Cargo volume prediction method, device and equipment of logistics network and computer storage medium

Abstract

The application provides a cargo volume prediction method, device and equipment of a logistics network and a computer storage medium. The cargo volume prediction method of the application comprises the following steps: receiving a cargo volume prediction request of a logistics network, obtaining a target site identifier corresponding to the cargo volume prediction request of the logistics network and target cargo volume information associated with the target site identifier; processing the target cargo volume information through a single-site long short-term memory network prediction model in a preset prediction model to obtain a first prediction value; processing the target cargo volume information through a full-site long short-term memory network prediction model in the preset prediction model to obtain a second prediction value; determining the prediction weights of the first prediction value and the second prediction value respectively, weighting the first prediction value and the second prediction value according to the prediction weights respectively to obtain a cargo volume prediction value; and the cargo volume prediction is performed through weighting of the preset prediction model in the application, so that the cargo volume prediction is more accurate.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of machine learning, in particular to a logistics network cargo volume prediction method, device, equipment and computer storage medium. BACKGROUND

[0002] With the development of economy, the logistics market also ushered in rapid growth, the rapid expansion of logistics scale and the increasingly complete and complex logistics network are important manifestations. As one of the important reference means for logistics site pre-resource planning, cargo volume prediction becomes increasingly important.

[0003] Current cargo volume prediction is mostly achieved through prediction models, such as time series models such as exponential smoothing models, or machine learning models such as random forests; such cargo volume prediction methods often make separate predictions for different sites in the logistics network, making it difficult to capture and utilize the information that these sites are interconnected and influence each other, and the average prediction accuracy of all sites in the entire logistics network is also not satisfactory, which does not meet the needs of digital and fine collaborative management of logistics sites in the logistics network. SUMMARY

[0004] The present application provides a logistics network cargo volume prediction method, device, equipment and computer storage medium, aiming to solve the technical problem that the existing cannot accurately predict the cargo volume information of all sites in the logistics network at the same time, leading to difficulty in realizing fine collaborative management of the logistics network.

[0005] In one aspect, the present application provides a logistics network cargo volume prediction method, which comprises the following steps:

[0006] Receiving a cargo volume prediction request of a logistics network, obtaining a target site identifier corresponding to the cargo volume prediction request of the logistics network, and target cargo volume information associated with the target site identifier;

[0007] Processing the target cargo volume information through a single-site long short-term memory network prediction model in a preset prediction model to obtain a first prediction value, and processing the target cargo volume information through a full-site long short-term memory network prediction model in the preset prediction model to obtain a second prediction value;

[0008] Determining the prediction weights of the first prediction value and the second prediction value respectively, weighting the first prediction value and the second prediction value according to the respective prediction weights to obtain a cargo volume prediction value.

[0009] In another aspect, the present application also provides a logistics network cargo volume prediction device, which comprises:

[0010] The request receiving module is configured to receive a cargo volume prediction request of a logistics network, obtain a target site identifier corresponding to the cargo volume prediction request of the logistics network, and obtain target cargo volume information associated with the target site identifier.

[0011] The processing prediction module is configured to process the target cargo volume information by using a single-site long short-term memory network prediction model in a preset prediction model to obtain a first prediction value, and process the target cargo volume information by using a full-site long short-term memory network prediction model in the preset prediction model to obtain a second prediction value.

[0012] The weighting determination module is configured to determine a prediction weight of each of the first prediction value and the second prediction value, and weight the first prediction value and the second prediction value according to the prediction weight of each of the first prediction value and the second prediction value to obtain a cargo volume prediction value.

[0013] In another aspect, the present application further provides a cargo volume prediction device of a logistics network, which comprises:

[0014] One or more processors;

[0015] A memory; and

[0016] One or more application programs, wherein the one or more application programs are stored in the memory and configured to be executed by the processor to implement the cargo volume prediction method of the logistics network.

[0017] In another aspect, the present application further provides a computer storage medium having a computer program stored thereon, wherein the computer program is loaded by a processor to execute the steps in the cargo volume prediction method of the logistics network.

[0018] The logistics network cargo volume prediction method provided in the application comprises the following steps: receiving a logistics network cargo volume prediction request, obtaining a target site identifier corresponding to the logistics network cargo volume prediction request and target cargo volume information associated with the target site identifier, processing the target cargo volume information through a single-site long short-term memory network prediction model in a preset prediction model to obtain a first prediction value, processing the target cargo volume information through a full-site long short-term memory network prediction model in the preset prediction model to obtain a second prediction value, determining prediction weights of the first prediction value and the second prediction value respectively, and weighting the first prediction value and the second prediction value according to the prediction weights respectively to obtain a cargo volume prediction value. In the embodiment of the application, the preset prediction model is used for cargo volume prediction through the single-site long short-term memory network prediction model and the full-site long short-term memory network prediction model in the preset prediction model to obtain the first prediction value and the second prediction value. Further, the prediction weights of the first prediction value and the second prediction value are set for weighting processing considering the mutual association and influence information between various stations, and the final cargo volume prediction value is obtained. Such a cargo volume prediction method makes the cargo volume prediction more accurate, so as to realize digital and fine management of the stations in the logistics network. BRIEF DESCRIPTION OF DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative effort on the basis of these drawings.

[0020] Figure 1 is a scene schematic diagram of the logistics network cargo volume prediction provided in the embodiment of the application;

[0021] Figure 2 is an embodiment flowchart of constructing a preset prediction model in the logistics network cargo volume prediction method in the embodiment of the application;

[0022] Figure 3 is an embodiment flowchart of the logistics network cargo volume prediction method in the embodiment of the application;

[0023] Figure 4 is an embodiment flowchart of prompting the preset prediction model update in the logistics network cargo volume prediction method provided in the embodiment of the application;

[0024] Figure 5 is an embodiment flowchart of cargo volume prediction report analysis in the logistics network cargo volume prediction method provided in the embodiment of the application;

[0025] Figure 6is an embodiment structure schematic diagram of a cargo volume prediction device of a logistics network provided in the embodiment of the present application.

[0026] Figure 7 is an embodiment structure schematic diagram of a cargo volume prediction device of a logistics network provided in the embodiment of the present application. DETAILED DESCRIPTION

[0027] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of the present application.

[0028] In the description of the present application, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and therefore cannot be understood as indicating or implying that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present application. In addition, the terms "first", "second" are only for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the technical features indicated. Therefore, the features limited by "first", "second" can explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of "a plurality of" is two or more, unless otherwise explicitly and specifically limited.

[0029] In the present application, the word "exemplary" is used to mean "serving as an example, instance, or illustration." Any implementation described as "exemplary" in the present application is not necessarily to be construed as preferred or advantageous over other implementations. The following description is presented to enable any person skilled in the art to make and use the present application. In the following description, for the purposes of explanation, numerous details are set forth. It should be appreciated that the present application can be practiced in other embodiments that depart from these specific details. In other instances, detailed descriptions of well-known structures and processes are not presented in order to avoid obscuring the description of the present application. Thus, the present application is not intended to be limited to the embodiments shown, but is to be accorded the widest scope consistent with the principles and features presented herein.

[0030] The embodiments of the present application provide a logistics network cargo volume prediction method, device, equipment and computer storage medium, which are described in detail below.

[0031] The logistics network cargo volume prediction method in the embodiment of the application is applied to a logistics network cargo volume prediction device. The logistics network cargo volume prediction device is arranged in a logistics network cargo volume prediction equipment. The logistics network cargo volume prediction equipment is provided with one or more processors, a memory, and one or more application programs. The one or more application programs are stored in the memory and configured to be executed by the processor to implement the logistics network cargo volume prediction method. The logistics network cargo volume prediction equipment can be a terminal such as a mobile phone or a tablet computer. The logistics network cargo volume prediction equipment can also be a server or a service cluster composed of multiple servers.

[0032] As shown in Figure 1 , the logistics network cargo volume prediction scenario of the embodiment of the application includes a logistics network cargo volume prediction equipment 100 (the logistics network cargo volume prediction equipment 100 is integrated with a logistics network cargo volume prediction device). The logistics network cargo volume prediction equipment 100 runs a computer storage medium corresponding to the logistics network cargo volume prediction to perform the steps of the logistics network cargo volume prediction. Figure 1 It can be understood that

[0033] the logistics network cargo volume prediction equipment in the logistics network cargo volume prediction scenario or the device included in the logistics network cargo volume prediction equipment does not constitute a limitation on the embodiment of the application, that is, the number and types of devices included in the logistics network cargo volume prediction scenario or the number and types of devices included in each device do not affect the overall implementation of the technical solution in the embodiment of the application and can be regarded as equivalent replacements or derivatives of the technical solution claimed in the embodiment of the application. Figure 1 The logistics network cargo volume prediction equipment 100 in the embodiment of the application is mainly used to receive a logistics network cargo volume prediction request, obtain a target site identifier corresponding to the logistics network cargo volume prediction request and target cargo volume information associated with the target site identifier, process the target cargo volume information through a single-site long short-term memory network prediction model in a preset prediction model to obtain a first prediction value, process the target cargo volume information through a full-site long short-term memory network prediction model in the preset prediction model to obtain a second prediction value, determine prediction weights of the first prediction value and the second prediction value respectively, and weight the first prediction value and the second prediction value according to the respective prediction weights to obtain a cargo volume prediction value.

[0034]

[0035] ​The logistics network cargo volume prediction device 100 in the embodiment of the application can be a logistics network cargo volume prediction device, a logistics network cargo volume prediction device network or a logistics network cargo volume prediction device cluster composed of a plurality of logistics network cargo volume prediction devices. For example, the logistics network cargo volume prediction device 100 described in the embodiment of the application includes but is not limited to a computer, a network host, a single network logistics network cargo volume prediction device, a plurality of network logistics network cargo volume prediction devices, or a cloud logistics network cargo volume prediction device composed of a plurality of logistics network cargo volume prediction devices. The cloud logistics network cargo volume prediction device is composed of a large number of computers or network logistics network cargo volume prediction devices based on cloud computing.

[0036] Those skilled in the art can understand that, Figure 1 The application environment shown in the embodiment of the application is only one application scenario of the application scheme, and does not constitute a limitation on the application scenarios of the application scheme. Other application environments can include more or fewer logistics network cargo volume prediction devices or network connection relationships of logistics network cargo volume prediction devices than those shown in the embodiment of the application. Figure 1 The application environment shown in the embodiment of the application is only one application scenario of the application scheme, and does not constitute a limitation on the application scenarios of the application scheme. Other application environments can include more or fewer logistics network cargo volume prediction devices or network connection relationships of logistics network cargo volume prediction devices than those shown in the embodiment of the application. Figure 1 The application environment shown in the embodiment of the application is only one application scenario of the application scheme, and does not constitute a limitation on the application scenarios of the application scheme. Other application environments can include more or fewer logistics network cargo volume prediction devices or network connection relationships of logistics network cargo volume prediction devices than those shown in the embodiment of the application.

[0037] In addition, the logistics network cargo volume prediction device 100 in the logistics network cargo volume prediction scenario of the application can be provided with a display device, or the logistics network cargo volume prediction device 100 is not provided with a display device and is in communication connection with an external display device 200. The display device 200 is used to output the result of the execution of the logistics network cargo volume prediction method in the logistics network cargo volume prediction device. The logistics network cargo volume prediction device 100 can access a background database 300 (the background database can be a local storage of the logistics network cargo volume prediction device, and the background database can also be set in the cloud). The background database 300 stores information related to the logistics network cargo volume prediction.

[0038] It should be noted that, Figure 1 The logistics network cargo volume prediction scenario shown in the embodiment of the application is only one example. The logistics network cargo volume prediction scenario described in the embodiment of the application is used to more clearly illustrate the technical scheme of the embodiment of the application, and does not constitute a limitation on the technical scheme provided by the embodiment of the application.

[0039] Based on the above logistics network cargo volume prediction scenario, an embodiment of a logistics network cargo volume prediction method is provided. The logistics network cargo volume prediction method in the embodiment includes:

[0040] receiving a cargo volume prediction request of a logistics network, obtaining a target site identifier corresponding to the cargo volume prediction request of the logistics network, and target cargo volume information associated with the target site identifier;

[0041] processing the target cargo volume information through a single-site long short-term memory network prediction model in a preset prediction model to obtain a first prediction value, and processing the target cargo volume information through a full-site long short-term memory network prediction model in the preset prediction model to obtain a second prediction value;

[0042] determining prediction weights of the first prediction value and the second prediction value respectively, weighting the first prediction value and the second prediction value according to the prediction weights respectively to obtain a cargo volume prediction value.

[0043] In the logistics network cargo volume prediction method of the embodiment, the logistics network cargo volume prediction is realized based on a preset prediction model. The preset prediction model includes a single-site long short-term memory network prediction model corresponding to each site and a full-site long short-term memory network prediction model of the logistics network. The preset prediction model is combined through the single-site long short-term memory network prediction model and the full-site long short-term memory network prediction model for comprehensive prediction, so that the cargo volume prediction is more accurate. In the embodiment, the preset prediction model is constructed before the logistics network cargo volume prediction method is executed. Specifically:

[0044] As shown in Figure 2 , the preset prediction model is constructed in the logistics network cargo volume prediction method of the embodiment. Figure 2 is an embodiment flowchart of constructing the preset prediction model in the logistics network cargo volume prediction method of the embodiment.

[0045] The step of constructing the preset prediction model in the logistics network cargo volume prediction method of the embodiment includes 201-205.

[0046] 201, receiving a prediction model training instruction, collecting historical cargo volume information of each site in the logistics network, taking the historical cargo volume information as training samples and dividing them into a training set and a test set, wherein the training set includes a training subset of each site, and the test set includes a test subset of each site.

[0047] The logistics network cargo volume prediction method in the embodiment is applied to a logistics network cargo volume prediction device. The type of the logistics network cargo volume prediction device is not specifically limited. The logistics network cargo volume prediction device can be a server or a terminal.

[0048] The cargo volume prediction device receives a prediction model training instruction, where the triggering manner of the prediction model training instruction is not specifically limited, that is, the prediction model training instruction can be triggered by a user, for example, the user inputs the key information "model training" in the display interface of the cargo volume prediction device to trigger the prediction model training instruction; in addition, the prediction model training instruction can also be triggered by the cargo volume prediction device, for example, the cargo volume prediction device is pre-configured to automatically trigger the prediction model training instruction when the training sample is updated, and the cargo volume prediction device monitors the training sample, and the cargo volume prediction device automatically triggers the prediction model training instruction when detecting that the training sample is updated.

[0049] After receiving the prediction model training instruction, the cargo volume prediction device collects historical cargo volume information of each site in the logistics network, where the historical cargo volume information refers to the cargo volume of each site in a historical time period and time information corresponding to the cargo volume of each site, for example, the historical cargo volume information of a site numbered 1 in the logistics network is xxx tons at xxx time.

[0050] The cargo volume prediction device takes the historical cargo volume information as a training sample, and divides each historical cargo volume information into a training set and a test set, that is, the training set and the test set can be divided according to a proportion, for example, the proportion of the training set to the test set is 9:1, and the cargo volume prediction device divides each historical cargo volume information according to time intervals to divide 9:1 in turn, and in this embodiment, the training set contains training subsets of each site, and the test set contains test subsets of each site.

[0051] 202, a single-site long short-term memory network prediction model is established according to the training subsets of each site, and a full-site long short-term memory network prediction model is established according to the training set of the full site.

[0052] The cargo volume prediction device establishes a single-site long short-term memory network prediction model according to the training subsets of each site, that is, the cargo volume prediction device respectively trains the training subsets of each site in the training set to obtain a single-site long short-term memory network prediction model; the cargo volume prediction device establishes a full-site long short-term memory network prediction model according to the training set of the full site; specifically, including:

[0053] (1) sorting the training subsets of each site according to historical time sequence to form historical time sequences of each site, determining input variables based on the corresponding week date attribution of the historical time sequences of each site and whether it is a weekday, and training a preset long short-term memory network model through the input variables of each site to obtain a single-site long short-term memory network prediction model of each site;

[0054] (2), aligning the training set of the logistics network according to the time sequence according to the date to form a multi-dimensional time sequence, forming a two-dimensional matrix data by the two-dimensional matrix data, training the long short-term memory network model, and obtaining the long short-term memory network prediction model of the whole site of the logistics network.

[0055] 203, by the single-site long short-term memory network prediction model and the whole-site long short-term memory network prediction model respectively processing each site of the test set in the test set, obtaining the first prediction value and the second prediction value.

[0056] The cargo volume prediction device processes each test subset in the test set by the single-site long short-term memory network prediction model and the whole-site long short-term memory network prediction model respectively, and obtains the first prediction value and the second prediction value, that is, in this embodiment, each test subset is tested by the single-site long short-term memory network prediction model and the whole-site long short-term memory network prediction model to obtain the first prediction value and the second prediction value corresponding to each site.

[0057] 204, according to the first prediction value and the second prediction value and the actual value on the test set of the corresponding date, linear regression processing is performed to obtain a regression model, and the prediction weight of the first prediction value and the second prediction value is determined according to the regression model.

[0058] The cargo volume prediction device performs linear regression processing according to the first prediction value and the second prediction value, obtains a regression model, and determines the prediction weight of the first prediction value and the second prediction value according to the regression model. Step 204 specifically includes:

[0059] (1), for each site, extracting the actual value on the test set of the corresponding date of the first prediction value and the second prediction value in each test subset;

[0060] (2), the first prediction value and the second prediction value are used as explanatory variables, the first actual value and the second actual value are used as target variables, and the explanatory variables and the target variables are linearly regressed to obtain a regression model;

[0061] (3), obtaining the regression coefficient of the regression model, and taking the regression coefficient as the prediction weight corresponding to the first prediction value and the second prediction value.

[0062] That is, in this embodiment, the cargo volume prediction device performs linear regression to determine the weight of the first prediction value and the second prediction value, so that the cargo volume prediction considers different factors and performs weighted processing, so that the cargo volume prediction result is more accurate.

[0063] 205 , encapsulate the single-site LSTM network prediction model, the full-site LSTM network prediction model, and the prediction weights to form a preset prediction model.

[0064] The cargo volume prediction device encapsulates the single-site long-short-term memory network prediction model, the full-site long-short-term memory network prediction model and the prediction weight to form a preset prediction model. In this embodiment, the model is pre-trained based on the historical cargo volume information of each site to form a preset prediction model. The single-site long-short-term memory network prediction model makes independent predictions based on the data of each site, and can fully learn the time-varying patterns of its own cargo volume, so as to better capture and predict its own cargo volume fluctuations; the full-site long-short-term memory network prediction model is based on the data of all sites in the logistics network as model input. The model learns the spatial correlation between the cargo volumes of all sites in the logistics network by simultaneously training the data of each site, thereby mining and using the mutual correlation and influence of the cargo volume information between the sites. Specifically, the cargo volume forecasting device models all the stations in the logistics network one by one and jointly models them based on the preset long-short-term memory network model. By effectively utilizing the time and space correlation information of the data of different stations in the logistics network, it gives each logistics station a prediction value based on the modeling dimensions of time correlation and space correlation as the intermediate value, and then appropriately weights it to generate the final logistics network cargo volume forecast value, making the cargo volume forecast more accurate.

[0065] For ease of understanding, this embodiment provides specific steps for constructing a preset prediction model, including:

[0066] Step 1: The cargo volume forecasting device collects historical cargo volume information of all logistics sites in the logistics network to be forecasted based on forecasting needs, and forms time series data of the original cargo volume of all logistics sites;

[0067] Step 2: The cargo volume forecasting device divides the time series data of cargo volume at all logistics sites into a training set and a test set;

[0068] Step 3: The cargo volume forecasting device sorts the training subset for each logistics site by time to form a historical time series. Based on the input variables of each site's historical time series, the corresponding day of the week, and whether the time series is formed on a weekday, the cargo volume forecasting device uses a long short-term memory network model for training. It then generates a unique long short-term memory network model for each site, also known as a single-site long short-term memory network prediction model.

[0069] Step four: the training set of the cargo volume prediction device is aligned by date to form a two-dimensional matrix data composed of multi-dimensional time series, and the cargo volume prediction device trains the overall long short-term memory network model through the two-dimensional matrix data to give the long short-term memory network model for the cargo volume prediction of the entire logistics network, also called the full-site long short-term memory network prediction model;

[0070] Step five: the cargo volume prediction device applies the models trained in steps three and four to the test set for each station to obtain the prediction values of each station according to the separate long short-term memory network model prediction model and the prediction values according to the overall long short-term memory network model prediction model, which are respectively denoted as the first prediction value and the second prediction value; for each station, the two types of prediction values generated are used as the explanatory variables, and the true values in the test set are used as the target variables to perform linear regression, and the regression coefficients of the obtained regression model are used as the prediction weights of the first prediction value and the second prediction value of each station in the final prediction value; at this time, the fitting value on the regression model is used as the cargo volume prediction value on the test set; the actual cargo volume value of the target site at the time corresponding to the cargo volume prediction value is obtained, and a ratio operation is performed between the actual cargo volume value and the cargo volume prediction value to obtain a prediction deviation value; if the prediction deviation value exceeds a preset deviation range, a prompt information is output to prompt updating the preset prediction model.

[0071] (6) Based on historical data, the long short-term memory network models obtained in steps three and four are applied to each station to obtain the first prediction value and the second prediction value of each station, and the first prediction value and the second prediction value of each station are weighted according to the prediction weights of the two types of prediction values of each station obtained in step five to obtain the final prediction value of the cargo volume of each station.

[0072] It can be understood that:

[0073] In step one, the historical cargo volume information includes the sorting site code, the sorting site operation of the waybill and its cargo volume, and the time of the waybill in the sorting site operation. The total waybill cargo volume can be aggregated by site and day dimension (or finer time dimension) according to the sorting site code and the operation time in the sorting site, and the statistical results are sorted in chronological order to obtain a time series composed of daily operation cargo volume records of each sorting site before the predicted day, and finally obtain the original time series data of all sites;

[0074] In step two, the training set and the test set of the time series data of the cargo volume of all logistics sites are divided by forward division of the time series, and by selecting appropriate training set and test set sizes and the step length of the input historical data required for long short-term memory network model training, specific division time nodes can be obtained;

[0075] The specific method for training the long short-term memory network model for each logistics site in step three on the training set based on the historical time series of each site and the input variables such as the week date attribution and whether it is a weekday time series is as follows: first, generate the time series data required for training the long short-term memory network model for each site: select the historical cargo volume of the given site during the period of the training set, and generate the week date attribution time series corresponding to the time series according to the date, (specifically, if it is Monday, it is recorded as 1, and other dates such as Tuesday, Wednesday to Sunday are sequentially similar), then generate the whether it is a weekday time series corresponding to the time series (if it is a weekday, it is recorded as 1; if it is a non-working day, it is recorded as 0), if there are other related time series variables, they can also be similarly introduced into the model as needed; then the above three or more time series are used as the input feature data of the long short-term memory network model prediction model for each site, and the long short-term memory network model prediction model for each site is trained;

[0076] The specific method for training the long short-term memory network model for each logistics site in step three on the training set based on the historical time series of each site and the input variables such as the week date attribution and whether it is a weekday time series is as follows: first, generate the time series data required for training the long short-term memory network model for each site: select the historical cargo volume of the given site during the period of the training set, and generate the week date attribution time series corresponding to the time series according to the date, (specifically, if it is Monday, it is recorded as 1, and other dates such as Tuesday, Wednesday to Sunday are sequentially similar), then generate the whether it is a weekday time series corresponding to the time series (if it is a weekday, it is recorded as 1; if it is a non-working day, it is recorded as 0), if there are other related time series variables, they can also be similarly introduced into the model as needed; then the above three or more time series are used as the input feature data of the long short-term memory network model prediction model for each site, and the long short-term memory network model prediction model for each site is trained;

[0077] The specific method for training the long short-term memory network model for each logistics site in step three on the training set based on the historical time series of each site and the input variables such as the week date attribution and whether it is a weekday time series is as follows: first, generate the time series data required for training the long short-term memory network model for each site: select the historical cargo volume of the given site during the period of the training set, and generate the week date attribution time series corresponding to the time series according to the date, (specifically, if it is Monday, it is recorded as 1, and other dates such as Tuesday, Wednesday to Sunday are sequentially similar), then generate the whether it is a weekday time series corresponding to the time series (if it is a weekday, it is recorded as 1; if it is a non-working day, it is recorded as 0), if there are other related time series variables, they can also be similarly introduced into the model as needed; then the above three or more time series are used as the input feature data of the long short-term memory network model prediction model for each site, and the long short-term memory network model prediction model for each site is trained;

[0078] The specific method for training the long short-term memory network model for each logistics site in step three on the training set based on the historical time series of each site and the input variables such as the week date attribution and whether it is a weekday time series is as follows: first, generate the time series data required for training the long short-term memory network model for each site: select the historical cargo volume of the given site during the period of the training set, and generate the week date attribution time series corresponding to the time series according to the date, (specifically, if it is Monday, it is recorded as 1, and other dates such as Tuesday, Wednesday to Sunday are sequentially similar), then generate the whether it is a weekday time series corresponding to the time series (if it is a weekday, it is recorded as 1; if it is a non-working day, it is recorded as 0), if there are other related time series variables, they can also be similarly introduced into the model as needed; then the above three or more time series are used as the input feature data of the long short-term memory network model prediction model for each site, and the long short-term memory network model prediction model for each site is trained;

[0079] In the embodiment of the present application, a prediction model is preset, and cargo volume prediction is performed through the single-site long-short-term memory network prediction model and the full-site long-short-term memory network prediction model in the preset prediction model to obtain a first prediction value and a second prediction value. Furthermore, considering the correlation information and influence information between each station, the prediction weights of the first prediction value and the second prediction value are set for weighted processing to obtain the final cargo volume prediction value. This cargo volume prediction method makes the cargo volume prediction more accurate, so as to realize the digitalization and refined management of sites in the logistics network.

[0080] like Figure 3 As shown, Figure 3 This is a flow chart of an embodiment of a method for predicting cargo volume in a logistics network in an embodiment of the present application.

[0081] In some embodiments of the present application, a method for predicting cargo volume in a logistics network includes the following steps 301 to 303:

[0082] 301 : Receive a cargo quantity prediction request from a logistics network, and obtain a target site identifier corresponding to the cargo quantity prediction request from the logistics network, and target cargo quantity information associated with the target site identifier.

[0083] The cargo volume prediction device receives the cargo volume prediction request from the logistics network, wherein the triggering method of the cargo volume prediction request is not specifically limited, that is, the cargo volume prediction request can be actively triggered by the user, for example, the user enters: "xxx site" in the display interface of the cargo volume prediction device and clicks the cargo volume prediction button to actively trigger the cargo volume prediction request; in addition, the cargo volume prediction request can also be automatically triggered by the cargo volume prediction device, for example, the cargo volume prediction device is pre-set to automatically trigger the cargo volume prediction request at dawn every day, then the cargo volume prediction device automatically triggers the cargo volume prediction request when it detects the time at dawn every day.

[0084] The cargo volume prediction device obtains the target site identifier corresponding to the cargo volume prediction request of the logistics network, wherein the target site identifier refers to identification information that uniquely identifies the site, such as the site number, site location information, etc. The cargo volume prediction device obtains the target cargo volume information associated with the target site identifier, and the target cargo volume information refers to information related to the cargo volume prediction, such as historical cargo volume and the time corresponding to the historical cargo volume.

[0085] 302. Process the target cargo quantity information through the single-site long short-term memory network prediction model in the preset prediction model to obtain a first prediction value, and process the target cargo quantity information through the full-site long short-term memory network prediction model in the preset prediction model to obtain a second prediction value.

[0086] The cargo volume prediction device inputs target cargo volume information into a preset prediction model, processes the target cargo volume information through a single-site long short-term memory network prediction model in the preset prediction model, and obtains a first prediction value; further, the cargo volume prediction device processes the target cargo volume information through a full-site long short-term memory network prediction model in the preset prediction model, and obtains a second prediction value; step 302 specifically includes:

[0087] (1) inputting the target cargo volume information into a preset prediction model, extracting first feature information in the target cargo volume information through a single-site long short-term memory network prediction model in the preset prediction model, and processing to obtain a first prediction value;

[0088] (2) extracting second feature information in the target cargo volume information through a full-site long short-term memory network prediction model in the preset prediction model, and processing to obtain a second prediction value.

[0089] That is, in the embodiment, the cargo volume prediction device inputs the target cargo volume information into a preset prediction model, extracts first feature information in the target cargo volume information through a single-site long short-term memory network prediction model in the preset prediction model, and processes to obtain a first prediction value; wherein the first feature information is time feature information and cargo information feature information obtained after the single-site long short-term memory network prediction model pre-processes the target cargo volume information; the cargo volume prediction device extracts second feature information in the target cargo volume information through a full-site long short-term memory network prediction model in the preset prediction model, and processes to obtain a second prediction value; wherein the second feature information is time feature information and cargo information feature information obtained after the full-site long short-term memory network prediction model pre-processes the target cargo volume information.

[0090] 303, determining the prediction weight of each of the first prediction value and the second prediction value, weighting the first prediction value and the second prediction value according to the prediction weight of each, and obtaining a cargo volume prediction value.

[0091] The cargo volume prediction device determines the prediction weight of each of the first prediction value and the second prediction value, that is, the cargo volume prediction device takes the coefficient of the regression model as the prediction weight, and the cargo volume prediction device weights the first prediction value and the second prediction value according to the prediction weight of each, and obtains a cargo volume prediction value.

[0092] The preset prediction model in the embodiment of the application is used for cargo volume prediction through a single-site long short-term memory network prediction model and a full-site long short-term memory network prediction model in the preset prediction model to obtain a first prediction value and a second prediction value. Further, considering the associated information and influence information between each station, the first prediction value and the second prediction value are weighted by setting respective prediction weights of the first prediction value and the second prediction value for weighted processing to obtain a final cargo volume prediction value. Such a cargo volume prediction manner makes the cargo volume prediction more accurate to realize digitalization and fine management of the stations in the logistics network.

[0093] The prediction algorithm based on the long short-term memory neural network model in the embodiment of the application simultaneously mines and effectively models the spatial correlation and the temporal correlation between different stations in the logistics network. First, the long short-term memory network model is used to model each station separately to give a first prediction value of each station from the modeling dimension of temporal dependence. Then, the long short-term memory network model is used to jointly model all stations in the logistics network to give a second prediction value of each station from the modeling dimension of spatial dependence. Finally, the spatiotemporal correlation information in the historical data of the stations is fully mined and utilized in the same prediction framework through appropriate weighting, so that the prediction of each station in the logistics network not only has interpretability but also has good robustness.

[0094] Compared with existing algorithms applied to complex logistics networks, the logistics network cargo volume prediction method based on the long short-term memory neural network in the embodiment of the application has obvious improvement in the average prediction accuracy of the cargo volume of all stations in the network, thereby better promoting the support of cargo volume prediction for front-end resource planning in the complex logistics network scenario and being conducive to fine management of different stations in the logistics network.

[0095] Reference Figure 4 , Figure 4 FIG. 1 is a flowchart of an embodiment of the logistics network cargo volume prediction method provided in the embodiment of the application.

[0096] In some embodiments of the application, the cargo volume prediction device prompting the preset prediction model update includes the following steps 401-402.

[0097] 401, obtain the actual cargo volume value of the target station corresponding to the target station identifier at the time corresponding to the cargo volume prediction value, and perform ratio operation on the actual cargo volume value and the cargo volume prediction value to obtain a prediction deviation value.

[0098] The cargo volume prediction device obtains an actual cargo volume value of a target site corresponding to the target site identification at a time corresponding to the cargo volume prediction value, that is, the cargo volume prediction device obtains the actual cargo volume value at the time corresponding to the cargo volume prediction value after obtaining the cargo volume prediction value, and performs ratio operation on the actual cargo volume value and the cargo volume prediction value to obtain a prediction deviation value.

[0099] The cargo volume prediction device has a preset deviation range, which is an allowed error range. The preset deviation range can be set according to a specific scenario. For example, the preset deviation range is 0.8-1.2. The cargo volume prediction device compares the prediction deviation value with the preset deviation range to determine whether the prediction deviation value exceeds the preset deviation range. If the prediction deviation value does not exceed the preset deviation range, no processing is performed.

[0100] 402. If the prediction deviation value exceeds the preset deviation range, prompt information is output to prompt updating of the preset prediction model.

[0101] If the prediction deviation value exceeds the preset deviation range, the cargo volume prediction device outputs prompt information to prompt updating of the preset prediction model. In this embodiment, the cargo volume prediction device compares the cargo volume prediction value with the actual cargo volume value to determine the prediction deviation value. The cargo volume prediction device determines whether to update the preset prediction model according to the size of the prediction deviation value to ensure the accuracy of the prediction of the preset prediction model.

[0102] Referring to Figure 5 , Figure 5 FIG. 1 is a flowchart of an embodiment of cargo volume prediction report analysis in a cargo volume prediction method of a logistics network provided in this embodiment.

[0103] In some embodiments of this application, it is specified that the visual display of the cargo volume prediction value further includes the following steps 501-502:

[0104] 501. The cargo volume prediction value is converted into a line graph, and the line graph is added to a preset cargo volume analysis template to generate a cargo volume prediction report of the logistics network.

[0105] The cargo volume prediction device converts the cargo volume prediction value into a line graph, that is, the cargo volume prediction device sets the horizontal coordinate of the line graph as time and the vertical coordinate as the cargo volume prediction value, and inputs the cargo volume information and the corresponding time into the coordinates to form the line graph.

[0106] The cargo volume prediction device includes a preset cargo volume analysis template, which includes analysis information of different cargo volume changes. The cargo volume prediction device adds the line graph to the preset cargo volume analysis template to generate a cargo volume prediction report of the logistics network.

[0107] 502. Send the cargo volume forecast report to a preset terminal, so that a user corresponding to the preset terminal can view cargo volume forecast analysis information.

[0108] The cargo volume forecasting device sends the cargo volume forecast report to a pre-set terminal, which is a terminal corresponding to a pre-set cargo volume analyst, so that the user corresponding to the pre-set terminal can view the cargo volume forecast analysis information. In this embodiment, the cargo volume forecasting device outputs relevant information of the cargo volume forecast in a visual manner and sends it to the corresponding terminal, which can realize the supervision of cargo volume forecasts, enable the cargo volume forecast results to be used in actual production, and facilitate user operation.

[0109] like Figure 6 As shown, Figure 6 The figure is a schematic diagram of the structure of an embodiment of a cargo volume prediction device for a logistics network.

[0110] In order to better implement the method for predicting cargo volume in a logistics network in the embodiment of the present application, based on the method for predicting cargo volume in a logistics network, the embodiment of the present application further provides a device for predicting cargo volume in a logistics network. The device for predicting cargo volume in a logistics network includes the following modules 601-603:

[0111] The request receiving module 601 is configured to receive a cargo quantity forecast request from a logistics network, obtain a target site identifier corresponding to the cargo quantity forecast request from the logistics network, and obtain target cargo quantity information associated with the target site identifier;

[0112] A processing and prediction module 602 is configured to process the target cargo volume information using a single-site long short-term memory network prediction model in a preset prediction model to obtain a first prediction value, and to process the target cargo volume information using a full-site long short-term memory network prediction model in the preset prediction model to obtain a second prediction value;

[0113] The weighted determination module 603 is used to determine the respective prediction weights of the first prediction value and the second prediction value, and weight the first prediction value and the second prediction value according to their respective prediction weights to obtain a cargo volume prediction value.

[0114] In some embodiments of the present application, the cargo volume prediction device for the logistics network includes:

[0115] receiving a prediction model training instruction, collecting historical cargo volume information of each site in the logistics network, using the historical cargo volume information as training samples and dividing the data into a training set and a test set, wherein the training set includes a training subset of each site, and the test set includes a test subset of each site;

[0116] A single-site LSTM prediction model is established based on the training subset of each site, and a full-site LSTM prediction model is established based on the training set of all sites.

[0117] obtaining a first prediction value and a second prediction value by processing each test subset of each site in the test set by the single-site long short-term memory network prediction model and the all-site long short-term memory network prediction model respectively;

[0118] obtaining a regression model by performing linear regression processing on the first prediction value, the second prediction value and actual values on the test set of the corresponding date, and determining a prediction weight of the first prediction value and the second prediction value respectively according to the regression model;

[0119] encapsulating the single-site long short-term memory network prediction model, the all-site long short-term memory network prediction model and the prediction weight to form a preset prediction model.

[0120] In some embodiments of the present application, the establishment of the single-site long short-term memory network prediction model according to the training subset of each site and the establishment of the all-site long short-term memory network prediction model according to the training set of the logistics network in the logistics network cargo volume prediction device comprises:

[0121] sorting the training subset of each site according to the historical time sequence to form a historical time sequence of each site, determining input variables based on the corresponding week date belonging and whether it is a weekday of the historical time sequence of each site, and training a preset long short-term memory network model through the input variables of each site to obtain a single-site long short-term memory network prediction model of each site;

[0122] aligning the training set of the logistics network according to the time sequence according to the date to form a multi-dimensional time sequence, forming a two-dimensional matrix data composed of the multi-dimensional time sequence, training a preset long short-term memory network model through the two-dimensional matrix data to obtain an all-site long short-term memory network prediction model of the logistics network.

[0123] In some embodiments of the present application, the linear regression processing on the first prediction value, the second prediction value and the actual values on the test set of the corresponding date to obtain a regression model, and the determination of the prediction weight of the first prediction value and the second prediction value according to the regression model in the logistics network cargo volume prediction device comprises:

[0124] for each site, extracting the first actual value corresponding to the first prediction value in each test subset and the second actual value corresponding to the second prediction value on the test set of the corresponding date;

[0125] taking the actual value as a target variable, and performing linear regression on the explanatory variable and the target variable to obtain a regression model;

[0126] obtaining a regression coefficient of the regression model, and taking the regression coefficient as the prediction weight corresponding to the first prediction value and the second prediction value.

[0127] In some embodiments of the present application, the processing prediction module 602 comprises:

[0128] The target cargo volume information is input into a preset prediction model, first feature information in the target cargo volume information is extracted by a single-site long short-term memory network prediction model in the preset prediction model, and a first prediction value is obtained by processing;

[0129] Second feature information in the target cargo volume information is extracted by a full-site long short-term memory network prediction model in the preset prediction model, and a second prediction value is obtained by processing.

[0130] In some embodiments of the present application, the cargo volume prediction device of the logistics network comprises:

[0131] The actual cargo volume value of the target site corresponding to the target site identifier at the time corresponding to the cargo volume prediction value is obtained, the actual cargo volume value is subjected to ratio operation with the cargo volume prediction value, and a prediction deviation value is obtained;

[0132] If the prediction deviation value exceeds a preset deviation range, a prompt information is output to prompt updating the preset prediction model.

[0133] In some embodiments of the present application, the cargo volume prediction device of the logistics network comprises:

[0134] The cargo volume prediction value is converted into a line graph, the line graph is added to a preset cargo volume analysis template, and a cargo volume prediction report of the logistics network is generated;

[0135] The cargo volume prediction report is sent to a preset terminal, so that a user corresponding to the preset terminal can view cargo volume prediction analysis information.

[0136] In the logistics network cargo volume prediction device in the embodiments of the present application, all modules involved in the logistics network cargo volume prediction method are set in the model. Each module in the embodiments of the present application can perform the operation of model training in the preset prediction model, and can also realize the operation of application of the preset prediction model. The operations performed by each module in the embodiments can be as follows:

[0137] The request receiving module is configured to receive a logistics network cargo volume prediction request, obtain a target site identifier corresponding to the logistics network cargo volume prediction request, and obtain target cargo volume information associated with the target site identifier;

[0138] The model training module is configured to establish a single-site long short-term memory network prediction model according to a training subset of each site, for subsequent calculation of a first prediction value; and establish a full-site long short-term memory network prediction model according to a training set of all sites, for subsequent calculation of a second prediction value;

[0139] a weighting determination module configured to determine a prediction weight of each of the first prediction value and the second prediction value, obtain a first prediction value by processing the target cargo volume information through a single-site long short-term memory network prediction model in a preset prediction model, obtain a second prediction value by processing the target cargo volume information through a full-site long short-term memory network prediction model in the preset prediction model, perform linear regression on explanatory variables including the first prediction value and the second prediction value and a target variable including an actual value corresponding to the explanatory variables, and obtain a regression model, and obtain a regression coefficient of the regression model as the prediction weight corresponding to the first prediction value and the second prediction value; and a prediction deviation value is obtained by performing ratio operation on the cargo volume prediction value obtained by weighting and the actual cargo volume value, and a prompt information is output if the prediction deviation value exceeds a preset deviation range, to prompt further updating of the preset prediction model in the model training module.

[0140] a processing prediction module configured to obtain a first prediction value by processing the target cargo volume information through a single-site long short-term memory network prediction model in a preset prediction model, obtain a second prediction value by processing the target cargo volume information through a full-site long short-term memory network prediction model in the preset prediction model, and obtain a cargo volume prediction value by weighting the first prediction value and the second prediction value according to the prediction weight of each of the first prediction value and the second prediction value obtained by the weighting determination module.

[0141] The preset prediction model in the cargo volume prediction device for a logistics network provided in the embodiments of the present application performs cargo volume prediction through a single-site long short-term memory network prediction model and a full-site long short-term memory network prediction model in the preset prediction model to obtain a first prediction value and a second prediction value, further, the prediction weight of each of the first prediction value and the second prediction value is set for weighting processing to obtain a final cargo volume prediction value, and such a cargo volume prediction manner makes the cargo volume prediction more accurate to realize digitalization and fine management of sites in a logistics network.

[0142] The embodiments of the present application further provide a cargo volume prediction device for a logistics network, as shown in Figure 7 Figure 7 An embodiment structure schematic diagram of the cargo volume prediction device for a logistics network provided in the embodiments of the present application.

[0143] The cargo volume prediction device for a logistics network integrates any one of the cargo volume prediction devices for a logistics network provided in the embodiments of the present application, and the cargo volume prediction device for a logistics network includes:

[0144] one or more processors;

[0145] a memory; and​

[0146] one or more application programs, wherein the one or more application programs are stored in the memory and are configured to perform the steps of the method for predicting the cargo volume of the logistics network in any of the embodiments of the method for predicting the cargo volume of the logistics network.

[0147] In particular, the logistics network cargo volume prediction device can include a processor 701 having one or more processing cores, a memory 702 having one or more computer storage media, a power supply 703, an input unit 704, and the like. Those skilled in the art can understand that the logistics network cargo volume prediction device structure shown in the above-mentioned embodiments does not constitute a limitation on the logistics network cargo volume prediction device, and can include more or fewer components than those shown, or combine certain components, or different component arrangements. Among them: Figure 7

[0148] The processor 701 is the control center of the logistics network cargo volume prediction device, and connects various parts of the logistics network cargo volume prediction device through various interfaces and lines. By running or executing software programs and / or modules stored in the memory 702, and calling data stored in the memory 702, the processor 701 performs various functions of the logistics network cargo volume prediction device and processes data, thereby overall monitoring the logistics network cargo volume prediction device. Optionally, the processor 701 can include one or more processing cores; preferably, the processor 701 can integrate an application processor and a modem processor, wherein the application processor mainly processes the operating system, user interface, and application program, etc., and the modem processor mainly processes wireless communication. It can be understood that the above-mentioned modem processor can also not be integrated into the processor 701.

[0149] The memory 702 can be used to store software programs and modules. The processor 701 performs various functions and data processing by running the software programs and modules stored in the memory 702. The memory 702 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application program required by a function (such as a sound playing function, a training playing function, etc.), and the like; the data storage area can store data created according to the use of the logistics network cargo volume prediction device, etc. In addition, the memory 702 can include a high-speed random access memory, and can also include a non-volatile memory, such as at least one magnetic disk storage device, a flash memory device, or other volatile solid-state memory device. Accordingly, the memory 702 can also include a memory controller to provide the processor 701 with access to the memory 702.

[0150] ​The logistics network cargo volume prediction device further comprises a power supply 703 for powering the various components. Preferably, the power supply 703 is logically connected to the processor 701 through a power management system, so that the power management system can be used to manage charging, discharging, power consumption management, and the like. The power supply 703 can further comprise one or more DC or AC power sources, a recharging system, a power failure detection circuit, a power converter or inverter, a power status indicator, and the like.

[0151] The logistics network cargo volume prediction device can further comprise an input unit 704 for receiving input numerical or character information, and generating keyboard, mouse, joystick, optical or trackball signal inputs related to user settings and function controls.

[0152] Although not shown, the logistics network cargo volume prediction device can further comprise a display unit, and the like, which will not be described here. In the present embodiment, the processor 701 of the logistics network cargo volume prediction device loads one or more executable files corresponding to the processes of one or more application programs into the memory 702, and runs the application programs stored in the memory 702 according to the following instructions, thereby implementing various functions, such as:

[0153] receiving a logistics network cargo volume prediction request, obtaining a target site identifier corresponding to the logistics network cargo volume prediction request, and target cargo volume information associated with the target site identifier;

[0154] processing the target cargo volume information through a single-site long short-term memory network prediction model in a preset prediction model to obtain a first prediction value, and processing the target cargo volume information through a full-site long short-term memory network prediction model in the preset prediction model to obtain a second prediction value;

[0155] determining prediction weights of the first prediction value and the second prediction value, respectively, weighting the first prediction value and the second prediction value according to the respective prediction weights to obtain a cargo volume prediction value.

[0156] Those of ordinary skill in the art will understand that all or part of the steps in the various methods of the above embodiments can be completed by instructions, or by instructions controlling relevant hardware, which can be stored in a computer storage medium and loaded and executed by a processor.

[0157] To this end, the embodiment of the present application provides a computer storage medium, which can include a read only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc. A computer program is stored on the computer storage medium, and the computer program is loaded by a processor to execute the steps in any of the freight volume prediction methods of the logistics network provided by the embodiment of the present application. For example, the computer program loaded by the processor can execute the following steps:

[0158] receiving a freight volume prediction request of a logistics network, obtaining a target site identifier corresponding to the freight volume prediction request of the logistics network, and target freight volume information associated with the target site identifier;

[0159] processing the target freight volume information by a single-site long short-term memory network prediction model in a preset prediction model to obtain a first prediction value, and processing the target freight volume information by a full-site long short-term memory network prediction model in the preset prediction model to obtain a second prediction value;

[0160] determining prediction weights of the first prediction value and the second prediction value respectively, weighting the first prediction value and the second prediction value according to the prediction weights respectively to obtain a freight volume prediction value.

[0161] In the above embodiments, the description of each embodiment has its own focus, and the parts not described in detail in a certain embodiment can be referred to the detailed description of other embodiments above, which will not be repeated here.

[0162] In specific implementation, each unit or structure above can be implemented as an independent entity, or can be combined as the same or several entities, and the specific implementation of each unit or structure above can be referred to the method embodiments above, which will not be repeated here.

[0163] The specific implementation of each operation above can be referred to the embodiments above, which will not be repeated here.

[0164] The above has introduced in detail the freight volume prediction method of the logistics network provided by the embodiment of the present application, and the principle and implementation mode of the present application have been described by applying specific examples; the above embodiment description is only for helping to understand the method and core idea of the present application; meanwhile, for those skilled in the art, according to the idea of the present application, the specific implementation mode and application range will be changed, and the above description should not be understood as the limitation of the present application.

Claims

1. A method for predicting cargo volume in a logistics network, characterized in that: The cargo volume forecasting method of the logistics network includes: Receiving a cargo volume forecast request from a logistics network, obtaining a target site identifier corresponding to the cargo volume forecast request from the logistics network, and target cargo volume information associated with the target site identifier; the target site identifier is identification information that uniquely identifies a site; The target cargo volume information is processed by a single-site long-short-term memory network prediction model in a preset prediction model to obtain a first prediction value, and the target cargo volume information is processed by a full-site long-short-term memory network prediction model in the preset prediction model to obtain a second prediction value; wherein the single-site long-short-term memory network prediction model is obtained by model training based on a training subset of sites in a training set, and the full-site long-short-term memory network prediction model is obtained by model training based on a training set of all sites; Determining respective prediction weights of the first prediction value and the second prediction value, and weighting the first prediction value and the second prediction value according to their respective prediction weights to obtain a cargo volume prediction value; Obtaining an actual cargo volume value of the target site corresponding to the target site identifier at the time corresponding to the cargo volume forecast value, performing a ratio operation on the actual cargo volume value and the cargo volume forecast value to obtain a forecast deviation value; If the prediction deviation value exceeds the preset deviation range, a prompt message is output to prompt the user to update the preset prediction model.

2. The method for predicting cargo volume in a logistics network according to claim 1, characterized in that: Before obtaining a first prediction value by processing the target cargo quantity information using a single-site long short-term memory network prediction model in a preset prediction model and obtaining a second prediction value by processing the target cargo quantity information using a full-site long short-term memory network prediction model in the preset prediction model, the method includes: receiving a prediction model training instruction, collecting historical cargo volume information of each site in the logistics network, using the historical cargo volume information as a sample and dividing it into a training set and a test set, wherein the training set includes a training subset of each site, and the test set includes a test subset of each site; A single-site LSTM prediction model is established based on the training subset of each site, and a full-site LSTM prediction model is established based on the training set of all sites. Processing a test subset of each site in the test set using the single-site long short-term memory network prediction model and the full-site long short-term memory network prediction model respectively to obtain a first prediction value and a second prediction value; Performing linear regression processing on the first predicted value, the second predicted value, and the actual values ​​of the test set on the corresponding date to obtain a regression model, and determining the prediction weights of the first predicted value and the second predicted value respectively according to the regression model; The single-site long short-term memory network prediction model, the full-site long short-term memory network prediction model and the prediction weights are encapsulated to form a preset prediction model.

3. The method for predicting cargo volume in a logistics network according to claim 2, characterized in that: The method of establishing a single-site LSTM prediction model based on the training subset of each site and establishing a full-site LSTM prediction model based on the training set of all sites includes: Sort the training subsets of each venue in historical chronological order to form a historical time series for each venue. Determine input variables based on the historical time series of each venue, as well as the corresponding day of the week and whether it is a weekday. Train a preset long-short-term memory network model using the input variables of each venue to obtain a single-venue long-short-term memory network prediction model for each venue. The training set of the logistics network is aligned according to the time series and date to form a multidimensional time series. The two-dimensional matrix data composed of the multidimensional time series is used to train the preset long short-term memory network model to obtain the full-site long short-term memory network prediction model of the logistics network.

4. The method for predicting cargo volume in a logistics network according to claim 2, wherein: Performing linear regression processing based on the first predicted value and the second predicted value and actual values ​​on the test set of the corresponding date to obtain a regression model, and determining prediction weights of the first predicted value and the second predicted value based on the regression model, including: For each site, extract the first predicted value and the second predicted value in each test subset and the actual value on the test set of the corresponding date; Taking the first predicted value and the second predicted value as explanatory variables, taking the actual value as the target variable, and performing linear regression on the explanatory variables and the target variable to obtain a regression model; Obtain a regression coefficient of the regression model, and use the regression coefficient as a prediction weight corresponding to the first prediction value and the second prediction value.

5. The method for predicting cargo volume in a logistics network according to claim 1, wherein: The step of processing the target cargo quantity information by using a single-site long short-term memory network prediction model in a preset prediction model to obtain a first prediction value, and processing the target cargo quantity information by using a full-site long short-term memory network prediction model in the preset prediction model to obtain a second prediction value includes: Inputting the target cargo volume information into a preset prediction model, extracting first feature information from the target cargo volume information through a single-site long short-term memory network prediction model in the preset prediction model, and processing the extracted feature information to obtain a first prediction value; The second feature information in the target cargo quantity information is extracted and processed through the full-site long short-term memory network prediction model in the preset prediction model to obtain a second prediction value.

6. The method for predicting cargo volume in a logistics network according to any one of claims 1 to 5, characterized in that: After determining respective prediction weights of the first prediction value and the second prediction value, and weighting the first prediction value and the second prediction value according to their respective prediction weights to obtain a cargo volume prediction value, the method includes: Converting the cargo volume forecast value into a broken line graph, adding the broken line graph to a preset cargo volume analysis template, and generating a cargo volume forecast report for the logistics network; The cargo volume forecast report is sent to a preset terminal so that a user corresponding to the preset terminal can view cargo volume forecast analysis information.

7. A cargo volume prediction device for a logistics network, characterized in that: The cargo volume prediction device of the logistics network includes: a request receiving module configured to receive a cargo volume forecast request from a logistics network, obtain a target site identifier corresponding to the cargo volume forecast request from the logistics network, and obtain target cargo volume information associated with the target site identifier; the target site identifier being identification information uniquely identifying the site; a processing prediction module, configured to process the target cargo volume information using a single-site long short-term memory network prediction model in a preset prediction model to obtain a first prediction value, and to process the target cargo volume information using a full-site long short-term memory network prediction model in the preset prediction model to obtain a second prediction value; wherein the single-site long short-term memory network prediction model is trained based on a training subset of sites in a training set, and the full-site long short-term memory network prediction model is trained based on a training set of all sites; A weighted determination module is used to determine the respective prediction weights of the first prediction value and the second prediction value, weight the first prediction value and the second prediction value according to their respective prediction weights to obtain a cargo volume prediction value; obtain the actual cargo volume value of the target site corresponding to the target site identifier at the time corresponding to the cargo volume prediction value, perform a ratio operation on the actual cargo volume value and the cargo volume prediction value to obtain a prediction deviation value; if the prediction deviation value exceeds a preset deviation range, output a prompt message to prompt an update of the preset prediction model.

8. A cargo volume forecasting device for a logistics network, characterized in that: The cargo volume prediction device of the logistics network includes: one or more processors; memory; and, One or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the processor to implement the cargo volume forecasting method for a logistics network according to any one of claims 1 to 6.

9. A computer storage medium, characterized in that A computer program is stored thereon, and the computer program is loaded by a processor to execute the steps in the method for predicting cargo volume in a logistics network according to any one of claims 1 to 6.

Images