Logistics vehicle scheduling method and system

Abstract

The invention relates to the technical field of artificial intelligence and logistics management, and particularly provides a logistics vehicle scheduling method and system, and the method comprises the following steps: S1, employing the big data analysis and machine learning technology to carry out the prediction of future order demands, and providing input for a scheduling engine; s2, the optimal scheduling engine generates an optimal scheduling scheme through multi-objective optimization and reinforcement learning; s3, adopting an Internet of Things technology to realize vehicle scheduling and dynamic adjustment; and S4, the cloud computing platform provides computing resources and distributed storage. Compared with the prior art, the real-time data can be collected through the Internet of Things technology, the complex scheduling problem can be efficiently solved by relying on the powerful computing capacity of cloud computing, and the high efficiency and flexibility of a scheduling scheme are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of artificial intelligence and logistics management, and specifically provides a logistics vehicle scheduling method and system. Background Art

[0002] The Vehicle Routing Problem (VRP) is a classic problem in the field of logistics and supply chain, which involves optimizing the configuration of vehicles, routes and orders under various constraints. Traditional scheduling methods include rule-driven optimization and single-objective linear programming, but when dealing with multi-objective problems (such as simultaneous optimization of transportation costs, delivery time and service levels), they often face problems such as high computational complexity and poor adaptability.

[0003] In recent years, multi-objective evolutionary algorithms (MOEA) such as NSGA-II have been widely used due to their high efficiency in solving multi-objective optimization problems; at the same time, deep reinforcement learning (such as PPO and A3C algorithms based on the Actor-Critic framework) provides new ideas for real-time optimization due to its adaptability in dynamic environments.

[0004] Internet of Things technology makes real-time monitoring of logistics vehicles possible, but existing solutions do not make sufficient use of real-time data. In addition, large-scale scheduling problems place higher demands on computing resources, and traditional local computing can hardly meet real-time requirements. The elastic expansion capability of cloud computing has become the key to solving this problem. Summary of the invention

[0005] The present invention aims at solving the above-mentioned deficiencies of the prior art and provides a logistics vehicle scheduling method with strong practicality.

[0006] A further technical task of the present invention is to provide a logistics vehicle dispatching system that is reasonably designed, safe and applicable.

[0007] The technical solution adopted by the present invention to solve its technical problem is:

[0008] A logistics vehicle scheduling method comprises the following steps:

[0009] S1. Use big data analysis and machine learning technology to predict future order demand and provide input for the scheduling engine;

[0010] S2, the optimization scheduling engine generates the optimal scheduling solution through multi-objective optimization and reinforcement learning;

[0011] S3. Use Internet of Things technology to achieve dynamic adjustment of vehicle dispatching;

[0012] S4. Cloud computing platform provides computing resources and distributed storage.

[0013] Furthermore, in step S1, it includes:

[0014] (1) Collect order history data, user behavior data, weather information, and traffic flow data for data collection;

[0015] (2) Perform data cleaning, normalization, feature selection and dimensionality reduction on the data;

[0016] (3) Select models based on different scenarios. Time series models are used for short-term forecasts, while regression models are used for long-term forecasts.

[0017] (4) Combining Bayesian optimization and grid search to adjust hyperparameters and perform prediction optimization;

[0018] (5) Generate order demand distribution within the time window.

[0019] Furthermore, in step S2, it includes:

[0020] (1) Model the problem and construct a multi-objective function that includes cost, time, and service level. Constraints include vehicle capacity, time window restrictions, and order priority.

[0021] (2) Generate an initial solution using coding techniques;

[0022] (3) Calculate the fitness of each solution based on non-dominated sorting;

[0023] (4) Apply crossover and mutation operators to perform legacy operations;

[0024] (5) Use deep reinforcement learning algorithms to dynamically adjust the solutions generated by NSGA-II;

[0025] (6) Generate a Pareto optimal solution that satisfies all constraints as the final scheduling solution.

[0026] Furthermore, in step S3, it includes:

[0027] (1) Use GPS modules, RFID sensors, and smart dashboards to collect data on vehicle location, load, and speed;

[0028] (2) Using the Dijkstra algorithm based on the shortest path, reinforcement learning is called in real time to re-optimize the scheduling plan based on the current state;

[0029] (3) Use MQTT protocol and edge computing technology to store monitoring data on the cloud computing platform for subsequent analysis by users.

[0030] Furthermore, in step S4, it includes:

[0031] (1) Using Hadoop framework for distributed computing;

[0032] (2) Using HDFS to store large-scale scheduling data;

[0033] (3) Use the MapReduce model for large-scale parallel task processing;

[0034] (4) Dynamically adjust computing resources to optimize model training and inference efficiency.

[0035] A logistics vehicle dispatching system, comprising a demand forecasting module, an optimization dispatching engine module, a real-time monitoring and feedback module and a cloud computing platform module;

[0036] The demand forecasting module is used to use big data analysis and machine learning technology to predict future order demand and provide input for the scheduling engine;

[0037] The optimization scheduling engine module is used for the optimization scheduling engine to generate the optimal scheduling solution through multi-objective optimization and reinforcement learning;

[0038] The real-time monitoring and feedback module is used to realize dynamic adjustment of vehicle scheduling and operation by adopting Internet of Things technology;

[0039] The cloud computing platform module is used to provide computing resources and distributed storage.

[0040] Furthermore, in the demand forecasting module, it includes:

[0041] (1) Collect order history data, user behavior data, weather information, and traffic flow data for data collection;

[0042] (2) Perform data cleaning, normalization, feature selection and dimensionality reduction on the data;

[0043] (3) Select models based on different scenarios. Time series models are used for short-term forecasts, while regression models are used for long-term forecasts.

[0044] (4) Combining Bayesian optimization and grid search to adjust hyperparameters and perform prediction optimization;

[0045] (5) Generate order demand distribution within the time window.

[0046] Furthermore, the optimization scheduling engine module includes:

[0047] (1) Model the problem and construct a multi-objective function that includes cost, time, and service level. Constraints include vehicle capacity, time window restrictions, and order priority.

[0048] (2) Generate an initial solution using coding techniques;

[0049] (3) Calculate the fitness of each solution based on non-dominated sorting;

[0050] (4) Apply crossover and mutation operators to perform legacy operations;

[0051] (5) Use deep reinforcement learning algorithms to dynamically adjust the solutions generated by NSGA-II;

[0052] (6) Generate a Pareto optimal solution that satisfies all constraints as the final scheduling solution.

[0053] Furthermore, in the real-time monitoring and feedback module, it includes:

[0054] (1) Use GPS modules, RFID sensors, and smart dashboards to collect data on vehicle location, load, and speed;

[0055] (2) Using the Dijkstra algorithm based on the shortest path, reinforcement learning is called in real time to re-optimize the scheduling plan based on the current state;

[0056] (3) Use MQTT protocol and edge computing technology to store monitoring data on the cloud computing platform for subsequent analysis by users.

[0057] Furthermore, the cloud computing platform module includes:

[0058] (1) Using Hadoop framework for distributed computing;

[0059] (2) Using HDFS to store large-scale scheduling data;

[0060] (3) Use the MapReduce model for large-scale parallel task processing;

[0061] (4) Dynamically adjust computing resources to optimize model training and inference efficiency.

[0062] Compared with the prior art, the logistics vehicle scheduling method and system of the present invention has the following outstanding beneficial effects:

[0063] The present invention achieves a multi-objective balance among transportation cost, delivery time and service level through NSGA-II and reinforcement learning technology; dynamically adjusts the scheduling plan by using real-time monitoring and feedback mechanism to adapt to the complex and changing logistics environment; relies on the cloud computing platform to support large-scale parallel computing and model training, significantly improving scheduling efficiency; and the application of Internet of Things technology makes the logistics network more intelligent and visualized, improving customer satisfaction and operational transparency. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0065] Attached Figure 1 It is a flow chart of a logistics vehicle scheduling method. DETAILED DESCRIPTION

[0066] In order to enable those skilled in the art to better understand the solution of the present invention, the present invention is further described in detail below in conjunction with specific implementation methods. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0067] A best embodiment is given below:

[0068] like Figure 1 As shown, a logistics vehicle scheduling method in this embodiment has the following steps:

[0069] S1. Use big data analysis and machine learning technology to predict future order demand and provide input for the scheduling engine;

[0070] include:

[0071] (1) Collect order history data, user behavior data, weather information, and traffic flow data for data collection;

[0072] (2) Perform data cleaning, normalization, feature selection and dimensionality reduction on the data;

[0073] (3) Select models based on different scenarios. Time series models are used for short-term forecasts, while regression models are used for long-term forecasts.

[0074] (4) Combining Bayesian optimization and grid search to adjust hyperparameters and perform prediction optimization;

[0075] (5) Generate order demand distribution within the time window.

[0076] S2, the optimization scheduling engine generates the optimal scheduling solution through multi-objective optimization and reinforcement learning;

[0077] include:

[0078] (1) Model the problem and construct a multi-objective function that includes cost, time, and service level. Constraints include vehicle capacity, time window restrictions, and order priority.

[0079] (2) Generate an initial solution using coding techniques;

[0080] (3) Calculate the fitness of each solution based on non-dominated sorting;

[0081] (4) Apply crossover and mutation operators to perform legacy operations;

[0082] (5) Use deep reinforcement learning algorithms to dynamically adjust the solutions generated by NSGA-II;

[0083] (6) Generate a Pareto optimal solution that satisfies all constraints as the final scheduling solution.

[0084] S3. Use Internet of Things technology to achieve dynamic adjustment of vehicle dispatching;

[0085] (1) Use GPS modules, RFID sensors, and smart dashboards to collect data on vehicle location, load, and speed;

[0086] (2) Using the Dijkstra algorithm based on the shortest path, reinforcement learning is called in real time to re-optimize the scheduling plan based on the current state;

[0087] (3) Use MQTT protocol and edge computing technology to store monitoring data on the cloud computing platform for subsequent analysis by users.

[0088] S4, cloud computing platform provides computing resources and distributed storage;

[0089] include:

[0090] (1) Using Hadoop framework for distributed computing;

[0091] (2) Using HDFS to store large-scale scheduling data;

[0092] (3) Use the MapReduce model for large-scale parallel task processing;

[0093] (4) Dynamically adjust computing resources to optimize model training and inference efficiency.

[0094] Based on the above method, a logistics vehicle scheduling system in this embodiment includes a demand forecasting module, an optimization scheduling engine module, a real-time monitoring and feedback module, and a cloud computing platform module;

[0095] The demand forecasting module is used to predict future order demand using big data analysis and machine learning techniques, providing input for the scheduling engine;

[0096] The optimization scheduling engine module is used to optimize the scheduling engine to generate the optimal scheduling solution through multi-objective optimization and reinforcement learning;

[0097] The real-time monitoring and feedback module is used to realize dynamic adjustment of vehicle dispatch and operation by adopting Internet of Things technology;

[0098] The cloud computing platform module is used to provide computing resources and distributed storage.

[0099] Among them, the demand forecasting module includes:

[0100] (1) Collect order history data, user behavior data, weather information, and traffic flow data for data collection;

[0101] (2) Perform data cleaning, normalization, feature selection and dimensionality reduction on the data;

[0102] (3) Select models based on different scenarios. Time series models are used for short-term forecasts, while regression models are used for long-term forecasts.

[0103] (4) Combining Bayesian optimization and grid search to adjust hyperparameters and perform prediction optimization;

[0104] (5) Generate order demand distribution within the time window.

[0105] The optimization scheduling engine module includes:

[0106] (1) Model the problem and construct a multi-objective function that includes cost, time, and service level. Constraints include vehicle capacity, time window restrictions, and order priority.

[0107] (2) Generate an initial solution using coding techniques;

[0108] (3) Calculate the fitness of each solution based on non-dominated sorting;

[0109] (4) Apply crossover and mutation operators to perform legacy operations;

[0110] (5) Use deep reinforcement learning algorithms to dynamically adjust the solutions generated by NSGA-II;

[0111] (6) Generate a Pareto optimal solution that satisfies all constraints as the final scheduling solution.

[0112] The real-time monitoring and feedback module includes:

[0113] (1) Use GPS modules, RFID sensors, and smart dashboards to collect data on vehicle location, load, and speed;

[0114] (2) Using the Dijkstra algorithm based on the shortest path, reinforcement learning is called in real time to re-optimize the scheduling plan based on the current state;

[0115] (3) Use MQTT protocol and edge computing technology to store monitoring data on the cloud computing platform for subsequent analysis by users.

[0116] The cloud computing platform module includes:

[0117] (1) Using Hadoop framework for distributed computing;

[0118] (2) Using HDFS to store large-scale scheduling data;

[0119] (3) Use the MapReduce model for large-scale parallel task processing;

[0120] (4) Dynamically adjust computing resources to optimize model training and inference efficiency.

[0121] The above-mentioned specific implementations are only specific cases of the present invention. The patent protection scope of the present invention includes but is not limited to the above-mentioned specific implementations. Any technical solutions that conform to the above-mentioned specific implementations of the present invention and any appropriate changes or substitutions made by ordinary technicians in the relevant technical field shall fall within the patent protection scope of the present invention.

[0122] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A logistics vehicle scheduling method, characterized in that: The steps are as follows: S1. Use big data analysis and machine learning technology to predict future order demand and provide input for the scheduling engine; S2, the optimization scheduling engine generates the optimal scheduling solution through multi-objective optimization and reinforcement learning; S3. Use Internet of Things technology to achieve dynamic adjustment of vehicle dispatching; S4. Cloud computing platform provides computing resources and distributed storage.

2. A logistics vehicle dispatching method according to claim 1, characterized in that: In step S1, it includes: (1) Collect order history data, user behavior data, weather information, and traffic flow data for data collection; (2) Perform data cleaning, normalization, feature selection and dimensionality reduction on the data; (3) Select models based on different scenarios. Time series models are used for short-term forecasts, while regression models are used for long-term forecasts. (4) Combining Bayesian optimization and grid search to adjust hyperparameters and perform prediction optimization; (5) Generate order demand distribution within the time window.

3. A logistics vehicle dispatching method according to claim 2, characterized in that: In step S2, it includes: (1) Model the problem and construct a multi-objective function that includes cost, time, and service level. Constraints include vehicle capacity, time window restrictions, and order priority. (2) Generate an initial solution using coding techniques; (3) Calculate the fitness of each solution based on non-dominated sorting; (4) Apply crossover and mutation operators to perform legacy operations; (5) Use deep reinforcement learning algorithms to dynamically adjust the solutions generated by NSGA-II; (6) Generate a Pareto optimal solution that satisfies all constraints as the final scheduling solution.

4. A logistics vehicle dispatching method according to claim 3, characterized in that: In step S3, it includes: (1) Use GPS modules, RFID sensors, and smart dashboards to collect data on vehicle location, load, and speed; (2) Using the Dijkstra algorithm based on the shortest path, reinforcement learning is called in real time to re-optimize the scheduling plan based on the current state; (3) Use MQTT protocol and edge computing technology to store monitoring data on the cloud computing platform for subsequent analysis by users.

5. A logistics vehicle dispatching method according to claim 4, characterized in that: In step S4, it includes: (1) Using Hadoop framework for distributed computing; (2) Using HDFS to store large-scale scheduling data; (3) Use the MapReduce model for large-scale parallel task processing; (4) Dynamically adjust computing resources to optimize model training and inference efficiency.

6. A logistics vehicle dispatching system, characterized in that: It includes demand forecasting module, optimization scheduling engine module, real-time monitoring and feedback module and cloud computing platform module; The demand forecasting module is used to use big data analysis and machine learning technology to predict future order demand and provide input for the scheduling engine; The optimization scheduling engine module is used for the optimization scheduling engine to generate the optimal scheduling solution through multi-objective optimization and reinforcement learning; The real-time monitoring and feedback module is used to realize dynamic adjustment of vehicle scheduling and operation by adopting Internet of Things technology; The cloud computing platform module is used to provide computing resources and distributed storage.

7. A logistics vehicle dispatching system according to claim 6, characterized in that: In the demand forecasting module, it includes: (1) Collect order history data, user behavior data, weather information, and traffic flow data for data collection; (2) Perform data cleaning, normalization, feature selection and dimensionality reduction on the data; (3) Select models based on different scenarios. Time series models are used for short-term forecasts, while regression models are used for long-term forecasts. (4) Combining Bayesian optimization and grid search to adjust hyperparameters and perform prediction optimization; (5) Generate order demand distribution within the time window.

8. A logistics vehicle dispatching system according to claim 7, characterized in that: The optimization scheduling engine module includes: (1) Model the problem and construct a multi-objective function that includes cost, time, and service level. Constraints include vehicle capacity, time window restrictions, and order priority. (2) Generate an initial solution using coding techniques; (3) Calculate the fitness of each solution based on non-dominated sorting; (4) Apply crossover and mutation operators to perform legacy operations; (5) Use deep reinforcement learning algorithms to dynamically adjust the solutions generated by NSGA-II; (6) Generate a Pareto optimal solution that satisfies all constraints as the final scheduling solution.

9. A logistics vehicle dispatching system according to claim 8, characterized in that: The real-time monitoring and feedback module includes: (1) Use GPS modules, RFID sensors, and smart dashboards to collect data on vehicle location, load, and speed; (2) Using the Dijkstra algorithm based on the shortest path, reinforcement learning is called in real time to re-optimize the scheduling plan based on the current state; (3) Use MQTT protocol and edge computing technology to store monitoring data on the cloud computing platform for subsequent analysis by users.

10. A logistics vehicle dispatching system according to claim 9, characterized in that: The cloud computing platform module includes: (1) Using Hadoop framework for distributed computing; (2) Using HDFS to store large-scale scheduling data; (3) Use the MapReduce model for large-scale parallel task processing; (4) Dynamically adjust computing resources to optimize model training and inference efficiency.

Images