Delivery time adjustment method, device and electronic equipment

Abstract

The embodiments of the present application provide a delivery time adjustment method, device, electronic device and computer-readable storage medium, which relate to the computer field. The embodiments of the present application implement a target adjustment information corresponding to the first estimated delivery time of a target order based on a reinforcement learning model and according to the characteristics of the order parameter information of the target order. Since the preset adjustment model makes full use of the relevant order parameter information of the target order in the prediction process, combined with the characteristics of the order parameter information of the target order, the first estimated delivery time is adjusted to improve the accuracy of the delivery time prediction.

Description

Technical Field

[0001] The present application relates to the field of computer technology. Specifically, the present application relates to a delivery time adjustment method, device, electronic device and computer-readable storage medium. Background Art

[0002] With the continuous development of Internet technology, online ordering has gradually become a new choice for the public. In the online ordering consumption mode, the instant delivery of ordered goods brings users a convenient consumption experience.

[0003] In scenarios such as ordering goods online, after receiving an order from a user to purchase goods through a client, the shopping platform can determine the estimated order delivery time based on the relevant information of the order (such as the preparation time, order distance, and other information). Since the estimated delivery time is an estimated delivery time based on the relevant information of the order, the estimated delivery time inevitably has accuracy issues. In related technologies, the estimated delivery time can be adjusted to ensure a more accurate delivery time displayed to the user. However, in related technologies, the accuracy of the adjusted estimated delivery time is low. Summary of the invention

[0004] The purpose of this application is to solve at least one of the above-mentioned technical defects, especially the technical defect of low accuracy in estimating delivery time.

[0005] According to one aspect of the present application, a delivery time adjustment method is provided, the method comprising: obtaining order parameter information of a target order and a first estimated delivery time;

[0006] Inputting the order parameter information and the first estimated delivery time into a preset adjustment model to obtain a first weight value corresponding to each preset candidate adjustment information; the first weight value indicates a probability corresponding to each candidate adjustment information;

[0007] The candidate adjustment information includes a first adjustment operation and an operation duration of the first adjustment operation; the first adjustment operation includes an increase or decrease operation for the first estimated delivery time; the preset adjustment model is obtained by reinforcement learning of a training sample set;

[0008] Target adjustment information is determined based on the first weight value and the candidate adjustment information.

[0009] Optionally, before obtaining the order parameter information of the target order and the first estimated delivery time, the method includes:

[0010] Obtain a training sample set; wherein the training sample set includes: sample order parameter information, sample estimated delivery time, and sample actual delivery time of the sample order;

[0011] For each of the sample orders, input the sample order parameter information and the sample estimated delivery time into the initial training model to obtain a prediction weight value for each of the candidate adjustment information;

[0012] Determining a first delivery accuracy rate of the sample order after performing the candidate adjustment operation according to the predicted weight value and the actual delivery time of the sample;

[0013] Determining a training loss value according to the first delivery accuracy rate;

[0014] Based on the sample order information and the training loss value corresponding to the sample order, the neural network model is repeatedly trained until the preset adjustment model that meets the training end conditions is obtained.

[0015] Optionally, determining a training loss value according to the first delivery accuracy rate includes:

[0016] For each of the sample orders, an accuracy improvement value is determined according to the first delivery accuracy of the sample order and the actual delivery accuracy of the sample order, wherein the actual delivery accuracy of the sample order is determined according to the sample actual delivery time of the sample order;

[0017] The training loss value is determined based on the accuracy improvement value and the correction value, wherein the correction value is determined based on the relationship between the predicted adjustment information and the first preset adjustment information, and the predicted adjustment information is determined based on each of the candidate adjustment information and its corresponding predicted weight value.

[0018] Optionally, the data relationship corresponding to the training loss value is as follows:

[0019]

[0020] Wherein, L(θ|s) represents the training loss value, represents the gradient value of x; E(x) represents the expected value of x;

[0021] S represents the sample order, A represents the set of candidate adjustment information,

[0022] pθ(A|s) represents the probability of S executing each candidate adjustment information in A;

[0023] r(A) represents the first delivery accuracy rate;

[0024] b(s) represents the actual delivery accuracy;

[0025] λ*l(A) represents the correction value, λ represents the preset adjustment intensity coefficient,

[0026] l(A) represents the correction term, l(A) = max((t(A) - t u ),0)+max(-(t(A)-t l ), 0), t(A) represents the predicted adjustment time of the predicted adjustment information, [t l ,t u ] represents the preset adjustment time interval of the first preset adjustment information.

[0027] Optionally, the order parameter information includes at least one of the following:

[0028] Number of blocking orders;

[0029] Estimated lead time for orders;

[0030] Average order preparation time in history;

[0031] Order delivery distance;

[0032] Average delivery time of orders in history.

[0033] Optionally, the method further includes:

[0034] Determining a second estimated delivery time according to the first estimated delivery time and the target adjustment information;

[0035] The second estimated delivery time is sent to the client, and the client is instructed to display the second estimated delivery time on a user interface.

[0036] Optionally, the determining the target adjustment information based on the first weight value and the candidate adjustment information includes at least one of the following:

[0037] Determine the candidate adjustment information having the largest corresponding first weight value as the target adjustment information;

[0038] According to each of the candidate adjustment information and its respectively corresponding first weight value, weighted processing is performed on the operation duration of each of the first adjustment operations to obtain the target adjustment information.

[0039] According to one aspect of the present application, a delivery time adjustment method is provided, the method comprising: receiving a second estimated delivery time sent by a server, the second estimated delivery time being determined according to a first estimated delivery time and target adjustment information, the target adjustment information being determined based on each candidate adjustment information and its corresponding first weight value, the first weight value corresponding to each candidate adjustment information being determined by a preset adjustment model based on order parameter information and the first estimated delivery time;

[0040] The first weight value indicates the probability corresponding to each of the candidate adjustment information;

[0041] The candidate adjustment information includes a first adjustment operation and an operation duration of the first adjustment operation; the first adjustment operation includes an increase or decrease operation for the first estimated delivery time; the preset adjustment model is obtained by machine learning of a training sample set;

[0042] The second estimated delivery time is displayed on the user interface.

[0043] According to another aspect of the present application, a delivery time adjustment device is provided, the device comprising:

[0044] An acquisition module, used to acquire order parameter information and a first estimated delivery time of a target order;

[0045] A prediction module, configured to input the order parameter information and the first estimated delivery time into a preset adjustment model to obtain a first weight value corresponding to each preset candidate adjustment information; the first weight value indicates a probability corresponding to each candidate adjustment information;

[0046] The candidate adjustment information includes a first adjustment operation and an operation duration of the first adjustment operation; the first adjustment operation includes an increase or decrease operation for the first estimated delivery time; the preset adjustment model is obtained by reinforcement learning of a training sample set;

[0047] A determination module is used to determine target adjustment information based on the first weight value and the candidate adjustment information.

[0048] According to another aspect of the present application, a delivery time adjustment device is provided, the device comprising:

[0049] a receiving module, configured to receive a second estimated delivery time sent by a server, wherein the second estimated delivery time is determined according to the first estimated delivery time and target adjustment information, wherein the target adjustment information is determined based on each candidate adjustment information and its corresponding first weight value, and the first weight value corresponding to each candidate adjustment information is determined by a preset adjustment model based on order parameter information and the first estimated delivery time;

[0050] The first weight value indicates the probability corresponding to each of the candidate adjustment information;

[0051] The candidate adjustment information includes a first adjustment operation and an operation duration of the first adjustment operation; the first adjustment operation includes an increase or decrease operation for the first estimated delivery time; the preset adjustment model is obtained by machine learning of a training sample set;

[0052] A display module is used to display the second estimated delivery time on a user interface.

[0053] According to another aspect of the present application, an electronic device is provided, the electronic device comprising:

[0054] one or more processors;

[0055] Memory;

[0056] One or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the one or more processors, and the one or more programs are configured to: execute the delivery time adjustment method described in any one of the first or second aspects.

[0057] For example, in a third aspect of the present application, a computing device is provided, including: a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus;

[0058] The memory is used to store at least one executable instruction, and the executable instruction enables the processor to perform operations corresponding to the delivery time adjustment method as described in any one of the first or second aspects of the present application.

[0059] According to another aspect of the present application, a computer-readable storage medium is provided.

[0060] When the computer program is executed by a processor, the delivery time adjustment method described in any one of the first or second aspects is implemented.

[0061] For example, the fourth aspect of an embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, the delivery time adjustment method described in any one of the first or second aspects of the present application is implemented.

[0062] According to one aspect of the present application, a computer program product or a computer program is provided, the computer program product or the computer program includes computer instructions, and the computer instructions are stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the method provided in various optional implementations of the first aspect or the second aspect.

[0063] The beneficial effects of the technical solution provided by this application are:

[0064] The embodiment of the present application obtains order parameter information and a first estimated delivery time of a target order; inputs the order parameter information and the first estimated delivery time into a preset adjustment model to obtain a first weight value corresponding to each preset candidate adjustment information; determines the target adjustment information based on the first weight value and the candidate adjustment information; and implements a reinforcement learning model based on the characteristics of the order parameter information of the target order to predict the target adjustment information corresponding to the first estimated delivery time of the target order. Since the preset adjustment model makes full use of the relevant order parameter information of the target order in the prediction process, and combines the characteristics of the order parameter information of the target order to adjust the first estimated delivery time, the accuracy of the delivery time prediction is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in describing the embodiments of the present application are briefly introduced below.

[0066] Figure 1 A system architecture diagram of a delivery time adjustment method provided in an embodiment of the present application;

[0067] Figure 2 A schematic diagram of the framework structure of a delivery time adjustment system provided in an embodiment of the present application;

[0068] Figure 3 One of the flow charts of a delivery time adjustment method provided in an embodiment of the present application;

[0069] Figure 4 A second flow chart of a delivery time adjustment method provided in an embodiment of the present application;

[0070] Figure 5 One of the structural schematic diagrams of a delivery time adjustment device provided in an embodiment of the present application;

[0071] Figure 6 A second structural diagram of a delivery time adjustment device provided in an embodiment of the present application;

[0072] Figure 7 A schematic diagram of the structure of an electronic device for adjusting delivery duration provided in an embodiment of the present application. DETAILED DESCRIPTION

[0073] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and cannot be interpreted as limiting the present application.

[0074] It will be understood by those skilled in the art that, unless expressly stated, the singular forms "one", "said", and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present application refers to the presence of the features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof. It should be understood that when we refer to an element as being "connected" or "coupled" to another element, it may be directly connected or coupled to the other element, or there may be an intermediate element. In addition, the "connection" or "coupling" used herein may include wireless connection or wireless coupling. The term "and / or" used herein includes all or any unit and all combinations of one or more associated listed items.

[0075] In order to make the objectives, technical solutions and advantages of the present application clearer, the implementation methods of the present application will be further described in detail below with reference to the accompanying drawings.

[0076] At least part of the content of the delivery time adjustment method provided in the embodiment of the present application involves fields such as machine learning in the field of artificial intelligence, and also involves various fields of cloud technology, such as cloud computing and cloud services in cloud technology, and related data computing and processing fields in the field of big data.

[0077] Artificial Intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technology in computer science that attempts to understand the essence of intelligence and produce a new intelligent machine that can respond in a similar way to human intelligence. Artificial intelligence is to study the design principles and implementation methods of various intelligent machines so that machines have the functions of perception, reasoning and decision-making.

[0078] Artificial intelligence technology is a comprehensive discipline that covers a wide range of fields, including both hardware-level and software-level technologies. The basic technologies of artificial intelligence generally include sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics and other technologies. Artificial intelligence software technology mainly includes computer vision technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0079] Machine Learning (ML) is a multi-disciplinary subject involving probability theory, statistics, approximation theory, convex analysis, algorithm complexity theory and other disciplines. It specializes in studying how computers simulate or implement human learning behavior to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is the core of artificial intelligence and the fundamental way to make computers intelligent. Its applications are spread across all areas of artificial intelligence. Machine learning and deep learning usually include artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and self-learning.

[0080] To further illustrate the technical solution provided by the embodiment of the present application, this is described in detail below in conjunction with the accompanying drawings and specific implementation methods. Although the embodiment of the present application provides the method operation steps shown in the following embodiments or drawings, more or fewer operation steps may be included in the method based on routine or no creative labor. In the steps where there is no necessary causal relationship logically, the execution order of these steps is not limited to the execution order provided in the embodiment of the present application.

[0081] First combine Figure 1 , which is a system architecture diagram of the delivery time adjustment method provided in the embodiment of the present application. The system may include a server 101 and a terminal cluster, wherein the server 101 may be considered as a background server that provides delivery time adjustment processing.

[0082] The terminal cluster may include: terminal 102, terminal 103, terminal 104, ..., wherein the terminal is installed with a client supporting delivery time adjustment processing or commodity trading, for example, the terminal may include a delivery person terminal and a user terminal. There may be a communication connection between the terminals, for example, there is a communication connection between terminal 102 and terminal 103, and there is a communication connection between terminal 103 and terminal 104.

[0083] At the same time, server 101 can provide services for the terminal cluster through the communication connection function. Any terminal in the terminal cluster can have a communication connection with server 101, for example, there is a communication connection between terminal 102 and server 101, and there is a communication connection between terminal 103 and server 101. The above-mentioned communication connection does not limit the connection method, and can be directly or indirectly connected through wired communication, directly or indirectly connected through wireless communication, or through other methods.

[0084] The network for the communication connection may be a wide area network or a local area network, or a combination of the two. This application does not limit this.

[0085] The delivery time adjustment method of the embodiment of the present application can be executed on the server side or on the terminal side, and the execution subject is not limited in the embodiment of the present application. During the delivery time adjustment process, the user can trigger an order operation through the client installed on the terminal device, and the server adjusts the first estimated delivery time (i.e., performs an increase or decrease process) according to the order parameter information and the first estimated delivery time, and sends the adjusted second estimated delivery time to the terminal, and then the terminal displays the second estimated delivery time to the user through the client.

[0086] Therefore, the method provided in the embodiment of the present application can be executed by a computer device, and the computer device includes but is not limited to a terminal (including the above-mentioned user terminal) or a server (including the above-mentioned server 101). The above-mentioned server can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The above-mentioned terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, a smart speaker, a smart watch, etc., but is not limited to this. The terminal and the server can be directly or indirectly connected via wired or wireless communication, and this application is not limited here.

[0087] Of course, the method provided in the embodiment of the present application is not limited to Figure 1 The application scenarios shown can also be used in other possible application scenarios, and the embodiments of the present application are not limited thereto. Figure 1 The functions that can be implemented by each device in the application scenario shown will be described in the subsequent method embodiments, and will not be described in detail here.

[0088] As an optional implementation, Figure 2 A schematic diagram of the framework structure of a delivery time adjustment system provided by the present application is shown in FIG. Figure 2 As shown in , the system includes two parts: offline and online. The offline part mainly performs model training. The online part mainly uses the model prediction function to adjust the initial estimated delivery time (first estimated delivery time) of the order based on the actual order situation of the user.

[0089] like Figure 1 As shown, the processing performed by the online part of the embodiment of the present application includes:

[0090] (1) After the user triggers an order operation through the client, obtain the order parameter information of the target order;

[0091] (2) Obtaining an initial estimated delivery time. For the convenience of description, the initial estimated delivery time may be referred to as a first estimated delivery time. The first estimated delivery time may be predicted by a learning model for predicting delivery time.

[0092] (3) Through the preset adjustment model of the embodiment of the present application, based on the order parameter information and the first estimated delivery time, the first weight value corresponding to each candidate adjustment information for adjusting the first estimated delivery time is predicted, and based on the first weight value and the candidate adjustment information, the target adjustment information for adjusting the first estimated delivery time is output.

[0093] (4) The first estimated delivery time is adjusted based on the target adjustment information to obtain a second estimated delivery time, i.e., the estimated delivery time displayed to the user.

[0094] The processing performed by the offline part of the embodiment of the present application:

[0095] The idea of ​​reinforcement learning is adopted to train the initial model to obtain the preset adjustment model:

[0096] Among them, the reinforcement learning model can adopt reinforcement learning (RL), deep reinforcement learning (Deep Q Network, DQN), deep deterministic policy gradient (DDPG), deep reinforcement learning algorithm (Actor-Critic Algorithm, A3C) and other models, which are not limited here.

[0097] In the embodiment of the present application, the Action and Reward of reinforcement learning are specifically designed as follows:

[0098] Action: candidate adjustment information for the target order, including the operation of adding or subtracting the first estimated delivery time and the corresponding operation duration.

[0099] Reward: The delivery accuracy of the target order after the candidate adjustment operation is performed on the target order based on each candidate adjustment information.

[0100] The embodiment of the present application combines reinforcement learning training, and uses the adjustment operation (addition or subtraction operation) of the first estimated delivery time as the behavior action in reinforcement learning. The delivery accuracy of the target order under the adjustment of each behavior action is used as the reward Reward of reinforcement learning. By maximizing the reward Reward, the behavior Action corresponding to the target order is determined, that is, the operation of adding or subtracting the first estimated delivery time, and the corresponding operation duration.

[0101] The embodiment of the present application provides a possible implementation method, which can be executed by any electronic device. Optionally, any electronic device can be a server device with the ability to adjust the delivery time, or a device or chip integrated in these devices. Figure 3 As shown, it is one of the flow charts of a delivery time adjustment method provided in an embodiment of the present application, and the method includes the following steps:

[0102] Step S301: Obtain order parameter information and a first estimated delivery time of a target order.

[0103] In the embodiment of the present application, the target order may include an order corresponding to an order operation triggered by a user (buyer user) through a user client. Optionally, in a specific scenario, the target order may include an order for a user to purchase items in an online store, a meal order, etc.

[0104] The order parameter information represents the relevant parameter information involved in the target order. The order parameter information may specifically include user identification information (Identity document, ID), information about the items or meals ordered by the user, the amount paid for the order, the number of blocked orders, the estimated preparation time for the order, the historical average preparation time for the order, the order delivery distance, the historical average delivery time for the order, and other information.

[0105] Optionally, information such as user ID, ordered items or meal information, order payment amount, order delivery distance, etc. in the order parameter information can be obtained by querying relevant information of the target order after the user triggers the order operation; information such as the number of blocked orders, historical average preparation time for orders, historical average delivery time for orders, etc. in the order parameter information can be obtained by querying the database; the estimated preparation time for the order in the order parameter information can be estimated and determined based on the number of items ordered by the user and the time to prepare each item.

[0106] The first estimated delivery time indicates the estimated delivery time of the target order. The first estimated delivery time can be estimated based on the average delivery time of the order history. In addition, the first estimated delivery time can be predicted based on the order parameter information such as the number of blocked orders, the estimated order preparation time, the average order preparation time in history, the order delivery distance, and the average order delivery time in history, through a learning model for predicting the time; wherein the learning model for predicting the time can be pre-trained for the actual delivery time of the historical orders learned in the actual training.

[0107] Step S302: input the order parameter information and the first estimated delivery time into a preset adjustment model to obtain a first weight value corresponding to each preset candidate adjustment information.

[0108] The first weight value indicates the probability corresponding to each candidate adjustment information, and the probability is the probability of executing the candidate adjustment information on the first estimated delivery time.

[0109] The candidate adjustment information includes a first adjustment operation and the operation duration of the first adjustment operation; the first adjustment operation includes an overtime operation or a subtraction operation for the first estimated delivery time; an overtime operation includes, for example, an operation of adding 1 minute or 2 minutes to the first estimated delivery time, and a subtraction operation includes, for example, an operation of subtracting 1 minute or 2 minutes from the first estimated delivery time.

[0110] The preset adjustment model is obtained by performing reinforcement learning on the training sample set.

[0111] In the embodiment of the present application, the preset adjustment model is a pre-trained model. Optionally, the preset adjustment model can be obtained based on reinforcement learning training. Specifically, in the embodiment of the present application, reinforcement learning includes actions and rewards, and the specific design is as follows:

[0112] Action: candidate adjustment information for the target order, the candidate adjustment information includes a first adjustment operation and the operation duration of the first adjustment operation; the first adjustment operation includes an overtime operation or a time reduction operation for the first estimated delivery time; for example, the candidate adjustment information may include, for example, 1 minute overtime, 2 minutes overtime, 1 minute overtime, 2 minutes overtime, etc. The candidate adjustment information may be pre-set according to the time requirement for the actual delivery time. For example, in the actual order delivery process, the first estimated delivery time is 20 minutes, and the time requirement for the actual delivery time is not more than 30 minutes. Then, the operation duration of the overtime operation of the candidate adjustment information is less than 10 minutes. In addition, the operation duration in the candidate adjustment information may be directly set, for example, the operation duration of the overtime operation may be set to 5 minutes to 15 minutes, etc.

[0113] Reward: The delivery accuracy of the target order after the candidate adjustment operation is performed on the target order based on each candidate adjustment information.

[0114] The embodiment of the present application introduces a reinforcement learning training method, and uses the adjustment operation (addition or subtraction operation) of the first estimated delivery time as the behavior action in reinforcement learning; the delivery accuracy of the target order under the adjustment of each behavior action is used as the reward Reward of reinforcement learning; and the optimal behavior Action corresponding to the target order is determined by maximizing the reward Reward, that is, the addition or subtraction operation of the first estimated delivery time and the corresponding operation duration are determined through the optimal delivery accuracy.

[0115] In the actual prediction process, the first weight value corresponding to each candidate adjustment information can be predicted based on the preset adjustment model, the order parameter information and the first estimated delivery time; the first weight value is the probability value corresponding to each candidate adjustment information. In this way, when determining the target adjustment information corresponding to the first estimated delivery time, it can be determined based on each candidate adjustment information and its corresponding first weight value.

[0116] Step S303: Determine target adjustment information based on the first weight value and the candidate adjustment information.

[0117] Specifically, when determining the target adjustment information of the first estimated delivery time, the candidate adjustment information with the largest first weight value may be determined as the target adjustment information. In addition, based on the first weight value corresponding to each candidate adjustment information, one of the candidate adjustment information may be extracted from the candidate adjustment information as the target adjustment information. It is understandable that the probability of each candidate adjustment information being extracted is the first weight value corresponding to the candidate adjustment information.

[0118] In addition, the operation duration of each first adjustment operation may be weighted according to each candidate adjustment information and its corresponding first weight value, and the target adjustment information may be determined according to the weighted operation duration. For example, the candidate adjustment information and its corresponding first weight values ​​are:

[0119] Candidate adjustment information 1: 3 minutes of overtime, the corresponding first weight value is 0.2;

[0120] Candidate adjustment information 2: 5 minutes of overtime, the corresponding first weight value is 0.5;

[0121] Candidate adjustment information 3: 7 minutes of overtime, the corresponding first weight value is 0.3;

[0122] The operation duration of each candidate adjustment information is weighted and calculated as: 3*0.2+5*0.5+7*0.3=5.2; therefore, the target adjustment information after weighted processing is an additional time of 5.2 minutes.

[0123] For another example, when there are time-adding and time-reducing operations in the candidate adjustment information, when weighted calculation is performed on the operation duration, the operation duration of the corresponding time-reducing operation may be subtracted. For example, the candidate adjustment information and its corresponding first weight values ​​are respectively:

[0124] Candidate adjustment information 1: 3 minutes of overtime, the corresponding first weight value is 0.2;

[0125] Candidate adjustment information 2: time reduction of 1 minute, the corresponding first weight value is 0.5;

[0126] Candidate adjustment information 3: 2 minutes of overtime, the corresponding first weight value is 0.3;

[0127] The operation duration of each candidate adjustment information is weighted and calculated as: 3*0.2-1*0.5+2*0.3=0.7; therefore, the target adjustment information after weighted processing is an additional time of 0.7 minutes.

[0128] Based on the above method, combined with Example 1, the application scenario of the embodiment of the present application is described:

[0129] In the scenario where the user orders food through an online store, after the user triggers the ordering operation through the client, the client will usually display the estimated delivery time or the estimated delivery time in the user interface. It should be noted that in the embodiment of the present application, the above-mentioned client can be the client of the delivery person, that is, the estimated delivery time is displayed in the user interface of the delivery end; the above-mentioned client can also be the client of the user, that is, the estimated delivery time is displayed in the user interface of the user end. In addition, the estimated delivery time displayed in the above-mentioned user interface is obtained by adjusting the initial estimated delivery time (the first estimated delivery time in the embodiment of the present application) through the method of the embodiment of the present application, that is, it is obtained by adding or subtracting the adjustment operation on the initial estimated delivery time.

[0130] Specifically, the embodiments of the present application can obtain order parameter information of the order after the user triggers the order operation. For example, the order parameter information may include user ID, ordered food information, order payment amount, number of blocked orders, estimated order delivery time, historical average order delivery time, order delivery distance, historical average order delivery time and other information; and obtain the first estimated delivery time of the order, and then input the order parameter information of the order and the first estimated delivery time into a preset adjustment model to obtain a first weight value corresponding to each candidate adjustment information; and then determine the target adjustment information according to the first weight value corresponding to each candidate adjustment information; and adjust the first estimated delivery time according to the target adjustment information to obtain the estimated delivery time displayed to the user.

[0131] The embodiment of the present application obtains order parameter information and a first estimated delivery time of a target order; inputs the order parameter information and the first estimated delivery time into a preset adjustment model to obtain a first weight value corresponding to each preset candidate adjustment information; determines the target adjustment information based on the first weight value and the candidate adjustment information; and implements a reinforcement learning model based on the characteristics of the order parameter information of the target order to predict the target adjustment information corresponding to the first estimated delivery time of the target order. Since the preset adjustment model makes full use of the relevant order parameter information of the target order in the prediction process, and combines the characteristics of the order parameter information of the target order to adjust the first estimated delivery time, the accuracy of the delivery time prediction is improved.

[0132] In another embodiment of the present application, before obtaining the order parameter information of the target order and the first estimated delivery time, the method includes:

[0133] Obtain a training sample set; wherein the training sample set includes: sample order parameter information, sample estimated delivery time, and sample actual delivery time of the sample order;

[0134] For each of the sample orders, the sample order parameter information and the sample estimated delivery time are input into the initial training model to obtain a prediction weight value for each of the candidate adjustment information;

[0135] Determining a first delivery accuracy rate of the sample order after performing the candidate adjustment operation according to the predicted weight value and the actual delivery time of the sample;

[0136] Determining a training loss value according to the first delivery accuracy rate;

[0137] Based on the sample order information and the training loss value corresponding to the sample order, the neural network model is repeatedly trained until the preset adjustment model that meets the training end conditions is obtained.

[0138] Before obtaining the order parameter information of the target order and the first estimated delivery time, the embodiment of the present application also includes a training process of a preset adjustment model.

[0139] Specifically, the training sample set may include multiple sample orders, and the sample orders may be historical delivery orders. During the training process, the preset adjustment model of the present application is obtained by learning the actual delivery conditions of the historical delivery orders.

[0140] During the training process, for each sample order, the sample order parameter information and the sample estimated delivery time are input into the initial training model to obtain the prediction weight value of each candidate adjustment information. Then, based on the prediction weight value and the sample actual delivery time, the first delivery accuracy of the sample order after executing the candidate adjustment operation is determined.

[0141] In the embodiment of the present application, on-time delivery of an order can be understood as the difference between the actual delivery time of the sample order and the estimated delivery time displayed to the user is within a preset time range. For the convenience of description, T is used. s Indicates the actual delivery time; T u Indicates the estimated delivery time displayed to the user; t c represents the difference between the actual delivery time and the estimated delivery time displayed to the user; then t c =T s -T u When t c If the sample order is within the preset time range, it is considered that the sample order is delivered on time. c When it is within the range of [-20min, 0min], the order can be considered to be delivered on time.

[0142] Among them, the estimated delivery time T displayed to the user u To perform a first adjustment operation on the first estimated delivery time, that is, the delivery time after the time-addition operation or time-reduction operation, t0 is used to represent the first estimated delivery time, and Δt is used to represent the operation time corresponding to the time-addition or time-reduction operation. Then, T u It can be expressed as T u =t0+Δt.

[0143] It can be understood that for each candidate adjustment information, if the delivery time of the sample order is within the time range of on-time delivery under the adjustment operation of the candidate adjustment information, that is, when t cIf the delivery time of the sample order is not within the preset time range, then the first delivery accuracy of the sample order that performs the candidate adjustment operation is the first weight value corresponding to the candidate adjustment information. If the delivery time of the sample order is not within the time range for on-time delivery under the adjustment operation of the candidate adjustment information, then the first delivery accuracy of the sample order that performs the candidate adjustment operation is 0.

[0144] For example, if the candidate adjustment information is an overtime of 2 minutes, and the first weight value corresponding to the candidate adjustment information is 0.3; if the sample order can be delivered on time under the adjustment operation of 2 minutes of overtime, then the first delivery accuracy corresponding to the candidate adjustment information is 30%; if the sample order cannot be delivered on time under the adjustment operation of 2 minutes of overtime, then the first delivery accuracy corresponding to the candidate adjustment information is 0.

[0145] Furthermore, after determining the first delivery accuracy, a training loss value may be determined based on the first delivery accuracy.

[0146] In another embodiment of the present application, determining the training loss value according to the first delivery accuracy rate includes:

[0147] For each of the sample orders, an accuracy improvement value is determined according to the first delivery accuracy of the sample order and the actual delivery accuracy of the sample order, wherein the actual delivery accuracy of the sample order is determined according to the sample actual delivery time of the sample order;

[0148] The training loss value is determined based on the accuracy improvement value and the correction value, wherein the correction value is determined based on the relationship between the predicted adjustment information and the first preset adjustment information, and the predicted adjustment information is determined based on each of the candidate adjustment information and its corresponding predicted weight value.

[0149] Specifically, in the actual delivery process, if the sample order is delivered on time, the actual delivery accuracy of the sample order is 1; if the sample order is not delivered on time, the actual delivery accuracy of the sample order is 0. When the difference between the actual delivery time of the sample order and the estimated delivery time displayed to the user is within the preset time range, the sample order is deemed to be delivered on time. The estimated delivery time displayed to the user for the sample order can be determined based on the first estimated delivery time of the sample order and the preset adjustment operation time, that is, for the sample order, the estimated delivery time T displayed to the user is u =t0+Δt; wherein Δt is preset.

[0150] Furthermore, the accuracy improvement value is determined according to the first delivery accuracy of the sample order and the actual delivery accuracy of the sample order. The first delivery accuracy is represented by r(A); the actual delivery accuracy is represented by b(s); then, the accuracy improvement value is r(A)-b(s).

[0151] In addition, in an embodiment of the present application, when determining the loss value, a correction value can also be determined based on the relationship between the predicted adjustment information and the first preset adjustment information; and the training loss value is determined based on the accuracy improvement value and the correction value.

[0152] The first preset adjustment information indicates a time range of a preset operation duration of an overtime operation or a subtraction operation. For example, the first preset adjustment information may be [6min, 10min].

[0153] The predicted adjustment information is determined based on each of the candidate adjustment information and its corresponding prediction weight value. Specifically, the predicted adjustment information can be the candidate adjustment information with the largest corresponding prediction weight value; it can also be the adjustment information determined after weighted processing of the candidate adjustment information according to the prediction weight value.

[0154] It can be understood that the correction value is determined based on whether the operation duration of the overtime operation or the time reduction operation of the predicted adjustment information is within the time range of the preset operation duration of the overtime operation or the time reduction operation (the first preset adjustment information); that is, when the operation duration of the predicted adjustment information is within the time range of the operation duration of the first preset adjustment information, the correction value is 0, that is, no correction is required; when the operation duration of the predicted adjustment information is not within the time range of the operation duration of the first preset adjustment information, the correction value is required for correction. The determination of the correction value can refer to the specific data relationship embodiment below.

[0155] In another embodiment of the present application, the data relationship corresponding to the training loss value is as follows:

[0156]

[0157] Wherein, L(θ|s) represents the training loss value, represents the gradient value of x; E(x) represents the expected value of x;

[0158] S represents the sample order, A represents the set of candidate adjustment information,

[0159] pθ(A|s) represents the probability of S executing each candidate adjustment information in A;

[0160] r(A) represents the first delivery accuracy rate;

[0161] b(s) represents the actual delivery accuracy;

[0162] λ*l(A) represents the correction value, λ represents the preset adjustment intensity coefficient,

[0163] l(A) represents the correction term, l(A) = max((t(A) - t u ),0)+max(-(t(A)-t l ), 0), t(A) represents the predicted adjustment time of the predicted adjustment information, [t l ,t u ] represents the preset adjustment time interval of the first preset adjustment information.

[0164] In the embodiment of the present application, during the training process, due to Characterizes the improvement of delivery accuracy, so it can maximize or minimize To achieve better and better prediction results of the preset adjustment model.

[0165] Furthermore, the initial training model is repeatedly trained. If the preset training end condition is met, the training is terminated and the model at the end of the training is used as the preset adjustment model. If the training end condition is not met, the model parameters of the initial training model are adjusted, and the adjusted model is continued to be trained based on each of the training samples.

[0166] In another embodiment of the present application, the method further includes:

[0167] Determining a second estimated delivery time according to the first estimated delivery time and the target adjustment information;

[0168] The second estimated delivery time is sent to the client, and the client is instructed to display the second estimated delivery time on a user interface.

[0169] Specifically, the second estimated delivery time indicates the estimated delivery time displayed to the user after the first estimated delivery time is adjusted by the target adjustment information. The estimated delivery time displayed to the user is represented by T u Indicates that t0 represents the first estimated delivery time, and Δt represents the operation time corresponding to the overtime or overtime operation. Then, T u It can be expressed as T u =t0+Δt.

[0170] Furthermore, the second estimated delivery time may be sent to the client, instructing the client to display the second estimated delivery time on the user interface. The client may be the client of the delivery person, i.e., the delivery end; or the client of the user, i.e., the user end.

[0171] In another embodiment of the present application, determining the target adjustment information based on the first weight value and the candidate adjustment information includes at least one of the following:

[0172] Determine the candidate adjustment information having the largest corresponding first weight value as the target adjustment information;

[0173] According to each of the candidate adjustment information and its respectively corresponding first weight value, weighted processing is performed on the operation duration of each of the first adjustment operations to obtain the target adjustment information.

[0174] Specifically, when determining the target adjustment information for the first estimated delivery time, the candidate adjustment information with the largest first weight value may be determined as the target adjustment information. Alternatively, based on the first weight value corresponding to each candidate adjustment information, one of the candidate adjustment information may be extracted from the candidate adjustment information as the target adjustment information. It is understood that the probability of each candidate adjustment information being extracted is the first weight value corresponding to the candidate adjustment information.

[0175] In addition, the operation duration of each of the first adjustment operations may be weighted according to each of the candidate adjustment information and its corresponding first weight value, and the target adjustment information may be determined according to the weighted operation duration.

[0176] The embodiment of the present application obtains order parameter information and a first estimated delivery time of a target order; inputs the order parameter information and the first estimated delivery time into a preset adjustment model to obtain a first weight value corresponding to each preset candidate adjustment information; determines the target adjustment information based on the first weight value and the candidate adjustment information; and implements a reinforcement learning model based on the characteristics of the order parameter information of the target order to predict the target adjustment information corresponding to the first estimated delivery time of the target order. Since the preset adjustment model makes full use of the relevant order parameter information of the target order in the prediction process, and combines the characteristics of the order parameter information of the target order to adjust the first estimated delivery time, the accuracy of the delivery time prediction is improved.

[0177] The present application embodiment provides a possible implementation method, which can be executed by any electronic device. Optionally, any electronic device can be a terminal device with the ability to adjust the delivery time, or a device or chip integrated in these devices. Figure 4 As shown, it is one of the flow charts of a delivery time adjustment method provided in an embodiment of the present application, and the method includes the following steps:

[0178] Step S401: receiving a second estimated delivery time sent by a server, wherein the second estimated delivery time is determined based on the first estimated delivery time and target adjustment information, wherein the target adjustment information is determined based on each candidate adjustment information and its corresponding first weight value, and the first weight value corresponding to each candidate adjustment information is determined based on order parameter information and the first estimated delivery time through a preset adjustment model;

[0179] The first weight value indicates the probability corresponding to each of the candidate adjustment information;

[0180] The candidate adjustment information includes a first adjustment operation and an operation duration of the first adjustment operation; the first adjustment operation includes an increase or decrease operation for the first estimated delivery time; the preset adjustment model is obtained by machine learning of a training sample set;

[0181] Specifically, the order parameter information represents the relevant parameter information involved in the target order. The order parameter information may specifically include user identification information (Identity document, ID), information on the items or meals ordered, the amount paid for the order, the number of blocked orders, the estimated preparation time for the order, the historical average preparation time for the order, the order delivery distance, the historical average delivery time for the order, and other information.

[0182] Information such as user ID, ordered items or meal information, order payment amount, order delivery distance, etc. in the order parameter information can be obtained by querying relevant information of the target order after the user triggers the order operation; information such as the number of blocked orders, historical average preparation time for orders, historical average delivery time for orders, etc. in the order parameter information can be obtained by querying the database; the estimated preparation time for the order in the order parameter information can be estimated and determined based on the quantity of items ordered by the user and the time to prepare each item.

[0183] The first estimated delivery time represents the estimated delivery time of the target order, wherein the first estimated delivery time can be estimated based on the average delivery time of the order history. In addition, it can also be predicted by a learning model for predicting the time based on order parameter information such as the number of blocked orders, the estimated order preparation time, the average order preparation time in history, the order delivery distance, and the average order delivery time in history; wherein the learning model for predicting the time can be pre-trained for the actual delivery time of the historical orders learned in actual training.

[0184] In the embodiment of the present application, the preset adjustment model is a model obtained by pre-training. Optionally, the preset adjustment model can be obtained by training based on the idea of ​​reinforcement learning. Specifically, in the embodiment of the present application, the Action and Reward of reinforcement learning are specifically designed as follows:

[0185] Action: candidate adjustment information for the target order, the candidate adjustment information includes a first adjustment operation and the operation duration of the first adjustment operation; the first adjustment operation includes an overtime operation or a time reduction operation for the first estimated delivery time; for example, the candidate adjustment information may include, for example, 1 minute overtime, 2 minutes overtime, 1 minute overtime, 2 minutes overtime, etc. The candidate adjustment information may be pre-set according to the time requirement for the actual delivery time. For example, in the actual order delivery process, the first estimated delivery time is 20 minutes, and the time requirement for the actual delivery time is not more than 30 minutes. Then, the operation duration of the overtime operation of the candidate adjustment information is less than 10 minutes. In addition, the operation duration in the candidate adjustment information may be directly set, for example, the operation duration of the overtime operation may be set to 5 minutes to 15 minutes, etc.

[0186] Reward: The delivery accuracy of the target order after the candidate adjustment operation is performed on the target order based on each candidate adjustment information.

[0187] The embodiment of the present application combines reinforcement learning training, and uses the adjustment operation (addition or subtraction operation) of the first estimated delivery time as the behavior action in reinforcement learning. The delivery accuracy of the target order under the adjustment of each behavior action is used as the reward Reward of reinforcement learning. By maximizing the reward Reward, the behavior Action corresponding to the target order, that is, the operation of adding or subtracting the first estimated delivery time, and the corresponding operation duration are determined.

[0188] During the actual prediction process, the preset adjustment model can predict the first weight value corresponding to each candidate adjustment information, that is, the probability value corresponding to each candidate adjustment information, based on the order parameter information and the first estimated delivery time. Then, when determining the target adjustment information corresponding to the first estimated delivery time, it can be determined based on each candidate adjustment information and its corresponding first weight value.

[0189] Specifically, when determining the target adjustment information for the first estimated delivery time, the candidate adjustment information with the largest first weight value may be determined as the target adjustment information. Alternatively, based on the first weight value corresponding to each candidate adjustment information, one of the candidate adjustment information may be extracted from the candidate adjustment information as the target adjustment information. It is understood that the probability of each candidate adjustment information being extracted is the first weight value corresponding to the candidate adjustment information.

[0190] In addition, the operation duration of each of the first adjustment operations may be weighted according to each of the candidate adjustment information and its corresponding first weight value, and the target adjustment information may be determined according to the weighted operation duration.

[0191] The second estimated delivery time represents the estimated delivery time displayed to the user after the target adjustment information is adjusted for the first estimated delivery time. The estimated delivery time displayed to the user is represented by T u Indicates that t0 represents the first estimated delivery time, and Δt represents the operation time corresponding to the overtime or overtime operation. Then, T u It can be expressed as T u =t0+Δt.

[0192] Step S402: Displaying the second estimated delivery time on the user interface.

[0193] The user interface may be a user interface of a client of the delivery person, or a user interface of a client of the user who places the order.

[0194] The embodiment of the present application obtains order parameter information and a first estimated delivery time of a target order; inputs the order parameter information and the first estimated delivery time into a preset adjustment model to obtain a first weight value corresponding to each preset candidate adjustment information; determines the target adjustment information based on the first weight value and the candidate adjustment information; and implements a reinforcement learning model based on the characteristics of the order parameter information of the target order to predict the target adjustment information corresponding to the first estimated delivery time of the target order. Since the preset adjustment model makes full use of the relevant order parameter information of the target order in the prediction process, and combines the characteristics of the order parameter information of the target order to adjust the first estimated delivery time, the accuracy of the delivery time prediction is improved.

[0195] The present application embodiment provides a delivery time adjustment device, such as Figure 5 As shown, the delivery time adjustment device 50 may include: an acquisition module 501, a prediction module 502 and a determination module 503, wherein:

[0196] An acquisition module 501 is used to acquire order parameter information and a first estimated delivery time of a target order;

[0197] Prediction module 502, used to input the order parameter information and the first estimated delivery time into a preset adjustment model to obtain a first weight value corresponding to each preset candidate adjustment information; the first weight value indicates the probability corresponding to each candidate adjustment information;

[0198] The candidate adjustment information includes a first adjustment operation and an operation duration of the first adjustment operation; the first adjustment operation includes an operation of adding time or reducing time for the first estimated delivery time; the preset adjustment model is obtained by performing reinforcement learning on a training sample set;

[0199] The determination module 503 is configured to determine target adjustment information based on the first weight value and the candidate adjustment information.

[0200] In another embodiment of the present application, the device further includes a training module, which is used to obtain a training sample set before obtaining the order parameter information of the target order and the first estimated delivery time; wherein the training sample set includes: sample order parameter information, sample estimated delivery time, and sample actual delivery time of the sample order;

[0201] For each of the sample orders, the sample order parameter information and the sample estimated delivery time are input into the initial training model to obtain a prediction weight value for each of the candidate adjustment information;

[0202] Determining a first delivery accuracy rate of the sample order after performing the candidate adjustment operation according to the predicted weight value and the actual delivery time of the sample;

[0203] Determining a training loss value according to the first delivery accuracy rate;

[0204] Based on the sample order information and the training loss value corresponding to the sample order, the neural network model is repeatedly trained until the preset adjustment model that meets the training end conditions is obtained.

[0205] In another embodiment of the present application, a training module is used to determine, for each of the sample orders, an accuracy improvement value according to the first delivery accuracy of the sample order and the actual delivery accuracy of the sample order, wherein the actual delivery accuracy of the sample order is determined according to the sample actual delivery time of the sample order;

[0206] The training loss value is determined based on the accuracy improvement value and the correction value, wherein the correction value is determined based on the relationship between the predicted adjustment information and the first preset adjustment information, and the predicted adjustment information is determined based on each of the candidate adjustment information and its corresponding predicted weight value.

[0207] In another embodiment of the present application, the data relationship corresponding to the training loss value is as follows:

[0208]

[0209] Wherein, L(θ|s) represents the training loss value, represents the gradient value of x; E(x) represents the expected value of x;

[0210] S represents the sample order, A represents the set of candidate adjustment information,

[0211] pθ(A|s) represents the probability of S executing each candidate adjustment information in A;

[0212] r(A) represents the first delivery accuracy rate;

[0213] b(s) represents the actual delivery accuracy;

[0214] λ*l(A) represents the correction value, λ represents the preset adjustment intensity coefficient,

[0215] l(A) represents the correction term, l(A) = max((t(A) - t u ),0)+max(-(t(A)-t l ), 0), t(A) represents the predicted adjustment time of the predicted adjustment information, [t l ,t u ] represents the preset adjustment time interval of the first preset adjustment information.

[0216] In another embodiment of the present application, the order parameter information includes at least one of the following:

[0217] Number of blocking orders;

[0218] Estimated lead time for orders;

[0219] Average order preparation time in history;

[0220] Order delivery distance;

[0221] Average delivery time of orders in history.

[0222] In another embodiment of the present application, the device further comprises:

[0223] a sending module, configured to determine a second estimated delivery time according to the first estimated delivery time and the target adjustment information;

[0224] The second estimated delivery time is sent to the client, and the client is instructed to display the second estimated delivery time on a user interface.

[0225] In another embodiment of the present application, the determination module is used to determine the candidate adjustment information with the largest corresponding first weight value as the target adjustment information;

[0226] or

[0227] According to each of the candidate adjustment information and its respectively corresponding first weight value, weighted processing is performed on the operation duration of each of the first adjustment operations to obtain the target adjustment information.

[0228] The present application embodiment provides a delivery time adjustment device, such as Figure 6As shown, the delivery time adjustment device 60 may include a receiving module 601 and a display module 602, wherein:

[0229] A receiving module 601 is used to receive a second estimated delivery time sent by a server, where the second estimated delivery time is determined based on the first estimated delivery time and target adjustment information, where the target adjustment information is determined based on each candidate adjustment information and its corresponding first weight value, and the first weight value corresponding to each candidate adjustment information is determined based on order parameter information and the first estimated delivery time through a preset adjustment model;

[0230] The first weight value indicates the probability corresponding to each of the candidate adjustment information;

[0231] The candidate adjustment information includes a first adjustment operation and an operation duration of the first adjustment operation; the first adjustment operation includes an increase or decrease operation for the first estimated delivery time; the preset adjustment model is obtained by machine learning of a training sample set;

[0232] The display module 602 is used to display the second estimated delivery time on the user interface.

[0233] The delivery time adjustment device of this embodiment can execute the delivery time adjustment method shown in the above embodiments of this application. The implementation principle is similar and will not be repeated here.

[0234] The embodiment of the present application obtains order parameter information and a first estimated delivery time of a target order; inputs the order parameter information and the first estimated delivery time into a preset adjustment model to obtain a first weight value corresponding to each preset candidate adjustment information; determines the target adjustment information based on the first weight value and the candidate adjustment information; and implements a reinforcement learning model based on the characteristics of the order parameter information of the target order to predict the target adjustment information corresponding to the first estimated delivery time of the target order. Since the preset adjustment model makes full use of the relevant order parameter information of the target order in the prediction process, and combines the characteristics of the order parameter information of the target order to adjust the first estimated delivery time, the accuracy of the delivery time prediction is improved.

[0235] An electronic device is provided in an embodiment of the present application, and the electronic device includes: a memory and a processor; at least one program, stored in the memory, for being executed by the processor, and compared with the prior art, it can achieve: the embodiment of the present application obtains order parameter information of the target order and a first estimated delivery time; inputs the order parameter information and the first estimated delivery time into a preset adjustment model to obtain a first weight value corresponding to each preset candidate adjustment information; determines the target adjustment information based on the first weight value and the candidate adjustment information; and implements a reinforcement learning model based on the characteristics of the order parameter information of the target order to predict the target adjustment information corresponding to the first estimated delivery time of the target order. Since the preset adjustment model makes full use of the relevant order parameter information of the target order in the prediction process, and adjusts the first estimated delivery time in combination with the characteristics of the order parameter information of the target order, the accuracy of the delivery time prediction is improved.

[0236] In an alternative embodiment, an electronic device is provided, such as Figure 7 As shown, Figure 7 The electronic device 4000 shown includes: a processor 4001 and a memory 4003. The processor 4001 and the memory 4003 are connected, such as through a bus 4002. Optionally, the electronic device 4000 may also include a transceiver 4004, which may be used for data interaction between the electronic device and other electronic devices, such as data transmission and / or data reception. It should be noted that in actual applications, the transceiver 4004 is not limited to one, and the structure of the electronic device 4000 does not constitute a limitation on the embodiments of the present application.

[0237] Processor 4001 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. It may implement or execute various exemplary logic blocks, modules and circuits described in conjunction with the disclosure of this application. Processor 4001 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0238] The bus 4002 may include a path to transmit information between the above components. The bus 4002 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. The bus 4002 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 7 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0239] The memory 4003 may be a ROM (Read Only Memory) or other types of static storage devices that can store static information and instructions, a RAM (Random Access Memory) or other types of dynamic storage devices that can store information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, optical disk storage (including compressed optical disk, laser disk, optical disk, digital versatile disk, Blu-ray disk, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited to these.

[0240] The memory 4003 is used to store the application code (computer program) for executing the solution of the present application, and the execution is controlled by the processor 4001. The processor 4001 is used to execute the application code stored in the memory 4003 to implement the contents shown in the above method embodiment.

[0241] The electronic devices include but are not limited to mobile phones, laptop computers, multimedia players, desktop computers, etc.

[0242] An embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored. When the computer-readable storage medium is run on a computer, the computer can execute the corresponding content in the aforementioned method embodiment.

[0243] The embodiment of the present application obtains order parameter information and a first estimated delivery time of a target order; inputs the order parameter information and the first estimated delivery time into a preset adjustment model to obtain a first weight value corresponding to each preset candidate adjustment information; determines the target adjustment information based on the first weight value and the candidate adjustment information; and implements a reinforcement learning model based on the characteristics of the order parameter information of the target order to predict the target adjustment information corresponding to the first estimated delivery time of the target order. Since the preset adjustment model makes full use of the relevant order parameter information of the target order in the prediction process, and combines the characteristics of the order parameter information of the target order to adjust the first estimated delivery time, the accuracy of the delivery time prediction is improved.

[0244] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0245] The above descriptions are only some embodiments of the present invention. It should be pointed out that, for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A delivery time adjustment method, characterized in that: include: Obtain order parameter information and the first estimated delivery time of the target order; Inputting the order parameter information and the first estimated delivery time into a preset adjustment model to obtain a first weight value corresponding to each preset candidate adjustment information; the first weight value indicates a probability corresponding to each candidate adjustment information; The candidate adjustment information includes a first adjustment operation and an operation duration of the first adjustment operation; the first adjustment operation includes an operation of adding time or reducing time for the first estimated delivery time; the preset adjustment model is obtained by performing reinforcement learning on a training sample set; Determining target adjustment information based on the first weight value and the candidate adjustment information; The preset adjustment model is: based on the sample order information of the sample orders in the training sample set and the training loss value corresponding to the sample orders, the initial training model is repeatedly trained to obtain a model that meets the training end condition; wherein the data relationship corresponding to the training loss value is as follows: Wherein, L(θ|s) represents the training loss value, represents the gradient value of x; E(x) represents the expected value of x; S represents the sample order, A represents the set of candidate adjustment information, pθ(A|s) represents the probability of S executing each candidate adjustment information in A; r(A) represents the first delivery accuracy of the sample order after executing the candidate adjustment operation, determined based on the prediction weight value of each candidate adjustment information and the actual delivery time of the sample order; the prediction weight value is predicted based on the initial training model; b(s) represents the actual delivery accuracy of the sample order; the actual delivery accuracy of the sample order is determined according to the sample actual delivery time of the sample order; λ*l(A) represents the correction value, λ represents the preset adjustment intensity coefficient, l(A) represents the correction term, l(A) = max((t(A) - t u ),0)+max(-(t(A)-t l ),0), t(A) represents the prediction adjustment time of the prediction adjustment information, [t l ,t u ] represents the preset adjustment time interval of the first preset adjustment information.

2. The delivery time adjustment method according to claim 1, characterized in that: Before obtaining the order parameter information of the target order and the first estimated delivery time, the method includes: Acquire a training sample set; wherein the training sample set includes sample order information of sample orders, and the sample order information includes: at least one of sample order parameter information, sample estimated delivery time, and sample actual delivery time; For each of the sample orders, the sample order parameter information and the sample estimated delivery time are input into the initial training model to obtain a prediction weight value for each of the candidate adjustment information; Determining a first delivery accuracy rate of the sample order after performing the candidate adjustment operation according to the predicted weight value and the actual delivery time of the sample; Determining a training loss value according to the first delivery accuracy rate; Based on the sample order information and the training loss value corresponding to the sample order, the neural network model corresponding to the initial training model is repeatedly trained until the preset adjustment model that meets the training end conditions is obtained.

3. The delivery time adjustment method according to claim 2, characterized in that: The determining of the training loss value according to the first delivery accuracy rate includes: For each of the sample orders, determining an accuracy improvement value according to the first delivery accuracy of the sample order and the actual delivery accuracy of the sample order; The training loss value is determined based on the accuracy improvement value and the correction value, wherein the correction value is determined based on the relationship between the predicted adjustment information and the first preset adjustment information, and the predicted adjustment information is determined based on each of the candidate adjustment information and its corresponding predicted weight value.

4. The delivery time adjustment method according to claim 1, characterized in that: The order parameter information includes at least one of the following: Number of blocking orders; Estimated lead time for orders; Average order preparation time in history; Order delivery distance; Average delivery time of orders in history.

5. The delivery time adjustment method according to claim 1, characterized in that: The method further comprises: Determining a second estimated delivery time according to the first estimated delivery time and the target adjustment information; The second estimated delivery time is sent to the client, and the client is instructed to display the second estimated delivery time on a user interface.

6. The delivery time adjustment method according to claim 1, characterized in that: The determining target adjustment information based on the first weight value and the candidate adjustment information includes at least one of the following: Determine the candidate adjustment information having the largest corresponding first weight value as the target adjustment information; According to each of the candidate adjustment information and its respectively corresponding first weight value, weighted processing is performed on the operation duration of each of the first adjustment operations to obtain the target adjustment information.

7. A delivery time adjustment method, characterized in that: include: receiving a second estimated delivery time sent by the server, where the second estimated delivery time is determined based on the first estimated delivery time and target adjustment information, where the target adjustment information is determined based on each candidate adjustment information and its corresponding first weight value, where the first weight value corresponding to each candidate adjustment information is determined based on the order parameter information and the first estimated delivery time through a preset adjustment model; The first weight value indicates the probability corresponding to each of the candidate adjustment information; The candidate adjustment information includes a first adjustment operation and an operation duration of the first adjustment operation; the first adjustment operation includes an increase or decrease operation for the first estimated delivery time; the preset adjustment model is obtained by machine learning of a training sample set; Displaying the second estimated delivery time on the user interface; The preset adjustment model is: based on the sample order information of the sample orders in the training sample set and the training loss value corresponding to the sample orders, the initial training model is repeatedly trained to obtain a model that meets the training end condition; wherein the data relationship corresponding to the training loss value is as follows: Wherein, L(θ|s) represents the training loss value, represents the gradient value of x; E(x) represents the expected value of x; S represents the sample order, A represents the set of candidate adjustment information, pθ(A|s) represents the probability of S executing each candidate adjustment information in A; r(A) represents the first delivery accuracy of the sample order after executing the candidate adjustment operation, determined based on the prediction weight value of each candidate adjustment information and the actual delivery time of the sample order; the prediction weight value is predicted based on the initial training model; b(s) represents the actual delivery accuracy of the sample order; the actual delivery accuracy of the sample order is determined according to the sample actual delivery time of the sample order; λ*l(A) represents the correction value, λ represents the preset adjustment intensity coefficient, l(A) represents the correction term, l(A) = max((t(A) - t u ),0)+max(-(t(A)-t l ),0), t(A) represents the prediction adjustment time of the prediction adjustment information, [t l ,t u ] represents the preset adjustment time interval of the first preset adjustment information.

8. A delivery time adjustment device, characterized in that: include: An acquisition module, used to acquire order parameter information and a first estimated delivery time of a target order; A prediction module, configured to input the order parameter information and the first estimated delivery time into a preset adjustment model to obtain a first weight value corresponding to each preset candidate adjustment information; the first weight value indicates a probability corresponding to each candidate adjustment information; The candidate adjustment information includes a first adjustment operation and an operation duration of the first adjustment operation; the first adjustment operation includes an operation of adding time or reducing time for the first estimated delivery time; the preset adjustment model is obtained by performing reinforcement learning on a training sample set; a determination module, configured to determine target adjustment information based on the first weight value and the candidate adjustment information; The preset adjustment model is: based on the sample order information of the sample orders in the training sample set and the training loss value corresponding to the sample orders, the initial training model is repeatedly trained to obtain a model that meets the training end condition; wherein the data relationship corresponding to the training loss value is as follows: Wherein, L(θ|s) represents the training loss value, represents the gradient value of x; E(x) represents the expected value of x; S represents the sample order, A represents the set of candidate adjustment information, pθ(A|s) represents the probability of S executing each candidate adjustment information in A; r(A) represents the first delivery accuracy of the sample order after executing the candidate adjustment operation, determined based on the prediction weight value of each candidate adjustment information and the actual delivery time of the sample order; the prediction weight value is predicted based on the initial training model; b(s) represents the actual delivery accuracy of the sample order; the actual delivery accuracy of the sample order is determined according to the sample actual delivery time of the sample order; λ*l(A) represents the correction value, λ represents the preset adjustment intensity coefficient, l(A) represents the correction term, l(A) = max((t(A) - t u ),0)+max(-(t(A)-t l ),0), t(A) represents the prediction adjustment time of the prediction adjustment information, [t l ,t u ] represents the preset adjustment time interval of the first preset adjustment information.

9. A delivery time adjustment device, characterized in that: include: a receiving module, configured to receive a second estimated delivery time sent by a server, wherein the second estimated delivery time is determined according to the first estimated delivery time and target adjustment information, wherein the target adjustment information is determined based on each candidate adjustment information and its corresponding first weight value, and the first weight value corresponding to each candidate adjustment information is determined by a preset adjustment model based on order parameter information and the first estimated delivery time; The first weight value indicates the probability corresponding to each of the candidate adjustment information; The candidate adjustment information includes a first adjustment operation and an operation duration of the first adjustment operation; the first adjustment operation includes an increase or decrease operation for the first estimated delivery time; the preset adjustment model is obtained by machine learning of a training sample set; A display module, configured to display the second estimated delivery time on a user interface; The preset adjustment model is: based on the sample order information of the sample orders in the training sample set and the training loss value corresponding to the sample orders, the initial training model is repeatedly trained to obtain a model that meets the training end condition; wherein the data relationship corresponding to the training loss value is as follows: Wherein, L(θ|s) represents the training loss value, represents the gradient value of x; E(x) represents the expected value of x; S represents the sample order, A represents the set of candidate adjustment information, pθ(A|s) represents the probability of S executing each candidate adjustment information in A; r(A) represents the first delivery accuracy of the sample order after executing the candidate adjustment operation, determined based on the prediction weight value of each candidate adjustment information and the actual delivery time of the sample order; the prediction weight value is predicted based on the initial training model; b(s) represents the actual delivery accuracy of the sample order; the actual delivery accuracy of the sample order is determined according to the sample actual delivery time of the sample order; λ*l(A) represents the correction value, λ represents the preset adjustment intensity coefficient, l(A) represents the correction term, l(A) = max((t(A) - t u ),0)+max(-(t(A)-t l ),0), t(A) represents the prediction adjustment time of the prediction adjustment information, [t l ,t u ] represents the preset adjustment time interval of the first preset adjustment information.

10. An electronic device, characterized in that: The electronic device comprises: one or more processors; Memory; One or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the one or more processors, and the one or more programs are configured to: execute the delivery time adjustment method according to any one of claims 1 to 7.

11. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the delivery time adjustment method according to any one of claims 1 to 7 is implemented.

Images