A method, system and computer program for predicting the power of tugboat berthing and unberthing operations
By combining long short-term memory network models and sensor data, the problem of accurate prediction in tugboat power management was solved, improving tugboat operation efficiency and safety, and achieving high-precision prediction of power demand.
Patent Information
- Application Number
- CN202410994811.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-24
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2044-07-24
AI Technical Summary
Traditional tugboat power management methods rely on experience and lack precise calculation and prediction mechanisms, resulting in low efficiency and safety risks. Furthermore, existing methods are unable to reflect changes in operational status in real time.
A tugboat power prediction model is constructed by combining a long short-term memory network model with sensor data and classical wind, wave and current resistance calculations. The power demand is predicted using real-time status information and historical data, and the model is dynamically optimized by identifying operating modes through cluster analysis.
It achieves accurate dynamic prediction under complex operating conditions, improves the efficiency and safety of tugboat operations, and has adaptive capabilities and dynamic optimization mechanisms.
Smart Images

Figure CN118536408B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of port transportation informationization, and in particular to a power prediction method for tugboat berthing and unberthing operations. Background Art
[0002] Tugboats are an indispensable part of port logistics and maritime transportation and play a vital role. Tugboats are mainly used to assist large ships in berthing and unberthing operations to ensure the safety and efficient management of ships. However, tugboats face various complex marine environments and meteorological conditions during their missions, such as wind, waves, and current speed, which directly affect the power requirements of tugboats.
[0003] The traditional tugboat power management method mainly relies on the crew's experience and intuition to judge and adjust, lacking an accurate calculation and prediction mechanism. This method is not only inefficient, but also easily leads to energy waste and operational safety risks. With the development of science and technology, although there are some power demand prediction methods based on physical models, these methods usually cannot reflect the actual state changes in tugboat operations in real time, and therefore it is difficult to meet the precise needs of power management in actual operations. In addition, although modern tugboats are equipped with various internal and external sensors to monitor ship status and environmental information, the effective integration and utilization of these data remains a challenge. How to accurately predict the actual power demand of tugboats under different operating conditions, optimize the operation process of a single tugboat and the coordinated operation process of multiple tugboats, and improve operational efficiency and safety has become an urgent problem to be solved in the industry. Summary of the invention
[0004] In view of the above-mentioned deficiencies in the current technology, the present invention provides a power prediction method for tugboat berthing and unberthing operations, which can solve the technical problem of accurately predicting the power of a tugboat under complex circumstances such as diverse operating modes and difficult power control.
[0005] To achieve the above object, the embodiments of the present invention adopt the following technical solutions:
[0006] A method for predicting the power of a tugboat during berthing and unberthing operations comprises the following steps:
[0007] Obtain tugboat historical status information, operation information and operation mode;
[0008] According to the historical status information, operation information and operation mode of the tugboat, the theoretical tugboat power parameters under the operation mode are obtained;
[0009] According to the historical status information, operation information, operation mode and power parameters of the tugboat, a time series model is constructed at time step t, which receives the status information, operation information and operation mode of the tugboat at the current stage, and predicts the power output result based on the model state at time step t-1 and combines the power parameters of the tugboat, and then trains the model to obtain the tugboat power prediction model;
[0010] Get the status information and operation information of the tugboat's current operation;
[0011] Identify the current operation mode of the tugboat according to the current operation status information and operation information of the tugboat;
[0012] The status information, operation information and operation mode of the tugboat's current operation are input into the tugboat power prediction model to obtain the tugboat power prediction result.
[0013] According to one aspect of the present invention, the long short-term memory network model includes: at time step t, the network receives an input vector composed of tugboat status information, operation information and operation mode at the current stage, and based on the time step The network model status is used to predict the power output results.
[0014] According to one aspect of the present invention, the tugboat historical status information, operation information and operation mode include: basic information of the tugboat, real-time speed and position of the tugboat, wind speed, wave height and flow rate, tugboat operation status, tugboat operation mode and operation vessel information.
[0015] According to one aspect of the present invention, the tugboat power parameters include the predicted total power of the tugboat and the implementation power of the tugboat in each operation phase of assisting the ship.
[0016] According to one aspect of the present invention, the predicted total power of the tugboat is obtained by the following calculation method:
[0017] in: Predict total power for tugboats; The theoretical total power required for the vessel to operate the tugboat service; is the safety factor considered due to climatic conditions, generally 1.2~1.5; CG is the gust factor; They are wind force, current force and wave force on the operating vessel respectively; is the operation safety factor, which is taken as 1.1~1.25.
[0018] According to one aspect of the present invention, the tugboat assists the ship in implementing the power of each operation stage including:
[0019] In the initial stage, the ship still has a certain speed, sufficient self-control main engine power and rudder efficiency. At this stage, the power for the tugboat to assist the ship's operation is mainly the power required for the tugboat to propel itself.
[0020] In the middle stage, the ship begins to enter the maneuvering waters or harbors. At this stage, the power used by the tugboat to assist the ship in operation is the predicted total power of the tugboat.
[0021] The final stage is when the ship turns and berths before entering the dock, or when the ship leaves the dock and turns. The power used by the tugboat to assist the ship in this stage is the predicted total power of the tugboat.
[0022] According to one aspect of the present invention, the method of identifying the current operation mode of the tugboat by using a clustering method according to the current operation status information and operation information of the tugboat includes:
[0023] According to the status information and operation information of the tugboat, a list of trajectory parameters of the tugboat at different times is constructed;
[0024] The first cluster analysis is performed on the trajectory parameter list of the tugboat at different times to identify whether the tugboat's operating mode is free navigation or other modes;
[0025] If the tugboat’s operation mode is identified as other modes, a second cluster analysis is performed for identification;
[0026] Obtain status information of ships around the tugboat and identify ships that the tugboat is assisting in operations;
[0027] The status information of the vessel assisted by the tugboat, the status information of the tugboat and the operation information are integrated and processed;
[0028] A second cluster analysis is performed on the fused data to identify the tugboat's operating mode as berthing or escort mode.
[0029] According to one aspect of the present invention, the method for predicting the power of a tugboat during berthing and unberthing operations further comprises the following steps:
[0030] Obtain actual power demand data after tugboat operation;
[0031] According to the actual power demand data after the tugboat operation, the tugboat power prediction model is dynamically adjusted and optimized.
[0032] A power prediction system for tugboat berthing and unberthing operations, comprising:
[0033] An information acquisition module is used to obtain the tugboat's historical status information, operation information and operation mode;
[0034] The power parameter acquisition module is used to obtain the theoretical tugboat power parameters under the operation mode according to the tugboat historical status information, operation information and operation mode;
[0035] A model building module is used to build a time series model for predicting the power output result based on the historical status information, operation information, operation mode and power parameters of the tugboat at time step t, receiving the status information, operation information and operation mode of the tugboat at the current stage, and training the model state at time step t-1 in combination with the power parameters of the tugboat to obtain a tugboat power prediction model;
[0036] The current information acquisition module is used to obtain the current operating status information and operation information of the tugboat;
[0037] An operation mode identification module is used to identify the current operation mode of the tugboat by using a clustering method according to the current operation status information and operation information of the tugboat;
[0038] The power prediction module is used to input the status information, operation information and operation mode of the tugboat's current operation into the tugboat power prediction model to obtain the tugboat power prediction result.
[0039] According to one aspect of the present invention, the power prediction system for tugboat berthing and unberthing operations further comprises:
[0040] The power data acquisition module is used to obtain the actual power demand data after the tugboat operation;
[0041] The model optimization module is used to dynamically adjust and optimize the tugboat power prediction model according to the actual power demand data after the tugboat operation.
[0042] A computer program, when executed, implements the steps of the method for predicting the power of a tugboat during berthing and unberthing operations as described above.
[0043] Advantages of the present invention:
[0044] The present invention provides a power prediction method for tugboat berthing and unberthing operations. The method uses real-time status information collected by sensors configured on the tugboat, combines classic wind, wave and current resistance calculation formulas and tugboat operation information, calculates the theoretical power demand of the tugboat, and uses a neural network to build a tugboat power prediction model to accurately predict the actual power demand of the tugboat. The method can solve the technical problem of accurately predicting the power of a tugboat under complex circumstances such as diverse operating modes and difficult power control.
[0045] This method has adaptive capabilities and dynamic optimization mechanisms, and can continuously learn and adjust the prediction model based on real-time data and historical operating experience to achieve high-precision prediction of tugboat power requirements.
[0046] This method can significantly improve the energy efficiency and safety of tugboat operations, and is of great significance for promoting the green development and intelligent upgrading of the shipping industry. BRIEF DESCRIPTION OF THE DRAWINGS
[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0048] Figure 1 This is a flow chart of a method for predicting power during tugboat berthing and unberthing operations according to Embodiment 1 of the present invention;
[0049] Figure 2 This is a flow chart of a method for predicting power during tugboat berthing and unberthing operations according to a second embodiment of the present invention. DETAILED DESCRIPTION
[0050] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0051] Embodiment 1
[0052] like Figure 1 As shown, a method for predicting the power of a tugboat during berthing and unberthing operations comprises the following steps:
[0053] S1: Get the historical status information of the tugboat.
[0054] The tugboat historical status information includes: basic information of the tugboat, real-time speed, position, wind speed, wave height and flow rate of the tugboat.
[0055] Various sensors installed on the tugboat are used to sense and collect internal and external status information of the tugboat, including various data during tugboat operations.
[0056] Sensors include: fuel flow meter, speed sensor, GPS positioning system, power sensor, speed sensor, power consumption monitoring sensor, external environment sensor, etc. External environment sensors include wind wave current detection, etc.
[0057] In the process of using shipboard sensors to obtain real-time power information of tugboats, whether diesel or electric tugboats, multiple sources of error may be encountered, which have a direct impact on the accuracy and reliability of the data.
[0058] Sources of error include: (1) The accuracy error of the sensor itself is inevitable, which depends on the design, manufacturing quality and calibration status of the sensor. (2) If the sensor is not calibrated properly, it will cause systematic deviations, further affecting the data quality. (3) Environmental factors such as temperature, humidity and seawater salinity may also affect sensor performance and measurement results. (4) During the data transmission process, problems such as packet loss, delay or signal interference may occur, resulting in incomplete or inaccurate received data. (5) Time synchronization error, that is, if the timestamp of the sensor data is inaccurate or there is a deviation in the time synchronization between different sensors, it may cause data alignment problems, which is particularly critical for power prediction models that rely on time series analysis. (6) Problems with human operation, including sensor installation, maintenance and data entry processes, may also introduce errors.
[0059] In order to reduce the impact of these errors on the accuracy of power prediction, a series of measures can be taken, including regular calibration of sensors, selection of high-quality sensors, implementation of strict data quality control and preprocessing processes, use of advanced data transmission and storage technologies, and consideration of measurement errors during model training. By identifying and addressing these potential sources of error, the accuracy and reliability of tugboat power prediction can be significantly improved.
[0060] During the sensor data collection and processing, it is crucial to ensure the accuracy and integrity of the data. This includes dealing with the problem of inconsistent data formats and ensuring that data from different sensors use a unified format and measurement unit for effective data integration. At the same time, in the face of the risk of data loss, the use of redundant storage and reliable data transmission protocols is a key strategy to reduce data loss. For lost data, you can consider using methods such as interpolation to estimate missing values. Ensuring that the data is correctly aligned in time and space is also a problem that needs attention. Especially when the sensor acquisition frequency is inconsistent or there is a timestamp error, a strict time synchronization protocol needs to be implemented, and appropriate techniques need to be used to align the data during the data preprocessing stage. In addition, sensor data may be affected by factors such as environmental noise or equipment aging, so data cleaning and outlier detection processes need to be implemented to improve data quality. Finally, it is very important to protect the security and privacy of data, which requires the use of encryption, access control, and data anonymization technologies during data transmission, storage, and processing.
[0061] S2: Obtain the tugboat's historical operation information and operation mode.
[0062] The tugboat historical operation information and operation mode include: basic information of the tugboat, real-time speed and position of the tugboat, wind speed, wave height and flow rate, tugboat operation status, tugboat operation mode, and operation vessel information.
[0063] AIS and shipborne radar can be used to obtain information on the status of surrounding ships, and shipborne wind, wave and current sensors can be used to obtain hydrological and meteorological information.
[0064] Tugboat operation information can be obtained from the port dispatch system.
[0065] The tugboat operation status includes free navigation, escort, and assisted berthing and unberthing.
[0066] Tugboat operations include towing and pushing.
[0067] Operating vessel information includes: length, width, draft, etc.
[0068] When obtaining information data from the port dispatching system, it is necessary to design an efficient data synchronization mechanism to ensure real-time acquisition of the latest task status and information, including the use of push or pull methods to obtain data. For systems with high real-time requirements, there should also be a robust error handling mechanism and data verification function to ensure that the received data is accurate and that abnormal situations can be discovered and handled in a timely manner.
[0069] The information data obtained from the port dispatching system must undergo rigorous preprocessing to clean and standardize the data to ensure its quality and consistency. Next, feature engineering is used to extract the most useful information for the prediction model. This step is the key to optimizing the accuracy of the prediction. While ensuring the real-time nature of the data, we also need to balance the accuracy of the prediction results, which may involve the selection and adjustment of the algorithm, as well as the optimization of system performance. This process not only requires the rapid processing of large amounts of data, but also ensures that valuable insights can be extracted from it to support accurate power demand forecasting.
[0070] When obtaining information data from the port dispatch system, data security cannot be ignored. Implementing effective encryption measures and adopting a secure authentication mechanism to prevent data from being intercepted or tampered with during transmission are indispensable steps to ensure data security. In addition, preventing unauthorized access and ensuring data integrity and privacy are also aspects that must be carefully considered. Through these measures, an efficient and secure data exchange environment can be provided.
[0071] S3: According to the historical status information, operation information and operation mode of the tugboat, the theoretical tugboat power parameters under the operation mode are obtained.
[0072] The tugboat power parameters include the predicted total power of the tugboat and the implementation power of the tugboat in assisting the ship in each operation phase.
[0073] The predicted total power of the tugboat is obtained by the following calculation method: in: Predict total power for tugboats; The theoretical total power required for the vessel to operate the tugboat service; is the safety factor considered due to climatic conditions, generally 1.2~1.5; CG is the gust factor; They are wind force, current force and wave force on the operating vessel respectively; is the operation safety factor, which is taken as 1.1~1.25.
[0074] According to the different operating states of the tugboat, such as pushing, towing and escorting, the operating resistance of the tugboat is calculated as the theoretical power demand, that is, the theoretical total power required by the tugboat to serve the operating ship.
[0075] Theoretically, the total power of a tugboat during berthing and unberthing operations should be equal to the sum of the forces acting on the maneuvering ship by wind, waves and currents. However, considering other important influencing factors, additional safety factors need to be considered. These influencing factors generally include: the tugboat must consider sufficient towing redundancy to stop the moving ship quickly; the tugboat does not always tow or push the ship in the right direction. For example, when the ship has a certain forward speed, the tugboat needs to consume a certain amount of power to follow the maneuvering ship, which requires additional power; the tugboat no longer has 100% of the original rated power due to wear and dirt; the tugboat maneuvering the bow and stern of the ship often cannot push and pull at the same time; the water flow driven by the tugboat's propeller impacting the maneuvering ship's hull will reduce the tugboat's towing efficiency; waves will also have a greater impact on the tugboat's efficiency, especially when the wind and waves are strong, and the impact of waves on the tugboat's power needs to be considered.
[0076] There are many methods for calculating the total power demand of tugboats for current berthing and unberthing tasks. The calculation formula is the calculation method for the theoretical total power demand of tugboats proposed by the International Maritime Organization.
[0077] Since the actual total power of the tugboat (including all bow thrusters) and the total effective power required to overcome natural conditions There is a difference between the two, and it is recommended to add an operation safety factor .
[0078] According to the above formula, the tugboat power required for the current task can be simply calculated and used as the basis for subsequent calculations.
[0079] The above theoretical calculations are not fixed and vary according to the various stages of the tugboat's operation in assisting the ship to enter or leave the port.
[0080] The tugboat assists the ship in various operation stages by:
[0081] (1) In the initial stage, the ship still has a certain speed, sufficient main engine power and rudder efficiency for self-control. At this stage, the tugboat does not need to provide too much towing force, but only needs to provide certain assistance to the ship. At this stage, the power for the tugboat to assist the ship's operation is mainly the power required for the tugboat to propel itself;
[0082] (2) In the middle stage, the ship begins to enter the maneuvering waters or harbors. At this stage, the ship's own power gradually decreases to zero, and it gradually loses power and steering effect. It needs the assistance of tugboats to offset the influence of external forces. The power of the tugboat to assist the ship in this stage is the predicted total power of the tugboat. According to the above Calculate by the calculation formula;
[0083] (3) The final stage is when the ship turns and berths before entering the dock, or when the ship leaves the dock and turns. In this stage, the ship's speed is generally zero, so more tugboat power is needed to assist in resisting the influence of external forces. The power of the tugboat to assist the ship in this stage is the predicted total power of the tugboat. According to the above Calculate by the calculation formula;
[0084] S4: Based on the historical status information, operation information, operation mode and power parameters of the tugboat, a time step t is constructed to receive the current stage of the tugboat status information, operation information and operation mode, and based on the time step The model state is combined with the time series model of the tugboat power parameters to predict the power output results and train them to obtain the tugboat power prediction model.
[0085] The tugboat power prediction problem has a typical time-related nature, so most time series models such as LSTM (Long Short-Term Memory Network) or other recurrent neural network (RNN) models can achieve good results. In this embodiment, the LSTM method is used as an embodiment to demonstrate the use of the tugboat power prediction method, and its time series prediction method is also applicable.
[0086] In this embodiment, LSTM is used to construct a neural network model for tugboat power prediction, which fully considers the front-end dependency of time series data, thereby improving the accuracy of power demand prediction.
[0087] LSTM is a special recurrent neural network (RNN) designed to solve the gradient vanishing or gradient exploding problems encountered by standard RNNs when processing long sequence data. LSTM can effectively maintain long-term dependency information by introducing three gates (input gate, forget gate, output gate) and a cell state. In the time series prediction of tugboat power, LSTM can use past state information and current input to predict future power requirements in tugboat operations.
[0088] The long short-term memory network model LSTM includes the following components: input gate, forget gate, output gate and cell state. Through these four gating mechanisms, LSTM can effectively handle long-term dependencies and perform well in various sequence learning tasks.
[0089] The long short-term memory network model LSTM includes: at time step t, the network receives the current stage tugboat operation state input vector , and based on the hidden state formed by the output power state of the previous tugboat operation And the previous node status To predict its future power output state.
[0090] The data obtained by the sensors can form a growing and expanding tugboat operation database, which can be combined with the tugboat's historical power data to correct the power formula and parameter system to improve the accuracy of power prediction. The historical data includes previous operating conditions, power records and environmental conditions.
[0091] The basic process of building a tugboat power prediction model based on LSTM is as follows:
[0092] (1) Data fusion and feature engineering.
[0093] Data fusion is to ensure that real-time status information can be effectively combined with theoretical power demand data, which may involve technologies such as timestamp alignment, data interpolation or synchronization processing to ensure the temporal consistency of the data.
[0094] Feature engineering is the data preparation for the LSTM method, which requires careful design of features to reflect the key factors of the tugboat power demand. This includes extracting useful information from the raw data, such as using statistical features of sliding windows (average, maximum, rate of change, etc.) and the theoretical power demand obtained from the resistance calculation formula.
[0095] (2) Accurate calculation of theoretical power requirements.
[0096] The result of this step comes from the calculation result of step S3, ensuring that the resistance calculation formula used is applicable to the current tugboat and operating conditions. It should be noted that different tugboat designs and different operating environments (such as water depth and sea conditions) may have a significant impact on the power demand.
[0097] (3) Design of LSTM model.
[0098] First, the sequence length needs to be selected. The length of the input sequence should be reasonably selected to capture the time dependence of the power demand. This needs to be determined through experiments to balance computational efficiency and prediction accuracy. Secondly, the model complexity needs to be determined according to the actual situation. The LSTM network structure should be designed according to the amount of available data and the complexity of the features. An overly complex model may lead to overfitting, especially when the amount of data is limited.
[0099] (4) Model training and verification.
[0100] Appropriate training strategies, such as batch training and early stopping, need to be adopted according to the actual situation to prevent overfitting and improve the generalization ability of the model. After each training, the performance of the model should be evaluated, using appropriate evaluation indicators such as mean square error (MSE), root mean square error (RMSE) or mean absolute error (MAE) to measure the prediction performance of the model, and cross-validation should be performed to ensure the robustness of the model.
[0101] When training the LSTM network model, the update method of the LSTM network model unit is as follows:
[0102] (1) Update the forget gate according to the state of the previous stage.
[0103] The forget gate is updated according to the status of the previous stage to decide which information needs to be discarded (such as different job tasks, etc.). The forget gate determines how much previous node status information will be retained.
[0104]
[0105] (2) Transmit tugboat operation information to the input gate.
[0106] The selected tugboat operation information is passed to the input gate of the LSTM model, which determines how much of the current input information will be stored in the node state, including some key operation information (ship’s navigation status, tugboat’s engine status, etc.) and environmental information (wind, wave and current information, etc.).
[0107]
[0108] (3) Update the node status through the information of the forget gate and the input gate.
[0109] Updating the node status of the LSTM network according to the latest input information is the core step of LSTM, which is updated through the information of the forget gate and the input gate.
[0110]
[0111] The output gate outputs the current prediction result.
[0112] The output gate controls how much of the node state information is used as the current hidden state. in, Represents the sigmoid function, which is used to compress the output of the gate between 0 and 1; Provides range adjustment for output values; ⊙ represents Hadamard product (element multiplication); Input vector for the tugboat operation status at the current stage; and is a hidden state formed based on the output power state of the previous tugboat operation and the previous node status; are the activation values of the forget gate, input gate, and output gate at time step t respectively; is the candidate node state at time step t; is the node state at time step t; and are the weights and bias parameters that the network needs to learn, are the weight matrices of the forget gate, input gate, cell state, and output gate, respectively. They are the bias vectors of forget gate, input gate, cell state, and output gate respectively.
[0113] By using these structures and formulas and reasonably adjusting the network structure of LSTM, it is possible to effectively process the time series data in tugboat power demand forecasting, capture long-term dependencies, and thus provide more accurate forecasts.
[0114] S5: Obtain the current operating status information and operation information of the tugboat.
[0115] S6: According to the current operating status information and operating information of the tugboat, a clustering method is used to identify the current operating mode of the tugboat.
[0116] The method of identifying the current operation mode of the tugboat by using a clustering method according to the current operation status information and operation information of the tugboat includes:
[0117] According to the status information and operation information of the tugboat, a list of trajectory parameters of the tugboat at different times is constructed;
[0118] The first cluster analysis is performed on the trajectory parameter list of the tugboat at different times to identify whether the tugboat's operating mode is free navigation or other modes;
[0119] If the tugboat’s operation mode is identified as other modes, a second cluster analysis is performed for identification;
[0120] Obtain status information of ships around the tugboat and identify ships that the tugboat is assisting in operations;
[0121] The status information of the vessel assisted by the tugboat, the status information of the tugboat and the operation information are integrated and processed;
[0122] A second cluster analysis is performed on the fused data to identify the tugboat's operating mode as berthing, unberthing or escort mode.
[0123] In this embodiment, the state parameters of each trajectory point include time, longitude and latitude, speed, heading, speed difference, heading difference, and the rate of change of speed and heading. Tugboat with track points The trajectory Among them, Ship track points:
[0124]
[0125] The first cluster analysis is used to distinguish the free navigation state of the tugboat. The free navigation state of the tugboat is different from the escort and berthing and unberthing states. The tugboat is small and flexible. In order to quickly reach the mission point to perform the mission, the speed of the tugboat in the free navigation state is relatively high. Based on this feature, the tugboat trajectory is clustered for the first time. This clustering identifies the tugboat trajectory as free navigation and other parts (including berthing and escort states). The K-means clustering algorithm can be used to select 6 state parameters except t The tugboat trajectories are clustered, and the clustering results are used to identify whether the tugboat is in a free navigation state according to the characteristics of the tugboat's free navigation state.
[0126] If the first cluster analysis identifies that the tugboat is not in a free-navigation state, a second cluster analysis is performed. The second cluster analysis is used to distinguish between the two states of the tugboat: escort and auxiliary berthing. Before the second clustering, the AIS data of large ships operating with tugboats is first found. The principle of finding data on large ships assisted by tugboats is to use the temporal and spatial relationship between tugboats and large ships to determine, identify the AIS data of large ships within 30 seconds before and after the tugboat AIS trajectory point, and then count the trajectory points of large ships within 400m from the tugboat within this time interval. Some of the ships counted may only be in the same temporal and spatial range as the tugboat for a short time, and do not necessarily have an operational relationship, so large ships with data statistics greater than 15 times that meet the above restrictions are considered to be ships assisting tugboat operations. The second step is to fuse the large ship data found for tugboat auxiliary operations with the tugboat data, and to judge the spatiotemporal relationship of the large ship data again. The point with the smallest time interval and the closest distance to the tugboat AIS data is fused with the current tugboat AIS data, so that the state data of the large ship is used as a state parameter of the tugboat; then the course and speed difference of the large ship is calculated for the fused data. Finally, the fused data is clustered for the second time using the K-means algorithm. After clustering, the classification of classes is identified based on the position difference between the tugboat escort state and the auxiliary berthing and unberthing state. Because the range of the tugboat auxiliary berthing and unberthing state is small, the class with a high position change rate is the escort state data, and vice versa, it is the berthing and unberthing state data. For the tugboat data that does not identify the auxiliary large ship, it is added to the free navigation state data.
[0127] At this point, the second cluster analysis is completed, and the data of tugboat free navigation, escort, berthing and unberthing status are identified. The restriction conditions are:
[0128] in represents the time threshold of the tugboat data, Represents the distance threshold of the tugboat data, Indicates the number of large ship track points that are temporally and spatially associated with the tugboat.
[0129] S7: Inputting the current operating status information, operating information and operating mode of the tugboat into the tugboat power prediction model to obtain the tugboat power prediction result.
[0130] The beneficial effects of this embodiment are: this method can effectively utilize the real-time data and classic physical models in tugboat operations, realize the combination of historical data of the ship and ideal formula, so as to make the power prediction more accurate; by adding various sensors, the real-time measurement of tugboat power can be realized, making the power management of tugboat more refined. This method not only helps to optimize the berthing and unberthing operation process of tugboat, but also improves the efficiency and safety of tugboat operations.
[0131] Embodiment 2
[0132] like Figure 2 As shown, a method for predicting the power of a tugboat during berthing and unberthing operations comprises the following steps:
[0133] S1: Get the historical status information of the tugboat.
[0134] The tugboat historical status information includes: basic information of the tugboat, real-time speed, position, wind speed, wave height and flow rate of the tugboat.
[0135] Various sensors installed on the tugboat are used to sense and collect internal and external status information of the tugboat, including various data during tugboat operations.
[0136] Sensors include: fuel flow meter, speed sensor, GPS positioning system, power sensor, speed sensor, power consumption monitoring sensor, external environment sensor, etc. External environment sensors include wind wave current detection, etc.
[0137] In the process of using shipboard sensors to obtain real-time power information of tugboats, whether diesel or electric tugboats, multiple sources of error may be encountered, which have a direct impact on the accuracy and reliability of the data.
[0138] Sources of error include: (1) The accuracy error of the sensor itself is inevitable, which depends on the design, manufacturing quality and calibration status of the sensor. (2) If the sensor is not calibrated properly, it will cause systematic deviations, further affecting the data quality. (3) Environmental factors such as temperature, humidity and seawater salinity may also affect sensor performance and measurement results. (4) During the data transmission process, problems such as packet loss, delay or signal interference may occur, resulting in incomplete or inaccurate received data. (5) Time synchronization error, that is, if the timestamp of the sensor data is inaccurate or there is a deviation in the time synchronization between different sensors, it may cause data alignment problems, which is particularly critical for power prediction models that rely on time series analysis. (6) Problems with human operation, including sensor installation, maintenance and data entry processes, may also introduce errors.
[0139] In order to reduce the impact of these errors on the accuracy of power prediction, a series of measures can be taken, including regular calibration of sensors, selection of high-quality sensors, implementation of strict data quality control and preprocessing processes, use of advanced data transmission and storage technologies, and consideration of measurement errors during model training. By identifying and addressing these potential sources of error, the accuracy and reliability of tugboat power prediction can be significantly improved.
[0140] During the sensor data collection and processing, it is crucial to ensure the accuracy and integrity of the data. This includes dealing with the problem of inconsistent data formats and ensuring that data from different sensors use a unified format and measurement unit for effective data integration. At the same time, in the face of the risk of data loss, the use of redundant storage and reliable data transmission protocols is a key strategy to reduce data loss. For lost data, you can consider using methods such as interpolation to estimate missing values. Ensuring that the data is correctly aligned in time and space is also a problem that needs attention. Especially when the sensor acquisition frequency is inconsistent or there is a timestamp error, a strict time synchronization protocol needs to be implemented, and appropriate techniques need to be used to align the data during the data preprocessing stage. In addition, sensor data may be affected by factors such as environmental noise or equipment aging, so data cleaning and outlier detection processes need to be implemented to improve data quality. Finally, it is very important to protect the security and privacy of data, which requires the use of encryption, access control, and data anonymization technologies during data transmission, storage, and processing.
[0141] S2: Obtain the tugboat's historical operation information and operation mode.
[0142] The tugboat historical operation information and operation mode include: basic information of the tugboat, real-time speed and position of the tugboat, wind speed, wave height and flow rate, tugboat operation status, tugboat operation mode, and operation vessel information.
[0143] AIS and shipborne radar can be used to obtain information on the status of surrounding ships, and shipborne wind, wave and current sensors can be used to obtain hydrological and meteorological information.
[0144] Tugboat operation information can be obtained from the port dispatch system.
[0145] The tugboat operation status includes free navigation, escort, and assisted berthing and unberthing.
[0146] Tugboat operations include towing and pushing.
[0147] Operating vessel information includes: length, width, draft, etc.
[0148] When obtaining information data from the port dispatching system, it is necessary to design an efficient data synchronization mechanism to ensure real-time acquisition of the latest task status and information, including the use of push or pull methods to obtain data. For systems with high real-time requirements, there should also be a robust error handling mechanism and data verification function to ensure that the received data is accurate and that abnormal situations can be discovered and handled in a timely manner.
[0149] The information data obtained from the port dispatching system must undergo rigorous preprocessing to clean and standardize the data to ensure its quality and consistency. Next, feature engineering is used to extract the most useful information for the prediction model. This step is the key to optimizing the accuracy of the prediction. While ensuring the real-time nature of the data, we also need to balance the accuracy of the prediction results, which may involve the selection and adjustment of the algorithm, as well as the optimization of system performance. This process not only requires the rapid processing of large amounts of data, but also ensures that valuable insights can be extracted from it to support accurate power demand forecasting.
[0150] When obtaining information data from the port dispatch system, data security cannot be ignored. Implementing effective encryption measures and adopting a secure authentication mechanism to prevent data from being intercepted or tampered with during transmission are indispensable steps to ensure data security. In addition, preventing unauthorized access and ensuring data integrity and privacy are also aspects that must be carefully considered. Through these measures, an efficient and secure data exchange environment can be provided.
[0151] S3: According to the historical status information, operation information and operation mode of the tugboat, the theoretical tugboat power parameters under the operation mode are obtained.
[0152] The tugboat power parameters include the predicted total power of the tugboat and the implementation power of the tugboat in assisting the ship in each operation phase.
[0153] The predicted total power of the tugboat is obtained by the following calculation method: in: Predict total power for tugboats; The theoretical total power required for the vessel to operate the tugboat service; is the safety factor considered due to climatic conditions, generally 1.2~1.5; CG is the gust factor; They are wind force, current force and wave force on the operating vessel respectively; is the operation safety factor, which is taken as 1.1~1.25.
[0154] According to the different operating states of the tugboat, such as pushing, towing and escorting, the operating resistance of the tugboat is calculated as the theoretical power demand, that is, the theoretical total power required by the tugboat to serve the operating ship.
[0155] Theoretically, the total power of a tugboat during berthing and unberthing operations should be equal to the sum of the forces acting on the maneuvering ship by wind, waves and currents. However, considering other important influencing factors, additional safety factors need to be considered. These influencing factors generally include: the tugboat must consider sufficient towing redundancy to stop the moving ship quickly; the tugboat does not always tow or push the ship in the right direction. For example, when the ship has a certain forward speed, the tugboat needs to consume a certain amount of power to follow the maneuvering ship, which requires additional power; the tugboat no longer has 100% of the original rated power due to wear and dirt; the tugboat maneuvering the bow and stern of the ship often cannot push and pull at the same time; the water flow driven by the tugboat's propeller impacting the maneuvering ship's hull will reduce the tugboat's towing efficiency; waves will also have a greater impact on the tugboat's efficiency, especially when the wind and waves are strong, and the impact of waves on the tugboat's power needs to be considered.
[0156] There are many methods for calculating the total power demand of tugboats for current berthing and unberthing tasks. The calculation formula is the calculation method for the theoretical total power demand of tugboats proposed by the International Maritime Organization.
[0157] Since the actual total power of the tugboat (including all bow thrusters) and the total effective power required to overcome natural conditions There is a difference between the two, and it is recommended to add an operation safety factor .
[0158] According to the above formula, the tugboat power required for the current task can be simply calculated and used as the basis for subsequent calculations.
[0159] The above theoretical calculations are not fixed and vary according to the various stages of the tugboat's operation in assisting the ship to enter or leave the port.
[0160] The tugboat assists the ship in various operation stages by:
[0161] (1) In the initial stage, the ship still has a certain speed, sufficient main engine power and rudder efficiency for self-control. At this stage, the tugboat does not need to provide too much towing force, but only needs to provide certain assistance to the ship. At this stage, the power for the tugboat to assist the ship's operation is mainly the power required for the tugboat to propel itself;
[0162] (2) In the middle stage, the ship begins to enter the maneuvering waters or harbors. At this stage, the ship's own power gradually decreases to zero, and it gradually loses power and steering effect. It needs the assistance of tugboats to offset the influence of external forces. The power of the tugboat to assist the ship in this stage is the predicted total power of the tugboat. According to the above Calculate by the calculation formula;
[0163] (3) The final stage is when the ship turns and berths before entering the dock, or when the ship leaves the dock and turns. In this stage, the ship's speed is generally zero, so more tugboat power is needed to assist in resisting the influence of external forces. The power of the tugboat to assist the ship in this stage is the predicted total power of the tugboat. According to the above Calculate by the calculation formula;
[0164] S4: According to the historical status information, operation information, operation mode and power parameters of the tugboat, a time series model is constructed at time step t, which receives the status information, operation information and operation mode of the tugboat in the current stage, and predicts the power output result based on the model state at time step t-1 and combines the power parameters of the tugboat, and trains the model to obtain the tugboat power prediction model.
[0165] The tugboat power prediction problem has a typical time-related nature, so most time series models such as LSTM (Long Short-Term Memory Network) or other recurrent neural network (RNN) models can achieve good results. In this embodiment, the LSTM method is used as an embodiment to demonstrate the use of the tugboat power prediction method, and its time series prediction method is also applicable.
[0166] In this embodiment, LSTM is used to construct a neural network model for tugboat power prediction, which fully considers the front-end dependency of time series data, thereby improving the accuracy of power demand prediction.
[0167] LSTM is a special recurrent neural network (RNN) designed to solve the gradient vanishing or gradient exploding problems encountered by standard RNNs when processing long sequence data. LSTM can effectively maintain long-term dependency information by introducing three gates (input gate, forget gate, output gate) and a cell state. In the time series prediction of tugboat power, LSTM can use past state information and current input to predict future power requirements in tugboat operations.
[0168] The long short-term memory network model LSTM includes the following components: input gate, forget gate, output gate and cell state. Through these four gating mechanisms, LSTM can effectively handle long-term dependencies and perform well in various sequence learning tasks.
[0169] The long short-term memory network model LSTM includes: , the network receives the current stage of the tugboat operation state input vector , and based on the hidden state formed by the output power state of the previous tugboat operation And the previous node status To predict its future power output state.
[0170] The data obtained by the sensors can form a growing and expanding tugboat operation database, which can be combined with the tugboat's historical power data to correct the power formula and parameter system to improve the accuracy of power prediction. The historical data includes previous operating conditions, power records and environmental conditions.
[0171] The basic process of building a tugboat power prediction model based on LSTM is as follows:
[0172] (1) Data fusion and feature engineering.
[0173] Data fusion is to ensure that real-time status information can be effectively combined with theoretical power demand data, which may involve technologies such as timestamp alignment, data interpolation or synchronization processing to ensure the temporal consistency of the data.
[0174] Feature engineering is the data preparation for the LSTM method, which requires careful design of features to reflect the key factors of the tugboat power demand. This includes extracting useful information from the raw data, such as using statistical features of sliding windows (average, maximum, rate of change, etc.) and the theoretical power demand obtained from the resistance calculation formula.
[0175] (2) Accurate calculation of theoretical power requirements.
[0176] The result of this step comes from the calculation result of step S3, ensuring that the resistance calculation formula used is applicable to the current tugboat and operating conditions. It should be noted that different tugboat designs and different operating environments (such as water depth and sea conditions) may have a significant impact on the power demand.
[0177] (3) Design of LSTM model.
[0178] First, the sequence length needs to be selected. The length of the input sequence should be reasonably selected to capture the time dependence of the power demand. This needs to be determined through experiments to balance computational efficiency and prediction accuracy. Secondly, the model complexity needs to be determined according to the actual situation. The LSTM network structure should be designed according to the amount of available data and the complexity of the features. An overly complex model may lead to overfitting, especially when the amount of data is limited.
[0179] (4) Model training and verification.
[0180] Appropriate training strategies, such as batch training and early stopping, need to be adopted according to the actual situation to prevent overfitting and improve the generalization ability of the model. After each training, the performance of the model should be evaluated, using appropriate evaluation indicators such as mean square error (MSE), root mean square error (RMSE) or mean absolute error (MAE) to measure the prediction performance of the model, and cross-validation should be performed to ensure the robustness of the model.
[0181] When training the LSTM network model, the update method of the LSTM network model unit is as follows:
[0182] (1) Update the forget gate according to the state of the previous stage.
[0183] The forget gate is updated according to the status of the previous stage to decide which information needs to be discarded (such as different job tasks, etc.). The forget gate determines how much previous node status information will be retained.
[0184] (2) Transmit tugboat operation information to the input gate.
[0185] The selected tugboat operation information is passed to the input gate of the LSTM model, which determines how much of the current input information will be stored in the node state, including some key operation information (ship’s navigation status, tugboat’s engine status, etc.) and environmental information (wind, wave and current information, etc.).
[0186] (3) Update the node status through the information of the forget gate and the input gate.
[0187] Updating the node status of the LSTM network according to the latest input information is the core step of LSTM, which is updated through the information of the forget gate and the input gate.
[0188] (4) The output gate outputs the current prediction result.
[0189] The output gate controls how much of the node state information is used as the current hidden state. in, Represents the sigmoid function, which is used to compress the output of the gate between 0 and 1; Provides range adjustment for output values; ⊙ represents Hadamard product (element multiplication); Input vector for the tugboat operation status at the current stage; and is a hidden state formed based on the output power state of the previous tugboat operation and the previous node status; are the activation values of the forget gate, input gate, and output gate at time step t respectively; is the candidate node state at time step t; is the node state at time step t; and are the weights and bias parameters that the network needs to learn, are the weight matrices of the forget gate, input gate, cell state, and output gate, respectively. They are the bias vectors of forget gate, input gate, cell state, and output gate respectively.
[0190] By using these structures and formulas and reasonably adjusting the network structure of LSTM, it is possible to effectively process the time series data in tugboat power demand forecasting, capture long-term dependencies, and thus provide more accurate forecasts.
[0191] S5: Obtain the current operating status information and operation information of the tugboat.
[0192] S6: According to the current operating status information and operating information of the tugboat, a clustering method is used to identify the current operating mode of the tugboat.
[0193] The method of identifying the current operation mode of the tugboat by using a clustering method according to the current operation status information and operation information of the tugboat includes:
[0194] According to the status information and operation information of the tugboat, a list of trajectory parameters of the tugboat at different times is constructed;
[0195] The first cluster analysis is performed on the trajectory parameter list of the tugboat at different times to identify whether the tugboat's operating mode is free navigation or other modes;
[0196] If the tugboat’s operation mode is identified as other modes, a second cluster analysis is performed for identification;
[0197] Obtain status information of ships around the tugboat and identify ships that the tugboat is assisting in operations;
[0198] The status information of the vessel assisted by the tugboat, the status information of the tugboat and the operation information are integrated and processed;
[0199] A second cluster analysis is performed on the fused data to identify the tugboat's operating mode as berthing, unberthing or escort mode.
[0200] In this embodiment, the state parameters of each trajectory point include time, longitude and latitude, speed, heading, speed difference, heading difference, and the rate of change of speed and heading. Tugboat with track points The trajectory Among them, Ship track points:
[0201]
[0202] The first cluster analysis is used to distinguish the free navigation state of the tugboat. The free navigation state of the tugboat is different from the escort and berthing and unberthing states. The tugboat is small and flexible. In order to quickly reach the mission point to perform the mission, the speed of the tugboat in the free navigation state is relatively high. Based on this feature, the tugboat trajectory is clustered for the first time. This clustering identifies the tugboat trajectory as free navigation and other parts (including berthing and escort states). The K-means clustering algorithm can be used to select 6 state parameters except t The tugboat trajectories are clustered, and the clustering results are used to identify whether the tugboat is in a free navigation state according to the characteristics of the tugboat's free navigation state.
[0203] If the first cluster analysis identifies that the tugboat is not in a free-navigation state, a second cluster analysis is performed. The second cluster analysis is used to distinguish between the two states of the tugboat: escort and auxiliary berthing. Before the second clustering, the AIS data of large ships operating with tugboats is first found. The principle of finding data on large ships assisted by tugboats is to use the temporal and spatial relationship between tugboats and large ships to determine, identify the AIS data of large ships within 30 seconds before and after the tugboat AIS trajectory point, and then count the trajectory points of large ships within 400m from the tugboat within this time interval. Some of the ships counted may only be in the same temporal and spatial range as the tugboat for a short time, and do not necessarily have an operational relationship, so large ships with data statistics greater than 15 times that meet the above restrictions are considered to be ships assisting tugboat operations. The second step is to fuse the large ship data found for tugboat auxiliary operations with the tugboat data, and to judge the spatiotemporal relationship of the large ship data again. The point with the smallest time interval and the closest distance to the tugboat AIS data is fused with the current tugboat AIS data, so that the state data of the large ship is used as a state parameter of the tugboat; then the course and speed difference of the large ship is calculated for the fused data. Finally, the fused data is clustered for the second time using the K-means algorithm. After clustering, the classification of classes is identified based on the position difference between the tugboat escort state and the auxiliary berthing and unberthing state. Because the range of the tugboat auxiliary berthing and unberthing state is small, the class with a high position change rate is the escort state data, and vice versa, it is the berthing and unberthing state data. For the tugboat data that does not identify the auxiliary large ship, it is added to the free navigation state data.
[0204] At this point, the second cluster analysis is completed, and the data of tugboat free navigation, escort, berthing and unberthing status are identified. The restriction conditions are:
[0205]
[0206] in represents the time threshold of the tugboat data, Represents the distance threshold of the tugboat data, Indicates the number of large ship track points that are temporally and spatially associated with the tugboat.
[0207] S7: Inputting the current operating status information, operating information and operating mode of the tugboat into the tugboat power prediction model to obtain the tugboat power prediction result.
[0208] S8: Acquire actual power demand data after the tugboat operation.
[0209] S9: Dynamically adjust and optimize the tugboat power prediction model based on the actual power demand data after the tugboat operation.
[0210] In this method, according to the actual needs of the project, two important contents are the automatic adjustment of parameters of the LSTM prediction network and the selection of data update strategy.
[0211] Automatic parameter adjustment includes two parts: model fine-tuning and online learning. Model fine-tuning involves regularly updating or fine-tuning model parameters using newly collected data after tugboat operations to adapt to changes in tugboat operations and environmental conditions. Online learning allows the model to continuously learn from new data without having to retrain from scratch.
[0212] Since the tugboat operation data in the database is dynamically updated, it is necessary to consider formulating strategies to regularly integrate the newly collected data into the training set, and dynamically adjust and optimize the tugboat power prediction model to reflect the latest operating conditions and power requirements. At the same time, the quality of the new data, including accuracy, completeness and consistency, should be ensured to maintain the model performance.
[0213] The beneficial effect of this embodiment is that the method has adaptive capabilities and a dynamic optimization mechanism, and can continuously learn and adjust the prediction model based on real-time data and historical operating experience to achieve high-precision prediction of tugboat power requirements.
[0214] Embodiment 3
[0215] A power prediction system for tugboat berthing and unberthing operations, comprising:
[0216] An information acquisition module is used to obtain the tugboat's historical status information, operation information and operation mode;
[0217] A power parameter acquisition module is used to obtain the power parameters of the tugboat according to the historical status information, operation information and operation mode of the tugboat;
[0218] A model building module is used to build and train a long short-term memory network model based on the tugboat's historical status information, operation information, operation mode, and tugboat power parameters to obtain a tugboat power prediction model;
[0219] The current information acquisition module is used to obtain the current operating status information and operation information of the tugboat;
[0220] An operation mode identification module is used to identify the current operation mode of the tugboat by using a clustering method according to the current operation status information and operation information of the tugboat;
[0221] The power prediction module is used to input the status information, operation information and operation mode of the tugboat's current operation into the tugboat power prediction model to obtain the tugboat power prediction result.
[0222] This system needs to communicate and collaborate with the port dispatching system, so it is necessary to define clear interface specifications, including data format, transmission method and security protocol, which is the basis for ensuring smooth communication between the two systems. On this basis, an efficient data synchronization mechanism is designed to ensure that the tugboat power prediction system obtains the latest task status and information in real time, including the use of push or pull methods to obtain data. For systems with high real-time requirements, there should also be a robust error handling mechanism and data verification function to ensure that the received data is accurate and abnormal situations can be discovered and handled in a timely manner.
[0223] Data processing and analysis are the core links in the system. First of all, all data received from the port dispatching system, including operation status, operation mode and basic information of the ship, must undergo strict pre-processing to clean and standardize the data to ensure its quality and consistency. Next, the most useful information for the prediction model is extracted through feature engineering. This step is the key to optimizing the accuracy of the prediction. While ensuring the real-time nature of the data, we also need to balance the accuracy of the prediction results, which may involve the selection and adjustment of the algorithm, as well as the optimization of system performance. This process not only requires the rapid processing of large amounts of data, but also ensures that valuable insights can be extracted from it to support accurate power demand forecasting.
[0224] Considering the particularity of the port production scheduling system, communication protocols and security occupy an important position in the tugboat power prediction system. Selecting a suitable communication protocol, such as MQTT, AMQP or WebSocket, is a prerequisite for ensuring efficient and stable data transmission. At the same time, data security cannot be ignored. Implementing end-to-end encryption measures and adopting a secure authentication mechanism to prevent data from being intercepted or tampered with during transmission are indispensable steps to ensure system security. In addition, preventing unauthorized access and ensuring data integrity and privacy are also aspects that must be carefully considered when designing the system. Through these measures, an efficient and secure data exchange environment can be provided for the tugboat power prediction system.
[0225] Preferably, the power prediction system for tugboat berthing and unberthing operations described in this embodiment further includes:
[0226] The power data acquisition module is used to obtain the actual power demand data after the tugboat operation;
[0227] The model optimization module is used to dynamically adjust and optimize the tugboat power prediction model according to the actual power demand data after the tugboat operation.
[0228] Embodiment 4
[0229] A computer program, when executed, implements the steps of the method for predicting the power of a tugboat berthing and unberthing operations as described in Embodiment 1 or 2.
[0230] Embodiment 5
[0231] A readable storage medium having stored thereon a computer program as described in Example 4, wherein when the computer program is executed, the steps of the method for predicting the power of tugboat berthing and unberthing operations as described in Example 1 or 2 are implemented.
[0232] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by any technician familiar with the art within the technical scope disclosed in the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.
Claims
1. A method for predicting the power of a tugboat during berthing and unberthing operations, characterized in that: The following steps are involved: Obtain tugboat historical status information, operation information and operation mode; According to the historical status information, operation information and operation mode of the tugboat, the power parameters of the tugboat in the operation mode are obtained; According to the historical status information, operation information, operation mode and power parameters of the tugboat, a time series model is constructed at time step t, which receives the status information, operation information and operation mode of the tugboat at the current stage, and predicts the power output result based on the model state at time step t-1 and combines the power parameters of the tugboat, and then trains the model to obtain the tugboat power prediction model; Get the status information and operation information of the tugboat's current operation; According to the current operating status information and operating information of the tugboat, the clustering method is used to identify the current operating mode of the tugboat; Inputting the current operating status information, operating information and operating mode of the tugboat into the tugboat power prediction model to obtain the tugboat power prediction result; The time series model is a long short-term memory network model; The long short-term memory network model includes the following components: input gate, forget gate, output gate and cell state; When training the LSTM network model, the LSTM network model unit is updated as follows: Update the forget gate according to the state of the previous stage; Passing tugboat operation information to the input gate; Update the node status through the information of forget gate and input gate; The output gate outputs the current prediction result; The tugboat power parameters include the predicted total power of the tugboat and the implementation power of the tugboat assisting the ship in each operation phase; The predicted total power of the tugboat is obtained by the following calculation method: in: Predict total power for tugboats; The theoretical total power required for the vessel to operate the tugboat service; is the safety factor considered due to climatic conditions, which is 1.2~1.5; CG is the gust factor; , , They are wind force, current force and wave force on the operating vessel respectively; is the operation safety factor, which is 1.1~1.25; The tugboat assists the ship in various operation stages by: In the initial stage, the ship still has speed, self-controlled main engine power and rudder effect. At this stage, the power used by the tugboat to assist the ship is the power required for the tugboat to propel itself. In the middle stage, the ship begins to enter the maneuvering waters or harbors. At this stage, the power used by the tugboat to assist the ship in operation is the predicted total power of the tugboat. The final stage is when the ship turns and berths before entering the dock, or when the ship leaves the dock and turns. The power used by the tugboat to assist the ship in this stage is the predicted total power of the tugboat.
2. The method for predicting the power of tugboat berthing and unberthing operations according to claim 1, characterized in that: The tugboat's status information, operation information and operation mode include: basic information of the tugboat, real-time speed and position of the tugboat, wind speed, wave height and flow rate, tugboat's operation status, tugboat's operation mode and operation vessel information.
3. The method for predicting the power of tugboat berthing and unberthing operations according to claim 1, characterized in that: The method of identifying the current operation mode of the tugboat by using a clustering method according to the current operation status information and operation information of the tugboat includes: According to the status information and operation information of the tugboat, a list of trajectory parameters of the tugboat at different times is constructed; The first cluster analysis is performed on the trajectory parameter list of the tugboat at different times to identify whether the tugboat's operating mode is free navigation or other modes; If the tugboat’s operation mode is identified as other modes, a second cluster analysis is performed for identification; Obtain status information of ships around the tugboat and identify ships that the tugboat is assisting in operations; The status information of the vessel assisted by the tugboat, the status information of the tugboat and the operation information are integrated and processed; A second cluster analysis is performed on the fused data to identify the tugboat's operating mode as berthing, unberthing or escort mode.
4. The method for predicting the power of a tugboat berthing and unberthing operations according to claim 1, characterized in that: The method for predicting the power of a tugboat during berthing and unberthing operations further comprises the following steps: Obtain actual power demand data after tugboat operation; According to the actual power demand data after the tugboat operation, the tugboat power prediction model is dynamically adjusted and optimized.
5. A power prediction system for tugboat berthing and unberthing operations, characterized in that: include: An information acquisition module is used to obtain the tugboat's historical status information, operation information and operation mode; A power parameter acquisition module is used to acquire the power parameters of the tugboat in the operation mode according to the historical status information, operation information and operation mode of the tugboat; A model building module is used to build a time series model for predicting the power output result based on the historical status information, operation information, operation mode and power parameters of the tugboat at time step t, receiving the status information, operation information and operation mode of the tugboat at the current stage, and training the model state at time step t-1 in combination with the power parameters of the tugboat to obtain a tugboat power prediction model; The current information acquisition module is used to obtain the current operating status information and operation information of the tugboat; An operation mode identification module is used to identify the current operation mode of the tugboat by using a clustering method according to the current operation status information and operation information of the tugboat; A power prediction module is used to input the current operating status information, operating information and operating mode of the tugboat into the tugboat power prediction model to obtain the tugboat power prediction result; The time series model is a long short-term memory network; The long short-term memory network model includes the following components: input gate, forget gate, output gate and cell state; When training the LSTM network model, the LSTM network model unit is updated as follows: Update the forget gate according to the state of the previous stage; Passing tugboat operation information to the input gate; Update the node status through the information of forget gate and input gate; The output gate outputs the current prediction result; The tugboat power parameters include the predicted total power of the tugboat and the implementation power of the tugboat assisting the ship in each operation phase; The predicted total power of the tugboat is obtained by the following calculation method: in: Predict total power for tugboats; The theoretical total power required for the vessel to operate the tugboat service; is the safety factor considered due to climatic conditions, which is 1.2~1.5; CG is the gust factor; , , They are wind force, current force and wave force on the operating vessel respectively; is the operation safety factor, which is 1.1~1.25; The tugboat assists the ship in various operation stages by: In the initial stage, the ship still has speed, self-controlled main engine power and rudder effect. At this stage, the power used by the tugboat to assist the ship is the power required for the tugboat to propel itself. In the middle stage, the ship begins to enter the maneuvering waters or harbors. At this stage, the power used by the tugboat to assist the ship in operation is the predicted total power of the tugboat. The final stage is when the ship turns and berths before entering the dock, or when the ship leaves the dock and turns. The power used by the tugboat to assist the ship in this stage is the predicted total power of the tugboat.
6. The power prediction system for tugboat berthing and unberthing operations according to claim 5, characterized in that: The power prediction system for tugboat berthing and unberthing operations also includes: The power data acquisition module is used to obtain the actual power demand data after the tugboat operation; The model optimization module is used to dynamically adjust and optimize the tugboat power prediction model according to the actual power demand data after the tugboat operation.
7. A computer program product, characterized in that When the computer program product is executed, the steps of the method for predicting the power of a tugboat berthing and unberthing operations as claimed in any one of claims 1 to 4 are implemented.
