A vehicle management method and system based on AI vehicle platform analysis

Abstract

The present application relates to the field of vehicle dispatch management technology, and specifically to a vehicle management method and system based on AI vehicle platform analysis, the method comprising: obtaining vehicle location and order information under instant delivery business through AI vehicle platform, analyzing the estimated tolerance time of each order, analyzing the order delivery success and rejection based on the estimated tolerance time, constructing the order rejection probability distribution, and then obtaining the time sensitivity of each order at each moment; analyzing the delivery priority expectation of each vehicle control strategy according to the time sensitivity of each order, further determining the operator reward factor of each operator of the ALNS algorithm, combining the weight coefficient of each operator at each iteration, calculating the selection probability of each operator during each iteration, using the ALNS algorithm to obtain the optimal vehicle dispatch strategy, and adjusting and managing the vehicle and order delivery routes. The present application can meet the actual vehicle dispatch needs and improve the convergence speed of the algorithm.

Description

Technical Field

[0001] The present application relates to the technical field of vehicle dispatching management, and specifically to a vehicle management method and system based on AI vehicle platform analysis. Background Art

[0002] With the acceleration of urbanization, traditional vehicle dispatching methods are facing more and more challenges. By integrating advanced sensors, communication technologies and artificial intelligence algorithms, AI vehicle platforms can achieve real-time monitoring and data analysis of vehicle status, thus providing new perspectives and solutions for vehicle dispatching management. In recent years, the instant delivery business has developed rapidly, and it is necessary to use reasonable dispatching methods to achieve multi-objective optimization under limited vehicle resources, such as minimizing delivery costs, maximizing revenue and maximizing customer satisfaction.

[0003] For combinatorial optimization problems such as multi-objective optimization, heuristic search algorithms are generally used to obtain approximate optimal solutions. Among them, the adaptive large neighborhood search algorithm (ALNS) has the advantages of large neighborhood search, adaptive mechanism and damage repair operation, and is applied to solve this type of combinatorial optimization problem. However, due to the complex and changeable environment of instant delivery business, the traditional ALNS algorithm adopts a roulette wheel method to select the damage operator and the repair operator. Since the operator selection of the roulette wheel follows a linear probability distribution, it cannot adapt to the diversity needs of the actual instant delivery business, which easily leads to the selection of only well-performing operators in the actual operator selection, resulting in the defects of falling into local optimality, slow adjustment speed and insufficient diversity. Summary of the invention

[0004] In order to solve the above technical problems, the purpose of this application is to provide a vehicle management method and system based on AI vehicle platform analysis. The technical solutions adopted are as follows:

[0005] The embodiment of the present application provides a vehicle management method based on AI vehicle platform analysis, comprising the following steps:

[0006] Obtain vehicle location information and delivery order information for instant delivery services through the AI ​​vehicle platform;

[0007] For each order placed at each moment, the estimated tolerance time of each order is determined by analyzing the product type, order time and expected delivery time of each order. Based on the estimated tolerance time, the order delivery success and rejection analysis is analyzed to construct the order rejection probability distribution. Combined with the estimated tolerance time of each order, the time sensitivity of each order placed at each moment is obtained.

[0008] The simulated annealing algorithm is used to obtain the initial solution of vehicle control, including the number of dispatched vehicles, the number of orders for each vehicle, and the order delivery route of each vehicle. The operators of the ALNS algorithm are selected. In each iteration, the delivery priority expectation of each vehicle control strategy is analyzed according to the average value of the time sensitivity of all orders that each vehicle needs to deliver at their expected delivery arrival time. Based on the difference between the delivery priority expectation of each vehicle control strategy and the delivery priority expectation of each vehicle control strategy after adjustment by each operator, the operator reward factor of each operator of the ALNS algorithm is determined. Combined with the weight coefficient of each operator in each iteration, the selection probability of each operator in each iteration is calculated.

[0009] The ALNS algorithm is used according to the initial solution and the selection probability of the operator in each iteration process to obtain the optimal vehicle scheduling strategy and adjust and manage the vehicle and order delivery routes.

[0010] Furthermore, the calculation formula for the estimated tolerance time of each order is:

[0011] ; Where U represents the estimated tolerance time of the current order, and They represent the estimated delivery arrival time and order time of the current order respectively. Indicates the preset order type scaling weight.

[0012] Furthermore, the commodity types include food and commodity, wherein the order type scaling weight of the food category is smaller than the order type scaling weight of the commodity category.

[0013] Furthermore, the probability distribution expression of order rejection is: ,in, is the rejection probability of the current order at time t.

[0014] Furthermore, the calculation formula of the time sensitivity is: ,in, Indicates the time sensitivity of the current order at time t.

[0015] Furthermore, the calculation formula of the delivery priority expectation of each vehicle control strategy is:

[0016] ; Where B represents the delivery priority expectation of the current vehicle control strategy, N represents the number of orders that need to be delivered, Represents the time sensitivity of the vth order at the estimated delivery arrival time E.

[0017] Furthermore, the calculation formula corresponding to the operator reward factor is:

[0018] ;in, Representation Operator The operator reward factor of Indicates that the current vehicle control strategy is implemented by the operator The adjusted delivery priority expectation, B represents the delivery priority expectation of the current vehicle control strategy.

[0019] Furthermore, the calculation process of the selection probability of each operator in each iteration process is: calculate the product of the weight coefficient of each operator and the operator reward factor in each iteration process, and take the ratio of the product of each operator in each iteration process to the cumulative sum of the products of all operators as the selection probability of each operator in each iteration process.

[0020] Furthermore, in the first iteration, the weight coefficients of each operator are the same and the sum of the weight coefficients of all operators is 1. During the iteration process, the weight coefficient of each operator is automatically adjusted by the ALNS algorithm.

[0021] An embodiment of the present application also provides a vehicle management system based on AI vehicle platform analysis, the system comprising a memory, a processor, and a computer program stored in the memory and running on the processor, and the processor implements the steps of any one of the above methods when executing the computer program.

[0022] From the above, it can be seen that the vehicle management method and system based on AI vehicle platform analysis provided by this application have at least the following beneficial effects:

[0023] This application obtains the location information and order information of each vehicle on the AI ​​vehicle platform, and constructs the time sensitivity of a single order based on the order situation and order type of a single order customer, reflecting the priority of a single order in real time. According to the initial delivery strategy, the delivery priority expectation under different operators is obtained to obtain the operator reward factor, reflecting the exploration ability of a single operator strategy on the solution space. Combined with the operator weight coefficient, the selection probability of each operator in the iteration process is obtained, and finally the vehicle scheduling management plan is obtained by using the algorithm, and it can be adjusted dynamically. Compared with the traditional ALNS algorithm of vehicle scheduling strategy, this application avoids the use of roulette operator selection strategy, combined with the time sensitivity of each order, the delivery priority expectation obtained is dynamically changing with time, and the reward factor of the operator is obtained. According to the operator reward factor, the dynamic adjustment of the selection probability is realized, so that the selection probability of the operator is closer to nonlinearity, meeting the actual vehicle scheduling needs and improving the convergence speed of the algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present application or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0025] Figure 1 A flowchart of the steps of a vehicle management method based on AI vehicle platform analysis provided in this application;

[0026] Figure 2 A schematic diagram of the process of obtaining the selection probability of each operator in each iteration process provided by this application. DETAILED DESCRIPTION

[0027] In order to further explain the technical means and effects adopted by this application to achieve the predetermined invention purpose, the following is a detailed description of the vehicle management method and system based on AI vehicle platform analysis proposed in this application, its specific implementation method, structure, features and effects, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0028] Unless otherwise specified and limited, terms such as "comprises", "comprising" or any other variants thereof are intended to cover non-exclusive inclusion, so that a circuit structure, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such articles or devices. In the absence of further restrictions, an element defined by the sentence "comprising a ..." does not exclude the presence of other identical elements in the article or device including the element. In addition, the term "and\or" used herein includes any and all combinations of one or more related listed items. All technical and scientific terms used herein have the same meaning as those commonly understood by technicians in the technical field of this application.

[0029] The following is a detailed description of a specific solution of a vehicle management method and system based on AI vehicle platform analysis provided by the present application in conjunction with the accompanying drawings.

[0030] See also Figure 1 , which shows a flow chart of a vehicle management method based on AI vehicle platform analysis provided by an embodiment of the present application, including the following steps:

[0031] Step 100: Obtain vehicle location information and delivery order information under the instant delivery service through the AI ​​vehicle platform.

[0032] This embodiment mainly focuses on the vehicle dispatch management problem of the AI ​​vehicle platform under the instant delivery business, so that the AI ​​vehicle platform can obtain the vehicle's location information, mainly through the positioning sensor. And the platform obtains the order information that needs vehicle delivery, where a single order information includes the merchant location, customer location, product type, order time and expected delivery time.

[0033] In this way, the order data that needs to be deployed by the AI ​​vehicle platform in a certain area on each dispatch day is continuously obtained. It should be noted that in this embodiment, the AI ​​vehicle platform is a system platform that includes the information of each delivery vehicle and order information of the instant delivery service.

[0034] Step 200: For each order placed at each moment, the estimated tolerance time of each order is determined by analyzing the product type, order time and expected delivery time of each order, and the order delivery success and rejection situations are analyzed based on the estimated tolerance time. The order rejection probability distribution is constructed, and the time sensitivity of each order placed at each moment is obtained in combination with the estimated tolerance time of each order.

[0035] When a customer places an order on the instant delivery service interface, the customer’s order information will be presented on the AI ​​vehicle platform. Therefore, it is necessary to effectively dispatch the vehicle based on the location information of the vehicle and order on the platform to achieve the maximum expected benefit with the minimum resource consumption.

[0036] In order to enhance the rights and interests of customers in the instant delivery business, customers are allowed to refuse orders. However, since customers' refusal of orders will cause certain cost losses and interfere with the vehicle scheduling of the subsequent vehicle platform, the impact of customer rejection of orders needs to be considered in the actual vehicle scheduling process.

[0037] Through actual data, we found that the main reason why customers reject orders is that the orders are not delivered in time, and customers have different tolerances for delivery delays for different order types. Therefore, it is necessary to adjust the vehicle dispatch strategy according to different order situations to improve customer satisfaction.

[0038] Since there are many orders in the service system of instant delivery business at the same time, and the delivery of orders is carried out in real time. In this embodiment, the orders are basically divided into two categories: one is commodity category, and the other is food category. Generally, customers have a higher time tolerance for commodity delivery orders, and have a lower time tolerance for food delivery orders. Therefore, in order to quantify the priority of each order, the time sensitivity of each order is analyzed based on the time data of each order placed at the current moment.

[0039] First, the estimated tolerance time for each order is determined by analyzing the order product type, order time, and expected delivery time. Taking the current order as an example, the calculation formula for the estimated tolerance time for the current order is:

[0040] ; Where U represents the estimated tolerance time of the current order, and They represent the estimated delivery arrival time and order time of the current order respectively. Indicates the preset order type scaling weight, which is obtained in the following way: the scaling weight of the food order type is less than the scaling weight of the commodity order type. In this embodiment, if the current order is food, The random number of the embodiment is 0.2. If the current order is a commodity The random number in this embodiment is 0.4.

[0041] If the estimated delivery arrival time and the time interval between the current order customer are shorter, the customer's expectations for the order will be higher. If the order type is commodity, the corresponding smaller order type will have a smaller scaling weight, and the estimated tolerance time for the order will be smaller. If the order type is food, the corresponding smaller order type will have a smaller scaling weight, and the estimated tolerance time for the order will be smaller.

[0042] Based on the estimated tolerance time and combined with the analysis of successful and rejected orders in the instant delivery business, this embodiment simplifies the overview to construct the probability distribution of order rejection over time. The order rejection probability distribution is the relationship between the delivery time and the probability of order rejection, where: The probability of order rejection between time periods is 0; The probability of order rejection between time periods is ; The rejection probability of an order that is longer than the estimated tolerance time of the order is 1, and the corresponding expression is:

[0043] ;in, represents the rejection probability of the current order at time t, U represents the estimated tolerance time of the current order, and They represent the estimated delivery arrival time and order time of the current order respectively.

[0044] It can be understood that when an order can be delivered before the estimated delivery arrival time, the rejection probability of the order is zero; when an order is delivered between the estimated delivery arrival time and the estimated tolerance time, the rejection probability of the order increases linearly with the increase of time; when an order exceeds the estimated tolerance time, the order will be completely rejected, which will directly cause cost losses.

[0045] As time changes, the time sensitivity A of the current order at each moment is obtained by combining the rejection probability distribution with the estimated tolerance time of the order:

[0046] ;in, represents the time sensitivity of the current order at time t, represents the rejection probability of the current order at time t, which is obtained through the order rejection probability distribution graph, represents the estimated tolerance time of the current order, and t represents the tth moment. It should be noted that if , it indicates that the customer's estimated tolerance time has been exceeded and the order has not yet been delivered to the customer. At this time, the time sensitivity of the current order is set to 10.

[0047] As the delivery time increases, the time sensitivity for different orders is different. If the current order is a food order with a high expected value, the interval between the estimated tolerance time and the expected delivery time is small. At this time, the time sensitivity of the order increases rapidly with the increase of time, and the time sensitivity value of the order at a single moment is large. On the contrary, if the current order is a commodity order with a small expected value, the interval between the estimated tolerance time and the expected delivery time is large, the time sensitivity of the order increases relatively slowly over time, and the time sensitivity value of the order at a single moment is small.

[0048] Step 300: Use the simulated annealing algorithm to obtain the initial solution of vehicle control, including the number of scheduled vehicles, the number of orders for each vehicle, and the order delivery route of each vehicle, and select the operators of the ALNS algorithm. In each iteration, analyze the delivery priority expectations of each vehicle control strategy based on the average value of the time sensitivity of all orders that need to be delivered by each vehicle at their estimated delivery arrival time.

[0049] The core idea of ​​the ALNS algorithm is large neighborhood search, which is mainly achieved through the destruction operator and the repair operator. The destruction operator deletes some nodes from the current vehicle distribution strategy and destroys the current feasible solution; the repair operator inserts some nodes into the distribution strategy after the destruction according to a certain set rule to generate a new solution. Through the destruction operator and the repair operator, it is possible to explore implicit solutions in a larger neighborhood within the current local range, and by continuously expanding the solution space, the optimal solution of the algorithm as a whole can be found.

[0050] Usually, the ALNS algorithm designs multiple groups of operators, each of which includes a destruction operator and a repair operator, and uses the roulette strategy to obtain the destruction operator and repair operator selected for the next exploration. However, the probability of roulette obeys a linear distribution, which limits the exploration of better operator combinations, making it easy to fall into local optimality and slow adjustment. Therefore, it is necessary to dynamically adjust the operator selection probability based on the time sensitivity of different orders.

[0051] In this embodiment, the most widely used destruction operators and repair operators are selected. Specifically, the destruction operators include: random destruction operators , Worst Damage Operator , Neighborhood destruction operator ; Repair operators include: random repair operators , Greedy repair operator , the nearest insertion operator The destruction operator and the repair operator form a pair of operators of the ALNS algorithm. In this embodiment, there are three pairs of operators of the ALNS algorithm. , , wherein the rules of the destruction operator and the repair operator are well-known techniques and will not be described in detail in this embodiment.

[0052] The simulated annealing algorithm is used to generate an initial control solution, that is, the initial vehicle control strategy, which specifically includes the number of vehicles n, the number of orders for each vehicle, and the route planning for the delivery of each vehicle's orders. The delivery priority expectation under each control strategy is calculated. Taking the current control strategy as an example, it is specifically:

[0053] ; Where B represents the delivery priority expectation of the current vehicle control strategy, N represents the number of orders that need to be delivered, represents the time sensitivity of the vth order at the estimated delivery arrival time E. It should be noted that, due to the above is a function of time. Only by determining the time t can the function value be determined. Therefore, in this embodiment, Through the above function about time A calculation is performed to determine the time sensitivity based on the estimated delivery arrival time of the order.

[0054] If the current control strategy contains more urgent delivery orders, the time sensitivity value corresponding to the urgent orders is higher. At this time, the expected value of delivery priority is larger, indicating that the current delivery strategy is more unreasonable and needs to be adjusted.

[0055] Step 400: Based on the difference between the delivery priority expectation of each vehicle control strategy and the delivery priority expectation of each vehicle control strategy after adjustment of each operator, determine the operator reward factor of each operator of the ALNS algorithm, and calculate the selection probability of each operator in each iteration process in combination with the weight coefficient of each operator in each iteration.

[0056] According to step 300, the delivery priority expectation of each vehicle control strategy can be obtained. Further, based on the difference between the delivery priority expectation of each vehicle control strategy and the delivery priority expectation of each vehicle control strategy after adjustment of each operator, the operator reward factor of each operator of the ALNS algorithm is determined. The calculation formula of the operator reward factor of each operator is:

[0057] ;in, Representation Operator The operator reward factor of , , Indicates that the current vehicle control strategy is implemented by the operator Adjusted delivery priority expectation, wherein the specific process of the ALNS algorithm optimizing the control strategy according to the operator is the prior art, and B represents the delivery priority expectation of the current vehicle control strategy.

[0058] It is understandable that if the operator After processing, the expected delivery priority of the current strategy changes significantly, indicating that the operator The processing method can explore a larger solution space, that is, the operator After processing, it is possible to explore a larger neighborhood and find the optimal delivery strategy for a large neighborhood within the scope of a larger vehicle delivery scheduling strategy, thereby obtaining the operator On the contrary, if the distribution priority expectation changes little after being processed by this operator, it means that a smaller solution space can be explored, that is, the possible vehicle scheduling strategy set is small, and only the local optimal distribution strategy within a small range is solved, then the operator reward weight corresponding to this operator is small.

[0059] According to the above process, the operator reward factor of each operator under the current vehicle control strategy can be obtained. Combined with the weight coefficient of each operator at each iteration in the ALNS algorithm process, the selection probability of each operator can be obtained.

[0060] According to the operator reward factor of each operator, combined with the weight coefficient of each operator in each iteration, the selection probability of each operator in each iteration process is calculated. Specifically, the product of the weight coefficient of each operator and the operator reward factor in each iteration process is calculated, and the ratio of the product of each operator in each iteration process to the cumulative sum of the products of all operators is used as the selection probability of each operator in each iteration process. The specific process is as follows: Figure 2As shown, in this embodiment, the calculation formula corresponding to the selection probability is:

[0061] ;in, is the operator in the kth iteration The probability of selection, For operator The operator reward factor of is the operator in the kth iteration The weight coefficients of the three operators are the same during the first iteration, and the sum of the weight coefficients of all operators is 1. In this embodiment, the weight coefficients of the three operators during the first iteration are However, as the iteration proceeds, ALNS will adjust the weight of the operator, giving higher weights to operators with better performance and higher scores. Therefore, in the subsequent iteration process, the weight coefficient of the operator is obtained by the ALNS algorithm. The process of the ALNS algorithm automatically adjusting the weight coefficient of the operator is a prior art and will not be repeated in this embodiment.

[0062] This embodiment uses the operator reward factor and the weight coefficient of the algorithm to obtain the selection probability of each operator in the iteration process. According to the different time sensitivities of the orders in the strategy, the reward factor corresponding to the operator with greater exploration ability is larger, so the probability of being selected is greater. Therefore, the operator is not only selected by the weight coefficient of the operator, thereby improving the diversity of operator selection and the quality of the solution.

[0063] Step 500: According to the initial solution and the selection probability of the operator in each iteration process, the ALNS algorithm is used to obtain the optimal vehicle scheduling strategy and adjust and manage the vehicle and order delivery routes.

[0064] According to step 400, the selection probability of different operators in each iteration process is obtained, thereby realizing the selection of different operator combinations. In the actual delivery process, the parameters of this embodiment are set as follows: the delivery cost of the vehicle is 1 yuan / km, the profit of successfully delivered orders is 10, and the profit of rejected orders is -20. The acceptance criterion of the iterative strategy adopts the Metropolis criterion of the simulated annealing algorithm. This embodiment takes minimizing the transportation cost as the objective function. It should be noted that the implementer of the objective function can set it according to the actual application scenario. In this embodiment, the cumulative sum of the delivery costs of all vehicles is used as the objective function to ensure the minimization of transportation costs.

[0065] Then, the simulated annealing algorithm is used to obtain the initial solution, and the initial solution and the selection probability of the operator in each iteration process are brought into the ALNS algorithm. The maximum number of iterations of the algorithm is set to 500 times, and the algorithm outputs the optimal vehicle scheduling strategy, including the number of vehicles that need to be scheduled, the number of orders for each vehicle, and the delivery route of each vehicle order. It should be noted that in the process of real-time regulation of the vehicle's delivery method, when a new order is added, the orders being delivered are not affected. For the newly added orders that need to be delivered, the above-mentioned method and process of this embodiment are used to perform vehicle scheduling management according to the vehicle location and order information. The specific implementation process of the ALNS algorithm and the simulated annealing algorithm are both well-known technologies in the field of scheduling optimization, and this implementation will not be repeated here.

[0066] By distributing the vehicle scheduling strategy of the AI ​​vehicle management platform to each vehicle, the scheduling management of the vehicles can be realized, which can achieve the maximum benefit effect at the lowest cost.

[0067] Based on the same inventive concept as the above method, an embodiment of the present application also provides a vehicle management system based on AI vehicle platform analysis, including a memory, a processor, and a computer program stored in the memory and running on the processor, and when the processor executes the computer program, the steps of any one of the above-mentioned vehicle management methods based on AI vehicle platform analysis are implemented.

[0068] It is to be understood that the sequence of the embodiments of the present application described above is for description only and does not represent the advantages and disadvantages of the embodiments. The above describes specific embodiments of the present specification. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0069] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0070] The above content is only an implementation method of the present application and is not intended to limit the scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly used in other related technical fields, are also included in the protection scope of the present application.

Claims

1. A vehicle management method based on AI vehicle platform analysis, characterized in that: The following steps are involved: Obtain vehicle location information and delivery order information for instant delivery services through the AI ​​vehicle platform; For each order placed at each moment, the estimated tolerance time of each order is determined by analyzing the product type, order time and expected delivery time of each order. Based on the estimated tolerance time, the order delivery success and rejection analysis is analyzed to construct the order rejection probability distribution. Combined with the estimated tolerance time of each order, the time sensitivity of each order placed at each moment is obtained. The simulated annealing algorithm is used to obtain the initial solution of vehicle control, including the number of dispatched vehicles, the number of orders for each vehicle, and the order delivery route of each vehicle. The operators of the ALNS algorithm are selected. In each iteration, the delivery priority expectation of each vehicle control strategy is analyzed according to the average value of the time sensitivity of all orders that each vehicle needs to deliver at their expected delivery arrival time. Based on the difference between the delivery priority expectation of each vehicle control strategy and the delivery priority expectation of each vehicle control strategy after adjustment by each operator, the operator reward factor of each operator of the ALNS algorithm is determined. Combined with the weight coefficient of each operator in each iteration, the selection probability of each operator in each iteration is calculated. The ALNS algorithm is used according to the initial solution and the selection probability of the operator in each iteration process to obtain the optimal vehicle scheduling strategy and adjust and manage the vehicle and order delivery routes; The expression of the order rejection probability distribution is: ,in, is the rejection probability of the current order at time t, and They represent the estimated delivery arrival time and order time of the current order respectively, and U represents the estimated tolerance time of the current order; The calculation formula of the time sensitivity is: ,in, Indicates the time sensitivity of the current order at time t; The calculation formula for the delivery priority expectation of each vehicle control strategy is: ; Where B represents the delivery priority expectation of the current vehicle control strategy, N represents the number of orders that need to be delivered, represents the time sensitivity of the vth order at the estimated delivery arrival time E; The calculation formula corresponding to the operator reward factor is: ;in, Representation Operator The operator reward factor of Indicates that the current vehicle control strategy is implemented by the operator The adjusted delivery priority expectation, B represents the delivery priority expectation of the current vehicle control strategy.

2. A vehicle management method based on AI vehicle platform analysis as claimed in claim 1, characterized in that: The calculation formula for the estimated tolerance time of each order is: ; Where U represents the estimated tolerance time of the current order, and They represent the estimated delivery arrival time and order time of the current order respectively. Indicates the preset order type scaling weight.

3. A vehicle management method based on AI vehicle platform analysis as claimed in claim 2, characterized in that: The commodity types include food and commodity, wherein the order type scaling weight of the food type is smaller than the order type scaling weight of the commodity type.

4. A vehicle management method based on AI vehicle platform analysis as claimed in claim 1, characterized in that: The calculation process of the selection probability of each operator in each iteration process is as follows: the product of the weight coefficient of each operator and the operator reward factor is calculated in each iteration process, and the ratio of the product of each operator in each iteration process to the cumulative sum of the products of all operators is used as the selection probability of each operator in each iteration process.

5. A vehicle management method based on AI vehicle platform analysis as claimed in claim 4, characterized in that: In the first iteration, the weight coefficients of each operator are the same and the sum of the weight coefficients of all operators is 1. During the iteration process, the weight coefficients of each operator are automatically adjusted by the ALNS algorithm.

6. A vehicle management system based on AI vehicle platform analysis, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.

Images