A method, system and management method for predicting the number of containers entering and leaving a terminal

Abstract

The present application provides a method, system and management method for predicting the number of incoming and outgoing containers at a terminal, which are applied to the technical field of predicting the number of incoming and outgoing containers at a terminal. By acquiring historical data including voyage numbers, route numbers and the total number of incoming and outgoing containers at a terminal, a bidirectional long short-term memory network regression model and a classification model are trained, and the two models are used to predict the total number of incoming and outgoing containers at a terminal and the proportion of different types of containers in each time period, thereby accurately predicting the specific number of incoming and outgoing containers of various types at a specific voyage and time point, significantly improving the efficiency and accuracy of terminal container management, and optimizing resource allocation and scheduling decisions.

Description

Technical Field

[0001] The present application relates to the technical field of prediction of the number of containers entering and leaving a terminal, and in particular to a method, system and management method for predicting the number of containers entering and leaving a terminal. Background Art

[0002] As a key node in shipping logistics, the operational efficiency of the terminal directly affects the cost and speed of international trade, and thus affects the stability of the global supply chain. In port management and logistics operations, it is crucial to accurately predict the number of containers entering and leaving the terminal.

[0003] However, the current forecasting methods have the limitation of being unable to update the forecasting model in real time to adapt to changes in terminal operations. At the same time, due to insufficient forecasting accuracy, it is difficult to achieve optimal allocation of terminal resources, resulting in resource shortages during peak periods and idle resources during trough periods, which seriously affects the overall operational efficiency of the terminal.

[0004] Based on this, a new method for predicting the number of container pickups at the terminal is needed. Summary of the invention

[0005] In view of this, the embodiments of this specification provide a method, system and management method for predicting the number of incoming and outgoing containers at a terminal, which are applied to the technical field of predicting the number of incoming and outgoing containers at a terminal. By analyzing ship-related information such as voyages and routes, not only the total number of incoming and outgoing containers at the terminal in various time periods is predicted, but also the number of various types of containers is predicted, providing detailed data support for the operation of the terminal, making the scheduling and management of the terminal more accurate and efficient, thereby optimizing resource allocation, significantly improving the overall smoothness of logistics operations, and improving the operating efficiency of the terminal.

[0006] The embodiments of this specification provide the following technical solutions:

[0007] The present invention provides a method for predicting the number of containers picked up at a terminal, including:

[0008] Acquire real-time data, including voyage number, route number, and total number of containers picked up and delivered at the terminal for each voyage;

[0009] The real-time data is input into a bidirectional long short-term memory network regression model to obtain the total amount of incoming boxes expected in each period;

[0010] Input the real-time data into a random forest classification model to obtain the probability of each type of container in each time period;

[0011] According to the total amount of containers expected to be picked up in each time period and the probability of each type of container in each time period, the number of containers of each type picked up in a specific voyage and at a specific time is obtained.

[0012] The embodiment of this specification also provides a system for predicting the number of containers entering and leaving a terminal, including:

[0013] A data acquisition module is used to acquire real-time data, including voyage number, route number and the total amount of containers picked up and delivered at the terminal for each voyage;

[0014] The prediction module is used to make predictions based on real-time data using a bidirectional long short-term memory network regression model and a random forest classification model to obtain the estimated total number of containers picked up in each period and the probability of each type of container in each period;

[0015] The data processing module is used to obtain the number of containers of each type picked up at a specific time on a specific voyage based on the total number of containers picked up and the probability of each type of container in each time period predicted by the prediction module.

[0016] The embodiment of the present specification also provides a method for managing container loading and unloading at a terminal, comprising: using any of the above-mentioned methods for predicting the number of container loading and unloading at a terminal to predict the number of container loading and unloading at a terminal;

[0017] After completing the prediction, converting the prediction data into a structured format;

[0018] Transmitting the forecast data in the structured format to the port operation management system through an API interface;

[0019] The port operation management system manages the containers at the terminal according to the received prediction data.

[0020] Compared with the prior art, the beneficial effects that can be achieved by at least one of the above technical solutions adopted in the embodiments of this specification include at least:

[0021] By obtaining historical data including voyage numbers, route numbers and the total number of containers picked up and delivered at the terminal, a bidirectional long short-term memory network regression model and a classification model are trained. These two models are used to predict the total number of containers picked up and delivered at the terminal and the proportion of different types of containers in each time period, thereby accurately predicting the specific number of containers picked up and delivered at a specific voyage and time point, significantly improving the efficiency and accuracy of terminal container management and optimizing resource allocation and scheduling decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0023] Figure 1 This is the full flow chart of the operation of the terminal container loading and unloading prediction method in this application;

[0024] Figure 2 It is a flow chart of the method for predicting the number of containers picked up at a terminal in this application. DETAILED DESCRIPTION

[0025] The embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0026] The following describes the implementation methods of the present application through specific examples, and those skilled in the art can easily understand other advantages and effects of the present application from the contents disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The present application can also be implemented or applied through other different specific implementation methods, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present application. It should be noted that, in the absence of conflict, the following embodiments and the features in the embodiments can be combined with each other. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without making creative work belong to the scope of protection of the present application.

[0027] It should be noted that various aspects of the embodiments within the scope of the appended claims are described below. It should be apparent that the aspects described herein can be embodied in a wide variety of forms, and any specific structure and / or function described herein is merely illustrative. Based on the present application, it should be understood by those skilled in the art that an aspect described herein can be implemented independently of any other aspect, and two or more of these aspects can be combined in various ways. For example, any number and aspect described herein can be used to implement the device and / or practice the method. In addition, other structures and / or functionalities other than one or more of the aspects described herein can be used to implement this device and / or practice this method.

[0028] It should also be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present application. The drawings only show components related to the present application rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.

[0029] Additionally, in the following description, specific details are provided to facilitate a thorough understanding of the examples. However, it will be understood by those skilled in the art that the examples can be practiced without these specific details.

[0030] In port management and logistics operations, accurate prediction of the number of containers arriving and departing plays a key role in achieving rational resource allocation, optimizing scheduling decisions, and improving operational efficiency. Through precise predictions, the terminal's loading and unloading equipment, human resources, and site usage can be planned in advance, thereby maximizing economic benefits while ensuring efficient port operations.

[0031] At present, the existing forecasting methods cannot accurately capture the dynamic changes in the number of containers, resulting in resource shortages during peak periods and idle resources during low periods, which seriously affects the overall operational efficiency of the terminal.

[0032] In view of this, the inventors discovered through research and improvement exploration that existing prediction schemes often process time series data from a single direction (such as from front to back), and are unable to simultaneously consider the impact of the past and future in the time series data, resulting in deficiencies in capturing long-term dependencies in the time series, limiting the accuracy of the prediction, and having difficulty adapting to dynamic changes in time series data, especially when the changing trends or patterns of the data exhibit different characteristics over time.

[0033] Based on this, the embodiment of this specification proposes a method for predicting the number of containers entering and leaving a terminal: Figure 2 As shown in the figure, the overall idea is: by obtaining ship-related information including voyage number, route number and the total number of containers picked up and delivered at the terminal in real time, the bidirectional long short-term memory network (BILSTM) regression model is used to predict the total number of containers picked up and delivered. At the same time, the random forest classification model is used to predict the proportion probability of different types of containers. Finally, the results of the two models are combined to obtain the specific number of containers picked up and delivered for a specific voyage within a specific time period. This will achieve accurate prediction of the number of containers picked up and delivered at the terminal, provide data support for more refined and reasonable operation of the subsequent terminal, optimize resource allocation and scheduling decisions, and significantly improve the efficiency and accuracy of container management at the terminal.

[0034] The technical solutions provided by various embodiments of the present application are described below in conjunction with the accompanying drawings.

[0035] like Figure 2 As shown, the embodiment of this specification provides a method for predicting the number of containers picked up at a terminal, including:

[0036] Acquire real-time data, including voyage number, route number, and total number of containers picked up and delivered at the terminal for each voyage;

[0037] The real-time data is input into a bidirectional long short-term memory network regression model to obtain the total amount of incoming boxes expected in each period;

[0038] Input the real-time data into a random forest classification model to obtain the probability of each type of container in each time period;

[0039] According to the total amount of containers expected to be picked up in each time period and the probability of each type of container in each time period, the number of containers of each type picked up in a specific voyage at a specific time is obtained.

[0040] Specifically, through the data interface, real-time voyage-related information used in training the model is obtained, and the acquired real-time data is preprocessed, including:

[0041] Remove outliers from the data, and / or perform standardization and feature engineering on the real-time data.

[0042] The processed real-time data is then fed into the Bidirectional Long Short-Term Memory (BILSTM) regression model to output the estimated total container volume for a specific voyage in different time periods. , where t represents each forecast period and j represents a specific voyage.

[0043] The processed real-time data is passed into the random forest classification model to obtain the proportion of each type of container in the total volume for a specific voyage , where t represents each forecast period, i corresponds to different container types, and j represents a specific voyage.

[0044] That is, the probability that the containers of each voyage belong to each category is obtained. The probability here can also be regarded as the proportion of the containers of each voyage belonging to each category.

[0045] Finally, the total container volume prediction value of the BILSTM regression model for a specific voyage is Probability values ​​from the random forest classification model Multiply them together to calculate the specific number of containers of various types picked up at a specific time on a specific voyage. .

[0046]

[0047] For example, if the total expected number of incoming containers is 100, and the random forest model predicts that the proportion of 20-foot containers is 30% (i.e. 0.3) and the proportion of refrigerated containers is 10% (i.e. 0.1):

[0048] Then, the number of 20-foot containers can be calculated as: number of 20-foot containers = 100*0.3 = 30;

[0049] The number of cold boxes can be calculated as: number of cold boxes = 100*0.1 = 10;

[0050] The number of refrigerated containers in a 20-foot container can be calculated as: number of refrigerated containers in a 20-foot container = 100*0.3*0.1=3;

[0051] This enables the number of all types of containers to be calculated, ensuring the comprehensiveness and accuracy of the forecast results.

[0052] This application starts with ship-related information such as voyages and routes, and on the basis of predicting the total number of containers picked up and received at the terminal in each time period, also predicts the number of different types of containers therein, so as to plan the loading and unloading equipment, human resources and site use of the terminal in advance, avoid resource shortages during peak periods or idle resources during trough periods, and improve overall efficiency, which plays an important role in improving the efficiency and reliability of the entire logistics chain.

[0053] In some embodiments, such as Figure 1 As shown, before the real-time data is input into the bidirectional long short-term memory network regression model, the bidirectional long short-term memory network regression model is trained, including:

[0054] Creating multiple bidirectional long short-term memory network models, and presetting the number of layers, convergence algorithm, and activation function of each bidirectional long short-term memory network regression model;

[0055] Using the historical data to train each bidirectional long short-term memory network regression model;

[0056] According to the prediction accuracy of the total number of containers picked up and delivered, the best one is selected from the trained bidirectional long short-term memory network regression models as the final bidirectional long short-term memory network regression model.

[0057] Specifically, in the BILSTM regression model, the dependent variables are the total number of containers delivered and the total number of containers picked up in each time period, and the independent variables are voyage-related information.

[0058] Create multiple Bidirectional Long Short-Term Memory (BILSTM) models with different configurations, including different numbers of layers, convergence algorithms, and activation functions.

[0059] The processed historical data is organized into tensor data to meet the input requirements of the deep learning model. The organized tensor data is then used to train each BILSTM model, and the best performing model is selected as the final regression model. This model has the smallest error and the best generalization ability in predicting the total number of container pickups, thus ensuring that the final selected BILSTM model has the best performance on historical data and can reliably predict the future number of container pickups.

[0060] When predicting the total number of boxes entering and exiting the entrance, this embodiment can implement bidirectional training from front to back and from back to front in the time dimension by inputting the processed data into the BILSTM model. Compared with traditional machine learning methods, this embodiment can better capture the long dependencies in the time series, and performs better when dealing with nonlinear problems, and has better prediction results.

[0061] In some embodiments, historical voyage related information is extracted from the data system, including but not limited to the voyage number, route number, and the total number of containers picked up and dropped off at the terminal for each voyage, etc., and the historical data is cleaned (such as removing outliers in the data and filling missing values) and the data is standardized and feature engineered to ensure the data quality and applicability of the input model.

[0062] The BILSTM regression model in this application is specially designed to capture long-term dependencies in time series data, and can learn from both the front-to-back and back-to-front directions, effectively improving the accuracy and robustness of the prediction of the number of containers in and out. Specifically, the total number of containers in and out in each time period is first generated as the output variable based on the historical voyage information, and then the processed multi-dimensional input features are fed into the BILSTM model for training.

[0063] In some embodiments, when training a bidirectional long short-term memory network regression model, a cross-validation method is used, and at least one of the following evaluation indicators is used to evaluate the performance of the bidirectional long short-term memory network regression model: mean square error, standard mean square error, or absolute value error to obtain the final bidirectional long short-term memory network regression model.

[0064] Specifically, during the training process of the bidirectional long short-term memory network regression model, cross-validation is used to evaluate the generalization ability of the model. In each step of the cross-validation, at least one of the following evaluation indicators is used to measure the regression effect of the model: mean square error, standard mean square error, absolute value error, etc.

[0065] In some embodiments, before inputting the real-time data into the random forest classification model, the random forest classification model is trained, including:

[0066] presetting a parameter grid according to the historical data;

[0067] Through grid search, each parameter combination in the preset parameter grid is traversed, and each parameter combination is trained using the historical data to obtain multiple trained classification models;

[0068] The model with the best classification effect is selected from multiple trained classification models as the final classification model.

[0069] Specifically, in the random forest classification model, the dependent variable is the container type for loading and unloading in each time period, and the independent variable is the voyage-related information.

[0070] Through grid search, we traverse different values ​​of parameters such as the maximum number of child nodes and maximum depth of the random forest classification model. For each parameter combination, we train a random forest model and use historical data to train these models. Finally, we select the model with the best classification effect as the final classification model for the actual container type prediction task.

[0071] In some embodiments, when training a random forest classification model, at least one of the following evaluation indicators is used to evaluate the performance of the random forest classification model: accuracy, precision, or F1 score.

[0072] Specifically, when training a random forest classification model, evaluation indicators such as Precision, Accuracy, and F1 Score are used to measure the performance of the model on the classification task, such as the ability to accurately identify positive samples, the overall correct classification ratio, and the balance between precision and recall, so as to ensure that the final selected random forest model has the best performance on the given data set and improve the prediction accuracy of the model in practical applications.

[0073] In some embodiments, in order to further refine the prediction of the number of different types of containers, a classification model is constructed using a random forest algorithm based on the total incoming and outgoing container volume obtained based on the BILSTM regression model.

[0074] Specifically, the proportion of each type of container (such as standard 20-foot container, 40-foot container, etc.) is classified and predicted, the model performance is optimized by setting different numbers and depth parameters of decision trees, and the cross-validation method is used to evaluate the generalization ability of the model on unseen data.

[0075] In some embodiments, the method for predicting the number of incoming and outgoing containers at a terminal further includes:

[0076] The forecast data for different time points, container types and / or voyages are counted to obtain the total forecast value.

[0077] Specifically, By accumulating the time t dimension, the type i dimension, and the voyage j dimension, we can obtain various high-order statistical data.

[0078] For example, by accumulating the number of containers picked up and dropped off at each time point output by the prediction model, the total prediction value for a specific time period (such as a day, a week, or a month) can be obtained, which helps port management understand the container flow in a specific time period and thus conduct effective resource planning and scheduling.

[0079] For example, for each type of container, the predicted number of incoming and outgoing containers is summarized to obtain the total predicted value for each container type in a specific time period, which helps to optimize yard space allocation, loading and unloading operation planning, and meet the storage and retrieval needs of specific types of containers, thereby improving port operation efficiency.

[0080] For example, for each voyage, the predicted number of containers loaded and unloaded is accumulated to obtain the total predicted value for each voyage within a specific time period, thereby providing an accurate plan for the ship's loading and unloading operations.

[0081] Through statistical analysis, ports can identify potential peak periods or high demand for specific types of containers, so as to take measures in advance to avoid resource shortages or operational delays and improve operational efficiency.

[0082] In some embodiments, the original prediction data and the predicted statistical data are organized into a structured format and sent to the port operation management system or any other relevant party through an API interface.

[0083] In the time dimension, the bidirectional long short-term memory network (BILSTM) of this application can learn from both the front-to-backward and the back-to-forward directions, effectively capturing the long-term dependencies in the time series and improving the accuracy and robustness of the prediction of the number of incoming and outgoing containers. In addition, the random forest classification model further refines the prediction of the number of different types of containers, ensuring the comprehensiveness and meticulousness of the prediction results. This invention achieves accurate prediction of the number of incoming and outgoing containers at the terminal, has the ability to respond quickly, and can provide real-time and reliable decision support data for the port operation management system, thereby effectively improving terminal operation efficiency and resource utilization, and optimizing the operation process of the entire logistics chain.

[0084] Based on the same inventive concept, the present application also provides a system for predicting the number of containers entering and leaving a terminal, including:

[0085] A data acquisition module is used to acquire real-time data, including voyage number, route number and the total amount of containers picked up and delivered at the terminal for each voyage;

[0086] The prediction module is used to make predictions based on real-time data using a bidirectional long short-term memory network regression model and a random forest classification model to obtain the estimated total number of containers picked up in each period and the probability of each type of container in each period;

[0087] The data processing module is used to obtain the number of containers of each type picked up at a specific time on a specific voyage based on the total number of containers picked up and the probability of each type of container in each time period predicted by the prediction module.

[0088] In some embodiments, the terminal container pick-up quantity prediction system further includes:

[0089] Model training module, used to train bidirectional long short-term memory network regression model and random forest classification model;

[0090] Among them, a bidirectional long short-term memory network regression model is trained to predict the total amount of container pickup at the terminal in each time period;

[0091] The random forest classification model is used to predict the proportion of different types of containers in the total number of containers picked up and delivered at the terminal in each time period.

[0092] In some embodiments, the terminal container pick-up quantity prediction system further includes:

[0093] The data statistics module is used to collect statistics on the forecast data at different time points, container types and / or voyages to obtain the total forecast value.

[0094] Based on the same inventive concept, the present application also provides a method for managing incoming and outgoing containers at a terminal, comprising: using any of the above-mentioned methods for predicting the number of incoming and outgoing containers at a terminal to predict the number of incoming and outgoing containers at a terminal;

[0095] After completing the prediction, converting the prediction data into a structured format;

[0096] Transmitting the forecast data in the structured format to the port operation management system through an API interface;

[0097] The port operation management system manages the containers at the terminal according to the received prediction data.

[0098] In this specification, the same or similar parts between the various embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the embodiments described later, the description is relatively simple, and the relevant parts can be referred to the partial description of the previous embodiments.

[0099] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A method for predicting the number of containers picked up at a terminal, characterized in that: include: Acquire real-time data, including voyage number, route number, and total number of containers picked up and delivered at the terminal for each voyage; The real-time data is input into a bidirectional long short-term memory network regression model to obtain the total amount of incoming boxes expected in each period; At the same time, the real-time data is input into a random forest classification model to obtain the probability of each type of container in each time period; According to the total amount of containers expected to be picked up in each period and the probability of each type of container in each period, the number of containers picked up in and out of a specific voyage at a specific time is obtained to avoid resource shortages during peak periods or idle resources during low periods.

2. The method for predicting the number of incoming and outgoing containers at a terminal according to claim 1, characterized in that: Before inputting the real-time data into the bidirectional long short-term memory network regression model, the bidirectional long short-term memory network regression model is trained, including: Creating multiple bidirectional long short-term memory network models, and presetting the number of layers, convergence algorithm, and activation function of each bidirectional long short-term memory network regression model; Use historical data to train each bidirectional LSTM regression model; According to the prediction accuracy of the total number of containers picked up and delivered, the best one is selected from the trained bidirectional long short-term memory network regression models as the final bidirectional long short-term memory network regression model.

3. The method for predicting the number of incoming and outgoing containers at a terminal according to claim 2, characterized in that: When training a bidirectional long short-term memory network regression model, a cross-validation method is used, and at least one of the following evaluation indicators is used to evaluate the performance of the bidirectional long short-term memory network regression model: mean square error, standard mean square error or absolute value error to obtain a final bidirectional long short-term memory network regression model.

4. The method for predicting the number of incoming and outgoing containers at a terminal according to claim 1, characterized in that: Before the real-time data is input into the random forest classification model, the random forest classification model is trained, including: Preset parameter grid based on historical data; Through grid search, each parameter combination in the preset parameter grid is traversed, and each parameter combination is trained using the historical data to obtain multiple trained classification models; The model with the best classification effect is selected from multiple trained classification models as the final classification model.

5. The method for predicting the number of incoming and outgoing containers at a terminal according to claim 4, characterized in that: When training a random forest classification model, at least one of the following evaluation indicators is used to evaluate the performance of the random forest classification model: accuracy, precision, or F1 score.

6. The method for predicting the number of incoming and outgoing containers at a terminal according to claim 1, characterized in that: The method for predicting the number of incoming and outgoing containers at a terminal further includes: The forecast data for different time points, container types and / or voyages are counted to obtain the total forecast value.

7. A system for predicting the number of containers entering and leaving a terminal, characterized in that: include: A data acquisition module is used to acquire real-time data, including voyage number, route number and the total amount of containers picked up and delivered at the terminal for each voyage; The prediction module is used to make predictions based on real-time data using a bidirectional long short-term memory network regression model and a random forest classification model to obtain the estimated total number of containers picked up in each period and the probability of each type of container in each period; The data processing module is used to obtain the number of containers of each type picked up at a specific time on a specific voyage based on the total number of containers picked up and the probability of each type of container in each time period predicted by the prediction module.

8. The system for predicting the number of incoming and outgoing containers at a terminal according to claim 7, characterized in that: The terminal container pick-up quantity prediction system also includes: Model training module, used to train bidirectional long short-term memory network regression model and random forest classification model; Among them, a bidirectional long short-term memory network regression model is trained to predict the total amount of container pickup at the terminal in each time period; The random forest classification model is used to predict the proportion of different types of containers in the total number of containers picked up and delivered at the terminal in each time period.

9. The system for predicting the number of incoming and outgoing containers at a terminal according to claim 7, characterized in that: The terminal container pick-up quantity prediction system also includes: The data statistics module is used to collect statistics on the forecast data at different time points, container types and / or voyages to obtain the total forecast value.

10. A container loading and unloading management method at a terminal, characterized in that: include: Using the method for predicting the number of incoming and outgoing containers at a terminal as claimed in any one of claims 1 to 6, predicting the number of incoming and outgoing containers at a terminal; After completing the prediction, convert the prediction data into a structured format; Transmitting the forecast data in the structured format to the port operation management system through an API interface; The port operation management system manages the containers at the terminal according to the received prediction data.

Images