Empty box management and control system of ship company

Abstract

The invention relates to the technical field of shipping management, and discloses a shipping company empty box management and control system, the system is composed of a data acquisition layer, a data processing layer, a machine learning layer, a decision optimization layer, an application integration layer and a user interface layer, the system can acquire and integrate multi-source data in real time through an Internet of Things equipment interface and an external data source interface, and the system is convenient to use. And a basis is provided for subsequent intelligent analysis. And by utilizing a machine learning technology, the system can accurately predict the empty box overstock condition and automatically generate a scientific allocation scheme. Meanwhile, the system also provides optimal path planning and anomaly detection and early warning functions, and ensures high efficiency and safety of empty box circulation. Through a standardized API interface and an automatic control module, the system realizes seamless integration with an existing information system of a ship company and comprehensive control of port automation equipment. The user interface layer provides an intuitive monitoring interface and rich data analysis reports, so that a ship company can make decisions and optimize conveniently.

Description

Technical Field

[0001] The present invention relates to the technical field of shipping management, and in particular to an empty container management and control system of a shipping company. Background Art

[0002] In the global shipping industry, empty container management has always been one of the major challenges facing shipping companies. As a key link in shipping logistics, the efficient management of empty containers is directly related to the operating efficiency and cost control of shipping companies. In traditional empty container management, the path planning of empty container circulation mainly relies on manual experience and simple rules, and it is impossible to automatically generate the optimal transportation path according to real-time conditions and optimization goals. At the same time, there is a lack of effective anomaly detection mechanisms, and it is impossible to timely discover and deal with potential risks in the empty container circulation process, which brings safety hazards to the operation of shipping companies.

[0003] Moreover, the traditional empty container management system lacks effective interfaces and communication methods with the existing information systems of shipping companies, resulting in information islands and data non-sharing. The control of port automation equipment mainly relies on manual operation and simple automation programs, which cannot achieve comprehensive automation control of empty container loading and unloading and storage. Summary of the invention

[0004] The object of the present invention is to provide an empty container management and control system for a shipping company to solve the problems raised in the above background technology.

[0005] To achieve the above-mentioned object, the present invention provides the following technical solutions: a shipping company empty container management and control system, comprising a data collection layer, a data processing layer, a machine learning layer, a decision optimization layer, an application integration layer and a user interface layer;

[0006] The data collection layer includes an IoT device interface and an external data source interface; the IoT device interface is used to connect and collect data from container tracking equipment, port automation equipment and ship dynamic systems; the external data source interface is used to obtain data from weather stations, port databases and shipping company ERP systems;

[0007] The data processing layer includes a data preprocessing module, a feature extraction module and a data storage module; the data preprocessing module is used to clean the collected raw data, including removing noise, duplicates and invalid data, and data standardization; the feature extraction module is used to extract features related to empty box management from the cleaned data; the data storage module is used to store the processed data in a database;

[0008] The machine learning layer includes a model training module, a model optimization module and a model deployment module; the model training module uses known historical data of empty box flow to train an empty box backlog prediction model, and the output of the empty box backlog prediction model is the prediction result of the future state of the empty box; the model optimization module is used to evaluate and optimize the trained model; the model deployment module is used to deploy the optimized model to the production environment;

[0009] The decision optimization layer includes an intelligent decision module, a path planning module and an anomaly detection and early warning module; the intelligent decision module is used to automatically generate an empty container allocation plan based on the prediction results of the empty container backlog prediction model; the path planning module is used to provide the optimal empty container transportation path planning; the anomaly detection and early warning module is used to monitor the empty container flow in real time and identify potential risks;

[0010] The application integration layer includes a system interface module and an automation control module; the system interface module is used to provide a standardized API interface for integration with the existing information system of the shipping company; the automation control module is used to communicate with the port automation equipment to achieve automated control of empty container loading and unloading and storage;

[0011] The user interface layer includes a real-time monitoring interface, a data analysis report and a user interaction module; the real-time monitoring interface is used to provide a real-time monitoring view of the empty container flow; the data analysis report is used to generate a detailed data analysis report; the user interaction module is used to provide a user interface for querying empty container information and adjusting optimization strategies.

[0012] Preferably, the data preprocessing module is implemented as follows:

[0013] Remove noise data: Use a threshold-based method to identify and process outlier data in the data set;

[0014] Remove duplicates: Use hash table algorithm to remove duplicate records in the data set;

[0015] Remove invalid records: Filter and delete invalid records in the dataset based on predefined rules for missing key fields and field values ​​that do not conform to the expected format.

[0016] Preferably, the data preprocessing module adopts the Z-score standardization method, subtracting the mean of each numerical data and dividing it by its standard deviation, so that the processed data conforms to the standard normal distribution. The formula for standardization processing is: Among them, X is the original data, μ is the mean of the original data, σ is the standard deviation of the original data, and Z is the standardized data.

[0017] Preferably, the step of acquiring the empty container backlog prediction model comprises:

[0018] Step 1: Collect the historical data of empty container circulation of shipping companies, including empty container location, status, transportation time, port operation status, ship dynamics, weather conditions and corresponding empty container backlog labels;

[0019] Step 2: Process the collected data through the data preprocessing module and feature extraction module to form a regular data set; further divide the data set into training set, validation set and test set, where the training set is used for model training, the validation set is used to verify and optimize the model effect, and the test set is used to verify the generalization ability of the model on unknown data sets;

[0020] Step 3: Select the linear regression model to build an empty container backlog prediction model. The algorithm formula of the linear regression model is:

[0021] y=β 0 +β 1 x 1 +β 2 x 2 +...+γ n x n +∈

[0022] Among them, y is the target variable, representing the empty container backlog, x 1 、x 2 , ..., x n is the input variable, representing the characteristic parameter that affects the empty container backlog; β 1 , β 2 , ..., β n is the regression coefficient, determined by model training; β 0 is the intercept term, which represents the basic value of the empty container backlog; ∈ is the error term; the purpose of model training is to estimate the regression coefficient β 0 , β 1 , β 1 , β 2 , ..., β n , so that the error between the predicted value and the true value is minimized; when initializing the model parameters, for each feature x in the linear regression model i The corresponding parameter β i Assign an initial value, which is set to a random number close to 0, and the intercept term β 0 An initial value is also given; these initial parameter values ​​will serve as the starting point for model training and will be continuously optimized and adjusted through subsequent training processes;

[0023] Step 4: Use the training set to train the empty box backlog prediction model, and continuously iterate and update the parameters of the linear regression model through the gradient descent algorithm to minimize the loss function value between the true value and the predicted value; in each iteration, adjust the value of the model parameter according to the gradient information of the loss function to the model parameter until the loss function converges or reaches the preset number of iterations; use the validation set to verify the trained model, and adjust the model parameters or structure according to the verification results; use the test set to comprehensively evaluate the optimized model through the set evaluation indicators to ensure that the model performance meets the expected standards;

[0024] Step 5: Save the trained and verified empty container backlog prediction model for subsequent use in the intelligent decision-making layer to predict the empty container backlog in real time.

[0025] Preferably, the intelligent decision model receives the future empty container backlog prediction result output by the empty container backlog prediction model, the result including the predicted empty container backlog amount at different time points and the possible backlog trend;

[0026] According to the prediction results, combined with the shipping company's operating strategy and cost considerations, the empty container allocation goals and constraints are set; the empty container allocation goals and constraints are converted into a mathematical optimization model, specifically:

[0027] Set the decision variable x ij represents the number of empty containers transferred from port i to port j, c ij represents the transportation cost from port i to port j, d j represents the predicted empty container demand at port j, s i represents the number of available empty containers at port i, and C represents the total transportation cost budget. The optimization problem can be expressed as:

[0028] Minimize∑ ij c ij * ij

[0029] The constraints are:

[0030] The empty container transfer volume cannot exceed the number of available empty containers at the port:

[0031] The number of empty containers transferred should meet the port's forecast demand: Among them, δ represents the error range of the operation;

[0032] The total shipping cost cannot exceed the budget:∑ ij c ij * ij ≤C;

[0033] A mathematical optimization model is constructed through the above algorithm formula to solve the optimal solution for empty container allocation;

[0034] Based on the solution results of the mathematical optimization model, an empty container allocation plan is generated; the plan includes a specific empty container transportation plan, transportation route, time schedule and related cost budget.

[0035] Preferably, on the basis of the empty container allocation plan generated by the intelligent decision-making module, the path planning module further refines the planning of the transportation path; according to the starting port, the end port and the transportation volume in the empty container allocation plan, the path planning module constructs a network diagram containing all possible transportation paths; the nodes in the network diagram represent ports, the edges represent the routes between the ports, and the weights of the edges are comprehensively determined according to the transportation cost, sailing time and safety of the routes;

[0036] The path planning module uses an intelligent optimization algorithm to search for the optimal path from the starting port to the destination port in the network graph.

[0037] Preferably, the anomaly detection and early warning module continuously monitors and analyzes real-time data during the empty container allocation process to detect any abnormal situation that deviates from the normal mode and issue an early warning signal in a timely manner.

[0038] Preferably, the system interface module includes an internal interface unit and an external interface unit, the internal interface unit is used to realize communication and data exchange between modules in the system, and the external interface unit is used for interaction between the system and an external data source.

[0039] Preferably, the automation control module designs corresponding control logic based on the decision results output by the intelligent decision-making module, including determining control objectives, formulating control strategies and setting control parameters, and further connects and communicates with various types of equipment through standard communication protocols, including sending control instructions, receiving equipment status feedback, and monitoring equipment operating status.

[0040] Compared with the prior art, the present invention has the following beneficial effects:

[0041] The present invention collects and integrates data from container tracking equipment, port automation equipment, ship dynamic systems, weather stations, port databases, and shipping company ERP systems in real time through the Internet of Things device interface and the external data source interface. This enables shipping companies to grasp the actual status of empty containers in a timely manner, thereby significantly improving the efficiency and accuracy of empty container management. Using the empty container backlog prediction model in the machine learning layer, the system can accurately predict the future demand and backlog of empty containers based on historical data. Combined with the intelligent decision-making module, shipping companies can automatically generate scientific empty container allocation plans to effectively avoid the waste of empty container resources and the increase in transportation costs.

[0042] The route planning module can automatically generate the optimal empty container transportation route planning based on real-time conditions and optimization goals, thereby improving transportation efficiency and reducing transportation costs. At the same time, the anomaly detection and early warning module can monitor the empty container flow in real time, promptly identify and handle potential risks, and ensure the safe operation of the shipping company.

[0043] The application integration layer provides a standardized API interface, which enables the system to be seamlessly integrated with the existing information system of the shipping company to achieve information sharing and data interoperability. The automation control module realizes the comprehensive automation control of empty container loading and unloading and storage, reducing manual intervention and improving operation efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is the overall structure diagram of the present invention;

[0045] Figure 2 This is the working principle diagram of the intelligent decision-making module;

[0046] Figure 3 The training flow chart of the empty box backlog prediction model. DETAILED DESCRIPTION

[0047] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0048] See also Figure 1-3 The present invention provides a technical solution: a shipping company empty container management and control system, including a data collection layer, a data processing layer, a machine learning layer, a decision optimization layer, an application integration layer and a user interface layer.

[0049] The data collection layer is responsible for collecting various data required in the empty container circulation process. This layer includes the IoT device interface and the external data source interface.

[0050] IoT device interface: This interface is connected to container tracking equipment, port automation equipment and ship dynamic systems to collect key information such as the location, status, and transportation route of empty containers in real time.

[0051] External data source interface: This interface obtains weather information, port operation data and shipping company business data related to empty container management from external data sources such as weather stations, port databases and shipping company ERP systems. These data are combined with data collected by IoT devices to provide comprehensive information support for the system.

[0052] The data processing layer is responsible for cleaning, feature extraction and storage of the collected raw data. The data processing layer includes:

[0053] Data preprocessing module: This module first removes noise, duplicates and invalid data from the data to ensure the accuracy and reliability of the data. Then, the data is standardized to make data from different sources and formats comparable, which is convenient for subsequent analysis and processing.

[0054] Feature extraction module: This module extracts features related to empty container management from the cleaned data, such as the demand for empty containers, backlog, and turnover speed. These features provide key inputs for subsequent machine learning model training and decision optimization.

[0055] Data storage module: This module stores the processed data in the database, providing data support for all levels of the system. The design of the database takes into account the accessibility, security and scalability of the data, ensuring that the system can run efficiently and stably.

[0056] The machine learning layer uses known historical data on empty container turnover to train an empty container backlog prediction model, and evaluates and optimizes the model. The machine learning layer includes:

[0057] Model training module: This module uses historical data to train an empty container backlog prediction model. The model uses advanced machine learning algorithms to accurately predict future empty container demand and backlog. Through continuous training and learning, the model's prediction accuracy will gradually improve.

[0058] Model optimization module: This module evaluates and optimizes the trained model. By comparing the model's prediction results with the actual data, the model's deficiencies are discovered and targeted adjustments and optimizations are made. The optimized model can better adapt to the actual operating environment and improve the accuracy of the prediction.

[0059] Model deployment module: This module deploys the optimized model to the production environment. Through standardized interfaces and processes, it ensures that the model can run stably and efficiently in actual operations. At the same time, the module also provides model monitoring and update functions to ensure that the model can continue to adapt to changes in the operating environment.

[0060] The decision optimization layer automatically generates empty container allocation plans based on the prediction results of the empty container backlog prediction model, and provides optimal empty container transportation path planning and anomaly detection and early warning functions. The decision optimization layer includes:

[0061] Intelligent decision-making module: This module automatically generates an empty container allocation plan based on the forecast results and the shipping company's operating strategy. The plan takes into account multiple factors such as the demand for empty containers, backlog, and transportation costs to ensure the scientificity and feasibility of the plan. At the same time, the module also provides plan adjustment and optimization functions to facilitate shipping companies to make adjustments based on actual conditions.

[0062] Path planning module: This module uses advanced path planning algorithms to automatically generate the optimal empty container transportation path based on real-time conditions and optimization goals. Path planning takes into account multiple factors such as transportation distance, transportation time, and transportation cost to ensure the rationality and economy of the path. At the same time, the module also provides path adjustment and optimization functions, which facilitates shipping companies to make flexible adjustments based on actual conditions.

[0063] Abnormal detection and early warning module: This module monitors the empty container flow in real time and promptly detects potential risks and abnormalities by comparing actual data with predicted data. Once an abnormality is found, the module will immediately trigger the early warning mechanism and notify relevant personnel to handle it. At the same time, the module also provides abnormal analysis and handling suggestions to help shipping companies deal with abnormal situations quickly and effectively.

[0064] The application integration layer is responsible for integrating this system with the existing information system of the shipping company to realize data sharing and exchange. At the same time, this layer also communicates with the port automation equipment to realize the automatic control of empty container loading and unloading and storage. The application integration layer includes:

[0065] System interface module: This module provides a standardized API interface for integration with the existing information system of the shipping company. Through the interface, the system can obtain the business data of the shipping company in real time and push the processed data to the shipping company's information system. This realizes data sharing and exchange, and improves the practicality and convenience of the system.

[0066] Automation control module: This module communicates with the port automation equipment to realize the automation control of empty container loading and unloading and storage. Through the control instructions of the module, the automation equipment can automatically complete the loading and unloading, storage and other operations of empty containers. It improves the operational efficiency and safety of the port and reduces labor costs.

[0067] The user interface layer provides a real-time monitoring interface, data analysis reports and user interaction modules, which facilitates shipping companies to grasp the empty container flow situation in real time, analyze data and adjust optimization strategies. The user interface layer includes:

[0068] Real-time monitoring interface: This interface provides a real-time monitoring view of the empty container flow, including the location, status, transportation route and other information of the empty containers. Through the interface, the shipping company can grasp the flow of empty containers in real time and promptly discover and handle abnormal situations.

[0069] Data analysis report: This module generates a detailed data analysis report based on the processed data. The report includes the analysis results of key indicators such as empty container demand, backlog, and turnover speed, as well as statistics and analysis of abnormal situations. Through the report, shipping companies can gain an in-depth understanding of the actual situation of empty container management and provide a basis for formulating optimization strategies.

[0070] User interaction module: This module provides a user interface for querying empty container information and adjusting optimization strategies. Through the interface, shipping companies can easily query information such as historical data, current status and forecast results of empty containers. At the same time, they can also adjust optimization strategies according to actual conditions, such as adjusting empty container allocation plans and transportation routes, etc. This improves the flexibility and practicality of the system.

[0071] The present invention will be further described below in conjunction with Examples 1 to 3:

[0072] Embodiment 1:

[0073] The data preprocessing module processes data in the following ways:

[0074] Remove noisy data:

[0075] Noise data, that is, outliers in the data set, may interfere with subsequent data analysis and model training, and even lead to erroneous conclusions. To eliminate these noise data, this module adopts a threshold-based judgment method. The specific operation is to first set a reasonable threshold range for each field in the data set according to the actual business scenario and data distribution characteristics. Then, traverse each record in the data set and perform threshold verification on the value of each field. Once it is found that the value of a field exceeds the preset threshold range, the record is determined to be noise data and is processed accordingly. The processing methods include directly deleting the record or marking it as an exception for subsequent special processing. This method can effectively remove noise data in the data set, thereby improving the accuracy and reliability of the data.

[0076] Remove duplicates:

[0077] During the data collection process, duplicate records may appear in the data set due to factors such as equipment failure and network delay. These duplicate records not only increase the burden of data processing, but may also interfere with subsequent data analysis and model training. To solve this problem, this module introduces a hash table algorithm for data deduplication. The hash table algorithm is an efficient means of data deduplication. It maps each record to a unique hash value and uses the uniqueness of the hash value to identify duplicate records. In specific implementation, each record in the data set is traversed, its hash value is calculated, and the hash value is stored in the hash table. If the hash value of a record already exists in the hash table, the record is determined to be a duplicate and deduplication is performed. The processing methods include directly deleting the record, or retaining one and recording the number of repetitions. This method can quickly remove duplicate items in the data set and ensure the uniqueness and consistency of the data.

[0078] Remove invalid records:

[0079] There may be invalid records in the data set due to collection errors or omissions. The key fields of these records may be missing or the field values ​​may not conform to the expected format. To eliminate these invalid records, this module filters and deletes them according to preset rules. The specific operation is to first define a set of expected formats of key fields and field values. These key fields and formats are carefully designed based on actual business scenarios and data requirements. Then, traverse each record in the data set and check each key field. Once it is found that the key field is missing or the field value does not conform to the expected format, the record is judged to be invalid and filtered and deleted. This method can effectively remove invalid records in the data set, improve the integrity and validity of the data, and provide more accurate and reliable data support for subsequent data analysis and model training.

[0080] In the data preprocessing module of the present invention, the Z-score standardization method is used to ensure the comparability of numerical data and the accuracy of model training. This method performs specific processing on each numerical data to conform to the standard normal distribution, thereby significantly improving the data processing efficiency and the generalization ability of the model. The following are the specific implementation steps of the standardization process:

[0081] For each numeric field in the data set, the mean (μ) and standard deviation (σ) of the original data need to be calculated. The mean is the sum of all values ​​divided by the number of values, reflecting the center location of the data; the standard deviation is the square root of the average of the sum of the squares of the differences between each value and the mean, measuring the degree of data dispersion.

[0082] Each value is converted using the Z-score standardization formula. The formula is: Among them, X is the original data, μ is the mean of the original data, σ is the standard deviation of the original data, and Z is the standardized data. This formula converts the original data into standard normal distribution data with a mean of 0 and a standard deviation of 1 by subtracting the mean and dividing by the standard deviation.

[0083] During the standardization process, each record in the data set is traversed and the above formula is applied to each numerical field for conversion. The converted data maintains the relative size relationship of the original data, while making the data between different fields comparable, which facilitates subsequent data analysis and model training.

[0084] Embodiment 2:

[0085] The steps of acquiring the empty container backlog prediction model of the present invention include:

[0086] Step 1: Data Collection

[0087] First, we collect historical data on empty container turnover from multiple shipping companies. These data cover the location and status of empty containers (such as available, in use, under repair, etc.), transportation time, port operation status (such as loading and unloading, storage, waiting for scheduling, etc.), ship dynamics (such as arrival, departure, sailing, etc.), weather conditions (such as wind, rainfall, visibility, etc.) and corresponding empty container backlog labels. The empty container backlog label is determined based on the number of empty containers in the historical records and the port's handling capacity, and is used to identify whether there is an empty container backlog in a certain period of time.

[0088] Step 2: Data preprocessing and segmentation

[0089] The collected data is processed by the data preprocessing module and feature extraction module to form a regular data set. Data preprocessing includes removing noise data, filling missing values, and handling outliers to ensure the accuracy and consistency of the data. Feature extraction is to extract features useful for empty container backlog prediction from the original data, such as empty container turnover rate, port operation efficiency, weather impact index, etc.

[0090] The processed data set is divided into training set, validation set and test set. The training set is used for model training, accounting for 70% of the total data set; the validation set is used for model effect verification and optimization, accounting for 15% of the total data set; the test set is used to verify the generalization ability of the model on unknown data sets, accounting for 15% of the total data set.

[0091] Step 3: Model construction and initialization

[0092] Select the linear regression model to build an empty container backlog prediction model. The algorithm formula of the linear regression model is:

[0093] y=β 0 +β 1 x 1+β 2 x 2 +...+β n x n +∈

[0094] Among them, y is the target variable, representing the empty container backlog, x 1 、x 2 , ..., x n is the input variable, representing the characteristic parameter that affects the empty container backlog; β 1 , β 2 , ..., β n is the regression coefficient, determined by model training; β 0 is the intercept term, which represents the basic value of the empty container backlog; ∈ is the error term; the purpose of model training is to estimate the regression coefficient β 0 , β 1 , β 1 , β 2 , ..., β n , so that the error between the predicted value and the true value is minimized.

[0095] Step 4: Model training and optimization

[0096] The model training process includes:

[0097] S1. Initialize model parameters: Before training begins, initialize each model parameter (including the intercept term β 0 and the regression coefficient β 1 , β 1 , β 2 , ..., β n ) is assigned a random initial value close to 0. These initial values ​​will serve as the starting point for the gradient descent algorithm.

[0098] S2. Calculate the predicted value: In each iteration, the predicted value is calculated according to the linear regression formula using the current model parameters and the input data in the training set.

[0099] S3. Calculate the loss function value: Next, calculate the loss function value between the true value and the predicted value. In linear regression, the commonly used loss function is the mean squared error (MSE), and its calculation formula is:

[0100]

[0101] Where m is the number of samples in the training set, y i is the true value of the i-th sample, y hat i is the model's predicted value for the i-th sample.

[0102] S4, calculate gradient and update parameters: Calculate the gradient of the loss function to the current model parameters. The gradient indicates the direction and magnitude of the model parameters that should be adjusted in order to reduce the value of the loss function. Based on this gradient information, the gradient descent algorithm is used to update the value of the model parameters. Specifically, for each parameter β j (including the intercept term β 0 ), and its update formula is as follows:

[0103]

[0104] Among them, η is the learning rate, which controls the step size of parameter update; is the loss function for the parameter β j The partial derivative of , that is, the gradient.

[0105] S5, iterative optimization: Repeat the above steps and continuously iterate and update the model parameters until the loss function converges to a stable minimum value or reaches the preset number of iterations. Through this process, the parameters of the linear regression model are gradually optimized to improve the prediction accuracy of the model.

[0106] Step 5: Save and apply the model

[0107] The trained and verified empty container backlog prediction model is saved for subsequent use in the intelligent decision-making layer to predict the empty container backlog in real time. The model can quickly and accurately predict the empty container backlog in the future based on the input empty container flow data.

[0108] Embodiment 3:

[0109] The intelligent decision-making module is used to solve the problem of empty container backlog at ports. It generates the optimal empty container allocation plan by predicting the future empty container backlog and combining the shipping company's operating strategy and cost considerations.

[0110] The intelligent decision-making module receives the future empty container backlog forecast results from the empty container backlog forecast model. For example, the forecast results show that in the next three months, the empty container backlog of Port A will continue to increase, while the empty container demand of Port B and Port C will increase and decrease respectively.

[0111] According to the prediction results and the shipping company's operating strategy, the intelligent decision-making module sets the goal of empty container allocation to minimize the empty container backlog at Port A, meet the empty container demand at Port B, and minimize transportation costs. Constraints include: the number of available empty containers at Port A, the empty container handling capacity of Port B and Port C, the availability of transportation resources, and the total transportation cost budget.

[0112] The intelligent decision-making module builds a mathematical optimization model. Decision variables x ij represents the number of empty containers transported from port i to port j. Transportation cost c ijDetermined based on actual transportation distance, transportation method and transportation rate. Forecast empty container demand d j Determined according to the results of the empty container backlog prediction model, and a certain deviation range δ is allowed j . Available empty boxes i The total transportation cost budget C is determined based on the actual inventory at the port.

[0113] The specific optimization algorithm formula is as follows:

[0114] Minimize∑ ij c ij *x ij

[0115] The constraints are:

[0116] The empty container transfer volume cannot exceed the number of available empty containers at the port:

[0117] The number of empty containers transferred should meet the port's forecast demand: Among them, δ represents the error range of the operation;

[0118] The total shipping cost cannot exceed the budget:∑ ij c ij * ij ≤C;

[0119] The intelligent decision-making module selects a suitable optimization algorithm to solve the mathematical optimization model. In this embodiment, a linear programming algorithm is used to solve the problem. Through the linear programming algorithm, an empty container allocation plan that meets the constraints and minimizes the total transportation cost can be obtained.

[0120] Finally, the intelligent decision-making module conducts feasibility checks and risk assessments on the generated empty container allocation plan. It checks whether the plan is consistent with the actual transportation resources and port operation capabilities, and evaluates the risks and uncertainties that may be encountered during the implementation of the plan. If the plan is feasible and the risk is controllable, it will be output to the relevant execution department or system; if the plan is not feasible or the risk is high, the plan will be adjusted and optimized until a satisfactory solution is obtained.

[0121] After the intelligent decision-making module has generated an empty container allocation plan, the route planning module will further refine the planning process of the transportation route. The following is a detailed description of the process and the functions of the anomaly detection and warning module:

[0122] The path planning module refines the transportation path:

[0123] ①Build a network diagram:

[0124] According to the starting port, the end port and the corresponding transportation volume in the empty container allocation plan, the path planning module first constructs a network diagram containing all possible transportation paths. In this network diagram, each node represents a port, and the edges between nodes represent the routes between ports. In order to fully reflect the actual situation of the route, the weight of the edge is determined after comprehensively considering multiple factors such as the transportation cost, sailing time and safety of the route.

[0125] ② Intelligent optimization algorithm searches for the optimal path:

[0126] Based on the constructed network graph, the path planning module uses intelligent optimization algorithms (such as Dijkstra algorithm, A* algorithm or genetic algorithm, etc.) to search for the optimal path from the starting port to the end port. This optimal path is obtained after comprehensively considering multiple goals such as cost, time and safety while meeting transportation needs. Through the search of intelligent optimization algorithms, it can be ensured that the selected path is optimal or approximately optimal overall.

[0127] Abnormal detection and early warning module functions:

[0128] ①Data monitoring:

[0129] The anomaly detection and early warning module monitors various data in the empty container allocation process in real time, including but not limited to the location, status, transportation time and port operation status of the empty containers. These data are obtained through IoT technology, sensors or manual input.

[0130] ②Abnormal analysis:

[0131] The module will conduct in-depth analysis of the monitored data and detect whether there are any abnormalities by comparing the current data with historical data or the preset normal range. For example, if the backlog of empty containers at a port suddenly increases, or the transportation time of a certain route is significantly extended, these may be considered abnormal situations.

[0132] ③ Early warning signal is issued:

[0133] Once an abnormal situation is detected, the anomaly detection and warning module will immediately issue an early warning signal. This early warning signal can be conveyed to relevant personnel in a variety of ways, such as email, SMS or pop-up window prompts within the system. The early warning signal contains detailed information about the abnormality and recommended countermeasures so that relevant personnel can respond quickly and handle the abnormal situation.

[0134] The system interface module consists of two parts: the internal interface unit and the external interface unit, which are responsible for communication and data exchange inside and outside the system respectively.

[0135] The main function of the internal interface unit is to ensure smooth communication and data exchange between modules in the system. For example, when the intelligent decision-making module generates an empty container allocation plan, it will pass this plan to the path planning module through the internal interface unit so that the latter can carry out detailed planning of the transportation path. Similarly, the automation control module will also receive the decision results of the intelligent decision-making module through the internal interface unit and design the control logic accordingly. The internal interface unit adopts an efficient data transmission protocol to ensure the fast and accurate transmission of data within the system.

[0136] The external interface unit is responsible for the interaction between the system and the external data source. For example, the system may need to obtain real-time weather information, port operation status or ship dynamics from external data sources, which is crucial for the formulation of empty container allocation plans. The external interface unit establishes a connection with the external data source through a standard communication protocol (such as HTTP, FTP or WebSocket, etc.) and obtains the required data regularly or on demand. At the same time, the external interface unit can also send the system's processing results or status information to external systems or users to achieve information sharing and collaboration.

[0137] The automation control module designs and implements the corresponding control logic based on the decision results output by the intelligent decision module. It includes:

[0138] Control logic design: The automation control module first determines the control target based on the decision results, for example, reducing empty container backlogs, improving transportation efficiency, or reducing transportation costs. Then, it formulates specific control strategies, such as optimizing transportation routes, adjusting ship schedules, or increasing loading and unloading resources. Finally, it sets control parameters, such as transportation time, loading and unloading efficiency, or cost budget, to ensure the effective implementation of the control strategy.

[0139] Equipment connection and communication: The automation control module connects and communicates with various devices through standard communication protocols (such as Modbus, OPCUA or MQTT, etc.). These devices include ships, loading and unloading machinery, transport vehicles, etc., which are key execution units in the empty container allocation process. The automation control module sends control instructions to these devices, such as start, stop, accelerate or decelerate, etc., to achieve remote control and monitoring of them. At the same time, it also receives equipment status feedback, such as current location, working status or fault information, so as to adjust the control strategy or handle faults in time.

[0140] Equipment operating status monitoring: In addition to sending control instructions and receiving status feedback, the automation control module also continuously monitors the operating status of the equipment. It evaluates the performance and service life of the equipment by analyzing the equipment's working data, operation logs or fault records, and provides decision support for equipment maintenance and care. This helps ensure stable operation of the equipment and extend its service life, thereby reducing operation and maintenance costs and safety risks during empty container allocation.

[0141] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0142] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A shipping company empty container management and control system, characterized by: It includes data collection layer, data processing layer, machine learning layer, decision optimization layer, application integration layer and user interface layer; The data acquisition layer includes an IoT device interface and an external data source interface; The IoT device interface is used to connect and collect data from container tracking equipment, port automation equipment and ship dynamic systems; the external data source interface is used to obtain data from weather stations, port databases and shipping company ERP systems; The data processing layer includes a data preprocessing module, a feature extraction module and a data storage module; the data preprocessing module is used to clean the collected raw data, including removing noise, duplicates and invalid data, and data standardization; the feature extraction module is used to extract features related to empty box management from the cleaned data; the data storage module is used to store the processed data in a database; The machine learning layer includes a model training module, a model optimization module and a model deployment module; the model training module uses known historical data of empty box flow to train an empty box backlog prediction model, and the output of the empty box backlog prediction model is the prediction result of the future state of the empty box; the model optimization module is used to evaluate and optimize the trained model; the model deployment module is used to deploy the optimized model to the production environment; The decision optimization layer includes an intelligent decision-making module, a path planning module, and an anomaly detection and early warning module; The intelligent decision-making module is used to automatically generate an empty container allocation plan based on the prediction results of the empty container backlog prediction model; The path planning module is used to provide the optimal empty container transportation path planning; the anomaly detection and early warning module is used to monitor the empty container flow in real time and identify potential risks; The application integration layer includes a system interface module and an automation control module; The system interface module is used to provide a standardized API interface for integration with the existing information system of the shipping company; the automation control module is used to communicate with the port automation equipment to achieve automated control of empty container loading and unloading and storage; The user interface layer includes a real-time monitoring interface, a data analysis report and a user interaction module; The real-time monitoring interface is used to provide a real-time monitoring view of the empty container flow; The data analysis report is used to generate a detailed data analysis report; The user interaction module is used to provide a user interface for querying empty box information and adjusting optimization strategies.

2. A shipping company empty container management and control system according to claim 1, characterized in that: The implementation of the data preprocessing module includes: Remove noise data: Use a threshold-based method to identify and process outlier data in the data set; Remove duplicates: Use hash table algorithm to remove duplicate records in the data set; Remove invalid records: Filter and delete invalid records in the dataset based on predefined rules for missing key fields and field values ​​that do not conform to the expected format.

3. A shipping company empty container management and control system according to claim 2, characterized in that: The data preprocessing module adopts the Z-score standardization method, subtracting the mean value from each numerical data and dividing it by its standard deviation, so that the processed data conforms to the standard normal distribution. The formula for standardization is: Among them, X is the original data, μ is the mean of the original data, σ is the standard deviation of the original data, and Z is the standardized data.

4. The empty container management and control system of a shipping company according to claim 1, characterized in that: The steps of obtaining the empty container backlog prediction model include: Step 1: Collect the historical data of empty container circulation of shipping companies, including empty container location, status, transportation time, port operation status, ship dynamics, weather conditions and corresponding empty container backlog labels; Step 2: Process the collected data through the data preprocessing module and feature extraction module to form a regular data set; further divide the data set into training set, validation set and test set, where the training set is used for model training, the validation set is used to verify and optimize the model effect, and the test set is used to verify the generalization ability of the model on unknown data sets; Step 3: Select the linear regression model to build an empty container backlog prediction model. The algorithm formula of the linear regression model is: y=β0+β1x1+β2x2+...+β n x n +∈ Among them, y is the target variable, representing the empty container backlog, x1, x2, ..., x n is the input variable, representing the characteristic parameters that affect the empty container backlog; β1, β2, ..., β n is the regression coefficient, which is determined by model training; β0 is the intercept term, which represents the basic value of the empty container backlog; ∈ is the error term; the purpose of model training is to estimate the regression coefficients β0, β1, β1, β2, ..., β n , so that the error between the predicted value and the true value is minimized; when initializing the model parameters, for each feature x in the linear regression model i The corresponding parameter β i Assign an initial value, which is set to a random number close to 0, and also assign an initial value to the intercept term β0; these initial parameter values ​​will serve as the starting point for model training and will be continuously optimized and adjusted through subsequent training processes; Step 4: Use the training set to train the empty box backlog prediction model, and continuously iterate and update the parameters of the linear regression model through the gradient descent algorithm to minimize the loss function value between the true value and the predicted value; in each iteration, adjust the value of the model parameter according to the gradient information of the loss function to the model parameter until the loss function converges or reaches the preset number of iterations; use the validation set to verify the trained model, and adjust the model parameters or structure according to the verification results; use the test set to comprehensively evaluate the optimized model through the set evaluation indicators to ensure that the model performance meets the expected standards; Step 5: Save the trained and verified empty container backlog prediction model for subsequent use in the intelligent decision-making layer to predict the empty container backlog in real time.

5. The empty container management and control system of a shipping company according to claim 1, characterized in that: The intelligent decision-making model receives the future empty container backlog prediction result output by the empty container backlog prediction model, which includes the predicted empty container backlog amount at different time points and the possible backlog trend; According to the prediction results, combined with the shipping company's operating strategy and cost considerations, the empty container allocation goals and constraints are set; the empty container allocation goals and constraints are converted into a mathematical optimization model, specifically: Set the decision variable x ij represents the number of empty containers transferred from port i to port j, c ij represents the transportation cost from port i to port j, d j represents the predicted empty container demand at port j, s i represents the number of available empty containers at port i, and C represents the total transportation cost budget. The optimization problem can be expressed as: Minimize∑ ij c ij *x ij The constraints are: The empty container transfer volume cannot exceed the number of available empty containers at the port: The number of empty containers transferred should meet the port's forecast demand: Among them, δ j Indicates the error range of the operation; The total shipping cost cannot exceed the budget: Σ ij c ij *x ij ≤C; A mathematical optimization model is constructed through the above algorithm formula to solve the optimal solution for empty container allocation; Based on the solution results of the mathematical optimization model, an empty container allocation plan is generated; the plan includes a specific empty container transportation plan, transportation route, time schedule and related cost budget.

6. A shipping company empty container management and control system according to claim 5, characterized in that: Based on the empty container allocation plan generated by the intelligent decision-making module, the path planning module further refines the planning of the transportation path; according to the starting port, the end port and the transportation volume in the empty container allocation plan, the path planning module constructs a network diagram containing all possible transportation paths; the nodes in the network diagram represent ports, the edges represent the routes between ports, and the weights of the edges are determined comprehensively based on the transportation cost, sailing time and safety of the routes; The path planning module uses an intelligent optimization algorithm to search for the optimal path from the starting port to the destination port in the network graph.

7. The empty container management and control system of a shipping company according to claim 1, characterized in that: The anomaly detection and early warning module continuously monitors and analyzes real-time data during the empty container allocation process to detect any abnormal situation that deviates from the normal mode and issue an early warning signal in a timely manner.

8. The empty container management and control system of a shipping company according to claim 1, characterized in that: The system interface module includes an internal interface unit and an external interface unit. The internal interface unit is used to realize communication and data exchange between modules in the system, and the external interface unit is used for interaction between the system and an external data source.

9. The empty container management and control system of a shipping company according to claim 1, characterized in that: The automation control module designs corresponding control logic based on the decision results output by the intelligent decision-making module, including determining control objectives, formulating control strategies and setting control parameters, and further connects and communicates with various devices through standard communication protocols, including sending control instructions, receiving device status feedback, and monitoring device operating status.

Images