Power generation boat driven unmanned boat energy supply method, system, equipment and medium

Abstract

The invention discloses an energy supply method, system, equipment and medium for driving an unmanned ship by a power generation ship, and relates to the technical field of energy management and intelligent scheduling of an unmanned system, and the method comprises the steps: collecting equipment parameters and a scheduling log during the operation of the unmanned ship, and laying the power generation ship with a self-power-generation assembly and multiple types of energy interfaces in a target sea area; a continuous energy consumption data sequence is constructed based on propulsion equipment output parameters and a task scheduling table, a time sequence prediction model is constructed by adopting an LSTM neural network, and the continuous energy consumption data sequence is input into the time sequence prediction model to predict an energy consumption trend in a future time period. Taking the prediction result, the unmanned ship task log, the historical scheduling data and the Euclidean distance between the unmanned ship and the power generation ship as input, substituting the input into a preset scheduling calculation function, and calculating energy complementation priority scores of the plurality of unmanned ships; and according to the energy complementing priority scores of the plurality of unmanned ships, performing energy complementing task sorting, issuing a scheduling instruction through a communication link, and triggering an inter-node response operation. According to the invention, resource allocation optimization is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned system energy management and intelligent scheduling, and specifically to a method, system, equipment and medium for supplying energy to an unmanned boat driven by a power boat. Background Art

[0002] With the development of intelligent marine equipment and unmanned systems, unmanned boats (UAVs) are finding widespread application in ocean mapping, environmental monitoring, border patrol, and maritime communications relay. Because UAVs typically operate over long distances, for extended periods, and in autonomous navigation, reliable energy resupply is crucial for their continued operation. Currently, common resupply methods include shore-based charging, battery replacement onboard a mothership, or recharging using self-contained solar cells. However, these methods have varying degrees of limitations.

[0003] Shore-based charging relies on proximity to ports or refueling stations, limiting its applicability. While battery replacement on motherships allows for remote refueling, it's costly, complex, and carries safety risks. Self-contained solar refueling is constrained by weather conditions and the surface area of unmanned vessels, making it unable to meet the energy demands of high-intensity, long-duration missions. Therefore, a more flexible, efficient, and intelligent offshore energy refueling mechanism is urgently needed.

[0004] With the rise of intelligent floating platform technology and distributed energy management, offshore floating platforms with autonomous power generation capabilities have become a viable energy replenishment node. Combining distributed scheduling algorithms with self-organizing networks of unmanned boats (UAVs) enables remote, dynamic, and multi-vessel coordinated energy replenishment. However, in actual deployments, challenges remain: the energy consumption of UAVs varies significantly over time, making effective scheduling difficult based solely on their current state; competition for energy resources among multiple boats is severe, lacking a unified assessment and prioritization mechanism; scheduling systems struggle to proactively allocate resources based on mission urgency and power trends; and there is no effective mechanism to respond to sudden power crises and communication disruptions.

[0005] Therefore, it is necessary to propose an unmanned boat energy replenishment method that integrates the time series prediction model, dynamic scheduling scoring function and edge execution mechanism to improve the energy autonomy capability and collaborative operation efficiency of unmanned systems in complex environments. Summary of the Invention

[0006] In view of the above-mentioned problems, the present invention is proposed.

[0007] Therefore, the technical problems solved by the present invention are: the existing unmanned boat energy supply scheduling method is unable to predict future energy consumption trends, fails to comprehensively consider task urgency and scheduling response factors, and has low efficiency in energy resource allocation; and how to realize a dynamic and adaptive unmanned boat energy supply scheduling mechanism based on the LSTM prediction model and multi-factor priority scoring function.

[0008] In order to solve the above technical problems, the present invention provides the following technical solutions: a method for supplying energy to an unmanned boat driven by a power boat, comprising collecting equipment parameters and scheduling logs during the operation of the unmanned boat, and deploying a power boat equipped with self-generating components and multiple types of energy interfaces in the target sea area.

[0009] A continuous energy consumption data sequence is constructed based on the output parameters of the propulsion equipment and the task scheduling table. An LSTM neural network is used to build a time series prediction model. The continuous energy consumption data sequence is input into the time series prediction model to predict the energy consumption trend in the future time period. The prediction results, along with the unmanned boat mission log, historical scheduling data, and the Euclidean distance between the unmanned boat and the power boat, are used as input and substituted into the preset scheduling calculation function to calculate the energy replenishment priority scores of multiple unmanned boats.

[0010] The recharging tasks are sorted according to the recharging priority scores of multiple unmanned boats, and scheduling instructions are issued through the communication link to trigger inter-node response operations.

[0011] Calculating the energy replenishment priority scores of multiple unmanned boats includes building a time series prediction model based on the LSTM neural network to predict the power change trend of the unmanned boats in the next 5 to 30 minutes in real time, and using the power forecast value at a specified time point as input variables together with the mission urgency level, historical scheduling response delay and Euclidean distance from the power generation boat to construct a multi-factor weighted scheduling function and dynamically calculate the energy replenishment priority score.

[0012] As a preferred solution of the energy replenishment method for the power boat driven unmanned boat described in the present invention, the collection of equipment parameters and scheduling logs during the operation of the unmanned boat includes collecting the real-time power, current, voltage and power percentage of the propulsion system through ship-borne sensors, and extracting the task number, task type and expected duration as scheduling log fields.

[0013] A time synchronization mechanism is set up to timestamp each set of data and sample it at fixed time intervals. The data is organized into a time series structure through a sliding window and cached in the edge computing node.

[0014] As a preferred solution of the energy replenishment method of the power boat driven unmanned boat described in the present invention, the deployment of the power boat in the target sea area includes selecting multiple self-generating floating boats with solar panels and wind turbines and deploying them in the preset route area. The power boat is equipped with a battery compartment and a dual-mode energy interface, and the interface type is dynamically matched by identifying the unmanned boat ID.

[0015] The energy interface includes a docking mechanical structure and an inductive charging structure.

[0016] The deployment strategy is based on the unmanned boat mission density distribution map and energy consumption hot zone statistics to execute positioning deployment.

[0017] As a preferred solution of the energy replenishment method for unmanned boat driven by power-generating boat described in the present invention, the construction of a continuous energy consumption data sequence includes pairing the collected propulsion motor power, power reduction rate, and attitude adjustment frequency as time series features with the target task start and end time, task category, and task load level in the task scheduling table to form a labeled data sample sequence.

[0018] The data sequence is standardized and interpolated to form a fixed-length sliding window input for time series modeling.

[0019] Building a time series prediction model involves using a three-layer stacked LSTM structure, where each layer contains 128 units and takes propulsion power, power, voltage, wind speed, ship speed, and mission identification as input dimensions.

[0020] The model uses mean square error as the loss function and Adam optimizer as the training optimization method. The initial model is trained offline on historical samples by the shore-based center. After training, the model is fine-tuned in real time locally on the unmanned boat through the edge update mechanism. The update frequency is once every 48 hours or automatically triggered when the error is greater than the update threshold condition.

[0021] A task number embedding mechanism is introduced to vectorize the task number field in the scheduling log and integrate it with the real-time collected propulsion equipment status parameters.

[0022] The introduction of the task number embedding mechanism includes establishing an embedding vector mapping table for the task number, and the embedding table is learned together with other network parameters during the model training process.

[0023] When constructing the input features for each time step, real-time propulsion power, voltage value, battery percentage, and attitude adjustment frequency are included as multi-dimensional continuous features. The continuous features are combined into a state vector, and the task embedding vector is concatenated with the state vector to form an enhanced input vector. The enhanced input vector is used as the standard input of the LSTM unit.

[0024] As a preferred solution of the energy replenishment method for the power boat driven unmanned boat described in the present invention, the prediction of energy consumption trends in future time periods includes inputting a labeled data sample sequence into a time series prediction model, the time series prediction model inputs the time change pattern of the sequence, completes dependent state propagation and output state calculation in the internal memory unit, and outputs an energy change trend sequence within the future prediction time period. The trend sequence outputs the estimated power values corresponding to multiple future time points based on the time point as the index.

[0025] After each prediction output is completed, the latest result is passed to the scheduling function for energy replenishment sorting, and the prediction result is stored in the local cache area for comparison of model errors and to assist in subsequent fine-tuning strategies.

[0026] As a preferred solution of the energy replenishment method for unmanned boats driven by power boats described in the present invention, the calculation of the energy replenishment priority scores of multiple unmanned boats includes inputting the power forecast value at a specified time point in the energy consumption trend curve within the next 5 to 30 minutes output by the LSTM model and the task urgency level corresponding to the unmanned boat task number, the average response delay in the last three scheduling results, and the Euclidean distance between the unmanned boat and the power boat as input variables into a preset scheduling calculation function to calculate the energy replenishment priority score.

[0027] As a preferred solution of the energy replenishment method for the power boat driven unmanned boat described in the present invention, the energy replenishment priority score calculation includes setting standardized parameters respectively, normalizing the standardized parameters, and calculating the energy replenishment priority score for each unmanned boat according to the weighted scheduling function. After all the energy replenishment priority scores are calculated, the energy replenishment priority scores are arranged in descending order.

[0028] Standardized parameters include maximum battery capacity, maximum response delay, and maximum dispatch distance.

[0029] As a preferred solution of the energy replenishment method for unmanned boats driven by power boats described in the present invention, the energy replenishment task sorting includes calculating the energy replenishment quantity threshold based on the resource capacity of the power boat, determining the maximum energy replenishment quantity of the unmanned boats according to the threshold, and sorting all the participating unmanned boats from large to small according to the energy replenishment priority. The sorting results are from front to back until the maximum energy replenishment quantity of the unmanned boats is obtained to obtain the energy replenishment scheduling authority for this round, and the power boats are scheduled and executed in sequence.

[0030] If an unmanned boat does not obtain the authority to schedule energy replenishment in this round and the battery level is lower than the safety threshold, it will send an emergency request signal, and will immediately jump the queue and update the sorting list according to the response rules.

[0031] As a preferred solution of the energy replenishment method for the power boat driven unmanned boat described in the present invention, the sending of scheduling instructions and triggering of inter-node response operations include sending a scheduling instruction package to the target unmanned boat through a satellite-to-ground link or a relay boat link. After receiving the package, the target unmanned boat performs status verification, navigation path confirmation and interface pre-configuration, and enters the energy replenishment waiting state within a specified time period.

[0032] When the power boat status is abnormal during the recharging period, the backup recharging action is executed through the backup relay node and the scheduling log is updated synchronously.

[0033] The instruction package includes the mission number, recharging start time, berthing position and backup node identification.

[0034] Another object of the present invention is to provide an energy supply system for an unmanned boat driven by a power boat, which can solve the problems of the current unmanned boat energy supply scheduling system, such as the lack of real-time prediction and dynamic sorting technology, including delayed energy supply response and uneven resource distribution, by constructing an energy consumption prediction model based on an LSTM neural network and combining a weighted scheduling calculation scheme with multi-source state parameters.

[0035] As a preferred solution of the energy supply system for power-generating boats driving unmanned boats described in the present invention, it includes: a data acquisition and scheduling equipment module, a model construction and calculation scoring module, and a task sequencing and instruction issuing module.

[0036] The data acquisition and scheduling equipment module is used to collect equipment parameters and scheduling logs when the unmanned boat is running, and deploy power-generating boats with self-generating components and multiple types of energy interfaces in the target sea area.

[0037] The model construction calculation scoring module is used to construct a continuous energy consumption data sequence based on the propulsion equipment output parameters and the task scheduling table, use the LSTM neural network to build a time series prediction model, input the continuous energy consumption data sequence into the time series prediction model to predict the energy consumption trend in the future time period, and use the prediction results and the unmanned boat mission log, historical scheduling data and the Euclidean distance between the unmanned boat and the power boat as input, substitute them into the preset scheduling calculation function, and calculate the energy replenishment priority scores of multiple unmanned boats.

[0038] The task sorting and issuing instruction module is used to sort the energy replenishment tasks according to the energy replenishment priority scores of multiple unmanned boats, issue scheduling instructions through the communication link and trigger inter-node response operations.

[0039] A computer device includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement a method for supplying energy to an unmanned boat driven by a power boat.

[0040] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of a method for supplying energy to an unmanned boat driven by a power boat.

[0041] The beneficial effects of this invention are as follows: The proposed method for recharging unmanned boats using power boats accurately predicts future energy consumption trends for unmanned boats by introducing an LSTM neural network prediction model, significantly improving the foresight of recharging scheduling. By constructing a multi-factor priority scoring function that integrates task urgency, response delay, and spatial distance, it achieves dynamic optimization of resource allocation. By setting a scheduling capacity threshold and supporting an emergency queue-jumping mechanism, the system's robustness in emergency scenarios is enhanced. This invention achieves improved scheduling efficiency, energy consumption adaptability, and multi-boat collaboration capabilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. 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 any creative work.

[0043] Figure 1 This is an overall flow chart of a method for supplying energy to an unmanned boat driven by a power boat, provided in the first embodiment of the present invention.

[0044] Figure 2 This is a system flow chart of a method for supplying energy to an unmanned boat driven by a power boat, provided in accordance with the second embodiment of the present invention. DETAILED DESCRIPTION

[0045] To make the above-mentioned objects, features, and advantages of the present invention more clearly understood, the following detailed description of the specific embodiments of the present invention is given in conjunction with the accompanying drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in this field without creative work should fall within the scope of protection of the present invention.

[0046] Example 1, with reference to Figure 1 , as one embodiment of the present invention, provides a method for supplying energy to an unmanned boat driven by a power boat, comprising:

[0047] S1: Collect the equipment parameters and scheduling logs of the unmanned boat during operation, and deploy power boats equipped with self-generating components and multiple types of energy interfaces in the target sea area.

[0048] The real-time power, current, voltage and power percentage of the propulsion system are collected through shipboard sensors, and the task number, task type and expected duration are extracted as scheduling log fields.

[0049] The scheduling log fields include structured log records of task number, type, start and end time, and task level information.

[0050] A time synchronization mechanism is set up to timestamp each set of data and sample it at fixed time intervals. The data is organized into a time series structure through a sliding window and cached in the edge computing node.

[0051] The time series structure includes data collected in continuous time periods organized into a tensor input form of a time series with a fixed step size and length.

[0052] Multiple self-generating floating boats with solar panels and wind turbines are selected and deployed in the preset route area. The power boats are equipped with battery compartments and dual-mode energy interfaces, and the interface type is dynamically matched by identifying the ID of the unmanned boat.

[0053] The energy interface includes a docking mechanical structure and an inductive charging structure.

[0054] The deployment strategy is based on the unmanned boat mission density distribution map and energy consumption hot zone statistics to execute positioning deployment.

[0055] S2: Based on the propulsion equipment output parameters and the task scheduling table, a continuous energy consumption data sequence is constructed. An LSTM neural network is used to build a time series prediction model. The continuous energy consumption data sequence is input into the time series prediction model to predict the energy consumption trend in the future time period. The prediction results and the unmanned boat mission log, historical scheduling data, and the Euclidean distance between the unmanned boat and the power boat are used as input and substituted into the preset scheduling calculation function to calculate the energy replenishment priority scores of multiple unmanned boats.

[0056] The collected propulsion motor power, power reduction rate, and attitude adjustment frequency are used as time series features and paired with the target task start and end time, task category, and task load level in the task scheduling table to form a labeled data sample sequence.

[0057] A preferred solution for forming a sequence of labeled data samples is:

[0058] The data sequence is standardized and interpolated to form a fixed-length sliding window input for time series modeling.

[0059] Building a time series prediction model involves using a three-layer stacked LSTM structure, where each layer contains 128 units and takes propulsion power, power, voltage, wind speed, ship speed, and mission identification as input dimensions.

[0060] The model uses mean square error as the loss function and Adam optimizer as the training optimization method. The initial model is trained offline on historical samples by the shore-based center. After training, the model is fine-tuned in real time locally on the unmanned boat through the edge update mechanism. The update frequency is once every 48 hours or automatically triggered when the error is greater than the update threshold condition.

[0061] An optimal solution for the model to use mean square error as the loss function is:

[0062]

[0063] Among them, L MSE Represents the mean square error loss function in model training, E hat,j represents the predicted power value at the jth time point, E true,jrepresents the actual power value at the jth time point, n represents the number of time series sample points, and j represents the time point.

[0064] The update threshold condition is set according to the experimenter's experience.

[0065] A task number embedding mechanism is introduced to vectorize the task number field in the scheduling log and integrate it with the real-time collected propulsion equipment status parameters;

[0066] The introduction of the task number embedding mechanism includes establishing an embedding vector mapping table for the task number. The embedding table is learned together with other network parameters during the model training process.

[0067] A preferred solution for introducing a task number embedding mechanism is:

[0068] Embed(task_id)→R d

[0069] Among them, task_id represents the current task number, Embed represents the embedding function, d represents the embedding vector dimension, R d represents a d-dimensional real vector space.

[0070] When constructing the input features for each time step, real-time propulsion power, voltage value, battery percentage, and attitude adjustment frequency are included as multi-dimensional continuous features. The continuous features are combined into a state vector, and the task embedding vector is concatenated with the state vector to form an enhanced input vector. The enhanced input vector is used as the standard input of the LSTM unit to update the state at each time step.

[0071] A preferred solution for combining continuous features into a state vector is:

[0072] x t =[p t ,v t ,soc t ,θ t ]∈R 4

[0073] Among them, p t represents the propulsion motor power value at time t, v t Indicates the voltage value at time t, soc t Indicates the percentage of power at time t, θ t Indicates the attitude adjustment frequency at time t.

[0074] A preferred approach to forming the enhanced input vector is:

[0075]

[0076] in, represents the final input vector at time t, Represents the vector concatenation operator, R 4+d Represents the dimension of the concatenated vector space.

[0077] A preferred solution for updating the state of a time step is:

[0078]

[0079] Among them, h t represents the hidden state of LSTM at time t, h t-1 Indicates the hidden state at the previous moment.

[0080] The labeled data sample sequence is input into the time series prediction model. The time series prediction model inputs the time change pattern of the sequence, completes the dependent state propagation and output state calculation in the internal memory unit, and outputs the energy change trend sequence within the future prediction time. The trend sequence outputs the power estimation values corresponding to multiple future time points based on the time point as the index.

[0081] After each prediction output is completed, the latest result is passed to the scheduling function for energy replenishment sorting, and the prediction result is stored in the local cache area for comparison of model errors and to assist in subsequent fine-tuning strategies.

[0082] The power consumption forecast value at a specified time point in the energy consumption trend curve within the next 5 to 30 minutes output by the LSTM model, the task urgency level corresponding to the unmanned boat task number, the average response delay in the last three scheduling results, and the Euclidean distance between the unmanned boat and the power boat are input as input variables into the preset scheduling calculation function to calculate the energy replenishment priority score.

[0083] An optimal solution for calculating the energy replenishment priority score is:

[0084] S i =γ1·e i +γ2·U i +γ3·r i +γ4·d i

[0085] Among them, S i represents the priority score of the i-th unmanned boat, γ1, γ2, γ3, γ4 represent the weighted coefficients of each scoring factor, e i It represents the power tension factor of the i-th unmanned boat, U i represents the mission urgency level of the i-th unmanned boat, r i represents the response delay factor of the i-th unmanned boat, d i Represents the distance factor between the i-th unmanned boat and the power boat.

[0086] The average response delay includes the average response time of the unmanned boat in the last three rounds of dispatch response.

[0087] The task urgency level includes level information corresponding to the task number, indicating the urgency of the task, and is usually an integer level.

[0088] respectively set the standardization parameters and perform normalization on the standardization parameters;

[0089] A preferred solution for normalizing the standardized parameters is:

[0090]

[0091] in, represents the estimated power value of the i-th unmanned boat at the predicted time point, P max Indicates the maximum charge capacity of the battery, R i R represents the average response time of the last three rounds of the i-th unmanned boat, max Indicates the maximum tolerance limit of response time, D i represents the Euclidean distance between the current i-th unmanned boat and the power boat, D max Indicates the maximum dispatch radius.

[0092] The energy replenishment priority score is calculated for each unmanned boat according to the weighted scheduling function. After all the energy replenishment priority scores are calculated, the energy replenishment priority scores are arranged in descending order.

[0093] Standardized parameters include maximum battery capacity, maximum response delay, and maximum dispatch distance.

[0094] S3: Sort the recharging tasks according to the recharging priority scores of multiple unmanned boats, issue scheduling instructions through the communication link and trigger inter-node response operations.

[0095] The energy replenishment quantity threshold is calculated based on the resource capacity of the power boat, and the maximum number of unmanned boats to be replenished is determined according to the threshold. According to the energy replenishment priority, all participating unmanned boats are sorted from large to small according to the energy replenishment priority. The sorting results are sorted from front to back until the maximum number of unmanned boats to be replenished obtain the energy replenishment scheduling authority for this round, and the power boats are dispatched and executed in turn.

[0096] A preferred solution for calculating the energy replenishment quantity threshold is:

[0097]

[0098] Among them, k represents the maximum number of energy-replenishing boats in the current cycle, E s represents the remaining dispatchable energy of the power boat, E represents the average energy consumption of a single unmanned boat, T w Indicates the duration of the current scheduling window, t crepresents the recharging time of a single unmanned boat, and |·| represents the rounding-down operator.

[0099] If an unmanned boat does not obtain the authority to schedule energy replenishment in this round and the battery level is lower than the safety threshold, it will send an emergency request signal, and will immediately jump the queue and update the sorting list according to the response rules.

[0100] The dispatch instruction package is sent to the target unmanned boat through the satellite-ground link or the relay boat link. After receiving it, the target unmanned boat performs status verification, navigation path confirmation and interface pre-configuration, and enters the energy replenishment waiting state within the specified time period.

[0101] When the power boat status is abnormal during the recharging period, the backup recharging action is executed through the backup relay node and the scheduling log is updated synchronously.

[0102] The instruction package includes the mission number, recharging start time, berthing position and backup node identification.

[0103] Example 2, reference Figure 2 , which is an embodiment of the present invention, provides an energy supply system for a power-generating boat-driven unmanned boat, including a data acquisition and scheduling device module 100, a model construction and calculation scoring module 200, and a task sequencing and instruction issuing module 300.

[0104] Among them, S4: the data collection and scheduling equipment module 100 is used to collect equipment parameters and scheduling logs when the unmanned boat is running, and deploy power boats with self-generating components and multiple types of energy interfaces in the target sea area.

[0105] It should also be noted that the collection results are pushed by the collection data scheduling device module 100 to the model construction calculation and scoring module 200. The model construction calculation and scoring module 200 constructs a standardized input sequence from the time series in a sliding window manner, and integrates the embedded vector representation of the task number to form an input sample with task perception capability for use by the three-layer stacked LSTM prediction model.

[0106] S5: The model construction calculation scoring module 200 is used to construct a continuous energy consumption data sequence based on the propulsion equipment output parameters and the task scheduling table, use the LSTM neural network to build a time series prediction model, input the continuous energy consumption data sequence into the time series prediction model to predict the energy consumption trend in the future time period, and use the prediction results and the unmanned boat task log, historical scheduling data and the Euclidean distance between the unmanned boat and the power generation boat as input, substitute them into the preset scheduling calculation function, and calculate the energy replenishment priority scores of multiple unmanned boats.

[0107] It should also be noted that the energy replenishment priority score values of each unmanned boat output by the model construction calculation scoring module 200 are arranged in descending order and passed to the task sorting and issuing instruction module 300. The task sorting and issuing instruction module 300 calculates the maximum number of energy replenishment boats in the current scheduling period according to the preset threshold conditions, and generates a scheduling list based on the sorting results.

[0108] S6: The task sorting and issuing instruction module 300 is used to sort the energy replenishment tasks according to the energy replenishment priority scores of multiple unmanned boats, issue scheduling instructions through the communication link and trigger inter-node response operations.

[0109] It should also be noted that the task sequencing and instruction issuing module 300 rearranges the priority according to the emergency strategy and immediately queues the energy replenishment task. After the energy replenishment action is completed, the scheduling log is updated and synchronized back to the data collection scheduling device module 100, forming a closed-loop collection-prediction-instruction execution process.

[0110] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, and other media that can store program code.

[0111] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0112] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic devices), a portable computer disk cartridge (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering, or processing in another suitable manner as necessary, and then stored in a computer memory.

[0113] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used: a discrete logic circuit having logic gate circuits for implementing logical functions on data signals, an application-specific integrated circuit having suitable combinational logic gate circuits, a programmable gate array (PGA), a field-programmable gate array (FPGA), etc. It should be noted that the above embodiments are merely illustrative of the technical solutions of the present invention and are not intended to be limiting. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced with equivalents without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications should be encompassed by the claims of the present invention.

Claims

1. A method for supplying energy to an unmanned boat driven by a power boat, characterized in that: include: Collect equipment parameters and scheduling logs during the operation of unmanned boats, and deploy power boats equipped with self-generating components and multiple types of energy interfaces in the target sea area; A continuous energy consumption data sequence is constructed based on the propulsion equipment output parameters and the task scheduling table. A time series prediction model is constructed using an LSTM neural network. The continuous energy consumption data sequence is input into the time series prediction model to predict energy consumption trends in future time periods. The prediction results, along with the unmanned boat mission log, historical scheduling data, and the Euclidean distance between the unmanned boat and the power boat, are used as inputs and substituted into a preset scheduling calculation function to calculate the energy replenishment priority scores for multiple unmanned boats. The recharging tasks are sorted according to the recharging priority scores of multiple unmanned boats, and scheduling instructions are issued through the communication link to trigger response operations between nodes; Calculating the energy replenishment priority scores of multiple unmanned boats includes building a time series prediction model based on the LSTM neural network to predict the power change trend of the unmanned boats in real time, and taking the power forecast value at a specified time point as well as the mission urgency level, historical scheduling response delay and Euclidean distance from the power generation boat as input variables to construct a multi-factor weighted scheduling function and dynamically calculate the energy replenishment priority score.

2. The energy supply method for driving an unmanned boat using a power boat as claimed in claim 1, characterized in that: The collection of equipment parameters and scheduling logs during the operation of the unmanned boat includes: The real-time power, current, voltage, and power percentage of the propulsion system are collected through onboard sensors, and the task number, task type, and estimated duration are extracted as scheduling log fields. A time synchronization mechanism is set up to timestamp each set of data and sample it at fixed time intervals. The data is organized into a time series structure through a sliding window and cached in the edge computing node.

3. The energy supply method for driving an unmanned boat using a power boat as claimed in claim 1, characterized in that: The deployment of power generation boats in the target sea area includes: Select multiple self-generating floating boats equipped with solar panels and wind turbines and deploy them in the preset route area. The power boats are equipped with battery compartments and dual-mode energy interfaces, and the interface type is dynamically matched by identifying the ID of the unmanned boat; The energy interface includes a docking mechanical structure and an inductive charging structure; The deployment strategy is based on the unmanned boat mission density distribution map and energy consumption hot zone statistics to execute positioning deployment.

4. The energy supply method for driving an unmanned boat using a power boat according to claim 1, 2 or 3, characterized in that: The constructing of a continuous energy consumption data sequence includes: The collected propulsion motor power, power reduction rate, and attitude adjustment frequency are used as time series features and paired with the target task start and end time, task category, and task load level in the task scheduling table to form a labeled data sample sequence; Normalize and interpolate the data sequence to form a fixed-length sliding window input for time series modeling; The time series prediction model is constructed using a three-layer stacked LSTM structure, with each layer containing 128 units, and taking propulsion power, electricity, voltage, wind speed, ship speed, and mission identification as input dimensions; The model uses mean square error as the loss function and Adam optimizer as the training optimization method. The initial model is trained offline on historical samples by the shore-based center. After training, the model is fine-tuned locally on the unmanned boat in real time through the edge update mechanism. The update frequency is once every 48 hours or automatically triggered when the error exceeds the update threshold condition. A task number embedding mechanism is introduced to vectorize the task number field in the scheduling log and integrate it with the real-time collected propulsion equipment status parameters; The introduction of the task number embedding mechanism includes establishing an embedding vector mapping table for the task number. The embedding table is learned together with other network parameters during the model training process. When constructing the input features for each time step, real-time propulsion power, voltage value, battery percentage, and attitude adjustment frequency are included as multi-dimensional continuous features. The continuous features are combined into a state vector, and the task embedding vector is concatenated with the state vector to form an enhanced input vector. The enhanced input vector is used as the standard input of the LSTM unit.

5. The energy supply method for driving an unmanned boat using a power boat as claimed in claim 4, characterized in that: The predicted energy consumption trend in the future time period includes: The labeled data sample sequence is input into the time series prediction model. The time series prediction model inputs the time change pattern of the sequence, completes the dependent state propagation and output state calculation in the internal memory unit, and outputs the energy change trend sequence within the future prediction time. The trend sequence outputs the power estimation value corresponding to multiple future time points based on the time point as the index; After each prediction output is completed, the latest result is passed to the scheduling function for energy replenishment sorting, and the prediction result is stored in the local cache area for comparison of model errors and to assist in subsequent fine-tuning strategies.

6. The energy supply method for an unmanned boat driven by a power boat according to claim 1, 2, 3 or 5, characterized in that: The calculation of the energy replenishment priority scores of the multiple unmanned boats includes: The power consumption forecast value at a specified time point in the energy consumption trend curve within the next 5 to 30 minutes output by the LSTM model, the task urgency level corresponding to the unmanned boat task number, the average response delay in the last three scheduling results, and the Euclidean distance between the unmanned boat and the power boat are input as input variables into the preset scheduling calculation function to calculate the energy replenishment priority score.

7. The energy supply method for driving an unmanned boat using a power boat as claimed in claim 6, characterized in that: The energy replenishment priority score calculation includes: Standardized parameters are set and normalized respectively, and the energy replenishment priority score is calculated for each unmanned boat according to the weighted scheduling function. After all the energy replenishment priority scores are calculated, the energy replenishment priority scores are arranged in descending order; Standardized parameters include maximum battery capacity, maximum response delay, and maximum dispatch distance.

8. The energy supply method for an unmanned boat driven by a power boat according to claim 1, 2, 3, 5 or 7, characterized in that: The energy replenishment task sequencing includes: The energy replenishment threshold is calculated based on the resource capacity of the power boats. The maximum number of unmanned boats to be recharged is determined based on the threshold. All participating unmanned boats are sorted from highest to lowest according to their recharge priority. The sorting results are sorted from the highest to the lowest, and the unmanned boats with the maximum number of recharges are assigned the energy replenishment scheduling authority for this round. The power boats are then scheduled in turn. If an unmanned boat does not obtain the authority to schedule energy replenishment in this round and the battery level is lower than the safety threshold, it will send an emergency request signal, and will immediately jump the queue and update the sorting list according to the response rules.

9. The method for supplying energy to an unmanned boat driven by a power boat according to claim 8, characterized in that: The sending of scheduling instructions and triggering of inter-node response operations include: The dispatch instruction packet is sent to the target unmanned vehicle via the satellite-to-ground link or the relay boat link. After receiving it, the target unmanned vehicle performs status verification, navigation path confirmation and interface pre-configuration, and enters the energy replenishment waiting state within the specified time period; When the power boat status is abnormal during the recharging period, the backup recharging action is executed through the backup relay node and the scheduling log is updated synchronously; The instruction package includes the mission number, recharging start time, berthing position and backup node identification.

10. An energy supply system for an unmanned boat driven by a power boat, characterized by: Data collection and scheduling equipment module (100), model building and calculation scoring module (200), task sequencing and instruction issuing module (300); The data acquisition and scheduling equipment module (100) is used to collect equipment parameters and scheduling logs during the operation of the unmanned boat, and to deploy a power boat equipped with self-generating components and multiple types of energy interfaces in the target sea area; The model construction calculation scoring module (200) is used to construct a continuous energy consumption data sequence based on the propulsion equipment output parameters and the task scheduling table, use the LSTM neural network to build a time series prediction model, input the continuous energy consumption data sequence into the time series prediction model to predict the energy consumption trend in the future time period, and use the prediction result and the unmanned boat task log, historical scheduling data and the Euclidean distance between the unmanned boat and the power generation boat as inputs, substitute them into a preset scheduling calculation function, and calculate the energy replenishment priority scores of multiple unmanned boats; The task sequencing and issuing instruction module (300) is used to sequence the energy replenishment tasks according to the energy replenishment priority scores of the plurality of unmanned boats, issue scheduling instructions through a communication link, and trigger inter-node response operations.

11. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method for supplying energy to an unmanned boat driven by a power boat according to any one of claims 1 to 9 are implemented.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for supplying energy to an unmanned boat driven by a power boat according to any one of claims 1 to 9 are implemented.

Images