Order Processing Method, Device, Electronic Device and Readable Storage Medium

Abstract

Embodiments of the present application provide an order processing method, apparatus, electronic device, and readable storage medium. Among them, the order processing method includes: obtaining a plurality of orders to be processed; inputting the order feature information of each of the plurality of orders to be processed into a combined parameter value prediction model to obtain parameter values for performing combined operations on the plurality of orders to be processed respectively; combining the plurality of orders to be processed according to the combined parameter values corresponding to each of the plurality of orders to be processed to obtain a plurality of order packages to be processed; adding the plurality of order packages to a scheduling pool for an order scheduling system to schedule the plurality of order packages to be processed. There is no need to keep a fixed order pressing time for each order, and orders are flexibly and dynamically combined at the order granularity. By using the combined parameter value prediction model to comprehensively consider the feature information of multiple orders, the obtained combination strategy is relatively reasonable, and order combination and scheduling are performed according to the combination strategy.

Description

Technical Field

[0001] Embodiments of the present application relate to the field of information processing technologies, and in particular, to an order processing method, apparatus, electronic device, and readable storage medium. Background Art

[0002] In current multi-objective decision-making problems in fields such as logistics distribution and path planning, a greedy algorithm is mostly used for objective optimization. For example, in the field of logistics distribution, it is set that before each order to be delivered is dispatched, it needs to wait for a certain period of time (order backlog) in the order pool. During this waiting period, other orders to be delivered similar to it are queried, and the similar orders are combined into an order package, and then dispatched in the form of the order package to save operating resources. That is to say, the backlog time of each order to be delivered is fixed and unchanged.

[0003] In the related art, in order to cooperate with the greedy algorithm to determine the optimal combination strategy, the backlog time of all orders in each order pool is a relatively fixed time. For example, the current backlog time of the order pool is 5 minutes. However, with the continuous operation of the business, the fixed backlog time will be adjusted to a certain extent according to the package combination rate of the orders in the order pool. Suppose based on the analysis of the historical order data in the order pool, the package combination rate of the orders in the order pool is only 10%, indicating that the order quantity is relatively sparse. In order to reduce the waiting time of users and improve the user experience, the backlog time of the order pool should be reduced, for example, adjusted from 5 minutes to 3 minutes. However, this will result in a shorter backlog time for each order in the order pool, making it more difficult to combine packages, and the package combination rate will be lower. As a result, reducing the backlog time further will lead to a vicious cycle, and ultimately it will be very difficult to combine the orders in the order pool.

[0004] As can be seen from the above analysis, it is not appropriate to set a fixed backlog time for each order in the related art, and it needs to be improved. Summary of the Invention

[0005] Embodiments of the present application provide an order processing method, apparatus, electronic device, and readable storage medium to achieve dynamic adjustment of the order combination strategy.

[0006] In the first aspect of the embodiments of the present application, an order processing method is provided, and the method includes:

[0007] Obtain a plurality of orders to be processed;

[0008] Input the order feature information of each of the plurality of orders to be processed into a combination parameter value prediction model to obtain parameter values for performing combination operations on the plurality of orders to be processed respectively;

[0009] Combine the plurality of orders to be processed according to the combination parameter values corresponding to each of the plurality of orders to be processed to obtain a plurality of order packages to be processed;

[0010] Add the multiple to-be-processed order packages to a scheduling pool for an order scheduling system to schedule the multiple to-be-processed order packages.

[0011] A second aspect of the embodiments of the present application provides an order processing device, which includes:

[0012] A first obtaining module, configured to obtain multiple to-be-processed orders;

[0013] A first input module, configured to input the order feature information of each of the multiple to-be-processed orders into a combined parameter value prediction model to obtain parameter values for performing combined operations on the multiple to-be-processed orders respectively;

[0014] A first combining module, configured to combine the multiple to-be-processed orders according to the combined parameter values corresponding to the multiple to-be-processed orders respectively to obtain multiple to-be-processed order packages;

[0015] A first adding module, configured to add the multiple to-be-processed order packages to a scheduling pool for an order scheduling system to schedule the multiple to-be-processed order packages.

[0016] Optionally, the device further includes:

[0017] A first package hold time determination module, configured to, when the combined parameter value corresponding to a to-be-processed order at least includes the hold time of the to-be-processed order, for each to-be-processed order package among the multiple to-be-processed order packages, determine the package hold time of the to-be-processed order package according to the hold time of each to-be-processed order in the to-be-processed order package;

[0018] A second package hold time determination module, configured to, when the order information of a to-be-processed order includes a maximum hold time, for each to-be-processed order package among the multiple to-be-processed order packages, determine the package hold time of the to-be-processed order package according to the maximum hold time of each to-be-processed order in the to-be-processed order package;

[0019] Wherein, the order scheduling system is configured to schedule the multiple to-be-processed order packages according to the package hold time of each of the multiple to-be-processed order packages.

[0020] A third aspect of the embodiments of the present application provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the steps in the method described in the first aspect of the present application are implemented.

[0021] A fourth aspect of the embodiments of the present application provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, and when the processor executes, the steps of the method described in the first aspect of the present application are implemented.

[0022] By using the order processing method provided in the embodiments of the present application, the feature information of multiple orders is processed through a combined parameter value prediction model, the combined operation parameters of each order among the multiple orders are dynamically determined, the combined strategy for the multiple orders is dynamically determined according to the combined operation parameters, at least some of the orders are combined into a to-be-processed order package according to the combined strategy, and then, the scheduling system performs scheduling processing on the to-be-processed order package in the form of a package. In this way, it is not necessary to keep a fixed order holding time for each order, realizing flexible and dynamic combination of orders at the order granularity, with the order holding time being flexible and variable, avoiding the situation of the decrease in the package combination rate caused by the fixed or continuously decreasing order holding time in the related art. Moreover, by using the combined parameter value prediction model, the feature information of multiple orders can be comprehensively considered, so the obtained combined strategy is relatively reasonable. Combining and then scheduling orders according to the relatively reasonable combined strategy reduces the waiting time of users and improves the user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for the description in the embodiments of the present application will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and those of ordinary skill in the art can obtain other drawings according to these drawings without creative efforts.

[0024] Figure 1 is a flowchart of a training method for a combined parameter value prediction model used in the order processing method proposed in an embodiment of the present application;

[0025] Figure 2 is an exemplary schematic diagram showing an order delivery scenario;

[0026] Figure 3 is a flowchart of an order processing method proposed in another embodiment of the present application;

[0027] Figure 4 is a flowchart of an order processing method proposed in yet another embodiment of the present application;

[0028] Figure 5 is a schematic diagram of an order processing device proposed in an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0029] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some, but not all, of the embodiments of the present application. All other embodiments obtained by those of ordinary skill in the art without creative efforts based on the embodiments in the present application belong to the scope of protection of the present application.

[0030] It should be noted that in this application, the technical solution proposed in this application applicable to multi-objective decision optimization problems will be exemplarily described by taking the delivery order in logistics distribution as an example. However, this should not be construed as a specific limitation of this application. For example, the technical solution proposed in this application applicable to multi-objective decision optimization problems is applicable not only to the delivery of logistics orders, but also to path planning, etc. The orders in this application are not limited to the orders in logistics distribution, but can also be taxi orders for transportation, etc.

[0031] In the multi-objective decision problems in current fields such as logistics distribution and path planning, the greedy algorithm is mostly used for objective optimization. For example, in the field of logistics distribution, it is set that before each order to be delivered is delivered, it needs to wait for a certain period of time (order pressing) in the order pool. During this waiting period, other orders to be delivered similar to it are queried, and the similar orders are combined into an order package, and then scheduled in the form of the order package to save operating resources. That is to say, the order pressing time for each order to be delivered is fixed and unchanged.

[0032] However, it is not appropriate to set a fixed order pressing time for each order, and it needs to be improved. The following is an example to illustrate the unreasonableness of this method.

[0033] Table 1

[0034] Waybill Number Time of Entering the Order Pool Delivery Time 1 12:00 12:03 2 12:01 12:04 3 12:05 12:08 … … … m … …

[0035] As shown in Table 1, the order pressing time for each order in the order pool corresponding to this area is fixed at 3 minutes. Among them, Order 1, Order 2, and Order 3 are similar in nature. If the order pressing time of each order is not considered, then Order 1, Order 2, and Order 3 can be combined into an order package and scheduled in the form of the package. However, because the order pressing time for each order is fixed, when Order 3 enters the order pool, Order 1 and 2 have already left the order pool, resulting in Order 3 not being able to be combined with Order 1 and Order 2 into an order package. Moreover, if Order 3 fails to be combined into a package with other orders at the end of its order pressing time, then for Order 3, these 3 minutes of order pressing time are meaningless, which means that these 3 minutes are wasted in vain. Not only is the scheduling cost not saved, but relatively, the waiting time of the user corresponding to Order 3 is increased in vain, reducing the user experience.

[0036] Moreover, to ensure the maximization of policy benefits, as the business continues to operate, the fixed order pressing time will be adjusted to some extent according to the package combination rate of the orders in the order pool. Suppose, based on the data analysis of historical orders in the order pool, the package combination rate of the orders in the order pool is only 10%, indicating that the order quantity is relatively sparse. To reduce the waiting time of users and improve the user experience, the order pressing time of the order pool should be reduced. For example, it is adjusted from 3 minutes to 2 minutes. However, this will result in a shorter order pressing time for each order in the order pool, making it more difficult to combine the orders in the entire order pool, and the package combination rate will be lower. For example, it will be more difficult for Order 3 to be combined with other orders, resulting in an even lower package combination rate when the order pressing time is further reduced, such as reducing the order pressing time from 2 minutes to 1 minute again. Obviously, Order 1 and Order 2 cannot be combined into a package at this time, and so on in a vicious cycle, ultimately making it very difficult to combine the orders in the order pool.

[0037] Based on the above technical deficiencies, this application proposes based on the idea of machine learning: analyzing and regressing a combined parameter value prediction model based on the combined data of historical orders, and predicting the combined parameters of batch orders through this combined parameter value prediction model. The combined parameters can include the order pressing time corresponding to each order respectively, and which orders each order should be combined with, etc. And according to the combined parameters output by the prediction model, dynamically combine the orders in the order pool, so that the order pressing time of each order in the batch orders in the order pool is flexibly variable. Instead of using the technical means that all orders in each order pool adopt a fixed order pressing time to wait for combination with other orders.

[0038] Reference Figure 1 , Figure 1 is a flowchart of the training method of the combined parameter value prediction model used in the order processing method proposed in an embodiment of this application, exemplarily showing the training process of the combined parameter value prediction model. As Figure 1 shown, the method includes the following steps:

[0039] S101, obtain the order feature information of each of multiple sample orders. The order feature information of a sample order includes the order attribute information of the sample order and the association information between the sample order and other sample orders.

[0040] The sample orders are representative historical orders selected from the order pool. The sample orders are divided into two categories: positive sample orders and negative sample orders. A positive sample order refers to an order that has been scheduled in the form of a package by being combined with other orders in the sample orders in history, and a negative sample order refers to an order that has not been combined with other orders in the sample orders in history and is still scheduled in the form of a single order.

[0041] Obtain the order feature information of each sample order, which can be extracted using a feature extraction model or through annotation. Among them, the feature extraction model is a model specifically trained on the training set S composed of sample orders based on the seed model f(x) for extracting the feature information of orders.

[0042] The order attribute information of a sample order characterizes the characteristics of the order itself. For example: the delivery starting point of the order, the delivery end point, the delivery distance, the density of merchants at the starting point, the density of users at the end point, etc.

[0043] The association information between a sample order and other sample orders characterizes the correlation between the feature information of this sample order and the feature information of other sample orders. For example: the combination probability of this sample order combined with other sample orders, whether this sample order and other sample orders are in the same boundary area.

[0044] S102. Determine the respective labels corresponding to multiple sample orders. The label corresponding to a sample order characterizes whether this sample order has been combined with other sample orders into a sample order package.

[0045] Since the sample orders are representative orders selected from historical orders, including some orders that have been combined into packages and some orders that have not been combined into packages. Therefore, the sample orders can be annotated to mark the orders that have been combined into packages and the orders that have not been combined into packages, that is, assign labels to the sample orders. This label not only characterizes whether each sample order has been combined into a package but also characterizes which orders among other sample orders each sample order has been combined with.

[0046] S103. Train a preset model based on the order feature information and corresponding labels of multiple sample orders to obtain a combined parameter value prediction model.

[0047] After obtaining the order feature information and corresponding labels of each sample order through the aforementioned steps S101 and S102, in this step, input the order feature information of multiple sample orders into the preset model ωf(n) for training, and optimize the preset model based on the labels, that is, optimize the parameter ω of the model. In other words, perform supervised learning on the preset model. By inputting the feature information of each sample order into the preset model and evaluating the output result of the preset model based on the label, that is, evaluating whether the prediction result of the preset model for each sample order conforms to the combined information characterized by the label corresponding to the sample order. If it conforms, give a positive evaluation to the model; if it does not conform, give a negative evaluation to the model, so that the model can be updated according to the evaluation result, and further enable the preset model to have the generalization ability to correctly predict the combined parameters of sample orders, thereby obtaining a combined parameter value prediction model.

[0048] When the combined parameter value prediction model has the most basic generalization ability, that is, the combined operation parameters output by the combined parameter value prediction model can already be used to combine multiple sample orders in the training set S. However, at this time, the combined operation parameters output by the combined parameter value prediction model are usually not yet optimal combined operation parameters and cannot meet the requirements of business operations, nor can they be applied to actual order processing scenarios.

[0049] Exemplarily, as Figure 2 shown, there are two paths between place A and place E, passing through place B, place C, and place D. Now there are order 7: to be delivered from place A to place B, order 8: to be delivered from place A to place C, order 9: to be delivered from place B to place C, order 10: to be delivered from place B to place D, order 11: to be delivered from place A to place E, and order 12: to be delivered from place B to place E. There are two deliverymen, Jia and Yi, at place A and one deliveryman Bing at place B.

[0050] Assuming that the degree of being on the same route is taken as the main indicator for order bundling, then the delivery routes of order 7, order 10, and order 12 are relatively on the same route, the delivery routes of order 8 and order 11 are relatively on the same route, while the delivery route of order 9 is not very on the same route as any of the other orders.

[0051] In order to enable all deliverymen to deliver the orders and deliver all orders from the departure place to the destination in the shortest time, in the scenario of the above example, deliveryman Jia should follow the route A→B→E, first deliver order 7 to place B, and then deliver order 12 to place E; deliveryman should follow the route A→D→E, deliver order 8 to place D, and then deliver order 11 to place E; deliveryman Bing should follow the route B→D→C, deliver order 10 to place D, and then deliver order 9 to place C, or, deliveryman Bing should follow the route B→C→D, deliver order 9 to place C, and then deliver order 10 to place D. That is, the optimal combination method for order 7 to order 12 is: order 7 and order 12 are combined into an order package, order 8 and order 11 are combined into an order package, and order 9 and order 10 are combined into an order package.

[0052] For the scenario of the above Figure 2 example, the combined operation parameters output by the combined parameter value prediction model at this time are very likely to indicate combining order 7, order 10, and order 12, combining order 8 and order 11, while order 9 fails to be combined into a package. Obviously, it cannot meet the business requirements.

[0053] Therefore, it is necessary to consider the policy returns corresponding to different combination strategies of the sample orders in the training set S, and determine the combination strategy whose policy return meets the preset conditions (for example: the optimal policy return) from them. That is to say, it is necessary to iteratively optimize the combination parameter value prediction model so that the combination parameter value prediction model will consider the policy returns corresponding to different combination strategies when outputting the combined operation parameters, until the combination strategy indicated by the combined operation parameters output by the combination parameter value prediction model meets the preset conditions.

[0054] In an alternative embodiment, after obtaining the combination parameter value prediction model through steps S101 to S103, the combination parameter value prediction model can be iteratively optimized in the following manner. The method includes the following steps:

[0055] S201, input the order feature information of each of a preset number of sample orders into the combination parameter value prediction model, and determine the sample parameter values for performing combination operations on the preset number of sample orders respectively.

[0056] The sample orders at this time should be selected from the historical orders combined according to the combined operation parameters output after prediction by the combination parameter value prediction model to form the training set S'.

[0057] For ease of understanding, the following will use the Figure 2 scenario in the above example for illustration, that is, assume that the training set S' is composed of order 7 to order 12, but it should be noted that this should not be construed as a specific limitation of the present application. This is only an exemplary illustration. Obviously, in practice, the training samples cannot be only 5 samples, nor can they be such simplified 5 samples.

[0058] Input the order feature information of each order in the training set S' composed of sample orders into the combination parameter value prediction model that has been trained and has generalization ability. The combination parameter value prediction model will output the sample parameter values of the combined operations of each order in the training set S'. The sample parameter values indicate the sample combination strategies of each order in the training set S', or the sample combination strategies of each order sample and the corresponding sample order pressing time.

[0059] For example, after inputting the delivery path information in the feature information of the above order 7 to order 12 into the combination parameter value prediction model, the combination strategy indicated by the combined operation parameters output after the combination parameter value prediction model converges is: combine order 7, order 10, and order 12, combine order 8 and order 11, and order 9 cannot be combined. At this time, the combined operation parameters output by the combination parameter value prediction model are the sample parameter values.

[0060] S202, apply dynamic interference to the sample parameter values to determine the target parameter values corresponding to the combination parameter value prediction model.

[0061] That is, interference is imposed on the combination strategies of each order in the training set S' output by the combination parameter value prediction model to dynamically adjust the combination parameters, and a certain number of different sample combination operation parameters will be obtained. According to the different sample combination operation parameters obtained by dynamic interference, select the sample combination operation parameters that meet the conditions (for example: optimal) from these different sample combination operation parameters.

[0062] In an alternative implementation, it can be achieved through the following steps:

[0063] S2021, adjust the sample parameter values multiple times to obtain the sample parameter values after multiple adjustments.

[0064] Construct an offline simulation system with the above sample order set S', dynamically adjust the combination parameters of each order in the sample training set S', and determine multiple adjusted combination strategies.

[0065] For example, with the above Figure 2 Example scenario, construct an offline simulation system, re-combine orders 7 to 12 freely to obtain other different combination strategies, such as: combine order 7 and order 12, combine order 8 and order 11, combine order 9 and order 10; or, combine order 7, order 8, and order 10, combine order 9 and order 12, order 11 cannot be combined, etc.

[0066] S2022, determine the order processing performance values corresponding to the sample parameter values and the sample parameter values after each adjustment respectively.

[0067] The order processing performance value represents the strategy benefits of different combination strategies corresponding to the order, such as the waiting time of the user, the delivery cost of the delivery person, the degree of convenience of the delivery routes of the orders in the order package, the operation and scheduling cost, etc.

[0068] Through the offline simulation system, simulate and calculate the combination strategies before the above adjustments and multiple adjusted combination strategies to determine the order processing performance values corresponding to each combination strategy, that is, the strategy benefits corresponding to each combination strategy. Specifically, the particle swarm algorithm can be used to perform the simulation calculations corresponding to different combination strategies (perform distributed calculations using the particle swarm search algorithm) to obtain the strategy benefits corresponding to different combination strategies.

[0069] For example, arrange the different combination strategies of the above orders 7 to 12 (including the combination strategies output by the combination parameter value prediction model and the different combination strategies obtained after imposing interference) according to Figure 2Simulate the delivery of the schematic scenarios and determine the benefits of the delivery methods corresponding to each combined strategy, such as the weighted average of the waiting times of each user, the weighted average of the delivery costs of the delivery staff, the weighted average of the degree of route compliance of the delivery routes of the orders in the order package, the weighted average of the operation scheduling costs, and so on.

[0070] In S2023, among the sample parameter values and the sample parameter values after multiple adjustments, the sample parameter values corresponding to the order processing performance values that meet the preset threshold are determined as the target parameter values corresponding to the combined parameter value prediction model.

[0071] Determine the combined strategies with strategy benefits greater than the preset threshold from all strategy benefits, that is, determine the orders that meet the preset threshold from all the order processing performance values, and then determine the combined parameters (i.e., sample parameter values) of the sample order as the target parameter values output by the combined parameter value prediction model.

[0072] For example, for the above orders 7 to 12, it is finally determined that the combined strategy of combining order 7 and order 12, combining order 8 and order 11, and combining order 9 and order 10 has the greatest strategy benefit, that is, the order processing performance values of orders 7 to 12 are the best. That is, the combined strategy of combining order 7 and order 12, combining order 8 and order 11, and combining order 9 and order 10 is the optimal combined strategy. Then, the combined operation parameters corresponding to the combined strategy of combining order 7 and order 12, combining order 8 and order 11, and combining order 9 and order 10 are determined as the target parameter values that the combined parameter value prediction model should output.

[0073] Reference Figure 3 , Figure 3 is a flowchart of an order processing method proposed in an embodiment of the present application, exemplarily showing the process of predicting the combined parameters of an order using a combined parameter value prediction model. As Figure 3 shown, the method includes the following steps:

[0074] S301, Obtain a plurality of orders to be processed.

[0075] The orders to be processed are the orders that enter the order pool and wait to be combined.

[0076] S302, Input the order feature information of each of the plurality of orders to be processed into the combined parameter value prediction model to obtain the parameter values for performing combined operations on the plurality of orders to be processed respectively.

[0077] Among them, the order feature information of the order to be processed is the same as the feature information of the sample order, and the extraction method of the feature information is also the same or similar. For specific details, please refer to the description of step S101 above and will not be elaborated here.

[0078] After inputting the feature information of each order into the combined parameter value prediction model, the combined parameter value prediction model processes the feature information of each order and outputs the parameter values of the combined operation for indicating the combined strategy of these multiple orders to be processed. The combined operation parameter values include the combined strategy indicating which orders among these multiple orders to be processed should be combined with which orders, or the combined operation parameter values include the combined strategy indicating which orders among these multiple orders to be processed should be combined with which orders and the order pressing time of each order.

[0079] Taking orders 1 to order m exemplified in Table 1 above as an example, where orders 1, 2, and 3 have similar order characteristics. When not considering the order pressing time of each order, the three orders can be combined. However, in the prior art, in order to cooperate with the greedy algorithm to determine the optimal combined strategy, when the order pressing time of each order is fixed at 3 minutes, order 1 can only be combined with order 2. When order 3 enters the order pool, orders 1 and 2 have already left the order pool and entered the scheduling pool, resulting in order 3 not being able to be combined with orders 1 and 2.

[0080] After inputting the respective feature information of orders 1 to order m into the combined parameter value prediction model obtained through training in steps S101 to S103, the combined parameter value prediction model will comprehensively consider the feature information of orders 1 to order m. The combined strategy of each order output by the combined parameter value prediction model is specifically shown in Table 2 below.

[0081] Table 2

[0082]

[0083]

[0084] It can be seen that the order pressing time of each order is not fixed, that is, the combined parameter value prediction model dynamically adjusts the combined strategy of each order according to the feature information of each order, and there is no need to set a fixed order pressing time for each order.

[0085] Specifically, in an optional implementation manner, in step S302, inputting multiple orders to be processed into the combined parameter value prediction model to obtain the parameter values for respectively performing combined operations on the multiple orders to be processed, including:

[0086] S3021, obtaining the parameter values for respectively performing combined operations on the multiple orders to be processed according to the target parameter values corresponding to the combined parameter value prediction model.

[0087] That is, when the combined parameter value prediction model makes a prediction, for the same batch of the same batch of pending orders, the output combined operation parameters are consistent with the target parameter values determined during pre-training, and the combination strategy indicated by the target parameter value is already the optimal combination strategy or the combination strategy that meets the preset conditions.

[0088] S303, combining the multiple pending orders according to the combination parameter values corresponding to the multiple pending orders respectively, to obtain multiple pending order packages.

[0089] After obtaining the combined parameter value (combination strategy, or order holding time and combination strategy) of each pending order, multiple pending orders are combined according to the combination strategy in the combined parameter value, thereby obtaining multiple pending order packages, each of which includes multiple pending orders with the same or similar properties. For example, each pending order package is composed of multiple delivery orders with a convenience rate of 90%. For example, combine order 1, order 2, and order 3 in the above example to obtain a pending order package.

[0090] S304, adding the multiple pending order packages to the scheduling pool, so that the order scheduling system can schedule the multiple pending order packages.

[0091] After combining multiple pending order packages, the order packages are placed in the scheduling pool (i.e., the scheduling system) for scheduling by the scheduling system in the form of order packages. For example, in the above example, after combining order 1, order 2, and order 3 to obtain a pending order package, the package is placed in the scheduling pool for scheduling by the scheduling system. For example, the information of the pending order package is sent to a delivery person terminal, so that the corresponding delivery person delivers each pending order included in the processing order package.

[0092] The characteristic information of multiple orders is processed by the combined parameter value prediction model, the combined operation parameters of each order in the multiple orders are dynamically determined, the combination strategy for multiple orders is dynamically determined according to the combined operation parameters, at least part of the orders are combined into a pending order package according to the combination strategy, and then the scheduling system schedules the pending order package in the form of a package. In this way, it is not necessary to keep a fixed order pressing time for each order, and the orders can be combined flexibly and dynamically with the order as the granularity. The order pressing time of the order is flexible and variable, avoiding the situation in the related art where the fixed or continuously reduced order pressing time causes the package closing rate to decrease. In addition, the combined parameter value prediction model can comprehensively consider the characteristic information of multiple orders, so the obtained combination strategy is more reasonable, and the orders are combined and then scheduled according to the more reasonable combination strategy, which reduces the waiting time of users and improves the user experience.

[0093] In an alternative implementation, after step S303, the combined order performance processing value according to the order and the target parameter value can also be used to update the combined parameter value prediction model, which specifically includes:

[0094] S501, determine the order processing performance value corresponding to each of the multiple orders to be processed.

[0095] That is, after combining multiple orders according to the target parameter value, determine the policy benefit of the combination strategy corresponding to each order to be processed.

[0096] S502, update the target parameter value corresponding to the combined parameter value prediction model according to the order processing performance value corresponding to each of the multiple orders to be processed and the parameter values for which the multiple orders to be processed perform the combination operation.

[0097] Continue to use the orders combined according to the target parameter value to form a new training set S″, and use the training set S″ to iteratively optimize the combined parameter value prediction model, that is, extract the feature information of the orders in the training set S″, input it into the combined parameter value prediction model, and update the target parameter value corresponding to the combined parameter value prediction model according to the parameter values for which the multiple orders to be processed perform the combination operation and the output result of the combined parameter value prediction model. Subsequently, the updated target parameter value can be used to determine whether each order participates in the combined package and which orders to combine with for the subsequent processing of the order.

[0098] In an alternative implementation, the order scheduling system schedules multiple orders to be processed according to the backlog time of each order to be processed package. Among them, the backlog time of the package refers to the waiting stay time of each order to be processed package. Therefore, after obtaining multiple orders to be processed packages in step S303, the method also needs to determine the backlog time of each order to be processed package, that is, it also includes the following steps:

[0099] S3031, when the combined parameter value corresponding to the order to be processed at least includes the backlog time of the order to be processed, for each order to be processed package among the multiple orders to be processed packages, determine the backlog time of the order to be processed package according to the backlog time of each order to be processed in the order to be processed package.

[0100] When the combined parameter value includes the order holding time for each order, after combining multiple orders to be processed according to the combined parameter value and obtaining an order package to be processed, for the order holding time of each order package to be processed, the shortest order holding time among the order holding times of each order in the order package to be processed is determined as the package order holding time of the order package to be processed. For example: After combining Order 1, Order 2, and Order 3 in the above example to obtain an order package to be processed, the order holding time of this order package to be processed is 1 minute, that is, at 12:06, the scheduling system schedules the order package composed of Order 1, Order 2, and Order 3.

[0101] S3032. When the order information of the order to be processed includes the maximum order holding time, for each order package to be processed among multiple order packages to be processed, according to the maximum order holding time of each order to be processed in the order package to be processed, the package order holding time of the order package to be processed is determined.

[0102] When the order information of each order to be processed itself includes the maximum order holding time initially set for the order to be processed itself, for an order package to be processed, the package order holding time of the order package to be processed should be determined according to the maximum order holding time initially set for each order to be processed in the order package to be processed.

[0103] For example, a certain order package to be processed includes Order 4, Order 5, and Order 6, and the order holding time of Order 4 is 1 minute, the order holding time of Order 5 is 2 minutes, and the order holding time of Order 6 is 3 minutes. Then the package order holding time of the order package to be processed composed of Order 4, Order 5, and Order 6 should be determined as the order holding time of Order 4, that is, the package order holding time of this order package to be processed is 1 minute.

[0104] The order scheduling system schedules the order package or the order. For example, scheduling multiple order packages to be processed can be to assign a delivery person to each order package to be processed, plan the delivery route, etc., which are not specifically limited in this application, but should all meet the requirement that when the order package to be processed ends at the package order holding time, scheduling processing is immediately performed.

[0105] In an optional implementation manner, after determining the package order holding time through the above step S3031 or step S3032, step S304 adds multiple order packages to be processed to the scheduling pool for the order scheduling system to schedule multiple order packages to be processed, including:

[0106] S3041 adds multiple order packages to be processed carrying the package order holding time to the scheduling pool for the order scheduling system to schedule multiple order packages to be processed according to the package order holding time carried by each of the multiple order packages to be processed; or

[0107] S3042. For each pending order package among multiple pending order packages, when the package hold time of the pending order package ends, add the pending order package to the scheduling pool for the order scheduling system to schedule the pending order package.

[0108] Specifically, after determining the package hold time of each pending order package through step S3031 or step S3032, add multiple pending order packages to the scheduling pool and schedule them according to the package hold time of each pending order package, such as specific scheduling operations like assigning appropriate deliverymen and planning corresponding delivery routes.

[0109] Of course, it is also possible to let the pending order package wait in the order pool to wait for whether new orders can be combined into the pending order package again. Obviously, the maximum waiting time is the package hold time of the pending order package. When the package hold time of the pending order package ends, add the pending order package to the scheduling pool for the order scheduling system to schedule the pending order package. For specific operations of scheduling, refer to the description above, such as assigning deliverymen and planning delivery routes.

[0110] When combining and processing multiple pending orders, some of the pending orders may not be successfully combined. Therefore, these orders that are not successfully combined should be scheduled separately. In an alternative embodiment, refer to Figure 4 , Figure 4 is a flowchart of an order processing method proposed in an embodiment of the present application, exemplarily showing the process of scheduling uncombined orders. As Figure 4 shown, after step S303, the method includes the following steps:

[0111] S401. For the remaining uncombined orders among multiple pending orders: Add the remaining order with the maximum hold time to the scheduling pool for the order scheduling system to schedule the remaining order package; or

[0112] S402. For the remaining uncombined orders among multiple pending orders: When the maximum hold time of the remaining order ends, add the remaining order to the scheduling pool for the order scheduling system to schedule the remaining order package.

[0113] Specifically, the uncombined pending orders also need to be scheduled. Similar to the combined pending order packages mentioned above, the uncombined pending orders can directly enter the scheduling pool for the scheduling system to schedule; of course, they can also wait in the order pool until the maximum hold time ends and then enter the scheduling pool for the scheduling system to schedule.

[0114] Based on the same inventive concept, an embodiment of the present application provides an order processing device. Refer toFigure 5 , Figure 5 is a schematic diagram of an order processing device provided by an embodiment of the present application. As Figure 5 shown, the device includes:

[0115] A first obtaining module 501, configured to obtain a plurality of orders to be processed;

[0116] A first input module 502, configured to input the order feature information of each of the plurality of orders to be processed into a combined parameter value prediction model, and obtain parameter values for performing combined operations on the plurality of orders to be processed respectively;

[0117] A first combining module 503, configured to combine the plurality of orders to be processed according to the combined parameter values corresponding to each of the plurality of orders to be processed, and obtain a plurality of order packages to be processed;

[0118] A first adding module 504, configured to add the plurality of order packages to be processed to a scheduling pool for an order scheduling system to schedule the plurality of order packages to be processed.

[0119] Optionally, the device further includes:

[0120] A first package hold time determination module, configured to, when the combined parameter value corresponding to an order to be processed at least includes the hold time of the order to be processed, for each order package among the plurality of order packages to be processed, determine the package hold time of the order package according to the hold time of each order to be processed in the order package;

[0121] A second package hold time determination module, configured to, when the order information of an order to be processed includes a maximum hold time, for each order package among the plurality of order packages to be processed, determine the package hold time of the order package according to the maximum hold time of each order to be processed in the order package;

[0122] Wherein, the order scheduling system is configured to schedule the plurality of order packages according to the package hold time of each of the plurality of order packages.

[0123] Optionally, the first adding module is further configured to add the plurality of order packages carrying the package hold time to the scheduling pool for the order scheduling system to schedule the plurality of order packages according to the package hold time carried by each of the plurality of order packages; or, is further configured to, for each order package among the plurality of order packages, add the order package to the scheduling pool when the package hold time of the order package ends, for the order scheduling system to schedule the order package.

[0124] Optionally, the device further includes:

[0125] A second acquisition module, configured to acquire the order feature information of each of multiple sample orders, where the order feature information of one sample order includes the order attribute information of this sample order and the association information between this sample order and other sample orders;

[0126] A first determination module, configured to determine the label corresponding to each of the multiple sample orders, where the label corresponding to one sample order represents whether this sample order and other sample orders have been combined into a sample order package;

[0127] A first training module, configured to train a preset model according to the order feature information and corresponding labels of each of the multiple sample orders to obtain the combined parameter value prediction model.

[0128] Optionally, the device further includes:

[0129] A second determination module, configured to input the order feature information of a preset number of sample orders into the combined parameter value prediction model to determine the sample parameter values for performing combination operations on the preset number of sample orders respectively;

[0130] A third determination module, configured to impose dynamic interference on the sample parameter values to determine the target parameter values corresponding to the combined parameter value prediction model;

[0131] The first input module includes:

[0132] A third acquisition module, configured to obtain the parameter values for performing combination operations on each of the multiple orders to be processed according to the target parameter values corresponding to the combined parameter value prediction model.

[0133] Optionally, the third acquisition module includes:

[0134] A first adjustment unit, configured to adjust the sample parameter values multiple times to obtain the sample parameter values after multiple adjustments;

[0135] A first determination unit, configured to determine the order processing performance values corresponding to the sample parameter values and the sample parameter values after each adjustment respectively;

[0136] A second determination unit, configured to determine the sample parameter values whose corresponding order processing performance values meet a preset threshold among the sample parameter values and the sample parameter values after multiple adjustments as the target parameter values corresponding to the combined parameter value prediction model.

[0137] Optionally, the device further includes:

[0138] A fourth determination module, configured to determine the order processing performance values corresponding to each of the multiple orders to be processed;

[0139] A first update module, configured to update a target parameter value corresponding to the combined parameter value prediction model according to the order processing performance values corresponding to the multiple orders to be processed respectively, and the parameter values of the combined operations performed by the multiple orders to be processed respectively.

[0140] Optionally, the first addition module is further configured to process the remaining orders that have not been combined among the multiple orders to be processed, including:

[0141] Adding the remaining order carrying the maximum order pressing time to the scheduling pool for the order scheduling system to schedule the remaining order package; or

[0142] At the end of the maximum order pressing time of the remaining order, adding the remaining order to the scheduling pool for the order scheduling system to schedule the remaining order package.

[0143] Based on the same inventive concept, another embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the steps in the method described in any one of the above embodiments of the present application are implemented.

[0144] Based on the same inventive concept, another embodiment of the present application provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, and when the processor executes, the steps in the method described in any one of the above embodiments of the present application are implemented.

[0145] For the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and for the related parts, reference can be made to the partial description of the method embodiment.

[0146] Each embodiment in this specification is described in a progressive manner, and the key point of each embodiment is to illustrate the differences from other embodiments. For the same or similar parts among the embodiments, reference can be made to each other.

[0147] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a device, or a computer program product. Therefore, the embodiments of the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the embodiments of the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0148] Embodiments of the present application are described with reference to the flowcharts and / or block diagrams of methods, terminal devices (systems), and computer program products according to embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, as well as the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing terminal device to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing terminal device generate a device for implementing the functions specified in one process Figure 1 one process or multiple processes and / or blocks Figure 1 or a device for implementing the functions specified in multiple blocks.

[0149] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing terminal device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including an instruction device that implements the functions specified in one process Figure 1 one process or multiple processes and / or blocks Figure 1 or a device for implementing the functions specified in multiple blocks.

[0150] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal device, such that a series of operation steps are executed on the computer or other programmable terminal device to generate a computer-implemented process, so that the instructions executed on the computer or other programmable terminal device provide steps for implementing the functions specified in one process Figure 1 one process or multiple processes and / or blocks Figure 1 or a device for implementing the functions specified in multiple blocks.

[0151] Although the preferred embodiments of the embodiments of the present application have been described, those skilled in the art can make additional changes and modifications to these embodiments once they know the basic creative concept. Therefore, the appended claims are intended to be construed as including the preferred embodiments and all changes and modifications that fall within the scope of the embodiments of the present application.

[0152] Finally, it should also be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or terminal device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or terminal device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, article or terminal device comprising the said element.

[0153] The above has introduced in detail an order processing method, device, electronic device and readable storage medium provided by the present application. Specific examples are used in this article to elaborate on the principle and implementation manner of the present application. The description of the above embodiments is only used to help understand the method and its core idea of the present application; at the same time, for those of ordinary skill in the art, according to the idea of the present application, there will be changes in the specific implementation manner and application scope. In summary, the content of this specification should not be construed as a limitation to the present application.

Claims

1. An order processing method, characterized in that, The method includes: Obtaining a plurality of orders to be processed; Inputting the order feature information of each of the plurality of orders to be processed into a combined parameter value prediction model to obtain parameter values for performing combined operations on the plurality of orders to be processed respectively; Combining the plurality of orders to be processed according to the respective combined parameter values corresponding to the plurality of orders to be processed to obtain a plurality of order packages to be processed; Adding the plurality of order packages to be processed to a scheduling pool for an order scheduling system to schedule the plurality of order packages to be processed; Before inputting the order feature information of each of the plurality of orders to be processed into the combined parameter value prediction model, the method further includes: Obtaining the order feature information of each of a plurality of sample orders, where the order feature information of a sample order includes the order attribute information of the sample order and the association information between the sample order and other sample orders; Determining the respective labels corresponding to the plurality of sample orders, where the label corresponding to a sample order represents whether the sample order and other sample orders have been combined into a sample order package; Training a preset model according to the order feature information and corresponding labels of the plurality of sample orders to obtain the combined parameter value prediction model; Inputting the order feature information of a preset number of sample orders into the combined parameter value prediction model to determine sample parameter values for performing combined operations on the preset number of sample orders respectively; Applying dynamic interference to the sample parameter values to determine the target parameter values corresponding to the combined parameter value prediction model; Inputting the plurality of orders to be processed into the combined parameter value prediction model to obtain parameter values for performing combined operations on the plurality of orders to be processed respectively, including: Obtaining the parameter values for performing combined operations on the plurality of orders to be processed respectively according to the target parameter values corresponding to the combined parameter value prediction model.

2. The method according to claim 1, wherein After obtaining the plurality of order packages to be processed, the method further includes: When the combined parameter value corresponding to an order to be processed at least includes the order's hold time, for each order package among the plurality of order packages to be processed, determining the package hold time of the order package according to the hold time of each order to be processed in the order package; When the order information of an order to be processed includes the maximum hold time, for each order package among the plurality of order packages to be processed, determining the package hold time of the order package according to the maximum hold time of each order to be processed in the order package; Wherein, the order scheduling system is used to schedule the plurality of order packages to be processed according to the respective package hold times of the plurality of order packages to be processed.

3. The method according to claim 2, wherein Adding the plurality of order packages to be processed to a scheduling pool for an order scheduling system to schedule the plurality of order packages to be processed, including: Add multiple to-be-processed order packages with package hold times to a scheduling pool for an order scheduling system to schedule the multiple to-be-processed order packages according to the package hold times carried by the multiple to-be-processed order packages respectively; or for each to-be-processed order package among the multiple to-be-processed order packages, add the to-be-processed order package to the scheduling pool when the package hold time of the to-be-processed order package ends, for the order scheduling system to schedule the to-be-processed order package.

4. The method according to claim 1, wherein Applying a dynamic interference to the sample parameter values to determine the target parameter values corresponding to the combined parameter value prediction model, including: Adjusting the sample parameter values multiple times to obtain the sample parameter values after multiple adjustments; Determining the order processing performance values corresponding to the sample parameter values and the sample parameter values after each adjustment respectively; Determining, among the sample parameter values and the sample parameter values after multiple adjustments, the sample parameter values whose corresponding order processing performance values meet a preset threshold as the target parameter values corresponding to the combined parameter value prediction model.

5. The method according to claim 1, wherein After obtaining the parameter values for the multiple to-be-processed orders to perform combined operations respectively according to the target parameter values corresponding to the combined parameter value prediction model, the method further includes: Determining the order processing performance values corresponding to the multiple to-be-processed orders respectively; Updating the target parameter values corresponding to the combined parameter value prediction model according to the order processing performance values corresponding to the multiple to-be-processed orders respectively and the parameter values for the multiple to-be-processed orders to perform combined operations respectively.

6. The method according to any one of claims 1-5, characterized in that, After combining the multiple to-be-processed orders to obtain multiple to-be-processed order packages, the method further includes: For the remaining orders among the multiple to-be-processed orders that are not combined: Adding the remaining order with the maximum hold time to the scheduling pool for the order scheduling system to schedule the remaining order; or adding the remaining order to the scheduling pool when the maximum hold time of the remaining order ends, for the order scheduling system to schedule the remaining order.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by a processor, it implements the steps in the method according to any one of claims 1-6.

8. An electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein, When the processor executes, it implements the steps of the method according to any one of claims 1-6.

Images