A method for predicting the transport of ships and vehicles based on the reserves of dangerous chemicals in the port

Abstract

The application discloses a kind of transport ships and vehicle scheduling method based on port dangerous chemical reserves prediction, it is related to dangerous chemical port storage and transportation safety management field, constructs dangerous chemical port information data resource base;Based on three kinds of machine learning model, a set of prediction models are constructed, the MAPE index is used to select the optimal model to predict the dangerous chemical reserves and the number of ships and vehicles arriving at the port in each time period in the future;Then a dynamic scheduling model is constructed in time periods, the operating mode is automatically determined according to the interval of the storage tank liquid level;Establish a reservation type arrival coordination mechanism, the port publishes the recommended arrival time of each ship and vehicle in advance, and orderly operation is realized.By setting the lower limit of loading and the upper limit of unloading as the intermediate threshold, combined with the dynamic prediction of the storage tank liquid level and the time period scheduling model, the risk of equipment damage, leakage or overflow can be effectively avoided before the storage tank reserves approach the critical threshold, and the safety and efficiency of dangerous chemical port storage and transportation are improved.

Description

Technical Field

[0001] This invention relates to the field of safety management of hazardous chemical port storage and transportation, and in particular to a method for scheduling transport vessels and vehicles based on the prediction of hazardous chemical reserves in ports. Background Technology

[0002] Currently, the scheduling of hazardous chemical port storage and transportation relies primarily on the experience of port personnel or on pre-established fixed schedules and procedures for arranging the loading and unloading of hazardous chemicals by ships and vehicles. This scheduling model lacks forward-looking forecasting and analysis of the port's hazardous chemical storage capacity, simply adopting a first-come, first-served approach. It fails to dynamically adjust operational plans based on the actual port operating conditions and storage tank capacity. This leads to a surge in the number of arriving ships and vehicles during peak port periods, causing existing ship and vehicle scheduling to be unable to keep up with actual demand. This results in prolonged waiting times for ships and vehicles at the port, severely reducing transportation efficiency, increasing logistics costs, and, more importantly, potentially creating a series of safety hazards that threaten port safety.

[0003] Due to the special nature of the goods stored and transported at hazardous chemical ports, ensuring the safety of hazardous chemical storage and transportation is of paramount importance. Liquid level management in port hazardous chemical storage tanks is a core element in ensuring safety and operational efficiency, with low-low and high-high liquid levels being two crucial thresholds. Low-low liquid level refers to the amount of hazardous chemicals stored in the tank falling below the safety lower limit, which may lead to pump dry running, evaporation of pipeline residues, equipment damage, or even leakage risks due to pressure imbalance. High-high liquid level refers to the tank approaching its maximum safe capacity; exceeding this level may cause serious accidents such as overflow, tank overpressure, or environmental pollution. Traditional port scheduling methods, lacking real-time monitoring and scientific forecasting of hazardous chemical storage levels, struggle to anticipate and adjust tank levels before they approach critical thresholds. This makes it difficult to effectively avoid various safety risks caused by abnormal fluctuations in storage levels and also increases logistics costs and management burdens due to hazardous chemical congestion.

[0004] To address the aforementioned technical challenges, there is an urgent need to research and develop a technical method capable of real-time monitoring and dynamic prediction of hazardous chemical storage levels at ports. By precisely controlling tank levels and planning ship arrival replenishment and vehicle transportation in advance, the risks of equipment damage, leakage, or overflow caused by abnormal hazardous chemical storage tank levels at ports can be effectively mitigated. This will ensure the safety of hazardous chemical storage and transportation at ports while improving port operational efficiency and reducing logistics costs. Summary of the Invention

[0005] To overcome the shortcomings of existing technologies, the purpose of this invention is to provide a method for scheduling transport vessels and vehicles based on the prediction of hazardous chemical storage in ports. This method uses a machine learning model to dynamically predict the storage of hazardous chemicals and automatically determines the operating mode based on the liquid level range of the storage tanks. This enables the rational allocation and orderly operation of transport resources, thereby improving port operating efficiency while ensuring the safety of hazardous chemical storage and transportation in ports.

[0006] To address the problems of the prior art, the technical solution of the present invention is as follows: A method for scheduling transport vessels and vehicles based on port hazardous chemical storage capacity prediction includes the following steps: S1. Construct a hazardous chemical port information data resource database: Obtain daily port time data, storage data, ship data, vehicle data, average ship unloading volume and average vehicle loading volume for various types of hazardous chemicals arriving at the port. The obtained data is preprocessed to obtain usable multi-source data. The usable multi-source data for several days constitutes the hazardous chemical port information data resource database. Furthermore, data on hazardous chemical port times, storage volumes of each hazardous chemical, transport vessels, and vehicles are acquired. The acquired port time data per working day is divided into multiple time units, based on the working hours of port personnel. Flow meters and tank gauging are used to measure the storage volume of each hazardous chemical. Port terminal monitoring and radar monitoring systems record the arrival and departure times, trajectories, and loading status of each vessel in real time, with surveillance cameras ensuring comprehensive coverage of the terminal. Barrier gate recording devices are used to record the license plate numbers, vehicle information, hazardous chemical information carried, and port entry / exit times of arriving vehicles. These barrier gate devices are located at port entrances and exits, and relevant information is collected by cameras adjacent to the gates. Simultaneously, the port's back-end system continuously monitors the safety threshold parameters of each hazardous chemical storage tank: storage volumes corresponding to low-low liquid levels, storage volumes corresponding to high-high liquid levels, the minimum allowable storage volume for loading, and the maximum allowable storage volume for unloading. After data preprocessing, usable data is obtained, including the numerical changes of the storage of various hazardous chemicals in different time periods of each working day, the number and time of arrival of ships and vehicles in each time period, the information and quantity of hazardous chemicals carried or required by ships and vehicles, and the average daily cargo load of various types of hazardous chemicals on ships and the average cargo loading of vehicles.

[0007] S2. Hazardous chemical storage prediction based on multi-source data: Based on multi-source data from the hazardous chemical port information database, LSTM, LightGBM and Prophet models are constructed. The MAPE (Mean Absolute Percentage Error) is used as the model evaluation index, and the model with the smallest MAPE is selected as the final prediction method to predict the hazardous chemical storage data for the next day and the information data of the expected arriving transporters.

[0008] Furthermore, the system uses arriving vessels, the quantity of each hazardous chemical in the current hazardous chemical storage tanks, the number of vehicles, the arrival time of vessels and vehicles, the quantity of each hazardous chemical transported by vessels, and the quantity of hazardous chemicals required by vehicles as characteristic data. Then, it determines the prediction method, forming a set of measurement method models [LSTM, LightGBM, Prophet] based on machine learning. LSTM excels at processing time-series data, accurately predicting storage capacity to aid in scheduling decisions; LightGBM has fast training speed and low memory usage, efficiently supporting real-time scheduling calculations; and Prophet automatically captures data trends and cycles, accurately predicting storage capacity to assist in scheduling. Next, it determines the evaluation index, setting MAPE as the evaluation index for the method model set. The method that calculates the minimum MAPE value (min) in the measurement method model set is selected as the final prediction method. The model predicts hazardous chemical storage data and information on expected arriving transporters. The hazardous chemical storage data refers to the hazardous chemical storage volume before operations begin the next day; the information on expected arriving transporters includes the number of vessels and vehicles arriving in different time periods of the next day, as well as the average cargo load of various types of hazardous chemical vessels and the average cargo loading of vehicles for the next day.

[0009] S3. Construct a time-segmented dynamic scheduling model: S31. Select a prediction model and prediction method. Based on the predicted data of hazardous chemical reserves for the next day and the information of the expected arriving transporters, divide the day into K equal-length time periods. S32. Based on the hazardous chemical storage situation, calculate the number of ships and vehicles that can operate in each time period; S33. During the k-th time period, k takes values ​​in the range [1, K]. Calculate the vessels and vehicles that can be operated and suggest the start time for operation. S34. Remaining time rescheduling of transportation resources and appointment-based arrival coordination methods.

[0010] Furthermore, based on the predicted hazardous chemical storage volume for the next day and the information on expected arriving transporters, this invention constructs a time-segmented dynamic scheduling model. This model divides a day into multiple equal-length time periods. For each time period, the operational mode (priority unloading, priority loading, or simultaneous operation) is determined based on the hazardous chemical storage volume at the start of that period (priority unloading zone between low-low liquid levels and the lower limit of loading allowance, priority loading zone between the upper limit of unloading allowance and high-high liquid levels, and the parallel operation zone between the two). Based on this, and considering the number of ships and vehicles waiting for operation in each time period, the unloading and loading speeds, and the length of the divided time, the actual number of ships and vehicles available for operation in each time period is calculated, and transport resources that cannot be scheduled are automatically carried over to the next time period. Through iterative calculations across time periods, the suggested start time and estimated completion time for all ships and vehicles are finally obtained. For transportation resources that cannot operate during the current time slot due to storage limitations or insufficient time, the system prioritizes their placement in the waiting queue for subsequent time slots. If resources remain after all time slots have ended for the day, they are rescheduled using the remaining available time in each time slot, and the impact on subsequent time slots is recalculated to select the option with the fewest remaining resources. If there are still remaining vessels or vehicles, they are scheduled to operate in the first time slot of the following day. Through this dynamic scheduling, the system enables scheduled arrival and orderly operation of transportation resources, improving port operational efficiency and safety.

[0011] Compared with the prior art, the advantages of the present invention are as follows: 1. This invention integrates and preprocesses multi-source data by constructing a hazardous chemical port information data resource database. It cleans and processes data acquired from various data acquisition devices such as port terminal monitoring systems, radar monitoring systems, gate recording equipment, flow meters, and tank gauging devices, providing a complete data foundation for scheduling decisions.

[0012] 2. This invention constructs a model set containing three machine learning models—LSTM, LightGBM, and Prophet—through a multi-model prediction mechanism. The MAPE index is used to evaluate and compare the three models, and the model with the smallest MAPE is selected as the final prediction method. Different prediction strategies are adopted according to the uniform distribution characteristics of the data, which effectively improves the accuracy of hazardous chemical reserve prediction and provides a scientific basis for scheduling decisions.

[0013] 3. This invention uses a time-segmented dynamic scheduling model to automatically determine the operation mode based on the range of the tank liquid level. When the storage volume is in the priority unloading zone, the unloading operation is arranged first; when the storage volume is in the priority loading zone, the loading operation is arranged first; and when the storage volume is in the parallel operation zone, the unloading and loading operations are arranged simultaneously, thereby realizing the rational allocation of transportation resources and maximizing the utilization of port operation capacity.

[0014] 4. This invention fully utilizes the remaining workable time in each time period through cross-time period iterative calculation and remaining resource rescheduling mechanism, and re-incorporates transportation resources that could not be arranged in time into the scheduling scheme, selects the scheme with the fewest remaining resources as the final scheme, maximizes the utilization of port operation capacity, and reduces resource idleness and waiting time for ships and vehicles.

[0015] 5. This invention utilizes a pre-arranged arrival coordination mechanism, whereby the port authority announces the suggested start times for vessels and vehicles in advance based on the final scheduling plan. Transport operators then rationally arrange their navigation plans and transport routes according to these suggested times, thereby achieving pre-arranged arrival and orderly operation of transport resources, reducing anchorage waiting time and port congestion, and improving port operational efficiency.

[0016] 6. This invention sets a lower limit for loading. Q low and the upper limit of unloading Q high As an intermediate threshold, ensuring that the tank level does not reach the low-low level. Q min and high-high liquid level Q max Under the premise of safety boundaries, scheduling is carried out by combining dynamic prediction of tank liquid level and time-sharing scheduling model. Predictions and adjustments are made before the storage capacity of tanks approaches the critical threshold, effectively avoiding the risk of equipment damage, leakage or overflow, and improving the safety of hazardous chemical port storage and transportation.

[0017] 7. This invention acquires hazardous chemical port time data, hazardous chemical storage volumes, transport vessel data, and vehicle data. Then, through multi-source data preprocessing and integration, it uses a machine learning model to predict hazardous chemical storage volumes for different time periods of the next day, as well as the expected number of arriving vessels and vehicles. It calculates the input completion time for each transport vessel and the output completion time for each vehicle. Based on real-time hazardous chemical storage volume predictions, it quickly generates optimal vessel and vehicle allocation schemes to guide daily port operations and improve the safety, efficiency, and economy of hazardous chemical transportation and storage at the port. Attached Figure Description

[0018] Figure 1 This is a flowchart of the overall process of the method of the present invention.

[0019] Figure 2 This is a flowchart of the LSTM model prediction method of the present invention.

[0020] Figure 3 This is a flowchart of the LightGBM model prediction method of the present invention.

[0021] Figure 4 This is a flowchart of the Prophet model prediction method of the present invention.

[0022] Figure 5 A comparison chart of the prediction results of the LSTM model, LightGBM model, and Prophet model of this invention.

[0023] Figure 6 This is a line graph showing the predicted hazardous chemical reserves of the present invention when the data is uniform.

[0024] Figure 7 This is a line graph showing the predicted hazardous chemical reserves of the present invention when the data is uneven.

[0025] Figure 8 This is a schematic diagram of the hazardous chemical storage tank of the present invention.

[0026] Figure 9 This is a schematic diagram illustrating an application example of the present invention. Detailed Implementation

[0027] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0028] A method for scheduling transport vessels and vehicles based on port hazardous chemical storage capacity prediction includes the following steps: S1. Construct a hazardous chemical port information data resource database: Obtain daily port time data, storage data, ship data, vehicle data, average ship unloading volume and average vehicle loading volume for various types of hazardous chemicals arriving at the port. The obtained data is preprocessed to obtain usable multi-source data. The usable multi-source data for several days constitutes the hazardous chemical port information data resource database. S2. Hazardous chemical storage prediction based on multi-source data: Based on multi-source data from the hazardous chemical port information database, LSTM, LightGBM and Prophet models are constructed. The average absolute percentage error (MAPE) is used as the model evaluation index. The model with the smallest MAPE is selected as the final prediction method to predict the hazardous chemical storage data for the next day and the information data of the expected arriving transporters. S3. Construct a time-segmented dynamic scheduling model: S31. Select a prediction model and prediction method. Based on the predicted data of hazardous chemical reserves for the next day and the information of the expected arriving transporters, divide the day into K equal-length time periods. S32. Based on the hazardous chemical storage situation, calculate the number of ships and vehicles that can operate in each time period; S33. During the k-th time period, k takes values ​​in the range [1, K]. Calculate the vessels and vehicles that can be operated and suggest the start time for operation. S34. Remaining time rescheduling of transportation resources and appointment-based arrival coordination methods.

[0029] The specific process of this invention application is described in detail below: S1. Construct a hazardous chemical port information database that considers hazardous chemical reserves. Obtain hazardous chemical port time data, determine time data units, and divide the day into sections based on the work schedule of port staff. K The 24 hours of the day are divided into time units, that is, the entire day is divided into... K part.

[0030] When unloading refined oil or chemical products from ships, the ship's onboard pumps are used to pressurize and transport them into tanks. The fixed unloading speed for hazardous chemicals at ports is [missing information]. v 1, v The value of 1 is typically 100 m³ / h to 200 m³ / h (the specific value depends on the actual situation at the port). Measurement is performed by gauging the storage tanks, and the change in hazardous chemical volume is measured as follows: Q c When hazardous chemicals are loaded onto hazardous chemical vehicles, they are transported from storage tanks to the loading arms on the loading platform via pipelines. The loading speed of hazardous chemicals is fixed. v 2, v The value of 2 is typically 60 m³ / h to 120 m³ / h (the specific value depends on the actual situation of the port). A flow meter is used for measurement, and the change in the hazardous chemical flow rate is... Q d And under normal circumstances, the unloading speed is greater than the loading speed.

[0031] Ship unloading – The number of ships entering the port is recorded in real time using port terminal monitoring systems and radar monitoring systems. m、 The port entry time, departure time, route, and information on the hazardous chemicals carried by each vessel (including the type and quantity of hazardous chemicals) are monitored by surveillance cameras installed at key locations such as high points on the dock, at crossings, and around warehouses. Radar systems are placed at high points in the port or on specific monitoring towers to ensure that every corner of the dock is covered by surveillance.

[0032] Vehicle loading – The number of vehicles entering the port is recorded using a barrier gate recording device erected at the entrance and exit of the port storage yard in the form of a horizontal bar. n When a vehicle enters the port storage area, the gate recording device will automatically open and record the vehicle's license plate number, vehicle information, and information on the hazardous chemicals carried by the vehicle (including the type and quantity of hazardous chemicals required by the vehicle). When the vehicle begins loading, the monitoring equipment on the crane position will record the start time of the operation and the time when the vehicle completes the operation plan and prepares to leave the crane position.

[0033] After cleaning and processing the raw data, usable multi-source data is obtained: K Hazardous chemical storage values ​​for each time period Qi Number of ships arriving at port daily m and the number of vehicles arriving at the port n The arrival times of ships and vehicles, the types and quantities of hazardous chemicals carried by ships and vehicles, the types and quantities of hazardous chemicals required for vehicle loading, and the average daily unloading volume of various hazardous chemicals by ships and the average daily loading volume by vehicles; among which, Average unloading volume of ships (m³ / vessel): Calculated based on the ratio of total daily unloading volume of vessels to the number of vehicles arriving at the port; Average cargo volume per vehicle (m³ / vehicle): Calculated based on the ratio of the total daily cargo loading volume to the number of vehicles arriving at the port.

[0034] S2. Hazardous chemical reserve prediction and application process based on multi-source data Based on the available multi-source data, a set of measurement method models [LSTM, LightGBM, Prophet] is formed using machine learning. Figures 2-4 The flowchart illustrates the prediction method. LSTM excels at processing time-series data, accurately predicting reserves to aid in scheduling decisions; LightGBM offers fast training speed and low memory usage, efficiently supporting real-time scheduling calculations; and Prophet automatically captures data trends and cycles, accurately predicting reserves to assist in scheduling. Next, the evaluation metrics are determined. MAPE is set as the evaluation metric for the method model set. The method that calculates the minimum MAPE value in the model set is selected as the final prediction method. This predicts the hazardous chemical reserves for the next day and the information on expected arriving transporters. The hazardous chemical reserves data refers to the quantity (in m³) of hazardous chemicals in each storage tank within the port before port operations begin on the next day, i.e., before ships enter the port to unload and vehicles to load. The expected arriving transporter information includes the number of ships expected to arrive at the port at different times of the day. M (Unit: ships) and number of vehicles N (Unit: vehicles) and the average unloading volume of various hazardous chemicals by ships in the coming day. (m³ / ship) and average cargo volume per vehicle (m³ / vehicle).

[0035] (1) LSTM (Long Short-Term Memory) is a special type of recurrent neural network (RNN) mainly used for processing and predicting time series data. Based on the available data, the data is divided into training and testing sets according to a certain ratio. The training set is used for model training, and the testing set is used to evaluate model performance. Next, an LSTM model is built, consisting of an input layer, an LSTM layer, a fully connected layer, and an output layer. The input layer is used to input historical port data, including time-segmented data: the values ​​of hazardous chemicals in each hazardous chemical storage tank at each time segment of each day over the past several days. Q i Number of ships arriving at port m Ship unloading volume Q c Number of vehicles arriving at the port n Vehicle workload Q d Average daily unloading volume of various hazardous chemicals by ships and average cargo volume per vehicle The LSTM layer is used to analyze long-term patterns in the time series; the fully connected layer is used to transform the patterns extracted by the LSTM into prediction logic; and the output layer outputs data on hazardous chemical reserves and information on expected arriving transporters. Next, the model is trained using training data, comparing predicted data with actual values ​​and calculating the error. The model parameters are then adjusted based on the error to reduce prediction errors in subsequent predictions. The model learns repeatedly from the data until its predictions stabilize. Then, the model is evaluated using test data, calculating the model's performance index (MAPE). The prediction results are then visualized and analyzed, and the accuracy of the predictions is evaluated by comparing actual data with predicted reserves. Finally, the trained LSTM model is applied to predict hazardous chemical reserves and information on expected arriving transporters.

[0036] (2) LightGBM (Light Gradient - Boosting Machine) has a fast training speed and low memory consumption, and efficiently supports real-time scheduling. The usable data is divided into training and testing sets according to a certain ratio. The training set is used for model training, and the testing set is used to evaluate model performance. In the model building stage, feature engineering is first required to convert the port's historical data into a format suitable for tree model processing, including time-segmented features: the values ​​of hazardous chemical storage tanks for each time period of each day over the past several days. Q i Number of ships arriving at port m Ship unloading volume Q c Number of vehicles arriving at the port n Vehicle workload Qd Average daily unloading volume of various hazardous chemicals by ships and average cargo volume per vehicle The time-series features are then identified. Next, the LightGBM model is initialized, including a weak learner, and hyperparameters are set, such as the learning rate (controlling the step size of parameter updates in each iteration), the number of trees (determining model complexity and fitting ability), and the number of leaf nodes (limiting the complexity of each decision tree). Through combinations of hundreds of decision trees, all possible scene rules are covered. The model is then trained iteratively using the training set data. In each iteration, the error of the current model (the difference between the actual value and the current model's prediction) is calculated, and a new tree is created specifically to correct the error from the previous iteration. This process of generating new trees continues in iterative optimization until the error no longer decreases significantly. Finally, the model is evaluated and optimized using the test set data. Evaluation metrics (such as accuracy and mean squared error, MSE) are calculated. Based on the evaluation results, hyperparameters are adjusted, such as increasing the number of trees or adjusting the learning rate, to improve model performance. Finally, the trained LightGBM model is used to predict new data, and the prediction results of all decision trees are summed to obtain the final hazardous chemical storage prediction data and the information of the expected arriving transporters.

[0037] (3) Prophet automatically captures data trends and cycles to accurately predict storage capacity and assist in scheduling. Usable data is divided into training and testing sets according to a certain proportion. The training set is used for model training, and the testing set is used to evaluate model performance. The model building phase adopts a modular design approach, decomposing the core elements of the time series through an additive model (existing model). The input data consists of the historical operational time-series characteristics of the port, including: the numerical values ​​of hazardous chemical storage tanks recorded daily for different time periods over the past several days. Q i Number of ships arriving at port m Ship unloading volume Q c Number of vehicles arriving at the port n Vehicle workload Q d Average daily unloading volume of various hazardous chemicals by ships and average cargo volume per vehicle Key indicators such as...

[0038] Next, the model is trained using the training set data. The training process can involve both automatic fitting and manual intervention. Automatic fitting includes trend fitting: identifying overall upward or downward trends in storage (e.g., detecting a% monthly increase in storage in tank A); cycle capture: automatically detecting fixed fluctuation patterns daily, weekly, and yearly (e.g., a decrease of b% in storage at 10 AM every day); and event response: learning the impact of specific events (e.g., a decrease of c% in tank consumption during holidays). Manual intervention can involve adding port-specific events (e.g., "concentrated chemical shipments on the 25th of each month") or setting seasonal parameters (e.g., forcing the model to focus on "daily cycles" rather than annual cycles). The model is then evaluated and optimized. Test set data is used to evaluate the trained model, and evaluation metrics (such as Mean Absolute Percentage Error (MAPE) and Symmetric Root Mean Square Error (sRMSE)) are calculated, with a focus on analyzing prediction bias during peak periods. The model's prediction results are then visualized by overlaying predicted values ​​with actual data on time-series curves to verify the matching degree of trend inflection points, cyclical peaks, or troughs. Based on the evaluation results, the model is optimized by adjusting hyperparameters such as the Fourier series order and the prior scale of variable points through grid search to make the prediction patterns more realistic. Finally, the trained Prophet model is used to analyze hazardous chemical storage data for the next day. Q j The number of ships expected to arrive at port at different times throughout the next day M and the expected number of vehicles arriving in Hong Kong N Average cargo unloading volume of ships in the next day and average cargo volume per vehicle Make predictions.

[0039] (4) The best prediction method is selected by using the MAPE evaluation index. MAPE measures the average relative error between the predicted value and the actual value as a percentage, which intuitively reflects the accuracy of the prediction. It is not affected by the amount of data and is convenient for comparing the performance of models under different scales of data. The MAPE calculation formula is as follows: ; in, It is the first i The true value of each sample It is the first i The predicted value for each sample, x It refers to the number of samples.

[0040] The steps for calculating MAPE are as follows: ① Calculate the absolute value of the difference between the predicted value and the true value for each sample, and divide it by the true value to obtain the relative error for each sample.

[0041] ② Sum the relative errors of all samples.

[0042] ③ Divide the sum by the sample size. X Then multiply by 100% to get the MAPE value.

[0043] The smaller the MAPE value, the higher the accuracy of the prediction model. The MAPE values ​​were calculated for three model sets: LSTM, LightGBM, and Prophet. The method that calculates the minimum MAPE value (min) in the model set was selected as the final prediction method to predict the hazardous chemical storage data and the expected number of ships and vehicles arriving at port for different time periods of the next day.

[0044] S3. Construct a transportation resource allocation model based on the predicted values ​​of hazardous chemical reserves. S31. Selecting the prediction model and prediction method After the model is selected, when the predicted hazardous chemical data of the prediction model are uniform (e.g.) Figure 6 Taking hazardous chemical reserves forecasting as an example, the previous day's data can be used to forecast the hazardous chemical reserves for the following day, as well as the expected number of ships and vehicles arriving at the port; when the data is uneven (e.g. Figure 7 For example, the forecast of hazardous chemical reserves can be obtained by using last week's data to predict the hazardous chemical reserves for the following week, as well as the expected number of ships and vehicles arriving at the port.

[0045] Figure 5 The chart shows a comparison of prediction results from different models (taking hazardous chemical reserves prediction as an example). Based on the comparison of MAPE values, it is assumed that the LSTM machine learning model has a smaller MAPE value, meaning its prediction results are more accurate. Therefore, the LSTM machine learning model is selected for prediction.

[0046] S32. Based on the hazardous chemical storage situation, calculate the number of vessels and vehicles that can operate in each time period. When unloading refined oil or chemical products from ships, the ship's onboard pumps are used to pressurize and transport them into tanks. The fixed unloading speed for hazardous chemicals at ports is [missing information]. v 1, v The value of 1 is typically 100 m³ / h to 200 m³ / h (the specific value depends on the actual situation at the port). Measurement is performed by gauging the storage tanks, and the change in hazardous chemical volume is measured as follows: Q c When hazardous chemicals are loaded onto vehicles, they are transported from storage tanks to the loading arms on the loading platform via pipelines. The loading speed of hazardous chemicals is fixed. v 2, v The value of 2 is typically 60 m³ / h to 120 m³ / h (the specific value depends on the actual situation of the port). A flow meter is used for measurement, and the change in the hazardous chemical flow rate is... Q d And under normal circumstances, the unloading speed is greater than the loading speed.

[0047] Known unloading speed v 1. Loading speed v 2. The volume of hazardous chemicals in the storage tank at the high-high level is: Q max The amount of hazardous chemicals in the storage tank at the low-low level is: Q min Divide the next day into K There are several equal-length time intervals, each with a length of [length missing]. ΔT , ,like Figure 9 As shown, using the LSTM model, it is assumed that various... k ( k =1, 2, ..., K ) within different time periods M k Several ships arrived at the port. N k Two vehicles arrived at the port; before the ships and vehicles began operations, the amount of a certain hazardous chemical A in the port's storage tanks that day was [amount missing]. Q 0 and the average unloading volume of the ship on that day were The average cargo load of the vehicle is Port operation start time t =0 (hours), the number of ships that had not completed unloading in the previous time period. R 0 (vehicles), the number of vehicles that were not loaded in the previous period. S 0 (vehicles). The following lists each time period. k Perform the calculation: Step 1: Calculate the first... k Total number of ships and vehicles waiting within a given time period In the k The number of ships and vehicles waiting to operate within each time period are as follows: ; in, In the first k The number of ships waiting for operations within a given time period; In the first k The number of ships predicted by the model to arrive at the port within a given time period; In the first k - The number of vessels that could not be scheduled for operations within a given time period.

[0048] ; in, In the first k The number of vehicles waiting for work within a given time period; In the first kThe number of vehicles predicted by the model to arrive at the port within a given time period; In the first k - The number of vehicles that could not be scheduled for work within a time period.

[0049] Step 2: Calculate the first... k The maximum number of ships and vehicles that can operate within a given time period. Because the day is divided into K There are 1 time periods, each with a length of 1 / 2. ΔT Assuming ships and vehicles operate continuously, then in ΔT The maximum number of vessels that can operate within the specified time period is (rounded down to the nearest integer): ; The maximum number of vehicles that can operate is (rounded down to the nearest integer): ; Step 3: According to the first k Hazardous chemical storage levels before the start of operations in the specified time period Q k Within the specified range, determine the operating mode and calculate. k Number of vessels that can operate within a given time period and number of vehicles

[0050] like Figure 8 As shown, low-low liquid level Q min This refers to a situation where the level of hazardous chemicals in a storage tank is below the safety lower limit, which may lead to pump dry running, volatilization of pipeline residues, or equipment damage, and even leakage risks due to pressure imbalance; high-high liquid level. Q max This refers to a storage tank approaching its maximum safe capacity, exceeding which may cause overflow, tank overpressure, or environmental pollution. Definition: When the storage capacity of hazardous chemicals Q i Between low and low liquid levels Q min and the minimum storage allowed for loading Q low The values ​​are related, that is, 0 < Q min < Q i ≤ Q low When unloading is carried out, priority will be given to unloading operations (within this range, since the storage of hazardous chemicals is about to reach the low-low level of the storage tank, if loading continues, it may trigger an alarm, so unloading will be given priority in this range). When the storage capacity of hazardous chemicals Q i Given the maximum storage allowed for unloadingQ high and high-high liquid level values Q max Between, that is Q low < Q high ≤ Q i ≤ Q max When loading operations are carried out, priority will be given to loading operations (within this range, since the storage of hazardous chemicals is about to reach the high-high liquid level of the storage tank, if unloading continues, it may trigger an alarm, so loading operations will be given priority in this range). when Q low < Q i < Q high This period is defined as a time when parallel operations can be carried out simultaneously, within which ship unloading and vehicle loading can occur concurrently. Q low as well as Q high The numerical determination can be set according to port safety regulations and actual port operations. The port's back-end system can obtain the high-high liquid level readings for each hazardous chemical storage tank. Q max The amount of low-low liquid level is Q min Minimum storage allowed for loading Q low Maximum storage capacity allowed for unloading Q high .

[0051] Since machine learning models only predict the total number of ships and vehicles arriving within each time period, and the precise arrival time of each ship and vehicle can vary depending on the actual situation, this invention adopts a conservative strategy, initially assuming that all arriving ships and vehicles arrive at each... k The operation begins immediately at the start time within the time period. Specifically, when... k When =1, the hazardous chemical reserves before the first phase of operations begins are the hazardous chemical reserves predicted by the model, that is... Q k = Q 1= Q j ;when k =2, 3… K At that time, the first k Before the start of operations in this phase, the storage capacity of hazardous chemicals was [number missing]. k The remaining volume in the storage tank after the vessels and vehicles that can operate in stage -1 have completed their operations is obtained, that is... .

[0052] (1) Case 1: When 0 < Q min < Q k ≤ Q low If the load is below the minimum allowable loading limit, it should be unloaded from the ship first.

[0053] ① Calculate if the hazardous chemical storage capacity is increased to Q low Minimum unloading volume required Therefore, the required number of ships is (rounded up to the nearest integer): ; ② Calculate the actual increase in hazardous chemical storage capacity during this period. Q low Number of ships required: ; like m If 1=0, then unloading from the ship and loading from the truck are both impossible during this time period; that is, there is no work during this time period, and the process proceeds to the next time period. , No. k The hazardous chemical storage level before the start of operations in time period +1 was: Q k+1 = Q k Proceed to step four.

[0054] ③If m If 1 ≠ 0, then the volume of hazardous chemicals unloaded by the ship is... This can increase the storage capacity of hazardous chemicals to Q low The actual time used was After the ship unloads, the storage tank contains the following hazardous chemicals: So the first k The remaining time within each time period is .

[0055] 1) If Then in the first k No loading operations were carried out during this period, , No. k The hazardous chemical storage level before the start of operations in time period +1 was: At this time, the first k The duration of the +1 time period is: ( This means that this is a task that spans multiple time periods, that is, the first... k The time period for +1 can be less than the length of time for the task. ΔT And the first kThe actual usable operating time for the +1 time period is calculated by subtracting the hazardous chemical storage capacity from the total length of the two time periods. Q low (Time used), then the first k The maximum number of vessels that can operate within the +1 time period is The maximum number of vehicles that can be operated is Proceed to step four.

[0056] 2) and Q k '< Q low In the k Loading cannot be carried out for the remaining time within the specified time period, and unloading operations cannot be arranged. Therefore... , No. k The hazardous chemical storage level before the start of operations in time period +1 was: Proceed to step four.

[0057] 3) If and Q k '≥ Q low In the k During the remaining time period, unloading and loading can be carried out in parallel, proceeding to scenario three, with the initial hazardous chemical storage capacity being... Q k '; Remaining work time Remaining work time The maximum number of vessels that can operate inside is Remaining work time The maximum number of vehicles that can operate inside is Number of remaining waiting ships Number of vehicles remaining waiting Calculate the number of vessels still available for operation during this time period according to scenario three. m 2 and number of vehicles n 2. Then in the first... k During the time period, the number of vessels and vehicles that can operate are respectively , No. k The hazardous chemical storage level before the start of operations in time period +1 was: .

[0058] (2) Scenario 2: When Q high ≤ Q k ≤ Q max If the amount exceeds the maximum allowable storage capacity for unloading, priority will be given to loading it onto trucks.

[0059] ① Calculate if the hazardous chemical storage capacity is reduced to Q high Minimum required loading volume Therefore, the required number of vehicles is (rounded up to the nearest integer): ; ② Calculate the actual reduction of hazardous chemical reserves during this period to [a certain level]. Q high Number of vehicles required: ; like n If 1=0, then loading cannot be done on trucks and unloading cannot be done on ships during this time period; that is, there is no work during this time period, and the next time period begins. , No. k The hazardous chemical storage level before the start of operations in time period +1 was: Q k+1 = Q k Proceed to step four.

[0060] ③If n 1≠0, then the volume of hazardous chemicals carried by the vehicle is This can reduce the stockpile of hazardous chemicals to [a certain level]. Q high The actual time used was The amount of hazardous chemicals stored in the storage tank after unloading is So the first k The remaining time within each time period is .

[0061] 1) If Then in the first k No unloading operations were carried out during this period, , n z = n 1, no. k The hazardous chemical storage level before the start of operations in time period +1 was: At this time, the first k The duration of the +1 time period is: ( This means that this is a task that spans multiple time periods, that is, the first... k The time period for +1 can be less than the length of time for the task. ΔT And the first k The actual usable operating time for the +1 time period is calculated by subtracting the decrease in hazardous chemical storage from the total time of the two time periods. Q high (Time used), then the first k The maximum number of vessels that can operate within the +1 time period is The maximum number of vehicles that can be operated is Proceed to step four.

[0062] 2) If And Q k '≥ Q high Then in the first k Unloading operations cannot be carried out for the remaining time within the specified time period, and loading operations cannot be arranged. Therefore... , n z = n 1, no. k The hazardous chemical storage level before the start of operations in time period +1 was: Proceed to step four.

[0063] 3) If And Q k '< Q high In the k During the remaining time period, unloading and loading can be carried out in parallel, proceeding to scenario three, with the initial hazardous chemical storage capacity being... Q k ', Remaining work time Remaining work time The maximum number of vessels that can operate inside is Remaining work time The maximum number of vehicles that can operate inside is Number of remaining waiting ships Number of vehicles remaining waiting Calculate the number of vessels still available for operation during this time period according to scenario three. m 2 and number of vehicles n 2. Then in the first... k During the time period, the number of vessels and vehicles that can operate are respectively , No. k The hazardous chemical storage level before the start of operations in time period +1 was: .

[0064] (3) Scenario 3: When Q low < Q k < Q high During this period, operations are conducted in parallel, allowing for simultaneous unloading and loading. Under this assumption, the tank's storage capacity changes linearly over the time period, with extreme values ​​occurring at the beginning or end. Since the initial storage capacity is already within the safe range, ensuring the final storage capacity remains within the safe limit is sufficient to guarantee overall safety.

[0065] ① The calculation does not consider the safety boundary ( Q min , Q max The maximum possible number of vessels and vehicles operating at that time: ; ; ②Assuming continuous and uninterrupted operation of ships and vehicles, calculate the first... k Hazardous chemical reserves at the end of the time period: ; ③According to Q end With reserves boundary ( Q low , Q high Adjust the relationship between ) 1) If Q low < Q end < Q high Then take .

[0066] 2) If Q high ≤ Q end < Q max This indicates that the ship has unloaded too much cargo, so the number of vehicles that can be arranged for operations is... Find the first k The number of vessels that can operate during the time period is: ; 3) If Q min < Q end ≤ Q low This indicates that the vehicles are overloaded, therefore the number of vessels that can be deployed for operations is... Find the first k The number of vehicles that can operate during the time period is: ; In this case, the first k The hazardous chemical storage level before the start of operations in time period +1 was: .

[0067] Step 4: Update the k The number of ships and vehicles that could not be scheduled for operations within the specified time period is updated iteratively. kNumber of vessels that can operate during the +1 time period and number of vehicles

[0068] ①No. k The number of vessels that could not be scheduled for operations within the specified time period is:

[0069] ②No. k The number of vehicles that could not be scheduled for work within the specified time period is:

[0070] ③ Iterate within the time unit:

[0071] in t k , t k +1 are the first k and the k +1 start time point for a time period.

[0072] First calculate the... k Number of vessels that could not be scheduled for operations within the specified time period R k and number of vehicles S k and the k +1 Hazardous chemical storage before the start of operations Q k+1 Then, following step three, determine the first... k +1 time period hazardous chemical storage Q k+1 The work area, and calculate the first step according to the process in step three. k Number of vessels that can operate during the +1 time period and number of vehicles Repeat the above process until the iteration reaches the [number]th iteration. K A time period.

[0073] S33, in the k Within the time period, calculate the recommended start time for available vessels and vehicles to begin operations. The number of ships or vehicles required to reach the storage limit for hazardous chemicals is rounded up to the nearest integer. Therefore, scenarios one and two involve cross-time period operations, which means that... or In scenario three, the number of vessels or vehicles is rounded down and does not involve cross-time period operations. Therefore, the formula for calculating the recommended start time for operable vessels and vehicles is as follows: (1) If the first k- If a task involves operations spanning multiple time periods within a given time period, then in the first time period... k The start time of the operation within the time period ist k-1 + t 1, of which t k-1 For the first k -1 starting time point of a time period; when it occurs The situation, that is, the first k - The storage capacity of hazardous chemicals has not yet increased to the required level within a certain time period. Q low ,at this time t 1= t C1 When it appears The situation is also the first k - The storage capacity of hazardous chemicals has not decreased to [a certain level] within a certain time period. Q high ,at this time t 1= t D1。

[0074] ①No. k Recommended start time for each operable vessel within the specified time period ; in, T C In the first k The recommended start time for each operable vessel within the time period; for the a-th operable vessel in phase k, a Take 1 to Integers; t c The time taken for a ship to unload its cargo.

[0075] Then the first k Within the time period, the first The time point for the operational vessels to complete the operation is T Cl : ; ②No. k Recommended start time for each available vehicle within the specified time period. ; in, T D In the first k The recommended start time for each operational vehicle within a given time period; for the b-th operational vehicle in stage k, b Take 1 to Integers; t D The time taken to load goods onto a vehicle.

[0076] Then the firstk Within the time period, the first The timeframe for a single operational vehicle to complete the operation is as follows: T Dl : ; (2) If the first k- If no cross-time period operations are involved within time period 1, then in the first time period... k The start time of the operation within the time period is t k ,in t k For the first k The start time of each time period.

[0077] ①No. k Recommended start time for each operable vessel within the specified time period ; in, T C In the first k The recommended start time for each operable vessel within the time period; for the a-th operable vessel in phase k, a Take 1 to Integers; t c The time taken for a ship to unload.

[0078] Then the first k Within the time period, the first The time point for the operational vessels to complete the operation is T Cl : ; ②No. k Recommended start time for each available vehicle within the specified time period. ; in, T D In the first k The recommended start time for each operational vehicle within a given time period; for the b-th operational vehicle in stage k, b Take 1 to Integers; t D The time taken to load goods onto a vehicle.

[0079] Then the first k Within the time period, the first The timeframe for a single operational vehicle to complete the operation is as follows: T Dl : ; S34. Remaining Time Rescheduling of Transportation Resources and Methods for Coordinating Scheduled Arrivals So, for ships, the first... k The remaining workable time for each time period is T Ck : ; For vehicles, the first k The remaining workable time for each time period is T Dk : ; During the last period of the day, that is, the K There are remaining ships during the time period. R K There are no remaining vehicles for loading and the current stock of hazardous chemicals is... Q K It has exceeded the reserve limit. Q high Unloading operations may continue, but must not exceed the safety boundary. Q max The number of vessels that can be arranged for operation is Similarly, if it is within the last time period of the day, that is, the... K There are remaining vehicles during the time period. S K There are no remaining ships unloading cargo and the current hazardous chemical storage capacity is [not specified]. Q K It is below the reserve boundary. Q low Loading operations can continue, but must not fall below the safety boundary. Q min The number of vehicles that can be arranged for operation is The permitted operating times for each vessel and vehicle are detailed in Section 3 of this chapter.

[0080] During the last period of the day, that is, the K There are remaining ships during the time period. R K and remaining vehicles S K No homework was assigned, and the first K If all available vessels and vehicles are scheduled for operation within a given time slot, you can check if there are any remaining operating times within each time slot of the day suitable for unloading or loading operations. ,or If such a situation exists, vessels or vehicles can be rescheduled for that time period. However, since rescheduling vessels or vehicles for a new time period will affect the hazardous materials storage capacity for the next time period, the availability of vessels and vehicles for operation in subsequent time periods must be recalculated according to the above scheduling calculation process to determine the remaining vessels for the new round of scheduling. R K and vehicles S K Then, the final scheduling plan is selected by comparing and selecting the plan with the fewest remaining ships or vehicles. If there are unavoidable remaining ships and vehicles, they will be scheduled to be carried out in the first time period of the next day.

[0081] Based on the final scheduling plan, the port authority publishes suggested start times for each vessel and vehicle in each time period. Vessels and vehicles can choose suitable arrival times based on their own sailing plans or transport schedules, achieving pre-arranged arrival. For vessels or vehicles that have not arrived at the port before the suggested start time (e.g., 2 hours in advance), the port authority can proactively contact the transport company. The transport company can adjust its speed or departure time according to the actual sailing progress to arrive within the suggested time window, thereby reducing anchorage waiting and port congestion. For transport resources that cannot be scheduled for operation in the current time period (i.e., vessels and vehicles left over from the previous time period), the system automatically adds them to the waiting queue for the next time period and schedules them for priority operation in the next time period. Simultaneously, the suggested start time is recalculated, and a delay notification and a new suggested operation time are sent to the transport company. Through this process, the port can achieve pre-arranged arrival and orderly operation of transport resources based on hazardous chemical storage prediction and dynamic scheduling calculations, improving port operational efficiency and safety.

[0082] Application Cases To make it easier to understand, let's illustrate with an example. Suppose we divide a day into four time periods, that is... K =4, then ΔT =6h; known unloading speed v 1 = 200 m³ / h, loading speed is v 2 = 100 m³ / h; the average transport volume of hazardous chemical A per ship. The required quantity of hazardous chemical A for each vehicle So, the unloading time per ship t C =1000 / 200=5h, loading time per vehicle t D =50 / 100=0.5h; Low-low liquid level Q min =500m³, high-high liquid level Q max =5000m³; Minimum allowable loading capacity Q low=1000m³, maximum allowable unloading capacity Q high =4000m³. Predicted on a certain day using a machine learning model. t When =0, the current storage quantity of hazardous chemical A Q 0 = 800m³ R 0 = 0 ships S 0 = 0 vehicles. The estimated number of ships and vehicles arriving at the port during different time periods on that day is as follows: Table 1. Forecast of the number of ships and vehicles for each time period.

[0083] 1. Calculate the number of vessels and vehicles that can be operated. (1) k =1 (0:00~6:00) Q 1= Q 0 = 800m³, which belongs to case one. ① Calculate the total number of ships and vehicles waiting in the first time period. ; ; ② Calculate the maximum number of vessels and vehicles that can operate during the first time period: ; ; ③ Calculate the actual increase in hazardous chemical storage capacity during this period. Q low Number of ships required: ; The amount of hazardous chemicals unloaded by the ship was C 1 = 1000m³, Time taken t 1 = 5h, the current volume of hazardous chemicals in the storage tank is Q 1' = 1800m³; Time remaining in the first time period ΔT =1h, the maximum number of ships that can operate in the remaining time is The maximum number of vehicles that can be operated is The current total number of waiting ships is The total number of waiting vehicles is Now, proceed to scenario three.

[0084] ④ The calculation does not consider the safety boundary ( Q min , Q max The maximum possible number of vessels and vehicles operating at that time: ; ; ⑤ Assuming continuous and uninterrupted operation of ships and vehicles, calculate the hazardous chemical storage capacity at the end of the remaining workable time in the first time period: ; because Q low < Q end < Q high ,but m 2=0, n 2=2, then Number of remaining ships R 1=1, number of vehicles remaining S 1 = 6 Q 2 = 1700 m³.

[0085] (2) k =2 (6:00~12:00), Q 2 = 1700m³, which belongs to case three. ①Then the total number of waiting ships during this time period is The total number of waiting vehicles is

[0086] ② Without considering safety boundaries ( Q min , Q max The maximum possible number of vessels and vehicles operating at that time: ; ; ③ Assuming continuous and uninterrupted operation of ships and vehicles, calculate the hazardous chemical storage capacity at the end of the second time period: ; because Q low < Q end < Q high ,So Number of remaining ships R 2=1, number of vehicles remaining S 2=0, Q 3 = 2250 m³. The loading time during this period is 9 × 0.5 = 4.5 hours, with 1.5 hours of downtime.

[0087] (3) k =3 (12:00~18:00), Q3 = 2250m³, which belongs to case three. ①Then the total number of waiting ships during this time period is The total number of waiting vehicles is

[0088] ② Without considering safety boundaries ( Q min , Q max The maximum possible number of vessels and vehicles operating at that time: ; ; ③ Assuming continuous and uninterrupted operation of ships and vehicles, calculate the hazardous chemical storage capacity at the end of the third time period: ; because Q low < Q end < Q high ,So Number of remaining ships R 3=1, number of vehicles remaining S 3=3, Q 4 = 2650 m³.

[0089] (4) k =4 (18:00~24:00), Q 4 = 2650m³, which falls under case three. ①Then the total number of waiting ships during this time period is The total number of waiting vehicles is

[0090] ② Without considering safety boundaries ( Q min , Q max The maximum possible number of vessels and vehicles operating at that time: ; ; ③ Assuming continuous and uninterrupted operation of ships and vehicles, calculate the hazardous chemical storage capacity at the end of the fourth time period: ; because Q low < Q end < Q high ,So Number of remaining shipsR 4=0, number of vehicles remaining S 4 = 1.

[0091] (5) Reschedule of remaining vehicles For the vehicles, there is only a 1.5-hour idle period in the second time slot. The remaining vehicles are inserted into this time slot, and one additional vehicle is added to the original 9 vehicles in the second time slot, so that the total number of vehicles loaded in this time slot becomes 10.

[0092] ① Recalculate the second time period Assuming continuous operation of ships and vehicles, calculate the hazardous chemical storage capacity at the end of the second time period: ; because Q low < Q end < Q high ,That The loading time becomes 10 × 0.5 = 5 hours, and the loading idle time becomes 1 hour.

[0093] ② Recalculate the third time period only Q 3. If there is a change, then , Q low < Q end < Q high ,So Number of remaining ships R 3=1, number of vehicles remaining S 3=3, Q 4 = 2600m³.

[0094] ③ Recalculate the 4th time period only Q 4. If there is a change, then Q low < Q end < Q high ,So Number of remaining ships R 4=0, number of vehicles remaining S 4=0, at this time the storage tank contains 3000m³ of hazardous chemicals.

[0095] In summary, the recommended start times for all vessels and vehicles are as follows (Tables 2 and 3). Vessels or vehicles that can arrive within each time period should choose an appropriate time to arrive at the port. The port will remind the transporter of the arrival time 2 hours in advance based on the user's choice.

[0096] Table 2 Recommended Schedule for Vessel Operations

[0097] Table 3 Recommended vehicle start timetable

[0098] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for scheduling transport vessels and vehicles based on port hazardous chemical storage capacity prediction, characterized in that, Includes the following steps: S1. Construct a hazardous chemical port information data resource database: Obtain daily port time data, storage data, ship data, vehicle data, average ship unloading volume and average vehicle loading volume for various types of hazardous chemicals arriving at the port. The obtained data is preprocessed to obtain usable multi-source data. The usable multi-source data for several days constitutes the hazardous chemical port information data resource database. S2. Hazardous chemical storage prediction based on multi-source data: Based on multi-source data from the hazardous chemical port information database, LSTM, LightGBM and Prophet models are constructed. The average absolute percentage error (MAPE) is used as the model evaluation index. The model with the smallest MAPE is selected as the final prediction method to predict the hazardous chemical storage data for the next day and the information data of the expected arriving transporters. S3. Construct a time-segmented dynamic scheduling model: S31. Select a prediction model and prediction method. Based on the predicted hazardous chemical storage data for the next day and the information on the expected arriving transporters, divide the day into... K A time period of equal length; S32. Based on the hazardous chemical storage situation, calculate the number of ships and vehicles that can operate in each time period; S33, in the k Within the time period, k takes values ​​in the range [1, K]. Calculate the vessels and vehicles that can be operated and suggest the start time for operations. S34. Remaining time rescheduling of transportation resources and appointment-based arrival coordination methods.

2. The method for scheduling transport vessels and vehicles based on port hazardous chemical storage prediction according to claim 1, characterized in that, In S1, the construction of a hazardous chemicals port information database includes: Hazardous materials port time data: The 24 hours of the day are divided into K time periods according to the working hours of port operators; Hazardous chemical storage data: Measurements are taken using flow meters and tank gauging. Vessel data: Port terminal monitoring system and radar monitoring system are used to record the vessel's entry time, departure time, navigation trajectory, and the types and quantities of hazardous chemicals loaded. Vehicle data: The data is recorded using the gate at the port of entry, including the vehicle's license plate number, vehicle information, and the type and quantity of hazardous chemicals carried by the hazardous chemical vehicle. Average ship unloading volume and average vehicle loading volume: These are calculated based on the ratio of total daily ship unloading volume to the number of ships, and the ratio of total daily vehicle loading volume to the number of vehicles.

3. The method for scheduling transport vessels and vehicles based on port hazardous chemical storage prediction according to claim 2, characterized in that, The port terminal monitoring system includes surveillance cameras installed at high points on the terminal, at intersections, and around warehouses; The radar monitoring system includes radar devices installed at high points in the port or on specific monitoring towers. The gate recording device includes a camera that is erected as a horizontal bar at the entrance and exit of the port storage yard.

4. The method for scheduling transport vessels and vehicles based on port hazardous chemical storage prediction according to claim 1, characterized in that, In S2, the hazardous chemical storage data is: the hazardous chemical storage data of each hazardous chemical storage tank in the port before the port operations begin the next day, that is, before the ships enter the port to unload cargo and before the hazardous chemical vehicles enter the port to load cargo. The estimated arrival information includes the number of ships and vehicles carrying hazardous chemicals expected to arrive at the port at different times of the day, as well as the average unloading volume of ships and the average loading volume of vehicles carrying various types of hazardous chemicals for the next day.

5. The method for scheduling transport vessels and vehicles based on port hazardous chemical storage prediction according to claim 1, characterized in that, In S31, a prediction model is selected, and the model with the smallest MAPE is chosen. The MAPE calculation formula is as follows: ； in, It is the first i The true value of each sample It is the first i The predicted value for each sample, where x is the number of samples; After model selection, the prediction method is as follows: When the predicted hazardous chemical data is uniform, the data from the previous day can be used to predict the hazardous chemical storage data and the number of ships and vehicles expected to arrive at the port for the following day. When the data is uneven, the data from the previous week will be used to predict the hazardous chemical reserves and the number of ships and vehicles expected to arrive at the port in the following week.

6. The method for scheduling transport vessels and vehicles based on port hazardous chemical storage prediction according to claim 1, characterized in that, In S32, Step 1: Calculate the first... k The total number of ships and vehicles waiting within a given time period; Step 2: Calculate the first... k The maximum number of vessels and vehicles that can operate within a given time period; Step 3: According to the first k Before commencing operations within a specific time period, determine the location of hazardous chemical storage areas, define the operational mode, and calculate... k The number of vessels and vehicles that can operate within a given time period; Step 4: Update the k The number of ships and vehicles that could not be scheduled for operations within the specified time period is updated iteratively. k The number of vessels and vehicles that can operate during the +1 time period.

7. The method for scheduling transport vessels and vehicles based on port hazardous chemical storage prediction according to claim 6, characterized in that, In step three of S32, the day is divided into K equal-length time periods, each with a length of ΔT, where ΔT is greater than the time for unloading a single ship and loading a single truck. The operational mode for each time period is determined based on the range in which the hazardous chemical storage volume is located at the beginning of the time period. Case 1, when 0 < Q min < Q i ≤ Q low When necessary, unloading operations should be carried out first; Scenario 1, when Q low < Q high ≤ Q i ≤ Q max Loading operations should be carried out in a timely manner. Scenario 1, when Q low < Q i < Q high At the same time, unloading of ships and loading of hazardous chemical vehicles can be carried out simultaneously; in, Q i Data on hazardous chemical reserves; Q min Low-low level refers to the level of hazardous chemicals in the storage tank being lower than the safety limit, which may lead to pump running dry, evaporation of pipeline residues or equipment damage, or even leakage risk due to pressure imbalance. Q low The minimum storage quantity allowed for loading; Q high The maximum permitted storage capacity for unloading; Q max The high-high level indicates that the storage tank is close to its maximum safe capacity. Exceeding this capacity may cause overflow, tank overpressure, or environmental pollution.

8. The method for scheduling transport vessels and vehicles based on port hazardous chemical storage prediction according to claim 1, characterized in that, In S33, the first k- For vessels and vehicles, the recommended start time calculation process is as follows: (This section discusses both time-spanning and non-time-spanning operations.) First calculate the first k Recommended start time for each operable vessel within the specified time period. T C Recommended start time for each operational vehicle T D ; Calculate the next k Within the time period, the first The time point for the operational vessels to complete the operation is T Cl , No. The time point when the vehicle can complete the operation is T Dl .

9. The method for scheduling transport vessels and vehicles based on port hazardous chemical storage prediction according to claim 1, characterized in that, In S34, the remaining time rescheduling of transportation resources involves the process of rescheduling transportation resources that remain after all time slots have ended for the day. First, calculate the remaining workable time of the vessel in the k-th time period. T Ck The remaining working time of the vessel is T Dk ; During the last time period of the day: if there are remaining ships but no remaining hazardous chemical vehicles and the current hazardous chemical storage has exceeded the maximum storage allowed for unloading, unloading operations will continue, but the level must not exceed the high-high liquid level; if there are remaining vehicles but no remaining ships for unloading and the current hazardous chemical storage has fallen below the minimum storage allowed for loading, loading operations will continue, but the level must not fall below the low-low liquid level. In the last time period of the day: if there is remaining workable time that meets the work conditions, the remaining resources are scheduled into that time period, the impact of subsequent time periods is recalculated, and finally the scheme with the fewest remaining resources is selected as the final scheduling scheme. If there are still remaining resources that cannot be scheduled during the last time slot of the day, they will be scheduled for the first time slot of the next day. If there are any remaining ships and vehicles, they will be scheduled for operation in the first time slot of the next day.

10. The method for scheduling transport vessels and vehicles based on port hazardous chemical storage prediction according to claim 1, characterized in that, In S34, the port can proactively contact the transporter if the vessel or vehicle has not arrived at the port before the recommended start time of operations. The transporter can then adjust the speed or departure time according to the actual sailing progress to arrive within the recommended time window. For transportation resources that cannot be scheduled for operation in the current time period, the system will automatically add them to the waiting queue of the next time period and schedule them for priority operation in the next time period. At the same time, the system will recalculate the suggested start time and send a delay notice and a new suggested operation time to the transportation party.

Images