Data processing method and device, equipment and medium

Abstract

The invention discloses a data processing method and device, equipment and a medium, and generally relates to the technical field of data processing. The method comprises the steps that under the condition that ship data are received, first data are acquired according to the ship data, the ship data comprise real-time data of multiple ships, and the first data comprise the ship data and feature data; the feature data are other parameter data corresponding to ship features of the ship data and different from the parameter information of the ship data; the first data are input into a target model, the first data are processed, target data corresponding to the first data are obtained, a target ship corresponding to the target data is determined, the target model comprises at least three target sub-models, and the at least three first sub-models are sub-models in the first model; the target data is data corresponding to a target action, and the target ship is a ship executing the target action.

Description

Technical Field

[0001] The present disclosure generally relates to the field of data processing technology, and more particularly to a data processing method, apparatus, device and medium. Background Art

[0002] As global marine resources are gradually depleted, illegal marine activities pose a serious threat to marine ecosystems and legal fisheries. Marine departments of various countries and international organizations are actively looking for technical means to monitor and combat illegal marine activities.

[0003] The Automatic Identification System (AIS) is a global marine ship monitoring system that can collect various parameters of marine ships in real time, such as ship position, ship speed, ship heading, etc. Therefore, the system is widely used in the safety management of marine ships.

[0004] However, due to the huge gap between the basic nature of the data collected by AIS and the richness of illegal maritime activities, there are still some monitoring loopholes in the process of managing illegal maritime activities. Summary of the invention

[0005] In view of the above-mentioned defects or deficiencies in the relevant technology, it is expected to provide a data processing method, device, equipment and medium that can solve the huge gap between the basic nature of the data collected by the AIS and the richness of illegal marine activities. There are still some monitoring loopholes in the process of managing illegal marine activities, and greatly improve the accuracy and timeliness of ship marine activity monitoring.

[0006] In a first aspect, a data processing method is provided, the method comprising: In the case of receiving ship data, first data is acquired according to the ship data, the ship data including real-time data of a plurality of ships, the first data including the ship data and the characteristic data, the characteristic data being other parameter data corresponding to the ship characteristics of the ship data and different from the parameter information of the ship data; The first data is input into a target model, the first data is processed, target data corresponding to the first data is obtained, and a target ship corresponding to the target data is determined, the target model includes at least three target sub-models, the at least three first sub-models are sub-models in the first model, the target data is data corresponding to a target action, and the target ship is a ship that performs a target action.

[0007] In the present application, when ship data reflecting real-time data of multiple ships is received, first data is obtained according to the ship data (the first data includes ship data and feature data, wherein the feature data is other parameter data corresponding to the ship features of the ship data and different from the parameter information of the ship data); then, the first data is input into the target model, the first data is processed, the target data corresponding to the first data is obtained, and the target ship corresponding to the target data is determined (the target model includes at least three target sub-models, the at least three first sub-models are sub-models in the first model, the target data is the data corresponding to the target action, and the target ship is the ship that performs the target action). In this way, by processing and judging the ship data of multiple ships at the same time through multiple models, the accuracy of identifying illegal actions of ships on the ocean can be greatly improved, and the generalization of judgment in the process of judging illegal actions of complex marine ship targets can be greatly improved.

[0008] In a second aspect, a data processing system is provided, the system comprising: a data acquisition module, configured to acquire first data according to the ship data when receiving the ship data, wherein the ship data includes real-time data of a plurality of ships, and the first data includes the ship data and the characteristic data, wherein the characteristic data is other parameter data corresponding to the ship characteristics of the ship data and different from the parameter information of the ship data; a data calculation module, used for inputting the first data into a target model, processing the first data, obtaining target data corresponding to the first data, and determining a target ship corresponding to the target data, wherein the target model includes at least three target sub-models, the at least three first sub-models are sub-models in the first model, the target data are data corresponding to a target action, and the target ship is a ship that performs a target action.

[0009] In a third aspect, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method described in the first aspect is implemented.

[0010] In a fourth aspect, a computer-readable storage medium is provided, on which a computer program is stored, characterized in that when the program is executed by a processor, the method described in the first aspect is implemented.

[0011] In a fifth aspect, a computer program product is provided. The computer program product includes instructions, and the instructions are executed by a processor to implement the method described in the first aspect.

[0012] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Other features, objects and advantages of the present application will become more apparent by reading the detailed description of non-limiting embodiments made with reference to the following drawings: Figure 1 One of the flow charts of the data processing method provided in the embodiment of the present application; Figure 2 The second flowchart of the data processing method provided in the embodiment of the present application; Figure 3 A schematic diagram of the structure of a data processing device provided in an embodiment of the present application; Figure 4 A schematic diagram of the structure of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0014] The present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the relevant invention, rather than to limit the invention. It is also necessary to explain that, for ease of description, only the parts related to the invention are shown in the accompanying drawings.

[0015] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0016] As global fishery resources are gradually depleted, illegal fishing has become a serious threat to the marine ecosystem and the legal fishing industry. Fisheries departments of various countries and international organizations are actively looking for technical means to monitor and combat illegal fishing.

[0017] The Automatic Identification System (AIS) is a global marine ship monitoring system that can collect various parameters of marine ships in real time, such as ship position, ship speed, ship heading, etc. Therefore, the system is widely used in the safety management of marine ships.

[0018] However, due to the gap between the basic nature of the data collected by the AIS and the richness of illegal fishing activities, there are still some monitoring loopholes in the process of managing illegal fishing activities. These monitoring loopholes specifically include: 1) Illegal fishing activities are often rich in action forms, highly flexible and highly concealed, and the basic parameters of the ship are actually difficult to accurately identify illegal fishing activities; 2) The activities of ships at sea are rich and complex, and the basic parameters of the ship combined with traditional algorithms are actually difficult to accurately identify illegal fishing activities, which can easily lead to false alarms and missed reports. The simple thresholds set by the rules may not be suitable for different sea areas, seasons or specific fishing conditions, resulting in legal fishing activities being misjudged or illegal fishing activities being missed.

[0019] Based on the above problems, related technologies have developed machine learning models to identify illegal fishing behaviors. However, 1) the machine learning models in related technologies (such as decision trees and support vector machines) are often relatively simple, while the behaviors of illegal fishing vessels are often complex and changeable. Therefore, the generalization ability of the single machine learning model is limited and it is easily affected by data noise, and it cannot provide stable and efficient identification of illegal fishing behaviors in a variety of different scenarios; 2) The system based on rules or a single model lacks continuous optimization capabilities and fails to dynamically update the model or rules with newly labeled data, resulting in the inability of the system performance to continuously improve with the arrival of new data and to adapt to changes in illegal fishing behaviors.

[0020] Based on this, the present application proposes a data processing method, device, equipment and medium, which can solve the huge gap between the basic nature of the data collected by the AIS and the richness of illegal marine activities. There are still some monitoring loopholes in the process of managing illegal marine activities, which greatly improves the accuracy and timeliness of ship marine activity monitoring.

[0021] Figure 1 FIG. 1 is a flow chart of a data processing method provided by an embodiment of the present application. The execution subject of the method may be a data processing system. Figure 1 As shown, the method includes the following steps 301 and 302: Step 301: when receiving ship data, obtaining first data according to the ship data.

[0022] In an embodiment of the present application, the above-mentioned ship data includes real-time data of multiple ships.

[0023] It can be understood that in the embodiment of the present application, the above-mentioned ship data is direct parameter data of a ship in a communicative state on the ocean acquired in real time.

[0024] In one example, the multiple ships are ships connected to an Automatic Identification System (AIS).

[0025] Exemplarily, the above-mentioned ship data includes: ship dynamic data and ship static data.

[0026] In the embodiment of the present application, the above-mentioned ship data is the acquired original ship data. Further, the original ship data refers to the data that has not been processed, and is directly transmitted by the ship through communication, or the data processing system directly acquires from the ship.

[0027] In one example, the above dynamic data may include: ship speed data, ship heading data, ship position data, and ship timestamp data.

[0028] In one example, the static data may include: ship type data, ship identification number data.

[0029] In an embodiment of the present application, the first data includes ship data and the characteristic data, and the characteristic data is other parameter data corresponding to the ship characteristics of the ship data and different from the parameter information of the ship data.

[0030] It can be understood from the foregoing that, since the ship data is raw data, it is not sufficient to fully characterize all behaviors of the ship. Furthermore, other data of the ship can be indirectly obtained based on the raw data of the ship directly received.

[0031] For example, based on the unit time of a ship and the distance within the unit time, the speed of the ship in the unit time can be indirectly obtained, while in the directly obtained ship data, the speed obtained is the real-time speed of the ship at every moment, rather than the speed within the unit time. For another example, based on the distance and direction of a ship in the unit time, the angle of the ship's navigation segment can be determined, and the destination area and other information of the ship's subsequent navigation process can be determined based on the segment angle within the unit time. In the directly obtained ship data, the speed obtained is the real-time navigation angle of the ship at every moment, rather than the segment angle within the unit time.

[0032] Based on this, the other parameters of the ship indirectly obtained are the above-mentioned characteristic data. This characteristic data is different from the ship data and is associated with the ship data. It is derived from the ship data and is necessary data for determining the ship's behavior. That is, this characteristic data is associated with and corresponds to the ship data, but is also different.

[0033] Optionally, in an embodiment of the present application, a specific method for obtaining the first data is as follows. Exemplarily, in the case of receiving the ship data, obtaining the first data according to the ship data includes: determining the characteristic parameters corresponding to the predetermined action characteristic information according to the ship data and the predetermined action characteristic information; obtaining the characteristic data based on the ship data and the characteristic parameters; and fusing the characteristic data and the ship data to generate the first data.

[0034] Exemplarily, the above-mentioned predetermined action characteristic information may be preset or user-defined, which is not limited in the embodiments of the present application.

[0035] It can be understood that the above-mentioned predetermined action characteristic information is related to the current scene requirement. For example, assuming that the current scene requirement is that there are many ships sailing in a certain water area, and the speed of ships sailing in the water area needs to be limited, then the above-mentioned predetermined action characteristic information is the sailing speed of the ship in the water area, and the characteristic parameter corresponding to the predetermined action characteristic information can also be the sailing speed of the ship in the water area. Then, it can be based on the characteristic parameter corresponding to the predetermined action characteristic information.

[0036] It should be noted that the predetermined action characteristic information may be a single information or multiple information. In the case where the predetermined action characteristic information is a single information, the characteristic parameter corresponding to the predetermined action characteristic information may include one parameter or multiple parameters. Accordingly, in the case where the predetermined action characteristic information is multiple information, the characteristic parameter corresponding to the predetermined action characteristic information may include multiple parameters, and the number of the multiple parameters may be the same as the number of the multiple information, or may be greater than the number of the multiple information.

[0037] Exemplarily, based on the above characteristic parameters, the ship data can be processed, and then the characteristic parameters corresponding to the characteristic information are obtained. For example, if the characteristic parameter is the navigation speed of the ship within the water area, the navigation time and navigation distance within the above water area in the ship data are collected, and the navigation distance and navigation time are compared, and finally the specific value of the navigation speed of the ship within the water area is obtained as characteristic data.

[0038] Exemplarily, the above-mentioned feature data may include dynamic feature data, spatial feature data and temporal feature data.

[0039] In one example, the dynamic feature data may include: the ship's speed, heading, acceleration, steering angle, etc. In the case where it is necessary to detect whether a ship is fishing in an illegal area, the dynamic feature data may be used to reflect the behavior pattern of the ship.

[0040] In one example, the spatial feature data may include: ship location information, trajectory curvature, ship turning frequency, etc. When it is necessary to detect whether a ship is fishing in an illegal area, the spatial feature data may be used to analyze whether the ship's motion trajectory meets the characteristics of fishing operations.

[0041] In one example, the time characteristic data may include: the ship's stay time, sailing time, etc. When it is necessary to detect whether the ship is fishing in an illegal area, the time characteristic data can be used to determine whether the ship stays in a specific area for a long time, and then infer whether it is fishing.

[0042] For example, after acquiring the characteristic data and the ship data, these data need to be packaged and integrated to form complete data information reflecting the current navigation status of the ship, and the data information is used as the first data.

[0043] like Figure 2 As shown, the above specific methods of obtaining the first data all belong to the steps of data preprocessing.

[0044] Step 302: input the first data into a target model, process the first data, obtain target data corresponding to the first data, and determine a target ship corresponding to the target data.

[0045] In an embodiment of the present application, the target model includes at least three target sub-models, the at least three first sub-models are sub-models in the first model, the target data are data corresponding to the target action, and the target ship is a ship that performs the target action.

[0046] Exemplarily, first data reflecting the navigation status of a ship is input into a target model, and the target model processes the first data to obtain target data reflecting various action information of the first data, and determines a target ship corresponding to the target data based on the target data.

[0047] In an embodiment of the present application, at least three target sub-models in the above-mentioned target model, each target sub-model can be used to process the above-mentioned first data. At the same time, the three target sub-models are different, that is, they process the first data in different ways.

[0048] Furthermore, after at least three target sub-models process the first data, at least three processing results are obtained, and the three processing results are fused to obtain target data corresponding to the first data.

[0049] It is understandable that in the related technologies before the embodiment of the present application, the ship data is often analyzed by a single data processing model. However, the results obtained by such analysis are prone to inaccuracy. In the embodiment of the present application, the first data is analyzed by at least three processing models. Since each target sub-model has differences in scene usage characteristics and calculation accuracy, the first data is processed by at least three target sub-models at the same time. After the result data is obtained, the result data is fused to obtain the final target data.

[0050] In the embodiment of the present application, the above target action can be matched with the above predetermined action characteristic information. It can be understood that, assuming that the target action is illegal fishing in waters, the target action corresponds to multiple predetermined action characteristic information, such as illegal purse seine, illegal dragnet, illegal area navigation, etc. Only when at least one action characteristic information among the multiple predetermined action characteristics is met, can the target action made by the current ship be explained, and the ship also belongs to the target ship.

[0051] Illustratively, all or some of the multiple ships mentioned above may be target ships, which is not limited in the embodiments of the present application.

[0052] Exemplarily, the process of obtaining the target data corresponding to the first data can be: taking the result data obtained from at least three target sub-models as the basis; or taking the average value of the result data obtained from at least three target sub-models as the basis; or taking the weighted average of the result data obtained from at least three target sub-models as the basis, which is not limited to this in the embodiments of the present application.

[0053] Exemplarily, the first model includes all models for processing the first data.

[0054] In one example, the number of first sub-models in the first model may be greater than the number of target sub-models in the target model, or may be equal to the number of target sub-models in the target model, which is not limited in the embodiments of the present application.

[0055] Furthermore, the process of selecting the first sub-model from the first model as the target sub-model and the process of calculating the target data through at least three target sub-models are described in detail in the subsequent description and are not repeated here.

[0056] In the method provided in the embodiment of the present application, when ship data reflecting real-time data of multiple ships is received, first data is obtained according to the ship data (the first data includes ship data and feature data, wherein the feature data is other parameter data corresponding to the ship features of the ship data and different from the parameter information of the ship data); then, the first data is input into the target model, the first data is processed, the target data corresponding to the first data is obtained, and the target ship corresponding to the target data is determined (the target model includes at least three target sub-models, the above-mentioned at least three first sub-models are sub-models in the first model, the target data is the data corresponding to the target action, and the target ship is the ship that performs the target action). In this way, by processing and judging the ship data of multiple ships at the same time through multiple models, the accuracy of identifying illegal actions of ships on the ocean can be greatly improved, and the generalization of judgment in the process of judging illegal actions of complex marine ship targets can be greatly improved.

[0057] In another embodiment of the present application, a specific implementation method is provided for confirming the target data and the target ship corresponding to the target data by setting the calculation weight of the target sub-model in advance after processing the first data. Exemplarily, the specific implementation of "processing the first data, obtaining the target data corresponding to the first data, and determining the target ship corresponding to the target data" mentioned above includes: processing the first data through each of the target sub-models, obtaining the target sub-data corresponding to each of the target sub-models; processing the target sub-data corresponding to each of the target sub-models according to the calculation weight of each of the target sub-models.

[0058] Exemplarily, the calculation weight of each of the above-mentioned target sub-models is a weight determined according to the data processing accuracy of each of the above-mentioned target sub-models.

[0059] Exemplarily, the above calculation weights may be preset or customized by the user in advance, and this application does not limit this.

[0060] It can be understood that after determining at least three target sub-models in the target model, the above-mentioned first data can be processed for at least three groups of target sub-models, and then the target sub-data can be obtained. Then, the target sub-data are processed according to the calculation weights of each target sub-model preset in advance.

[0061] Exemplarily, each set of target sub-data includes data results of each ship among the above-mentioned multiple ships, and the data results are used to determine whether the action of each ship corresponds to the target action.

[0062] Exemplarily, the target model includes at least three target sub-models, wherein each sub-model can be used to process data of a ship on waters. Three different target sub-models are introduced below, and all of these three target sub-models are models for analyzing fishing action behaviors on waters.

[0063] In one example, the target sub-model may be a support vector machine (SVM) model. A support vector machine is a machine learning model based on a maximum margin classifier, which is applicable to binary classification problems. It can divide data into different categories by finding an optimal segmentation hyperplane. In an embodiment of the present application, SVM is used to determine whether a ship is fishing based on the characteristic data of the ship's trajectory.

[0064] Furthermore, the training steps of the SVM model are as follows: 1) Construct feature space: Input the dynamic, spatial and temporal features of the ship into the SVM model as the input vector of the training sample. 2) Define the objective function: By constructing the objective function, find the hyperplane that can maximize the classification interval so as to effectively distinguish fishing behavior from non-fishing behavior. 3) Model optimization: Use kernel functions (such as Gaussian kernel, linear kernel or polynomial kernel) to map nonlinear features to improve the accuracy of classification. 4) Model evaluation: Use cross-validation to evaluate the SVM model, and optimize the model performance by adjusting the penalty parameters and kernel function parameters.

[0065] In one example, the target sub-model may be a random forest model. Random forest is a model based on ensemble learning, which generates multiple decision trees and combines their output results to improve classification accuracy and noise resistance. Random forest has strong robustness when processing high-dimensional feature spaces.

[0066] Furthermore, the training steps of the random forest model are as follows: 1) Construct a training sample set: randomly select multiple sub-sample sets from the labeled fishing behavior dataset, and each sub-sample set is used to train a decision tree; 2) Construct a decision tree: for each decision tree, divide the data according to the characteristics in the sample, and gradually generate a tree structure for classification; 3) Integrated output: when new ship trajectory data is input, each decision tree in the random forest will make an independent prediction, and finally the final classification result is determined by a voting mechanism; 4) Model evaluation and optimization: by adjusting parameters such as the number of trees and the depth of the trees, the model performance is further optimized to ensure that it can accurately identify complex fishing behavior patterns.

[0067] In one example, the target sub-model may be a neural network model. A neural network model is a model that can simulate the connection of brain neurons and is particularly suitable for processing nonlinear and high-dimensional data. The embodiment of the present application uses a multi-layer feedforward neural network to capture complex patterns in ship behavior data.

[0068] Furthermore, the training steps of the neural network model are as follows: 1) Network architecture design: Construct a multi-layer neural network, including an input layer, several hidden layers and an output layer. The input layer is used to receive the trajectory characteristics of the ship, and the output layer is used to give the classification results; 2) Back propagation algorithm: Using the back propagation algorithm, by minimizing the loss function, the weights of the network are gradually updated, so that the model can gradually learn the relationship between ship behavior and fishing behavior; 3) Activation function selection: In order to increase the nonlinear representation ability of the network, activation functions such as ReLU, Sigmoid or Tanh are used in the hidden layer 4) Model evaluation and optimization: By adjusting the hyperparameters such as the number of network layers, the number of nodes and the learning rate, and using cross-validation to optimize the generalization ability of the model, ensure that good classification performance can be maintained on unseen data.

[0069] Exemplarily, when the above-mentioned first model only includes three first sub-models, these three first sub-models will all be selected as target sub-models and used as the final target model.

[0070] The following will describe the use process of the above target model through Example 1.

[0071] Example 1: Assume that the above data processing system is an AIS system, which includes three target sub-models, and the three target sub-models are SVM model, Random Forest model, and Neural Network model. The target action to be judged is illegal fishing action on the ocean, and accordingly, the target ship is also an illegal fishing ship that performs illegal fishing action on the ocean.

[0072] When the AIS system receives the first data, the three target sub-models predict the ship data respectively to obtain preliminary prediction results. Then, the system performs weighted average of the prediction results of the three models according to the pre-assigned calculation weights to calculate the final classification result.

[0073] Specifically, the process of determining the target data and the target ship according to the calculation weight and the target model is as follows: Step A1: Obtain the prediction results of three models: Use the support vector machine (SVM) model, random forest model and neural network model to predict the new data respectively, and output three preliminary prediction results, which are , and The prediction results are expressed as probability values, representing the likelihood that the vessel will be identified as fishing.

[0074] Step A2: Weighted average based on weights: According to the accuracy of each model on the training set , and The calculation weights of the three preset models are: , and , calculate the weighted average prediction score The weighted average formula is as follows:

[0075] like Figure 2 As shown, the above steps A1 to A2 are the process of inputting the first data for data preprocessing into the three models.

[0076] In another embodiment of the present application, a specific implementation method for setting the calculation weight of the target sub-model in advance is also provided. Exemplarily, before the above-mentioned "obtaining the target feature data in the above-mentioned ship data when receiving the ship data", the above-mentioned method includes: obtaining multiple groups of original ship data and multiple groups of target ship data; inputting the above-mentioned multiple groups of original ship data into each first sub-model in the above-mentioned first model, generating multiple groups of test arrays corresponding to the above-mentioned multiple groups of original ship data for each first sub-model; comparing the above-mentioned multiple groups of test data of each first sub-model with the above-mentioned multiple groups of target ship data, and determining the difference between the multiple groups of test data of each first sub-model and the above-mentioned multiple groups of target ship data; when the above-mentioned difference is greater than the preset first threshold value, adjusting each of the above-mentioned first sub-models according to the above-mentioned difference, so that the difference between the test data generated by each of the above-mentioned first sub-models and the target ship data is less than the preset first threshold value, and the first sub-model in the above-mentioned first model is ranked according to the difference corresponding to each of the above-mentioned first sub-models; and determining the calculation weight of each of the above-mentioned first sub-models according to the ranking of the above-mentioned first sub-models.

[0077] Exemplarily, the target ship data corresponds one-to-one to the original ship data.

[0078] Exemplarily, the above-mentioned preset first threshold may be preset in advance or may be user-defined.

[0079] Exemplarily, the above-mentioned preset first threshold can be used to indicate whether the training results of each first sub-model in the first model meet user requirements. When the difference between the above-mentioned test data and the target data is less than the preset first threshold, the training can be stopped to generate a qualified first sub-model model. Otherwise, it is necessary to input multiple groups of original arrays again for training until the difference between the test data and the target data is less than the preset threshold.

[0080] It is understandable that after the first sub-model adaptation model is constructed, it is necessary to train the first sub-model so that the first sub-model can accurately complete the above-mentioned process of processing the first data into target data.

[0081] Furthermore, multiple groups of original ship data for training and target ship data corresponding to the original arrays can be preset in advance, and then the multiple groups of original ship data are input into each first sub-model to generate multiple groups of test arrays.

[0082] It is understandable that during the training process, the difference between the generated test array and the target ship array corresponding to the original ship array that meets user requirements may be greater than the preset first threshold. In this case, it is necessary to input multiple groups of original ship arrays into the first sub-model again and continue training until the difference between the test data and the target ship data is less than the preset threshold.

[0083] Furthermore, the above difference is used to indicate the error value between the test data output by each first sub-model and the multiple groups of target ship data corresponding to the multiple groups of original propagation data after multiple groups of original ship data are inputted during the training process. This error value is used to characterize the degree of training of the current first sub-model. As described above, when the error value between the test data and the target ship data of the first sub-model is less than or equal to the first threshold value, it means that the first sub-model has been trained and the first sub-model can be used to perform the task of judging whether the first data contains the target data and then judge the target ship; when the error value between the test data and the target ship data of the first sub-model is greater than the first threshold value, it means that the first sub-model still needs to be trained and the first sub-model cannot be used to perform the task of judging whether the first data contains the target data and then judge the target ship.

[0084] After the training is completed, even if the difference between the test data and the target ship data generated by each first sub-model is less than the preset first threshold, the difference between the test data and the target ship data generated by different first sub-models is still different. The different first sub-models can be ranked from small to large according to the difference. The smaller the difference, the higher the accuracy, the higher the ranking, and the greater the corresponding calculation weight; the larger the difference, the lower the accuracy, the lower the ranking, and the smaller the corresponding calculation weight.

[0085] The following example 2 illustrates the process of confirming and calculating weights: Example 2: The performance of each first sub-model during the training process may be different. In order to fully utilize the advantages of each first sub-model, a corresponding weight may be assigned according to the accuracy of each first sub-model during the training process.

[0086] The principle of weight distribution is: the model with high accuracy has a larger weight; the model with low accuracy has a smaller weight.

[0087] The specific calculation formula is:

[0088] in, Indicates The weight of the model, For the The accuracy of the model on the training set.

[0089] It is understandable that, since the data processing system is constantly collecting new ship data, when new ship data appears and the ship corresponding to the new ship data is finally confirmed to belong to the target data, it is necessary to expand the training samples and add the new ship data and its target data to the original training samples, that is, to expand the training samples and train the first sub-model in the first model to update each first sub-model in the first model.

[0090] In one example, the data processing system can incrementally train the existing model based on the new training samples, update the model's computational weights and model parameters, and gradually adjust the model's adaptability to the new data. This update method can avoid completely retraining the model, saving computing resources while improving the model's generalization ability.

[0091] Furthermore, after incremental training is completed, the system will re-evaluate the performance of the model on the validation set to ensure that the model's accuracy, precision, recall and other indicators have not dropped significantly. If the evaluation results show that the model performance has improved, the system will automatically deploy the updated model; otherwise, continue to adjust the parameters during incremental training to ensure the optimization effect of the model.

[0092] In another example, the data processing system may choose to retrain the model over time. Retraining means training the model from scratch using a new dataset to ensure that the model captures the latest behavior patterns.

[0093] Further, such as Figure 2 As shown, the specific implementation steps of retraining are as follows: 1) Training database update: Combine newly collected labeled data with historical data to form a complete training sample database containing the latest behavior characteristics; 2) Full training: Each first sub-model is fully retrained to optimize the parameters and structure of the model so that it can adapt to the target action characteristics of the new period; 3) Model evaluation and deployment: Through cross-validation and model evaluation, ensure that the retrained model achieves better performance indicators on the validation set. After the evaluation is completed, the system will automatically replace the old model and deploy the new model for real-time prediction.

[0094] It is understandable that the frequency and period of the first sub-model update can be dynamically adjusted according to the changes in the target action. Generally, the data processing system will periodically evaluate the first sub-model and trigger incremental training or retraining when necessary. In addition, if the data processing system detects that the recognition accuracy rate has dropped or a large amount of new data is marked as the target action, the data processing system will also speed up the model update frequency to ensure the accuracy and timeliness of the prediction. In another embodiment of the present application, a specific implementation method for performing noise reduction and completion processing on the ship data in advance is also provided. Exemplarily, before the "obtaining first data according to the above ship data" mentioned above, the above method includes: performing noise reduction processing on the above ship data to generate second data; performing data completion processing on the above second data through data interpolation operation to generate the above first data.

[0095] It is understandable that since the acquired ship data is the most original data, the data is likely to contain a lot of noise. If the most original data is directly input into the target model, it is very likely that the result obtained by the target model will be an erroneous result. Therefore, the ship data needs to be processed first and then the processed ship data is input into the target model.

[0096] Exemplarily, the ship data may be subjected to two processing methods, namely, noise reduction processing and completion processing, in order to improve the referenceability and accuracy of the ship data.

[0097] The following describes the process of noise reduction and completion through Example 3: Example 3: Assume that the above data processing system is an AIS system, and the target action that needs to be judged is illegal fishing action on the ocean. Correspondingly, the target ship is also an illegal fishing ship that performs illegal fishing action on the ocean.

[0098] After receiving the ship data, data cleaning, also known as data noise reduction, is performed first to ensure the integrity and accuracy of the data. The noise in AIS data usually comes from signal interference, equipment failure, transmission errors, etc. These data noises may manifest as: sudden changes in speed and heading; invalid or incorrect longitude and latitude information; missing or duplicate data records. The specific steps of data cleaning are as follows: Step B1: Filter invalid data: Eliminate unreasonable longitude and latitude information (such as points outside the sea area) and data points with abnormal speed and heading.

[0099] Step B2: Remove duplicate records: At the same time point, there may be duplicate AIS signals, and the system will remove duplicates based on timestamps and ship identification codes.

[0100] Step B3: Processing mutation data: If the change in speed or heading exceeds the normal range (such as the speed exceeds the reasonable limit, the heading exceeds 0 to 360 degrees), it is judged as an abnormal data point and is removed or corrected.

[0101] Next, data completion is performed. Due to unstable signals or transmission delays, AIS data may have missing points, resulting in discontinuity of the ship's trajectory. In order to compensate for this problem, this system uses linear interpolation or spline interpolation to restore the integrity of the trajectory. The interpolation steps mainly include the following two steps: Step C1: Track point interpolation: For missing track points, interpolation calculation is performed based on adjacent valid data points to restore the reasonable track of the ship in this time period.

[0102] Step C2: Feature interpolation: For missing dynamic data such as speed and heading, linear interpolation or other fitting methods are used to fill in the missing data using the trends of adjacent time periods.

[0103] like Figure 2 As shown, the above-mentioned data cleaning steps B1 to B3 and data completion steps C1 to C2 are all data preprocessing steps.

[0104] In another embodiment of the present application, a specific implementation method of selecting at least three first sub-models in the first model as target sub-models is also provided. Exemplarily, the aforementioned “inputting the first data into the target model, processing the first data, and obtaining the target data in the ship data” includes: selecting the first model according to the parameter characteristics of the first data, determining at least one first sub-model in the first model as the target model; inputting the first data into the target model; processing the first data through the first model, and obtaining the target data in the ship data.

[0105] Exemplarily, the first model mentioned above includes at least three data processing models.

[0106] Exemplarily, different data processing models correspond to different parameter characteristics.

[0107] It is understandable that in the embodiment of the present application, since many types of first sub-models can be set, different first sub-models have different processing capabilities for the same data, and accordingly, the processing preferences of each first sub-model are also different. For example, some first sub-models have a higher accuracy rate in recognizing specific actions of ships than other first sub-models, and some first sub-models have a higher accuracy rate in recognizing the distribution of specific protected fish schools in the ocean than other first sub-models. In this case, assuming that the number of first sub-models in the first model is greater than 3, it is necessary to screen the first sub-models in the first model according to the parameter characteristics of the current first data, and select the first sub-model that is more suitable for the current first data for processing the first data.

[0108] It should be noted that the scenario requirements provided in this application are that first, the number of first sub-models in the first model is greater than 3, and second, the amount of first data in the current data processing system is large, and it is hoped that a faster method will be used to determine the target ship. If all first sub-models are used to process the first data with a large amount of data, it may result in large resource consumption and a long time consumption, so screening is required.

[0109] Further, assuming that the current scenario is not that the number of first sub-models in the first model is greater than 3, and the amount of first data in the current data processing system is large, and it is hoped that a faster method will be used to determine the target ship, then all first sub-models can be used to process the first data. Assuming that the first model in the current scenario only includes 3 first sub-models, there is no need to screen, and all first sub-models are directly used to process the first data; and assuming that the current demand for whether to quickly determine the target ship is not strong, and the demand is to determine the target ship as accurately as possible, then the first model in the current scenario includes 5 first sub-models, and there is no need to screen, and all first sub-models are directly used to process the first data.

[0110] Optionally, in an embodiment of the present application, the first model is screened according to parameter characteristics of the first data, and at least one model in the first model is determined to be a target model. The data processing method provided in the present application includes: screening the first model according to parameter characteristics of the first data, and the first data corresponds to data processing characteristics of at least three first sub-models in the first model; screening the first model according to data characteristics of the data group, and determining that at least three first sub-models in the first model are target sub-models.

[0111] Exemplarily, the above-mentioned inputting of the first data into the target model includes in the data processing method provided in the present application: inputting the above-mentioned first data into the above-mentioned at least three first sub-models as target sub-models respectively.

[0112] Optionally, in an embodiment of the present application, the target data corresponding to the first data is obtained, and the target ship corresponding to the target data is determined. The data processing method provided in the embodiment of the present application includes: obtaining third data corresponding to the first data; comparing each third sub-data in the third data with the preset second threshold, and filtering out the target data in the third sub-data; and determining the target ship according to the target data.

[0113] Exemplarily, the third data includes a plurality of third sub-data, each of which is used to indicate the action data of each ship data.

[0114] Exemplarily, the action data includes data of the target action and data of non-target actions.

[0115] It is understandable that the non-target action is an action that is not the target action. For example, if the target action is an illegal fishing action, the non-target action is an action that is not the illegal fishing action.

[0116] Exemplarily, the target data is the third sub-data exceeding a preset second threshold.

[0117] It is understandable that after the above-mentioned first data is processed by at least three first sub-models in the first model, the third data is obtained through the model, and the third data needs to be further compared with the preset second threshold to filter out the target data.

[0118] Exemplarily, the above-mentioned preset second threshold may be preset by the data processing system, or may be preset by a user.

[0119] Furthermore, in order to adapt to the changes in illegal fishing behavior patterns in different sea areas or seasons, the data processing system can dynamically adjust the preset second threshold. For example, assuming that the system is used to detect illegal fishing vessels, during the peak period of illegal fishing, the system can appropriately lower the preset second threshold to increase sensitivity to suspicious vessels; and during periods of less fishing activity, the preset second threshold can be increased to reduce the number of false alarms. In addition, the data processing system can also adaptively adjust the preset second threshold based on feedback from historical data to improve the overall recognition accuracy.

[0120] Furthermore, in combination with the above example 1, the process of confirming the target ship is described as follows: According to the above step A3: determine whether the weighted result (i.e. the above third data) obtained by each first sub-model exceeds the preset second threshold T: the system calculates the final weighted average prediction score is compared with a preset second threshold. , then the vessel's behavior is judged to be fishing behavior. If , it is determined as non-fishing behavior. The preset second threshold is usually set based on experience and historical data to balance the false positive rate and the false negative rate.

[0121] Step A4: Output prediction results: If it is determined to be fishing behavior, the data processing system will enter the next early warning mechanism to confirm that the ship is the target ship; otherwise, the ship's trajectory will be recorded as ordinary maritime navigation behavior and no alarm will be triggered.

[0122] like Figure 2 As shown, the above step A3 is a process of inputting the first data after data preprocessing into the three first sub-models. The above step A4 is a process of outputting the prediction result.

[0123] For example, Figure 2 As shown, after the target ship is confirmed, the real-time alarm mechanism can be activated, that is, an early warning can be issued.

[0124] Furthermore, the system triggers alarms in the following ways: 1) Email alarm: Sends an alarm notification to the email system of the relevant department, including the identification code, current location, trajectory characteristics, and the system's analysis of its behavior of the suspicious ship. 2) SMS alarm: When a ship is identified as a possible target ship, the system will send a real-time SMS notification to the management personnel or law enforcement department to remind them to take timely measures. 3) Monitoring platform alarm: When the system is integrated into the offshore monitoring platform, when the target action is identified, the system can pop up an alarm on the platform and mark the location of the ship and its recent navigation trajectory on the map. 4) Other notification methods: According to needs, the system can also notify relevant personnel through application push, telephone alarm, etc.

[0125] Among them, the content of the early warning information may include the following: The early warning information not only includes immediate alarms, but also detailed analysis reports so that law enforcement agencies can make accurate judgments. The content of the early warning information mainly includes: 1) Basic information of the ship: including the identification code, type, flag state, captain and other basic information of the target ship; 2) Trajectory characteristics: The system will list in detail the AIS trajectory of the target ship in the recent period, including its position, speed, and course changes; 3) Behavior analysis: The system will analyze the behavior pattern of the target ship and point out whether it meets the typical target action characteristics. For example, whether it stays in a certain sea area for a long time, whether there are frequent turns or deceleration behaviors, etc.; 4) Prediction results and confidence: The system outputs the weighted average prediction score of the target ship being identified as fishing behavior , as well as the individual prediction results and weights of each target sub-model, help managers understand the basis for the system's judgment 5) Risk assessment: Based on the historical behavior and current behavior pattern of the target ship, the system conducts a level assessment of the risk of the target action of the target ship, which is convenient for management departments to arrange law enforcement resources according to priority.

[0126] Furthermore, tracking and recording can be performed after the warning. After the data management system issues a warning, the relevant departments can further track and record the target ship. The data processing system supports continuous monitoring of the target ship and stores its behavior records in the database. Through long-term tracking records, the system can accumulate rich behavior data for further analysis of the activity patterns of the target ship and improve the accuracy of future identification.

[0127] The data processing system can also compare the tracking data with the actual law enforcement results. If it is confirmed that the target ship is indeed engaged in the target action, the system will feed the result back to the model training database for subsequent model updates and training. If the ship is proven not to be engaged in illegal activities, the system will also record this information to further optimize the prediction accuracy of the model.

[0128] The present application also provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the training rule determination method described in the present application is implemented. For example, the following can be executed: Figure 1 The individual steps of the method are shown.

[0129] The present invention provides a computer program product, which includes instructions that are executed by a processor when the processor is running. Figure 1 The individual steps of the method are shown.

[0130] It should be noted that although the operations of the method of the present invention are described in a particular order in the drawings, this does not require or imply that the operations must be performed in this particular order or that all illustrated operations must be performed to achieve desired results.

[0131] Figure 3 This is a block diagram of a data processing system according to an embodiment of the present application. Figure 3 The device includes a data acquisition module 601 and a computing unit 602.

[0132] The data acquisition module 601 is used to obtain first data according to the ship data when receiving the ship data, wherein the ship data includes real-time data of multiple ships, and the first data includes the ship data and the characteristic data, wherein the characteristic data is other parameter data corresponding to the ship characteristics of the ship data and different from the parameter information of the ship data; The computing unit 602 is used to input the first data into a target model, process the first data, obtain target data corresponding to the first data, and determine a target ship corresponding to the target data, wherein the target model includes at least three target sub-models, the at least three first sub-models are sub-models in the first model, the target data are data corresponding to a target action, and the target ship is a ship that performs a target action.

[0133] Exemplarily, the data acquisition module 601 is mainly responsible for receiving ship data from the data processing system, and processing signal decoding and forwarding.

[0134] Exemplarily, the computing unit 602, including the CPU and GPU, is responsible for model training, incremental updates, and prediction tasks. For the training of large data sets and complex neural networks, the processing efficiency can be improved through a distributed computing framework (such as Hadoop, Spark).

[0135] In one embodiment, the data processing system also includes a model management module, which is used to process the first data through each target sub-model to obtain the target sub-data corresponding to each target sub-model; and process the target sub-data corresponding to each target sub-model according to the calculated weight of each target sub-model, wherein the calculated weight of each target sub-model is a weight determined according to the data processing accuracy of each target sub-model.

[0136] Exemplarily, the model management module of the data processing system is designed to efficiently manage and update each first sub-model in the first model to ensure that the data processing system maintains optimal performance when responding to the target action. The functions of this module include: 1) Model training interface: Provides an interface for model training, supporting full training and incremental training. Full training is used for model initialization and regular retraining, while incremental training is suitable for lightweight updates when there is new labeled data. 2) Model update mechanism: When the new labeled data is confirmed, the system will regularly start the incremental training mechanism to locally update the existing model. After each model update, the system will automatically perform performance evaluation and decide whether to deploy a new model based on the evaluation results. 3) Model evaluation module: After incremental training or retraining is completed, the system will use the validation set to evaluate the three models. Evaluation indicators include accuracy, precision, recall, F1 score, etc. The system will also dynamically adjust the weights of each model based on historical data to ensure that the weighted average prediction can reflect the latest model performance. 4) Model backup and rollback: The system supports version management for each updated model and saves a copy of the model after each training. If the new model performs poorly in actual applications, the system can quickly roll back to the previous stable version to ensure the stability and reliability of recognition.

[0137] In one embodiment, the model management module is specifically used to: obtain multiple groups of original ship data and multiple groups of target ship data, wherein the target ship data corresponds one-to-one to the original ship data; input the multiple groups of original ship data into each first sub-model in the first model, and generate multiple groups of test arrays corresponding to the multiple groups of original ship data for each first sub-model; compare the multiple groups of test data of each first sub-model with the multiple groups of target ship data, and determine the difference between the multiple groups of test data of each first sub-model and the multiple groups of target ship data; when the difference is greater than a preset first threshold, adjust each first sub-model according to the difference, so that the difference between the test data generated by each first sub-model and the target ship data is less than the preset first threshold, and the first sub-model in the first model is ranked according to the difference corresponding to each first sub-model; and determine the calculation weight of each first sub-model according to the ranking of the first sub-model.

[0138] In one embodiment, the computing unit is specifically used to determine characteristic parameters corresponding to the predetermined action characteristic information according to the ship data and the predetermined action characteristic information; obtain the characteristic data based on the ship data and the characteristic parameters; and fuse the characteristic data and the ship data to generate first data.

[0139] In one embodiment, the data processing system further includes a data preprocessing module. The data preprocessing module is used to perform noise reduction processing on the ship data to generate second data, and perform data completion processing on the second data through data interpolation operation to generate the first data.

[0140] Exemplarily, the data preprocessing module implements the functions of data cleaning, interpolation and feature extraction to ensure that the data input to the model is of high quality and consistency.

[0141] In one embodiment, the computing unit is specifically used to screen the first model according to parameter characteristics of the first data, determine at least one first sub-model in the first model as a target model, and different first sub-models correspond to different parameter characteristics; input the first data into the target model; and the model management module is used to process the first data through the first model to obtain target data in the ship data.

[0142] In one embodiment, the computing unit is specifically used to screen the first model according to parameter characteristics of the first data, the first data corresponding to data processing characteristics of at least three first sub-models in the first model; screen the first model according to data characteristics of the data group, and determine that at least three first sub-models in the first model are target sub-models; and input the first data into the at least three first sub-models as target sub-models.

[0143] In one embodiment, the computing unit is specifically used to obtain third data corresponding to the first data, the third data including multiple third sub-data, each third sub-data being used to indicate action data of each ship data, the action data including data of the target action and data of the non-target action; comparing each third sub-data in the third data with the preset second threshold, filtering out target data in the third sub-data, the target data being the third sub-data exceeding the preset second threshold; and determining the target ship according to the target data.

[0144] Exemplarily, the data processing system further includes: 1) Data storage module: This module stores the cleaned, interpolated and preprocessed AIS data, as well as the feature data set used for model training. Data storage can use an efficient database system, such as a relational database (MySQL, PostgreSQL) or a NoSQL database (such as MongoDB, Cassandra).

[0145] 2) Communication module: This module realizes the communication between the system and the remote terminal, ensuring that the early warning information can be transmitted to the relevant regulatory authorities in a timely manner.

[0146] 3) Early warning module The early warning module is one of the core functions of the system, responsible for issuing an alarm when illegal fishing activities are detected and notifying relevant departments through multiple communication channels.

[0147] The early warning module consists of the following sub-modules: Alarm trigger: When the weighted average prediction result exceeds the preset threshold, this module is responsible for activating the early warning signal, recording the information of suspicious ships and issuing early warning instructions to the system.

[0148] Notification mechanism: The early warning module integrates multiple communication methods to send early warning information to relevant departments and personnel in the fastest way. Supported communication methods include: Email: The warning information will be sent to the administrator's mailbox through the mail server configured in the system background.

[0149] SMS notification: The early warning module integrates the SMS gateway to send alarm messages to the managers’ mobile phones.

[0150] Monitoring platform integration: The system can be integrated with offshore monitoring platforms or other fishery management systems. When an early warning is triggered, an alarm event is automatically generated on the platform, marking the current position of the vessel and its navigation track.

[0151] Information feedback module: After the warning information is issued, the system also needs to receive feedback from management departments or law enforcement personnel. These feedbacks can be confirmation of the actual behavior of the ship (such as confirmation of illegal fishing after law enforcement), or correction of false alarms (such as the ship was not involved in fishing operations). The system will adjust according to the feedback information and optimize the model and threshold.

[0152] Alarm log: The system automatically records each warning event, including basic information of the ship, identification results, alarm time, warning content, notification method and recipient, etc. Through the alarm log, relevant departments can track the alarm history and improve the alarm accuracy of the system through data analysis.

[0153] 4) User interface and management background In order to facilitate user operation and management, the system provides a friendly user interface and management background. Users can monitor, configure and manage the system through a graphical interface. The interface design is simple and intuitive, helping users to efficiently grasp the system status and perform operations.

[0154] Real-time monitoring panel: This panel displays the current monitored ship track data and highlights the movement track and alarm status of suspicious ships through colors or marks. Users can view the detailed information of each ship and the system's analysis report on its behavior in real time.

[0155] Alarm management interface: When the system triggers an alarm, users can view detailed alarm information through the alarm management interface and confirm or reject the system's alarm results. Managers can classify the alarm information and feed back the processing results to the system to help optimize the model.

[0156] Model management interface: This interface allows administrators to manually initiate model training, update, evaluation, and rollback operations. Administrators can view the current status, historical performance, and detailed information of the training dataset for each model through the interface.

[0157] Log and report module: The system provides comprehensive logging and reporting functions, through which users can view the history of system operation, including alarm logs, model update logs, data processing logs, etc. The system can also generate detailed reports to help users evaluate system performance and optimize management decisions.

[0158] Threshold and parameter configuration: Users can adjust the system's threshold, alarm parameters, model weights, etc. in the background interface to meet the needs of illegal fishing detection in different sea areas, seasons or special circumstances. The system supports customized configuration schemes, allowing users to switch configurations according to different time periods and geographical locations.

[0159] It should be understood that the units recorded in the data processing system correspond to the steps in the method described in the accompanying drawings. Therefore, the operations and features described above for the method are also applicable to the data processing system and the units contained therein, and will not be repeated here. The data processing system can be pre-implemented in a browser or other security application of a computer device, or can be loaded into a browser or its security application of a computer device by downloading or the like. The corresponding units in the data processing system can cooperate with the units in the computer device to implement the solution of the embodiment of the present application.

[0160] For the several modules or units mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of the present disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. On the contrary, the features and functions of one module or unit described above can be further divided into multiple modules or units to be embodied.

[0161] It should be noted that for details not disclosed in the data processing system of the embodiments of the present application, please refer to the details disclosed in the above embodiments of the present application, which will not be repeated here.

[0162] Reference below Figure 4 , Figure 4 A schematic diagram of the structure of a computer device suitable for implementing the embodiments of the present application is shown. Figure 4As shown, the computer system 1700 includes a central processing unit (CPU) 1701, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 1702 or the program loaded from the storage part 1708 to the random access memory (RAM) 1703. In the RAM 1703, various programs and data required for the operation instructions of the system are also stored. The CPU 1701, the ROM 1702, and the RAM 1703 are connected to each other through a bus 1704. An input / output (I / O) interface 1705 is also connected to the bus 1704.

[0163] The following components are connected to the I / O interface 1705: an input section 1706 including a keyboard, a mouse, etc.; an output section 1707 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 1708 including a hard disk, etc.; and a communication section 1709 including a network interface card such as a LAN card, a modem, etc. The communication section 1709 performs communication processing via a network such as the Internet. A drive 1710 is also connected to the I / O interface 1705 as needed. A removable medium 1711, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 1710 as needed, so that a computer program read therefrom is installed into the storage section 1708 as needed.

[0164] In particular, according to an embodiment of the present application, the above reference flow chart Figure 1 The described process can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through the communication part 1709, and / or installed from a removable medium 1711. When the computer program is executed by the central processing unit (CPU) 1701, the above-mentioned functions defined in the system of the present application are executed.

[0165] It should be noted that the computer-readable medium shown in the present application may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, 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), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present application, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium such as a computer-readable storage medium that can send, propagate or transmit a program for use by or in conjunction with an instruction execution system, device or device. The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0166] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, functions and operating instructions of the system, method and computer program product according to various embodiments of the present application. In this regard, each box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the aforementioned module, program segment or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the box can also occur in a sequence different from that marked in the accompanying drawings. For example, the boxes represented by two connections can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flow chart, and the combination of the boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operating instruction, or can be implemented with a combination of dedicated hardware and computer instructions.

[0167] The units or modules involved in the embodiments described in the present application may be implemented by software or hardware. The units or modules described may also be set in a processor. For example, it may be described as: a processor includes a first receiving module, a second receiving module, and a sending module. The names of these units or modules do not, in some cases, constitute limitations on the units or modules themselves.

[0168] As another aspect, the present application further provides a computer-readable storage medium, which may be included in the electronic device described in the above embodiment, or may exist independently without being assembled into the electronic device. The above computer-readable storage medium stores one or more programs, and when the above programs are used by one or more processors to execute the data processing method described in the present application.

[0169] The above description is only a preferred embodiment of the present application and an explanation of the technical principles used. Those skilled in the art should understand that the scope of disclosure involved in the present application is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the aforementioned disclosed concept. For example, the above features are replaced with the technical features with similar functions disclosed in the present application (but not limited to) by each other to form a technical solution.

Claims

1. A data processing method, characterized in that: include: In the case of receiving ship data, first data is acquired according to the ship data, the ship data including real-time data of a plurality of ships, the first data including the ship data and the characteristic data, the characteristic data being other parameter data corresponding to the ship characteristics of the ship data and different from the parameter information of the ship data; The first data is input into a target model, the first data is processed, target data corresponding to the first data is obtained, and a target ship corresponding to the target data is determined, the target model includes at least three target sub-models, the at least three first sub-models are sub-models in the first model, the target data is data corresponding to a target action, and the target ship is a ship that performs a target action.

2. The method according to claim 1, characterized in that The processing of the first data, acquiring target data corresponding to the first data, and determining a target ship corresponding to the target data includes: Processing the first data through each of the target sub-models to obtain target sub-data corresponding to each of the target sub-models; The target sub-data corresponding to each target sub-model is processed according to the calculated weight of each target sub-model, wherein the calculated weight of each target sub-model is a weight determined according to the data processing accuracy of each target sub-model.

3. The method according to claim 1, characterized in that In the case of receiving the ship data, before acquiring the target feature data in the ship data, the method includes: Acquire multiple sets of original ship data and multiple sets of target ship data, wherein the target ship data corresponds one-to-one to the original ship data; Inputting the multiple groups of original ship data into each first sub-model in the first model, and generating multiple groups of test arrays corresponding to the multiple groups of original ship data for each first sub-model; Comparing the multiple groups of test data of each first sub-model with the multiple groups of target ship data, and determining the difference between the multiple groups of test data of each first sub-model and the multiple groups of target ship data; In the case where the difference is greater than a preset first threshold, adjusting each of the first sub-models according to the difference, so that the difference between the test data generated by each of the first sub-models and the target ship data is less than the preset first threshold, and ranking the first sub-models in the first model according to the difference corresponding to each of the first sub-models; The calculation weight of each first sub-model is determined according to the ranking of the first sub-model.

4. The method according to claim 1, characterized in that: When receiving the ship data, acquiring the first data according to the ship data includes: Determining characteristic parameters corresponding to the predetermined action characteristic information according to the ship data and the predetermined action characteristic information; Based on the ship data and characteristic parameters, acquiring the characteristic data; The feature data and the ship data are integrated to generate first data.

5. The method according to claim 1, characterized in that Before acquiring the first data according to the ship data, the method further includes: Performing noise reduction processing on the ship data to generate second data; The first data is generated by performing data interpolation processing on the second data.

6. The method according to claim 1, characterized in that The step of inputting the first data into a target model, processing the first data, and acquiring target data in the ship data includes: According to the parameter characteristics of the first data, the first model is screened, and at least one first sub-model in the first model is determined as a target model, where different first sub-models correspond to different parameter characteristics; inputting the first data into a target model; The first data is processed by the first model to obtain target data in the ship data.

7. The method according to claim 6, characterized in that The step of screening the first models according to the parameter characteristics of the first data and determining at least one model in the first models as a target model includes: Selecting the first model according to parameter characteristics of the first data, wherein the first data corresponds to data processing characteristics of at least three first sub-models in the first model; According to the data features of the data group, the first model is screened to determine at least three first sub-models in the first model as target sub-models; The step of inputting the first data into a target model comprises: The first data is respectively input into the at least three first sub-models as target sub-models.

8. The method according to claim 1, characterized in that The acquiring target data corresponding to the first data and determining the target ship corresponding to the target data includes: Acquire third data corresponding to the first data, the third data including a plurality of third sub-data, each third sub-data being used to indicate action data of each ship data, the action data including data of the target action and data of non-target actions; Comparing each third sub-data in the third data with the preset second threshold, and filtering out target data in the third sub-data, wherein the target data is the third sub-data exceeding the preset second threshold; The target ship is determined according to the target data.

9. A data processing system, characterized in that: The device comprises: a data acquisition module, configured to acquire first data according to the ship data when receiving the ship data, wherein the ship data includes real-time data of a plurality of ships, and the first data includes the ship data and the characteristic data, wherein the characteristic data is other parameter data corresponding to the ship characteristics of the ship data and different from the parameter information of the ship data; a data calculation module, used for inputting the first data into a target model, processing the first data, obtaining target data corresponding to the first data, and determining a target ship corresponding to the target data, wherein the target model includes at least three target sub-models, the at least three first sub-models are sub-models in the first model, the target data are data corresponding to a target action, and the target ship is a ship that performs a target action.

10. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the method according to any one of claims 1 to 8 is implemented.

11. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 8 is implemented.

12. A computer program product, comprising instructions, characterized in that: When the instructions are executed by a processor, the method according to any one of claims 1 to 8 is implemented.

Images