Artificial intelligence-based ship behavior identification and dynamic route planning method and system

Abstract

The invention relates to a ship behavior recognition and dynamic route planning method and system based on artificial intelligence, and belongs to the technical field of ship route dynamic planning, and the method comprises the steps: obtaining ship behavior data based on ship data through employing a machine learning and mode recognition technology; the ship behavior data is analyzed by combining a Bayesian network and an anomaly detection algorithm, a risk index is calculated, and then ship behavior risk assessment is carried out; an artificial intelligence optimization algorithm is utilized, the environment data and the ship behavior data are integrated, and a route is generated; in the ship sailing process, ship data and environment data in the ship sailing process are monitored in real time, and the route is adjusted based on the real-time ship data and environment data. According to the invention, the sailing time and the fuel consumption are reduced, the shipping efficiency is improved, the adaptability and the response capability of the ship are improved, and the continuity and the stability of sailing are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of ship path dynamic planning, and in particular to a ship behavior recognition and dynamic route planning method and system based on artificial intelligence. Background Art

[0002] Automatic identification system (AIS) and other ship monitoring systems have become popular. These systems can provide real-time ship dynamic information, such as a real-time virtual enhanced simulation system and method for ship driving based on AIS disclosed in the Chinese patent application publication number CN103631148A. The system consists of director monitoring software, AIS comprehensive processing software, ship driving display and control software, ship power model solving software, electronic chart software, navigation radar software and ship driving vision software. At the beginning of the simulation, the director monitoring software releases the simulation task, and other simulation software simulates according to the simulation task. The director monitoring software records the entire simulation process, and finally replays, evaluates, and analyzes the entire simulation process, and gives the evaluation results.

[0003] However, there is still much room for improvement in the use of these data in existing technologies. For example, how to accurately identify the normal and potential abnormal behaviors of ships from massive data, and how to provide optimal route planning for ships based on complex environmental factors are challenges that need to be urgently addressed in the current navigation field. Summary of the invention

[0004] The present invention provides a method and system for ship behavior recognition and dynamic route planning based on artificial intelligence to solve the problems in the prior art of being unable to accurately identify the normal behavior and potential abnormal behavior of ships, and being unable to plan for ships according to complex environmental factors and adjust the optimal route planning in real time.

[0005] In order to solve the above technical problems, the technical solution provided by the present invention is: The present invention relates to a method for ship behavior recognition and dynamic route planning based on artificial intelligence, which comprises the following steps: S1. Use artificial intelligence optimization algorithms to generate routes based on cost, safety and environmental impact; S2. The ship sails according to the route and collects ship data and environmental data during the voyage; S3. Preprocessing the vessel data and environmental data; S4. Based on the ship data, machine learning and pattern recognition technology are used to obtain ship behavior data; S5. Use anomaly detection algorithms to analyze ship behavior data and calculate risk indexes, and then conduct risk assessments on ship behavior to distinguish between normal and abnormal ship behavior; S6. Adjust the route based on risk assessment of vessel behavior and analysis of environmental data.

[0006] Preferably, the S4 uses machine learning and pattern recognition technology to obtain ship behavior data, classify ship behavior, distinguish normal from abnormal behavior, and then form ship behavior data. The specific steps include: S4.1. A single-layer LSTM network with 128 neurons is used as the feature learning network, and the historical data of the ship is brought into the feature learning network for training; S4.2. Input the collected ship data into the trained feature learning network, use the LSTM network to predict future behavior, and output the ship behavior data.

[0007] Preferably, the specific steps of S5 using anomaly detection algorithm to analyze ship behavior data and calculate risk index are: S5.1. Use the Gaussian mixture model as the anomaly detection model and the historical ship behavior data as the training set. Use the training set to train the anomaly detection model and obtain the anomaly threshold of each behavior data. S5.2. Input the ship behavior data into the trained anomaly detection model, combine the isolation forest and DBSCAN algorithms to detect abnormal behavior and distinguish normal behavior from abnormal behavior; S5.3. Based on normal behavior and abnormal behavior, the random forest algorithm is used to score the ship behavior to form a risk index. Based on the risk index, it is determined whether the ship behavior is risky, and the ship behavior risk assessment is completed.

[0008] Preferably, the specific steps of S1 using artificial intelligence optimization algorithm to generate a route based on comprehensive cost, safety and environmental impact are: S1.1. Plan the navigation route based on the ship's starting point, necessary waypoints and obstacle avoidance areas; S1.2. A cost function is constructed based on the voyage time, fuel consumption and risk index. The cost function is expressed as: C = w1·T + w2·F + w3·R, In the formula, C represents the total cost, T represents the voyage time, F represents the fuel consumption, R represents the risk index, w1, w2 and w3 represent the weights, w1+w2+w3=1; S1.3. Use the hybrid A* algorithm to calculate the cost of the planned route through iterative calculation, and select several routes with lower costs as alternative routes; S1.4. Use numerical weather forecast models and ocean dynamics models to evaluate the safety and environmental impact of alternative routes. Based on the cost, safety and environmental impact of alternative routes, score the alternative routes respectively. Select one route from the alternative routes as the route for the vessel's current voyage according to the score. The score calculation formula is: , In the formula, Score （r） For alternative routes r Rating, w C The weight of the cost, w S is the safety weight, w E is the weight of environmental impact, w C + w S + w E =1, C （r） For alternative routes r The cost, S （r） For alternative routes r The safety factor, S max is the maximum safety factor among all alternative routes. S min is the minimum safety factor among all alternative routes. E （r） For alternative routes r The environmental impact factor, E max is the maximum value of the environmental impact coefficient among all alternative routes, E min It is the minimum value of the environmental impact coefficient among all alternative routes.

[0009] Preferably, the S6 adjusts the route based on the risk assessment of the vessel behavior in combination with the environmental data analysis, and the specific steps include: S6.1. If the behavior of the vessel is judged to be normal in S5, proceed to S6.2; otherwise, recalculate the route in the same manner as in S1; S6.2. Use time series analysis and pattern recognition techniques to identify trends in environmental parameters; S6.3. With reference to the changing trends of environmental parameters, use geographic information systems and risk maps, and adopt fast convergence optimization algorithms to assess potential risks under current environmental conditions and calculate risk coefficients; S6.4. When the risk factor exceeds the threshold, the route is recalculated according to S1.

[0010] 6. The method for ship behavior recognition and dynamic route planning based on artificial intelligence according to claim 1 is characterized in that: the pre-processing of the ship data and environmental data in S3 specifically includes: S3.1. Use checksum or hash function to detect errors in data transmission or entry and delete erroneous data; S3.2. Use the Z-score model to identify outliers. For detected outliers, use the average value of adjacent data points to replace them. S3.3. Identify duplicate data by comparing key values ​​or record tuples and remove duplicate data; S3.4. Identify missing values ​​in the data. When the missing values ​​are in the middle of continuous data, use linear interpolation to fill in the missing values ​​for a short period of time. When the missing values ​​are at the beginning or end of continuous data, use linear extrapolation based on previous and subsequent trends to fill in the missing values. S3.5. Integrate vessel data and environmental data; S3.6. Normalize and standardize the vessel data and environmental data, scaling all numerical data to between 0 and 1, and making the data have zero mean and unit variance; S3.7. Extract feature values ​​from the vessel data and the environmental data, including time domain features and frequency domain features; S3.8. Use Kalman filter to fuse the position data from different sensors.

[0011] The present invention also relates to a system for implementing the above-mentioned artificial intelligence-based ship behavior recognition and dynamic route planning method, characterized in that it comprises: The route planning module is used to generate routes using artificial intelligence optimization algorithms, integrated environmental data and ship behavior data; Data collection module, used to collect vessel data and environmental data; A data processing module, used for pre-processing vessel data and environmental data; A behavior recognition module for acquiring vessel behavior data using machine learning and pattern recognition techniques; Anomaly detection module, which is used to analyze the ship behavior data and calculate the risk index using anomaly detection algorithms, and then conduct risk assessment on the ship behavior and distinguish between normal and abnormal ship behaviors; The dynamic update module is used to summarize the ship's navigation process, monitor the ship's data and environmental data in real time during the navigation process, and adjust the route based on the real-time ship data and environmental data.

[0012] Preferably, it also includes a user interface module for feeding back the route generated by the route planning module and the route adjusted by the dynamic update module to the crew.

[0013] Compared with the prior art, the technical solution provided by the present invention has the following beneficial effects: 1. The artificial intelligence-based ship behavior recognition and dynamic route planning method involved in the present invention utilizes an artificial intelligence optimization algorithm to comprehensively consider cost, safety and environmental impact, that is, by constructing a cost function, the cost of planning a navigation route is calculated by iterative calculation, and several navigation routes with lower costs are screened as alternative routes. In combination with the safety and environmental impact of the alternative routes, a route is selected from the alternative routes and a route is generated, thereby reducing navigation time and fuel consumption and improving shipping efficiency.

[0014] 2. The artificial intelligence-based ship behavior recognition and dynamic route planning method involved in the present invention collects ship data and environmental data in real time, monitors abnormal behaviors such as ship deviation from the route, and evaluates potential risks under current environmental conditions. The route is adjusted based on the abnormal behavior of the ship and the predicted potential risks to adapt to emergencies or environmental changes, thereby improving the adaptability and responsiveness of the ship and ensuring the continuity and stability of navigation.

[0015] 3. The artificial intelligence-based ship behavior recognition and dynamic route planning method involved in the present invention uses machine learning and pattern recognition technology to obtain ship behavior data based on ship data; then combines the anomaly detection algorithm to analyze the ship behavior data, calculate the risk index, and then evaluate the ship behavior risk. It can deeply analyze the behavior pattern of the ship and identify unusual behavior or potential risk events in a timely manner.

[0016] 4. The artificial intelligence-based ship behavior recognition and dynamic route planning method involved in the present invention performs data preprocessing by replacing abnormal values ​​in the collected data, deleting duplicate data, filling missing data, integrating data, normalizing and standardizing data, etc., and performs ship behavior recognition and navigation route planning and adjustment based on the processed data, providing high-quality input data for subsequent analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a flow chart of the ship behavior recognition and dynamic route planning method based on artificial intelligence. DETAILED DESCRIPTION

[0018] In order to further understand the content of the present invention, the present invention is described in detail in conjunction with examples. The following examples are used to illustrate the present invention but are not used to limit the scope of the present invention.

[0019] Refer to the attached Figure 1As shown, the present invention relates to a method for ship behavior recognition and dynamic route planning based on artificial intelligence, which comprises the following steps: S1. Generate routes using artificial intelligence optimization algorithms. The specific steps are: S1.1. Plan the navigation route based on the ship's starting point (latitude / longitude), necessary waypoints, and obstacle avoidance areas (such as waterway traffic separation zones, reef areas, etc.); S1.2. Based on the voyage time, fuel consumption and risk index, a multi-objective optimization method, such as the weighted sum method or the Pareto frontier method, is used to construct a cost function. The cost function is expressed as: C = w1·T + w2·F + w3·R, In the formula, C represents the total cost, T represents the sailing time, F represents the fuel consumption, R represents the risk index, w1, w2 and w3 represent the weights respectively, w1+w2+w3=1; By optimizing this function, planners can find the best route that saves time, fuel and avoids risks. The construction of this cost function reflects the shipping industry's comprehensive goal of pursuing economic benefits and environmental sustainability while ensuring safety. S1.3. Use the hybrid A* algorithm, set the heuristic factors α=1.0 and β=2.0, or use the genetic algorithm (GA), configure the population size to 100, the crossover rate to 0.8, and the mutation rate to 0.02, calculate the cost of the planned route through iterative calculation, and select several routes with lower costs as alternative routes, with at least 3 alternative routes; S1.4. Numerical weather forecast models and ocean dynamics models are used to evaluate the safety and environmental impact of alternative routes. The numerical weather forecast model uses the Global Forecast System (GFS) model. GFS predicts the global atmospheric state by solving a set of equations related to atmospheric dynamics, thermodynamics, radiation and humidity. The ocean dynamics model uses the ocean-atmosphere circulation coupling system plus the wave, tide and current simulation system model. The model constructs a complete ocean dynamics model by analyzing the impact of wind-driven circulation and thermohaline circulation on climate change, and simulating the joint influence of waves, tides and currents on ocean dynamics. Based on the cost, safety and environmental impact of the alternative routes, the alternative routes are scored respectively, and one route is selected from the alternative routes according to the score as the route for the ship's current voyage. The scoring calculation formula is: , In the formula, Score （r） For alternative routes r Rating, w C The weight of the cost, w Sis the safety weight, w E is the weight of environmental impact, w C + w S + w E =1, C （r） For alternative routes r The cost, S （r） For alternative routes r The safety factor, S max is the maximum safety factor among all alternative routes. S min is the minimum safety factor among all alternative routes. E （r） For alternative routes r The environmental impact factor, E max is the maximum value of the environmental impact coefficient among all alternative routes, E min It is the minimum value of the environmental impact coefficient among all alternative routes.

[0020] S2. The vessel sails according to the route and collects vessel data and environmental data during the voyage. The vessel data and environmental data include: receiving AIS signals from the vessel, extracting basic information of the vessel, MMSI number, ship name, location, heading, speed, etc.; in addition to AIS data, various sensors deployed in the marine environment are also important data sources. These sensors include anemometers, wave meters, tide gauges, water temperature meters, etc., which can provide real-time data about the marine environment; in order to have a more comprehensive understanding of the behavior of the vessel and the marine environment, additional information is obtained from third-party organizations such as meteorological stations, satellite data providers, and IMO through the third-party data integration submodule, such as weather forecasts, current maps, and fishing activity areas.

[0021] S3. Preprocess the vessel data and environmental data, including: S3.1. Use checksum or hash function (such as MD5 or SHA-256) to detect errors in data transmission or entry, and delete the erroneous data; S3.2. Use the Z-score model to identify outliers, where the Z-score threshold is set to 3, indicating that data points exceeding 3 standard deviations are considered outliers. For outliers, the average value of adjacent data points is used to replace them. S3.3. Identify duplicate data by comparing key values ​​or record tuples, use matching algorithms such as fuzzy matching to handle near-duplicate data, and eliminate duplicate and near-duplicate data; S3.4. Identify missing values ​​in the data. Any value containing NaN (non-number) or infinity is considered a missing value. When the missing value is in the middle of continuous data, use linear interpolation to fill in the missing value for a short period of time. When the missing value is at the beginning or end of continuous data, use linear extrapolation based on previous and subsequent trends to fill in the missing value. S3.5. Integrate the vessel data and environmental data. The integration includes time synchronization and unit unification. All sensor data are synchronized to UTC time, allowing a maximum deviation of 1 second. All speed-related measurements are converted to knots, and wind speed is converted to meters per second (m / s). S3.6. Normalize and standardize the vessel data and environmental data, scale all numerical data to between 0 and 1, and make the data have zero mean and unit variance. The normalization formula is: N= (X - Xmin) / (Xmax - Xmin), In the formula, N is the new value generated after normalization, X is the original data, Xmin is the minimum value of the original data, and Xmax is the maximum value of the original data; The standardized formula is: , In the formula, Z is the new value generated after standardization, and X is the original data. is the mean value of the feature, is the standard deviation of the feature; S3.7. Extract feature values ​​from vessel data and environmental data, including time domain features and frequency domain features. Time domain features refer to extracting statistical features such as mean, variance, maximum value, minimum value and volatility from time series data. Frequency domain features refer to applying fast Fourier transform (FFT) to analyze periodic features and energy distribution. S3.8. Use Kalman filter to fuse the position data from different sensors to reduce noise and improve the accuracy of position estimation; After data processing, use relational databases such as MySQL to store structured data, and non-relational databases such as MongoDB to store semi-structured data. Back up data regularly, adopt a strategy that combines incremental backup and full backup, and set up a disaster recovery plan to ensure that data can be quickly restored in the event of a system failure.

[0022] S4. Based on the ship data, machine learning and pattern recognition technology are used to obtain ship behavior data. The specific steps include: S4.1. A single-layer LSTM network with 128 neurons is used as the feature learning network, and the historical data of the ship is brought into the feature learning network for training; S4.2. Receive pre-processed JSON format data through REST API, and then input the collected ship data into the trained feature learning network. Each data contains timestamp (UNIX time), GPS coordinates (latitude and longitude), speed (knots), heading (degrees), ship identification code (IMO number), etc. The data is updated once a second, and the LSTM network is used to predict future behavior and output ship behavior data. The ship behavior data includes but is not limited to navigation speed, turning angle, acceleration, route stability (standard deviation), ship type, ship size, etc.

[0023] S5. Use anomaly detection algorithms to analyze ship behavior data, calculate risk index, and then conduct risk assessment on ship behavior. The specific steps are as follows: S5.1. Use the Gaussian mixture model as the anomaly detection model and the historical ship behavior data in the past 30 days as the training set. Use the training set to train the anomaly detection model and obtain the anomaly threshold of each behavior data. The model is trained once a week to reflect the latest behavior pattern. S5.2. Input the ship behavior data into the trained anomaly detection model, and combine the isolation forest and DBSCAN algorithms to detect abnormal behaviors and distinguish normal behaviors from abnormal behaviors. The number of samples of the isolation forest is set to 2000, the neighborhood radius (Eps) of DBSCAN is set to 0.5, and the minimum number of points (MinPts) is set to 5. Each time a new abnormal behavior is detected, it is added to the training set. Whenever at least 100 new abnormal cases are collected, the anomaly detection model is retrained. S5.3. Based on normal behavior and abnormal behavior, the random forest algorithm is used to score the ship behavior to form a risk index. Based on the risk index, it is determined whether the ship behavior has risks. If the risk index is higher than 0.75, it is considered abnormal, and the ship behavior risk assessment is completed.

[0024] S6. Adjust the route based on the risk assessment of vessel behavior and combined with environmental data analysis, including: S6.1. When the risk assessment of vessel behavior shows normal, enter S6.2. When the risk assessment of vessel behavior shows abnormality, the system automatically alarms after it is considered abnormal. For example, a route deviation threshold is set. When the vessel deviates from the route and the deviation is greater than the route deviation threshold, resulting in a risk index higher than 0.75, it is considered abnormal. After the vessel behavior is considered abnormal, the route is recalculated according to S1.1 to S1.4. S6.2. Use time series analysis and pattern recognition techniques to identify trends in environmental parameters; S6.3. With reference to the changing trends of environmental parameters, use geographic information systems and risk maps, and adopt fast convergence optimization algorithms to assess potential risks under current environmental conditions and calculate risk coefficients; S6.4. When the risk factor exceeds the threshold, the route is recalculated according to S1.1 to S1.4; The purpose of recalculating the route in S6.1 and S6.4 is to avoid dangerous areas or abnormal behavior of the ship. After generating a new route, the system will also provide navigation suggestions, such as slowing down or stopping navigation, or even coordinating with other ships to avoid obstacles. The system will feed back the adjusted route and navigation suggestions to the crew, allowing manual intervention to ensure that the final decision and operation are in line with the actual situation and experience judgment.

[0025] The present invention also relates to a system for implementing the above-mentioned artificial intelligence-based ship behavior recognition and dynamic route planning method, which comprises: The route planning module is used to generate routes by using artificial intelligence optimization algorithms to comprehensively consider cost, safety and environmental impact; A data collection module for collecting vessel data and environmental data during navigation; A data processing module, used for pre-processing vessel data and environmental data; A behavior recognition module for acquiring vessel behavior data using machine learning and pattern recognition techniques; Anomaly detection module, which is used to analyze the ship behavior data and calculate the risk index using anomaly detection algorithms, and then conduct risk assessment on the ship behavior and distinguish between normal and abnormal ship behaviors; Dynamic update module, based on risk assessment of vessel behavior, combined with environmental data analysis, to adjust the route; The user interface module is used to feed back the route generated by the route planning module and the route adjusted by the dynamic update module to the crew and perform manual intervention.

[0026] The present invention is described in detail above in conjunction with the embodiments, but the contents described are only preferred embodiments of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made according to the scope of application of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A method for ship behavior recognition and dynamic route planning based on artificial intelligence, characterized in that: It includes the following steps: S1. Use artificial intelligence optimization algorithms to generate routes based on cost, safety and environmental impact; S2. The ship sails according to the route and collects ship data and environmental data during the voyage; S3. Preprocessing the vessel data and environmental data; S4. Based on the ship data, machine learning and pattern recognition technology are used to obtain ship behavior data; S5. Use anomaly detection algorithms to analyze ship behavior data and calculate risk indexes, and then conduct risk assessments on ship behavior to distinguish between normal and abnormal ship behavior; S6. Adjust the route based on risk assessment of vessel behavior and analysis of environmental data.

2. The method for ship behavior recognition and dynamic route planning based on artificial intelligence according to claim 1, characterized in that: The S4 uses machine learning and pattern recognition technology to obtain ship behavior data, classify ship behavior, distinguish normal from abnormal behavior, and then form ship behavior data. The specific steps include: S4.

1. A single-layer LSTM network with 128 neurons is used as the feature learning network, and the historical data of the ship is brought into the feature learning network for training; S4.

2. Input the collected ship data into the trained feature learning network, use the LSTM network to predict future behavior, and output the ship behavior data.

3. The method for ship behavior recognition and dynamic route planning based on artificial intelligence according to claim 1, characterized in that: The specific steps of S5 using anomaly detection algorithm to analyze ship behavior data and calculate risk index are: S5.

1. Use the Gaussian mixture model as the anomaly detection model and the historical ship behavior data as the training set. Use the training set to train the anomaly detection model and obtain the anomaly threshold of each behavior data. S5.

2. Input the ship behavior data into the trained anomaly detection model, combine the isolation forest and DBSCAN algorithms to detect abnormal behavior and distinguish normal behavior from abnormal behavior; S5.

3. Based on normal behavior and abnormal behavior, the random forest algorithm is used to score the ship behavior to form a risk index. Based on the risk index, it is determined whether the ship behavior is risky, and the ship behavior risk assessment is completed.

4. The method for ship behavior recognition and dynamic route planning based on artificial intelligence according to claim 1, characterized in that: The specific steps of S1 using artificial intelligence optimization algorithm to generate routes based on cost, safety and environmental impact are as follows: S1.

1. Plan the navigation route based on the ship's starting point, necessary waypoints and obstacle avoidance areas; S1.

2. A cost function is constructed based on the voyage time, fuel consumption and risk index. The cost function is expressed as: C = w1·T + w2·F + w3·R, In the formula, C represents the total cost, T represents the voyage time, F represents the fuel consumption, R represents the risk index, w1, w2 and w3 represent the weights, w1+w2+w3=1; S1.

3. Use the hybrid A* algorithm to calculate the cost of the planned route through iterative calculation, and select several routes with lower costs as alternative routes; S1.

4. Use numerical weather forecast models and ocean dynamics models to evaluate the safety and environmental impact of alternative routes. Based on the cost, safety and environmental impact of alternative routes, score the alternative routes respectively. Select one route from the alternative routes as the route for the vessel's current voyage according to the score. The score calculation formula is: ， In the formula, Score （r） For alternative routes r Rating, w C The weight of the cost, w S is the safety weight, w E is the weight of environmental impact, w C + w S + w E =1, C （r） For alternative routes r The cost, S （r） For alternative routes r The safety factor, S max is the maximum safety factor among all alternative routes. S min is the minimum safety factor among all alternative routes. E （r） For alternative routes r The environmental impact factor, E max is the maximum value of the environmental impact coefficient among all alternative routes, E min It is the minimum value of the environmental impact coefficient among all alternative routes.

5. The method for ship behavior recognition and dynamic route planning based on artificial intelligence according to claim 1, characterized in that: The S6 adjusts the route based on the risk assessment of the vessel's behavior and the analysis of environmental data. The specific steps include: S6.

1. If the behavior of the vessel is determined to be normal in S5, proceed to S6.2; otherwise, recalculate the route in the same manner as in S1; S6.

2. Use time series analysis and pattern recognition techniques to identify trends in environmental parameters; S6.

3. With reference to the changing trends of environmental parameters, use geographic information systems and risk maps, and adopt fast convergence optimization algorithms to assess potential risks under current environmental conditions and calculate risk coefficients; S6.

4. When the risk factor exceeds the threshold, the route is recalculated according to S1.

6. The method for ship behavior recognition and dynamic route planning based on artificial intelligence according to claim 1, characterized in that: The preprocessing of the vessel data and the environmental data in S3 specifically includes: S3.

1. Use checksum or hash function to detect errors in data transmission or entry and delete erroneous data; S3.

2. Use the Z-score model to identify outliers. For detected outliers, use the average value of adjacent data points to replace them. S3.

3. Identify duplicate data by comparing key values ​​or record tuples and remove duplicate data; S3.

4. Identify missing values ​​in the data. When the missing values ​​are in the middle of continuous data, use linear interpolation to fill in the missing values ​​for a short period of time. When the missing values ​​are at the beginning or end of continuous data, use linear extrapolation based on previous and subsequent trends to fill in the missing values. S3.

5. Integrate vessel data and environmental data; S3.

6. Normalize and standardize the vessel data and environmental data, scaling all numerical data to between 0 and 1, and making the data have zero mean and unit variance; S3.

7. Extract feature values ​​from the vessel data and the environmental data, including time domain features and frequency domain features; S3.

8. Use Kalman filter to fuse the position data from different sensors.

7. A system for implementing the artificial intelligence-based ship behavior recognition and dynamic route planning method of claim 1, characterized in that: It includes: The route planning module is used to generate routes using artificial intelligence optimization algorithms, integrated environmental data and ship behavior data; Data collection module, used to collect vessel data and environmental data; A data processing module, used for pre-processing vessel data and environmental data; A behavior recognition module for acquiring vessel behavior data using machine learning and pattern recognition techniques; Anomaly detection module, which is used to analyze the ship behavior data and calculate the risk index using anomaly detection algorithms, and then conduct risk assessment on the ship behavior and distinguish between normal and abnormal ship behaviors; The dynamic update module is used to summarize the ship's navigation process, monitor the ship's data and environmental data in real time during the navigation process, and adjust the route based on the real-time ship data and environmental data.

8. The artificial intelligence-based ship behavior recognition and dynamic route planning system according to claim 7 is characterized by: It also includes a user interface module, which is used to feed back the route generated by the route planning module and the route adjusted by the dynamic update module to the crew.

Images