Time probability distribution model training method, delivery time acquisition method and device

Abstract

The present disclosure provides a time probability distribution model training method, a delivery time acquisition method, and a device. The training method includes: obtaining a training sample; the training sample is formed by order information of historical delivery orders and delivery personnel information of delivery personnel associated with the historical delivery orders, and the training sample corresponds to a first delivery time; obtaining multiple initial delivery times set for the training sample; inputting the training sample and the multiple initial delivery times into an initial time probability distribution model, and obtaining the delivery probabilities corresponding to the multiple initial delivery times output by the initial time probability distribution model; determining a second delivery time among the multiple initial delivery times based on the delivery probabilities; and using the trained initial time probability distribution model as a target time probability distribution model when the second delivery time matches the first delivery time. The present disclosure can improve the accuracy of estimated delivery times.

Description

Technical Field

[0001] The embodiments of the present disclosure relate to the field of Internet technology, and in particular to a time probability distribution model training method, a delivery time acquisition method and a device. Background Art

[0002] In instant delivery scenarios, when users place an order, the platform displays the estimated time of arrival (ETA) for the order. The ETA plays a crucial role in the entire business. As the platform's fulfillment commitment to users, the ETA impacts the user experience. Furthermore, the dispatch system uses the ETA as a scheduling constraint, influencing the dispatch system's order assignments and overall delivery efficiency. Furthermore, delivery personnel's performance time is strongly correlated with the ETA, impacting their delivery experience. Overall, the ETA, as the first step in the delivery process, has a direct impact on both user experience and efficiency. There is a business need to use the ETA to adjust the distribution of delivery time deviations (ETA minus actual delivery time) to optimize both user experience and efficiency. User experience, delivery personnel stress, and system efficiency vary under different delivery time deviations. Excessive deviations result in poor user experience. However, when the deviation is too small, the user experience is poor, delivery personnel stress increases, and the dispatch system may assign more on-the-go orders to delivery personnel, resulting in lower overall system efficiency. Summary of the Invention

[0003] The embodiments of the present disclosure provide a time probability distribution model training method, a delivery time acquisition method and a device, which are used to improve the accuracy of delivery time estimation, improve the user and delivery personnel experience, and improve the overall efficiency of the system.

[0004] According to a first aspect of an embodiment of the present disclosure, a time probability distribution model training method is provided, comprising:

[0005] Obtaining a training sample; the training sample is formed by order information of a historical delivery order and delivery personnel information of a delivery personnel associated with the historical delivery order, and the training sample corresponds to a first delivery time;

[0006] Obtaining multiple initial delivery times set for the training samples;

[0007] Inputting the training sample and the multiple initial delivery times into an initial time probability distribution model, and obtaining delivery probabilities corresponding to the multiple initial delivery times output by the initial time probability distribution model;

[0008] determining a second delivery time among the multiple initial delivery times according to the delivery probability;

[0009] When the second delivery time matches the first delivery time, the trained initial time probability distribution model is used as the target time probability distribution model.

[0010] Optionally, obtaining a training sample includes:

[0011] Obtaining basic order information of the historical delivery order and business party information of the business party associated with the historical delivery order, and using the basic order information and the business party information as the order information;

[0012] Obtaining delivery personnel information of delivery personnel within a preset distance range from the business party's location;

[0013] Among them, the basic order information includes the order time, the time period of the order time, the delivery distance, the order price and the location information of the recipient; the business party information includes the business party location information and the order time information; the delivery personnel information includes the location information of the delivery personnel, the delivery information of the order held by the delivery personnel, the delivery time information and the delivery capacity information.

[0014] Optionally, inputting the training sample and the multiple initial delivery times into an initial time probability distribution model, and obtaining delivery probabilities corresponding to the multiple initial delivery times output by the initial time probability distribution model, includes:

[0015] inputting the training sample and the plurality of initial delivery times into the initial time probability distribution model;

[0016] Determining, based on the initial time probability distribution model, an order coding vector corresponding to the order information and a delivery personnel coding vector corresponding to the delivery personnel information;

[0017] Based on the initial time probability distribution model and the order coding vector and the delivery personnel coding vector, the delivery probabilities corresponding to the multiple initial delivery times are predicted.

[0018] Optionally, determining a second delivery time among the multiple initial delivery times according to the delivery probability includes:

[0019] According to the time differences between the multiple initial delivery times and the first delivery time, obtaining at least one initial delivery time whose time difference is within a preset time range;

[0020] An initial delivery time with the highest delivery probability among the at least one initial delivery time is obtained, and the initial delivery time with the highest delivery probability is used as the second delivery time.

[0021] According to a second aspect of an embodiment of the present disclosure, a method for obtaining a delivery time is provided, comprising:

[0022] After receiving the target order, obtaining target order information associated with the target order and target delivery information of target delivery personnel within a preset distance range of the target business party where the target order is located;

[0023] Obtain multiple third delivery times set for the target order;

[0024] Inputting the target order information, the target delivery information, and the plurality of third delivery times into a target time probability distribution model, and obtaining target delivery probabilities corresponding to the plurality of third delivery times output by the target time probability distribution model;

[0025] According to the target delivery probability, a target delivery time corresponding to the target order is obtained by filtering from the multiple third delivery times.

[0026] Optionally, the filtering, based on the target delivery probability, to obtain the target delivery time corresponding to the target order from the plurality of third delivery times includes:

[0027] Obtaining at least one fourth delivery time within a preset time range from the plurality of third delivery times;

[0028] According to the target delivery probability, a delivery time with the highest delivery probability is selected from the at least one fourth delivery time, and the delivery time with the highest delivery probability is used as the target delivery time.

[0029] According to a third aspect of an embodiment of the present disclosure, a time probability distribution model training device is provided, comprising:

[0030] A training sample acquisition module, configured to acquire a training sample; the training sample is formed by order information of a historical delivery order and delivery personnel information of a delivery personnel associated with the historical delivery order, the training sample corresponding to a first delivery time;

[0031] An initial time acquisition module, configured to acquire a plurality of initial delivery times set for the training samples;

[0032] a delivery probability acquisition module, configured to input the training sample and the multiple initial delivery times into an initial time probability distribution model, and obtain delivery probabilities corresponding to the multiple initial delivery times output by the initial time probability distribution model;

[0033] a second time determination module, configured to determine a second delivery time among the plurality of initial delivery times according to the delivery probability;

[0034] A target model acquisition module is configured to use the trained initial time probability distribution model as a target time probability distribution model when the second delivery time matches the first delivery time.

[0035] Optionally, the training sample acquisition module includes:

[0036] An order information acquisition unit, configured to acquire basic order information of the historical delivery order and business party information of the business party associated with the historical delivery order, and use the basic order information and the business party information as the order information;

[0037] a delivery personnel information acquisition unit, configured to acquire delivery personnel information of delivery personnel within a preset distance range from the business party's location;

[0038] Among them, the basic order information includes the order time, the time period of the order time, the delivery distance, the order price and the location information of the recipient; the business party information includes the business party location information and the order time information; the delivery personnel information includes the location information of the delivery personnel, the delivery information of the order held by the delivery personnel, the delivery time information and the delivery capacity information.

[0039] Optionally, the delivery probability acquisition module includes:

[0040] a training sample input unit, configured to input the training sample and the plurality of initial delivery times into the initial time probability distribution model;

[0041] a coding vector determining unit, configured to determine, based on the initial time probability distribution model, an order coding vector corresponding to the order information and a delivery personnel coding vector corresponding to the delivery personnel information;

[0042] The delivery probability prediction unit is used to predict the delivery probabilities corresponding to the multiple initial delivery times based on the initial time probability distribution model according to the order coding vector and the delivery personnel coding vector.

[0043] Optionally, the second time determination module includes:

[0044] A first time acquisition unit is configured to acquire, based on the time differences between the multiple initial delivery times and the first delivery time, at least one initial delivery time whose time difference is within a first preset time range;

[0045] The second time acquisition unit is configured to acquire an initial delivery time with the highest delivery probability from the at least one initial delivery time, and use the initial delivery time with the highest delivery probability as the second delivery time.

[0046] According to a fourth aspect of an embodiment of the present disclosure, there is provided a delivery time acquisition device, comprising:

[0047] A target order information acquisition module is used to, after receiving a target order, acquire target order information associated with the target order and target delivery information of a target delivery person within a preset distance range of a target business party where the target order is located;

[0048] A third delivery time acquisition module, configured to acquire a plurality of third delivery times set for the target order;

[0049] a target delivery probability acquisition module, configured to input the target order information, the target delivery information, and the plurality of third delivery times into a target time probability distribution model, and acquire target delivery probabilities corresponding to the plurality of third delivery times output by the target time probability distribution model;

[0050] The target delivery time screening module is used to screen the target delivery time corresponding to the target order from the multiple third delivery times according to the target delivery probability.

[0051] Optionally, the target delivery time screening module includes:

[0052] a fourth delivery time acquiring unit, configured to acquire at least one fourth delivery time within a preset time range from among the plurality of third delivery times;

[0053] The target delivery time acquisition unit is configured to select a delivery time with the highest delivery probability from the at least one fourth delivery time according to the target delivery probability, and use the delivery time with the highest delivery probability as the target delivery time.

[0054] According to a fifth aspect of an embodiment of the present disclosure, there is provided an electronic device, including:

[0055] A processor, a memory, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method for training a time probability distribution model described in any one of the above items or the method for obtaining delivery time described in any one of the above items is implemented.

[0056] According to the sixth aspect of an embodiment of the present disclosure, a readable storage medium is provided. When the instructions in the storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute any of the above-mentioned time probability distribution model training methods or any of the above-mentioned delivery time acquisition methods.

[0057] The embodiments of the present disclosure provide a time probability distribution model training method, a delivery time acquisition method and an apparatus, by acquiring a training sample, the training sample being formed by order information of a historical delivery order and delivery personnel information of a delivery personnel associated with the historical delivery order, the training sample corresponding to a first delivery time, acquiring multiple initial delivery times set for the training sample, inputting the training sample and the multiple initial delivery times into an initial time probability distribution model, and acquiring the delivery probabilities corresponding to the multiple initial delivery times output by the initial time probability distribution model, determining the second delivery time among the multiple initial delivery times based on the delivery probabilities, and using the trained initial time probability distribution model as the target time probability distribution model when the second delivery time matches the first delivery time. The embodiments of the present disclosure can finely characterize the delivery time deviation by finely characterizing the probability distribution of delivery time given different ETAs during the modeling process, thereby improving the accuracy of the estimated delivery time, and improving the overall efficiency of the system while improving the user and delivery personnel experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the following briefly introduces the drawings required for use in the description of the embodiments of the present disclosure. Obviously, the drawings described below are only some embodiments of the embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0059] Figure 1 A flowchart of a method for training a temporal probability distribution model according to an embodiment of the present disclosure;

[0060] Figure 2 A flowchart of a method for obtaining delivery time according to an embodiment of the present disclosure;

[0061] Figure 3 A schematic diagram of the structure of a time probability distribution model training device provided by an embodiment of the present disclosure;

[0062] Figure 4 A schematic diagram of the structure of a delivery time acquisition device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0063] The following will be combined with the drawings in the embodiments of the present disclosure to clearly and completely describe the technical solutions in the embodiments of the present disclosure. Obviously, the described embodiments are part of the embodiments of the present disclosure, but not all of them. Based on the embodiments of the present disclosure, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the embodiments of the present disclosure.

[0064] Example 1

[0065] Reference Figure 1 , shows a flow chart of the steps of a time probability distribution model training method provided by an embodiment of the present disclosure, such as Figure 1 As shown, the time probability distribution model training method may specifically include the following steps:

[0066] Step 101: Acquire a training sample; the training sample is formed by order information of a historical delivery order and delivery personnel information of a delivery personnel associated with the historical delivery order, and the training sample corresponds to a first delivery time.

[0067] The embodiments of the present disclosure may be applied to a scenario where a time probability distribution model is trained by combining order information associated with a delivery order and delivery personnel information associated with the delivery order.

[0068] In this embodiment, training samples are formed from order information of historical delivery orders and delivery personnel information associated with the historical delivery orders. The order information may include basic order information and business party information associated with the historical delivery orders. Delivery personnel information may include delivery personnel within a preset distance range from the business party. The process of obtaining training samples is described in detail in conjunction with the following specific implementation methods.

[0069] In a specific implementation of the present disclosure, step 101 may include:

[0070] Sub-step A1: Obtain basic order information of the historical delivery order and business party information of the business party associated with the historical delivery order, and use the basic order information and the business party information as the order information.

[0071] In this embodiment, the basic order information may include the order time (i.e., the moment the order is placed, such as 1:30 a.m., etc.), the time period of the order (such as morning, noon, etc.), the delivery distance (i.e., the distance between the merchant where the order is placed and the ordering user), the order price (i.e., the total price of historical delivery orders) and the location information of the acquirer (i.e., where the ordering user receives the order).

[0072] Business party information refers to the relevant information of the business party (such as the merchant) of the historical delivery order, which may include the business party's location information and order time information.

[0073] When training samples are needed, historical delivery orders (i.e., orders that have been successfully delivered) can be obtained from the historical order platform. Then, the basic order information of the historical delivery orders and the business party information associated with the historical delivery orders can be obtained by combining the historical delivery orders and the business party information associated with the business party. The basic order information and the business party information can be used as the order information of the historical delivery orders.

[0074] Sub-step A2: Acquire delivery personnel information of delivery personnel who are within a preset distance range from the business party's location.

[0075] The delivery personnel information refers to the information associated with the personnel who delivered the historical orders (ie, the delivery personnel) and the information associated with the delivery personnel who are within a preset distance range from the business party.

[0076] In this example, the delivery personnel information may include the delivery personnel's location information, delivery information of the orders held by the delivery personnel (delivery order information of the orders held, delivery distance information, etc.), delivery punctuality information (i.e., the delivery personnel's historical punctuality information) and delivery capacity information (i.e., the delivery personnel's ability to deliver orders).

[0077] It is understandable that the number of training samples can be 800, 1000, etc., specifically, it can be determined according to business needs, and this embodiment does not impose any limitation thereto. Each training sample corresponds to a historical delivery order.

[0078] The first delivery time refers to the actual delivery time of a historical delivery order, that is, the moment when the historical delivery order was actually delivered. For example, if the historical delivery order is Order A, and Order A takes a total of 30 minutes from the time the order is placed to the time the order is accepted by the orderer, then the first delivery time is 30 minutes.

[0079] It can be understood that the above examples are merely examples listed for better understanding the technical solutions provided by the embodiments of the present disclosure, and are not intended to be the sole limitation on the embodiments.

[0080] After the training samples are obtained, step 102 is performed.

[0081] Step 102: Acquire multiple initial delivery times set for the training samples.

[0082] The initial delivery time refers to the delivery time pre-assigned by the business personnel to each training sample. In this embodiment, the business personnel can select several times from 1 to 90 minutes as the initial delivery time, such as 8 minutes from 1 to 10 minutes, 15 minutes from 11 to 20 minutes, ..., 87 minutes from 80 to 90 minutes, etc., as the initial delivery time.

[0083] It can be understood that the above examples are merely examples listed for better understanding the technical solutions of the embodiments of the present disclosure, and are not intended to be the sole limitation on the embodiments.

[0084] After obtaining the training samples, business personnel can assign multiple initial delivery times to each training sample.

[0085] The embodiments of the present disclosure estimate the probability distribution of the waybill delivery time under a given ETA by giving the waybill delivery time (i.e., the initial delivery time). By integrating the ETA value into the model, it is possible to characterize the delivery time deviation and achieve the purpose of directional optimization of the ETA deviation.

[0086] After obtaining multiple initial delivery times set for the training samples, step 103 is executed.

[0087] Step 103: Input the training sample and the multiple initial delivery times into an initial time probability distribution model, and obtain the delivery probabilities corresponding to the multiple initial delivery times output by the initial time probability distribution model.

[0088] The initial time probability distribution model refers to a time probability distribution model that has not been trained.

[0089] The delivery probability refers to the probability corresponding to each initial delivery time output by the initial time probability distribution model, that is, the probability of delivering the historical delivery order to the ordering user at each initial delivery time.

[0090] After obtaining the training samples and multiple initial delivery times corresponding to the training samples, the training samples and the multiple initial delivery times can be input into the initial time probability distribution model, so that the initial time probability distribution model can output the delivery probabilities corresponding to the multiple initial delivery times. Specifically, this process can be described in detail in combination with the following specific implementation method.

[0091] In another specific implementation of the present disclosure, step 103 may include:

[0092] Sub-step B1: Inputting the training sample and the multiple initial delivery times into the initial time probability distribution model.

[0093] In this embodiment, after obtaining the training samples and the multiple initial delivery times corresponding to the training samples, the training samples and the multiple initial delivery times may be input into the initial time probability distribution model, and then sub-step B2 is executed.

[0094] Sub-step B2: Based on the initial time probability distribution model, determine the order coding vector corresponding to the order information and the delivery personnel coding vector corresponding to the delivery personnel information.

[0095] The order encoding vector refers to the vector generated by encoding the order information in the training sample.

[0096] The delivery personnel encoding vector refers to the vector generated by encoding the delivery personnel information in the training sample.

[0097] After inputting the training samples into the initial temporal probability distribution model, the model can be used to encode the order information and delivery person information, respectively, to obtain order encoding vectors and delivery person encoding vectors. Specifically, during the feature encoding of the delivery person, the delivery person's feature information is fused with the information of m (m is a positive integer greater than or equal to 1) delivery waybills carried by the delivery person. The correlation between the waybills carried by the delivery person determines the order and time of their delivery. Self-Attention encoding is used to fuse and characterize the correlation of the waybills, resulting in a vector representation of each delivery person, characterizing the characteristics of the delivery person and the waybills carried by them.

[0098] After obtaining the order code vector and the delivery personnel code vector, execute sub-step B3.

[0099] Sub-step B3: Based on the initial time probability distribution model, according to the order code vector and the delivery personnel code vector, the delivery probabilities corresponding to the multiple initial delivery times are predicted.

[0100] After obtaining the order code vector and the delivery person code vector, the initial time probability distribution model can be used to predict the delivery probabilities corresponding to multiple initial delivery times. Specifically, to characterize the correlation between the current waybill and the relevant delivery person, an attention operation is performed on the vector code of the current waybill and the vector representation of the delivery person to assess the probability of the current waybill being delivered by each delivery person. This can more accurately characterize the delivery status of the waybill. After all this fused information is processed and activated using the softmax function, the probability of the delivery time of the waybill being between 1 and 90 minutes is obtained, given an ETA value of 30 to 90 minutes.

[0101] The embodiments of the present disclosure introduce real-time basic information at the time of estimation (i.e., the orders to be estimated, and information related to the delivery personnel within a preset range (such as the orders carried by the delivery personnel and the order and time of pick-up and delivery of each order)), and use the Attention mechanism to automatically fuse and characterize the information, which can better estimate the delivery status of the waybill and improve the accuracy of the ETA estimation.

[0102] After obtaining the delivery probabilities corresponding to the multiple initial delivery times output by the initial time probability distribution model, step 104 is executed.

[0103] Step 104: Determine a second delivery time among the multiple initial delivery times according to the delivery probability.

[0104] The second delivery time refers to a time that satisfies the delivery probability prediction and is selected from multiple initial delivery times.

[0105] After obtaining the delivery probabilities corresponding to the multiple initial delivery times output by the initial time probability distribution model, the second delivery time among the multiple initial delivery times can be determined based on the delivery probabilities. Specifically, this can be described in detail in combination with the following specific implementation method.

[0106] In another specific implementation of the present disclosure, the above step 104 may specifically include:

[0107] Sub-step C1: according to the time differences between the multiple initial delivery times and the first delivery time, obtaining at least one initial delivery time whose time difference is within a preset time range.

[0108] In this embodiment, the preset time range refers to the range pre-set by the business personnel for filtering the given delivery time based on the time difference between the given delivery time and the actual delivery time. The specific value of the preset time range can be determined according to business needs, and this embodiment does not impose any restrictions on this.

[0109] In this example, the time differences between multiple initial delivery times and the first delivery time may be calculated first, and then, based on the time differences, at least one initial delivery time whose time difference is within a preset time range may be selected from the multiple initial delivery times.

[0110] After obtaining at least one initial delivery time whose time difference is within the preset range, sub-step C3 is executed.

[0111] Sub-step C2: obtaining the initial delivery time with the highest delivery probability among the at least one initial delivery time, and using the initial delivery time with the highest delivery probability as the second delivery time.

[0112] After obtaining at least one initial delivery time whose time difference is within a preset range, the initial delivery time with the highest delivery probability can be selected from the at least one initial delivery time and used as the second delivery time.

[0113] In this embodiment, the delivery time is screened by combining the target optimization function. By modifying the target function to be compatible with the future optimization target of ETA, the scalability of the entire solution is guaranteed. The target optimization function can be expressed as the following formula (1):

[0114]

[0115] In formula (1), F represents the cumulative probability distribution (CDF) of delivery time. The optimization objective is to ensure that the delivery time deviation falls within the interval [a, b] and the actual time. At the same time, the ETA value cannot be too large, taking into account the on-time performance parameters pre-set by the business personnel. During the decision-making process, the x value with the largest J(x) is selected as the ETA. λ and η are hyperparameters used to weight the various objectives. The objective function can be modified to accommodate future ETA optimization goals, ensuring the scalability of the entire solution.

[0116] In the embodiments of the present disclosure, by combining the objective function and the probability distribution of delivery time, ETA decisions are made with the ETA optimization goal as the guide, and the deviation of the estimated delivery time is optimized in a targeted manner. The ETA estimation is also performed while taking into account the punctuality parameters set by the business personnel. This can accurately characterize the delivery time deviation and improve the accuracy of the ETA estimation.

[0117] After the second delivery time is obtained, step 105 is executed.

[0118] Step 105: When the second delivery time matches the first delivery time, the trained initial time probability distribution model is used as the target time probability distribution model.

[0119] After obtaining the second delivery time, it can be determined whether the second delivery time matches the first delivery time.

[0120] If the second delivery time matches the first delivery time, the trained initial time probability distribution model is used as the target time probability distribution model. For example, the on-time rate parameter pre-set by the business personnel is 8 minutes (that is, the estimated delivery time deviates from the first delivery time by no more than 8 minutes). When the second delivery time differs from the first delivery time by no more than 8 minutes, it can be determined that the second delivery time matches the first delivery time.

[0121] The disclosed embodiment finely depicts the probability distribution of delivery time given different ETAs during the modeling process, thereby finely depicting the delivery time deviation and improving the accuracy of the estimated delivery time.

[0122] The time probability distribution model training method provided by the embodiment of the present disclosure obtains a training sample, the training sample is formed by the order information of the historical delivery order and the delivery personnel information of the delivery personnel associated with the historical delivery order, the training sample corresponds to a first delivery time, and multiple initial delivery times set for the training sample are obtained. The training sample and the multiple initial delivery times are input into the initial time probability distribution model, and the delivery probabilities corresponding to the multiple initial delivery times output by the initial time probability distribution model are obtained. According to the delivery probability, the second delivery time among the multiple initial delivery times is determined. When the second delivery time matches the first delivery time, the trained initial time probability distribution model is used as the target time probability distribution model. The embodiment of the present disclosure can finely characterize the delivery time deviation by finely characterizing the probability distribution of the delivery time under given different ETAs during the modeling process, thereby improving the accuracy of the estimated delivery time, improving the user and delivery personnel experience, and improving the overall efficiency of the system.

[0123] Example 2

[0124] Reference Figure 2 , which shows a flow chart of the steps of a method for obtaining delivery time provided by an embodiment of the present disclosure, such as Figure 2 As shown, the delivery time acquisition method may specifically include the following steps:

[0125] Step 201: After receiving a target order, obtain target order information associated with the target order and target delivery information of a target delivery person within a preset distance range of the target business party where the target order is located.

[0126] The embodiments of the present disclosure may be applied to a scenario in which a delivery time is assigned to a received target order.

[0127] The target order refers to the currently received order that requires an estimated delivery time. In this example, the target order can be a meal order, a medicine purchase order, etc. Specifically, it can be determined according to the actual situation, and this embodiment does not limit this.

[0128] The target order information refers to order information associated with the target order, wherein the target order information may include basic order information of the target order and business party information of the business party associated with the target order.

[0129] The target business party refers to the business party associated with the target order.

[0130] Target delivery information refers to the information of the target delivery personnel located within a preset distance range of the target business party, wherein the target delivery information may include delivery information of the orders held by the target delivery personnel, location information of the target delivery personnel, etc.

[0131] The specific process of obtaining the target order information and the target delivery information may be as shown in the above step 101, and will not be further described in this embodiment.

[0132] After receiving the target order, the target order information associated with the target order and the target delivery information of the target delivery personnel within the preset distance range of the target business party where the target order is located can be obtained, and then step 202 is executed.

[0133] Step 202: Acquire multiple third delivery times set for the target order.

[0134] The third delivery time refers to the delivery time set by the business personnel for the target order.

[0135] After receiving the target order, the business personnel may assign multiple delivery times to the target order, namely, the third delivery time.

[0136] After obtaining multiple third delivery times set for the target order, step 203 is executed.

[0137] Step 203: Input the target order information, the target delivery information and the multiple third delivery times into a target time probability distribution model, and obtain the target delivery probabilities corresponding to the multiple third delivery times output by the target time probability distribution model.

[0138] The target delivery probability refers to the delivery probability corresponding to the third delivery time output by the target time probability distribution model.

[0139] After obtaining the target order information, target delivery information and multiple third delivery times, the target order information, target delivery information and multiple third delivery times can be input into the target time probability distribution model, and the target delivery probabilities corresponding to the multiple third delivery times output by the target time probability distribution model can be obtained. The specific implementation method can be as described in step 103 in the above embodiment 1, and this embodiment will not be repeated here.

[0140] After obtaining the target delivery probabilities corresponding to the plurality of third delivery times, step 204 is executed.

[0141] Step 204: According to the target delivery probability, a target delivery time corresponding to the target order is obtained from the plurality of third delivery times.

[0142] The target delivery time is the time you want to allocate for the target order.

[0143] After obtaining the target delivery probabilities corresponding to multiple third delivery times, the target delivery time corresponding to the target order can be filtered out from the multiple third delivery times according to the target delivery probabilities. Specifically, it can be described in detail in combination with the following specific implementation method.

[0144] In another specific implementation of the present disclosure, the above step 204 may include:

[0145] Sub-step D1: Obtain at least one fourth delivery time within a preset time range from the plurality of third delivery times.

[0146] In this embodiment, the fourth delivery time refers to a delivery time that is selected from the plurality of third delivery times and is within a preset time range.

[0147] In this example, at least one fourth delivery time within a preset time range may be selected from the plurality of third delivery times, and then sub-step D2 may be performed.

[0148] Sub-step D2: According to the target delivery probability, select the delivery time with the highest delivery probability from the at least one fourth delivery time, and use the delivery time with the highest delivery probability as the target delivery time.

[0149] After the fourth delivery time is screened out, a delivery time with the highest delivery probability can be screened out from at least one fourth delivery time according to the target delivery probability, and the delivery time with the highest delivery probability can be used as the target delivery time of the target order.

[0150] In this embodiment, in tail-end scenarios, the uncertainty of delivery time is high. Current ETA architectures use regular time additions to ensure on-time delivery. The delivery time probability distribution estimation model in this solution provides more detailed information to characterize this uncertainty, enabling better decision-making in tail-end scenarios. For each tail-end scenario, the corresponding confidence quantile and decision parameters can be searched and determined, fully leveraging the model's predictions and reducing the use of regular time additions, resulting in greater accuracy and flexibility.

[0151] The delivery time acquisition method provided by the embodiment of the present disclosure obtains the target order information associated with the target order and the target delivery information of the target delivery personnel within a preset distance range of the target business party where the target order is located after receiving the target order, obtains multiple third delivery times set for the target order, inputs the target order information, target delivery information and multiple third delivery times into the target time probability distribution model, and obtains the target delivery probabilities corresponding to the multiple third delivery times output by the target time probability distribution model, and selects the target delivery time corresponding to the target order from the multiple third delivery times based on the target delivery probabilities. The embodiment of the present disclosure can finely characterize the delivery time deviation by finely characterizing the probability distribution of delivery time under given different ETAs during the modeling process, thereby improving the accuracy of the estimated delivery time, improving the user and delivery personnel experience, and improving the overall efficiency of the system.

[0152] Example 3

[0153] Reference Figure 3 , which shows a schematic structural diagram of a time probability distribution model training device provided by an embodiment of the present disclosure, such as Figure 3 As shown, the time probability distribution model training device may specifically include the following modules:

[0154] A training sample acquisition module 310 is configured to acquire a training sample; the training sample is formed by order information of a historical delivery order and delivery personnel information of a delivery personnel associated with the historical delivery order, and the training sample corresponds to a first delivery time;

[0155] An initial time acquisition module 320 is used to acquire multiple initial delivery times set for the training samples;

[0156] A delivery probability acquisition module 330 is configured to input the training sample and the multiple initial delivery times into an initial time probability distribution model, and obtain delivery probabilities corresponding to the multiple initial delivery times output by the initial time probability distribution model;

[0157] A second time determination module 340 is configured to determine a second delivery time among the multiple initial delivery times based on the delivery probability;

[0158] The target model acquisition module 350 is configured to use the trained initial time probability distribution model as a target time probability distribution model when the second delivery time matches the first delivery time.

[0159] Optionally, the training sample acquisition module 310 includes:

[0160] An order information acquisition unit, configured to acquire basic order information of the historical delivery order and business party information of the business party associated with the historical delivery order, and use the basic order information and the business party information as the order information;

[0161] a delivery personnel information acquisition unit, configured to acquire delivery personnel information of delivery personnel within a preset distance range from the business party's location;

[0162] Among them, the basic order information includes the order time, the time period of the order time, the delivery distance, the order price and the location information of the recipient; the business party information includes the business party location information and the order time information; the delivery personnel information includes the location information of the delivery personnel, the delivery information of the order held by the delivery personnel, the delivery time information and the delivery capacity information.

[0163] Optionally, the delivery probability acquisition module 330 includes:

[0164] a training sample input unit, configured to input the training sample and the plurality of initial delivery times into the initial time probability distribution model;

[0165] a coding vector determining unit, configured to determine, based on the initial time probability distribution model, an order coding vector corresponding to the order information and a delivery personnel coding vector corresponding to the delivery personnel information;

[0166] The delivery probability prediction unit is used to predict the delivery probabilities corresponding to the multiple initial delivery times based on the initial time probability distribution model according to the order coding vector and the delivery personnel coding vector.

[0167] Optionally, the second time determination module 340 includes:

[0168] A first time acquisition unit is configured to acquire, based on the time differences between the multiple initial delivery times and the first delivery time, at least one initial delivery time whose time difference is within a first preset time range;

[0169] The second time acquisition unit is configured to acquire an initial delivery time with the highest delivery probability from the at least one initial delivery time, and use the initial delivery time with the highest delivery probability as the second delivery time.

[0170] The time probability distribution model training device provided by the embodiment of the present disclosure obtains training samples, the training samples are formed by the order information of historical delivery orders and the delivery personnel information of the delivery personnel associated with the historical delivery orders, the training samples correspond to a first delivery time, multiple initial delivery times set for the training samples are obtained, the training samples and the multiple initial delivery times are input into the initial time probability distribution model, and the delivery probabilities corresponding to the multiple initial delivery times output by the initial time probability distribution model are obtained, and the second delivery time among the multiple initial delivery times is determined according to the delivery probability, and when the second delivery time matches the first delivery time, the trained initial time probability distribution model is used as the target time probability distribution model. The embodiment of the present disclosure can finely characterize the delivery time deviation by finely characterizing the probability distribution of delivery time under given different ETAs during the modeling process, thereby improving the accuracy of the estimated delivery time, improving the user and delivery personnel experience, and improving the overall efficiency of the system.

[0171] Example 4

[0172] Reference Figure 4 , which shows a schematic structural diagram of a delivery time acquisition device provided by an embodiment of the present disclosure, such as Figure 4 As shown, the delivery time acquisition method may specifically include the following modules:

[0173] The target order information acquisition module 410 is configured to, after receiving a target order, acquire target order information associated with the target order and target delivery information of a target delivery person within a preset distance range of the target business party of the target order;

[0174] A third delivery time acquisition module 420 is configured to acquire a plurality of third delivery times set for the target order;

[0175] a target delivery probability acquisition module 430 for inputting the target order information, the target delivery information, and the plurality of third delivery times into a target time probability distribution model, and acquiring target delivery probabilities corresponding to the plurality of third delivery times output by the target time probability distribution model;

[0176] The target delivery time screening module 440 is configured to screen the target delivery time corresponding to the target order from the plurality of third delivery times according to the target delivery probability.

[0177] Optionally, the target delivery time screening module 440 includes:

[0178] a fourth delivery time acquiring unit, configured to acquire at least one fourth delivery time within a preset time range from among the plurality of third delivery times;

[0179] The target delivery time acquisition unit is configured to select a delivery time with the highest delivery probability from the at least one fourth delivery time according to the target delivery probability, and use the delivery time with the highest delivery probability as the target delivery time.

[0180] The delivery time acquisition device provided by the embodiment of the present disclosure obtains the target order information associated with the target order and the target delivery information of the target delivery personnel within a preset distance range of the target business party where the target order is located after receiving the target order, obtains multiple third delivery times set for the target order, inputs the target order information, target delivery information and multiple third delivery times into the target time probability distribution model, and obtains the target delivery probabilities corresponding to the multiple third delivery times output by the target time probability distribution model, and filters the target delivery time corresponding to the target order from the multiple third delivery times based on the target delivery probabilities. The embodiment of the present disclosure can finely characterize the delivery time deviation by finely characterizing the probability distribution of delivery time under given different ETAs during the modeling process, thereby improving the accuracy of the estimated delivery time, improving the user and delivery personnel experience, and improving the overall efficiency of the system.

[0181] An embodiment of the present disclosure also provides an electronic device, comprising: a processor, a memory, and a computer program stored on the memory and executable on the processor, wherein when the processor executes the program, the time probability distribution model training method of the aforementioned embodiment or the delivery time acquisition method of the aforementioned embodiment is implemented.

[0182] An embodiment of the present disclosure also provides a readable storage medium, which, when the instructions in the storage medium are executed by a processor of an electronic device, enables the electronic device to execute the time probability distribution model training method of the aforementioned embodiment, or the delivery time acquisition method of the aforementioned embodiment.

[0183] As for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0184] The algorithms and displays provided herein are not inherently related to any particular computer, virtual system, or other device. Various general-purpose systems may also be used together with the teachings herein. Based on the above description, it is apparent that the structure required for constructing such systems is suitable. In addition, the embodiments of the present disclosure are not directed to any specific programming language. It should be understood that various programming languages may be utilized to implement the contents of the embodiments of the present disclosure described herein, and the above description of specific languages is intended to disclose the best mode of implementation of the embodiments of the present disclosure.

[0185] In the description provided herein, numerous specific details are described. However, it is understood that embodiments of the present disclosure may be practiced without these specific details. In some instances, well-known methods, structures, and techniques are not shown in detail so as not to obscure the understanding of this description.

[0186] Similarly, it should be understood that in order to streamline the present disclosure and aid in understanding one or more of the various inventive aspects, in the above description of exemplary embodiments of the present disclosure, various features of the embodiments of the present disclosure are sometimes grouped together into a single embodiment, figure, or description thereof. However, this disclosed approach should not be interpreted as reflecting an intention that the claimed embodiments of the present disclosure require more features than are expressly recited in each claim. Rather, as reflected in the claims below, inventive aspects lie in less than all of the features of the individual embodiments disclosed above. Accordingly, the claims that follow the detailed description are hereby expressly incorporated into the detailed description, with each claim itself serving as a separate embodiment of the embodiments of the present disclosure.

[0187] Those skilled in the art will appreciate that the modules in the devices of the embodiments may be adaptively changed and arranged in one or more devices different from the embodiments. The modules or units or components in the embodiments may be combined into one module or unit or component, and further may be divided into a plurality of submodules or subunits or subcomponents. All features disclosed in this specification (including the accompanying claims, abstracts and drawings) and all processes or units of any method or device so disclosed may be combined in any combination, except that at least some of such features and / or processes or units are mutually exclusive. Unless expressly stated otherwise, each feature disclosed in this specification (including the accompanying claims, abstracts and drawings) may be replaced by an alternative feature providing the same, equivalent or similar purpose.

[0188] The various component embodiments of the embodiments of the present disclosure may be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof. Those skilled in the art will appreciate that a microprocessor or digital signal processor (DSP) may be used in practice to implement some or all of the functions of some or all of the components of the device for generating dynamic images according to the embodiments of the present disclosure. The embodiments of the present disclosure may also be implemented as a device or apparatus program for executing part or all of the methods described herein. Such a program implementing the embodiments of the present disclosure may be stored on a computer-readable medium, or may be in the form of one or more signals. Such a signal may be downloaded from an Internet website, or provided on a carrier signal, or provided in any other form.

[0189] It should be noted that the above embodiments illustrate rather than limit the embodiments of the present disclosure, and that those skilled in the art may devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference symbols placed between brackets should not be construed as limiting the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The embodiments of the present disclosure may be implemented by means of hardware comprising several different elements and by means of a suitably programmed computer. In a unit claim enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second, and third etc. does not indicate any order. These words may be interpreted as names.

[0190] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0191] The above description is merely a preferred embodiment of the embodiments of the present disclosure and is not intended to limit the embodiments of the present disclosure. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the embodiments of the present disclosure shall be included in the protection scope of the embodiments of the present disclosure.

[0192] The above description is merely a specific implementation of the embodiments of the present disclosure, but the scope of protection of the embodiments of the present disclosure is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the embodiments of the present disclosure should be included in the scope of protection of the embodiments of the present disclosure. Therefore, the scope of protection of the embodiments of the present disclosure should be based on the scope of protection of the claims.

Claims

1. A time probability distribution model training method, characterized in that: include: Get training samples; The training sample is formed by order information of historical delivery orders and delivery personnel information of delivery personnel associated with the historical delivery orders, and the training sample corresponds to a first delivery time; the first delivery time refers to the actual delivery time of the historical delivery order; Acquire multiple initial delivery times set for the training samples; the initial delivery time refers to the delivery time pre-assigned by business personnel to each training sample; Inputting the training sample and the multiple initial delivery times into an initial time probability distribution model, and obtaining delivery probabilities corresponding to the multiple initial delivery times output by the initial time probability distribution model; Determining a second delivery time from the multiple initial delivery times based on the delivery probability; the second delivery time is a time selected from the multiple initial delivery times that satisfies the delivery probability prediction; When the second delivery time matches the first delivery time, the trained initial time probability distribution model is used as the target time probability distribution model; the second delivery time matches the first delivery time means that the difference between the second delivery time and the first delivery time does not exceed the punctuality parameter pre-set by the business personnel.

2. The method according to claim 1, characterized in that The obtaining of training samples includes: Obtaining basic order information of the historical delivery order and business party information of the business party associated with the historical delivery order, and using the basic order information and the business party information as the order information; Obtaining delivery personnel information of delivery personnel within a preset distance range from the business party's location; Among them, the basic order information includes the order time, the time period of the order time, the delivery distance, the order price and the location information of the recipient; the business party information includes the business party location information and the order time information; the delivery personnel information includes the location information of the delivery personnel, the delivery information of the order held by the delivery personnel, the delivery time information and the delivery capacity information.

3. The method according to claim 1, characterized in that Inputting the training sample and the multiple initial delivery times into an initial time probability distribution model, and obtaining delivery probabilities corresponding to the multiple initial delivery times output by the initial time probability distribution model, includes: inputting the training sample and the plurality of initial delivery times into the initial time probability distribution model; Determining, based on the initial time probability distribution model, an order coding vector corresponding to the order information and a delivery personnel coding vector corresponding to the delivery personnel information; Based on the initial time probability distribution model and the order coding vector and the delivery personnel coding vector, the delivery probabilities corresponding to the multiple initial delivery times are predicted.

4. The method according to claim 1, wherein Determining a second delivery time among the multiple initial delivery times according to the delivery probability includes: According to the time differences between the multiple initial delivery times and the first delivery time, obtaining at least one initial delivery time whose time difference is within a preset time range; An initial delivery time with the highest delivery probability among the at least one initial delivery time is obtained, and the initial delivery time with the highest delivery probability is used as the second delivery time.

5. A method for obtaining delivery time, characterized in that: include: After receiving the target order, obtaining target order information associated with the target order and target delivery information of target delivery personnel within a preset distance range of the target business party where the target order is located; Obtain multiple third delivery times set for the target order; Inputting the target order information, the target delivery information, and the plurality of third delivery times into a target time probability distribution model, and obtaining target delivery probabilities corresponding to the plurality of third delivery times output by the target time probability distribution model; Filtering the target delivery time corresponding to the target order from the plurality of third delivery times according to the target delivery probability; Wherein, the target time probability distribution model is obtained by training using the time probability distribution model training method according to any one of claims 1 to 4.

6. The method according to claim 5, characterized in that The step of filtering, based on the target delivery probability, from the plurality of third delivery times to obtain a target delivery time corresponding to the target order includes: Obtaining at least one fourth delivery time within a preset time range from the plurality of third delivery times; According to the target delivery probability, a delivery time with the highest delivery probability is selected from the at least one fourth delivery time, and the delivery time with the highest delivery probability is used as the target delivery time.

7. A time probability distribution model training device, characterized in that: include: A training sample acquisition module is used to obtain training samples; The training sample is formed by order information of historical delivery orders and delivery personnel information of delivery personnel associated with the historical delivery orders, and the training sample corresponds to a first delivery time; the first delivery time refers to the actual delivery time of the historical delivery order; An initial time acquisition module, configured to acquire a plurality of initial delivery times set for the training samples; the initial delivery time refers to the delivery time pre-assigned by the business personnel to each training sample; a delivery probability acquisition module, configured to input the training sample and the multiple initial delivery times into an initial time probability distribution model, and obtain delivery probabilities corresponding to the multiple initial delivery times output by the initial time probability distribution model; A second time determination module is configured to determine a second delivery time from the plurality of initial delivery times based on the delivery probability; the second delivery time is a time selected from the plurality of initial delivery times that satisfies the delivery probability prediction; The target model acquisition module is used to use the trained initial time probability distribution model as the target time probability distribution model when the second delivery time matches the first delivery time; the second delivery time matches the first delivery time means that the difference between the second delivery time and the first delivery time does not exceed the punctuality parameter preset by the business personnel.

8. A delivery time acquisition device, characterized in that: include: A target order information acquisition module is used to, after receiving a target order, acquire target order information associated with the target order and target delivery information of a target delivery person within a preset distance range of a target business party where the target order is located; A third delivery time acquisition module, configured to acquire a plurality of third delivery times set for the target order; a target delivery probability acquisition module, configured to input the target order information, the target delivery information, and the plurality of third delivery times into a target time probability distribution model, and acquire target delivery probabilities corresponding to the plurality of third delivery times output by the target time probability distribution model; a target delivery time screening module, configured to screen, based on the target delivery probability, the target delivery time corresponding to the target order from the plurality of third delivery times; Wherein, the target time probability distribution model is obtained by training through the time probability distribution model training device described in claim 7.

9. An electronic device, characterized in that: include: A processor, a memory, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method for training a time probability distribution model according to any one of claims 1 to 4 or the method for obtaining delivery time according to any one of claims 5 to 6 is implemented.

10. A readable storage medium, characterized in that: When the instructions in the storage medium are executed by the processor of the electronic device, the electronic device is enabled to execute the time probability distribution model training method as described in any one of method claims 1 to 4, or the delivery time acquisition method as described in any one of claims 5 to 6.

Images