Marine moving target drift trajectory prediction method and device and electronic equipment

Abstract

The invention provides a sea moving target drift trajectory prediction method and device and electronic equipment. The method comprises the following steps: acquiring first motion information of a sea moving target in a current time period; inputting the first motion information into a trained LSTM model, and outputting a first preliminary drift trajectory prediction result of the maritime moving target in a prediction time period; taking the first preliminary drift trajectory prediction result as an expected result of a first reverse planning model, and performing reverse solution on the first reverse planning model to obtain a first final drift trajectory prediction result of the marine moving target in the prediction time period; according to the method, precise prediction of the marine moving target is realized through a method of combining an LSTM model and a reverse programming model.

Description

Technical Field

[0001] The present invention belongs to the technical field of ocean monitoring and data analysis, and in particular relates to a method, a device and an electronic device for predicting the drift trajectory of a moving target at sea. Background Art

[0002] The drift trajectory of moving targets at sea is affected by many factors, including wind direction, wind speed, ocean currents, tides, waves, and the target's own motion characteristics.

[0003] Traditional trajectory prediction methods are usually based on physical models, such as ocean flow models and wave models. Although these models can reflect the physical mechanism to a certain extent, their computational complexity is high and it is difficult to fully consider all influencing factors, resulting in limited prediction accuracy. Summary of the invention

[0004] In a first aspect, an embodiment of the present invention provides a method for predicting the drift trajectory of a moving target at sea, comprising: obtaining first motion information of a moving target at sea in a current time period; inputting the first motion information into a trained LSTM model, and outputting a first preliminary drift trajectory prediction result of the moving target at sea in a prediction time period; using the first preliminary drift trajectory prediction result as the expected result of a first reverse planning model, and reversely solving the first reverse planning model to obtain a first final drift trajectory prediction result of the moving target at sea in the prediction time period.

[0005] In some embodiments, the first preliminary drift trajectory prediction result is taken as the expected result of the first reverse planning model, and the first reverse planning model is reversely solved to obtain the first final drift trajectory prediction result of the marine moving target in the prediction time period, including: constructing a first reverse planning model including an objective function and a first constraint condition, wherein the objective function is to minimize the difference between the drift trajectory prediction result and the expected result, and the first constraint condition is a first constraint on the motion parameters in the first motion information; taking the first preliminary drift trajectory prediction result output by the LSTM model as the expected result, and reversely solving the first final drift trajectory prediction result that satisfies the objective function and the first constraint condition.

[0006] In some embodiments, the method also includes: obtaining a first actual drift trajectory result of the marine moving target in a predicted time period; and determining the prediction accuracy of the trained LSTM model and the first reverse planning model based on the difference between the first actual drift trajectory result and the first final drift trajectory prediction result.

[0007] In some embodiments, before inputting the first motion information into the trained LSTM model, it also includes: obtaining second motion information of the marine moving target in a historical time period and the corresponding second actual drift trajectory result, the second operation information and the second actual drift trajectory result constitute a data set; training the LSTM model based on the data set to obtain the trained LSTM model.

[0008] In some embodiments, the first motion information and the second motion information include at least one of the following motion parameters: geographic location data, meteorological environment data, and heading data of the moving target at sea; the training of the LSTM model based on the data set also includes: dynamically adjusting the prediction time period of the LSTM model according to the heading data.

[0009] In some embodiments, the heading data includes at least one of the following: heading, speed; then the dynamic adjustment of the prediction time period of the LSTM model according to the heading data includes: when the heading change amplitude or the speed is less than the corresponding first preset threshold, adjusting the prediction time period of the LSTM model to a first time magnitude; when the heading change amplitude or the speed is greater than the corresponding first preset threshold, adjusting the prediction time period of the LSTM model to a second time magnitude, and the first time magnitude is greater than the second time magnitude.

[0010] In some embodiments, when the heading change amplitude or the speed is less than the corresponding first preset threshold, the prediction time period of the LSTM model is adjusted to the first time level, including: when the heading change amplitude or the speed is less than the corresponding first preset threshold, the prediction time period of the LSTM model is gradually adjusted to the first time level by a rolling prediction method.

[0011] In some embodiments, after obtaining the trained LSTM model, it also includes: inputting the second motion information into the trained LSTM model, and outputting a second preliminary drift trajectory prediction result of the marine motion target in the historical time period; constructing a second reverse planning model including an objective function and a second constraint condition, wherein the objective function is to minimize the difference between the drift trajectory prediction result and the expected result, and the second constraint condition is a second constraint on the motion parameters in the second motion information; taking the second preliminary drift trajectory prediction result output by the LSTM model as the expected result, and reversely solving the second final drift trajectory prediction result that satisfies the objective function and the second constraint condition; and determining that the trained LSTM model is available when it is determined that the difference between the second actual drift trajectory result and the second final drift trajectory prediction result is less than a second preset threshold.

[0012] In the second aspect, an embodiment of the present invention provides a device for predicting the drift trajectory of a moving target at sea, comprising: an information acquisition module, used to obtain the first motion information of the moving target at sea in the current time period; a preliminary prediction module, used to input the first motion information into a trained LSTM model, and output a first preliminary drift trajectory prediction result of the moving target at sea in the prediction time period; and a final prediction module, used to use the first preliminary drift trajectory prediction result as the expected result of the first reverse planning model, reversely solve the first reverse planning model, and obtain the first final drift trajectory prediction result of the moving target at sea in the prediction time period.

[0013] In a third aspect, an embodiment of the present invention provides an electronic device, including a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus; the memory is used to store computer programs; the processor is used to execute the program stored in the memory to implement any of the first aspect.

[0014] The beneficial effects brought by the present invention are as follows:

[0015] It can be seen from the above scheme that the embodiment of the present invention provides a method, device and electronic device for predicting the drift trajectory of a moving target at sea. First, the LSTM model is used to predict the preliminary drift trajectory prediction results of the moving target at sea in the future period of time, and then the preliminary drift trajectory prediction results output by the LSTM model are optimized and adjusted in combination with the reverse planning method to obtain the final drift trajectory prediction results, thereby improving the prediction accuracy and interpretability, and providing technical support for the fields of maritime traffic safety, marine environmental protection and emergency rescue. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A schematic diagram of a flow chart of a method for predicting drift trajectory of a moving target at sea provided by an embodiment of the present invention;

[0017] Figure 2 A flowchart of a training method for an LSTM model provided in an embodiment of the present invention;

[0018] Figure 3 A schematic flow chart of another method for predicting drift trajectory of a moving target at sea provided by an embodiment of the present invention;

[0019] Figure 4 A schematic diagram of the structure of a device for predicting drift trajectory of a moving target at sea provided by an embodiment of the present invention;

[0020] Figure 5 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is a part of the embodiment of the present invention, not all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0022] The drift trajectory of moving targets at sea is affected by many factors, including wind direction, wind speed, ocean currents, tides, waves, and the target's own motion characteristics. Traditional trajectory prediction methods are usually based on physical models, such as ocean flow models and wave models. Although these models can reflect the physical mechanism to a certain extent, their computational complexity is high and it is difficult to fully consider all influencing factors, resulting in limited prediction accuracy.

[0023] In view of the above technical problems, the technical concept of the present invention is that: with the rapid development of artificial intelligence and big data technology, data-driven prediction methods have gradually become a research hotspot. Among them, Long Short-Term Memory (LSTM) as a special recurrent neural network (RNN) can process sequence data with long-term dependencies and performs well in the field of time series prediction. However, when using the LSTM model alone to predict the drift trajectory of moving targets at sea, there are still some problems, such as difficulty in optimizing model parameters, high risk of overfitting, and lack of interpretability of prediction results. The reverse planning method is an optimization method based on objective functions and constraints. The optimal or suboptimal decision-making scheme is obtained by reverse solving, which can be used as a supplement to the above prediction method. Therefore, the present invention realizes accurate prediction of the drift trajectory of moving targets at sea by combining the LSTM model and the reverse planning method.

[0024] Figure 1 A schematic diagram of a flow chart of a method for predicting drift trajectory of a moving target at sea provided by an embodiment of the present invention. Figure 1 As shown, the drift trajectory prediction method of the marine moving target includes:

[0025] Step S101: Acquire the first motion information of the marine moving target in the current time period.

[0026] Specifically, the moving targets at sea are ships, buoys, drifting objects, etc.; the current time period refers to the current continuous period of time; the first motion information includes at least one of the following motion parameters: geographic location data, meteorological environment data and heading data of the movement of the sea target, wherein the geographic location data includes longitude, latitude, timestamp, etc., the meteorological environment data includes wind direction, wind speed, current speed, etc., and the heading data includes heading, speed, etc.

[0027] Taking a ship as an example, the ship's relevant sensors are used to collect its movement data over a continuous period of time. For example, the longitude and latitude of the ship and timestamp are obtained through Beidou equipment, and the meteorological environment of the ship including important environmental factors such as wind direction, wind speed, ocean currents, and waves are obtained through meteorological instruments. The ship's heading and speed are obtained through speed and direction measuring instruments.

[0028] Preferably, after obtaining the first motion information of the moving target at sea in the current time period, the first motion information is preprocessed, including data cleaning, data calibration and data format conversion, wherein data cleaning is used to eliminate outliers and duplicate records in the data, data calibration is used to align data from different sources in space and time to ensure the accuracy and consistency of the data, and data format conversion is used to convert the timestamp data collected by each sensor into a unified format to facilitate subsequent data analysis and model training.

[0029] Step S102: input the first motion information into the trained LSTM model, and output the first preliminary drift trajectory prediction result of the marine motion target in the prediction time period.

[0030] Specifically, the prediction time period can also be called the prediction cycle, which refers to a period of time in the future. In this step, the preprocessed first motion information is input into the trained LSTM model in chronological order, and the feature extraction is performed through the hidden layer of the LSTM model to obtain the motion feature vector of the marine motion target, and then the preliminary drift trajectory prediction result of the marine motion target in the future period of time is output. The input sequence of the LSTM model is the motion parameter data in a continuous period of time, and the output sequence is the preliminary drift trajectory result of the target in the future period of time.

[0031] Step S103: taking the first preliminary drift trajectory prediction result as the expected result of the first reverse planning model, performing reverse solution on the first reverse planning model to obtain a first final drift trajectory prediction result of the marine moving target in the prediction time period.

[0032] Specifically, the core of the reverse planning method is to reversely solve the optimal or suboptimal decision-making plan according to the requirements and constraints of the prediction target. In this step, after obtaining the first preliminary drift trajectory prediction result output by the LSTM model, the reverse planning method is used to optimize the first preliminary drift trajectory prediction result, that is, the first preliminary drift trajectory prediction result output by the LSTM model is used as the expected result, and the final drift trajectory prediction result of the marine moving target in the future period of time is reversely solved.

[0033] In some embodiments, step S103 includes: constructing a first reverse planning model including an objective function and a first constraint condition, wherein the objective function is to minimize the difference between the drift trajectory prediction result and the expected result, and the first constraint condition is a first constraint on the motion parameters in the first motion information; taking the first preliminary drift trajectory prediction result output by the LSTM model as the expected result, and inversely solving the first final drift trajectory prediction result that satisfies the objective function and the first constraint condition.

[0034] Specifically, this embodiment further defines that the first reverse planning model includes an objective function and a first constraint, wherein the objective function is determined according to the prediction requirement, the prediction requirement is to accurately predict the drift trajectory of the marine moving target within a period of time in the future, and the objective function is to minimize the error between the drift trajectory prediction result and the expected result, and the first constraint is a constraint on the current motion parameters of the marine moving target, including the motion characteristics of the marine moving target (speed, heading, etc.), environmental factors (wind direction, wind speed, ocean currents, waves, etc.); then, the prediction result of the LSTM model is used as the expected solution, and the reverse planning method is used for reverse solution to obtain the final solution that satisfies the objective function and the first constraint. Preferably, in the process of reverse solution using the reverse planning method, heuristic search, simulated annealing and other optimization algorithms can be used to improve the solution efficiency and accuracy.

[0035] In some embodiments, the method also includes: obtaining a first actual drift trajectory result of the marine moving target in a predicted time period; and determining the prediction accuracy of the trained LSTM model and the first reverse planning model based on the difference between the first actual drift trajectory result and the first final drift trajectory prediction result.

[0036] Specifically, the actual drift trajectory results of the moving targets at sea during the prediction period are observed, and the final drift trajectory prediction results predicted by the LSTM model and the first reverse planning model are compared with the actual drift trajectory results. The prediction error and accuracy of the prediction model composed of the LSTM model and the first reverse planning model are determined to evaluate the performance of the prediction method of this embodiment.

[0037] The drift trajectory prediction method for marine moving targets provided in this embodiment first uses the LSTM model to predict the preliminary drift trajectory prediction results of marine moving targets in the future period of time, and then optimizes and adjusts the preliminary drift trajectory prediction results output by the LSTM model in combination with the reverse planning method to obtain the final drift trajectory prediction results, thereby improving the prediction accuracy and interpretability; the performance of the prediction method of this embodiment is evaluated by verifying and evaluating the optimized final drift trajectory prediction results, calculating indicators such as prediction error and accuracy.

[0038] Based on the above embodiments, Figure 2 A flow chart of a training method for an LSTM model provided in an embodiment of the present invention, that is, before executing step S102, the following steps are also included:

[0039] Step S201: obtaining second motion information of a marine moving target in a historical time period and a corresponding second actual drift trajectory result, wherein the second motion information and the second actual drift trajectory result constitute a data set.

[0040] Step S202: Train the LSTM model based on the data set to obtain the trained LSTM model.

[0041] Specifically, before applying the LSTM model, the LSTM model needs to be trained first, and the LSTM model is trained using supervised learning. First, the historical motion data corresponding to multiple historical time periods and the corresponding actual drift trajectory results are obtained to form a data set of multiple samples; then, each sample in the data set is input into the LSTM model, and the parameters of the model are adjusted through the back propagation algorithm so that the drift trajectory results predicted by the LSTM model are as close as possible to the actual drift trajectory results. During the training process, methods such as cross-validation can be used to evaluate the performance of the model to prevent overfitting.

[0042] In some embodiments, the step S202 further includes the following steps:

[0043] Step S203: input the second motion information into the trained LSTM model, and output the second preliminary drift trajectory prediction result of the marine moving target in the historical time period.

[0044] Step S204: construct a second reverse planning model including an objective function and a second constraint condition, wherein the objective function is to minimize the difference between the drift trajectory prediction result and the expected result, and the second constraint condition is a second constraint on the motion parameters in the second motion information.

[0045] Step S205: taking the second preliminary drift trajectory prediction result output by the LSTM model as the expected result, and reversely solving the second final drift trajectory prediction result that satisfies the objective function and the second constraint condition.

[0046] Step S206: Determine whether the difference between the second actual drift trajectory result and the second final drift trajectory prediction result is less than a second preset threshold.

[0047] If yes, execute step S207; if no, return to execute step S201.

[0048] Step S207: Determine whether the trained LSTM model is available.

[0049] Specifically, after the LSTM model is trained, it is determined whether the LSTM model is available. The determination process is as follows: the historical motion data of the historical time period (i.e., the second motion information) is input into the trained LSTM model to predict the drift trajectory result of the historical time period (i.e., the second preliminary drift trajectory prediction result); a reverse planning model corresponding to the training phase (i.e., the second reverse planning model) is constructed, including an objective function and a second constraint condition. The objective function is the same as the objective function of the aforementioned embodiment, which is to minimize the difference between the drift trajectory prediction result and the expected result. The second constraint condition is to minimize the motion parameters corresponding to the marine motion target in the historical time period and the The second preliminary drift trajectory prediction result output by the LSTM model is taken as the expected result, and the second reverse planning model is reversely solved to obtain the second final drift trajectory prediction result that satisfies the objective function and the second constraint conditions; it is determined whether the difference between the actual drift trajectory result of the historical time period and the second final drift trajectory prediction result is less than a second preset threshold value, and if so, it is determined that the trained LSTM model is available, and steps S101-S103 can be executed; otherwise, the trained LSTM model is unavailable, and the step S201 is returned to be executed to obtain new historical motion data and trajectory, and continue to train the LSTM model.

[0050] In some embodiments, the first motion information and the second motion information include at least one of the following motion parameters: geographic location data, meteorological environment data, and heading data of the moving target at sea; then the step S202 also includes: dynamically adjusting the prediction time period of the LSTM model according to the heading data.

[0051] Specifically, in the process of constructing the data set and inputting it into the LSTM model for training, the prediction period of the LSTM model is dynamically adjusted according to key information such as the heading data of the moving targets at sea.

[0052] In some embodiments, the heading data includes at least one of the following: heading, speed; then the dynamic adjustment of the prediction time period of the LSTM model according to the heading data includes: when the heading change amplitude or the speed is less than the corresponding first preset threshold, adjusting the prediction time period of the LSTM model to a first time magnitude; when the heading change amplitude or the speed is greater than the corresponding first preset threshold, adjusting the prediction time period of the LSTM model to a second time magnitude, and the first time magnitude is greater than the second time magnitude.

[0053] In some embodiments, when the heading change amplitude or the speed is less than the corresponding first preset threshold, the prediction time period of the LSTM model is adjusted to the first time level, including: when the heading change amplitude or the speed is less than the corresponding preset value, the prediction time period of the LSTM model is gradually adjusted to the first time level by a rolling prediction method.

[0054] Specifically, when the speed of the target moving at sea is slow or the course change is small, a longer prediction period can be set, such as 3 to 5 days; conversely, when the speed of the target moving at sea is fast or the course change is large, the prediction period needs to be appropriately shortened, such as a few hours to 1 day, to improve the accuracy of the prediction results. When the prediction period is long, the rolling prediction method can be used to gradually update the prediction results to adapt to the changes in the target drift trajectory.

[0055] On the basis of the foregoing embodiments, the historical motion data and trajectory of the moving targets at sea are deeply learned by using the LSTM model to obtain a trained LSTM model for subsequent applications; further, the prediction results output by the LSTM model are compared with the actual trajectory through a reverse planning method to evaluate the error. Only when the error is small, the LSTM model is confirmed to be available, thereby further improving the accuracy of the prediction model.

[0056] Figure 3 A flow chart of another method for predicting the drift trajectory of a moving target at sea provided by an embodiment of the present invention, taking a ship at sea as an example, combined with Figure 3 The embodiment of the present invention is described in detail and can be divided into a training phase and an application phase.

[0057] First, the training phase proceeds as follows:

[0058] (1) Collect historical movement data and actual historical trajectory data: Use Beidou equipment to obtain the longitude and latitude and timestamp of the ship in each historical time period. Use meteorological instruments to obtain the meteorological environment of the ship in each historical time period, including important environmental factors such as wind direction, wind speed, ocean currents, and waves. Use speed and direction measuring instruments to obtain the ship's heading and speed in each historical time period.

[0059] (2) Preprocessing of historical motion data: After obtaining the historical motion data, these historical motion data are preprocessed, such as data cleaning to eliminate outliers and duplicate records, data calibration to ensure data accuracy and consistency, and data format conversion to unify the formats of various motion data to facilitate subsequent model training.

[0060] (3) Training the LSTM model: Construct a dataset of historical motion data and actual historical trajectories, and use the dataset to train the LSTM model. During the training process, dynamically adjust the prediction period of the LSTM model according to the speed or heading in the historical motion data, and finally obtain a trained LSTM model.

[0061] (4) Reverse planning optimization in the training phase: Construct a reverse planning model in the training phase, including the objective function and the constraint set of the training phase. The constraint set constrains the motion parameters and environmental parameters of the historical time period. Input the historical motion data into the trained LSTM model, predict the preliminary drift trajectory prediction result corresponding to the historical time period, and use it as the expected result of the reverse planning model in the training phase. Inversely solve the final drift trajectory prediction result that satisfies the objective function and the constraint set of the training phase.

[0062] (5) Compare the actual historical trajectory data and determine whether the difference is less than the threshold: Compare the final drift trajectory prediction result with the actual historical trajectory data to obtain the difference, and determine whether the difference is less than the threshold. If so, the LSTM model is determined to be available, that is, execute (6). Otherwise, continue to train the LSTM model, that is, return to execute (1).

[0063] (6) The trained LSTM model is available.

[0064] The application phase process is as follows:

[0065] (7) Collect the current movement data of the ship: obtain the latitude and longitude and timestamp of the ship in the current time period through Beidou equipment, obtain the meteorological environment of the ship in the current time period including important environmental factors such as wind direction, wind speed, ocean currents, waves, etc. through meteorological instruments, and obtain the heading and speed of the ship in the current time period through speed and direction measuring instruments.

[0066] (8) Current motion data preprocessing: data cleaning, data verification and data format conversion of the collected current motion data.

[0067] (9) Input the pre-processed current motion data into the trained LSTM model to obtain the preliminary drift trajectory prediction result: Input the pre-processed current motion data into the trained LSTM model to predict the preliminary drift trajectory prediction result of the prediction time period.

[0068] (10) Reverse planning optimization in the application phase: Construct a reverse planning model in the application phase, including the objective function and the constraint set of the application phase, which is a constraint on the motion parameters and environmental parameters of the current time period; take the preliminary drift trajectory prediction result predicted by the LSTM model as the expected result, and reversely solve the final drift trajectory prediction result that satisfies the objective function and the constraint set of the application phase.

[0069] (11) Calculate the accuracy of the prediction model: obtain the actual drift trajectory results of the ship in the prediction time period, and determine the prediction accuracy of the LSTM model and the reverse planning model in the application stage based on the difference between the actual drift trajectory results and the final drift trajectory prediction results output by the reverse planning model.

[0070] In summary, this embodiment is a drift trajectory prediction method for marine moving targets based on the combination of LSTM model and reverse planning method. The LSTM model is used to perform deep learning on various historical motion data of marine moving targets, extract the motion characteristics of the targets, and obtain preliminary drift trajectory prediction results; then, combined with the reverse planning method, according to the needs and constraints of the prediction targets, the preliminary drift trajectory prediction results output by the LSTM model are optimized and adjusted to obtain the final drift trajectory prediction results. In the whole process, various influencing factors are comprehensively utilized, and the LSTM model is selected to be suitable for processing sequence data with long-term dependencies, and the reverse planning algorithm is used to optimize the processing results of the LSTM model, so the prediction results are more accurate.

[0071] Figure 4 A schematic diagram of a device for predicting drift trajectory of a moving target at sea provided by an embodiment of the present invention is shown in FIG. Figure 4 As shown, the drift trajectory prediction device for a moving target at sea comprises:

[0072] The information acquisition module 401 is used to acquire the first motion information of the marine motion target in the current time period;

[0073] A preliminary prediction module 402 is used to input the first motion information into the trained LSTM model and output a first preliminary drift trajectory prediction result of the marine moving target in a prediction time period;

[0074] The final prediction module 403 is used to use the first preliminary drift trajectory prediction result as the expected result of the first reverse planning model, reversely solve the first reverse planning model, and obtain the first final drift trajectory prediction result of the marine moving target in the prediction time period.

[0075] In some embodiments, the final prediction module 403 is specifically used to:

[0076] Constructing a first reverse planning model including an objective function and a first constraint condition, wherein the objective function is to minimize the difference between a drift trajectory prediction result and an expected result, and the first constraint condition is a first constraint on a motion parameter in the first motion information;

[0077] The first preliminary drift trajectory prediction result output by the LSTM model is used as the expected result, and a first final drift trajectory prediction result that satisfies the objective function and the first constraint condition is reversely solved.

[0078] In some embodiments, the apparatus further includes an accuracy assessment module 404, wherein the accuracy assessment module 404 is configured to:

[0079] Obtaining a first actual drift trajectory result of the marine moving target in a predicted time period;

[0080] The prediction accuracy of the trained LSTM model and the first reverse planning model is determined according to the difference between the first actual drift trajectory result and the first final drift trajectory prediction result.

[0081] In some embodiments, the apparatus further includes a model training module 405, wherein the model training module 405 is configured to:

[0082] Acquire second motion information of the marine moving target in a historical time period and a corresponding second actual drift trajectory result, wherein the second motion information and the second actual drift trajectory result constitute a data set;

[0083] The LSTM model is trained based on the data set to obtain the trained LSTM model.

[0084] In some embodiments, the first motion information and the second motion information include at least one of the following motion parameters: geographic location data, meteorological environment data, and heading data of the marine motion target;

[0085] The model training module 405 is also used for:

[0086] The prediction time period of the LSTM model is dynamically adjusted according to the heading data.

[0087] In some embodiments, the heading data includes at least one of the following: heading and speed; the model training module 405 is specifically used to:

[0088] When the heading change amplitude or the speed is less than the corresponding first preset threshold, adjusting the prediction time period of the LSTM model to the first time magnitude;

[0089] When the heading change amplitude or the speed is greater than the corresponding first preset threshold, the prediction time period of the LSTM model is adjusted to a second time magnitude, and the first time magnitude is greater than the second time magnitude.

[0090] In some embodiments, the model training module 405 is further used to:

[0091] When the heading change amplitude or the speed is less than the corresponding first preset threshold, a rolling prediction method is used to gradually adjust the prediction time period of the LSTM model to reach the first time level.

[0092] In some embodiments, the model training module 405 is further used to:

[0093] Inputting the second motion information into the trained LSTM model, and outputting a second preliminary drift trajectory prediction result of the marine moving target in the historical time period;

[0094] Constructing a second reverse planning model including an objective function and a second constraint condition, wherein the objective function is to minimize the difference between the drift trajectory prediction result and the expected result, and the second constraint condition is a second constraint on the motion parameters in the second motion information;

[0095] Taking the second preliminary drift trajectory prediction result output by the LSTM model as the expected result, and reversely solving the second final drift trajectory prediction result that satisfies the objective function and the second constraint condition;

[0096] When it is determined that the difference between the second actual drift trajectory result and the second final drift trajectory prediction result is less than a second preset threshold, it is determined that the trained LSTM model is available.

[0097] Technicians in the relevant field can clearly understand that, for the convenience and brevity of description, the specific working process and corresponding beneficial effects of the marine moving target drift trajectory prediction device described above can refer to the corresponding process in the aforementioned method example and will not be repeated here.

[0098] like Figure 5 As shown, an embodiment of the present invention provides an electronic device, including a processor 501, a communication interface 502, a memory 503 and a communication bus 504, wherein the processor 501, the communication interface 502, and the memory 503 communicate with each other through the communication bus 504.

[0099] Memory 503, used for storing computer programs;

[0100] In one embodiment of the present invention, the processor 501 is used to implement the steps of the method for predicting the drift trajectory of a moving target at sea provided by any one of the aforementioned method embodiments when executing the program stored in the memory 503.

[0101] The implementation principle and technical effect of the electronic device provided by the embodiment of the present invention are similar to those of the above embodiment and will not be described in detail here.

[0102] The above-mentioned memory 503 can be an electronic memory such as a flash memory, an EEPROM (electrically erasable programmable read-only memory), an EPROM, a hard disk or a ROM. The memory 503 has a storage space for program codes for executing any method steps in the above-mentioned method. For example, the storage space for program codes may include various program codes for implementing various steps in the above method respectively. These program codes can be read from or written into one or more computer program products. These computer program products include program code carriers such as hard disks, compact disks (CDs), memory cards or floppy disks. Such computer program products are usually portable or fixed storage units. The storage unit may have a storage segment or storage space arranged similarly to the memory 503 in the above-mentioned electronic device. The program code can be compressed, for example, in an appropriate form. Generally, the storage unit includes a program for executing the method steps according to an embodiment of the present invention, that is, a code that can be read by a processor such as 501, which, when run by an electronic device, causes the electronic device to execute various steps in the method described above.

[0103] The embodiment of the present invention further provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method for predicting the drift trajectory of a moving target at sea are implemented.

[0104] The computer-readable storage medium may be included in the device / apparatus described in the above embodiment; or it may exist independently without being assembled into the device / apparatus. The above computer-readable storage medium carries one or more programs, and when the above one or more programs are executed, the method according to the embodiment of the present invention is implemented.

[0105] According to an embodiment of the present invention, the computer-readable storage medium may be a non-volatile computer-readable storage medium, for example, may include but is not limited to: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, the computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, apparatus, or device.

[0106] It should be noted that, in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0107] The above are preferred embodiments of the present invention. It should be pointed out that, for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A method for predicting drift trajectory of a moving target at sea, characterized in that: include: Obtain the first motion information of the marine moving target in the current time period; Inputting the first motion information into the trained LSTM model, and outputting a first preliminary drift trajectory prediction result of the marine moving target in the prediction time period; The first preliminary drift trajectory prediction result is used as the expected result of the first reverse planning model, and the first reverse planning model is reversely solved to obtain the first final drift trajectory prediction result of the marine moving target in the prediction time period.

2. The method according to claim 1, characterized in that The method of taking the first preliminary drift trajectory prediction result as the expected result of the first reverse planning model, performing reverse solving on the first reverse planning model, and obtaining the first final drift trajectory prediction result of the marine moving target in the prediction time period includes: Constructing a first reverse planning model including an objective function and a first constraint condition, wherein the objective function is to minimize the difference between a drift trajectory prediction result and an expected result, and the first constraint condition is a first constraint on a motion parameter in the first motion information; The first preliminary drift trajectory prediction result output by the LSTM model is used as the expected result, and a first final drift trajectory prediction result that satisfies the objective function and the first constraint condition is reversely solved.

3. The method according to claim 2, characterized in that The method further comprises: Obtaining a first actual drift trajectory result of the marine moving target in a predicted time period; The prediction accuracy of the trained LSTM model and the first reverse planning model is determined according to the difference between the first actual drift trajectory result and the first final drift trajectory prediction result.

4. The method according to any one of claims 1 to 3, characterized in that: Before inputting the first motion information into the trained LSTM model, the method further includes: Acquire second motion information of the marine moving target in a historical time period and a corresponding second actual drift trajectory result, wherein the second motion information and the second actual drift trajectory result constitute a data set; The LSTM model is trained based on the data set to obtain the trained LSTM model.

5. The method according to claim 4, characterized in that The first motion information and the second motion information include at least one of the following motion parameters: geographic location data, meteorological environment data, and heading data of the marine motion target; The training of the LSTM model based on the data set further includes: The prediction time period of the LSTM model is dynamically adjusted according to the heading data.

6. The method according to claim 5, characterized in that The heading data includes at least one of the following: heading and speed; then the dynamically adjusting the prediction time period of the LSTM model according to the heading data includes: When the heading change amplitude or the speed is less than the corresponding first preset threshold, adjusting the prediction time period of the LSTM model to the first time magnitude; When the heading change amplitude or the speed is greater than the corresponding first preset threshold, the prediction time period of the LSTM model is adjusted to a second time magnitude, and the first time magnitude is greater than the second time magnitude.

7. The method according to claim 6, characterized in that When the heading change amplitude or the speed is less than the corresponding first preset threshold, adjusting the prediction time period of the LSTM model to the first time magnitude includes: When the heading change amplitude or the speed is less than the corresponding first preset threshold, the prediction time period of the LSTM model is gradually adjusted to the first time level by using a rolling prediction method.

8. The method according to claim 4, characterized in that After obtaining the trained LSTM model, the method further includes: Inputting the second motion information into the trained LSTM model, and outputting a second preliminary drift trajectory prediction result of the marine moving target in the historical time period; Constructing a second reverse planning model including an objective function and a second constraint condition, wherein the objective function is to minimize the difference between the drift trajectory prediction result and the expected result, and the second constraint condition is a second constraint on the motion parameters in the second motion information; Taking the second preliminary drift trajectory prediction result as the desired result, reversely solving the second final drift trajectory prediction result that satisfies the objective function and the second constraint condition; When it is determined that the difference between the second actual drift trajectory result and the second final drift trajectory prediction result is less than a second preset threshold, it is determined that the trained LSTM model is available.

9. A device for predicting drift trajectory of a moving target at sea, characterized in that: include: An information acquisition module, used to acquire first motion information of a marine moving target in a current time period; A preliminary prediction module, used for inputting the first motion information into the trained LSTM model, and outputting a first preliminary drift trajectory prediction result of the marine motion target in a prediction time period; The final prediction module is used to use the first preliminary drift trajectory prediction result as the expected result of the first reverse planning model, reversely solve the first reverse planning model, and obtain the first final drift trajectory prediction result of the marine moving target in the prediction time period.

10. An electronic device, characterized in that: It includes a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus; Memory, used to store computer programs; The processor is used to implement the steps of the method for predicting the drift trajectory of a moving target at sea as described in any one of claims 1 to 8 when executing the program stored in the memory.

Images