Transport capacity scheduling method, device and equipment for cigarette logistics and storage medium

Abstract

The invention discloses a cigarette logistics transport capacity scheduling method, device and equipment and a storage medium, relates to the technical field of tobacco logistics, and realizes systematization and intelligentization improvement of a cigarette logistics transport capacity scheduling process. The accuracy and integrity of Beidou trajectory data are ensured through data cleaning and preprocessing, and a foundation is laid for subsequent analysis; a structured transport capacity resource pool is constructed through multi-source data fusion, vehicle states and business requirements are integrated, and decision support is enhanced; establishing a dual-objective optimization model, and balancing the transportation cost and the carbon emission under the condition of meeting load and time window constraints; an intelligent algorithm is adopted to solve and generate a Pareto optimal scheme, and scientific scheduling selection under multi-target tradeoff is provided; the path is dynamically adjusted based on real-time position data, and the adaptability and execution efficiency of the scheme are improved; and closed-loop management is realized through a scheme distribution mechanism. The vehicle resource utilization rate is optimized, the fuel oil and labor cost is reduced, and meanwhile the transportation safety is enhanced through accurate monitoring.

Description

Technical Field

[0001] This application relates to the field of tobacco logistics technology, and in particular to a method, apparatus, equipment and storage medium for scheduling the transportation capacity of cigarette logistics. Background Technology

[0002] With the innovation of logistics facilities and equipment, and the widespread application of information technology in logistics, the tobacco industry needs to improve work efficiency, reduce operating costs, optimize core business processes, and enhance market competitiveness through logistics strategies. From tobacco leaf processing to cigarette production, and from cigarette warehouses to retail outlets, the sheer number of logistics links and the stringent control required are characteristics of the tobacco industry's logistics.

[0003] Vehicle loading and unloading are crucial aspects of logistics, directly impacting costs. Customer demand for cigarettes is evolving towards diversification, smaller batches, higher frequency, and greater personalization. Simultaneously, the operational volume and complexity of tobacco logistics are increasing daily. Therefore, how to quickly and effectively load and unload vehicles based on customer orders across the country, combined with factors such as logistics routes, transportation vehicles, and cargo (cigarette cases) specifications, has become an urgent problem to be solved in Shanghai Tobacco's logistics operations.

[0004] Traditional cigarette logistics capacity management relies primarily on manual experience and simple data statistics, lacking in-depth mining and utilization of massive amounts of data during transportation. For example, the ability to collect and analyze real-time data such as vehicle location, driving status, and cargo status is insufficient, resulting in an inability to comprehensively and accurately grasp key operational indicators such as transportation efficiency, cost, and customer satisfaction. This extensive management model makes it difficult to achieve refined management of transportation capacity and optimal allocation of resources. Summary of the Invention

[0005] The main objective of this application is to provide a method, apparatus, equipment, and storage medium for scheduling the transportation capacity of cigarette logistics, so as to solve the problem that the existing management model of cigarette logistics capacity is difficult to achieve refined management of transportation capacity and optimal allocation of resources.

[0006] To achieve the above objectives, this application provides the following technical solution: A method for scheduling transportation capacity in cigarette logistics, wherein the method is applied to transport vehicles used in cigarette logistics, the transport vehicles being equipped with a BeiDou satellite navigation system, and the method includes: Step S1: The trajectory data of the transport vehicle is collected in real time through the BeiDou satellite navigation system, and the trajectory data is cleaned and preprocessed to obtain processed BeiDou trajectory data; Step S2: The processed Beidou trajectory data is associated and aligned with the order information and inventory information of cigarette logistics through data fusion to obtain a structured transportation capacity resource pool; Step S3: Establish a dual-objective heterogeneous vehicle route optimization model that minimizes transportation costs and carbon emissions based on a structured transportation capacity resource pool, and constrain the dual-objective heterogeneous vehicle route optimization model by the maximum vehicle load and customer time window. Step S4: The dual-objective heterogeneous vehicle path optimization model is transformed into a multi-objective mixed integer programming problem, and the multi-objective mixed integer programming problem is solved by a non-dominated sorting genetic algorithm to obtain a Pareto optimal scheduling scheme. Step S5: Select the Pareto optimal solution with the minimum total transportation cost, and dynamically adjust the route of the transport vehicle based on the real-time location data of the transport vehicle using a global optimization algorithm to obtain the final scheduling solution of the transport vehicle. Step S6: Send the final scheduling plan to the transport vehicle and the external monitoring terminal.

[0007] Beneficial effects of steps S1 to S6: This module achieves a systematic and intelligent upgrade to the cigarette logistics capacity scheduling process. Specifically, step S1, trajectory data cleaning and preprocessing, ensures the accuracy and integrity of BeiDou positioning information, laying a reliable data foundation for subsequent analysis; step S2, multi-source data fusion, constructs a structured capacity resource pool, integrating vehicle status and business needs, enhancing the information support capability for scheduling decisions; step S3, a dual-objective optimization model, balances transportation costs and carbon emissions while meeting load and time window constraints, promoting the coordinated development of logistics operations towards economic efficiency and environmental protection; step S4, intelligent algorithms solve for Pareto optimal solutions, providing scientific scheduling choices under multi-objective trade-offs; step S5, dynamic path adjustment, optimizes vehicle routes based on real-time location data, improving the adaptability and execution efficiency of scheduling schemes; and step S6, a scheme distribution mechanism, ensures that scheduling instructions are promptly transmitted to the execution and monitoring ends, achieving closed-loop management. Overall, these steps work together to significantly optimize vehicle resource utilization, reduce empty runs and waiting time, lower fuel consumption and labor costs, and enhance the safety of the transportation process and customer service levels through precise monitoring and rapid response, providing core technological support for the efficient and low-carbon operation of cigarette logistics.

[0008] As a further improvement to this application, step S6 involves sending the final scheduling plan to the transport vehicle and the external monitoring terminal, followed by: Step S10: Monitor the execution status of the final scheduling scheme and the information of newly arrived orders in real time using a sliding time window algorithm to obtain a real-time monitoring dataset; Step S20: Calculate the key performance indicator deviation of the dual-objective heterogeneous vehicle path optimization model based on the real-time monitoring dataset. When the key performance indicator deviation exceeds a preset deviation threshold, trigger a model re-solution signal. Step S30: Reconstruct the dual-objective heterogeneous vehicle path optimization model for the current time period based on the model re-solution signal to obtain the updated optimization model; Step S40: Repeat steps S4 to S5 with the updated optimized model as the subject of execution to obtain the updated scheduling scheme; Step S50: Send the updated scheduling plan to the transport vehicle and the external monitoring terminal.

[0009] Beneficial effects of steps S10 to S50: This section enhances the real-time response capability and continuous optimization level of the cigarette logistics capacity scheduling system by constructing a dynamic closed-loop optimization mechanism. Specifically, step S10's sliding time window monitoring mechanism continuously tracks the execution status of the scheduling plan and new order information, forming a real-time data feedback loop to provide precise input for dynamic system adjustments; step S20's key performance indicator deviation detection function establishes quantitative evaluation standards, automatically triggering a re-solution mechanism when the actual operating status deviates from the expected target, ensuring the scheduling system's sensitivity to abnormal operating conditions; step S30's model reconstruction capability rapidly adjusts and optimizes model parameters based on real-time data, enabling the scheduling strategy to adapt to changes in the transportation environment; and step S40's iterative solution process generates updated solutions by reusing core algorithms. This ensures technical consistency in the dynamic adjustment process; the scheme distribution mechanism in step S50 synchronizes the optimization results to the execution terminal in real time, forming a complete closed loop of decision-making-execution-feedback; the synergistic effect of this series of steps enables the scheduling system to have the ability to quickly respond to dynamic factors such as vehicle status fluctuations, new order tasks, and changes in traffic conditions during transportation. By continuously calibrating the scheduling scheme, the deviation between the plan and the actual situation is reduced, vehicle travel routes and resource utilization are optimized, and additional costs caused by environmental changes are reduced. At the same time, the closed-loop control improves the system's anti-interference ability and operational stability, providing continuous optimization and dynamic scheduling support for cigarette logistics.

[0010] As a further improvement to this application, step S1 involves collecting the trajectory data of the transport vehicle in real time through the BeiDou satellite navigation system, and cleaning and preprocessing the trajectory data to obtain processed BeiDou trajectory data, including: Step S11: Obtain the original trajectory data stream of the transport vehicle in real time through the API interface of the Beidou satellite navigation system; Step S12: Perform preliminary filtering on the original trajectory data stream using a sliding window algorithm to obtain preliminary filtered trajectory data; Step S13: Identify and remove outliers and noise data from the preliminary filtered trajectory data using the isolated forest algorithm to obtain the anomaly-cleaned trajectory data. Step S14: The missing values ​​of the abnormal cleaned trajectory data are filled by linear interpolation, and the trajectory is smoothed by Kalman filtering algorithm to obtain continuous and complete trajectory data, which is the processed BeiDou trajectory data.

[0011] Beneficial effects of steps S11 to S14: This section provides a high-quality data foundation for subsequent capacity scheduling decisions by constructing a systematic BeiDou trajectory data processing flow. Specifically, step S11's API interface real-time acquisition mechanism ensures the integrity and timeliness of the original trajectory data stream, laying a reliable data input source for the entire process; step S12's sliding window filtering algorithm effectively removes abnormal data points that are significantly beyond the physical range, improving the rationality and usability of the data; step S13 uses the isolated forest algorithm for outlier detection, identifying and removing noise and outliers in the trajectory, enhancing the accuracy and consistency of the data; step S14 fills in missing data segments through linear interpolation and combines it with the Kalman filter algorithm to smooth and optimize the trajectory, generating a continuous and complete trajectory sequence, improving the continuity of the data in both time and space dimensions. These steps gradually improve the quality of BeiDou trajectory data, eliminate errors and interference in the acquisition process, and provide stable and accurate data support for subsequent steps such as capacity resource pool construction and vehicle route optimization, thereby ensuring the reliability of the scheduling model and the overall system's operational efficiency.

[0012] As a further improvement to this application, step S2 involves associating and aligning the processed BeiDou trajectory data with the order information and inventory information of cigarette logistics through data fusion to obtain a structured transportation capacity resource pool, including: Step S21: Extract order information and inventory information from the external cigarette logistics information system, and perform data cleaning processing on missing values ​​and outliers to obtain preprocessed business data; Step S22: Align the preprocessed business data with the processed BeiDou trajectory data in terms of time series, and integrate them to obtain time-synchronized multi-source data; Step S23: Based on the unique identifier of the transport vehicle, perform association rule matching on the time-synchronized multi-source data to obtain a preliminary association dataset; Step S24: The trajectory data and business data in the preliminary associated dataset are fused using the Kalman filter algorithm to obtain fused vehicle business data. Step S25: Evaluate the real-time availability and load capacity of the fused vehicle business data using a logistic regression algorithm to obtain the vehicle capacity evaluation result; Step S26: Organize the vehicle capacity assessment results into a structured database format to obtain the structured capacity resource pool.

[0013] Beneficial effects of steps S21 to S26: This section constructs a high-quality structured transportation capacity resource pool through a systematic data fusion and processing process, providing core data support for subsequent intelligent scheduling decisions. Specifically, step S21, business data cleaning, ensures the accuracy and completeness of order and inventory information, eliminating noise and biases in the original data; step S22, time series alignment, synchronizes BeiDou trajectory data and business data in the time dimension, establishing a unified benchmark for multi-source data association; step S23, based on vehicle unique identifier association rule matching, effectively links discrete data sources, forming a preliminary integrated dataset; step S24, applies the Kalman filter algorithm to deeply fuse trajectory and business data, improving data consistency in the spatiotemporal dimension; step S25, uses the logistic regression algorithm to quantitatively evaluate vehicle real-time availability and load capacity, achieving standardized measurement of transportation resources; and step S26, organizes the evaluation results into a structured database, forming a transportation resource pool that can be directly called by the scheduling model. These steps gradually break down data silos, improve data interoperability, and construct a dynamic data system that accurately reflects the correlation between vehicle status, location information, and order demand, providing unified and reliable data input for subsequent stages such as vehicle route optimization and real-time scheduling, ensuring that the scheduling system makes decisions and optimizes based on real and consistent business status.

[0014] As a further improvement to this application, step S3 involves establishing a dual-objective heterogeneous vehicle route optimization model that minimizes transportation costs and carbon emissions based on a structured capacity resource pool, and constraining the dual-objective heterogeneous vehicle route optimization model by vehicle maximum load and customer time window, including: Step S31: Extract the vehicle model, vehicle load capacity, and current location information from the structured transportation capacity resource pool to construct a heterogeneous vehicle feature set; Step S32: Obtain external order requirements and determine customer location, customer time window, and cargo weight and volume based on the external order requirements, and integrate them to obtain a customer demand feature vector; Step S33: Using the minimization of transportation costs as the main objective function, the customer demand feature vector is input into the main objective function to calculate vehicle start-up cost, travel distance cost, and time cost; Step S34: Using the minimization of carbon emissions as a secondary objective function, calculate the carbon emissions based on the heterogeneous vehicle feature set; Step S35: Reconstruct the primary objective function, the secondary objective function, the maximum vehicle load, and the customer time window into a dual-objective heterogeneous vehicle path optimization model.

[0015] Beneficial effects of steps S31 to S35: This section establishes a dual-objective optimization framework for cigarette logistics capacity scheduling through a systematic model building process. Specifically, step S31 extracts a heterogeneous set of vehicle features to accurately capture the diversity of vehicle models, load capacities, and current locations, providing a realistic capacity resource foundation for the scheduling model; step S32 integrates customer demand feature vectors, clarifying customer locations, time windows, and cargo attributes to define specific input constraints for the optimization problem; step S33 prioritizes minimizing transportation costs, quantifying vehicle start-up costs, travel distance costs, and time costs to ensure the economic feasibility of the scheduling plan; step S34 introduces minimizing carbon emissions as a secondary objective, calculating the environmental impact based on vehicle characteristics to promote green logistics development; and step S35 reconstructs the dual-objective function with load and time window constraints into a complete optimization model, achieving scientific decision-making under multi-objective trade-offs. These steps work synergistically, enabling the scheduling model to simultaneously optimize transportation costs and carbon emissions, improve resource utilization, reduce operating expenses, and enhance the environmental friendliness of the plan, providing a core optimization engine for intelligent scheduling.

[0016] As a further improvement to this application, step S4 transforms the bi-objective heterogeneous vehicle path optimization model into a multi-objective mixed integer programming problem, and solves the multi-objective mixed integer programming problem using a non-dominated sorting genetic algorithm to obtain a Pareto optimal scheduling scheme, including: Step S41: Discretize the continuous variables in the dual-objective heterogeneous vehicle path optimization model, define the dual objective function and constraints of the dual-objective heterogeneous vehicle path optimization model as linear expressions, and integrate all linear expressions to obtain a multi-objective mixed integer programming problem. Step S42: Perform chromosome encoding on the multi-objective mixed integer programming problem to obtain the initial gene population; Step S43: Based on the initial gene population, perform fast non-dominated sorting using a non-dominated sorting genetic algorithm, and divide the individuals in the initial gene population into different frontier levels according to the Pareto rank. Step S44: Calculate the crowding distance for individuals in the stratified population, and combine it with the elite retention strategy to select high-quality individuals to enter the next generation of evolution. Step S45: Perform simulated binary crossover and polynomial mutation on all high-quality individuals in the population to obtain a new generation of individuals. Step S46: Perform fast non-dominated sorting and crowding distance calculation on all new generation population individuals again to iteratively update the Pareto solution set; Step S47: When the maximum number of iterations is reached or the quality of the Pareto solution set is stable, terminate the iteration and output the Pareto optimal scheduling scheme.

[0017] Beneficial effects of steps S41 to S47: This section achieves efficient solutions to multi-objective optimization problems through a systematic algorithmic process, providing a scientific decision-making basis for cigarette logistics capacity scheduling. Specifically, step S41 discretizes the continuous variables in the dual-objective heterogeneous vehicle routing optimization model and transforms them into linear expressions, constructing a standardized multi-objective mixed integer programming problem, thus adapting complex problems to an efficient solution framework; step S42 generates an initial gene population through chromosome encoding, laying a diverse starting point for the algorithm search; step S43 applies a non-dominated sorting genetic algorithm for rapid stratification, identifying the Pareto rank of the solution set and distinguishing the quality levels of the set; step S44 calculates the crowding distance and combines it with an elite strategy to select high-quality individuals, balancing the diversity and convergence of the solution set; step S45 generates a new generation of population through simulated binary crossover and polynomial mutation operations, promoting global exploration and local optimization of the solution space; step S46 iteratively updates the Pareto solution set, gradually improving the quality of the solution; and step S47 outputs a stable Pareto optimal solution based on a preset termination condition, ensuring reliable convergence of the algorithm. These steps work together to transform the multi-objective optimization problem into a computable form, generating a set of scheduling schemes that balance transportation costs and carbon emissions through intelligent algorithms. This enhances the flexibility and robustness of scheduling decisions, provides high-quality input for subsequent route adjustments, and improves overall optimization efficiency.

[0018] As a further improvement to this application, step S5 involves selecting the Pareto optimal solution with the minimum total transportation cost, and dynamically adjusting the transportation vehicle's path using a global optimization algorithm based on the vehicle's real-time location data to obtain the final scheduling scheme for the transportation vehicle, including: Step S51: Select the scheme with the minimum total transportation cost from the Pareto optimal scheduling schemes as the basic scheduling scheme; Step S52: Extract the vehicle route sequence and order allocation information from the basic scheduling scheme to obtain the initial route plan; Step S53: Obtain the real-time location data of the transport vehicle based on the BeiDou satellite navigation system, and align it with the initial route plan in time and space; Step S54: The aligned real-time position data and the initial path plan are input into the particle swarm optimization algorithm to perform global optimization with the goal of minimizing path deviation cost, and the adjusted path scheme is obtained iteratively. Step S55: The 2-opt local search algorithm eliminates the intersecting paths of the adjusted route scheme and shortens the travel distance to obtain the final scheduling scheme of the transport vehicles.

[0019] Beneficial effects of steps S51 to S55: This module significantly enhances the adaptability and execution efficiency of the scheduling scheme to the real-time operating environment through a systematic dynamic path adjustment process. Specifically, step S51 selects the most economically optimal basic scheduling scheme from the Pareto optimal solution set, providing a high-quality initial decision-making basis for dynamic adjustment; step S52 extracts the path sequence and order allocation information of the scheme to form an operable initial path plan; step S53 establishes a mapping relationship between the actual location and the predetermined route by aligning BeiDou positioning data with the planned path in time and space, providing an accurate time and space reference for path optimization; step S54 applies the particle swarm optimization algorithm to perform global optimization with the goal of minimizing path deviation cost, enabling rapid response to emergencies and path replanning; step S55 uses the 2-opt local search algorithm to fine-tune the optimized path, eliminating path intersections and shortening the total travel distance, further improving the economy and feasibility of the scheme. These steps gradually transform static optimization schemes into dynamic execution paths that adapt to real-time conditions. Through algorithmic collaborative optimization, the system enhances the vehicle's adaptability to dynamic factors such as traffic changes and order adjustments, reduces the deviation between actual driving and planned routes, lowers additional costs caused by environmental changes, and strengthens the real-time decision-making capabilities and overall robustness of the scheduling system.

[0020] To achieve the above objectives, this application also provides the following technical solutions: A capacity scheduling device for cigarette logistics, wherein the capacity scheduling device is applied to the capacity scheduling method described above, and the capacity scheduling device comprises: The BeiDou trajectory data processing module is used to collect the trajectory data of the transport vehicle in real time through the BeiDou satellite navigation system, and to clean and preprocess the trajectory data to obtain processed BeiDou trajectory data. The capacity resource pool construction module is used to associate and align the processed Beidou trajectory data with the order information and inventory information of cigarette logistics through data fusion to obtain a structured capacity resource pool; The optimization model building module is used to establish a dual-objective heterogeneous vehicle route optimization model that minimizes transportation costs and carbon emissions based on a structured transportation capacity resource pool, and to constrain the dual-objective heterogeneous vehicle route optimization model by the maximum vehicle load and the customer time window. The optimization model solving module is used to transform the dual-objective heterogeneous vehicle path optimization model into a multi-objective mixed integer programming problem, and solve the multi-objective mixed integer programming problem using a non-dominated sorting genetic algorithm to obtain a Pareto optimal scheduling scheme. The dynamic route adjustment module is used to select the Pareto optimal solution with the minimum total transportation cost, and to dynamically adjust the transportation vehicle's route based on the real-time location data of the transportation vehicle using a global optimization algorithm to obtain the final scheduling solution for the transportation vehicle. The final scheduling scheme sending module is used to send the final scheduling scheme to the transport vehicle and the external monitoring terminal.

[0021] To achieve the above objectives, this application also provides the following technical solutions: An electronic device includes a processor and a memory coupled to the processor, the memory storing program instructions executable by the processor; when the processor executes the program instructions stored in the memory, it implements the capacity scheduling method described above.

[0022] To achieve the above objectives, this application also provides the following technical solutions: A computer-readable storage medium storing program instructions that, when executed by a processor, can implement the capacity scheduling method described above. Attached Figure Description

[0023] Figure 1 This is a schematic flowchart illustrating the steps of one embodiment of a cigarette logistics capacity scheduling method according to this application. Figure 2 This is a schematic diagram of the functional modules of a capacity scheduling device for cigarette logistics according to an embodiment of this application; Figure 3 This is a schematic diagram of the structure of an embodiment of the electronic device of this application; Figure 4 This is a schematic diagram of the structure of one embodiment of the storage medium of this application. Detailed Implementation

[0024] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0025] The terms "first," "second," and "third" in this application are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined as "first," "second," or "third" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified. All directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of this application are only used to explain the relative positional relationships and movements between components in a specific orientation (as shown in the figures). If the specific orientation changes, the directional indications also change accordingly. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices.

[0026] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0027] like Figure 1 As shown, this embodiment provides a method for scheduling transportation capacity in cigarette logistics. This method is applied to the transportation vehicles used in cigarette logistics, and the vehicles are equipped with the BeiDou satellite navigation system.

[0028] Specifically, the capacity scheduling method includes the following steps: Step S1: Real-time trajectory data of transport vehicles is collected through the BeiDou satellite navigation system, and the trajectory data is cleaned and preprocessed to obtain processed BeiDou trajectory data.

[0029] Further, in step S1, the trajectory data of the transport vehicle is collected in real time through the BeiDou satellite navigation system, and the trajectory data is cleaned and preprocessed to obtain processed BeiDou trajectory data, including the following steps: Step S11: Obtain the original trajectory data stream of the transport vehicle in real time through the API interface of the Beidou satellite navigation system.

[0030] Preferably, the trajectory data stream of the transport vehicle can be obtained in real time through the standard API interface (such as HTTP / HTTPS protocol) provided by the BeiDou Navigation Satellite System. The data packet format follows the NMEA-0183 standard and includes information such as latitude and longitude coordinates (WGS-84 coordinate system), timestamp (UTC time), ground speed (unit: km / h), and heading angle.

[0031] Meanwhile, the data sampling frequency can be dynamically adjusted according to the vehicle's operating status to balance data accuracy and transmission load. For example, the sampling frequency for low speeds below 20 km / h can be set to 1 Hz, or once per second, to reduce data redundancy; the sampling frequency for medium-high speeds above 20 km / h can be set to 0.2 Hz, or once every 5 seconds, to ensure trajectory details during medium-high speed movement.

[0032] Step S12: The original trajectory data stream is initially filtered using a sliding window algorithm to obtain initially filtered trajectory data.

[0033] Preferably, the window size of the sliding window algorithm can be set to 10 consecutive data points, and for positional jumps, it can be based on the distance formula between adjacent points d=Haversine(x t ,x t−1 The displacement is calculated using the Haversine formula, which is used to calculate the spherical distance. The displacement threshold can be set to 100 meters per second. If the displacement exceeds the displacement threshold, it is marked as a jump point and removed.

[0034] Step S13: Isolation forest algorithm is used to identify and remove outliers and noise data in the initially filtered trajectory data to obtain the anomaly-cleaned trajectory data.

[0035] Preferably, the isolated forest algorithm can be trained using historical normal trajectory data, with the number of trees set to 100 and the subsampling size set to 256.

[0036] Preferably, trajectory points can be converted into feature vectors, including position, velocity, and acceleration, through front-to-back point difference calculation. Simultaneously, using the formula... Calculate the outlier score, where E(h(x)) is the expected path length of data point x, c(n) is the normalization factor for the average path length of the binary search tree, and the outlier score threshold can be set to 0.6. Data points with scores higher than the outlier score threshold are identified as outliers.

[0037] Step S14: Fill missing values ​​in the abnormal cleaning trajectory data by linear interpolation, and smooth the trajectory by Kalman filtering algorithm to obtain continuous and complete trajectory data, which is the processed BeiDou trajectory data.

[0038] Preferably, missing data segments can be processed using a linear interpolation algorithm, for missing point pt Through the missing point p t Before and after the effective point p t−1 and p t+1 calculate: Where t is the value at the current time, t-1 is the value at the previous time, and t+1 is the value at the next time.

[0039] Preferably, trajectory noise can be reduced using a Kalman filter algorithm. The model includes a state vector (based on position and velocity) and an observation vector (based on the original BeiDou position).

[0040] Specifically, the state equation is x t =Ax t−1 +Bu t +w t ;where x t Let A be the state vector at time t; let A be the state transition matrix based on the uniform velocity model; let B be the control input matrix, which can be set to 0 in this embodiment; w t For process noise, the covariance is set to 0.1.

[0041] Specifically, the observation equation is z t =Hx t +v t Where H is the observation matrix; v t To observe the noise, the covariance is set to 1.

[0042] Beneficial effects of steps S11 to S14: This section provides a high-quality data foundation for subsequent capacity scheduling decisions by constructing a systematic BeiDou trajectory data processing flow. Specifically, step S11's API interface real-time acquisition mechanism ensures the integrity and timeliness of the original trajectory data stream, laying a reliable data input source for the entire process; step S12's sliding window filtering algorithm effectively removes abnormal data points that are significantly beyond the physical range, improving the rationality and usability of the data; step S13 uses the isolated forest algorithm for outlier detection, identifying and removing noise and outliers in the trajectory, enhancing the accuracy and consistency of the data; step S14 fills in missing data segments through linear interpolation and combines it with the Kalman filter algorithm to smooth and optimize the trajectory, generating a continuous and complete trajectory sequence, improving the continuity of the data in both time and space dimensions. These steps gradually improve the quality of BeiDou trajectory data, eliminate errors and interference in the acquisition process, and provide stable and accurate data support for subsequent steps such as capacity resource pool construction and vehicle route optimization, thereby ensuring the reliability of the scheduling model and the overall system's operational efficiency.

[0043] Step S2 involves linking and aligning the processed BeiDou trajectory data with the order and inventory information of cigarette logistics through data fusion to obtain a structured transportation capacity resource pool.

[0044] Further, in step S2, the processed BeiDou trajectory data is correlated and aligned with the order information and inventory information of cigarette logistics through data fusion to obtain a structured transportation capacity resource pool, including the following steps: Step S21: Extract order information and inventory information from the external cigarette logistics information system, and perform data cleaning processing on missing and outlier values ​​to obtain preprocessed business data.

[0045] Preferably, order and inventory information can be extracted in batches from an external cigarette logistics information system via an API interface (such as a RESTful API). Order data fields include order ID, customer location, demand time window, cargo weight, and volume; inventory data fields include warehouse ID, cigarette type, available quantity, and storage status. The data extraction frequency is set to once every 5 minutes.

[0046] Preferably, missing value handling can be achieved by calculating the missing rate of each field and setting a missing rate threshold of 10%. If the missing rate is lower than the threshold, mean imputation (for numeric fields) or mode imputation (for categorical fields) is used; if the missing rate exceeds the threshold, the field is discarded or iterative imputation is performed using the EM algorithm.

[0047] Preferably, outlier detection can be based on the Z-score method, with the Z-score threshold set to ±3. For numerical fields (such as cargo weight), if the Z-score exceeds the threshold, it is considered an outlier and removed; for categorical fields, frequency analysis is used to remove rare values ​​with an occurrence frequency of less than 1%.

[0048] Step S22: Align the preprocessed business data with the processed BeiDou trajectory data in terms of time series and integrate them to obtain time-synchronized multi-source data.

[0049] Preferably, time series alignment can be achieved by using a dynamic time warping algorithm to align BeiDou trajectory data and business data with inconsistent timestamps. First, a unified time base is established using UTC time, and a time alignment accuracy threshold of ±30 seconds is set. If the time difference exceeds the threshold, a linear interpolation method is used to adjust the business data timestamps.

[0050] Preferably, the aligned data can be grouped according to the length of the time window and integrated into time series blocks. Each block contains a sequence of trajectory points and a corresponding snapshot of business data, generating a time-synchronized multi-source dataset.

[0051] Step S23: Based on the unique identifier of the transport vehicle, perform association rule matching on the time-synchronized multi-source data to obtain a preliminary association dataset.

[0052] Preferably, association rule mining can be performed using the Apriori algorithm, with vehicle unique identifiers (such as license plate numbers) as key fields. The minimum support threshold is set to 0.1, and the minimum confidence threshold is set to 0.7 to mine frequent itemsets between trajectory data and business data.

[0053] For example, if a correlation rule is constructed such that the distance between the trajectory point of vehicle X at time T and the customer's location of order Y is less than 100 meters, the spatial distance is calculated using the Haversine formula, and the distance threshold can be set to 100 meters. If the distance is less than 100 meters, it is considered a valid correlation.

[0054] Step S24: The trajectory data and business data in the preliminary associated dataset are fused using the Kalman filter algorithm to obtain the fused vehicle business data.

[0055] Preferably, in step S24, the trajectory data and service data are fused again using the Kalman filter algorithm, and the state vector is defined as x. t =[position x, position y, velocity, load], and the observation vector is zt=[BeiDou position x, BeiDou position y, service load value].

[0056] It is worth noting that step S24 involves reusing the Kalman filter algorithm, rather than using the Kalman filter algorithm that appeared for the first time above.

[0057] Specifically, the state transition matrix A, based on the uniform velocity model, can be set as follows: , where Δt is the sampling interval, for example, 5 seconds. The process noise covariance Q can be set as a diagonal matrix diag(0.1,0.1,0.05,0.1), and the observation noise covariance R can be set as diag(1.0,1.0,0.5).

[0058] Next, by predicting step x t∣t−1 =Ax t−1 And update steps: Kalman gain K t =P t∣t−1 H T HP t∣t−1 H T +R) −1 The data is iteratively fused, with H being the observation matrix, to obtain the smoothed vehicle state.

[0059] Step S25: The real-time availability and load capacity of the fused vehicle business data are evaluated using a logistic regression algorithm to obtain the vehicle capacity assessment result.

[0060] Preferably, the real-time availability and load capacity of the vehicle are evaluated using a logistic regression algorithm. The feature vector includes vehicle position stability, load change rate, and time window compliance, represented by speed variance. The labels are binary, with 1 for available and 0 for unavailable.

[0061] Preferably, the calculation process of the logistic regression algorithm is as follows: ① Feature standardization: Use Min-Max standardization to scale features to the [0,1] interval.

[0062] ② Model Training: Train the logistic regression model on historical data, using cross-entropy loss as the loss function. L=−N1∑i=1N[yilog(pi)+(1−yi)log(1−pi)], the optimizer uses gradient descent and the learning rate is set to 0.01.

[0063] ③ Evaluation threshold: The probability threshold is set to 0.6. If the predicted probability is higher than the threshold, the vehicle is available. The load capacity assessment is based on the regression output, and the error threshold is set to 5%.

[0064] Step S26: Organize the vehicle capacity assessment results into a structured database format to obtain a structured capacity resource pool.

[0065] Preferably, the structured database format can be used to construct the capacity resource pool using a relational database (such as MySQL). The table structure includes a vehicle table (fields: vehicle ID, real-time location, availability status, load capacity), an order table (fields: order ID, demand information), and a relational table (fields: timestamp, vehicle ID, order ID). The evaluation results are stored in partitions by vehicle ID, and an index is created to optimize query efficiency. The data update frequency is set to once every 30 seconds.

[0066] Beneficial effects of steps S21 to S26: This section constructs a high-quality structured transportation capacity resource pool through a systematic data fusion and processing process, providing core data support for subsequent intelligent scheduling decisions. Specifically, step S21, business data cleaning, ensures the accuracy and completeness of order and inventory information, eliminating noise and biases in the original data; step S22, time series alignment, synchronizes BeiDou trajectory data and business data in the time dimension, establishing a unified benchmark for multi-source data association; step S23, based on vehicle unique identifier association rule matching, effectively links discrete data sources, forming a preliminary integrated dataset; step S24, applies the Kalman filter algorithm to deeply fuse trajectory and business data, improving data consistency in the spatiotemporal dimension; step S25, uses the logistic regression algorithm to quantitatively evaluate vehicle real-time availability and load capacity, achieving standardized measurement of transportation resources; and step S26, organizes the evaluation results into a structured database, forming a transportation resource pool that can be directly called by the scheduling model. These steps gradually break down data silos, improve data interoperability, and construct a dynamic data system that accurately reflects the correlation between vehicle status, location information, and order demand, providing unified and reliable data input for subsequent stages such as vehicle route optimization and real-time scheduling, ensuring that the scheduling system makes decisions and optimizes based on real and consistent business status.

[0067] Step S3: Establish a dual-objective heterogeneous vehicle route optimization model that minimizes transportation costs and carbon emissions based on the structured transportation capacity resource pool, and constrain the dual-objective heterogeneous vehicle route optimization model by the maximum vehicle load and customer time window.

[0068] Further, step S3 involves establishing a dual-objective heterogeneous vehicle route optimization model that minimizes transportation costs and carbon emissions based on a structured transportation capacity resource pool. This model is constrained by vehicle maximum load capacity and customer time windows, and includes the following steps: Step S31: Extract the vehicle model, vehicle load capacity, and current location information from the structured transportation capacity resource pool to construct a heterogeneous vehicle feature set.

[0069] Preferably, static and dynamic vehicle data are extracted from a structured transportation capacity resource pool (such as a MySQL database) using SQL queries. Key fields include vehicle ID, vehicle type (such as medium-sized van, heavy-duty semi-trailer tractor), rated load capacity (unit: tons), and current latitude and longitude coordinates (obtained from BeiDou data).

[0070] Specifically, the feature engineering for the heterogeneous vehicle feature set is as follows: ① Vehicle Classification Coding: Based on load capacity, vehicles are discretized into M categories (the value of M is set according to the number of vehicle types). Load threshold ranges are set, for example: medium-sized vans (load capacity ≤ 13 tons), heavy-duty semi-trailer tractors (13 tons < load capacity ≤ 26.6 tons), etc. Load thresholds are set based on historical data distribution, and quantile methods (such as 25%, 50%, and 75th percentiles) are used to determine category boundaries.

[0071] ② Location feature processing: The current location data is used to calculate the Euclidean distance to the warehouse using the Haversine formula. The distance threshold is set to 100 kilometers. If the vehicle is more than 100 kilometers away from the warehouse, it is marked as unavailable.

[0072] ③ Feature vectorization: Each vehicle is represented as a feature vector v=[vehicle type code, load, current position X, current position Y, speed], where numerical features are standardized using Z-score.

[0073] Step S32: Obtain external order requirements and determine customer location, customer time window, and cargo weight and volume based on external order requirements, and integrate them to obtain customer demand feature vector.

[0074] Preferably, order data can be extracted in batches from an external order system (such as an ERP system) via an API interface. Fields include order ID, customer latitude and longitude, demand time window, and cargo weight and volume.

[0075] Preferably, the Haversine formula is also used to calculate the distance d between the customer and the warehouse. The distance threshold can be set to 500 kilometers. Orders exceeding the threshold need to be handled specially. The time window is converted into a standardized feature, such as a time window urgency score. The time window width threshold for the customer's time window is set to 2 hours. If it is greater than 2 hours, it is considered a loose time window. Otherwise, it is a tight time window. The time window urgency score is obtained by taking the reciprocal of the customer's time window. The score threshold is set to 0.5 to correspond to the 2-hour window.

[0076] Preferably, the demand vector is constructed by representing each order as a feature vector d=[customer X, customer Y, ρ, weight, volume], where weight and volume are transformed by logarithmic transformation log(1+x).

[0077] Step S33: Using the minimization of transportation costs as the main objective function, the customer demand feature vector is input into the main objective function to calculate vehicle start-up cost, travel distance cost, and time cost.

[0078] Preferably, the main objective function is defined using a linear combination model to minimize transportation costs. The main costs include vehicle startup costs, travel distance costs, and time costs, and are represented by the main objective function as follows: .

[0079] Where, x ijk To indicate whether vehicle k travels from i to j, set the value to 1 if yes, otherwise set it to 0; y jk For vehicle k, the value is 1 if it serves customer j, and 0 otherwise; C u Cost per unit distance, for example, 0.5 yuan / km; C k The start-up cost, for example, is 500 to 1000 yuan, which is adjusted in real time according to the type of transport vehicle.

[0080] Preferably, the cost parameters can be determined based on historical data regression analysis, with the travel distance cost coefficient C... u The time cost weight can be set to 0.1 yuan / minute by fitting actual fuel consumption data through linear regression.

[0081] Preferably, the gradient descent algorithm can be used to optimize the cost weights, the loss function is the mean squared error (MSE), and the threshold for the number of iterations can be set to 1000.

[0082] Step S34: With minimizing carbon emissions as the secondary objective function, calculate carbon emissions based on the heterogeneous vehicle feature set.

[0083] Preferably, carbon emissions can be derived using a fuel consumption calculation model. Assuming the vehicle travels at a constant speed (v = 40 km / h) with zero acceleration, the simplified fuel consumption formula is: .

[0084] Where λ is the fuel consumption coefficient, for example, 0.17L / km for a medium-sized van; Mm is the vehicle's weight; q is the load capacity; γ is the road coefficient, defaulting to 1; and carbon emissions are... ε is the carbon emission coefficient, ε=2.6765kg / L.

[0085] Next, we define the secondary objective function as minimizing total carbon emissions, then: .

[0086] The carbon emission threshold can be set to 0.05 kg per kilometer, and paths exceeding the threshold need to be optimized.

[0087] Step S35: The primary objective function, secondary objective function, vehicle maximum load, and customer time window are reconstructed into a dual-objective heterogeneous vehicle path optimization model.

[0088] Preferably, the dual objectives can be integrated into a weighted sum model or a Pareto optimization model, and the ε-constraint method can be used to handle the multiple objectives, transforming the carbon emission objective into a constraint: F2≤δ, where δ is the upper limit of carbon emissions, which can be set to 90% of the historical average of total carbon emissions. Then the objective function of the model is minF=w1×F1+w2×F2, where the weights w1 and w2 are set by the analytic hierarchy process (AHP), and the default values ​​can be set to w1=0.7, w2=0.3.

[0089] Beneficial effects of steps S31 to S35: This section establishes a dual-objective optimization framework for cigarette logistics capacity scheduling through a systematic model building process. Specifically, step S31 extracts a heterogeneous set of vehicle features to accurately capture the diversity of vehicle models, load capacities, and current locations, providing a realistic capacity resource foundation for the scheduling model; step S32 integrates customer demand feature vectors, clarifying customer locations, time windows, and cargo attributes to define specific input constraints for the optimization problem; step S33 prioritizes minimizing transportation costs, quantifying vehicle start-up costs, travel distance costs, and time costs to ensure the economic feasibility of the scheduling plan; step S34 introduces minimizing carbon emissions as a secondary objective, calculating the environmental impact based on vehicle characteristics to promote green logistics development; and step S35 reconstructs the dual-objective function with load and time window constraints into a complete optimization model, achieving scientific decision-making under multi-objective trade-offs. These steps work synergistically, enabling the scheduling model to simultaneously optimize transportation costs and carbon emissions, improve resource utilization, reduce operating expenses, and enhance the environmental friendliness of the plan, providing a core optimization engine for intelligent scheduling.

[0090] Step S4: The dual-objective heterogeneous vehicle path optimization model is transformed into a multi-objective mixed integer programming problem, and the multi-objective mixed integer programming problem is solved by a non-dominated sorting genetic algorithm to obtain the Pareto optimal scheduling scheme.

[0091] Further, in step S4, the dual-objective heterogeneous vehicle routing optimization model is transformed into a multi-objective mixed-integer programming problem, and the multi-objective mixed-integer programming problem is solved using a non-dominated sorting genetic algorithm to obtain a Pareto optimal scheduling scheme, including the following steps: Step S41: Discretize the continuous variables in the dual-objective heterogeneous vehicle path optimization model, define the dual objective function and constraints of the dual-objective heterogeneous vehicle path optimization model as linear expressions, and integrate all linear expressions to obtain a multi-objective mixed integer programming problem.

[0092] Preferably, discretization can discretize continuous variables (such as vehicle position coordinates and speed) in a dual-objective heterogeneous vehicle path optimization model into integer variables. The discretization granularity is set to 0.1 kilometers, that is, the position coordinates are retained to one decimal place and converted into integers (unit: 10 meters) by rounding to reduce the solution complexity.

[0093] Preferably, the linear expression transformation can convert the nonlinear objective function and constraints into a linear form. For example, as mentioned above, the objective functions F1 and F2 are linearized using a weighted sum method: minF = w1F1 + w2F2, with weights w1 and w2 set to 0.7 and 0.3, respectively.

[0094] Step S42: Perform chromosome encoding on the multi-objective mixed integer programming problem to obtain the initial gene population.

[0095] Preferably, the chromosome encoding scheme can use path representation for chromosome encoding, where each chromosome represents a complete vehicle path scheme. The chromosome length is N+K (N is the number of customers, K is the number of vehicles), and each gene position represents a customer node or vehicle starting point. For example, a chromosome sequence such as [0,1,2,0,3,4] indicates that a vehicle starts from warehouse 0, serves customers 1 and 2, returns to the warehouse, and then serves customers 3 and 4.

[0096] Preferably, the initial population can be generated using a random generation strategy, with a population size of 100. Constraints are applied during generation to ensure that each customer is served only once and that vehicle weight does not exceed limits. Feasible solutions are initialized using the scanning algorithm mentioned in the documentation to improve population quality.

[0097] Step S43: Based on the initial gene population, perform fast non-dominated sorting using a non-dominated sorting genetic algorithm, and divide the individuals in the initial gene population into different frontier levels according to the Pareto rank.

[0098] Preferably, the fast non-dominated sorting algorithm can be implemented using the fast non-dominated sorting process of NSGA-II. First, calculate the dominance count and dominance set for each individual: for individual i, the dominance count n... i This represents the number of individuals that dominate i, and the dominance set Si stores the individuals dominated by i. During sorting, the threshold for dividing the frontier hierarchy is based on the Pareto rank: the first frontier consists of non-dominated individuals (ni=0), and subsequent frontiers are updated iteratively.

[0099] Preferably, the population can be traversed and the space complexity optimized to O(MN). 2 ), where M is the target number, N is the population size, and the maximum frontier number is set to 10, with any excess being merged.

[0100] Step S44: Calculate the crowding distance for individuals in the stratified population, and combine it with the elite retention strategy to select high-quality individuals to enter the next generation of evolution.

[0101] Preferably, the crowding distance calculation can be performed by sorting individuals in each frontal layer according to each objective function value and calculating the crowding distance: .

[0102] Among them, f m The objective function value is M=2, representing transportation costs and carbon emissions. After distance standardization, the congestion threshold can be set to 1.5 times the average value.

[0103] Preferably, the elite retention strategy can combine crowding distance and Pareto level to select the top 50% of individuals to enter the next generation, i.e., the elite ratio is set to 50% of the population size.

[0104] Step S45: Perform simulated binary crossover and polynomial mutation on all high-quality individuals in the population to obtain a new generation of individuals.

[0105] Preferably, the simulated binary crossover can be performed using the SBX algorithm, and the crossover distribution index can be set to 10. The offspring generation formula is c1=0.5×[(1+β)p1+(1−β)p2], where β is a random factor, and the crossover probability can be set to 0.9.

[0106] Preferably, the variation distribution exponent of the polynomial mutation can be set to 20, and the mutation probability can be set to 0.1. The variable asynchronous length is given by the formula δ=min(1,∣2u−1∣ 1 / (20+1) Calculate ∫-1∞ u, where u is a uniformly random number in the range [0,1].

[0107] Step S46: Perform fast non-dominated sorting and crowding distance calculation on all new generation population individuals again to iteratively update the Pareto solution set.

[0108] Preferably, during the iteration process, the non-dominated sorting and crowding degree calculation are re-performed in each generation to update the Pareto solution set. The upper limit of the solution set size can be set to 100, and individuals with low crowding degree are removed when the size exceeds this limit.

[0109] Meanwhile, the rate of change of the solution set is monitored. If the improvement of the optimal solution is less than 0.0001 for 10 consecutive generations, the early stopping mechanism is triggered.

[0110] Step S47: When the maximum number of iterations is reached or the quality of the Pareto solution set is stable, terminate the iteration and output the Pareto optimal scheduling scheme.

[0111] Preferably, the maximum number of iterations is set to 1000, or the solution set quality stability threshold is 0.0001 for 10 consecutive generations.

[0112] Preferably, the Pareto optimal solution set is stored in list form, with each solution containing a vehicle route sequence, cost value, and carbon emission value.

[0113] Beneficial effects of steps S41 to S47: This section achieves efficient solutions to multi-objective optimization problems through a systematic algorithmic process, providing a scientific decision-making basis for cigarette logistics capacity scheduling. Specifically, step S41 discretizes the continuous variables in the dual-objective heterogeneous vehicle routing optimization model and transforms them into linear expressions, constructing a standardized multi-objective mixed integer programming problem, thus adapting complex problems to an efficient solution framework; step S42 generates an initial gene population through chromosome encoding, laying a diverse starting point for the algorithm search; step S43 applies a non-dominated sorting genetic algorithm for rapid stratification, identifying the Pareto rank of the solution set and distinguishing the quality levels of the set; step S44 calculates the crowding distance and combines it with an elite strategy to select high-quality individuals, balancing the diversity and convergence of the solution set; step S45 generates a new generation of population through simulated binary crossover and polynomial mutation operations, promoting global exploration and local optimization of the solution space; step S46 iteratively updates the Pareto solution set, gradually improving the quality of the solution; and step S47 outputs a stable Pareto optimal solution based on a preset termination condition, ensuring reliable convergence of the algorithm. These steps work together to transform the multi-objective optimization problem into a computable form, generating a set of scheduling schemes that balance transportation costs and carbon emissions through intelligent algorithms. This enhances the flexibility and robustness of scheduling decisions, provides high-quality input for subsequent route adjustments, and improves overall optimization efficiency.

[0114] Step S5: Select the Pareto optimal solution with the minimum total transportation cost, and dynamically adjust the routes of the transport vehicles based on the real-time location data of the transport vehicles using a global optimization algorithm to obtain the final scheduling solution for the transport vehicles.

[0115] Further, in step S5, the Pareto optimal solution with the minimum total transportation cost is selected, and the transportation vehicles are dynamically routed using a global optimization algorithm based on their real-time location data to obtain the final scheduling solution for the transportation vehicles. This includes the following steps: Step S51: Select the scheme with the minimum total transportation cost from the Pareto optimal scheduling schemes as the basic scheduling scheme.

[0116] Preferably, cost and carbon emission data for all non-dominated solutions are extracted from the Pareto optimal solution set, and a weighted sum method is used to calculate the comprehensive score for each solution. The weights can be set as cost weight w1 = 0.7 and carbon emission weight w2 = 0.3, and the comprehensive score formula is also S = w1F1 + w2F2. The solution with the lowest score is selected as the basic scheduling scheme; if multiple solutions have the same score, the solution with the lower carbon emission is preferred.

[0117] Step S52: Extract the vehicle route sequence and order allocation information from the basic scheduling scheme to obtain the initial route plan.

[0118] Preferably, the structured output of the basic scheduling scheme (such as JSON format) is parsed to extract key fields: vehicle route sequence (ordered list of customer nodes) and order allocation information (order ID, cargo weight, volume). The route sequence is stored in the form of a directed graph, where nodes represent customer locations, edges represent driving paths, and edge weights are distances (in kilometers).

[0119] Preferably, the route sequence can be converted into a time-window-constrained scheduling plan, with each node having an attached service time window. The service time can be set to a fixed value of 10 minutes per node. The initial route plan output is in Gantt chart format, including vehicle departure time, node service time, and estimated arrival time.

[0120] Step S53: Obtain real-time location data of the transport vehicle based on the BeiDou satellite navigation system and align it with the initial route plan in time and space.

[0121] Preferably, real-time location data can also be obtained in real time through the BeiDou Navigation Satellite System API interface, including latitude and longitude coordinates (WGS-84 coordinate system), timestamp (UTC time), and velocity. The sampling frequency can be set to 1 Hz, and the data format conforms to the NMEA-0183 standard.

[0122] Preferably, the alignment algorithm can also use a dynamic time warping algorithm to align the real-time position sequence with the time axis of the initial path plan. The alignment accuracy threshold can be set to ±30 seconds. If the time deviation exceeds the threshold, linear interpolation is used to adjust the planned time point. Spatial alignment is calculated using the Haversine formula to determine the spherical distance between the actual position and the planned node. The distance threshold can be set to 100 meters. If the distance exceeds the threshold, it is marked as a deviation event.

[0123] Step S54: Input the aligned real-time position data and the initial path plan into the particle swarm optimization algorithm, perform global optimization with the goal of minimizing path deviation cost, and iterate to obtain the adjusted path scheme.

[0124] Preferably, minimizing the path deviation cost can be characterized as C 偏离 =α×d 总偏离 +β×Δt 延迟 , where d 总偏离 Δt represents the cumulative distance difference between the actual path and the planned path. 延迟 The total delay duration can be represented by weighting coefficients α=0.6 and β=0.4.

[0125] Preferably, the particle swarm size is set to 50, the dimension is the number of path nodes, the position vector represents the node access order, the inertia weight of the particle swarm optimization algorithm is set to 0.7, and the learning factor is 1.5.

[0126] In step S55, the 2-opt local search algorithm eliminates intersecting paths in the adjusted route scheme and shortens the travel distance, thus obtaining the final scheduling scheme for the transport vehicles.

[0127] Preferably, the 2-opt local search algorithm eliminates intersecting paths and shortens the total travel distance by swapping two edges in the path. It iterates through all node pairs (i,j) in the path, swaps edges (i,i+1) and (j,j+1) to (i,j) and (i+1,j+1), and calculates the distance of the new path.

[0128] Beneficial effects of steps S51 to S55: This module significantly enhances the adaptability and execution efficiency of the scheduling scheme to the real-time operating environment through a systematic dynamic path adjustment process. Specifically, step S51 selects the most economically optimal basic scheduling scheme from the Pareto optimal solution set, providing a high-quality initial decision-making basis for dynamic adjustment; step S52 extracts the path sequence and order allocation information of the scheme to form an operable initial path plan; step S53 establishes a mapping relationship between the actual location and the predetermined route by aligning BeiDou positioning data with the planned path in time and space, providing an accurate time and space reference for path optimization; step S54 applies the particle swarm optimization algorithm to perform global optimization with the goal of minimizing path deviation cost, enabling rapid response to emergencies and path replanning; step S55 uses the 2-opt local search algorithm to fine-tune the optimized path, eliminating path intersections and shortening the total travel distance, further improving the economy and feasibility of the scheme. These steps gradually transform static optimization schemes into dynamic execution paths that adapt to real-time conditions. Through algorithmic collaborative optimization, the system enhances the vehicle's adaptability to dynamic factors such as traffic changes and order adjustments, reduces the deviation between actual driving and planned routes, lowers additional costs caused by environmental changes, and strengthens the real-time decision-making capabilities and overall robustness of the scheduling system.

[0129] Step S6: Send the final scheduling plan to the transport vehicles and the external monitoring terminal.

[0130] Beneficial effects of steps S1 to S6: This module achieves a systematic and intelligent upgrade to the cigarette logistics capacity scheduling process. Specifically, step S1, trajectory data cleaning and preprocessing, ensures the accuracy and integrity of BeiDou positioning information, laying a reliable data foundation for subsequent analysis; step S2, multi-source data fusion, constructs a structured capacity resource pool, integrating vehicle status and business needs, enhancing the information support capability for scheduling decisions; step S3, a dual-objective optimization model, balances transportation costs and carbon emissions while meeting load and time window constraints, promoting the coordinated development of logistics operations towards economic efficiency and environmental protection; step S4, intelligent algorithms solve for Pareto optimal solutions, providing scientific scheduling choices under multi-objective trade-offs; step S5, dynamic path adjustment, optimizes vehicle routes based on real-time location data, improving the adaptability and execution efficiency of scheduling schemes; and step S6, a scheme distribution mechanism, ensures that scheduling instructions are promptly transmitted to the execution and monitoring ends, achieving closed-loop management. Overall, these steps work together to significantly optimize vehicle resource utilization, reduce empty runs and waiting time, lower fuel consumption and labor costs, and enhance the safety of the transportation process and customer service levels through precise monitoring and rapid response, providing core technological support for the efficient and low-carbon operation of cigarette logistics.

[0131] Further, in step S6, the final scheduling plan is sent to the transport vehicles and the external monitoring terminal. This is followed by the following steps: Step S10: The execution status of the final scheduling scheme and the information of newly arrived orders are monitored in real time using a sliding time window algorithm to obtain a real-time monitoring dataset.

[0132] Preferably, a sliding time window algorithm can be used to collect the execution status of the final scheduling scheme and information on newly arrived orders in real time. The window size can be set to 10 minutes, and the window sliding step is 1 minute to ensure data continuity and real-time performance. The monitoring data includes vehicle location (obtained from the BeiDou system, with the sampling frequency dynamically adjusted: 1Hz for low speeds <20km / h and 0.2Hz for medium and high speeds ≥20km / h), driving speed, order completion progress (such as the number of customers served and the remaining distance), and the weight of goods and time window requirements for new orders.

[0133] Step S20: Calculate the deviation of key performance indicators of the dual-objective heterogeneous vehicle path optimization model based on the real-time monitoring dataset. When the deviation of key performance indicators exceeds the preset deviation threshold, trigger the model re-solution signal.

[0134] Preferably, the key performance indicators can be based on the objective function of a dual-objective heterogeneous vehicle routing optimization model, including total transportation cost F1 and carbon emissions F2. The actual values ​​are calculated from the monitoring dataset, while the planned values ​​are extracted from the initial scheduling scheme.

[0135] Preferably, the deviation calculation model then uses a percentage error formula to calculate the deviation of each indicator, with a preset deviation threshold set to 5%. If D > 5%, a model re-solution signal is triggered; otherwise, monitoring continues. The calculation process is executed every 5 minutes, updated using real-time data streams. Trigger signals include deviation type (such as cost overrun or carbon emission exceedance) and severity level.

[0136] Step S30: Reconstruct the dual-objective heterogeneous vehicle path optimization model for the current time period based on the model re-solution signal to obtain the updated optimization model.

[0137] Preferably, the reconstruction method can be performed by repeating steps S1 to S5.

[0138] Step S40: Repeat steps S4 to S5 with the updated optimized model as the subject of execution to obtain the updated scheduling scheme.

[0139] Step S50: Send the updated scheduling plan to the transport vehicles and the external monitoring terminal.

[0140] Beneficial effects of steps S10 to S50: This section enhances the real-time response capability and continuous optimization level of the cigarette logistics capacity scheduling system by constructing a dynamic closed-loop optimization mechanism. Specifically, step S10's sliding time window monitoring mechanism continuously tracks the execution status of the scheduling plan and new order information, forming a real-time data feedback loop to provide precise input for dynamic system adjustments; step S20's key performance indicator deviation detection function establishes quantitative evaluation standards, automatically triggering a re-solution mechanism when the actual operating status deviates from the expected target, ensuring the scheduling system's sensitivity to abnormal operating conditions; step S30's model reconstruction capability rapidly adjusts and optimizes model parameters based on real-time data, enabling the scheduling strategy to adapt to changes in the transportation environment; and step S40's iterative solution process generates updated solutions by reusing core algorithms. This ensures technical consistency in the dynamic adjustment process; the scheme distribution mechanism in step S50 synchronizes the optimization results to the execution terminal in real time, forming a complete closed loop of decision-making-execution-feedback; the synergistic effect of this series of steps enables the scheduling system to have the ability to quickly respond to dynamic factors such as vehicle status fluctuations, new order tasks, and changes in traffic conditions during transportation. By continuously calibrating the scheduling scheme, the deviation between the plan and the actual situation is reduced, vehicle travel routes and resource utilization are optimized, and additional costs caused by environmental changes are reduced. At the same time, the closed-loop control improves the system's anti-interference ability and operational stability, providing continuous optimization and dynamic scheduling support for cigarette logistics.

[0141] See Figure 2 This embodiment provides an example of a transportation capacity scheduling device for cigarette logistics. In this embodiment, the transportation capacity scheduling device is applied to the transportation capacity scheduling method as described in the above embodiment.

[0142] Specifically, the capacity scheduling device includes a BeiDou trajectory data processing module 1, a capacity resource pool construction module 2, an optimization model construction module 3, an optimization model solving module 4, a dynamic path adjustment module 5, and a final scheduling scheme sending module 6, which are electrically or signalally connected in sequence.

[0143] The system comprises three modules: a BeiDou trajectory data processing module 1, a capacity resource pool construction module 2, and an optimization model construction module 3. The BeiDou trajectory data processing module 1 collects real-time trajectory data of transport vehicles using the BeiDou satellite navigation system, and cleans and preprocesses the trajectory data to obtain processed BeiDou trajectory data. The capacity resource pool construction module 2 integrates and aligns the processed BeiDou trajectory data with order and inventory information from cigarette logistics through data fusion to obtain a structured capacity resource pool. The optimization model construction module 3 establishes a dual-objective heterogeneous vehicle route optimization model based on the structured capacity resource pool, minimizing transportation costs and carbon emissions. It also constrains the dual-objective heterogeneous vehicles by limiting their maximum load capacity and customer time windows. The system includes a path optimization model; the optimization model solving module 4 transforms the dual-objective heterogeneous vehicle path optimization model into a multi-objective mixed integer programming problem, and solves the multi-objective mixed integer programming problem using a non-dominated sorting genetic algorithm to obtain a Pareto optimal scheduling scheme; the dynamic path adjustment module 5 selects the Pareto optimal scheme with the minimum total transportation cost, and dynamically adjusts the transportation vehicles' paths based on the real-time location data of the transportation vehicles using a global optimization algorithm to obtain the final scheduling scheme for the transportation vehicles; and the final scheduling scheme sending module 6 sends the final scheduling scheme to the transportation vehicles and external monitoring terminals.

[0144] It should be noted that this embodiment is a functional module embodiment based on the above method embodiment. For additional content such as extensions, optimizations, limitations, examples, principle explanations, and beneficial effects of this embodiment, please refer to the above embodiments. This embodiment will not repeat them here.

[0145] Figure 3 This is a schematic diagram of the structure of an electronic device according to an embodiment of this application. Figure 3 As shown, the electronic device 7 includes a processor 71 and a memory 72 coupled to the processor 71.

[0146] The memory 72 stores program instructions for implementing the federated learning-based collaborative energy-saving method for government data clusters in any of the above embodiments.

[0147] The processor 71 is used to execute program instructions stored in the memory 72 for collaborative energy saving of government data clusters based on federated learning.

[0148] The processor 71 can also be referred to as a CPU (Central Processing Unit). The processor 71 may be an integrated circuit chip with signal processing capabilities. The processor 71 can also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. A general-purpose processor can be a microprocessor or any conventional processor.

[0149] Furthermore, Figure 4 This is a schematic diagram of the structure of a storage medium according to an embodiment of this application. See also: Figure 4 In this embodiment of the application, the storage medium 8 stores program instructions 81 capable of implementing all the above methods. These program instructions 81 can be stored in the storage medium in the form of a software product, including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks, or terminal devices such as computers, servers, mobile phones, and tablets.

[0150] In the several embodiments provided in this application, it should be understood that the disclosed systems, methods, and approaches can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between systems or units may be electrical, mechanical, signal, or other forms.

[0151] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated units described above can be implemented in hardware or as software functional units. The above are merely embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made based on the description and drawings of this application, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. A method for scheduling transportation capacity in cigarette logistics, wherein the method is applied to transport vehicles used in cigarette logistics, the transport vehicles being equipped with a BeiDou satellite navigation system, characterized in that, The capacity scheduling method includes: Step S1: The trajectory data of the transport vehicle is collected in real time through the BeiDou satellite navigation system, and the trajectory data is cleaned and preprocessed to obtain processed BeiDou trajectory data; Step S2: The processed Beidou trajectory data is associated and aligned with the order information and inventory information of cigarette logistics through data fusion to obtain a structured transportation capacity resource pool; Step S3: Establish a dual-objective heterogeneous vehicle route optimization model that minimizes transportation costs and carbon emissions based on a structured transportation capacity resource pool, and constrain the dual-objective heterogeneous vehicle route optimization model by the maximum vehicle load and customer time window. Step S4: The dual-objective heterogeneous vehicle path optimization model is transformed into a multi-objective mixed integer programming problem, and the multi-objective mixed integer programming problem is solved by a non-dominated sorting genetic algorithm to obtain a Pareto optimal scheduling scheme. Step S5: Select the Pareto optimal solution with the minimum total transportation cost, and dynamically adjust the route of the transport vehicle based on the real-time location data of the transport vehicle using a global optimization algorithm to obtain the final scheduling solution of the transport vehicle. Step S6: Send the final scheduling plan to the transport vehicle and the external monitoring terminal.

2. The capacity scheduling method according to claim 1, characterized in that, Step S6: Send the final scheduling plan to the transport vehicle and the external monitoring terminal, followed by: Step S10: Monitor the execution status of the final scheduling scheme and the information of newly arrived orders in real time using a sliding time window algorithm to obtain a real-time monitoring dataset; Step S20: Calculate the key performance indicator deviation of the dual-objective heterogeneous vehicle path optimization model based on the real-time monitoring dataset. When the key performance indicator deviation exceeds a preset deviation threshold, trigger a model re-solution signal. Step S30: Reconstruct the dual-objective heterogeneous vehicle path optimization model for the current time period based on the model re-solution signal to obtain the updated optimization model; Step S40: Repeat steps S4 to S5 with the updated optimized model as the subject of execution to obtain the updated scheduling scheme; Step S50: Send the updated scheduling plan to the transport vehicle and the external monitoring terminal.

3. The capacity scheduling method according to claim 1, characterized in that, Step S1: The trajectory data of the transport vehicle is collected in real time through the BeiDou satellite navigation system, and the trajectory data is cleaned and preprocessed to obtain processed BeiDou trajectory data, including: Step S11: Obtain the original trajectory data stream of the transport vehicle in real time through the API interface of the Beidou satellite navigation system; Step S12: Perform preliminary filtering on the original trajectory data stream using a sliding window algorithm to obtain preliminary filtered trajectory data; Step S13: Identify and remove outliers and noise data from the preliminary filtered trajectory data using the isolated forest algorithm to obtain the anomaly-cleaned trajectory data. Step S14: Fill missing values ​​in the abnormal cleaned trajectory data by linear interpolation and smooth the trajectory by Kalman filtering algorithm to obtain continuous and complete trajectory data, which is the processed BeiDou trajectory data.

4. The capacity scheduling method according to claim 1, characterized in that, Step S2 involves associating and aligning the processed BeiDou trajectory data with order and inventory information from cigarette logistics through data fusion to obtain a structured transportation capacity resource pool, including: Step S21: Extract order information and inventory information from the external cigarette logistics information system, and perform data cleaning processing on missing values ​​and outliers to obtain preprocessed business data; Step S22: Align the preprocessed business data with the processed BeiDou trajectory data in terms of time series, and integrate them to obtain time-synchronized multi-source data; Step S23: Based on the unique identifier of the transport vehicle, perform association rule matching on the time-synchronized multi-source data to obtain a preliminary association dataset; Step S24: The trajectory data and business data in the preliminary associated dataset are fused using the Kalman filter algorithm to obtain fused vehicle business data. Step S25: Evaluate the real-time availability and load capacity of the fused vehicle business data using a logistic regression algorithm to obtain the vehicle capacity evaluation result; Step S26: Organize the vehicle capacity assessment results into a structured database format to obtain the structured capacity resource pool.

5. The capacity scheduling method according to claim 1, characterized in that, Step S3: Establish a dual-objective heterogeneous vehicle route optimization model based on the structured transportation capacity resource pool to minimize transportation costs and carbon emissions. Constrain the dual-objective heterogeneous vehicle route optimization model by the vehicle's maximum load capacity and the customer's time window, including: Step S31: Extract the vehicle model, vehicle load capacity, and current location information from the structured transportation capacity resource pool to construct a heterogeneous vehicle feature set; Step S32: Obtain external order requirements and determine customer location, customer time window, and cargo weight and volume based on the external order requirements, and integrate them to obtain a customer demand feature vector; Step S33: Using the minimization of transportation costs as the main objective function, the customer demand feature vector is input into the main objective function to calculate vehicle start-up cost, travel distance cost, and time cost; Step S34: Using the minimization of carbon emissions as a secondary objective function, calculate the carbon emissions based on the heterogeneous vehicle feature set; Step S35: Reconstruct the primary objective function, the secondary objective function, the maximum vehicle load, and the customer time window into a dual-objective heterogeneous vehicle path optimization model.

6. The capacity scheduling method according to claim 1, characterized in that, Step S4: The dual-objective heterogeneous vehicle path optimization model is transformed into a multi-objective mixed integer programming problem, and the multi-objective mixed integer programming problem is solved using a non-dominated sorting genetic algorithm to obtain a Pareto optimal scheduling scheme, including: Step S41: Discretize the continuous variables in the dual-objective heterogeneous vehicle path optimization model, define the dual objective function and constraints of the dual-objective heterogeneous vehicle path optimization model as linear expressions, and integrate all linear expressions to obtain a multi-objective mixed integer programming problem. Step S42: Perform chromosome encoding on the multi-objective mixed integer programming problem to obtain the initial gene population; Step S43: Based on the initial gene population, perform fast non-dominated sorting using a non-dominated sorting genetic algorithm, and divide the individuals in the initial gene population into different frontier levels according to the Pareto level. Step S44: Calculate the crowding distance for individuals in the stratified population, and combine it with the elite retention strategy to select high-quality individuals to enter the next generation of evolution. Step S45: Perform simulated binary crossover and polynomial mutation on all high-quality individuals in the population to obtain a new generation of individuals. Step S46: Perform fast non-dominated sorting and crowding distance calculation on all new generation population individuals again to iteratively update the Pareto solution set; Step S47: When the maximum number of iterations is reached or the quality of the Pareto solution set is stable, terminate the iteration and output the Pareto optimal scheduling scheme.

7. The capacity scheduling method according to claim 1, characterized in that, Step S5: Select the Pareto optimal solution with the minimum total transportation cost, and dynamically adjust the routes of the transport vehicles based on their real-time location data using a global optimization algorithm to obtain the final scheduling scheme for the transport vehicles, including: Step S51: Select the scheme with the minimum total transportation cost from the Pareto optimal scheduling schemes as the basic scheduling scheme; Step S52: Extract the vehicle route sequence and order allocation information from the basic scheduling scheme to obtain the initial route plan; Step S53: Obtain the real-time location data of the transport vehicle based on the BeiDou satellite navigation system, and align it with the initial route plan in time and space; Step S54: The aligned real-time position data and the initial path plan are input into the particle swarm optimization algorithm to perform global optimization with the goal of minimizing path deviation cost, and the adjusted path scheme is obtained iteratively. Step S55: The 2-opt local search algorithm eliminates the intersecting paths of the adjusted route scheme and shortens the travel distance to obtain the final scheduling scheme of the transport vehicles.

8. A capacity scheduling device for cigarette logistics, wherein the capacity scheduling device is applied to the capacity scheduling method as described in any one of claims 1 to 7, characterized in that, The transportation capacity scheduling device includes: The BeiDou trajectory data processing module is used to collect the trajectory data of the transport vehicle in real time through the BeiDou satellite navigation system, and to clean and preprocess the trajectory data to obtain processed BeiDou trajectory data. The capacity resource pool construction module is used to associate and align the processed Beidou trajectory data with the order information and inventory information of cigarette logistics through data fusion to obtain a structured capacity resource pool; The optimization model building module is used to establish a dual-objective heterogeneous vehicle route optimization model that minimizes transportation costs and carbon emissions based on a structured transportation capacity resource pool, and to constrain the dual-objective heterogeneous vehicle route optimization model by the maximum vehicle load and the customer time window. The optimization model solving module is used to transform the dual-objective heterogeneous vehicle path optimization model into a multi-objective mixed integer programming problem, and solve the multi-objective mixed integer programming problem using a non-dominated sorting genetic algorithm to obtain a Pareto optimal scheduling scheme. The dynamic route adjustment module is used to select the Pareto optimal solution with the minimum total transportation cost, and to dynamically adjust the transportation vehicle's route based on the real-time location data of the transportation vehicle using a global optimization algorithm to obtain the final scheduling solution for the transportation vehicle. The final scheduling scheme sending module is used to send the final scheduling scheme to the transport vehicle and the external monitoring terminal.

9. An electronic device, characterized in that, The method includes a processor and a memory coupled to the processor, the memory storing program instructions executable by the processor; when the processor executes the program instructions stored in the memory, it implements the capacity scheduling method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores program instructions that, when executed by a processor, enable the capacity scheduling method as described in any one of claims 1 to 7.

Images