Dynamic truck scheduling method, system and equipment based on artificial intelligence and medium

Abstract

The invention relates to a truck dynamic scheduling method, system and device based on artificial intelligence and a medium, and the method comprises the steps: obtaining transportation multi-dimensional data, and carrying out the cleaning and feature extraction of the transportation multi-dimensional data, so as to generate a scheduling optimization data set; based on the scheduling optimization data set, utilizing a multi-objective optimization algorithm to generate an initial transportation path; obtaining order demand data, performing load prediction on the order demand data based on the time sequence prediction model to obtain a corresponding load prediction result, and adjusting the initial transportation path according to the load prediction result to generate a transportation optimization path; and obtaining a transport vehicle state, and generating a transport scheduling strategy through a reinforcement learning algorithm based on the transport optimization path and the transport vehicle state. The method has the effect of improving the vehicle scheduling efficiency.

Description

Technical Field

[0001] The present application relates to the technical field of logistics scheduling, and in particular, to a dynamic scheduling method, system, device and medium for freight trucks based on artificial intelligence. Background Art

[0002] Currently, the logistics transportation industry is gradually developing towards high efficiency and intelligence. Among them, freight truck scheduling, as the core link of logistics transportation, is directly related to the improvement of transportation efficiency and resource utilization rate. However, freight truck scheduling involves the comprehensive processing of multi-dimensional data, including order requirements, vehicle status, road condition information, etc., which poses high requirements for scheduling algorithms and real-time decision-making capabilities.

[0003] Existing freight truck scheduling usually relies on pre-set fixed rules or simple path planning methods, lacking the comprehensive optimization ability for multiple objectives, and it is also difficult to dynamically respond to real-time changing transportation demands and road condition information. In complex logistics scenarios, existing methods are difficult to simultaneously optimize transportation time, load utilization rate and energy consumption, lacking flexibility and easily leading to a reduction in transportation efficiency.

[0004] The above-mentioned existing technical solutions have the following defects: The existing freight truck scheduling methods lack comprehensive capabilities such as multi-objective optimization and real-time dynamic adjustment, resulting in low efficiency of vehicle scheduling, so there is room for improvement. Summary of the Invention

[0005] In order to improve the efficiency of vehicle scheduling, the present application provides a dynamic scheduling method, system, device and medium for freight trucks based on artificial intelligence.

[0006] The first invention object of the present application is achieved through the following technical solutions: A dynamic scheduling method for freight trucks based on artificial intelligence, the dynamic scheduling method for freight trucks based on artificial intelligence includes: Obtain multi-dimensional transportation data, clean and extract features from the multi-dimensional transportation data to generate a scheduling optimization data set; Based on the scheduling optimization data set, use a multi-objective optimization algorithm to generate an initial transportation path; Obtain order demand data, and based on a time series prediction model, perform load prediction on the order demand data to obtain a corresponding load prediction result, and then adjust the initial transportation path according to the load prediction result to generate a transportation optimization path; Obtain the status of transportation vehicles, and based on the transportation optimization path and the status of transportation vehicles, generate a transportation scheduling strategy through a reinforcement learning algorithm.

[0007] By adopting the above technical solutions, by obtaining multi-dimensional transportation data, cleaning and feature extraction of the multi-dimensional transportation data are carried out to generate a scheduling optimization data set, which can extract key feature information related to transportation tasks from multi-dimensional data, reduce the interference of redundant data on scheduling decisions, and thus improve the quality of data input and the accuracy of subsequent optimization; by generating an initial transportation path using a multi-objective optimization algorithm based on the scheduling optimization data set, multiple objective factors such as transportation time, energy consumption, and load utilization rate can be comprehensively considered to generate a transportation path that meets the global optimization requirements, thereby improving transportation efficiency and reducing transportation costs; by obtaining order demand data and performing load prediction on the order demand data based on a time series prediction model, the future changes in transportation demand can be predicted, providing a basis for dynamic adjustment of path planning, and thus enhancing the adaptability of the transportation system to changes in order demand; by obtaining the status of transportation vehicles, generating a transportation scheduling strategy through a reinforcement learning algorithm based on the optimized transportation path and the status of transportation vehicles, the vehicle scheduling plan can be dynamically optimized, and emergencies in transportation tasks can be responded to in real time, thereby enhancing the flexibility of scheduling and the completion efficiency of transportation tasks.

[0008] In one example, the present application can be further configured as follows: the generating an initial transportation path using a multi-objective optimization algorithm based on the scheduling optimization data set specifically includes: According to the scheduling optimization data set, multiple objective functions are constructed by minimizing transportation time, minimizing energy consumption, and maximizing load utilization rate; The multiple objective functions are combined using a weighted summation method to generate the initial transportation path.

[0009] By adopting the above technical solutions, by constructing multiple objective functions by minimizing transportation time, minimizing energy consumption, and maximizing load utilization rate according to the scheduling optimization data set, the optimization objectives of transportation tasks can be quantified from different dimensions, providing the calculation basis required by the multi-objective optimization algorithm, and thus ensuring that the optimization process can comprehensively balance time, cost, and resource utilization rate; by combining multiple objective functions using a weighted summation method to generate an initial transportation path, the balance of different optimization objectives can be achieved with a controllable weight ratio, ensuring that the selection of the initial path meets both efficiency requirements and economic and load balance requirements, and thus providing a high-quality initial path plan for subsequent scheduling.

[0010] In one example, the present application can be further configured as follows: the constructing multiple objective functions by minimizing transportation time, minimizing energy consumption, and maximizing load utilization rate specifically includes: According to the distance, road conditions information, and real-time traffic conditions of the transportation path, the estimated time of the transportation path is calculated, and then a transportation time function is constructed; Based on the energy efficiency model of the transportation vehicle, combined with the path length, load, and the dynamic performance of the vehicle, estimate the energy consumption of the transportation path, and then construct an energy consumption function; By comparing the actual load of the vehicle with the maximum load capacity, generate a load utilization rate index, and optimize the load situation of the vehicle through a task allocation scheme, and then construct a load utilization rate function.

[0011] By adopting the above technical solutions, by calculating the estimated time of the transportation path according to the distance, road conditions information, and real-time traffic conditions of the transportation path, and then constructing a transportation time function, it is possible to dynamically evaluate the time requirements of different paths, provide a time-optimal calculation basis for path selection, and thus effectively reduce delays during transportation; by based on the energy efficiency model of the transportation vehicle, combined with the path length, load, and the dynamic performance of the vehicle, estimate the energy consumption of the transportation path, and then construct an energy consumption function, it is possible to provide an energy efficiency optimization calculation basis for path selection, thereby reducing the use cost of transportation energy and reducing environmental impact; by comparing the actual load of the vehicle with the maximum load capacity, generate a load utilization rate index, and optimize the load situation of the vehicle through a task allocation scheme, and then construct a load utilization rate function, it is possible to effectively utilize the load capacity of the transportation vehicle during path planning, thereby reducing the occurrence of empty runs and improving resource utilization efficiency.

[0012] In one example of this application, it can be further configured as follows: The construction of the time series prediction model includes: Obtain historical order data, and extract key time series features from the historical order data, and then construct a time series feature set; Iteratively train a preset long short-term memory network according to the time series feature set, learn the laws such as seasonal fluctuations of historical order data and historical transportation load patterns, and minimize the prediction error by adjusting the model weights to obtain the time series prediction model.

[0013] By adopting the above technical solutions, by obtaining historical order data, and extracting key time series features from the historical order data, and then constructing a time series feature set, it is possible to utilize historical data to mine the seasonal laws and demand fluctuation characteristics of transportation tasks, provide high-quality input data for the time series prediction model, and thus improve the accuracy of the prediction model; by iteratively training a preset long short-term memory network according to the time series feature set, learning the laws such as seasonal fluctuations of historical order data and historical transportation load patterns, and minimizing the prediction error by adjusting the model weights to obtain the time series prediction model, it is possible to dynamically capture complex data patterns through a deep learning model, and thus further improve the accuracy of transportation demand prediction.

[0014] In one example, the present application can be further configured as follows: the order demand data is subjected to load prediction by the time series prediction model to obtain a corresponding load prediction result, and then the initial transportation path is adjusted according to the load prediction result to generate an optimized transportation path, which specifically includes: Input the order demand data into the time series prediction model to obtain the load demand of each transportation task within a preset time period, and then generate the load prediction result; According to the load prediction result, comprehensively adjust the load demand of the vehicle and the initial path to generate the optimized transportation path.

[0015] By adopting the above technical solution, by inputting the order demand data into the time series prediction model to obtain the load demand of each transportation task within a preset time period, and then generating the load prediction result, it is possible to identify future order load changes in advance, provide a basis for path adjustment, and thus effectively avoid transportation delays and resource waste caused by overloading or underloading; by comprehensively adjusting the load demand of the vehicle and the initial path according to the load prediction result to generate the optimized transportation path, it is possible to dynamically optimize path selection and task allocation based on the prediction result, thereby improving the response ability and overall efficiency of the transportation process.

[0016] In one example, the present application can be further configured as follows: based on the optimized transportation path and the state of the transportation vehicle, a transportation scheduling strategy is generated through a reinforcement learning algorithm, which specifically includes: Construct the state space of the reinforcement learning algorithm based on the state of the transportation vehicle, and construct the motion space of the reinforcement learning algorithm based on the optimized transportation path; Obtain the transportation efficiency, load utilization rate, and energy consumption of the transportation task, and then perform comprehensive calculations to define the corresponding reward function; Update the policy network of the reinforcement learning algorithm according to the reward function, and output the transportation scheduling strategy to verify the feasibility of future transportation tasks.

[0017] By adopting the above technical solution, by constructing the state space of the reinforcement learning algorithm based on the state of the transport vehicle and the action space of the reinforcement learning algorithm based on the optimized transport path, a reinforcement learning model can be constructed based on the real-time state in the actual transport task, providing a basis for the dynamic optimization of the scheduling strategy; by obtaining the transport efficiency, load utilization rate and energy consumption of the transport task and then performing comprehensive calculations to define the corresponding reward function, the advantages and disadvantages of different scheduling strategies can be quantitatively evaluated, providing a direction for the training of the reinforcement learning model; by updating the policy network of the reinforcement learning algorithm according to the reward function and outputting the transport scheduling strategy to verify the feasibility of future transport tasks, the efficient response to complex transport scenarios can be achieved through the continuously optimized scheduling strategy, thereby improving the overall scheduling quality of the transport task.

[0018] In one example, the present application can be further configured as follows: The truck dynamic scheduling method based on artificial intelligence further includes: Analyze the real-time data during the transportation process through an anomaly detection algorithm, and when an abnormal situation is detected, generate a warning message and trigger an emergency scheduling strategy.

[0019] By adopting the above technical solution, by analyzing the real-time data during the transportation process through an anomaly detection algorithm, and when an abnormal situation is detected, generating a warning message and triggering an emergency scheduling strategy, potential abnormal situations during the transportation process can be identified in real time, and response measures can be generated in a timely manner, thereby avoiding serious impacts of abnormal situations on the transport task and improving the safety and stability of the transportation process.

[0020] The above second inventive object of the present application is achieved by the following technical solutions: A truck dynamic scheduling system based on artificial intelligence, the truck dynamic scheduling system based on artificial intelligence includes: A data processing module, configured to obtain multi-dimensional transport data, clean and extract features from the multi-dimensional transport data to generate a scheduling optimization data set; A path planning module, configured to generate an initial transport path based on the scheduling optimization data set by using a multi-objective optimization algorithm; A load prediction module, configured to obtain order demand data, perform load prediction on the order demand data based on a time series prediction model to obtain a corresponding load prediction result, and then adjust the initial transport path according to the load prediction result to generate an optimized transport path; A scheduling strategy generation module, configured to obtain the state of the transport vehicle, and generate a transport scheduling strategy based on the optimized transport path and the state of the transport vehicle through a reinforcement learning algorithm.

[0021] By adopting the above technical solutions, by obtaining multi-dimensional transportation data, cleaning and feature extracting the multi-dimensional transportation data to generate a scheduling optimization data set, it is possible to extract key feature information related to transportation tasks from multi-dimensional data, reduce the interference of redundant data on scheduling decisions, thereby improving the quality of data input and the accuracy of subsequent optimization; by generating an initial transportation route based on the scheduling optimization data set using a multi-objective optimization algorithm, it is possible to comprehensively consider multi-objective factors such as transportation time, energy consumption, and load utilization rate, generate a transportation route that meets the global optimization requirements, thereby improving transportation efficiency and reducing transportation costs; by obtaining order demand data and predicting the load of the order demand data based on a time series prediction model, it is possible to predict future changes in transportation demand, provide a basis for dynamic adjustment of route planning, thereby enhancing the adaptability of the transportation system to changes in order demand; by obtaining the status of transportation vehicles, based on the optimized transportation route and the status of transportation vehicles, generating a transportation scheduling strategy through a reinforcement learning algorithm, it is possible to dynamically optimize the vehicle scheduling plan and respond in real time to emergencies in transportation tasks, thereby enhancing the flexibility of scheduling and the completion efficiency of transportation tasks.

[0022] The above object three of the present application is achieved by the following technical solutions: A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the above-mentioned truck dynamic scheduling method based on artificial intelligence are implemented.

[0023] The above object four of the present application is achieved by the following technical solutions: A computer-readable storage medium stores a computer program. When the computer program is executed by a processor, the steps of the above-mentioned truck dynamic scheduling method based on artificial intelligence are implemented.

[0024] In summary, the present application includes the following beneficial technical effects: 1. By obtaining multi-dimensional transportation data, cleaning and extracting features from the multi-dimensional transportation data to generate a scheduling optimization data set, key feature information related to transportation tasks can be extracted from multi-dimensional data, reducing the interference of redundant data on scheduling decisions, thereby improving the quality of data input and the accuracy of subsequent optimization; by generating an initial transportation route using a multi-objective optimization algorithm based on the scheduling optimization data set, multi-objective factors such as transportation time, energy consumption, and load utilization rate can be comprehensively considered, generating a transportation route that meets the global optimization requirements, thereby improving transportation efficiency and reducing transportation costs; by obtaining order demand data and predicting the load of the order demand data based on a time series prediction model, the future changes in transportation demand can be predicted, providing a basis for dynamic adjustment of route planning, thereby enhancing the adaptability of the transportation system to changes in order demand; by obtaining the status of transportation vehicles, generating a transportation scheduling strategy through a reinforcement learning algorithm based on the optimized transportation route and the status of transportation vehicles, the vehicle scheduling plan can be dynamically optimized, and emergencies in transportation tasks can be responded to in real time, thereby enhancing the flexibility of scheduling and the completion efficiency of transportation tasks. 2. By constructing multiple objective functions by minimizing transportation time, minimizing energy consumption, and maximizing load utilization rate according to the scheduling optimization data set, the optimization objectives of transportation tasks can be quantified from different dimensions, providing the calculation basis required by the multi-objective optimization algorithm, thereby ensuring that the optimization process can comprehensively balance time, cost, and resource utilization rate; by using the weighted summation method to combine multiple objective functions to generate an initial transportation route, the balance of different optimization objectives can be achieved with a controllable weight ratio, ensuring that the selection of the initial route meets both efficiency requirements and takes into account economy and load balance, thereby providing a high-quality initial route plan for subsequent scheduling. 3. By calculating the estimated time of the transportation route based on the distance, road conditions, and real-time traffic conditions of the transportation route, and then constructing a transportation time function, the time requirements of different routes can be dynamically evaluated, providing a time-optimal calculation basis for route selection, thereby effectively reducing delays during transportation; by estimating the energy consumption of the transportation route based on the energy efficiency model of the transportation tool, combining the route length, load, and dynamic performance of the vehicle, and then constructing an energy consumption function, a calculation basis for energy efficiency optimization can be provided for route selection, thereby reducing the use cost of transportation energy and reducing environmental impact; by comparing the actual load of the vehicle with the maximum load capacity, generating a load utilization rate index, and optimizing the load situation of the vehicle through a task allocation plan, and then constructing a load utilization rate function, the load capacity of the transportation tool can be effectively utilized during route planning, thereby reducing the occurrence of empty runs and improving resource utilization efficiency. Description of the Drawings

[0025] Figure 1 is a flowchart of a truck dynamic scheduling method based on artificial intelligence in an embodiment of the present application; Figure 2 It is the implementation flowchart of step S20 in the truck dynamic scheduling method based on artificial intelligence in an embodiment of the present application; Figure 3 It is the implementation flowchart of step S21 in the truck dynamic scheduling method based on artificial intelligence in an embodiment of the present application; Figure 4 It is the implementation flowchart of the construction of the time series prediction model in the truck dynamic scheduling method based on artificial intelligence in an embodiment of the present application; Figure 5 It is the implementation flowchart of step S30 in the truck dynamic scheduling method based on artificial intelligence in an embodiment of the present application; Figure 6 It is the implementation flowchart of step S40 in the truck dynamic scheduling method based on artificial intelligence in an embodiment of the present application; Figure 7 It is another implementation flowchart of the truck dynamic scheduling method based on artificial intelligence in an embodiment of the present application; Figure 8 It is a principle block diagram of a truck dynamic scheduling system based on artificial intelligence in an embodiment of the present application; Figure 9 It is a schematic diagram of the device in an embodiment of the present application. Detailed implementation manners

[0026] The present application will be further described in detail below with reference to the accompanying drawings.

[0027] In one embodiment, as Figure 1 shown, the present application discloses a truck dynamic scheduling method based on artificial intelligence, which specifically includes the following steps: S10: Obtain multi-dimensional transportation data, clean and extract features from the multi-dimensional transportation data to generate a scheduling optimization data set.

[0028] Specifically, the multi-dimensional transportation data includes order data, vehicle status data, real-time traffic data, and weather data. By accessing Internet of Things devices, data of various sensors during the transportation process are collected in real time. When cleaning, outliers and missing values in the data are corrected by mean filling or interpolation method. When extracting features, the multi-dimensional data is standardized and normalized based on specific target optimization requirements, and core features of the data such as road condition score, vehicle load capacity score, and order priority score are extracted to generate a data set for scheduling optimization.

[0029] S20: Based on the scheduling optimization data set, use a multi-objective optimization algorithm to generate an initial transportation route.

[0030] Specifically, the scheduling optimization data set is input into the optimization algorithm. By invoking path planning methods such as the dynamic programming algorithm or the genetic algorithm, in the case of known starting and ending points, a preliminary path plan is generated according to the order requirements and vehicle distribution. Combining the path optimization data in historical transportation tasks, the best connection paths in different transportation regions are dynamically selected. At the same time, the possible energy consumption, transportation time, and load utilization rate of the path are estimated as reference indicators for generating the initial transportation path.

[0031] S30: Obtain the order demand data, perform load prediction on the order demand data based on the time series prediction model to obtain the corresponding load prediction result, and then adjust the initial transportation path according to the load prediction result to generate an optimized transportation path.

[0032] Specifically, extract the quantity of goods, estimated delivery time, and delivery location of each order from the order demand data. Combining the characteristics of seasonal fluctuations in historical orders and peak demand during holidays, the order data is divided into two parts: short-term demand and long-term trend. The short-term demand is predicted through the time series prediction model to generate the load demand for each transportation task in the future time period. The initial path plan is adjusted according to the prediction result, and orders with larger loads are preferentially assigned to vehicles with higher load capacity. Try to avoid overly long transfer routes and unnecessary empty runs during path adjustment, and finally generate an optimized transportation path.

[0033] S40: Obtain the status of transportation vehicles, and generate a transportation scheduling strategy based on the optimized transportation path and the status of transportation vehicles through the reinforcement learning algorithm.

[0034] Specifically, obtain the real-time position, current load condition, and battery power or fuel remaining of the vehicle from the status of transportation vehicles. Combining the transfer nodes and end point positions in the optimized path, perform policy modeling on the driving sequence and task assignment of the vehicle, and gradually iterate through the reinforcement learning algorithm to generate the optimal scheduling strategy. The training process of the reinforcement learning algorithm simulates various scenarios in transportation, such as vehicle breakdowns, path interruptions, or newly added orders in real time. When the strategy is output, specific tasks and arrival time nodes are assigned to each vehicle to improve the overall transportation efficiency and scheduling flexibility.

[0035] In one embodiment, as Figure 2 shown, in step S20, that is, based on the scheduling optimization data set, use the multi-objective optimization algorithm to generate the initial transportation path, which specifically includes: S21: According to the scheduling optimization data set, construct multiple objective functions by minimizing transportation time, minimizing energy consumption, and maximizing load utilization rate.

[0036] Specifically, based on the historical transportation time records and real-time traffic information in the scheduling optimization dataset, an association model between transportation time and route selection is established. A transportation time function is constructed with the route length, average speed, and road condition score as input variables. Meanwhile, combining the energy consumption model and load characteristics of the vehicle, an energy consumption function is constructed with the route length, current vehicle load, and road condition information as inputs. Finally, by comparing the current load of the vehicle with its maximum load capacity, a load utilization rate index is generated, and the load situation of each vehicle is optimized in combination with the order allocation tasks, and finally an objective function is generated for the subsequent route planning of the multi-objective optimization algorithm.

[0037] S22: Use the weighted summation method to combine multiple objective functions to generate an initial transportation route.

[0038] Specifically, different weight parameters are set for the three objective functions of transportation time, energy consumption, and load utilization rate according to their importance. A comprehensive optimization objective value is generated through the weighted summation method. For different transportation scenarios, the weight allocation ratio can be dynamically adjusted. For example, in an emergency transportation scenario, the weight of transportation time is increased, while in a regular transportation task, the balance between energy consumption and load utilization rate is given priority. The comprehensive objective value is used to evaluate the quality of different routes and select the route with the highest score as the initial transportation route.

[0039] In one embodiment, as Figure 3 shown, in step S21, that is, multiple objective functions are constructed by minimizing transportation time, minimizing energy consumption, and maximizing load utilization rate, specifically including: S211: According to the distance, road condition information, and real-time traffic conditions of the transportation route, calculate the estimated time of the transportation route, and then construct a transportation time function.

[0040] Specifically, call the real-time traffic data interface to obtain the average vehicle speed and congestion score of different route segments. Taking the route distance as the reference variable, and combining the influencing factors such as construction areas and prohibited sections in the road condition information, calculate the estimated time of each route through the weighted average method. For the transfer points that need to be passed through in the route, further superimpose the estimated value of the transfer operation time to obtain a complete transportation time function for optimizing the route selection.

[0041] S212: Based on the energy efficiency model of the transportation tool, combine the route length, load, and dynamic performance of the vehicle to estimate the energy consumption of the transportation route, and then construct an energy consumption function.

[0042] Specifically, an energy consumption model is established based on the vehicle's energy efficiency characteristic curve, taking the path length, road condition information, and vehicle load as the main input variables. A piecewise function is used to fit the energy consumption changes under different load conditions, and the slope information in the path or the number of traffic lights is also incorporated into the model as a correction parameter. Finally, an estimated value of the comprehensive energy consumption of the path is obtained for constructing an energy consumption function.

[0043] S213: By comparing the actual load of the vehicle with the maximum load capacity, a load utilization rate index is generated, and the load situation of the vehicle is optimized through a task allocation scheme, thereby constructing a load utilization rate function.

[0044] Specifically, the load utilization rate is calculated according to the ratio of the current load of each vehicle to its maximum load. Tasks are assigned to those vehicles with a lower load utilization rate to optimize resource allocation. For multi-task scenarios, tasks with a large order demand are preferentially assigned to vehicles close to the maximum load, thereby reducing the probability of empty running and transportation costs, and constructing a comprehensive load utilization rate function for multi-objective optimization.

[0045] In one embodiment, as Figure 4 shown, in step S30, the construction of the time series prediction model includes: S301: Obtain historical order data, extract key time series features from the historical order data, and then construct a time series feature set.

[0046] Specifically, features related to time are extracted from the historical order data, including the creation time, completion time, delivery cycle, etc. of the order. Further analysis is combined with the geographical distribution of the order, the type of goods, and historical peak periods to extract seasonal features and periodic features. After processing these time series features through feature encoding and normalization methods, a time series feature set for training the time series prediction model is generated. At the same time, a hierarchical feature set is established according to different order types to improve the prediction accuracy.

[0047] S302: Iteratively train a preset long short-term memory network according to the time series feature set, learn the laws such as seasonal fluctuations of historical order data and historical transportation load patterns, and minimize the prediction error by adjusting the model weights to obtain a time series prediction model.

[0048] Specifically, the processed time series feature set is input into a long short-term memory network model for training. The input layer of the model takes multi-dimensional time series data as features, and the output layer is the load prediction result within a certain future time period. During the training process, the sliding window method is used to gradually input sequence data. The model continuously adjusts the weight and bias parameters through the backpropagation algorithm to minimize the prediction error. At the same time, an early stopping mechanism is introduced to avoid overfitting. The accuracy of the prediction model is evaluated through the validation set, and finally an optimized time series prediction model is obtained for the dynamic prediction of load demand.

[0049] In one embodiment, as Figure 5 shown, in step S30, that is, based on the time series prediction model, load prediction is performed on the order demand data to obtain the corresponding load prediction result, and then the initial transportation path is adjusted according to the load prediction result to generate an optimized transportation path, specifically including: S31: Input the order demand data into the time series prediction model to obtain the load demand of each transportation task within a preset time period, and then generate a load prediction result.

[0050] Specifically, the order demand data collected in real time is merged with the historical order data and input into the trained time series prediction model. The model predicts the load demand of each transportation task within a future period of time according to the time series characteristics of the input data. The prediction result includes the estimated load and order distribution of each vehicle at each time node. The load prediction result is used to evaluate whether the current vehicle task is reasonable and whether it is necessary to adjust the task allocation or path selection.

[0051] S32: According to the load prediction result, comprehensively adjust the load demand and initial path of the vehicle to generate an optimized transportation path.

[0052] Specifically, based on the load prediction result, calculate the difference between the actual load of each vehicle and the predicted demand. For the vehicles with predicted overloading, relieve the load pressure by reducing their task allocation amount or adjusting some tasks to other vehicles. For the vehicles with predicted empty load, reallocate the path to cover more order demands. At the same time, optimize the selection of transfer points in the path in combination with real-time road condition information to ensure that the adjusted path can not only meet the load demand of the vehicle but also improve the overall transportation efficiency, and finally generate an optimized transportation path.

[0053] In one embodiment, as Figure 6 shown, in step S40, that is, based on the optimized transportation path and the state of the transportation vehicle, a transportation scheduling strategy is generated through a reinforcement learning algorithm, specifically including: S41: Construct the state space of the reinforcement learning algorithm based on the state of the transportation vehicle, and construct the motion space of the reinforcement learning algorithm based on the optimized transportation path.

[0054] Specifically, the real-time position, load condition, energy consumption status, and current task progress of the vehicle are used as the main variables of the state space of reinforcement learning. At the same time, the transfer nodes and end positions in the optimized transportation route are introduced as key decision points to construct an action space for describing possible path adjustments, task assignments, and vehicle scheduling behaviors. Each action in the action space is directly associated with the variables in the state space, and an optimal action policy is generated through iterative training of the reinforcement learning algorithm.

[0055] S42: Obtain the transportation efficiency, load utilization rate, and energy consumption of the transportation task, and then perform comprehensive calculations to define the corresponding reward function.

[0056] Specifically, the results of each scheduling action are fed back into the reinforcement learning algorithm, and the reward function is calculated based on the actual changes in transportation efficiency, load utilization rate, and energy consumption. The calculation of transportation efficiency is based on the ratio of the task completion time to the expected completion time, the load utilization rate is based on the ratio of the current load to the maximum load, and the energy consumption is based on the path length traveled by the vehicle and the energy efficiency model. When performing comprehensive calculations, the reward values are weighted and summed according to the proportion of the largest weight for transportation efficiency, the second largest for load utilization rate, and the smallest for energy consumption, to guide the algorithm to optimize the scheduling strategy.

[0057] S43: Update the policy network of the reinforcement learning algorithm according to the reward function, and output the transportation scheduling strategy to verify the feasibility of future transportation tasks.

[0058] Specifically, by backpropagating the reward value of each scheduling action, the parameters of the policy network in the reinforcement learning model are updated. The policy network adjusts the probability distribution of action selection according to the new parameters, generates a scheduling strategy that better meets the reward target. The output scheduling strategy includes the task assignment, path selection, and time planning for each vehicle. At the same time, in the next round of tasks, the policy network is continuously optimized by verifying the effectiveness of the scheduling strategy to improve the overall scheduling efficiency and robustness of the system.

[0059] In one embodiment, as Figure 7 shown, that is, the dynamic scheduling method for trucks based on artificial intelligence further includes: S50: Analyze the real-time data during the transportation process through an anomaly detection algorithm, and when an abnormal situation is detected, generate a warning message and trigger an emergency scheduling strategy.

[0060] Specifically, by monitoring the real-time position, load status, energy consumption, and path deviation of the vehicle, using rule-based or machine learning anomaly detection algorithms to identify emergencies, such as the vehicle deviating from the path, running at high speed, or having abnormal energy consumption, generating corresponding warning messages, the system automatically triggers emergency dispatching strategies, such as reassigning the transportation task to a nearby standby vehicle or adjusting the current task path, ensuring that the transportation task can be successfully completed under abnormal circumstances, and storing the anomaly records for subsequent analysis and model optimization.

[0061] It should be understood that the magnitudes of the sequence numbers of the steps in the above embodiments do not imply the order of execution. The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present application.

[0062] In one embodiment, an AI-based truck dynamic dispatching system is provided, which corresponds one-to-one with the AI-based truck dynamic dispatching method in the above embodiment. As Figure 8 shown, the AI-based truck dynamic dispatching system includes a data processing module, a path planning module, a load prediction module, and a dispatching strategy generation module. The detailed description of each functional module is as follows: The data processing module is used to obtain multi-dimensional transportation data, clean and extract features from the multi-dimensional transportation data to generate a dispatching optimization data set; The path planning module is used to generate an initial transportation path based on the dispatching optimization data set by using a multi-objective optimization algorithm; The load prediction module is used to obtain order demand data, perform load prediction on the order demand data based on a time series prediction model to obtain corresponding load prediction results, and then adjust the initial transportation path according to the load prediction results to generate an optimized transportation path; The dispatching strategy generation module is used to obtain the status of transportation vehicles, and generate a transportation dispatching strategy based on the optimized transportation path and the status of transportation vehicles through a reinforcement learning algorithm.

[0063] Optionally, the path planning module specifically includes: The objective function construction sub-module is used to construct multiple objective functions by minimizing transportation time, minimizing energy consumption, and maximizing load utilization rate according to the dispatching optimization data set; The objective merging sub-module is used to merge multiple objective functions by using a weighted summation method to generate an initial transportation path.

[0064] Optionally, the objective function construction sub-module specifically includes: The transportation time optimization unit is used to calculate the estimated time of the transportation path according to the distance, road conditions, and real-time traffic conditions of the transportation path, and then construct a transportation time function; An energy consumption optimization unit, which is used to estimate the energy consumption of a transportation route based on the energy efficiency model of a transportation vehicle, in combination with the route length, load, and power performance of the vehicle, and then construct an energy consumption function; A load optimization unit, which is used to generate a load utilization rate index by comparing the actual load of the vehicle with the maximum load capacity, and optimize the load condition of the vehicle through a task allocation scheme, and then construct a load utilization rate function.

[0065] Optionally, the construction of the time series prediction model includes: A feature extraction module, which is used to obtain historical order data and extract key time series features from the historical order data, and then construct a time series feature set; A model training module, which is used to iteratively train a preset long short-term memory network according to the time series feature set, learn the laws such as seasonal fluctuations of historical order data and historical transportation load patterns, and minimize the prediction error by adjusting the model weights to obtain a time series prediction model.

[0066] Optionally, the load prediction module specifically includes: A data prediction sub-module, which is used to input order demand data into the time series prediction model to obtain the load demand of each transportation task within a preset time period, and then generate a load prediction result; A route adjustment sub-module, which is used to comprehensively adjust the load demand and initial route of the vehicle according to the load prediction result to generate an optimized transportation route.

[0067] Optionally, the scheduling policy generation module specifically includes: A state space construction sub-module, which is used to construct the state space of the reinforcement learning algorithm based on the state of the transportation vehicle and construct the motion space of the reinforcement learning algorithm based on the optimized transportation route; A reward function definition sub-module, which is used to obtain the transportation efficiency, load utilization rate, and energy consumption of the transportation task, and then perform comprehensive calculations to define the corresponding reward function; A policy generation sub-module, which is used to update the policy network of the reinforcement learning algorithm according to the reward function and output a transportation scheduling policy to verify the feasibility of future transportation tasks.

[0068] Optionally, the truck dynamic scheduling method based on artificial intelligence further includes: An anomaly detection module, which is used to analyze the real-time data during the transportation process through an anomaly detection algorithm, and generate a warning message and trigger an emergency scheduling policy when an anomaly is detected.

[0069] For the specific limitations of the AI - based truck dynamic scheduling system, reference can be made to the limitations of the AI - based truck dynamic scheduling method in the above text, which will not be elaborated here. Each module in the above - mentioned AI - based truck dynamic scheduling system can be implemented in whole or in part by software, hardware, or a combination thereof. The above - mentioned modules can be embedded in the processor of the computer device in hardware form or be independent of it, or be stored in the memory of the computer device in software form, so as to facilitate the processor to call and execute the operations corresponding to each of the above - mentioned modules.

[0070] In one embodiment, a computer device is provided. This computer device can be a server, and its internal structure diagram can be as Figure 9 shown. The computer device includes a processor, a memory, a network interface, and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non - volatile storage medium and an internal memory. The non - volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non - volatile storage medium. The network interface of the computer device is used to communicate with external terminals through a network connection. When the computer program is executed by the processor, it implements an AI - based truck dynamic scheduling method.

[0071] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the following steps are implemented: Obtain multi - dimensional transportation data, clean and extract features from the multi - dimensional transportation data to generate a scheduling optimization data set; Based on the scheduling optimization data set, use a multi - objective optimization algorithm to generate an initial transportation route; Obtain order demand data, and based on a time - series prediction model, perform load prediction on the order demand data to obtain a corresponding load prediction result, and then adjust the initial transportation route according to the load prediction result to generate an optimized transportation route; Obtain the status of transportation vehicles, and based on the optimized transportation route and the status of transportation vehicles, generate a transportation scheduling strategy through a reinforcement learning algorithm.

[0072] In one embodiment, a computer - readable storage medium is provided, on which a computer program is stored. When the computer program is executed by the processor, the following steps are implemented: Obtain multi - dimensional transportation data, clean and extract features from the multi - dimensional transportation data to generate a scheduling optimization data set; Based on the scheduling optimization data set, use a multi - objective optimization algorithm to generate an initial transportation route; Obtain order demand data, perform load prediction on the order demand data based on a time series prediction model to obtain corresponding load prediction results, and then adjust the initial transportation route according to the load prediction results to generate an optimized transportation route; Obtain the status of transportation vehicles, and generate a transportation scheduling strategy based on the optimized transportation route and the status of transportation vehicles through a reinforcement learning algorithm.

[0073] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, storage, database, or other medium used in the various embodiments provided in the present application can include non-volatile and / or volatile memories. Non-volatile memories can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memories can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and Rambus dynamic RAM (RDRAM), etc.

[0074] Those skilled in the art can clearly understand that for the convenience and brevity of description, only the above-mentioned division of each functional unit and module is used as an example. In actual applications, the above functions can be allocated to different functional units and modules according to needs, that is, the internal structure of the system can be divided into different functional units or modules to complete all or part of the functions described above.

[0075] The above embodiments are only used to illustrate the technical solutions of the present application, and are not intended to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A dynamic truck dispatching method based on artificial intelligence, characterized in that: The artificial intelligence-based dynamic truck dispatching method includes: Acquire multidimensional transportation data, and perform cleaning and feature extraction on the multidimensional transportation data to generate a scheduling optimization data set; Based on the scheduling optimization data set, an initial transportation path is generated using a multi-objective optimization algorithm; Acquire order demand data, and perform load forecasting on the order demand data based on a time series forecasting model to obtain corresponding load forecasting results, and then adjust the initial transportation path according to the load forecasting results to generate an optimized transportation path; The transport vehicle status is obtained, and based on the transport optimization path and the transport vehicle status, a transport scheduling strategy is generated through a reinforcement learning algorithm.

2. The method for dynamic dispatching of trucks based on artificial intelligence according to claim 1, characterized in that: The method of generating an initial transportation path based on the scheduling optimization data set using a multi-objective optimization algorithm specifically includes: According to the scheduling optimization data set, multiple objective functions are constructed by minimizing transportation time, minimizing energy consumption, and maximizing load utilization; A weighted summation method is adopted to combine multiple objective functions to generate the initial transportation path.

3. The method for dynamic dispatching of trucks based on artificial intelligence according to claim 2 is characterized in that: The objective functions are constructed by minimizing transportation time, minimizing energy consumption and maximizing load utilization, including: Calculate the estimated time of the transportation route based on the distance, road condition information and real-time traffic conditions of the transportation route, and then construct a transportation time function; Based on the energy efficiency model of the transportation tool, the energy consumption of the transportation route is estimated in combination with the route length, load and vehicle power performance, and then the energy consumption function is constructed; By comparing the actual load of the vehicle with the maximum load capacity, a load utilization index is generated, and the load of the vehicle is optimized through a task allocation scheme, thereby constructing a load utilization function.

4. The method for dynamic dispatching of trucks based on artificial intelligence according to claim 1, characterized in that: The construction of the time series prediction model includes: Acquire historical order data, extract key time series features from the historical order data, and then construct a time series feature set; The preset long short-term memory network is iteratively trained according to the time series feature set to learn the seasonal fluctuations of the historical order data, the historical transportation load pattern and other laws, and the model weights are adjusted to minimize the prediction error to obtain the time series prediction model.

5. The method for dynamic dispatching of trucks based on artificial intelligence according to claim 4 is characterized in that: The load forecasting of the order demand data based on the time series forecasting model to obtain a corresponding load forecasting result, and then adjusting the initial transportation path according to the load forecasting result to generate an optimized transportation path specifically includes: Inputting the order demand data into the time series forecasting model to obtain the load demand of each transportation task within a preset time period, and then generating the load forecasting result; According to the load prediction result, the load demand of the vehicle and the initial path are comprehensively adjusted to generate the optimized transportation path.

6. The method for dynamic dispatching of trucks based on artificial intelligence according to claim 1, characterized in that: The generating of the transport scheduling strategy based on the transport optimization path and the transport vehicle status by using a reinforcement learning algorithm specifically includes: Constructing the state space of the reinforcement learning algorithm based on the state of the transport vehicle, and constructing the motion space of the reinforcement learning algorithm based on the transport optimization path; Obtain the transportation efficiency, load utilization and energy consumption of the transportation task, and then perform comprehensive calculations to define the corresponding reward function; The policy network of the reinforcement learning algorithm is updated according to the reward function, and the transportation scheduling strategy is output to verify the feasibility of future transportation tasks.

7. The method for dynamic dispatching of trucks based on artificial intelligence according to claim 1, characterized in that: The method further comprises: The real-time data during transportation is analyzed through anomaly detection algorithms, and when anomalies are detected, warning information is generated and emergency dispatch strategies are triggered.

8. A dynamic truck dispatching system based on artificial intelligence, characterized in that: The artificial intelligence-based dynamic truck dispatching system includes: A data processing module is used to obtain multi-dimensional transportation data, clean and extract features of the multi-dimensional transportation data, so as to generate a scheduling optimization data set; A path planning module, used to generate an initial transportation path based on the scheduling optimization data set using a multi-objective optimization algorithm; A load forecasting module is used to obtain order demand data, and perform load forecasting on the order demand data based on a time series forecasting model to obtain a corresponding load forecasting result, and then adjust the initial transportation path according to the load forecasting result to generate an optimized transportation path; The scheduling strategy generation module is used to obtain the status of the transport vehicle and generate a transport scheduling strategy through a reinforcement learning algorithm based on the transport optimization path and the status of the transport vehicle.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the dynamic truck scheduling method based on artificial intelligence as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the dynamic truck scheduling method based on artificial intelligence as described in any one of claims 1 to 7 are implemented.

Images