data format of sensor A: sensor_id: A, timestamp: 20231101140000, temperature: 28.5; the data format of sensor B: sensor_id: B, time: 2023-11-01 14:05:00, salinity: 32.1; this embodiment needs to unify the format of these data for subsequent processing, the steps are as follows: First, unify the timestamp format and convert sensor A's 20231101140000 and sensor B's 2023-11-0114:05:00 into the standard ISO 8601 format, such as 2023-11-01T14:00:00 and 2023-11-01T14:05:00. Then reshape the data structure and convert the data into a unified table structure. Then convert the data type to ensure that the data types of the temperature and salinity columns are floating point types. For another example, suppose the minimum value of the wave height data collected by the unmanned ship is stored in the form of a string, such as "0.2m", and the maximum value is also stored in the form of a string, such as "1.5m". In order to perform subsequent numerical calculations, the data formatting needs to be processed as follows: remove the unit and extract the numerical part from the string, that is, remove the "m" character; convert the data type and convert the extracted string value into a floating point type; finally, "0.2m" will be converted to 0.2 (floating point type) and "1.5m" will be converted to 1.5 (floating point type); in this way, the wave height data can be used for subsequent statistical analysis or model training; For another example, suppose that the longitude and latitude information collected by some sensors is stored in separate "longitude" and "latitude" columns, and the map visualization tool requires that the longitude and latitude information be merged into a comma-separated string; data formatting requires merging these two columns of information, and the steps are as follows: String concatenation: connect the values ​​of the "longitude" column and the "latitude" column, separated by a comma; for example, if the longitude value is 110.5 and the latitude value is 20.8, the concatenated string is "110.5,20.8"; create a new column: create a new column, for example, named "location", and store the concatenated string in the column, so that the original longitude and latitude information is converted into a format that is more suitable for use by specific tools; Step S2: Based on the preprocessed multi-source ocean environment data, the number of unmanned ships is predicted using the improved random forest model; The basic model of this step adopts the random forest regression model. Random forest is an ensemble learning method that improves prediction accuracy and robustness by constructing and aggregating the prediction results of multiple decision trees. During training, each decision tree will randomly select a part of samples and features for construction, thereby reducing the risk of overfitting. For regression problems, the prediction result of random forest is the average of the prediction values ​​of all decision trees. When applied to the prediction of the number of unmanned boats, the traditional random forest regression model mainly relies on historical inspection data and some static environmental parameters (for example, historical average wind speed). The limitations of this method are: 1) Lack of fine perception of real-time environmental dynamics: Traditional models find it difficult to capture the impact of real-time and dynamically changing environmental factors on the demand for unmanned boats on the day of the inspection; for example, sudden changes in wind speed and changes in the direction of ocean currents may significantly affect the difficulty of inspections and the number of unmanned boats required; 2) Static weight allocation: During the prediction process, the prediction results of all decision trees are usually given the same weight for averaging, ignoring the differences in the prediction capabilities of different decision trees under different environmental conditions; for example, some decision trees may be better at predicting the demand for unmanned boats in bad weather, while others are better at predicting demand in calm weather; In order to overcome the above limitations, this method introduces the "environmental perception weight" mechanism on the basis of the random forest regression model; the core idea of ​​this mechanism is: according to the real-time, multi-source environmental information, the prediction weight of each decision tree is dynamically adjusted. If a decision tree has historically predicted the number of unmanned ships under conditions similar to the current environment more accurately, it will be given a higher weight, so that the decision tree with stronger prediction ability under the current environmental conditions will play a greater role in the final prediction, thereby improving the accuracy and robustness of the prediction, such as Figure 4 The specific implementation steps are as follows: S2.1: Obtain historical data and currently collected multi-source marine environmental data; Use historical data to train the traditional random forest regression model and obtain a pre-trained random forest regression model; S2.2: Calculate the environmental risk index Risk_Index and the change rate Change_Rate of each environmental data according to the multi-source marine environmental data, and construct the current environmental feature vector in combination with the calculation results of the environmental risk index Risk_Index and the change rate Change_Rate. This embodiment introduces more refined features that can reflect the real-time dynamic changes of the environment on the basis of traditional environmental features; This embodiment comprehensively considers multiple real-time environmental factors (wind speed, wave height, current speed, etc.) and their risk levels, quantifies the overall risk level of the current environment, and calculates the environmental risk index Risk_Index according to the following formula:

[0018] in, is the jth environmental data; is the weight of the jth environmental data, reflecting the impact of this factor on the inspection risk, which can be obtained through expert experience or historical data analysis; is the jth environmental data The corresponding risk scoring function is used to map the actual value of the environmental data to the interval [0,1]. The higher the value, the greater the risk. For example, the risk scoring function of wind speed can be a piecewise function: , in this piecewise function, Indicates wind speed The corresponding risk score; The environmental change rate reflects the speed and magnitude of changes in environmental parameters over a period of time. For example, the rate of change of wind speed and current speed in the past hour helps the model determine the stability of the environment and thus more accurately predict the demand for unmanned ships. The environmental change rate Change_Rate is calculated according to the following formula:

[0019] in, is the change rate of the kth environmental data; is the actual value of the kth environmental data at the current time t; is the time interval; Indicates the absolute value symbol; S2.3: Calculate the similarity Sim_i between the environmental features of the historical data used in the training of each decision tree and the current environmental feature vector. The higher the similarity, the more suitable the experience of the decision tree is for the current environment, and a higher weight should be given to it. In this embodiment, the similarity Sim_i is calculated according to the cosine similarity formula. At the same time, based on the similarity Sim_i, the historical prediction performance score Perf_i of each decision tree is evaluated. Perf_i represents the performance score of the i-th decision tree in predicting the number of unmanned ships under conditions similar to the current environment. This score aims to quantify the prediction accuracy of the decision tree in similar situations in the past. The higher the value of Perf_i, the stronger the prediction ability of the decision tree in similar environments in the past. Therefore, it should be given a higher weight in the current environment. The calculation formula is:

[0020] Among them, Sim(Env_current,Env_historical_j) is the similarity between the current environment feature vector Env_current and the environment feature Env_historical_j corresponding to the j-th historical data, which can be calculated using the same cosine similarity formula as that for calculating Sim_i. This part measures the similarity between the historical environment and the current environment. The higher the similarity, the greater the reference value of the historical record for evaluating the current performance; Prediction_i_j is the prediction result of the i-th decision tree in the j-th historical data; Actual_j is the actual number of unmanned ships in the j-th historical data; In the calculation formula of Perf_i, the numerator represents the sum of the environmental similarity and the performance score of the decision tree in the environment for all historical records. This part calculates a weighted historical performance score. The prediction performance of similar historical environments will be given a higher weight. The denominator represents the sum of the similarities between all historical records and the current environment. It is used to normalize the numerator to ensure that the value of Perf_i is within a reasonable range and eliminate the impact caused by the different number of historical records. The calculation formula of Perf_i is actually a weighted average performance score based on similarity. It takes into account all historical prediction records, but gives more weight to those historical records that are more similar to the current environment. For each historical record, the similarity between the current environment and the historical environment is calculated, and the prediction performance of the i-th decision tree in the historical environment is evaluated (obtained through error conversion). Then, the similarity and performance score are multiplied, and all historical records are weighted summed and normalized. S2.4: Perform a weighted combination of the similarity Sim_i and the historical prediction performance score Perf_i of each decision tree, and calculate the adaptive weight Weight_i of each decision tree according to the following formula:

[0021] in, is the adaptive weight of the i-th decision tree; is the weight coefficient of similarity, The weight coefficient for historical forecast performance scoring satisfies: ; S2.5: Use the pre-trained random forest regression model to predict the number of unmanned ships and obtain the prediction result Prediction_i of each decision tree; S2.6: Take the weighted average of the prediction results Prediction_i of all decision trees according to their adaptive weights Weight_i, and obtain the final prediction value of the number of unmanned ships Predicted_Count according to the following formula:

[0022] in, is the prediction result of the i-th decision tree; Indicates rounding up; The innovation of step S2 is that it integrates real-time environmental information into the prediction process of random forest, breaking the limitation of treating all decision trees of traditional random forest equally. By introducing environmental perception weights, the model can trust those "expert" decision trees with stronger prediction capabilities in the current environment, thereby improving the accuracy and adaptability of the prediction. This adaptive weight adjustment mechanism enables the model to better cope with the complex and changeable marine environment and provide a more reliable basis for the intelligent scheduling of unmanned ships. Calculation example: Assume that before a certain inspection mission, the following real-time environmental data are obtained through fixed monitoring stations and weather forecasts: average wind speed: 8m / s, average wave height: 1.2m, average current speed: 0.5m / s, wind speed change rate in the past hour: +0.5m / s, current speed change rate in the past hour: +0.1m / s; Calculation of environmental risk index: Assume that the weights w_wind=0.4, w_wave=0.3, w_current=0.3; the risk scoring function is as follows: S_wind(8)=0.8; S_wave(1.2)=0.7; S_current(0.5)=0.5; then Risk_Index=0.4×0.8+0.3×0.7+0.3×0.5=0.68; Calculation of environmental change rate: Assume that the normalized wind speed change rate is 0.6 and the ocean current speed change rate is 0.4; the average value of the comprehensive environmental change rate is 0.5; Adaptive weight adjustment: Assume that the random forest contains 50 decision trees. Consider two of them: Tree 1: Historically, its training samples were mainly concentrated under conditions of wind speed of 6-9 m / s and wave height of 1.0-1.5 m; Tree 2: Historically, its training samples were mainly concentrated under conditions of wind speed of 3-6 m / s and wave height of 0.5-1.0 m; Calculate the similarity between the current environment and the two tree training environments: Assume that Sim_1 is calculated to be 0.85 and Sim_2 is calculated to be 0.60, then the current environment is more similar to the training environment of Tree 1; Evaluate the historical prediction performance of the two trees: Assuming that in the past, when the environmental conditions were similar to the current environment, Tree1 had a smaller prediction error, Perf_1 = 0.9, and Tree 2 had a larger prediction error, Perf_2 = 0.7, then calculate the adaptive weights (assuming w_sim = 0.6, w_perf = 0.4): Weight_1 = 0.6 × 0.85 + 0.4 × 0.9 = 0.51 + 0.36 = 0.87; Weight_2 = 0.6 × 0.60 + 0.4 × 0.7 = 0.36 + 0.28 = 0.64; It can be seen that because the current environment is closer to the training environment of Tree 1 and Tree 1 has better historical prediction performance, Tree 1 is given a higher weight; Weighted average forecast example: Assuming that Tree 1 predicts that 4 unmanned ships are needed and Tree 2 predicts that 3 unmanned ships are needed, the weighted prediction results of the two trees are: (0.87 × 4) + (0.64 × 3) = 3.48 + 1.92 = 5.4; Add up the weighted prediction results of all 50 trees to get the final predicted number of unmanned ships. For example, if the final prediction result is 3.2, round it up and the final prediction is that 4 unmanned ships are needed. Step S3: Based on the pre-processed multi-source marine environment data, cage distribution information, cage inspection task requirements and the initial state of each unmanned ship, an improved genetic algorithm is used to generate several optimal inspection paths; The basic model of this step uses the genetic algorithm (GA) to optimize the inspection path of the unmanned ship. The genetic algorithm is a search heuristic algorithm that simulates natural selection and genetic mechanisms. It optimizes the individuals (paths) in the population through continuous iterations and eventually finds the optimal solution that satisfies the objective function. In the path optimization problem, each individual represents a possible inspection path, and through genetic operations such as selection, crossover and mutation, it continuously evolves to find a better path. Traditional genetic algorithms have the following limitations in unmanned ship path optimization: 1) Static target weights: Fixed weights are usually used to combine multiple optimization targets (e.g., time, energy consumption, safety), and the priority of each target cannot be dynamically adjusted according to real-time environmental changes and the state of the unmanned ship; for example, in bad weather, more emphasis should be placed on safety, while in calm weather, more emphasis can be placed on efficiency; 2) Fixed genetic operators: Genetic operators such as crossover probability and mutation probability are usually set to fixed values, which are difficult to adaptively adjust according to the evolutionary state of the population; for example, when the population diversity is low, the mutation probability should be increased to introduce new genes; when the population fitness improves slowly, the crossover probability should be increased to accelerate the spread of excellent genes; 3) Easy to fall into local optimality: Traditional genetic algorithms are prone to fall into local optimal solutions in complex search spaces and cannot find the global optimal path; In order to overcome the above limitations, this embodiment introduces an adaptive multi-objective optimization mechanism and a dynamic genetic operator adjustment strategy based on the traditional genetic algorithm, so that the algorithm can dynamically adjust the optimization objectives and genetic operations according to real-time environmental information and the state of the unmanned ship, thereby improving the search efficiency and the quality of the solution. Figure 5 The specific implementation steps are as follows: S3.1: Based on the cage distribution information, an initial population is generated by random generation or heuristic method, wherein the initial population is used to represent an initial inspection path set, including a number of individuals, each of which represents a possible inspection path; S3.2: Multi-objective optimization modeling: The unmanned ship inspection path optimization problem is modeled as a multi-objective optimization problem, and multiple conflicting objectives are considered at the same time; minimizing the inspection time, minimizing energy consumption, and maximizing inspection safety are respectively used as optimization objectives, three conflicting objective functions are set, and different objective function values ​​of each individual are calculated respectively; a) The goal of minimizing inspection time is to shorten the total time required to complete all inspection tasks. The objective function is:

[0023] in, For inspection path The inspection time objective function value; Indicates the inspection path The i-th flight segment in ; For inspection path The inspection time required for the i-th navigation segment; b) The goal of minimizing energy consumption is to reduce the total energy consumed by the unmanned ship during the inspection process. The objective function is:

[0024] in, For inspection path The energy consumption objective function value; For inspection path The energy required for the i-th flight segment; c) The goal of maximizing inspection safety is to improve the safety of the unmanned ship during the inspection process, for example, avoiding high-risk areas. The objective function is:

[0025] in, For inspection path The inspection safety objective function value; is the environmental risk index of the ith navigation segment, and a higher value indicates a greater risk; is the length of the i-th flight segment; For inspection path Total length; S3.3: According to the currently collected multi-source ocean environment data and the status of the unmanned ship, the weights of each objective function are adaptively adjusted and normalized to reflect the priority of the current task; When the environmental risk index is high, the weight of inspection safety is increased and the weight of inspection time is reduced. The calculation formula is:

[0026]

[0027] in, is the weight of the inspection safety objective function at the current time t; is the basic weight of the inspection safety objective function; is the weight adjustment range of the inspection safety objective function; is the weight of the inspection time objective function at the current time t; is the basic weight of the inspection time objective function; is the weight adjustment range of the inspection time objective function; is the Sigmoid function; is the environmental risk index at the current time t; When the battery power of the unmanned ship is low, the weight of minimizing energy consumption is increased, and the calculation formula is:

[0028] in, is the weight of the energy consumption objective function at the current time t; is the basic weight of the energy consumption objective function; is the weight adjustment range of the energy consumption objective function; is the remaining energy percentage of the drone, ranging from [0,1]; After adaptive adjustment , and Normalize them separately so that the sum of the three normalized weights is 1; S3.4: Evaluate the diversity and fitness of the population, and dynamically adjust the crossover probability P_cross and mutation probability P_mutate; a) Dynamic adjustment based on population diversity: In this embodiment, the average pairwise distance between individuals in the population can be used to evaluate population diversity, and the formula is as follows:

[0029] in, is the diversity index of the population at the current moment t; N is the population size; represents the distance between individuals i and j, for example, based on the Euclidean distance; When the diversity index of the population is greater than or equal to the preset diversity threshold, only the crossover probability P_cross is increased to promote the combination of excellent genes. The calculation formula is:

[0030] in, is the crossover probability at the current time t; is the minimum crossover probability; is the maximum crossover probability; sensitivity coefficient adjusted for crossover probability; is the preset diversity threshold, and the crossover probability is increased when it is greater than or equal to this threshold; When the population diversity is high, it means that there are different excellent genes in the population. Increasing the crossover probability helps to combine these excellent genes together to produce better offspring. When the diversity index of the population is less than the preset diversity threshold, only the mutation probability P_mutate is increased, new genes are introduced, and the local optimum is jumped out. The calculation formula is:

[0031] in, is the mutation probability at the current time t; is the minimum mutation probability; is the maximum mutation probability; Sensitivity coefficient adjusted for mutation probability; When the population diversity is low, it means that the individuals in the population are highly similar and are prone to fall into the local optimum. Increasing the mutation probability can introduce new genes, increase the diversity of the population, and help escape the local optimum. b) Dynamic adjustment based on population fitness: In this embodiment, the fitness of the population is evaluated by continuously observing the optimal fitness values ​​of several generations of the population, and the fitness stagnation index of the population is calculated according to the following formula:

[0032] in, is the fitness stagnation indicator of the population; is the optimal fitness value of the g-th generation population; To observe the algebraic interval; When the fitness stagnation index of the population is less than or equal to the preset fitness stagnation threshold, the crossover probability P_cross and mutation probability P_mutate are increased at the same time. The calculation formula is:

[0033]

[0034] in, is the basic mutation probability; is the adjustment range of the mutation probability; is the preset fitness stagnation threshold; is the average fitness value of the population in the past several generations; is the average fitness value of the current generation population. When it is lower than this threshold, the probability of crossover and mutation is increased; When the fitness improves slowly, it may mean that the algorithm is stuck in a local optimum. In this case, increasing the crossover probability can help to escape the local optimum. At the same time, when the fitness improvement stagnates, increasing the mutation probability can explore new search spaces and also escape the local optimum. S3.5: Population iteration: select a part of individuals from the current population as parents, and exchange some genes of the selected parent individuals according to the dynamically adjusted crossover probability P_cross to generate new offspring individuals; at the same time, randomly change the genes in the individuals according to the dynamically adjusted mutation probability P_mutate, introduce new genes, and increase population diversity; S3.6: Construct and update the Pareto optimal solution set: The algorithm of this embodiment maintains a Pareto optimal solution set during the iteration process, which includes multiple non-dominated solutions that perform well on different objectives; the non-dominated solutions in the population iteration process are added to the Pareto optimal solution set; at the same time, check whether there is a solution dominated by the newly added solution in the Pareto optimal solution set, and if so, remove the dominated solution; Definition of non-dominated solution: For any solution A in the solution set, there is no other solution B that is better than A in all objectives, that is, for all objectives i, F i (B)≤F i (A), and there exists at least one target j such that Fj (B) <F j (A); S3.7: Determine whether the preset termination iteration condition is met. If not, re-execute steps S3.2 to S3.6. If met, select several optimal inspection paths from the Pareto optimal solution set obtained by the last update according to the cage inspection task requirements. That is, the optimal inspection path finally obtained can be selected from the Pareto optimal solution set of the last iteration. The selection strategy includes: a) Based on manager preferences, managers can choose solutions that perform better on specific goals according to the focus of the current task. For example, if the current task has a high time requirement, a solution with a shorter inspection time can be selected; b) Fuzzy decision-making: Fuzzy logic and other methods can be used to sort and select solutions in the Pareto solution set according to the satisfaction level of each goal; for example, a path that reaches a certain satisfaction threshold in terms of time, energy consumption and safety can be selected; c) Sorting and selection based on preset indicators: A comprehensive indicator can be defined, such as the distance to the ideal point or the minimum requirement, to sort the solutions in the Pareto solution set, and then select the top several; The innovation of step S3 is to integrate the adaptive mechanism into the multi-objective genetic algorithm, so that the path optimization process can be dynamically adjusted according to the real-time environment and the state of the unmanned ship; through adaptive target weight adjustment, the algorithm can flexibly balance the importance of different optimization targets and generate an inspection path that better meets current needs; through dynamic genetic operator adjustment, the algorithm can improve search efficiency, avoid premature convergence, and better explore the solution space to find a better solution; Calculation example: 1) Scene setting and initial state: Set the inspection area to Figure 3 The cage group in the middle (3 rows and 10 columns); real-time environmental data include: average wind speed 8 m / s (large), average wave height: 1.2 m (medium), average current speed: 0.5 m / s (medium), wind direction: due east, current direction: southwest, environmental risk index: 0.68 (high, there may be factors that are not conducive to navigation, such as strong winds and waves); unmanned boat status: current power level: 90% (high); 2) Adaptive target weight adjustment: Assume that the initial weights of the algorithm are set to: Base_W_time = 0.5 (inspection time); Base_W_energy = 0.3 (energy consumption); Base_W_safety = 0.2 (safety); Dynamically adjust weights based on environmental risks: Since the current environmental risk index is 0.68 (high), the algorithm will adjust the weights of each goal to give more weight to safety; assuming the weight adjustment formula is as follows: ΔW_safety(Risk_Index) = 0.15 × Risk_Index; ΔW_time(Risk_Index) = 0.1 × Risk_Index; Calculate the adjusted weights: W_safety(t) = 0.2 + 0.15 × 0.68 = 0.302; W_time(t) = 0.5 - 0.1 × 0.68 = 0.432; W_energy(t) = 0.3 (Since the current power is sufficient, it is assumed that the power has little impact on the energy consumption weight. This is simplified here. The power impact factor can be added in actual applications); 3) Weight normalization: Sum_W = 0.302 + 0.432 + 0.3 = 1.034; W'_time = 0.432 / 1.034 ≈ 0.418; W'_energy = 0.3 / 1.034 ≈ 0.290; W'_safety = 0.302 / 1.034 ≈ 0.292; Result analysis: The normalized weights show that due to the high environmental risk, the weight of safety increased from the initial 0.2 to about 0.292, while the weight of inspection time decreased (from 0.5 to about 0.418); this reflects that in the current environment, the algorithm is more inclined to choose a safer path, even if it may take more time; 4) Dynamic genetic operator adjustment: Current population status: Assume that at a certain iteration of the algorithm, the population diversity is low, Diversity(t) = 0.2 (normalized to the interval [0,1]); Crossover probability adjustment: Assuming the initial crossover probability parameters are P_cross_min=0.6, P_cross_max=0.9, c_cross = 5, use the following formula for adjustment: P_cross(t) = P_cross_min + (P_cross_max - P_cross_min) × exp(-c_cross × Diversity(t)) = 0.6 + (0.9 - 0.6) × exp(-5 × 0.2) = 0.6 + 0.3 ×exp(-1) ≈ 0.6 + 0.3 × 0.368 ≈ 0.71; Result analysis: Due to the low population diversity, the algorithm adaptively increased the crossover probability from the initial minimum value of 0.6 to about 0.71; this helps promote gene recombination between different individuals in the population, generate new path plans, increase population diversity, and try to jump out of the possible local optimization trap; Mutation probability adjustment: Considering the low diversity of the current population and the possibility of falling into a local optimum, the algorithm should tend to increase the mutation probability to introduce new genes and further explore the search space; assuming that the initial mutation probability parameters are P_mutate_base = 0.01, ΔP_mutate = 0.09, Threshold_stagnation = 0.001 (assuming that the fitness stagnation threshold is very small, it means that mutation is increased even if the fitness is slightly stagnant); assuming that the current fitness stagnation index Stagnation is 0.0005 (less than the threshold), use the following formula for adjustment: P_mutate(t) = P_mutate_base + ΔP_mutate × Sigmoid(Threshold_stagnation - Stagnation); Assuming that the Sigmoid function outputs close to 1 here (because Threshold_stagnation - Stagnation is positive and close to the threshold), P_mutate(t) ≈ 0.01 + 0.09 × 1 = 0.10; Result analysis: Due to the low diversity of the population and possible signs of fitness stagnation, the mutation probability increased significantly from the base value of 0.01 to 0.10; this means that the algorithm will randomly modify individuals more frequently, introduce new genes, try to jump out of the local optimum, and explore a wider solution space; 5) Pareto optimal solution set example: After multiple iterations of optimization, the algorithm may obtain the Pareto optimal solution set shown in Table 1. Each path represents a non-dominated solution that balances inspection time, energy consumption, and safety: Table 1 Examples of Pareto optimal solution sets

[0035] Non-dominated analysis of Pareto solutions: Path 1 (38, 45%, 0.88): Compared with Path 2 (42, 40%, 0.92): Path 1 takes less time, but consumes more energy and is slightly less secure; Compared with Path 3 (35, 50%, 0.85): Path 1 takes longer, consumes less energy, and is more secure; Compared with Path 4 (45, 38%, 0.90): Path 1 is shorter, has higher energy consumption, and is slightly less secure; Conclusion: No other path is better than path 1 in all three objectives, so path 1 is non-dominated; Path 2 (42, 40%, 0.92): Compared with path 1 (38, 45%, 0.88): path 2 takes longer, but consumes less energy and is more secure; Compared with Path 3 (35, 50%, 0.85): Path 2 takes longer, consumes less energy, and is more secure; Compared with path 4 (45, 38%, 0.90): path 2 takes less time, consumes slightly more energy, and is more secure; Conclusion: No other path is better than path 2 in all three objectives, so path 2 is non-dominated; Path 3 (35, 50%, 0.85): Compared with path 1 (38, 45%, 0.88): path 3 takes less time, but consumes more energy and has lower security; Compared with path 2 (42, 40%, 0.92): path 3 takes less time, consumes more energy, and is less secure; Compared with path 4 (45, 38%, 0.90): path 3 takes less time, consumes more energy, and is less secure; Conclusion: No other path is better than path 3 in all three objectives, so path 3 is non-dominated; Path 4 (45, 38%, 0.90): Compared with path 1 (38, 45%, 0.88): path 4 takes longer, consumes less energy, and is slightly more secure; Compared with Path 2 (42, 40%, 0.92): Path 4 takes longer, consumes slightly less energy, and is slightly less secure; Compared with Path 3 (35, 50%, 0.85): Path 4 takes longer, consumes less energy, and is more secure; Conclusion: No other path is better than path 4 in all three objectives, so path 4 is non-dominated; 6) Selection and application of Pareto solution: The above Pareto optimal solution set provides decision makers with a variety of inspection plans, each of which weighs different objectives. The final path to choose depends on the current specific needs and preferences; For example, if the current task has a high time requirement, such as the need to complete the inspection in a short time to deal with an emergency, path 3 may be selected, although it has higher energy consumption and slightly lower safety, but its inspection time is the shortest; If the unmanned ship has a low battery level, or wants to reduce energy consumption for economic reasons, it may choose Path 4, which takes more time but has the lowest energy consumption; If safety is to be considered, in the case of high risk in the current marine environment (high wind speed, high waves, etc.), safety becomes an important consideration. Therefore, path 2 may be preferred because it has the highest safety score among all Pareto solutions; If you want to find a balance, Path 1 provides a relatively balanced solution, achieving a good compromise between time, energy consumption, and security; In addition, in practical applications, the most appropriate path can be dynamically selected from the Pareto optimal solution set according to real-time environmental changes and the status of the unmanned ship, for example: During the inspection, if the weather suddenly deteriorates, the wind and waves increase, and the environmental risk index increases, the currently executed path can be switched to a path with higher security in the Pareto solution set (such as path 2 or path 4); if the battery power of the unmanned boat drops rapidly during the inspection, it can be switched to a path with lower energy consumption (such as path 4 or path 2); In general, the Pareto optimal solution set provides a set of excellent solutions that balance different objectives, providing decision makers with a flexible choice space to choose the most appropriate path according to the specific situation; Step S4: intelligently assigning a number of optimal inspection paths to each unmanned ship, and each unmanned ship performs inspections in the marine cage group area according to the optimal inspection path assigned to it; The core goal of this step is to efficiently assign specific inspection tasks to the most suitable unmanned ship based on the prediction results of the number of unmanned ships and the paths planned by the improved genetic algorithm, combined with the status information of the currently available unmanned ships, and generate corresponding inspection instructions, so as to maximize inspection efficiency, ensure inspection safety, and optimize the utilization rate of unmanned ship resources, and realize intelligent decision-making of task allocation; specifically, it includes the following steps: S4.1: Obtain the prediction result of the number of unmanned ships Predicted_Count and P optimal inspection paths, where P is a positive integer; The prediction result of the number of unmanned ships is obtained by the above-mentioned improved random forest model. For example, it is predicted that 3 unmanned ships are needed. The path optimization result is a number of optimal inspection paths obtained from the Pareto optimal inspection path set generated by the above-mentioned improved genetic algorithm. Each optimal inspection path represents a non-dominated solution that weighs different optimization objectives (time, energy consumption, safety). Each path contains detailed waypoint information and corresponding optimization target evaluation values, including the estimated inspection time, estimated energy consumption and safety score. These evaluation values ​​are directly derived from the calculation results of the objective function. At the same time, this step also needs to obtain the list of available unmanned ships and their status. The current set of unmanned ships available for dispatch is recorded as U = {USV 1 , USV 2 , ..., USV n}; Each unmanned ship USV i The status information includes: ID_i: unique identifier of the unmanned ship; Location_i: current location (latitude and longitude); Battery_Level_i: remaining battery percentage; Performance_State_i: performance state score (e.g., based on sensor health and history, ranging from 0 to 1, with 1 being the best); Idle_Time_i: the time interval from the end of the last task to the current time; Available_Time_i: estimated working time based on current power and historical energy consumption; S4.2: If Predicted_Count < P, that is, the predicted number of unmanned ships is less than the number of optimal inspection paths, then based on any one of the manager preference, fuzzy decision or preset indicator sorting and selection, Predicted_Count paths are selected from the P optimal inspection paths as the paths to be allocated; If Predicted_Count ≥ P, that is, the predicted number of unmanned ships is greater than or equal to the number of optimal inspection paths, then P optimal inspection paths are used as the paths to be allocated; The set of paths to be allocated is recorded as Path unassigned = {Path 1 , Path 2 , ..., Path m}; S4.3: Calculate the matching scores between all available unmanned ships and the paths to be assigned. This embodiment adopts an optimization allocation strategy based on multi-attribute matching. This strategy comprehensively considers the characteristics of the Pareto optimal path and the state of the unmanned ship, and calculates a matching score for each possible unmanned ship-path to be assigned. Finally, the task is assigned according to the score. While taking into account efficiency, this strategy also considers the balanced use of unmanned ships to avoid overuse of some unmanned ships. More importantly, this strategy allows the most appropriate path on the Pareto frontier to be selected for allocation according to the preferences of the manager or system. Specifically, for unmanned ships and the path to be assigned , the matching score between the two is calculated according to the following formula:

[0036] in, represents the matching score between the i-th unmanned ship and the j-th path to be assigned; , , , , and They are distance adaptability weight, time adaptability weight, energy adaptability weight, path safety weight, unmanned ship performance weight and unmanned ship idle time weight. The sum of all weights is 1. The weights can be determined through expert experience or data-driven methods. These weights can be adjusted according to the global task goals. For example, if the current task focuses more on time efficiency, the weight can be appropriately increased. ; It represents the distance score between the i-th unmanned ship and the j-th path to be assigned, ranging from 0 to 1. The closer the distance, the higher the score. The formula is:

[0037] represents the normalized distance between the position of the i-th unmanned ship and the starting point of the j-th path to be assigned; It represents the time adaptability score between the i-th unmanned ship and the j-th path to be assigned, and evaluates the matching degree between the expected inspection time of the path and the available time of the unmanned ship. The higher the score, the better the matching degree. The formula is:

[0038] For path The estimated inspection time is the output of the above inspection time objective function; is the time scale parameter; It represents the energy adaptability score between the i-th unmanned ship and the j-th path to be assigned, and evaluates the matching degree between the expected energy consumption of the path and the remaining power of the unmanned ship. The score is higher when the remaining power is sufficient to complete the task. The formula is:

[0039] For path The estimated energy consumption is the output of the above energy consumption objective function; is the energy conversion coefficient; represents the safety score of the jth path to be assigned, and directly adopts the output of the above inspection safety objective function to ensure the consistency of the task allocation strategy and the path optimization goal; It represents the performance score of the i-th unmanned ship, and directly uses the pre-acquired performance state score Performance_State_i of the unmanned ship; It represents the idle time score of the i-th unmanned ship. The longer the idle time, the higher the score. It is encouraged to assign tasks to unmanned ships with longer idle time. The formula is:

[0040] is the idle time parameter; After each unmanned ship is calculated, repeat the following steps until all the paths to be assigned are assigned: a) Find the current highest-scoring drone ship-path pair: (USV best ,Path best )=argmax{USV i ∈U,Path j ∈Path unassigned} Score(USV i ,Path j ), Set the path best Assigned to USV best ; b) Remove USV from the set of available unmanned vessels U best ; From the path set to be assigned Path unassigned Remove Path best ; That is, the unmanned ship-path pairing is performed based on the highest matching score. After one pairing is completed, the paired unmanned ship and the path are removed from the allocation sequence. The pairing process is repeated to assign all the paths to be assigned to each unmanned ship one by one. Finally, each unmanned boat conducts inspections in the marine cage group area according to the optimal inspection route assigned to it; The innovation of step S4 is that it makes full use of the Pareto optimal solution set provided by the path optimization model to achieve smarter and more flexible task allocation; through the weighted scoring mechanism, it comprehensively considers multiple attributes of the state Pareto optimal path of the unmanned ship (time, energy consumption, safety), so that the allocation decision is no longer the optimization of a single goal, but the result of multi-objective trade-offs; especially when the number of unmanned ships is limited, it can select the most appropriate path from the Pareto frontier according to the preset preferences for allocation, which greatly improves the rationality and efficiency of task allocation; in addition, the allocation strategy also considers the balanced use of unmanned ships, avoiding waste and excessive loss of resources; Secondly, in this embodiment, during the unmanned boat inspection process, the operating status of the unmanned boat is monitored in real time, including location information, remaining power, sensor data, etc., and the monitoring data is fed back to the central control screen; at the same time, the real-time multi-source marine environment data, the operating status of the unmanned boat, the inspection path, the cage information (the geographical location, number, culture species, release time and other related attribute information of the cage) and the inspection results (the analyzed cage damage, floating object type and location, abnormal behavior of fish schools, etc., which can be associated with the time, location, and the unmanned boat performing the inspection) are visualized on the central control screen, and the inspection results can realize the unified monitoring of information; the visual display can display the various data collected, processed and analyzed by this embodiment in an intuitive and easy-to-understand manner, providing real-time monitoring information and decision support for management personnel; For example, in Figure 3 On the large screen of the management center, an electronic map of the entire cage culture area can be displayed in real time. The map dynamically displays the location (indicated by a blue arrow) and trajectory (indicated by a blue thin line) of the unmanned boat being inspected. The unmanned boat that deviates from the planned path is highlighted with a red arrow and a flashing alarm icon. Clicking on the icon of an unmanned boat can view its current power, speed, and surrounding wind speed, wave height and other environmental information. The map is marked with a red exclamation mark icon to indicate the damaged cages detected by the unmanned boat. Clicking on the icon can view the damaged photos and videos taken by the unmanned boat as well as the time. In addition, this embodiment can also perform abnormal monitoring and alarm based on the data displayed visually and updated in real time: a) Path deviation calculation: Compare the real-time position of the unmanned ship with the planned inspection path, and calculate the vertical distance from the current position of the unmanned ship to the nearest path segment as the path deviation; Compare the calculated path deviation with the preset allowable deviation threshold Allowed_Deviation. When the deviation exceeds the threshold but is still within the acceptable range (for example, within 1.5 times of Allowed_Deviation), give a warning prompt on the visualization interface and record the deviation event; When the deviation exceeds the serious deviation threshold (for example, more than 2 times of Allowed_Deviation), immediately trigger a warning and highlight the unmanned ship on the visualization interface; b) Battery monitoring and warning: Set multiple battery warning thresholds, for example: Low battery warning: Battery_Level < Threshold_Low (for example 30%); Critical low battery warning: Battery_Level < Threshold_Critical (for example 15%); The low battery warning triggers a warning prompt on the visualization interface and records the event; The critical low battery warning triggers an emergency warning and gives a prominent prompt on the visualization interface; At the same time, when a battery alarm occurs, start the return journey or task reassignment process; Estimate the remaining endurance time of the unmanned ship based on the current battery level and historical energy consumption data to assist in judging whether an early return is needed; c) Sensor data monitoring and anomaly detection: Set reasonable maximum and minimum thresholds for key sensor data. When the sensor data exceeds the set threshold range, it is determined as an anomaly; Monitor the change rate of sensor data within a short period of time. When the change rate exceeds the preset threshold, it is determined as a mutation; When sensor data is abnormal, trigger a warning prompt on the visualization interface, record the abnormal data, and evaluate whether it is necessary to adjust the inspection plan or conduct remote diagnosis on the unmanned ship; If a certain sensor continuously reports abnormal data or the data is a fixed value, it is determined that the sensor has failed, trigger an emergency warning, and notify the maintenance personnel; d) Communication status monitoring: Adopt a heartbeat mechanism. The unmanned ship regularly sends heartbeat packets to the central control system; Set the communication timeout threshold Communication_Timeout (for example 120 seconds); If no heartbeat packet from the unmanned ship is received within the Communication_Timeout time, it is considered that the communication is interrupted; When the communication is interrupted, immediately trigger an emergency warning and give a prominent prompt on the visualization interface; Start the preset safety measures, such as automatically attempting to re-establish the connection, or instruct the unmanned ship to enter the safe mode (for example, stop the current task and maintain the current position or return automatically); In addition, during the inspection process, the real-time environmental data and status collected by the unmanned ship itself are continuously monitored to determine whether the inspection plan needs to be adjusted; if the environmental conditions change significantly (such as wind speed exceeds the set threshold) or the state of the unmanned ship is abnormal, the steps in the above-mentioned unmanned ship intelligent inspection method are re-executed based on the real-time data collected by the unmanned ship itself, the inspection path is re-evaluated and adjusted, and the inspection path and task allocation of the unmanned ship are dynamically adjusted to ensure the safety and efficiency of the inspection; Specifically, you can set a variety of dynamic adjustment trigger rules: a) Significant changes in environmental conditions: For example, when the wind speed detected by the unmanned ship's own sensors exceeds the preset safety threshold or the direction of the ocean current changes suddenly, multiple levels of thresholds can be set to trigger different degrees of adjustment; b) Abnormal status of the unmanned ship: for example, the battery of the unmanned ship is too low, a fault occurs, the ship deviates too far from the path, or the communication is interrupted; c) Receiving manual intervention instructions: The manager manually initiates adjustment instructions through the visual interface; Based on the above trigger conditions and current situation, select the appropriate adjustment strategy and execute it: a) Route replanning: When environmental conditions are severe, replan a safer route, such as avoiding high wind speed areas or strong current areas; b) Task redistribution: When an unmanned ship fails or runs out of power, the remaining tasks can be assigned to other unmanned ships to redistribute tasks; c) Adjust inspection parameters: for example, adjust the speed of the unmanned boat, the frequency of sensor acquisition, etc. d) Instruct to return home or enter safety mode: In extreme cases, the unmanned ship can be instructed to return home early or enter a preset safety mode; This method can comprehensively utilize the multi-source marine environmental data collected by fixed monitoring stations deployed in the cage group area and the unmanned ship's own sensors to achieve pre-planning of the unmanned ship's inspection path, intelligent task allocation, and dynamic adjustment of the inspection process, thereby improving inspection efficiency, ensuring inspection safety, optimizing resource allocation, and enhancing the scientific level of marine aquaculture management.

[0041] The same or similar reference numerals correspond to the same or similar components; The terms used to describe the positional relationship in the drawings are only used for illustrative purposes and should not be construed as limiting the present invention; Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the embodiments of the present invention. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the embodiments here. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the claims of the present invention.

Claims

1. An intelligent inspection method for marine cage unmanned boats based on multi-source data and environmental perception, characterized in that: The following steps are involved: S1: Collect and pre-process multi-source marine environmental data in the marine cage group area, and obtain cage distribution information, cage inspection task requirements, and the initial status of each unmanned vessel; S2: Based on the pre-processed multi-source ocean environment data, the number of unmanned ships is predicted using an improved random forest model; the improved random forest model introduces an adaptive weight adjustment mechanism based on environmental perception, and the prediction weight of each decision tree is dynamically adjusted according to the real-time multi-source ocean environment data; S3: Based on the pre-processed multi-source marine environment data, cage distribution information, cage inspection task requirements and the initial status of each unmanned ship, an improved genetic algorithm is used to generate several optimal inspection paths; the improved genetic algorithm adaptively adjusts the optimization target and the genetic operator according to the real-time multi-source marine environment data and the status of the unmanned ship; S4: intelligently assigning a number of optimal inspection paths to each unmanned ship, and each unmanned ship performs inspections in the marine cage group area according to the optimal inspection path assigned to it.

2. According to claim 1, a method for intelligent inspection of marine cage unmanned boats based on multi-source data and environmental perception is characterized in that: In the step S1, a plurality of fixed monitoring stations deployed in the marine cage group area and sensors pre-installed on each unmanned vessel are used to collect multi-source marine environmental data of the marine cage group area; The multi-source ocean environment data at least includes: wind speed, wind direction, current speed, current direction, wave height, wave period, wave speed, current speed, current direction and environmental meteorological data; The initial status of each unmanned ship is collected using sensors pre-installed on each unmanned ship.

3. The method for intelligent inspection of marine cage unmanned boats based on multi-source data and environmental perception according to claim 1 is characterized in that: In step S1, the preprocessing includes: Outlier detection and processing: Based on the 3σ principle, outlier detection is performed on all collected data, and the detected outliers are deleted or replaced with the average value of several adjacent normal data; where σ is the standard deviation of the data; Missing value interpolation: use linear interpolation to interpolate missing data; Data standardization: Z-score standardization is performed on all data that have undergone outlier detection and processing, as well as missing value interpolation; Data formatting: convert all data that have undergone data standardization into data of a pre-set unified type and format to complete preprocessing.

4. The method for intelligent inspection of marine cage unmanned boats based on multi-source data and environmental perception according to claim 1 is characterized in that: In step S2, based on the pre-processed multi-source ocean environment data, the improved random forest model is used to predict the number of unmanned ships, including the following steps: S2.1: Obtain historical data and currently collected multi-source marine environmental data; Use historical data to train the traditional random forest regression model and obtain a pre-trained random forest regression model; S2.2: Calculate the environmental risk index Risk_Index and the change rate Change_Rate of each environmental data according to the multi-source marine environmental data, and construct the current environmental feature vector in combination with the calculation results of the environmental risk index Risk_Index and the change rate Change_Rate; S2.3: Calculate the similarity Sim_i between the environmental features of the historical data used in training each decision tree and the current environmental feature vector; and based on the similarity Sim_i, evaluate the historical prediction performance score Perf_i of each decision tree; S2.4: Perform a weighted combination of the similarity Sim_i of each decision tree and the historical prediction performance score Perf_i to obtain the adaptive weight Weight_i of each decision tree; S2.5: Use the pre-trained random forest regression model to predict the number of unmanned ships and obtain the prediction result Prediction_i of each decision tree; S2.6: Take the weighted average of the prediction results Prediction_i of all decision trees according to their adaptive weights Weight_i to obtain the final prediction value Predicted_Count of the number of unmanned ships.

5. The method for intelligent inspection of marine cage unmanned boats based on multi-source data and environmental perception according to claim 4 is characterized in that: In step S2.2, the environmental risk index Risk_Index is calculated according to the following formula: in, is the jth environmental data; is the weight of the jth environmental data; is the jth environmental data The corresponding risk scoring function is used to map the actual value of the environmental data to the interval [0,1]; The environmental change rate Change_Rate is calculated according to the following formula: in, is the change rate of the kth environmental data; is the actual value of the kth environmental data at the current time t; is the time interval; Indicates the absolute value symbol; In the step S2.3, the similarity Sim_i is calculated according to the cosine similarity formula; The historical prediction performance score Perf_i of each decision tree is evaluated according to the following formula: Among them, Perf_i is the historical prediction performance score of the i-th decision tree; Sim(Env_current,Env_historical_j) is the similarity between the current environmental feature vector Env_current and the environmental feature Env_historical_j corresponding to the j-th historical data; Prediction_i_j is the prediction result of the i-th decision tree in the j-th historical data; Actual_j is the actual number of unmanned ships in the j-th historical data; In step S2.4, the adaptive weight Weight_i of each decision tree is calculated according to the following formula: in, is the adaptive weight of the i-th decision tree; is the weight coefficient of similarity, The weight coefficient for historical forecast performance scoring satisfies: ; In step S2.6, the final predicted value of the number of unmanned ships Predicted_Count is obtained according to the following formula: in, is the prediction result of the i-th decision tree; Indicates rounding up.

6. The method for intelligent inspection of marine cage unmanned boats based on multi-source data and environmental perception according to claim 1 is characterized in that: In step S3, using the improved genetic algorithm to generate the optimal inspection path for each unmanned ship includes the following steps: S3.1: Based on the cage distribution information, an initial population is generated by random generation or heuristic method, wherein the initial population is used to represent an initial inspection path set, including a number of individuals, each of which represents a possible inspection path; S3.2: Minimizing inspection time, minimizing energy consumption, and maximizing inspection safety are respectively taken as optimization goals, three conflicting objective functions are set, and different objective function values ​​of each individual are calculated respectively; S3.3: According to the currently collected multi-source ocean environment data and the status of the unmanned ship, the weights of each objective function are adaptively adjusted and the weights are normalized; S3.4: Evaluate the diversity and fitness of the population, and dynamically adjust the crossover probability P_cross and mutation probability P_mutate; S3.5: Population iteration: select a part of individuals from the current population as parents, and exchange some genes of the selected parent individuals according to the dynamically adjusted crossover probability P_cross to generate new offspring individuals; at the same time, randomly change the genes in the individuals according to the dynamically adjusted mutation probability P_mutate, introduce new genes, and increase population diversity; S3.6: Construct and update the Pareto optimal solution set: add the non-dominated solutions in the population iteration process to the Pareto optimal solution set; at the same time, check whether there is a solution dominated by the newly added solution in the Pareto optimal solution set, and if so, remove the dominated solution; S3.7: Determine whether the preset termination iteration condition is met. If not, re-execute steps S3.2~S3.6; if met, select several optimal inspection paths from the Pareto optimal solution set obtained from the last update according to the cage inspection task requirements.

7. The method for intelligent inspection of marine cage unmanned boats based on multi-source data and environmental perception according to claim 6 is characterized in that: In step S3.2, the inspection time objective function is: in, For inspection path The inspection time objective function value; Indicates the inspection path The i-th flight segment in ; For inspection path The inspection time required for the i-th navigation segment; The energy consumption objective function is: in, For inspection path The energy consumption objective function value; For inspection path The energy required for the i-th flight segment; The inspection safety objective function is: in, For inspection path The inspection safety objective function value; is the environmental risk index of the ith navigation segment; is the length of the i-th flight segment; For inspection path The total length of 8. The method for intelligent inspection of marine cage unmanned boats based on multi-source data and environmental perception according to claim 6 is characterized in that: In step S3.3, the weight of the inspection safety objective function and the weight of the inspection time objective function are adaptively adjusted according to the multi-source marine environment data currently collected. The calculation formula is: in, is the weight of the inspection safety objective function at the current time t; is the basic weight of the inspection safety objective function; is the weight adjustment range of the inspection safety objective function; is the weight of the inspection time objective function at the current time t; is the basic weight of the inspection time objective function; is the weight adjustment range of the inspection time objective function; is the Sigmoid function; is the environmental risk index at the current time t; The weight of the energy consumption objective function is adaptively adjusted according to the current state of the unmanned ship. The calculation formula is: in, is the weight of the energy consumption objective function at the current time t; is the basic weight of the energy consumption objective function; is the weight adjustment range of the energy consumption objective function; is the remaining energy percentage of the drone, ranging from [0,1]; After adaptive adjustment , and Normalization is performed separately so that the sum of the three normalized weights is 1.

9. The method for intelligent inspection of marine cage unmanned boats based on multi-source data and environmental perception according to claim 6 is characterized in that: In step S3.4, the diversity of the population is evaluated according to the following formula: in, is the diversity index of the population at the current moment t; N is the population size; represents the distance between individual i and individual j; By continuously observing the optimal fitness values ​​of several generations of populations, the fitness of the population is evaluated, and the fitness stagnation index of the population is calculated according to the following formula: in, is the fitness stagnation indicator of the population; is the optimal fitness value of the g-th generation population; To observe the algebraic interval; When the diversity index of the population is greater than or equal to the preset diversity threshold, only the crossover probability P_cross is increased, and the calculation formula is: in, is the crossover probability at the current time t; is the minimum crossover probability; is the maximum crossover probability; sensitivity coefficient adjusted for crossover probability; is the preset diversity threshold; When the diversity index of the population is less than the preset diversity threshold, only the mutation probability P_mutate is increased, and the calculation formula is: in, is the mutation probability at the current time t; is the minimum mutation probability; is the maximum mutation probability; Sensitivity coefficient adjusted for mutation probability; When the fitness stagnation index of the population is less than or equal to the preset fitness stagnation threshold, the crossover probability P_cross and mutation probability P_mutate are increased at the same time. The calculation formula is: in, is the basic mutation probability; is the adjustment range of the mutation probability; is the preset fitness stagnation threshold; is the average fitness value of the population in the past several generations; is the average fitness value of the current generation population.

10. The method for intelligent inspection of marine cage unmanned boats based on multi-source data and environmental perception according to claim 1, characterized in that: In step S4, a plurality of optimal inspection paths are intelligently allocated to each unmanned ship, including the following steps: S4.1: Obtain the prediction result of the number of unmanned ships Predicted_Count and P optimal inspection paths, where P is a positive integer; S4.2: If Predicted_Count < P, that is, the predicted number of unmanned ships is less than the number of optimal inspection paths, then based on any one of the manager preference, fuzzy decision or preset indicator sorting and selection, Predicted_Count paths are selected from the P optimal inspection paths as the paths to be allocated; If Predicted_Count ≥ P, that is, the predicted number of unmanned ships is greater than or equal to the number of optimal inspection paths, then P optimal inspection paths are used as the paths to be allocated; S4.3: Calculate the matching score Score between an unmanned ship and each path to be assigned according to the following formula: in, represents the matching score between the i-th unmanned ship and the j-th path to be assigned; , , , , and They are distance adaptability weight, time adaptability weight, energy adaptability weight, path safety weight, unmanned ship performance weight and unmanned ship idle time weight; represents the distance score between the i-th unmanned ship and the j-th path to be assigned; represents the time adaptability score between the i-th unmanned ship and the j-th path to be assigned; represents the energy adaptability score between the i-th unmanned ship and the j-th path to be assigned; represents the safety score of the jth path to be assigned; represents the performance score of the i-th unmanned ship; represents the idle time score of the i-th unmanned ship; Assign the to-be-assigned path with the highest matching score to the unmanned ship; S4.4: Repeat step S4.3 to assign other paths to be assigned to other unmanned ships one by one.

Images