Intelligent loading scheduling method based on big data driving

Abstract

The invention discloses an intelligent loading scheduling method based on big data driving, which establishes a more efficient intelligent loading scheduling system by collecting and processing data related to a transportation task, can respond to various dynamic changes in the transportation task in real time, optimizes resource configuration, reduces transportation cost, and improves transportation efficiency and service quality. Through the combination of multi-objective optimization and an intelligent scheduling algorithm, the transport task can be efficiently and accurately completed in different environments, and the method has a wide application prospect and a good market value.

Description

Technical Field

[0001] The present invention relates to the field of logistics and transportation scheduling, and in particular to an intelligent loading scheduling method driven by big data. Background Art

[0002] With the development of modern logistics industry, transportation scheduling faces increasingly complex challenges. Traditional transportation scheduling methods mostly rely on manual experience and fixed rules, lack flexibility and intelligence, and are difficult to effectively respond to the real-time changing transportation environment. Specifically, the existing transportation scheduling system has the following main problems:

[0003] Low efficiency of manual dispatch: Traditional transportation dispatch usually relies on dispatchers to assign tasks based on experience and pre-set rules, lacking real-time data support. This method is not only inefficient, but also easily affected by human factors, resulting in waste of transportation resources and task delays.

[0004] Lack of real-time scheduling decisions: Many existing scheduling systems cannot obtain real-time external environmental information (such as traffic flow, weather changes, etc.), so they cannot adjust transportation routes and task allocation in a timely manner. This makes the transportation process vulnerable to emergencies or environmental changes, resulting in reduced transportation efficiency.

[0005] Insufficient resource optimization: Traditional scheduling systems usually only focus on a single goal (such as time or cost), but ignore the balance between multiple goals, such as the comprehensive optimization of multiple factors such as transportation cost, resource consumption, transportation time, loading status, etc. This single-goal optimization method cannot meet the comprehensive requirements of efficiency, cost and resource utilization in modern transportation business.

[0006] Lack of intelligence in route planning: Existing route planning algorithms are usually based on fixed road network models or simple traffic forecasts, and cannot respond to the impact of complex traffic flow, weather changes, emergencies, and other factors in real time. Therefore, it is easy to choose unreasonable transportation routes, resulting in longer transportation time and waste of vehicle resources.

[0007] Insufficient use of big data: With the popularization of sensors, IoT technology and smart devices, modern transportation systems have accumulated a large amount of real-time data. However, traditional dispatching systems lack effective big data processing and analysis capabilities and cannot make full use of this data to achieve smarter decision-making and dispatching.

[0008] Therefore, how to make full use of big data technology, optimize transportation scheduling through real-time monitoring and intelligent algorithms, and improve transportation efficiency and resource utilization has become a key issue to be solved in the field of modern logistics. In order to solve the above problems, the present invention proposes an intelligent loading scheduling method based on big data drive, which improves transportation efficiency, reduces transportation costs, and enhances the intelligence level of the entire transportation process through real-time data collection, dynamic analysis and optimized scheduling. Summary of the invention

[0009] In order to solve the problems existing in the prior art, the present invention provides an intelligent loading scheduling method based on big data drive, comprising the following steps: Step S1: Collect and pre-process historical transport task data of transport vehicles, where the transport task data includes order data, vehicle data and route data as static data and weather data and traffic flow data as dynamic data; Step S2: extracting features from the static data and the dynamic data respectively, obtaining static features and dynamic features, and performing feature dimension reduction on the two; using the reduced-dimensional weather data or traffic flow data as labels to label the static data; Step S3: further extract the static features and dynamic features after dimensionality reduction through the time series model, obtain the weather change features and traffic flow change features in a specific time period, optimize the time series model based on the labeled data in step S2, and obtain the optimal time series model; Step S4: constructing a multi-objective optimization function based on the transportation cost, transportation time, loading rate and risk system of the historical transportation tasks of the transportation vehicle, and obtaining a set of initial solutions of the multi-objective optimization function through the Pareto frontier; Step S5: Based on the initial solution, the multi-objective optimization function is optimized again through Q-learning to obtain the optimal solution, and then the loading scheduling plan of the vehicle under the optimal solution is obtained; Step S6: Loading the loading scheduling plan of the vehicle under the optimal solution to the real-time transport vehicle, the loading scheduling plan including loading plan, path planning and task scheduling; Step S7: Acquire the dynamic data of the vehicle in step S6 in real time, execute steps S2 to S5 again, and obtain the loading scheduling plan of the vehicle under the updated optimal solution.

[0010] Furthermore, the transportation task data in step S1 specifically refers to: Order data, including type, amount, order priority, delivery date, customer type, and shipping costs; Vehicle data, including the vehicle’s current location, real-time speed, and vehicle loading status; Path data, including node information on the transportation path; Traffic data, including route congestion information; Environmental data, including weather data and temperature data.

[0011] Furthermore, the multi-objective optimization function in step S4 is specifically: Constructing a multi-objective optimization function , expressed as: ; ; ; ; ; in: represents the minimization function, Indicates time period Inside vehicle Execute historical loading tasks The decision variables; Indicates time period Inside vehicle Execute historical loading tasks operating costs; Indicates time period Inside vehicle Execute historical loading tasks Path distance; , , , Represents the weight of each objective function; Indicates the total transportation time; represents the total transportation time of the vehicle; represents the total transportation cost; Indicates vehicle The loading rate is used to further express the load-bearing weight or the volume that can be accommodated; represents the risk factor; Indicates vehicle Execute historical loading tasks The risk factor is: N represents the total number of vehicles; M represents the total number of loading tasks; For multi-objective optimization functions The constraints are expressed as: Each vehicle cannot perform multiple tasks at the same time: ; Each task can only be assigned once: ; Vehicle time limit: ; Loading Limits: ; A set of initial solutions of the multi-objective optimization function F is obtained through the Pareto front.

[0012] Furthermore, in step S5, the process of re-optimizing the multi-objective optimization function based on the initial solution through Q-learning is specifically as follows: Step S51: define state space; The state space represents the current situation of the environment, including static and dynamic characteristics; Step S52: define the action space; The action space represents the different weight combinations that the system can take so that the sum of the weights of each objective function is 1; it is expressed as: ; Step S53: Setting the reward function , expressed as: ; ; ; ; in, Indicates the actual time to complete the task. Indicates the time limit of the task. represents the time reward parameter; Indicates the actual operating cost; represents the maximum acceptable cost; represents the cost reward parameter; Indicates the actual loading rate; Indicates the theoretical maximum loading rate; Indicates the loading rate reward parameter; represents the risk reward parameter; Indicates the actual risk value; Indicates the maximum possible risk value; Indicates the penalty value for violating the constraint; Step S54: strategy update; Continuously optimize the weight distribution of each objective function to maximize the final reward.

[0013] The technical solution of the present invention can effectively improve the level of automation and intelligence in the transportation process, reduce human intervention, reduce transportation costs, improve resource utilization, and ensure that transportation tasks can be efficiently performed in a dynamic and uncertain environment.

[0014] The beneficial effect of the present invention is that compared with the prior art, the present invention collects and analyzes big data related to transportation tasks in real time, uses multi-objective optimization and reinforcement learning, builds an intelligent loading scheduling model, and dynamically optimizes loading plans, path planning, and task allocation. During the execution process, the system uses a real-time feedback mechanism combined with a dynamic adjustment strategy to effectively respond to traffic congestion, weather changes, and emergencies, and ensure the efficient completion of transportation tasks in complex environments. At the same time, by improving loading efficiency, optimizing resource allocation, and reducing transportation costs, it achieves maximum transportation efficiency, optimal resource utilization, and improved service quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 The present invention is a flowchart of an intelligent loading scheduling method based on big data drive. DETAILED DESCRIPTION

[0016] In order to enable those skilled in the art to better understand the present invention, the technical solution of the present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0017] An intelligent loading scheduling method based on big data drive includes the following steps: Step S1: Collect and pre-process historical transport task data of transport vehicles, where the transport task data includes order data, vehicle data and route data as static data and weather data and traffic flow data as dynamic data; Step S2: extracting features from the static data and the dynamic data respectively, obtaining static features and dynamic features, and performing feature dimension reduction on the two; using the reduced-dimensional weather data or traffic flow data as labels to label the static data; Step S3: further extract the static features and dynamic features after dimensionality reduction through the time series model, obtain the weather change features and traffic flow change features in a specific time period, optimize the time series model based on the labeled data in step S2, and obtain the optimal time series model; Step S4: constructing a multi-objective optimization function based on the transportation cost, transportation time, loading rate and risk system of the historical transportation tasks of the transportation vehicle, and obtaining a set of initial solutions of the multi-objective optimization function through the Pareto frontier; Step S5: Based on the initial solution, the multi-objective optimization function is optimized again through Q-learning to obtain the optimal solution, and then the loading scheduling plan of the vehicle under the optimal solution is obtained; Step S6: Loading the loading scheduling plan of the vehicle under the optimal solution to the real-time transport vehicle, the loading scheduling plan including loading plan, path planning and task scheduling; Step S7: Acquire the dynamic data of the vehicle in step S6 in real time, execute steps S2 to S5 again, and obtain the loading scheduling plan of the vehicle under the updated optimal solution.

[0018] Furthermore, the transportation task data in step S1 specifically refers to: Order data, including type, amount, order priority, delivery date, customer type, and shipping costs; Vehicle data, including the vehicle’s current location, real-time speed, and vehicle loading status; Path data, including node information on the transportation path; Traffic data, including route congestion information; Environmental data, including weather data and temperature data.

[0019] Furthermore, the multi-objective optimization function in step S4 is specifically: Constructing a multi-objective optimization function , expressed as: ; ; ; ; ; in: represents the minimization function, Indicates time period Inside vehicle Execute historical loading tasks The decision variables; Indicates time period Inside vehicle Execute historical loading tasks operating costs; Indicates time period Inside vehicle Execute historical loading tasks Path distance; , , , Represents the weight of each objective function; Indicates the total transportation time; represents the total transportation time of the vehicle; represents the total transportation cost; Indicates vehicle The loading rate is used to further express the load-bearing weight or the volume that can be accommodated; represents the risk factor; Indicates vehicle Execute historical loading tasks The risk factor is: N represents the total number of vehicles; M represents the total number of loading tasks; For multi-objective optimization function The constraints are expressed as: Each vehicle cannot perform multiple tasks at the same time: ; Each task can only be assigned once: ; Vehicle time limit: ; Loading Limits: ; A set of initial solutions of the multi-objective optimization function F is obtained through the Pareto front.

[0020] Furthermore, in step S5, the process of re-optimizing the multi-objective optimization function based on the initial solution through Q-learning is specifically as follows: Step S51: define state space; The state space represents the current situation of the environment, including static and dynamic characteristics; Step S52: define the action space; The action space represents the different weight combinations that the system can take so that the sum of the weights of each objective function is 1; it is expressed as: ; Step S53: Setting the reward function , expressed as: ; ; ; ; in, Indicates the actual time to complete the task. Indicates the time limit of the task. represents the time reward parameter; Indicates the actual operating cost; represents the maximum acceptable cost; represents the cost reward parameter; Indicates the actual loading rate; Indicates the theoretical maximum loading rate; Indicates the loading rate reward parameter; represents the risk reward parameter; Indicates the actual risk value; Indicates the maximum possible risk value; represents the penalty value for violating the constraint; Step S54: strategy update; Continuously optimize the weight distribution of each objective function to maximize the final reward.

[0021] The above embodiments are merely illustrative of the principles and effects of the present invention, and are not intended to limit the present invention. Anyone familiar with the technology may modify or change the above embodiments without violating the spirit and scope of the present invention. Therefore, all equivalent modifications or changes made by those with ordinary knowledge in the technical field without departing from the spirit and technical ideas disclosed by the present invention should still be covered by the claims of the present invention.

Claims

1. An intelligent loading scheduling method based on big data drive, characterized in that: The following steps are involved: Step S1: Collect and pre-process historical transport task data of transport vehicles, where the transport task data includes order data, vehicle data and route data as static data and weather data and traffic flow data as dynamic data; Step S2: extracting features from the static data and the dynamic data respectively, obtaining static features and dynamic features, and performing feature dimension reduction on the two; using the reduced-dimensional weather data or traffic flow data as labels to label the static data; Step S3: further extract the static features and dynamic features after dimensionality reduction through the time series model, obtain the weather change features and traffic flow change features in a specific time period, optimize the time series model based on the annotated data in step S2, and obtain the optimal time series model; Step S4: constructing a multi-objective optimization function based on the transportation cost, transportation time, loading rate and risk system of the historical transportation tasks of the transportation vehicle, and obtaining a set of initial solutions of the multi-objective optimization function through the Pareto frontier; Step S5: Based on the initial solution, the multi-objective optimization function is optimized again through Q-learning to obtain the optimal solution, and then the loading scheduling plan of the vehicle under the optimal solution is obtained; Step S6: Loading the loading scheduling plan of the vehicle under the optimal solution to the real-time transport vehicle, the loading scheduling plan including loading plan, path planning and task scheduling; Step S7: Acquire the dynamic data of the vehicle in step S6 in real time, execute steps S2 to S5 again, and obtain the loading scheduling plan of the vehicle under the updated optimal solution.

2. The intelligent loading scheduling method based on big data drive according to claim 1 is characterized in that: The transportation task data in step S1 specifically refers to the following within the transportation time: Order data, including type, amount, order priority, delivery date, customer type, and shipping costs; Vehicle data, including the vehicle’s current location, real-time speed, and vehicle loading status; Path data, including node information on the transportation path; Traffic data, including route congestion information; Environmental data, including weather data and temperature data.

3. The intelligent loading scheduling method based on big data drive according to claim 2 is characterized in that: The multi-objective optimization function in step S4 is specifically: Constructing a multi-objective optimization function , expressed as: ； ； ； ； ； in: represents the minimization function, Indicates time period Inside vehicle Execute historical loading tasks The decision variables; Indicates time period Inside vehicle Execute historical loading tasks operating costs; Indicates time period Inside vehicle Execute historical loading tasks Path distance; , , , Represents the weight of each objective function; Indicates the total transportation time; Indicates the total transportation time of the vehicle; represents the total transportation cost; Indicates vehicle The loading rate is used to further express the load-bearing weight or the volume that can be accommodated; represents the risk factor; Indicates vehicle Execute historical loading tasks The risk factor is: N represents the total number of vehicles; M represents the total number of loading tasks; For multi-objective optimization functions The constraints are expressed as: Each vehicle cannot perform multiple tasks at the same time: ； Each task can only be assigned once: ； Vehicle time limit: ； Loading Limits: ； A set of initial solutions of the multi-objective optimization function F is obtained through the Pareto front.

4. The intelligent loading scheduling method based on big data drive according to claim 3 is characterized in that: The process of re-optimizing the multi-objective optimization function based on the initial solution through Q-learning in step S5 is specifically as follows: Step S51: define the state space; The state space represents the current situation of the environment, including static and dynamic characteristics; Step S52: define the action space; The action space represents the different weight combinations that the system can take so that the sum of the weights of each objective function is 1; it is expressed as: ； Step S53: Setting the reward function , expressed as: ； ； ； ； in, Indicates the actual time to complete the task. Indicates the time limit of the task. represents the time reward parameter; Indicates the actual operating cost; represents the maximum acceptable cost; represents the cost reward parameter; Indicates the actual loading rate; Indicates the theoretical maximum loading rate; Indicates the loading rate reward parameter; represents the risk reward parameter; Indicates the actual risk value; Indicates the maximum possible risk value; Indicates the penalty value for violating the constraint; Step S54: strategy update; Continuously optimize the weight distribution of each objective function to maximize the final reward.

Images