Method and apparatus for sharing item delivery

Abstract

The disclosure provides a shared object delivery method and device, and relates to the technical field of smart cities. The method comprises the following steps: obtaining a plurality of historical delivery strategies of a shared object in a delivery range and shared records of the shared object under each historical delivery strategy, determining a delivery effect evaluation value of each historical delivery strategy according to the number of shared failures in the shared records, training a value network according to each historical delivery strategy and the delivery effect evaluation value, and then predicting the delivery effect evaluation value of each candidate delivery strategy of the shared object in the delivery range by using the trained value network, and finally delivering the shared object according to the target delivery strategy determined according to the evaluation value. Therefore, the delivery of the shared object is more reasonable, the utilization rate of the shared object and the user experience are improved, and the problem of imbalance between supply and demand of the shared object in different regions and time periods is solved.

Description

Technical Field

[0001] This disclosure relates to the field of smart city technology, and more particularly to the field of deep reinforcement learning technology, specifically to a method and apparatus for deploying shared items. Background Technology

[0002] While shared goods offer great convenience to users, they also present the problem of uneven supply and demand across different areas and time periods. Taking shared bicycles as an example, the significant differences in demand across areas and time periods lead to situations where some areas experience bicycle oversupply and congestion, while others suffer from insufficient supply. This not only results in low bicycle utilization but also severely impacts the user experience. Therefore, how to optimize the deployment of shared goods and improve their utilization and user experience is a problem worthy of further research. Summary of the Invention

[0003] This disclosure aims to at least partially address one of the technical problems in the related art.

[0004] Therefore, the first objective of this disclosure is to propose a method for deploying shared items, which trains a value network based on multiple historical deployment strategies for shared items within the deployment range and the sharing records of shared items under each historical deployment strategy. The trained value network is then used to predict the evaluation value of the deployment effect of each candidate deployment strategy for shared items within the deployment range, so as to determine the target deployment strategy. Based on the target deployment strategy, shared items are deployed, making the deployment of shared items more reasonable and improving the utilization rate of shared items and user experience.

[0005] The second objective of this disclosure is to propose a device for distributing shared goods.

[0006] The third objective of this disclosure is to provide an electronic device.

[0007] The fourth objective of this disclosure is to provide a non-transitory computer-readable storage medium.

[0008] The fifth objective of this disclosure is to provide a computer program product.

[0009] To achieve the above objectives, a first aspect of this disclosure provides a method for deploying shared items, comprising:

[0010] Obtain multiple historical deployment strategies for the shared items within the deployment range, as well as the sharing records of the shared items under each historical deployment strategy;

[0011] For any given historical delivery strategy, the delivery performance evaluation value of the corresponding historical delivery strategy is determined based on the number of sharing failures in the shared records.

[0012] Based on the historical delivery strategies and their performance evaluation values, a value network is trained.

[0013] A trained value network is used to predict the evaluation value of the delivery effect of each candidate delivery strategy for the shared items within the delivery range, so as to determine the target delivery strategy based on the evaluation value.

[0014] The shared items are deployed according to the target deployment strategy.

[0015] Optionally, as a first possible implementation of the first aspect, the shared record includes query records and borrowing records; the step of determining the delivery effect evaluation value of any historical delivery strategy based on the number of sharing failures in the shared record includes:

[0016] For any given historical delivery strategy, determine the query operations recorded within the statistical period following the execution of the corresponding historical delivery strategy from the query records.

[0017] If the client that performs the query operation does not perform the lending operation after the query operation, or if the lending record determines that the client failed to perform the lending operation after the query operation, then the query operation is determined to be a lending failure, so as to determine the number of lending failures.

[0018] Based on the lending records, query the return operations recorded within the statistical period after the execution of the corresponding historical delivery strategy to determine the number of return failures;

[0019] The evaluation value of the delivery effect of the corresponding historical delivery strategy is determined based on the number of loan failures and / or the number of return failures.

[0020] Optionally, as a second possible implementation of the first aspect, the step of querying the return operations recorded within the statistical period after the execution of the corresponding historical delivery strategy, based on the lending records, to determine the number of return failures, includes:

[0021] Based on the lending records, query the return operations recorded within the statistical period after the execution of the corresponding historical delivery strategy;

[0022] For the returned operation found, determine the partition from the multiple partitions divided by the delivery range to execute the corresponding returned operation;

[0023] If the number of shared items in the corresponding partition exceeds the quantity threshold after the return operation is performed, the return operation will be determined as a return failure.

[0024] Determine the number of failed returns from the returned operations retrieved.

[0025] Optionally, as a third possible implementation of the first aspect, the step of employing a trained value network to predict the evaluation value of the delivery effect of each candidate delivery strategy for the shared item within the delivery range, and determining the target delivery strategy based on the evaluation value, includes:

[0026] Based on the initial deployment volume of the shared items in each of the multiple partitions divided by the deployment range, an initial candidate deployment strategy is determined.

[0027] A trained value network is used to predict the initial candidate delivery strategy, and the delivery effect evaluation value corresponding to the initial candidate delivery strategy is used as a reference evaluation value.

[0028] Multiple rounds of strategy updates are performed based on the initial candidate delivery strategy;

[0029] Each time a strategy update is performed, the trained value network is used to predict the delivery effect of at least one candidate delivery strategy after the update.

[0030] If the evaluation value of the delivery performance of at least one candidate delivery strategy after this round of updates is less than the reference evaluation value, then the evaluation value of the delivery performance of at least one candidate delivery strategy after this round of updates will be used as the reference evaluation value, and the subsequent strategy update process will continue.

[0031] After each round of strategy updates is completed, the candidate delivery strategy corresponding to the reference evaluation value is taken as the target delivery strategy.

[0032] Optionally, as a fourth possible implementation of the first aspect, the step of performing multiple rounds of policy updates based on the initial candidate delivery strategy includes:

[0033] For each round of policy updates, a target partition is determined from multiple partitions;

[0034] The distribution volume of shared items is increased and decreased by a set step size for the target partition to obtain two candidate distribution strategies obtained in this round of strategy update.

[0035] Optionally, as a fifth possible implementation of the first aspect, determining a target partition from multiple partitions for each round of policy update includes:

[0036] For each round of strategy update, if the number of deliveries to the target partition used in the previous round is lower than the minimum or higher than the maximum, then the next partition adjacent to the target partition used in the previous round will be used as the target partition for this round.

[0037] If the number of targets deployed in the previous round was not lower than the lower limit and not higher than the upper limit, then the target partitions deployed in the previous round will be used as the target partitions for this round.

[0038] The shared item deployment method provided in this disclosure acquires multiple historical deployment strategies for shared items within the deployment range, as well as sharing records for each historical deployment strategy. For any given historical deployment strategy, the method determines the deployment effectiveness evaluation value based on the number of sharing failures in the sharing records. Then, based on each historical deployment strategy and its evaluation value, a value network is trained. This trained value network then predicts the deployment effectiveness evaluation value for each candidate deployment strategy within the deployment range. After determining the target deployment strategy based on the evaluation value, the shared item is deployed according to the target deployment strategy. This allows for more rational deployment of shared items, improves utilization and user experience, and addresses the problem of uneven supply and demand of shared items in different regions and time periods.

[0039] To achieve the above objectives, a second aspect of this disclosure provides a device for distributing shared items, comprising:

[0040] The acquisition module is used to acquire multiple historical deployment strategies of the shared items within the deployment range, as well as the sharing records of the shared items under each historical deployment strategy;

[0041] The determination module is used to determine the evaluation value of the delivery effect of any historical delivery strategy based on the number of sharing failures in the shared records.

[0042] The training module is used to train the value network based on the historical delivery strategies and their performance evaluation values.

[0043] The prediction module is used to use a trained value network to predict the evaluation value of the delivery effect of each candidate delivery strategy for the shared items within the delivery range, so as to determine the target delivery strategy based on the evaluation value.

[0044] The delivery module is used to deliver the shared items according to the target delivery strategy.

[0045] Optionally, as a first possible implementation of the second aspect, the shared records include query records and borrowing records; the determining module includes:

[0046] The first determining unit is used to determine, from the query records, the query operations recorded within the statistical period after the execution of the corresponding historical delivery strategy, for any given historical delivery strategy.

[0047] The second determining unit is configured to determine the query operation as a failed loan operation if the client executing the query operation does not perform a loan operation after the query operation, or if the loan record determines that the client failed to perform a loan operation after the query operation, in order to determine the number of failed loan operations.

[0048] The query unit is used to query the return operations recorded within the statistical period after the execution of the corresponding historical delivery strategy, based on the borrowing records, in order to determine the number of return failures.

[0049] The third determining unit is used to determine the evaluation value of the delivery effect of the corresponding historical delivery strategy based on the number of loan failures and / or the number of return failures.

[0050] Alternatively, as a second possible implementation of the second aspect, the query unit is further configured to:

[0051] Based on the lending records, query the return operations recorded within the statistical period after the execution of the corresponding historical delivery strategy;

[0052] For the returned operation found, determine the partition from the multiple partitions divided by the delivery range to execute the corresponding returned operation;

[0053] If the number of shared items in the corresponding partition exceeds the quantity threshold after the return operation is performed, the return operation will be determined as a return failure.

[0054] Determine the number of failed returns from the returned operations retrieved.

[0055] Alternatively, as a third possible implementation of the second aspect, the prediction module includes:

[0056] The fourth determining unit is used to determine an initial candidate deployment strategy based on the initial deployment quantity of the shared items in each of the multiple partitions divided by the deployment range.

[0057] The first prediction unit is used to predict the initial candidate delivery strategy using a trained value network, so as to use the delivery effect evaluation value corresponding to the initial candidate delivery strategy as a reference evaluation value.

[0058] The first processing unit is used to perform multiple rounds of policy updates based on the initial candidate delivery strategy;

[0059] The second prediction unit is used to predict the delivery effect of at least one candidate delivery strategy after the current round of strategy update using the trained value network whenever a round of strategy update is performed.

[0060] The second processing unit is used to take the evaluation value of the delivery effect of at least one candidate delivery strategy after the current round of updates as the reference evaluation value and continue to execute the subsequent strategy update process if the evaluation value of the delivery effect of at least one candidate delivery strategy after the current round of updates is less than the reference evaluation value.

[0061] The third processing unit is used to execute the strategy update for each round and use the candidate delivery strategy corresponding to the reference evaluation value as the target delivery strategy.

[0062] Optionally, as a fourth possible implementation of the second aspect, the first processing unit is further configured to:

[0063] For each round of policy updates, a target partition is determined from multiple partitions;

[0064] The distribution volume of shared items is increased and decreased by a set step size for the target partition to obtain two candidate distribution strategies obtained in this round of strategy update.

[0065] Optionally, as a fifth possible implementation of the second aspect, the first processing unit is further configured to:

[0066] For each round of strategy update, if the number of deliveries to the target partition used in the previous round is lower than the minimum or higher than the maximum, then the next partition adjacent to the target partition used in the previous round will be used as the target partition for this round.

[0067] If the number of targets deployed in the previous round was not lower than the lower limit and not higher than the upper limit, then the target partitions deployed in the previous round will be used as the target partitions for this round.

[0068] The shared item delivery device provided in this embodiment acquires multiple historical delivery strategies for shared items within the delivery range, as well as sharing records for each historical delivery strategy. For any given historical delivery strategy, it determines the delivery effectiveness evaluation value based on the number of sharing failures in the sharing records. Then, based on each historical delivery strategy and its evaluation value, it trains a value network. This trained value network then predicts the delivery effectiveness evaluation value for each candidate delivery strategy within the delivery range. After determining the target delivery strategy based on the evaluation value, it delivers the shared items according to the target delivery strategy. This allows for more rational delivery of shared items, improves utilization and user experience, and addresses the problem of uneven supply and demand of shared items in different areas and time periods.

[0069] To achieve the above objectives, a third aspect of this disclosure provides an electronic device comprising:

[0070] At least one processor; and

[0071] A memory communicatively connected to the at least one processor; wherein,

[0072] The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method described in the first aspect.

[0073] To achieve the above objectives, a fourth aspect of this disclosure provides a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause the computer to perform the method described in the first aspect.

[0074] To achieve the above objectives, a fifth aspect of this disclosure provides a computer program product including a computer program that, when executed by a processor, implements the method described in the first aspect.

[0075] Additional aspects and advantages of this disclosure will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this disclosure. Attached Figure Description

[0076] The above and / or additional aspects and advantages of this disclosure will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, in which:

[0077] Figure 1 A flowchart illustrating a method for deploying shared items provided in an embodiment of this disclosure;

[0078] Figure 2 This is a schematic diagram of a gridded delivery range processing provided in an embodiment of the present disclosure;

[0079] Figure 3 This is a schematic diagram of a value network structure provided in an embodiment of the present disclosure;

[0080] Figure 4 This is a flowchart illustrating another method for deploying shared items provided in an embodiment of this disclosure;

[0081] Figure 5 This is a schematic diagram of a process for determining the number of times a return has failed, provided as an embodiment of the present disclosure.

[0082] Figure 6 This is a flowchart illustrating another method for deploying shared items provided in an embodiment of this disclosure;

[0083] Figure 7 This is a schematic diagram illustrating a process for performing multiple rounds of policy updates based on an initial candidate delivery strategy, as provided in an embodiment of this disclosure.

[0084] Figure 8 This is a schematic diagram of a process for determining a target partition from multiple partitions, provided as an embodiment of the present disclosure.

[0085] Figure 9 This is a schematic diagram illustrating the principle of a method for deploying shared items in a given scenario, as provided in an embodiment of this disclosure.

[0086] Figure 10 A schematic diagram of the structure of a shared item delivery device provided in this disclosure embodiment; and

[0087] Figure 11 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this disclosure. Detailed Implementation

[0088] Embodiments of this disclosure are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this disclosure, and should not be construed as limiting this disclosure.

[0089] The following describes a method and apparatus for deploying shared items according to embodiments of the present disclosure, with reference to the accompanying drawings.

[0090] Figure 1 This is a flowchart illustrating a method for deploying shared items provided in an embodiment of this disclosure.

[0091] While shared goods offer great convenience to users, they also present the problem of uneven supply and demand across different areas and time periods. Take shared bicycles as an example: while they provide significant convenience for urban residents, the vast differences in demand across different areas often lead to an oversupply and bicycle accumulation in some areas, while other areas experience a shortage. This not only results in low bicycle utilization but also significantly impacts the user experience. To improve bicycle utilization and enhance the user experience, bicycle operators periodically redistribute bicycles within the city. Existing bicycle redistribution strategies primarily rely on human experience or algorithms, such as algorithms that deploy bicycles based on the proportion of bicycle usage in each area relative to the total citywide usage, and algorithms that use optimization models to calculate bicycle redistribution strategies.

[0092] The algorithm for deploying bicycles based on the proportion of bicycle usage in each area to the total number of bicycles in the city is based on historical bicycle usage data. It calculates the expected net bicycle lending volume for each area in the future. If a region has a high net bicycle lending volume, more bicycles are deployed in that region according to the designed proportion calculation formula; conversely, fewer bicycles are deployed in that region. For regions with a negative net bicycle lending volume, no bicycles are even deployed. However, this algorithm only considers the expected net bicycle lending volume for each region in the future and ignores the actual bicycle lending and returning situation. The algorithm that uses an optimization model to calculate the bicycle redistribution strategy is based on historical bicycle usage data. It estimates how many bicycles need to be deployed in each region to ensure that the region will not experience a shortage of bicycles or an excessive number of bicycles causing congestion for as long as possible. The number of bicycles that meet the above conditions for each region is the bicycle redistribution target. However, this algorithm treats each region as independent and considers them separately, ignoring the correlation between regions. For example, if a large number of bikes borrowed from a certain region are frequently returned to a certain destination region, then insufficient bike deployment in that region will inevitably lead to a significant reduction in the number of bikes returned in the corresponding destination region in the future, thus affecting the number of bikes in that destination region in the future. Moreover, using an optimization model to simultaneously optimize bike deployment in each region results in a problem that is difficult to define, has an excessively large solution space, and high solution complexity.

[0093] To address the aforementioned issues, this disclosure provides a method for deploying shared items. This method trains a value network based on multiple historical deployment strategies for shared items within the deployment range and the sharing records of shared items under each historical deployment strategy. The trained value network then predicts the deployment effect evaluation value for each candidate deployment strategy within the deployment range to determine the target deployment strategy. Based on the target deployment strategy, shared items are deployed, making the deployment of shared items more rational, improving the utilization rate of shared items and the user experience. Figure 1 As shown, the method for deploying this shared item includes the following steps:

[0094] Step 101: Obtain multiple historical deployment strategies for shared items within the deployment range, as well as the sharing records of shared items under each historical deployment strategy.

[0095] It should be noted that the method for distributing shared items provided in this embodiment can be executed by a device for distributing shared items. This device can be an electronic device or can be configured within an electronic device. The electronic device can be any stationary or mobile computing device capable of data processing, such as a laptop, smartphone, wearable device, or other mobile computing device; or a desktop computer or other stationary computing device; or a server; or other types of computing devices. This embodiment does not impose any limitations on this.

[0096] In this embodiment, the shared items can be any item whose usage needs can be staggered and can be flexibly switched between different users, such as shared bicycles, shared electric vehicles, shared cars, and shared power banks. The deployment area can be any region, and the shape of the region can be regular or irregular; this embodiment does not impose any restrictions on this. The sharing record can include at least one of the following: user query records for shared item locations, user borrowing records for shared items, and user returning records for shared items. It is understood that when a user needs to borrow a shared item, they will open the corresponding APP (Application) to query the locations of shared items available for borrowing near their location, thus creating a query record. When the user arrives at the location of the shared item and borrows it, a corresponding borrowing record will be created; and when the user returns the item after use, a corresponding returning record will be created.

[0097] It should be noted that although this embodiment does not limit the shape of the delivery range, the delivery range can be gridded to facilitate the determination of the delivery effect evaluation value of the corresponding historical delivery strategy in subsequent operations. As a possible implementation, a four-element array {id, s_time, s_lng, s_lat} can be generated for query records, where id represents the user ID registered in the APP, s_time represents the time the user opened the APP to perform the query, s_lng represents the longitude of the user when performing the query, and s_lat represents the latitude of the user when performing the query. Similarly, a four-element array {id, b_time, b_lng, b_lat} can be generated for lending records, where id also represents the user ID registered in the APP, b_time represents the time the user lent out the item, b_lng represents the longitude of the user when performing the lending out item, and b_lat represents the latitude of the user when performing the lending out item. Similarly, a four-element array {id, r_time, r_lng, r_lat} can be generated for each return record. Here, id represents the user ID registered in the app, r_time represents the time the user performed the return operation, r_lng represents the longitude of the user's return operation, and r_lat represents the latitude of the user's lending operation. Based on the lending and return records, multiple historical lending location data can be obtained: (b_lng1, b_lat1), (b_lng2, b_lat2), ..., (b_lng...). n ,b_lat nand multiple historical return location data (r_lng1, r_lat1), (r_lng2, r_lat2), ... (r_lng n ,r_lat n This allows us to determine the coverage area of ​​the shared items, i.e., the deployment area. For example... Figure 2 As shown, the delivery area is usually an irregular shape. Therefore, a regular rectangle can be used to circumscribe this delivery area, resulting in a circumscribed rectangle. The latitude and longitude of the four vertices of the circumscribed rectangle are: (lng...) min ,lat max ), (lng min ,lat min ), (lng max ,lat min ), (lng max ,lat max ),in:

[0098] lng min =min{b_lng1,b_lng2,…b_lng n ,r_lng1,r_lng2…r_lng n};

[0099] lng max =max{b_lng1,b_lng2,…b_lng n ,r_lng1,r_lng2…r_lng n};

[0100] lat min =min{b_lat1,b_lat2,…b_lat n ,r_lat1,r_lat2…r_lat n};

[0101] lat max =max{b_lat1,b_lat2,…b_lat n ,r_lat1,r_lat2…r_lat n}

[0102] The circumscribed rectangle can then be divided according to a set length to obtain multiple grids, thus completing the gridding process of the deployment area. The specific value of the set length is not limited in this embodiment. Optionally, it can be set based on manual experience. For example, the set length can be set to Δlng = 0.005, Δlat = 0.005, that is, the distance between the centers of the two grids can be set to approximately 500 meters. Alternatively, it can be dynamically adjusted according to actual application needs. This embodiment does not impose any restrictions on this.

[0103] In this embodiment of the disclosure, the shared item delivery device can acquire multiple historical delivery strategies for shared items within the delivery range, as well as the sharing records of shared items under each historical delivery strategy, and then execute subsequent steps. It should be noted that the shared item delivery device in this embodiment of the disclosure can acquire multiple historical delivery strategies for shared items within the delivery range, as well as the sharing records of shared items under each historical delivery strategy, through various public, legal, and compliant means.

[0104] In one possible implementation of this disclosure, the shared item delivery device can acquire multiple historical delivery strategies for shared items within the delivery range, as well as the sharing records of shared items under each historical delivery strategy, through online or offline collection methods. For example, after authorization from the shared item operator, the shared item delivery device can collect multiple historical delivery strategies for shared items within the delivery range, as well as the sharing records of shared items under each historical delivery strategy, in real time. Alternatively, after authorization from the shared item operator, it can also acquire multiple historical delivery strategies for shared items within the delivery range, as well as the sharing records of shared items under each historical delivery strategy, offline manually, etc. This disclosure does not impose any limitations on this aspect.

[0105] In another possible implementation of this disclosure, the shared item delivery device can acquire multiple historical delivery strategies for shared items within the delivery range, as well as the sharing records of shared items under each historical delivery strategy, through network transmission or physical copying. For example, after being authorized by the shared item operator, the shared item delivery device can acquire multiple historical delivery strategies for shared items within the delivery range, as well as the sharing records of shared items under each historical delivery strategy, from other devices through network transmission or physical copying.

[0106] Step 102: For any historical delivery strategy, determine the delivery performance evaluation value of the corresponding historical delivery strategy based on the number of sharing failures in the sharing record.

[0107] In this embodiment of the disclosure, since the sharing record may include at least one of the following: a query record of a user querying the location of a shared item, a borrowing record of a user lending out a shared item, and a return record of a user returning a shared item, sharing failure in the sharing record may include at least one of borrowing failure and return failure. Borrowing failure can include two scenarios: the user did not perform a borrowing operation after performing a query operation, or the borrowing operation failed. Return failure can include two scenarios: the return operation failed, or the return operation succeeded but resulted in the accumulation of shared items. As a possible implementation, if the user does not perform a borrowing operation after performing a query operation, it can be determined as a borrowing failure; similarly, if the user fails to return a shared item after performing a return operation, this return operation can be determined as a return failure; and if the user successfully returns a shared item after performing a borrowing operation, but this results in the accumulation of shared items and reduces the utilization rate of the shared items, this return operation can also be determined as a return failure. Therefore, based on at least one of the shared records—query records, borrowing records, and return records—the number of borrowing operations and / or the number of failed return operations can be determined by querying borrowing and / or return operations within any given time period. It should be noted that failed returns should not only include cases where the client fails to perform the return operation, but also cases where the client successfully performs the return operation but causes an accumulation of shared items. This is because when the number of shared items within the delivery area is already sufficient, if users return shared items to the delivery area, it will cause an accumulation of shared items, reducing the utilization rate of shared items. In other words, to prevent an excessive number of shared items within the delivery area, resulting in an accumulation of shared items and reduced utilization, a quantity threshold can be set. The number of shared items should not exceed this threshold; otherwise, even if the user successfully returns the shared item after performing the return operation, the return operation can still be identified as a failed return. The specific value of the quantity threshold is not limited in this embodiment. Optionally, it can be set based on human experience. For example, the quantity threshold can be set to 500. Alternatively, it can be dynamically adjusted according to actual application needs. This embodiment does not impose any restrictions on this.

[0108] In this embodiment, the shared item delivery device can determine the delivery effectiveness evaluation value of any historical delivery strategy by statistically analyzing the number of sharing failures in the sharing records. The delivery effectiveness evaluation value is used to measure the merits of the corresponding historical delivery strategy. Optionally, the delivery effectiveness evaluation value can be represented by the sum of the number of sharing failures within any time period under any delivery strategy; thus, the smaller the delivery effectiveness evaluation value, the better the delivery strategy.

[0109] In one possible implementation of this disclosure, for any historical delivery strategy, the delivery effect evaluation value of the corresponding historical delivery strategy can be determined by directly counting the number of sharing failures within any time period of the delivery scope.

[0110] In another possible implementation of this disclosure, for any historical delivery strategy, the delivery range can first be gridded to obtain multiple grids. Then, the delivery effectiveness evaluation value of the corresponding historical delivery strategy can be determined by counting the number of sharing failures of each grid in any time period and summing the results. For example, for any historical delivery strategy, assuming that M networks are obtained after gridding the delivery range, and these M grids are represented as g1, g2, ... g M This allows for the statistical analysis of any grid g within the delivery range. j The number of sharing failures within any given time period E. j To determine the performance evaluation value of the corresponding historical campaign strategies.

[0111] Step 103: Train the value network based on each historical delivery strategy and its performance evaluation value.

[0112] In this embodiment, after determining the evaluation values ​​of the delivery effectiveness of each historical delivery strategy, a value network can be trained based on each historical delivery strategy and its evaluation values. The value network is used to predict the evaluation value of the delivery effectiveness of any delivery strategy. As one possible implementation, each historical delivery strategy and its evaluation values ​​can be used as training samples to train the value network, resulting in a trained value network. Specifically, for any historical delivery strategy, any time period E can be divided into N time slices with a step size of Δt, i.e., E = t1∪t2∪…∪t. N Then, state transition samples are obtained by considering each time slice separately to train the value network. For example, when the time period E is from 8:00 AM to 12:00 PM, and Δt is 10 minutes, the four hours from 8:00 AM to 12:00 PM can be divided into time slices of 8:00-8:10, 8:10-8:20, ..., 11:50-12:00, with a step size of Δt = 10 minutes. Then, each 10-minute period is considered separately. It is understandable that under any historical delivery strategy, any grid g within the delivery range... j At any given time slice t k Each has a state S jk ={D k V j 1 Vk 2}, where D k Let t be an M-dimensional vector used to represent t. k The distribution of shared items within the delivery range, i.e., each element of the vector represents the number of shared items within its corresponding grid; V j 1 For grid g j The corresponding one-hot encoding; V k 2 For time slice t k The corresponding one-hot encoding. From time slice t k To the next time slice t k+1 Because of users' behaviors of lending, using, and returning shared items, Grid g j The state transition is S j(k+1) ={D k+1 V j 1 V k+1 2}, and simultaneously generate a mesh g j Instant sharing within the r jk That is, the grid at t k to t k+1 During this period, the number of times borrowing failed and the number of times returning failed are recorded, thus generating a state transition tuple {S}. jk S j(k+1) r jk Therefore, for each grid g1, g2, ... g M Each time slice t1, t2, ... t N-1 Each of these can generate a state transition tuple, thus yielding a state transition sample P = {S}. jk S j(k+1) r jk};j=1:M;k=1:(N-1). Since the above state transition samples exist for any historical delivery strategy, the Bellman iteration formula V(S) can be used. jk )=V(S j(k+1) )+γ×r jk Train a value network V(·), where γ∈[0,1] is a manually set decay parameter. The closer γ is to 0, the easier the value network V(·) is to train; the closer γ is to 1, the more reasonable the value network V(·) is. Optionally, γ can be set to 0.9.

[0113] To clearly illustrate the training process of the value network, embodiments of this disclosure provide, as follows: Figure 3 The value network structure shown is as follows: Figure 3The value network V(·) shown contains 3 layers. The first layer has 1 neuron, the second layer has 2×(M+M+N) neurons, and the third layer has (M+M+N) neurons, thus the activation function σ(w2×σ(w1×S)) is used. jk +b1)+b2)=σ(w2×σ(w1×S j(k+1) +b1)+b2)+γ×r jk Here, w1, w2, b1, b2 are the parameters of the neural network. Parameter training can be achieved using existing frameworks, such as pyTorch.

[0114] It should be noted that the value network V(·) is a function from a state to the expected sharing failure, indicating that starting from this state and ending at any time interval E, the expected sharing corresponding to that state fails. Therefore, each grid g j The number of sharing failures within any given time period E. j =V(S) j0 ), where S j0 For grid g j The state at the beginning of any time interval E, i.e., when t = 0.

[0115] Step 104: Using a trained value network, predict the evaluation value of the delivery effect of each candidate delivery strategy for shared items within the delivery range, and determine the target delivery strategy based on the evaluation value.

[0116] It should be noted that the candidate deployment strategies in this step differ from the historical deployment strategies. These strategies may include an initial deployment strategy obtained by evenly distributing the total number of shared items within the deployment range at the start of any time period E (t=0) to each grid within the deployment range, and a strategy obtained by increasing or decreasing the number of any grid g based on this initial deployment strategy. j At least one deployment strategy is derived from the number of shared items deployed within the site.

[0117] In this embodiment of the disclosure, a trained value network can be used to predict the evaluation value of the delivery effect of each candidate delivery strategy within the delivery range of the shared item, so as to obtain the evaluation value of the delivery effect of each candidate delivery strategy, and determine the target delivery strategy based on the evaluation value of the delivery effect of each candidate delivery strategy. Optionally, when the evaluation value of the delivery effect is represented by the sum of the number of sharing failures within any time period under any delivery strategy, since the smaller the evaluation value of the delivery effect of the delivery strategy, the better the delivery strategy, the candidate delivery strategy with the smallest evaluation value among the candidate delivery strategies can be taken as the target delivery strategy.

[0118] Step 105: Deploy shared items according to the target deployment strategy.

[0119] In this embodiment of the disclosure, after determining the target deployment strategy, shared items can be deployed according to the target deployment strategy to make the deployment of shared items more reasonable, improve the utilization rate of shared items and user experience, and solve the problem of uneven supply and demand of shared items in different regions and time periods.

[0120] The shared item deployment method provided in this disclosure acquires multiple historical deployment strategies for shared items within the deployment range, as well as sharing records for each historical deployment strategy. For any given historical deployment strategy, the method determines the deployment effectiveness evaluation value based on the number of sharing failures in the sharing records. Then, based on each historical deployment strategy and its evaluation value, a value network is trained. This trained value network then predicts the deployment effectiveness evaluation value for each candidate deployment strategy within the deployment range. After determining the target deployment strategy based on the evaluation value, the shared item is deployed according to the target deployment strategy. This allows for more rational deployment of shared items, improves utilization and user experience, and addresses the problem of uneven supply and demand of shared items in different regions and time periods.

[0121] As can be seen from the above analysis, in this embodiment of the disclosure, the evaluation value of the delivery effect of any historical delivery strategy can be determined based on the number of sharing failures in the sharing record. In order to clearly explain how this disclosure determines the evaluation value of the delivery effect of the corresponding historical delivery strategy based on the number of sharing failures in the sharing record, this disclosure also provides a method for delivering shared items.

[0122] Figure 4 This is a flowchart illustrating another method for deploying shared items provided in an embodiment of this disclosure. In this embodiment, the sharing record includes query records and borrowing records.

[0123] like Figure 4 As shown, the method for deploying shared items may include the following steps:

[0124] Step 401: Obtain multiple historical deployment strategies for shared items within the deployment range, as well as the sharing records of shared items under each historical deployment strategy.

[0125] In this embodiment of the disclosure, the sharing record includes query records and borrowing records. The borrowing record records both the borrowing operation of the shared item and the return operation of the shared item.

[0126] It should be noted that the other execution processes in this step can be referred to the execution process of step 101 in the previous embodiment, and the principle is the same, so they will not be repeated here.

[0127] Step 402: For any historical delivery strategy, determine the query operations recorded within the statistical period after the execution of the corresponding historical delivery strategy from the query records.

[0128] In this embodiment, for any historical delivery strategy, the query operations recorded within a statistical period after the execution of the corresponding historical delivery strategy can be determined from the query records. This embodiment does not limit the specific value of the statistical period; optionally, it can be set based on human experience, for example, a statistical period of 4 hours. Alternatively, it can be dynamically adjusted according to actual application needs; this embodiment does not impose any restrictions on this.

[0129] It is understandable that, since each historical deployment strategy refers to the historical deployment strategies of shared items within the deployment scope, for any historical deployment strategy, the query operations recorded within the statistical period after the execution of the corresponding historical deployment strategy are determined from the query records. In other words, for any historical deployment strategy, the query operations recorded within any time period of the deployment scope are considered to be the query operations recorded within the statistical period after the execution of the corresponding historical deployment strategy of shared items within the deployment scope.

[0130] Step 403: If the client that performed the query operation did not perform the lending operation after the query operation, or if the lending record indicates that the client failed to perform the lending operation after the query operation, then the query operation is determined to be a lending failure, so as to determine the number of lending failures.

[0131] In this embodiment of the disclosure, if the client that performs the query operation does not perform the borrowing operation after the query operation, or if the borrowing record in the shared record indicates that the client's borrowing operation failed after the query operation, then the query operation can be determined as a borrowing failure, and the number of borrowing failures can be counted accordingly. In other words, a borrowing failure can include both situations where the user does not perform the borrowing operation after performing the query operation and where the borrowing operation fails.

[0132] It is understandable that after a user performs a query, if there are no available shared items to borrow nearby, the client may not perform a borrowing operation. Conversely, if the shared item becomes unavailable after a query, the client may fail to borrow the item. Optionally, when a client fails to borrow the item after a query, the borrowing record can document this failure as a borrowing failure. Therefore, the borrowing record can be used to determine if the client failed to borrow the item after the query.

[0133] Step 404: Based on the lending records, query the return operations recorded within the statistical period after the execution of the corresponding historical delivery strategy to determine the number of return failures.

[0134] In this embodiment of the disclosure, since the borrowing record records both the borrowing operation of the shared item and the return operation of the shared item, the return operation recorded within the statistical period after the execution of the corresponding historical delivery strategy can be queried based on the borrowing record in the sharing record, thereby determining the number of return failures.

[0135] It is understandable that when a user returns a shared item, the return operation may fail due to various reasons, such as the item being unable to be returned properly (e.g., a shared bicycle lock not locking correctly) or the item being outside the designated return area. Optionally, when a client fails to return a shared item, the borrowing record can record this failure as a return operation failure. Therefore, the number of failed returns can be determined based on the number of failed returns recorded in the borrowing record.

[0136] It should be noted that "return failure" should not only include situations where the client fails to perform the return operation, but also situations where the client successfully performs the return operation but causes shared items to accumulate. This is because, although the user successfully returns the shared item, the accumulation of shared items reduces the utilization rate of the shared items. Therefore, situations where the client successfully performs the return operation but causes shared items to accumulate should be considered as return failures.

[0137] Step 405: Determine the evaluation value of the corresponding historical delivery strategy based on the number of failed lending and / or failed return.

[0138] In this embodiment of the disclosure, the evaluation value of the delivery effect of the corresponding historical delivery strategy can be determined based on the number of failed lending and / or failed return as determined by the above steps. Optionally, the number of failed lending can be used as the evaluation value of the delivery effect of the corresponding historical delivery strategy, or the number of failed return can be used as the evaluation value of the delivery effect of the corresponding historical delivery strategy, or the sum of the number of failed lending and failed return can be used as the evaluation value of the delivery effect of the corresponding historical delivery strategy. This embodiment of the disclosure does not impose any restrictions on this.

[0139] Step 406: Train the value network based on each historical delivery strategy and its performance evaluation value.

[0140] Step 407: Using a trained value network, predict the evaluation value of the delivery effect of each candidate delivery strategy for shared items within the delivery range, and determine the target delivery strategy based on the evaluation value.

[0141] Step 408: Deploy shared items according to the target deployment strategy.

[0142] It should be noted that the execution process of steps 406-408 can refer to the execution process of steps 103-105 in the previous embodiment, and the principle is the same, so it will not be repeated here.

[0143] The shared item deployment method provided in this embodiment determines the number of query operations recorded within a statistical period after the execution of any historical deployment strategy. If the client executing the query operation does not perform a borrowing operation after the query, or if the borrowing records indicate that the client's borrowing operation failed after the query, the query operation is identified as a borrowing failure, thus determining the number of borrowing failures. Then, based on the borrowing records, the method queries the number of return operations recorded within the statistical period after the execution of the corresponding historical deployment strategy to determine the number of return failures. Finally, based on the number of borrowing failures and / or return failures, the deployment effectiveness evaluation value of the corresponding historical deployment strategy is determined. Therefore, it is possible to determine the deployment effectiveness evaluation value of a corresponding historical deployment strategy based on the number of borrowing failures and / or return failures in the sharing records of shared items under each historical deployment strategy.

[0144] The above analysis shows that return failures should not only include situations where the client fails to perform the return operation, but also situations where the client successfully performs the return operation but causes shared items to accumulate. To clearly explain how to determine the number of return failures when it includes both situations where the client fails to perform the return operation and situations where the client successfully performs the return operation but causes shared items to accumulate, this disclosure provides... Figure 5 The flowchart shown illustrates the process of determining the number of failed returns. Figure 5 As shown, determining the number of failed returns may include the following steps:

[0145] Step 501: Based on the lending records, query the return operations recorded within the statistical period after the execution of the corresponding historical delivery strategy.

[0146] Here, since the borrowing records contain the return operations of shared items, the return operations recorded within the statistical period after the execution of the corresponding historical deployment strategy can be queried based on the borrowing records in the sharing records to obtain the corresponding return operations.

[0147] Step 502: For the returned operation found in the query, determine the partition from the multiple partitions divided by the delivery range to execute the corresponding returned operation.

[0148] Since each historical delivery strategy refers to the historical delivery strategies of shared items within the delivery range, multiple partitions can be divided from the delivery range for any given historical delivery strategy. One possible implementation is to divide the delivery range into multiple partitions by performing grid processing. The specific implementation process can be found in step 101 of the above embodiment, where the principle is the same, and will not be repeated here. It is understood that the multiple grids obtained in step 101 represent the multiple partitions at this point.

[0149] Here, for each returned operation found, the partition to which the corresponding returned operation will be executed can be determined from among the multiple partitions divided by the delivery area. The number of returned operations found can be one or more, and this embodiment does not limit this. For example, suppose the multiple partitions divided by the delivery area are represented as g1, g2, ... g M If any return operation found in the query is executed in partition g1, then the partition that executed the return operation is partition g1. Therefore, for the return operation found in the query, the partition that executed the corresponding return operation can be determined.

[0150] Step 503: If the number of shared items in the corresponding partition exceeds the quantity threshold after the return operation is performed, the return operation is determined to be a failure.

[0151] Here, after determining the partition to perform the corresponding return operation, the failure of the return operation can be determined by counting the number of shared items contained in the corresponding partition after the operation is completed. It should be noted that all return operations here refer to those successfully executed by the client. Optionally, if the number of shared items in the corresponding partition after the return operation exceeds a quantity threshold, the return operation can be determined as a failure. This embodiment of the disclosure does not limit the specific value of the quantity threshold. Optionally, it can be set based on experience; for example, the quantity threshold can be set to 500, meaning the number of shared items in a single partition cannot exceed 500. Otherwise, the shared items in that partition are considered to be piling up, and the return operation fails. Alternatively, it can be dynamically adjusted according to actual application needs; this embodiment of the disclosure does not impose any restrictions on this.

[0152] Step 504: Determine the number of failed returns from the retrieved return operations.

[0153] It is understandable that since there are two possible outcomes in the return operation, namely successful return and failed return, the failure includes not only the client failing to perform the return operation, but also the client successfully performing the return operation but causing the shared items to accumulate. Therefore, the number of failed returns can be determined from the returned operation records.

[0154] In summary, by querying return operations recorded within a statistical period following the execution of the corresponding historical deployment strategy based on borrowing records, this system identifies the partition from which the return operation was executed, based on multiple partitions defined by the deployment area. If the number of shared items in the corresponding partition exceeds a threshold after the return operation, the return operation is marked as a failure. Furthermore, the number of failed returns is determined from the retrieved records. This allows for the marking of a failed return operation even when a successful return operation on the client side results in an accumulation of shared items, thereby improving the utilization rate of shared items and preventing their accumulation.

[0155] As can be seen from the above analysis, the embodiments of this disclosure can use a trained value network to predict the evaluation value of the delivery effect of each candidate delivery strategy for shared items within the delivery range, so as to determine the target delivery strategy based on the evaluation value. In order to clearly explain how this disclosure uses a trained value network to predict the evaluation value of the delivery effect of each candidate delivery strategy for shared items within the delivery range, so as to determine the target delivery strategy based on the evaluation value, this disclosure also provides a method for delivering shared items.

[0156] Figure 6 This is a flowchart illustrating another method for deploying shared items provided in an embodiment of this disclosure.

[0157] like Figure 6 As shown, the method for deploying shared items may include the following steps:

[0158] Step 601: Obtain multiple historical deployment strategies for shared items within the deployment range, as well as the sharing records of shared items under each historical deployment strategy.

[0159] Step 602: For any historical delivery strategy, determine the delivery performance evaluation value of the corresponding historical delivery strategy based on the number of sharing failures in the sharing record.

[0160] Step 603: Train the value network based on each historical delivery strategy and its performance evaluation value.

[0161] It should be noted that the execution process of steps 601-603 can be referred to the execution process of steps 101-103 in the above embodiment, and the principle is the same, so it will not be repeated here.

[0162] Step 604: Determine the initial candidate deployment strategy based on the initial deployment volume of shared items in each of the multiple partitions divided by the deployment scope.

[0163] It is understood that in this embodiment of the disclosure, the grid mentioned in the above steps refers to the partition here, so that an initial candidate deployment strategy can be determined based on the initial deployment quantity of shared items in each of the multiple partitions divided by the deployment range. As one possible implementation, the total initial deployment quantity of shared items in each of the multiple partitions divided by the deployment range can be calculated first, and then the initial candidate deployment strategy can be obtained by distributing the same number of shared items equally to each partition.

[0164] Step 605: The trained value network is used to predict the initial candidate delivery strategy, and the delivery effect evaluation value corresponding to the initial candidate delivery strategy is used as the reference evaluation value.

[0165] In this embodiment, a trained value network can be used to predict the initial candidate delivery strategy, thereby obtaining the delivery effect evaluation value corresponding to the initial candidate delivery strategy, and using this delivery effect evaluation value as a reference evaluation value for subsequent steps. Optionally, the trained value network can be a value network V(·), which predicts the initial candidate delivery strategy Θ. The calculation formula for the delivery effect evaluation value Ω corresponding to the initial candidate delivery strategy is as follows:

[0166] Ω=∑ j l j =∑ j V(S j0 )

[0167] Where Ω represents the evaluation value of the delivery effect corresponding to the initial candidate delivery strategy; l j Used to indicate the number of sharing failures within any given time period E; S j0 ={Θ,V j 1 V0 2} is used to represent partition g j In the state at time slice t=0, Θ represents the initial candidate delivery strategy, which can be understood as the distribution of shared items within the delivery range at time slice t=0, that is, the quantity of shared items delivered in each of the multiple partitions divided by the delivery range at t=0; V j 1 For partition g j The corresponding one-hot encoding; V k 2 This is the one-hot encoding corresponding to time slice t=0.

[0168] Step 606: Perform multiple rounds of strategy updates based on the initial candidate delivery strategy.

[0169] In this embodiment of the disclosure, multiple rounds of policy updates can be performed based on an initial candidate delivery strategy. As one possible implementation, multiple rounds of policy updates can be performed by changing the delivery quantity of shared items in each of the multiple partitions divided by the delivery scope, based on the initial candidate delivery strategy. This updates the initial candidate delivery strategy multiple times to obtain at least one updated candidate delivery strategy.

[0170] Step 607: Whenever a strategy update is performed, the trained value network is used to predict the delivery effect of at least one candidate delivery strategy after the current update.

[0171] In this embodiment of the disclosure, each time a strategy update is performed, at least one candidate delivery strategy after the update can be obtained. The trained value network can then be used to predict the delivery effect of the at least one candidate delivery strategy after the update, so as to obtain the delivery effect evaluation value of the at least one candidate delivery strategy after the update.

[0172] Step 608: If the evaluation value of the delivery effect of at least one candidate delivery strategy after this round of updates is less than the reference evaluation value, then the evaluation value of the delivery effect of at least one candidate delivery strategy after this round of updates will be used as the reference evaluation value, and the subsequent strategy update process will continue.

[0173] In this embodiment, if the evaluation value of the delivery effect of at least one candidate delivery strategy after the current update is less than the reference evaluation value, then the evaluation value of the delivery effect of at least one candidate delivery strategy after the current update can be used as the reference evaluation value, and the subsequent strategy update process can continue. That is, each time a strategy update is performed, a trained value network is used to predict the delivery effect of the new candidate delivery strategy, and the evaluation value of the delivery effect of the new candidate delivery strategy is compared with the reference evaluation value. If the evaluation value of the new candidate delivery strategy is less than the reference evaluation value, then the reference evaluation value is updated to the new candidate delivery strategy. Therefore, after each round of strategy updates is completed, the obtained reference evaluation value is the minimum evaluation value among the delivery effect evaluation values ​​of all candidate delivery strategies, and the candidate delivery strategy corresponding to this reference evaluation value is the optimal delivery strategy among all candidate delivery strategies.

[0174] Step 609: After all rounds of strategy updates are completed, the candidate delivery strategy corresponding to the reference evaluation value is used as the target delivery strategy.

[0175] In this embodiment of the disclosure, after each round of strategy updates is completed, since the obtained reference evaluation value is the smallest evaluation value among the evaluation values ​​of the delivery effect of each candidate delivery strategy, and the candidate delivery strategy corresponding to the reference evaluation value is the optimal delivery strategy among each candidate delivery strategy, the candidate delivery strategy corresponding to the reference evaluation value can be used as the target delivery strategy.

[0176] Step 610: Deploy shared items according to the target deployment strategy.

[0177] It should be noted that the execution process of step 610 can be referred to the execution process of step 105 in the above embodiment, and the principle is the same, so it will not be repeated here.

[0178] The shared item deployment method provided in this disclosure determines initial candidate deployment strategies based on the initial deployment quantity of shared items in each of the multiple partitions divided by the deployment area. A trained value network is then used to predict the deployment effect of these initial candidate strategies, using the deployment effect evaluation value corresponding to each strategy as a reference evaluation value. Multiple rounds of strategy updates are then performed based on these initial candidate strategies. Each time a strategy update is performed, the trained value network predicts the deployment effect of at least one candidate strategy after the update. If the deployment effect evaluation value of at least one candidate strategy after the update is less than the reference evaluation value, this value is used as the reference evaluation value, and subsequent strategy updates continue. After all rounds of strategy updates are completed, the candidate deployment strategy corresponding to the reference evaluation value is selected as the target deployment strategy. Therefore, by updating the reference evaluation value, a target deployment strategy can be determined from multiple candidate strategies.

[0179] As can be seen from the above analysis, the embodiments of this disclosure can perform multiple rounds of policy updates based on the initial candidate delivery strategy. To clearly explain how this disclosure performs multiple rounds of policy updates based on the initial candidate delivery strategy, the embodiments of this disclosure provide... Figure 7 The diagram illustrates the process of performing multiple rounds of policy updates based on the initial candidate delivery strategy. Figure 7 As shown, performing multiple rounds of policy updates based on the initial candidate delivery strategy may include the following steps:

[0180] Step 701: For each round of policy update, determine a target partition from multiple partitions.

[0181] Here, for each round of strategy update, a target partition can be determined from multiple partitions, and multiple rounds of strategy updates can be performed by changing the distribution of shared items in the target partition.

[0182] Step 702: Increase and decrease the amount of shared items deployed in the target partition by a set step size to obtain two candidate deployment strategies obtained in this round of strategy update.

[0183] Here, based on the initial candidate deployment strategy, two candidate deployment strategies can be obtained in this round of strategy updates by increasing and decreasing the deployment amount of shared items in the target partition using a set step size. The specific value of the set step size is not limited in this embodiment; optionally, it can be set based on manual experience. For example, the set step size can be set to 1, or it can be dynamically adjusted according to actual application needs. This embodiment does not impose any restrictions on this. For example, assuming the target partition is partition g... j If the step size is set to 1, then the target partition g can be divided based on the initial candidate delivery strategy Θ. j One more shared item is added to the in-system deployment, resulting in a candidate deployment strategy Θ obtained from this round of strategy updates. + and the target partition g j The number of shared items deployed within the platform is reduced by one to obtain another candidate deployment strategy obtained from this round of strategy updates. - .

[0184] In summary, by identifying a target partition from multiple partitions for each round of strategy updates, and then increasing or decreasing the distribution of shared items in that target partition by a set step size, two candidate distribution strategies are obtained for this round of strategy updates. Therefore, based on the initial candidate distribution strategies, the initial candidate distribution strategies can be updated by identifying the target partition from multiple partitions and increasing or decreasing the distribution of shared items in that target partition by a set step size.

[0185] To make it clear Figure 7 In the illustrated embodiment, step 701, which involves determining a target partition from multiple partitions for each round of policy updates, is provided in this embodiment. Figure 8 The flowchart shown illustrates the process of determining a target partition from multiple partitions. Figure 8 As shown, determining a target partition from multiple partitions may include the following steps:

[0186] Step 801: For each round of strategy update, if the delivery volume of the target partition used in the previous round is lower than the minimum quantity or higher than the maximum quantity, then the next partition adjacent to the target partition used in the previous round will be used as the target partition for this round.

[0187] Here, for each round of strategy updates, if the number of targets allocated to the target partition in the previous round is lower than the lower limit or higher than the upper limit, then the next adjacent partition to the target partition in the previous round will be used as the target partition for this round. The specific values ​​of the upper and lower limits are not limited in this embodiment. Optionally, they can be set based on experience; for example, the upper limit can be set to 500 and the lower limit to 10. Alternatively, they can be dynamically adjusted according to actual application needs. This embodiment does not impose any restrictions on this. For example, suppose the target partition is partition g. j With the maximum number set at 500 and the minimum at 10, the strategy can be updated for each round, taking the target partition g from the previous round as an example. j If the number of deployments is less than 10 or more than 500, the next partition adjacent to the target partition used in the previous round will be used, for example, partition g. j+1 This will be the target zone for this round.

[0188] Step 802: If the number of deliveries for the target partition used in the previous round is not lower than the minimum quantity and not higher than the maximum quantity, then the target partition used in the previous round will be used as the target partition for this round.

[0189] Conversely, if the delivery volume of the target partition used in the previous round was neither lower than the minimum quantity nor higher than the maximum quantity, then the target partition used in the previous round will be used as the target partition for this round. Similarly, the target partition will be used as partition g. j Taking a maximum quantity of 500 and a minimum quantity of 10 as an example, in the previous round, the target partition g was used... j If the number of deployments is not less than 10 and not more than 500, the target partition used in the previous round, i.e., partition g, can be used. j This will be the target zone for this round.

[0190] In summary, by updating the strategy for each round, if the delivery volume of the target partition used in the previous round is lower than the minimum quantity or higher than the maximum quantity, the next adjacent partition of the target partition used in the previous round will be used as the target partition for this round. If the delivery volume of the target partition used in the previous round is neither lower than the minimum quantity nor higher than the maximum quantity, the target partition used in the previous round will be used as the target partition for this round. This allows for the setting of the stopping conditions for a round of strategy updates.

[0191] To illustrate the above embodiments more clearly, examples are given below.

[0192] like Figure 9As shown, after obtaining multiple historical deployment strategies for shared items within the deployment range, and the sharing records of shared items under each historical deployment strategy, the deployment range of each historical deployment strategy can be first gridded to obtain multiple partitions. Based on the sharing records of shared items under each historical deployment strategy, the deployment effect evaluation value of the corresponding historical deployment strategy is determined. Then, based on the multiple partitions divided by the deployment range, each historical deployment strategy, and its deployment effect evaluation value, a value network is trained. The trained value network is then used to determine the target deployment strategy using a gradient descent search algorithm on the initial candidate deployment strategies, so that shared items can be deployed according to the target deployment strategy. The input to the gradient descent search algorithm can be the multiple partitions divided by the deployment range and the trained value network, and the output is the target deployment strategy.

[0193] As one possible implementation, the gradient descent search algorithm can be:

[0194] Step 1: Generate an initial candidate deployment strategy Θ as the target deployment strategy. For example, first calculate the initial deployment quantity of shared items in each of the multiple partitions divided by the deployment range, and then obtain the initial candidate deployment strategy Θ by distributing the same number of shared items equally to each partition.

[0195] Step 2: Under the initial candidate delivery strategy Θ, use the trained value network to calculate the delivery performance evaluation value Ω corresponding to the initial candidate delivery strategy, for example: Ω = ∑ j V(S j0 )| Θ ;

[0196] Step 3: Set ΔΩ = +∞;

[0197] Step 4: When ΔΩ is greater than 0:

[0198] From g j =g1 to g j =g M (M is the number of zones divided into within the delivery area) All of them perform the following operations:

[0199] Based on the initial candidate delivery strategy Θ, the step size is increased by g. j The number of shared items deployed within the site is used to obtain an updated candidate deployment strategy. + ;

[0200] In the delivery strategy Θ + Using a trained value network, the delivery strategy Θ is calculated. + The corresponding evaluation value Ω for campaign performance + ,For example:

[0201] Based on the initial candidate delivery strategy Θ, the step size is reduced by g. j The number of shared items deployed within the site is used to obtain another updated candidate deployment strategy. - ;

[0202] In the delivery strategy Θ - Using a trained value network, the delivery strategy Θ is calculated. - The corresponding evaluation value Ω for campaign performance - ,For example:

[0203] Select Ω, Ω + and Ω - The minimum value in the range is used as the reference evaluation value Ω. * For example: Ω * =min{Ω,Ω + ,Ω -};

[0204] If Ω + =Ω * ,

[0205] Then update the target delivery strategy to Θ + ;

[0206] If Ω - =Ω * ,

[0207] Then update the target delivery strategy to Θ - ;

[0208] If Ω = Ω * ,

[0209] The target delivery strategy remains at Θ;

[0210] Update ΔΩ=Ω-Ω * ;

[0211] Ω * The value is assigned to Ω, i.e., Ω = Ω * ;

[0212] Output the target delivery strategy Θ.

[0213] In summary, by acquiring multiple historical deployment strategies for shared items within the deployment area, and the sharing records of shared items under each historical deployment strategy, the deployment area of ​​each historical deployment strategy is gridded to obtain multiple partitions. Based on the sharing records of shared items under each historical deployment strategy, the deployment effect evaluation value of the corresponding historical deployment strategy is determined. Then, based on the multiple partitions of the deployment area, each historical deployment strategy, and its deployment effect evaluation value, a value network is trained. This trained value network is then used to determine the target deployment strategy from the initial candidate deployment strategies using a gradient descent search algorithm. Based on the target deployment strategy, shared items are deployed. This allows for more rational deployment of shared items, improves the utilization rate of shared items and user experience, and solves the problem of uneven supply and demand of shared items in different areas and time periods.

[0214] To implement the above embodiments, this disclosure also proposes a device for distributing shared items.

[0215] Figure 10 This is a schematic diagram of a shared item delivery device provided in an embodiment of the present disclosure.

[0216] like Figure 10 As shown, the shared item delivery device includes: an acquisition module 11, a determination module 12, a training module 13, a prediction module 14, and a delivery module 15.

[0217] The acquisition module 11 is used to acquire multiple historical deployment strategies for shared items within the deployment range, as well as the sharing records of shared items under each historical deployment strategy;

[0218] Module 12 is used to determine the evaluation value of the delivery effect of any historical delivery strategy based on the number of sharing failures in the sharing record.

[0219] Training module 13 is used to train the value network based on each historical delivery strategy and its delivery effect evaluation value;

[0220] Prediction module 14 is used to use a trained value network to predict the evaluation value of the delivery effect of each candidate delivery strategy for shared items within the delivery range, so as to determine the target delivery strategy based on the evaluation value.

[0221] The delivery module 15 is used to deliver shared items according to the target delivery strategy.

[0222] Furthermore, in one possible implementation of this disclosure embodiment, the shared records include query records and borrowing records; the determining module 12 includes:

[0223] The first determining unit 1201 is used to determine, from the query records, the query operations recorded within the statistical period after the execution of the corresponding historical delivery strategy for any given historical delivery strategy.

[0224] The second determining unit 1202 is used to determine the query operation as a failed loan operation if the client that performed the query operation did not perform the loan operation after the query operation, or if the client failed to perform the loan operation after the query operation according to the loan record, so as to determine the number of failed loans.

[0225] The query unit 1203 is used to query the return operations recorded within the statistical period after the execution of the corresponding historical delivery strategy based on the lending records, so as to determine the number of return failures.

[0226] The third determining unit 1204 is used to determine the evaluation value of the delivery effect of the corresponding historical delivery strategy based on the number of loan failures and / or return failures.

[0227] Furthermore, in one possible implementation of this disclosure, the query unit 1203 is further configured to:

[0228] Based on the lending records, query the return operations recorded within the statistical period after the execution of the corresponding historical delivery strategy;

[0229] For the returned operations found in the query, determine the partition from the multiple partitions divided by the delivery range to execute the corresponding returned operation;

[0230] If the number of shared items in the corresponding partition exceeds the quantity threshold after the return operation is performed, the return operation will be determined as a failure.

[0231] Determine the number of failed returns from the returned operations retrieved.

[0232] Furthermore, in one possible implementation of this disclosure embodiment, the prediction module 14 includes:

[0233] The fourth determining unit 1401 is used to determine the initial candidate deployment strategy based on the initial deployment quantity of shared items in each of the multiple partitions divided by the deployment range.

[0234] The first prediction unit 1402 is used to predict the initial candidate delivery strategy using a trained value network, so as to use the delivery effect evaluation value corresponding to the initial candidate delivery strategy as a reference evaluation value.

[0235] The first processing unit 1403 is used to perform multiple rounds of policy updates based on the initial candidate delivery policy;

[0236] The second prediction unit 1404 is used to predict the delivery effect of at least one candidate delivery strategy after the current round of strategy update by using a trained value network whenever a round of strategy update is performed.

[0237] The second processing unit 1405 is used to take the evaluation value of the delivery effect of at least one candidate delivery strategy after the current round of updates as the reference evaluation value and continue to execute the subsequent strategy update process if the evaluation value of the delivery effect of at least one candidate delivery strategy after the current round of updates is less than the reference evaluation value.

[0238] The third processing unit 1406 is used to execute the strategy update after each round and take the candidate delivery strategy corresponding to the reference evaluation value as the target delivery strategy.

[0239] Furthermore, in one possible implementation of this disclosure, the first processing unit 1403 is further configured to:

[0240] For each round of policy updates, a target partition is determined from multiple partitions;

[0241] The distribution of shared items is increased and decreased by a set step size for the target partition to obtain two candidate distribution strategies for this round of strategy update.

[0242] Furthermore, in one possible implementation of this disclosure, the first processing unit 1403 is further configured to:

[0243] For each round of strategy update, if the number of deliveries to the target partition used in the previous round is lower than the minimum or higher than the maximum, then the next partition adjacent to the target partition used in the previous round will be used as the target partition for this round.

[0244] If the number of targets deployed in the previous round was not lower than the minimum number and not higher than the maximum number, then the target region deployed in the previous round will be used as the target region for this round.

[0245] It should be noted that the explanation of the above-described method for distributing shared items also applies to the device for distributing shared items in this embodiment, and will not be repeated here.

[0246] The shared item delivery device provided in this embodiment acquires multiple historical delivery strategies for shared items within the delivery range, as well as sharing records for each historical delivery strategy. For any given historical delivery strategy, it determines the delivery effectiveness evaluation value based on the number of sharing failures in the sharing records. Then, based on each historical delivery strategy and its evaluation value, it trains a value network. This trained value network then predicts the delivery effectiveness evaluation value for each candidate delivery strategy within the delivery range. After determining the target delivery strategy based on the evaluation value, it delivers the shared items according to the target delivery strategy. This allows for more rational delivery of shared items, improves utilization and user experience, and addresses the problem of uneven supply and demand of shared items in different areas and time periods.

[0247] To implement the above embodiments, this disclosure also proposes an electronic device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to execute the method for delivering shared items proposed in any of the foregoing embodiments of this disclosure.

[0248] Figure 11 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present disclosure, which can realize the present disclosure. Figure 1-10 The process of the illustrated embodiment is as follows: Figure 11 As shown, the electronic device may include: a housing 1101, a processor 1102, a memory 1103, a circuit board 1104, and a power circuit 1105. The circuit board 1104 is disposed inside the space enclosed by the housing 1101, and the processor 1102 and the memory 1103 are disposed on the circuit board 1104. The power circuit 1105 is used to supply power to the various circuits or devices of the electronic device. The memory 1103 is used to store executable program code. The processor 1102 runs a program corresponding to the executable program code by reading the executable program code stored in the memory 1103, for executing the shared item delivery method described in any of the foregoing embodiments.

[0249] For details on the specific execution process of the above steps by processor 1102, and the steps further executed by processor 1102 through running executable program code, please refer to this disclosure. Figure 1-10 The description of the illustrated embodiments will not be repeated here.

[0250] To implement the above embodiments, this disclosure also proposes a computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause the computer to execute the method for delivering shared items proposed in any of the foregoing embodiments of this disclosure.

[0251] To implement the above embodiments, this disclosure also proposes a computer program product, including a computer program that, when executed by a processor, implements the method for deploying shared items proposed in any of the foregoing embodiments of this disclosure.

[0252] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this disclosure. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0253] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this disclosure, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0254] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing custom logic functions or processes, and the scope of preferred embodiments of this disclosure includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as will be understood by those skilled in the art to which embodiments of this disclosure pertain.

[0255] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0256] It should be understood that various parts of this disclosure can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0257] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0258] Furthermore, the functional units in the various embodiments of this disclosure can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0259] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of the present disclosure have been shown and described above, it is to be understood that the above embodiments are exemplary and should not be construed as limiting the present disclosure. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present disclosure.

Claims

1. A method for deploying shared items, characterized in that, Includes the following steps: The system obtains multiple historical deployment strategies for the shared items within the deployment range, as well as the sharing records of the shared items under each historical deployment strategy. The sharing records include query records and borrowing records, and the borrowing records include borrowing operations and return operations. For any historical delivery strategy, the delivery effect evaluation value of the corresponding historical delivery strategy is determined based on the number of sharing failures in the sharing record. The sharing failures include borrowing failures and returning failures. Borrowing failures include failing to execute the borrowing operation and not executing the borrowing operation after executing the query operation. Returning failures include failing to execute the returning operation and accumulating shared items after successfully executing the returning operation. Based on the historical deployment strategies and their deployment effect evaluation values, a value network is trained. The value network is a function from a state to the expected sharing failure. The state includes the distribution characteristics of shared items within the deployment range, the grid area characteristics within the deployment range, and the time slice characteristics obtained by splitting the target time period. The value network is used to represent the expected sharing failure corresponding to the state from the start of the state to the end of any time period E. A trained value network is used to predict the evaluation value of the delivery effect of each candidate delivery strategy for the shared items within the delivery range, so as to determine the target delivery strategy based on the evaluation value. The shared items are deployed according to the target deployment strategy.

2. The method according to claim 1, characterized in that, For any given historical delivery strategy, determining the delivery performance evaluation value of the corresponding historical delivery strategy based on the number of sharing failures in the sharing record includes: For any given historical delivery strategy, determine the query operations recorded within the statistical period following the execution of the corresponding historical delivery strategy from the query records. If the client that performs the query operation does not perform the lending operation after the query operation, or if the lending record determines that the client failed to perform the lending operation after the query operation, then the query operation is determined to be a lending failure, so as to determine the number of lending failures. Based on the lending records, query the return operations recorded within the statistical period after the execution of the corresponding historical delivery strategy to determine the number of return failures; The evaluation value of the delivery effect of the corresponding historical delivery strategy is determined based on the number of loan failures and / or the number of return failures.

3. The method according to claim 2, characterized in that, The step of querying the return operations recorded within the statistical period after the execution of the corresponding historical delivery strategy, based on the lending records, to determine the number of failed returns includes: Based on the lending records, query the return operations recorded within the statistical period after the execution of the corresponding historical delivery strategy; For the returned operation found, determine the partition from the multiple partitions divided by the delivery range to execute the corresponding returned operation; If the number of shared items in the corresponding partition exceeds the quantity threshold after the return operation is performed, the return operation will be determined as a return failure. Determine the number of failed returns from the returned operations retrieved.

4. The method according to claim 2, characterized in that, The step of employing a trained value network to predict the evaluation value of the delivery effect of each candidate delivery strategy for the shared items within the delivery range, and determining the target delivery strategy based on the evaluation value, includes: Based on the initial deployment volume of the shared items in each of the multiple partitions divided by the deployment range, an initial candidate deployment strategy is determined. A trained value network is used to predict the initial candidate delivery strategy, and the delivery effect evaluation value corresponding to the initial candidate delivery strategy is used as a reference evaluation value. Multiple rounds of strategy updates are performed based on the initial candidate delivery strategy; Each time a strategy update is performed, the trained value network is used to predict the delivery effect of at least one candidate delivery strategy after the update. If the evaluation value of the delivery performance of at least one candidate delivery strategy after this round of updates is less than the reference evaluation value, then the evaluation value of the delivery performance of at least one candidate delivery strategy after this round of updates will be used as the reference evaluation value, and the subsequent strategy update process will continue. After each round of strategy updates is completed, the candidate delivery strategy corresponding to the reference evaluation value is taken as the target delivery strategy.

5. The method according to claim 4, characterized in that, The process of performing multiple rounds of strategy updates based on the initial candidate delivery strategy includes: For each round of policy updates, a target partition is determined from multiple partitions; The distribution volume of shared items is increased and decreased by a set step size for the target partition to obtain two candidate distribution strategies obtained in this round of strategy update.

6. The method according to claim 5, characterized in that, For each round of policy updates, determining a target partition from multiple partitions includes: For each round of strategy update, if the number of deliveries to the target partition used in the previous round is lower than the minimum or higher than the maximum, then the next partition adjacent to the target partition used in the previous round will be used as the target partition for this round. If the number of targets deployed in the previous round was not lower than the lower limit and not higher than the upper limit, then the target partitions deployed in the previous round will be used as the target partitions for this round.

7. A device for dispensing shared goods, characterized in that, include: The acquisition module is used to acquire multiple historical deployment strategies of the shared items within the deployment range, as well as the sharing records of the shared items under each historical deployment strategy. The sharing records include query records and borrowing records, and the borrowing records include borrowing operations and return operations. The determination module is used to determine the evaluation value of the delivery effect of any historical delivery strategy based on the number of sharing failures in the sharing record. The sharing failures include borrowing failures and returning failures. Borrowing failures include failing to execute the borrowing operation and not executing the borrowing operation after executing the query operation. Returning failures include failing to execute the returning operation and accumulating shared items after successfully executing the returning operation. The training module is used to train a value network based on the historical deployment strategies and their deployment effect evaluation values. The value network is a function from a state to the expected sharing failure. The state includes the distribution characteristics of shared items within the deployment range, the grid area characteristics within the deployment range, and the time slice characteristics obtained by splitting the target time period. The value network is used to represent the expected sharing failure corresponding to the state from the start of the state to the end of any time period E. The prediction module is used to use a trained value network to predict the evaluation value of the delivery effect of each candidate delivery strategy for the shared items within the delivery range, so as to determine the target delivery strategy based on the evaluation value. The delivery module is used to deliver the shared items according to the target delivery strategy.

8. The apparatus according to claim 7, characterized in that, The determining module includes: The first determining unit is used to determine, from the query records, the query operations recorded within the statistical period after the execution of the corresponding historical delivery strategy, for any given historical delivery strategy. The second determining unit is configured to determine the query operation as a failed loan operation if the client executing the query operation does not perform a loan operation after the query operation, or if the loan record determines that the client failed to perform a loan operation after the query operation, in order to determine the number of failed loan operations. The query unit is used to query the return operations recorded within the statistical period after the execution of the corresponding historical delivery strategy, based on the borrowing records, in order to determine the number of return failures. The third determining unit is used to determine the evaluation value of the delivery effect of the corresponding historical delivery strategy based on the number of loan failures and / or the number of return failures.

9. The apparatus according to claim 8, characterized in that, The query unit is also used for: Based on the lending records, query the return operations recorded within the statistical period after the execution of the corresponding historical delivery strategy; For the returned operation found, determine the partition from the multiple partitions divided by the delivery range to execute the corresponding returned operation; If the number of shared items in the corresponding partition exceeds the quantity threshold after the return operation is performed, the return operation will be determined as a return failure. Determine the number of failed returns from the returned operations retrieved.

10. The apparatus according to claim 8, characterized in that, The prediction module includes: The fourth determining unit is used to determine an initial candidate deployment strategy based on the initial deployment quantity of the shared items in each of the multiple partitions divided by the deployment range. The first prediction unit is used to predict the initial candidate delivery strategy using a trained value network, so as to use the delivery effect evaluation value corresponding to the initial candidate delivery strategy as a reference evaluation value. The first processing unit is used to perform multiple rounds of policy updates based on the initial candidate delivery strategy; The second prediction unit is used to predict the delivery effect of at least one candidate delivery strategy after the current round of strategy update using the trained value network whenever a round of strategy update is performed. The second processing unit is used to take the evaluation value of the delivery effect of at least one candidate delivery strategy after the current round of updates as the reference evaluation value and continue to execute the subsequent strategy update process if the evaluation value of the delivery effect of at least one candidate delivery strategy after the current round of updates is less than the reference evaluation value. The third processing unit is used to execute the strategy update for each round and use the candidate delivery strategy corresponding to the reference evaluation value as the target delivery strategy.

11. The apparatus according to claim 10, characterized in that, The first processing unit is further configured to: For each round of policy updates, a target partition is determined from multiple partitions; The distribution volume of shared items is increased and decreased by a set step size for the target partition to obtain two candidate distribution strategies obtained in this round of strategy update.

12. The apparatus according to claim 11, characterized in that, The first processing unit is further configured to: For each round of strategy update, if the number of deliveries to the target partition used in the previous round is lower than the minimum or higher than the maximum, then the next partition adjacent to the target partition used in the previous round will be used as the target partition for this round. If the number of targets deployed in the previous round was not lower than the lower limit and not higher than the upper limit, then the target partitions deployed in the previous round will be used as the target partitions for this round.

13. An electronic device, comprising: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to cause the at least one processor to perform the method of any one of claims 1-6.

14. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to perform the method according to any one of claims 1-6.

15. A computer program product comprising a computer program that, when executed by a processor, implements the method of any one of claims 1-6.

Images