Method, device and medium for predicting cargo volume at a specific location

Abstract

The application provides a specific place cargo volume prediction method and device, computer equipment and storage medium. The method comprises the following steps: obtaining a plurality of similar days of a to-be-predicted day, obtaining the actual loading capacity of vehicles corresponding to the to-be-predicted day according to the non-departure vehicle loading capacity of the plurality of similar days, and obtaining the daily cargo volume prediction value of the to-be-predicted day according to the arrived cargo volume, the loaded cargo volume and the non-departure loading capacity prediction value of the to-be-predicted day. The method can determine a plurality of similar days from a plurality of historical days, and predict based on the data of the similar days, thereby improving the accuracy of obtaining the cargo volume prediction.

Description

Technical Field

[0001] The present application relates to the field of logistics technology, and in particular to a method, device, computer equipment, and storage medium for predicting cargo volume at a specific location. Background Art

[0002] With the rapid development of the commodity economy, logistics business volume has maintained rapid growth. For logistics companies, cargo volume forecasting is an indispensable part, which affects the cost of logistics infrastructure deployment and the efficiency of decision-making.

[0003] In current technology, future cargo volume is generally predicted through historical cargo volume data. However, the factors affecting cargo volume are relatively complex, and the prediction value obtained based on historical cargo volume data is less accurate. Summary of the Invention

[0004] Based on this, it is necessary to provide a method, device, computer equipment and storage medium for predicting cargo volume in a specific place to address the technical problem of low accuracy of cargo volume prediction values ​​in current technology.

[0005] A method for predicting cargo volume at a specific location, the method comprising:

[0006] Determining a plurality of similar days corresponding to the day to be predicted from a plurality of historical days; the plurality of similar days being obtained based on a comparison result of feature vector data of the plurality of historical days and the day to be predicted; the feature vector data including date feature information of the plurality of historical days and the day to be predicted;

[0007] Obtaining a predicted value of the loading capacity of the undelivered vehicles corresponding to the predicted day based on the actual loading capacity of the vehicles on the multiple similar days;

[0008] The daily cargo volume forecast value for the day to be predicted is obtained based on the arrived cargo volume, the in-transit cargo volume and the predicted loading volume of the vehicles that have not yet been dispatched on the day to be predicted.

[0009] In one embodiment, determining a plurality of similar days corresponding to the day to be predicted from a plurality of historical days includes:

[0010] Acquire feature vector data corresponding to the plurality of historical days and the day to be predicted; normalize the feature vector data corresponding to each historical day and the feature vector data of the day to be predicted;

[0011] The multiple similar days are obtained based on the normalized vector distances between the feature vectors of each of the historical days and the day to be predicted.

[0012] In one of the embodiments, the obtaining the similar days based on the vector distance between the normalized feature vectors of the historical days and the day to be predicted comprises:

[0013] obtaining the Euclidean distance between the vector of the historical day and the vector of the day to be predicted;

[0014] sorting the Euclidean distances of the historical days, and screening the historical days satisfying the sorting threshold to obtain the similar days.

[0015] In one of the embodiments, the obtaining the predicted value of the vehicle loading of the day to be predicted based on the actual vehicle loadings of the similar days comprises:

[0016] obtaining at least one flow direction of each of the similar days; the flow direction represents the direction from the starting point to the destination of the transport vehicle;

[0017] obtaining the actual vehicle loading of each flow direction of each of the similar days;

[0018] obtaining the predicted value of the vehicle loading of each flow direction of the day to be predicted based on the average of the actual vehicle loadings of each flow direction of the similar days;

[0019] summing the predicted values of the vehicle loadings of each flow direction of the day to be predicted to obtain the predicted value of the vehicle loading of the day to be predicted.

[0020] In one of the embodiments, after the obtaining the predicted value of the daily freight volume of the day to be predicted based on the arrived freight volume, the in-transit freight volume and the predicted value of the vehicle loading of the day to be predicted, the method further comprises:

[0021] obtaining the freight volume deviation of each of the similar days based on the predicted total freight volume and the actual total freight volume of each of the similar days;

[0022] adjusting the predicted value of the daily freight volume of the day to be predicted based on the weighted average of the freight volume deviations of the similar days to obtain the final value of the predicted value of the daily freight volume of the day to be predicted.

[0023] In one of the embodiments, before the obtaining the freight volume deviation of each of the similar days based on the predicted total freight volume and the actual total freight volume of each of the similar days, the method further comprises:

[0024] obtaining the predicted total freight volume of each of the similar days based on the actual arrived freight volume, the actual in-transit freight volume and the predicted value of the vehicle loading of the day to be predicted of each of the similar days.

[0025] In one embodiment, after obtaining the daily cargo volume forecast value for the to-be-forecasted day based on the arrived cargo volume, the in-transit cargo volume, and the predicted load volume of the vehicles not yet dispatched for the to-be-forecasted day, the method further includes:

[0026] According to the actual vehicle loads on the multiple similar days, a predicted value of the load of the undelivered vehicles corresponding to the real-time prediction time on the predicted day is obtained;

[0027] Obtain the predicted values ​​of the volume of cargo that has arrived, the volume of cargo in transit, and the loading volume of vehicles that have not yet been dispatched at the real-time prediction time, and obtain the daily cargo volume forecast value for the day to be predicted at the real-time prediction time.

[0028] In one embodiment, after obtaining the predicted values ​​of the volume of arrived goods, the volume of goods in transit, and the load volume of vehicles not yet dispatched at the real-time prediction time, and obtaining the predicted value of the daily volume of goods for the to-be-predicted day at the real-time prediction time, the method further comprises:

[0029] Obtaining cargo volume deviations for each of the multiple similar days at multiple data update times; the cargo volume deviations for each data update time are obtained based on the total estimated cargo volume and the total actual cargo volume for each of the similar days at the data update time;

[0030] Obtaining, from the plurality of data update times of the similar days, a data update time closest to the real-time prediction time of the day to be predicted as a target data update time corresponding to each of the similar days;

[0031] Obtaining cargo volume deviations at the target data update time corresponding to each of the similar days;

[0032] According to the weighted average value of the cargo volume deviations at the target data update time corresponding to each of the similar days, the daily cargo volume forecast value at the real-time forecast time is adjusted to obtain the daily cargo volume forecast final value corresponding to the real-time forecast time.

[0033] A device for predicting cargo volume at a specific location, comprising:

[0034] A similar day determination module is configured to determine, from a plurality of historical days, a plurality of similar days corresponding to the day to be predicted; the plurality of similar days is obtained based on a comparison result of feature vector data of the plurality of historical days and the day to be predicted; the feature vector data includes date feature information of the plurality of historical days and the day to be predicted;

[0035] A vehicle loading prediction module is used to obtain a predicted value of the loading capacity of the undelivered vehicles corresponding to the predicted day based on the actual loading capacity of the vehicles on the multiple similar days;

[0036] a daily cargo volume prediction module configured to obtain a daily cargo volume prediction value of the day to be predicted according to the arrived cargo volume, the in-transit cargo volume and the predicted loading volume of the vehicle not yet departed on the day to be predicted.

[0037] A computer device comprising a memory and a processor, the memory storing a computer program, and the processor implementing the steps of the cargo volume prediction method of a specific site in any of the above embodiments when executing the computer program.

[0038] A computer readable storage medium storing a computer program, the computer program implementing the steps of the cargo volume prediction method of a specific site in any of the above embodiments when executed by a processor.

[0039] The cargo volume prediction method, device, computer device and storage medium of a specific site described above, by obtaining a plurality of similar days of the day to be predicted, obtaining the predicted loading volume of the vehicle not yet departed corresponding to the day to be predicted according to the actual loading volume of the vehicle of the plurality of similar days, and obtaining the daily cargo volume prediction value of the day to be predicted according to the arrived cargo volume, the in-transit cargo volume and the predicted loading volume of the vehicle not yet departed on the day to be predicted. The scheme of the present disclosure determines a plurality of similar days from a plurality of historical days, and makes a prediction based on the data of the similar days, thereby improving the accuracy of obtaining the cargo volume prediction. BRIEF DESCRIPTION OF DRAWINGS

[0040] Figure 1 A flowchart of the cargo volume prediction method of a specific site in one embodiment;

[0041] Figure 2 A flowchart of the cargo volume prediction method of a specific site in another embodiment;

[0042] Figure 3 A flowchart of the cargo volume prediction method of a specific site in another embodiment;

[0043] Figure 4 A flowchart of the cargo volume prediction method of a specific site in another embodiment;

[0044] Figure 5 A block diagram of the cargo volume prediction device of a specific site in one embodiment;

[0045] Figure 6 An internal structure diagram of the computer device in one embodiment. DETAILED DESCRIPTION

[0046] In order to make the purposes, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application.

[0047] It should be understood that the specific site can be a site for cargo collection and transportation planning, for example, it can be a logistics sorting site, which can be used to obtain from various flows and plan the transfer or distribution service according to the cargo volume of the arrival. The measurement method of the cargo volume can be weight, number, etc.

[0048] In one embodiment, as shown in Figure 1 A specific site cargo volume prediction method is provided, and the embodiment is exemplified by applying the method to a server. It should be understood that the method can also be applied to a terminal, and can also be applied to a system including a terminal and a server, and can be realized through the interaction of the terminal and the server. In the embodiment, the method includes the following steps:

[0049] Step S101, determining a plurality of similar days corresponding to the to-be-predicted day from a plurality of historical days.

[0050] Among them, the plurality of similar days can be obtained according to the comparison result of the feature vector data of the plurality of historical days and the to-be-predicted day of the specific site. The feature vector data includes the date feature information of the plurality of historical days and the to-be-predicted day, which can be data that affects the daily cargo volume of the specific site, such as whether it is a working day, whether it is a holiday, whether it is an abnormal weather, whether it is a promotion activity day, etc. The date feature information, cargo volume information, vehicle scheduling information, vehicle loading volume information, etc. of each date of the specific site can be stored in the storage module of the server or the storage device accessible by the server. The comparison between the feature vector data of the plurality of historical days and the to-be-predicted day of the specific site can be the sample distance of the feature vector data of each historical day to the to-be-predicted day, and the comparison is performed by sample similarity measurement, and the plurality of similar days that meet the requirements of similarity are determined from the plurality of historical days.

[0051] In a specific implementation, the server can obtain the feature vector data of the plurality of historical days and the to-be-predicted day of the specific site from the storage module or the storage device, obtain the comparison result of the feature vector data of each historical date and the to-be-predicted day, and determine the plurality of similar days that meet the requirements therefrom.

[0052] Step S102, obtaining the to-be-predicted day corresponding to the non-departure vehicle loading volume prediction value according to the vehicle actual loading volume of the plurality of similar days.

[0053] Among them, the vehicle actual loading volume can be obtained from the data of the historical day, which can be the vehicle loading volume actually arrived at the specific site in the historical day. The non-departure vehicle can be a vehicle that plans to depart to the specific site on the to-be-predicted day but has not yet departed. The plurality of similar days and the to-be-predicted day have similarity, and the server can determine the to-be-predicted day corresponding to the non-departure vehicle loading volume prediction value according to the vehicle actual loading volume of the plurality of similar days.

[0054] For example, the server may perform weighted processing based on the weights of multiple similar days and the actual loading capacities of the vehicles corresponding to each of the days to obtain a predicted value of the loading capacities of the vehicles that have not yet been dispatched corresponding to the day to be predicted.

[0055] Step S103, obtaining a daily cargo volume forecast value for the day to be forecasted based on the arrived cargo volume, the in-transit cargo volume, and the predicted loading volume of vehicles not yet dispatched for the day to be forecasted.

[0056] When calculating cargo volume statistics at a specific location, the volume fluctuates in real time as it moves through the transportation system. The daily volume at a specific location can include arrived cargo, in-transit cargo, and the load on vehicles that haven't yet departed. The logistics capacity system can record data such as arrived cargo and in-transit cargo for the forecasted day in real time.

[0057] In the specific implementation, the server can obtain the arrived cargo volume and the in-transit cargo volume in real time from the logistics capacity system, and obtain the daily cargo volume forecast for the predicted day based on the predicted value of the loading volume of vehicles that have not yet been dispatched calculated based on the data of multiple similar days.

[0058] In the aforementioned method for predicting cargo volume at a specific location, a forecast value for the load capacity of vehicles not yet dispatched corresponding to the predicted day is obtained by obtaining multiple similar days to the predicted day and, based on the actual vehicle loads on these similar days, obtaining a forecast value for the load capacity of vehicles not yet dispatched for the predicted day. Furthermore, a forecast value for the daily cargo volume for the predicted day is obtained based on the arrived cargo volume, the loaded cargo volume, and the forecast value for the load capacity of vehicles not yet dispatched for the predicted day. The disclosed solution determines multiple similar days from multiple historical days and performs predictions based on data from these similar days, thereby improving the accuracy of cargo volume forecasts.

[0059] In one embodiment, the step of determining a plurality of similar days corresponding to the to-be-predicted day from a plurality of historical days in step S202 includes:

[0060] Acquire feature vector data corresponding to multiple historical days and days to be predicted; normalize the feature vector data corresponding to each historical day and the feature vector data of the day to be predicted; and obtain multiple similar days based on the vector distance between the normalized feature vectors of each historical day and the day to be predicted.

[0061] In this embodiment, the server may obtain multiple similar days based on the vector distances between the feature vector data corresponding to each historical day and the day to be predicted.

[0062] Specifically, the server can obtain feature vector data corresponding to multiple historical days and days to be predicted. The feature vector data can be data that affects the daily cargo volume of a specific venue, such as whether it is a working day, a holiday, abnormal weather, or a promotion day. For feature vector data composed of different data sources and data formats, the server can perform normalization processing on the feature vector data corresponding to each historical day and the feature vector data of the day to be predicted, respectively, to balance the contribution of each feature data. The normalization processing can be mean-variance normalization, maximum normalization, and other processing methods that can achieve the purpose of the present invention. For example, the formula for mean-variance normalization can be as follows, where μ is the mean of all sample data and σ is the standard deviation of all sample data:

[0063]

[0064] After normalization, the server obtains the feature vectors of each historical day and the feature vector of the day to be predicted. Using a similarity algorithm, the server calculates the vector distance between the feature vectors of each historical day and the feature vector of the day to be predicted. Based on the obtained vector distances, multiple similar days are identified from multiple historical days according to metrics such as the vector distance size and the number of similar days required for daily cargo volume forecasting. Algorithms for obtaining vector distances include Euclidean distance, Manhattan distance, Chebyshev distance, Mahalanobis distance, correlation distance, and information entropy.

[0065] The solution of the above embodiment normalizes the feature vectors of multiple historical days and multiple similar days, and obtains multiple similar days based on the vector distance between the feature vectors of each historical day and the day to be predicted, thereby realizing the vectorization of complex indicators that affect daily cargo volume and improving the accuracy of the obtained similar days.

[0066] In one embodiment, the step of obtaining multiple similar days based on the normalized vector distances between the feature vectors of each historical day and the day to be predicted includes:

[0067] Obtain the Euclidean distance between the vectors of each historical day and the day to be predicted; sort the Euclidean distances corresponding to each historical day, and filter out historical days that meet the sorting threshold as multiple similar days.

[0068] In this embodiment, the vector distance between each historical day and the vector of the day to be predicted can be the Euclidean distance. The server can use the Euclidean distance to calculate the vector distance. The Euclidean distance refers to the true distance between two points in the P-dimensional space, or the natural length of the vector (i.e., the distance from the point to the origin). Assuming that the feature data of the current day to be predicted has P dimensions, the calculation model can be expressed as:

[0069]

[0070] Among them, EUCLID(x,y) represents the Euclidean distance between the feature vector of the day to be predicted and the feature vector of a certain historical day; x i Represents the value of the i-th element of the daily data x to be predicted; y i Represents the i-th element value of a sample data y.

[0071] After obtaining the Euclidean distances between the vectors of each historical day and the day to be predicted, the server can sort the Euclidean distances corresponding to each historical day. This sorting can be in descending or ascending order. The sorting threshold can be the number of similar days required for daily cargo volume prediction. The server can then select the K historical days with the smallest Euclidean distances as multiple similar days.

[0072] In some embodiments, the server may use Euclidean distance to perform a K-nearest neighbor algorithm to determine multiple historically similar days. The K-nearest neighbor algorithm means that if a majority of the K nearest samples (i.e., the most adjacent samples in the feature space) near a sample belong to a certain category, then the sample also belongs to that category.

[0073] The solution of the above embodiment obtains the Euclidean distance between the feature vectors of each historical day and the day to be predicted, and then obtains multiple similar days based on the sorting threshold, thereby achieving vectorized processing of complex indicators that affect daily cargo volume and improving the accuracy of obtaining similar days.

[0074] In one embodiment, Figure 2 As shown, the step of obtaining the predicted value of the loading volume of the undeparted vehicles corresponding to the predicted day according to the loading volume of the undeparted vehicles on multiple similar days in step S102 includes:

[0075] Step S201, obtaining at least one flow direction corresponding to each of a plurality of similar days;

[0076] Step S202 , obtaining the actual vehicle load of each flow direction in at least one flow direction on each similar day.

[0077] Step S203, obtaining a predicted value of the undelivered vehicle load in each direction on the predicted day based on the average of the actual vehicle load in each direction on multiple similar days;

[0078] Step S204 , summing up the predicted values ​​of the load capacity of the vehicles that have not yet been dispatched in each direction on the day to be predicted, to obtain the predicted value of the load capacity of the vehicles that have not yet been dispatched on the day to be predicted.

[0079] In this embodiment, the flow direction represents the direction from the starting point to the destination of the transport vehicle. Each vehicle can have multiple flow directions, and the quantity and characteristics of the goods in the same flow direction on each similar day are generally consistent. The server can obtain from the storage at least one flow direction corresponding to each of the multiple similar days, and the actual vehicle load of each flow direction in at least one flow direction on each similar day. The actual vehicle load can be obtained from the storage module or accessible storage device of the server. The server can obtain the average value of the actual vehicle load of each flow direction on multiple similar days to obtain the predicted value of the load of the undeparted vehicles in each flow direction, and then add the predicted value of the load of the undeparted vehicles in each flow direction on the day to be predicted to obtain the predicted value of the load of the undeparted vehicles on the day to be predicted.

[0080] For example, at least one flow direction corresponding to each similar day may include three flow directions a1, a2, and a3. The server can obtain the actual vehicle loading corresponding to each similar day at a1, a2, and a3 respectively, and obtain the predicted values ​​C1, C2, and C3 of the undispatched vehicle loading corresponding to a1, a2, and a3 respectively based on the average value of the actual vehicle loading at a1, a2, and a3. C1, C2, and C3 are added together to obtain the predicted value of the undispatched vehicle loading on the day to be predicted.

[0081] The solution of the above embodiment obtains the actual vehicle loading capacity corresponding to each flow direction on multiple similar days, obtains the predicted value of the loading capacity of the undeparted vehicles in each flow direction on the day to be predicted, and adds them up to obtain the predicted value of the loading capacity of the undeparted vehicles on the day to be predicted. When predicting the loading capacity of the undeparted vehicles, the influence of each flow direction is fully considered, thereby improving the accuracy of the obtained loading capacity of the undeparted vehicles on the day to be predicted.

[0082] In one embodiment, the steps after obtaining the predicted value of the daily cargo volume for the forecast day based on the predicted values ​​of the arrived cargo volume, the in-transit cargo volume, and the load volume of vehicles not yet dispatched for the forecast day include:

[0083] Based on the estimated total cargo volume and the actual total cargo volume of each similar day among multiple similar days, the cargo volume deviation corresponding to each similar day is obtained; based on the weighted average of the cargo volume deviations corresponding to multiple similar days, the daily cargo volume forecast value of the predicted day is adjusted to obtain the final daily cargo volume forecast value of the predicted day.

[0084] In this embodiment, the server's storage module or accessible storage device may also store the estimated total cargo volume for each similar day. This estimated total cargo volume may be the cargo volume predicted based on the available data at the time the similar day was used as the day to be predicted. Based on the cargo volume deviations for similar days, the server may adjust the obtained daily cargo volume forecast for the current day to be predicted, thereby improving the accuracy of the obtained cargo volume forecast for the day to be predicted.

[0085] In some embodiments, the cargo volume deviation of each similar day can be the difference between the total expected cargo volume and the total actual cargo volume of each similar day. The server can use the weighted average of the cargo volume deviations of each similar day as an adjustment parameter, and add it to the obtained daily cargo volume forecast value of the day to be predicted to obtain the final value of the daily cargo volume forecast of the day to be predicted. For example, if the day to be predicted has K similar days, the server can obtain K cargo volume deviations and the average value avg(X k ), according to the daily cargo volume forecast value / (1+avg(X k ))Get the final value of daily cargo volume forecast.

[0086] In some embodiments, the weights corresponding to each similar day can be the same, or the weights can be set according to the size of the vector distance between each similar day and the day to be predicted. The smaller the vector distance, the larger the weight value corresponding to the similar day, so as to improve the accuracy of the calculation.

[0087] The solution of the above embodiment obtains the cargo volume deviation of each similar day, adjusts the daily cargo volume forecast value of the day to be forecasted, and obtains the final value of the daily cargo volume forecast, thereby improving the accuracy of the daily cargo volume forecast.

[0088] In one embodiment, the steps before obtaining the cargo volume deviation corresponding to each similar day based on the estimated total cargo volume and the actual total cargo volume of each similar day include:

[0089] The actual amount of cargo that has arrived, the actual amount of cargo in transit, and the predicted value of the loading capacity of vehicles that have not yet been dispatched on each of multiple similar days are obtained to obtain the total expected cargo volume for each similar day.

[0090] In this embodiment, the data corresponding to each similar day may include an estimated total cargo volume. The estimated total cargo volume can be obtained by the server based on the sum of the actual arrived cargo volume, the actual in-transit cargo volume, and the predicted load volume of vehicles not yet dispatched in the transportation capacity system when the similar day is used as the prediction day.

[0091] The solution of the above embodiment obtains the estimated total cargo volume of each similar day through the actual arrived cargo volume, the actual in-transit cargo volume and the predicted value of the loading volume of vehicles not yet dispatched on each similar day, so as to obtain the cargo volume deviation of each similar day.

[0092] In one embodiment, after obtaining the daily cargo volume forecast value for the forecast day based on the forecast values ​​of the arrived cargo volume, the in-transit cargo volume, and the load volume of vehicles not yet dispatched for the forecast day, the method further includes:

[0093] According to the actual vehicle loading volume on multiple similar days, the predicted value of the loading volume of the vehicles that have not yet been dispatched corresponding to the real-time prediction time on the predicted day is obtained; the arrived cargo volume, the cargo volume in transit and the predicted value of the loading volume of the vehicles that have not yet been dispatched at the real-time prediction time are obtained to obtain the daily cargo volume forecast value on the predicted day at the real-time prediction time.

[0094] In this embodiment, transportation to a specific location can be continuous, cargo volume data can be updated in real time, and the server can generate real-time daily cargo volume forecasts. The data update time can be the time at which data was acquired for each historical day. Each data update time can correspond to actual arrived cargo volume, actual in-transit cargo volume, and predicted values ​​for the load volume of undeparted vehicles. The server can obtain the data update time closest to the real-time forecast time from multiple data update times for each similar day and use it as the target data update time for each similar day.

[0095] To calculate the predicted value of the number of vehicles loaded at the target time, the server can obtain the actual loading capacity of each vehicle based on the actual loading capacity of vehicles on similar days. Using this as a sample, the server can calculate the predicted value of the number of vehicles loaded at the target time based on the real-time number of vehicles loaded at the target time recorded by the transportation system. For example, the server can obtain the actual loading capacity of all vehicles on similar days, obtain the average of the actual loading capacity of each vehicle, and use this as a sample to calculate the predicted value of the number of vehicles loaded at the target time.

[0096] The server can obtain the actual amount of cargo that has arrived and the actual amount of cargo in transit corresponding to the real-time prediction time from the transportation capacity system, and combine it with the corresponding predicted value of the loading capacity of vehicles that have not yet been dispatched to obtain the daily cargo volume forecast value for the real-time prediction time, thereby realizing real-time prediction of daily cargo volume and improving the real-time performance of obtaining daily cargo volume data.

[0097] In one embodiment, after obtaining the predicted values ​​of the volume of arrived goods, the volume of goods in transit, and the load volume of vehicles not yet dispatched at the real-time prediction time, and obtaining the predicted value of the daily volume of goods on the predicted day at the real-time prediction time, the method further includes:

[0098] Obtain the cargo volume deviation of each similar day at multiple data update times among multiple similar days respectively; obtain the data update time closest to the real-time prediction time of the to-be-predicted day among the multiple data update times of each similar day as the target data update time corresponding to each similar day; obtain the cargo volume deviation of the target data update time corresponding to each similar day; adjust the daily cargo volume prediction value at the real-time prediction time according to the weighted average of the cargo volume deviations of the target data update time corresponding to each similar day, and obtain the final daily cargo volume prediction value corresponding to the real-time prediction time.

[0099] In this embodiment, the server can perform deviation adjustments on the daily cargo volume forecast values ​​obtained in real time. The data update time can be the time when the data was obtained on each historical day. For example, the server can update the data at a frequency of f hours / time and obtain the daily cargo volume forecast value corresponding to each data update time. On each historical day, the server can obtain N = 24 / f total estimated cargo volumes and N = 24 / f total actual cargo volumes. The cargo volume deviation corresponding to each data update time can be obtained based on the difference between the estimated total cargo volume and the actual total cargo volume on similar days at the data update time.

[0100] For example, when f = 4, the server can obtain six data update times for each historical day, along with the corresponding estimated and actual cargo volumes, and obtain the cargo volume deviation at each data update time. When forecasting cargo volume at time t1 within the forecasted day, the server can obtain the data update time closest to time t1 from the data update times corresponding to multiple similar days as the target data update time for that similar day, and obtain the cargo volume deviation corresponding to that target data update time.

[0101] If the day to be predicted includes K similar days, each similar day includes 6 data update times, where the data update times closest to the t1 time of the K similar days and the day to be predicted are t k1 , t k2 …t kk , the server can obtain t k1 , t k2 …t kk The corresponding cargo volume deviation.

[0102] The server can obtain t k1 , t k2 …t kk The weighted average of the corresponding cargo volume deviations is used to adjust the daily cargo volume forecast value at the real-time forecast time to obtain the final daily cargo volume forecast value corresponding to the real-time forecast time.

[0103] In some embodiments, the server can configure the algorithm execution frequency and data update frequency f based on factors such as business demand frequency and computational time, to conserve system computing resources and improve real-time prediction efficiency. For example, the time-consuming similar day identification algorithm can be run only once per day, while the time-consuming real-time prediction module can be run more frequently based on business needs.

[0104] The solution of the above embodiment obtains the cargo volume deviation corresponding to each data update time to obtain the cargo volume deviation corresponding to the real-time prediction time, and adjusts the daily cargo volume forecast value corresponding to the real-time prediction time according to the cargo volume deviation corresponding to the real-time prediction time to obtain the daily cargo volume forecast final value corresponding to the real-time prediction time, thereby improving the real-time and accuracy of obtaining the daily cargo volume forecast value.

[0105] In one embodiment, Figure 3 As shown, a method for predicting cargo volume at a specific location is provided, the method comprising:

[0106] Step S301, obtain feature vector data corresponding to multiple historical days and the day to be predicted; normalize the feature vector data corresponding to each historical day and the feature vector data of the day to be predicted respectively; obtain the Euclidean distance between the vectors of each historical day and the day to be predicted; sort the Euclidean distances corresponding to each historical day, and filter out historical days that meet the sorting threshold as multiple similar days; the feature vector data includes date feature information of the multiple historical days and the day to be predicted.

[0107] Step S302, obtaining at least one flow direction corresponding to each of multiple similar days; the flow direction represents the direction from the starting point to the destination of the transport vehicle; obtaining the actual vehicle load of each flow direction in at least one flow direction of each similar day; obtaining the predicted value of the load of the undeparted vehicles of each flow direction on the to-be-predicted day based on the average value of the actual vehicle load of each flow direction on multiple similar days; adding up the predicted value of the load of the undeparted vehicles of each flow direction on the to-be-predicted day to obtain the predicted value of the load of the undeparted vehicles of the to-be-predicted day.

[0108] Step S303: Obtain a daily cargo volume forecast value for the day to be predicted based on the predicted values ​​of the cargo volume that has arrived, the cargo volume that is in transit, and the load volume of vehicles that have not yet been dispatched.

[0109] Step S304: Obtain the actual arrived cargo volume, actual in-transit cargo volume, and predicted loading volume of vehicles not yet dispatched for each of the multiple similar days to obtain the estimated total cargo volume for each of the multiple similar days; obtain the cargo volume deviation corresponding to each of the multiple similar days based on the estimated total cargo volume and the actual total cargo volume; adjust the daily cargo volume forecast value for the day to be forecasted based on the weighted average of the cargo volume deviations corresponding to the multiple similar days to obtain the final daily cargo volume forecast value for the day to be forecasted.

[0110] In the above embodiment, multiple similar days are obtained through the Euclidean distance of the feature vector data from each historical day to the day to be predicted, and the predicted value of the loading capacity of the vehicles that have not been dispatched on the day to be predicted is obtained according to the actual loading capacity of the vehicles in each flow direction on each similar day. Then, the daily cargo volume forecast value of the day to be predicted is obtained in combination with the arrived cargo volume and the in-transit cargo volume on the day to be predicted. The daily cargo volume forecast value is corrected in combination with the cargo volume deviation of each similar day. The scheme disclosed in the present invention determines multiple similar days from multiple historical days, and makes predictions based on the data of similar days, thereby improving the relevance of the obtained benchmark data for cargo volume prediction, and corrects the daily cargo volume forecast value by the cargo volume deviation of similar days, thereby further improving the accuracy of the daily cargo volume prediction.

[0111] In one embodiment, Figure 4 As shown, a method for predicting cargo volume at a specific location is provided, the method comprising:

[0112] Step S401, obtain feature vector data corresponding to multiple historical days and the day to be predicted; normalize the feature vector data corresponding to each historical day and the feature vector data of the day to be predicted respectively; obtain the Euclidean distance between the vectors of each historical day and the day to be predicted; sort the Euclidean distances corresponding to each historical day, and filter out historical days that meet the sorting threshold as multiple similar days; the feature vector data includes date feature information of the multiple historical days and the day to be predicted.

[0113] Step S402, based on the actual vehicle loading capacity of multiple similar days, obtain the predicted value of the loading capacity of the vehicles that have not yet been dispatched corresponding to the real-time prediction time on the predicted day; obtain the arrived cargo volume, the cargo volume in transit and the predicted value of the loading capacity of the vehicles that have not yet been dispatched at the real-time prediction time, and obtain the daily cargo volume forecast value at the real-time prediction time on the predicted day.

[0114] Step S403 , respectively obtaining cargo volume deviations of each of the multiple similar days at multiple data update times; the cargo volume deviations at each data update time are obtained based on the total estimated cargo volume and the total actual cargo volume of each similar day at the data update time.

[0115] Step S404: Obtain the data update time closest to the real-time forecast time of the forecasted day among the multiple data update times of each similar day as the target data update time corresponding to each similar day; obtain the cargo volume deviation of the target data update time corresponding to each similar day.

[0116] Step S405 , adjusting the daily cargo volume forecast value at the real-time forecast time according to the weighted average of the cargo volume deviations at the target data update time corresponding to each similar day, to obtain the final daily cargo volume forecast value corresponding to the real-time forecast time.

[0117] In the above embodiment, multiple similar days are obtained through the Euclidean distance of the feature vector data from each historical day to the day to be predicted, and the predicted value of the loading capacity of the vehicles that have not been dispatched at the real-time prediction time is obtained according to the actual loading capacity of the vehicles on each similar day. Then, the cargo volume forecast value of the real-time prediction time of the day to be predicted is obtained in combination with the arrived cargo volume and the in-transit cargo volume corresponding to the real-time prediction time of the day to be predicted. The cargo volume forecast value of the day is corrected in combination with the cargo volume deviation corresponding to the real-time prediction time of each similar day. The scheme disclosed in the present invention determines multiple similar days from multiple historical days, performs cargo volume forecast and deviation correction based on the data of similar days, improves the relevance and accuracy of the obtained benchmark data for cargo volume forecast, and performs real-time cargo volume forecast based on the data of the closest similar day, further improving the real-time performance of daily cargo volume forecast.

[0118] It should be understood that although Figure 1-4 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. In addition, Figure 1-4 At least part of the steps may include multiple steps or multiple stages. These steps or stages are not necessarily performed at the same time, but can be performed at different times. The order of execution of these steps or stages is not necessarily one by one, but can be performed in turn or alternately with other steps or at least part of the steps or stages in other steps.

[0119] In one embodiment, Figure 5 As shown, a device for predicting cargo volume at a specific location is provided, and the device 500 includes:

[0120] A similar day determination module 501 is configured to determine, from a plurality of historical days, a plurality of similar days corresponding to the day to be predicted; the plurality of similar days is obtained based on a comparison result of feature vector data of the plurality of historical days and the day to be predicted; the feature vector data includes date feature information of the plurality of historical days and the day to be predicted;

[0121] The vehicle loading prediction module 502 is configured to obtain a predicted value of the loading of the undelivered vehicles corresponding to the predicted day based on the actual loading of the vehicles on the multiple similar days;

[0122] The daily cargo volume prediction module 503 is used to obtain the daily cargo volume prediction value of the day to be predicted based on the arrived cargo volume, the in-transit cargo volume and the predicted loading volume of the vehicles that have not yet been dispatched on the day to be predicted.

[0123] In one embodiment, the similar day determining module 501 comprises: a normalization unit configured to obtain feature vector data corresponding to the plurality of historical days and the day to be predicted respectively; normalize the feature vector data corresponding to each historical day and the feature vector data of the day to be predicted respectively; a similar day obtaining unit configured to obtain the plurality of similar days based on the vector distance between the normalized feature vector of each historical day and the day to be predicted.

[0124] In one embodiment, the normalization unit comprises: an Euclidean distance sub-unit configured to obtain the Euclidean distance between the vector of each historical day and the day to be predicted; sort the Euclidean distances corresponding to each historical day, and filter the historical days satisfying the sorting threshold to obtain the plurality of similar days.

[0125] In one embodiment, the vehicle loading prediction module 502 comprises: a vehicle loading prediction unit configured to obtain at least one flow direction corresponding to each similar day in the plurality of similar days; the flow direction represents the direction from the starting point to the destination of the transport vehicle; obtain the actual vehicle loading of each flow direction in the at least one flow direction of each similar day; obtain the predicted value of the actual vehicle loading of each flow direction of the day to be predicted according to the average value of the actual vehicle loading of each flow direction of the plurality of similar days; and obtain the predicted value of the actual vehicle loading of the day to be predicted by summing the predicted value of the actual vehicle loading of each flow direction of the day to be predicted.

[0126] In one embodiment, the device 500 further comprises: a cargo volume deviation unit configured to obtain the cargo volume deviation corresponding to each similar day in the plurality of similar days according to the predicted total cargo volume and the actual total cargo volume of each similar day; and a day cargo volume final value unit configured to adjust the day cargo volume prediction value of the day to be predicted according to the weighted average value of the cargo volume deviation corresponding to each similar day in the plurality of similar days, to obtain the final day cargo volume prediction value of the day to be predicted.

[0127] In one embodiment, the cargo volume deviation unit is further configured to obtain the actual arrived cargo volume, the actual in-transit cargo volume, and the predicted value of the actual vehicle loading of each similar day in the plurality of similar days, to obtain the predicted total cargo volume of each similar day.

[0128] In one embodiment, the device 500 further comprises: a real-time cargo volume prediction device configured to obtain the predicted value of the actual vehicle loading at a real-time prediction time of the day to be predicted according to the actual vehicle loading of the plurality of similar days; obtain the arrived cargo volume, the in-transit cargo volume, and the predicted value of the actual vehicle loading at the real-time prediction time, to obtain the day cargo volume prediction value of the day to be predicted at the real-time prediction time.

[0129] In one embodiment, the real-time cargo volume prediction device also includes: a real-time cargo volume prediction final value unit, which is used to respectively obtain the cargo volume deviation of each similar day in the multiple similar days at multiple data update times; the cargo volume deviation at each data update time is obtained according to the expected total cargo volume and the actual total cargo volume of each similar day at the data update time; obtain the data update time closest to the real-time prediction time of the day to be predicted among the multiple data update times of each similar day, as the target data update time corresponding to each similar day; obtain the cargo volume deviation of the target data update time corresponding to each similar day; adjust the daily cargo volume prediction value of the real-time prediction time according to the weighted average of the cargo volume deviations of the target data update time corresponding to each similar day, and obtain the daily cargo volume prediction final value corresponding to the real-time prediction time.

[0130] The specific limitations of the location-specific cargo volume forecasting device can be found in the limitations of the location-specific cargo volume forecasting method described above and will not be further elaborated here. Each module within the location-specific cargo volume forecasting device described above may be implemented in whole or in part via software, hardware, or a combination thereof. Each of these modules may be embedded in or independent of a processor within a computer device in hardware form, or may be stored in a computer device memory in software form, allowing the processor to invoke and execute the corresponding operations of each module.

[0131] The method for predicting the cargo volume of a specific location provided by the present application can be applied to a computer device, which can be a server, and its internal structure diagram can be as shown below: Figure 5 As shown. The computer device includes a processor, memory, and a network interface connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the computer device is used to store historical cargo volume and date feature data. The network interface of the computer device is used to communicate with an external terminal via a network connection. When executed by the processor, the computer program implements a method for predicting cargo volume at a specific location.

[0132] Those skilled in the art will understand that Figure 5 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0133] In one embodiment, a computer device is provided, comprising a memory and a processor, the memory has stored therein a computer program, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0134] In one embodiment, a computer readable storage medium is provided, having stored thereon a computer program, the computer program being executed by a processor to implement the steps in the above-mentioned method embodiments.

[0135] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by a computer program instructing related hardware, and the computer program can be stored in a non-volatile computer readable storage medium. When the computer program is executed, it can include the processes of the above-mentioned embodiments. Any reference to memory, storage, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory or optical memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0136] Any combination of the technical features of the above embodiments can be made. In order to make the description simple, all possible combinations of the technical features in the above embodiments are not described, however, as long as the combination of the technical features does not exist, it should be considered as the scope of the present application.

[0137] The above embodiments only express several embodiments of the present application, and the description is more specific and detailed, but it should not be understood as a limitation on the scope of the patent. It should be pointed out that for ordinary skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are within the scope of the present application. Therefore, the scope of protection of the patent of the present application should be subject to the appended claims.

Claims

1. A method for predicting cargo volume at a specific location, characterized in that: The method comprises: Determining a plurality of similar days corresponding to the day to be predicted from a plurality of historical days; the plurality of similar days being obtained based on a comparison result of feature vector data of the plurality of historical days and the day to be predicted; the feature vector data including date feature information of the plurality of historical days and the day to be predicted; Obtain at least one flow direction corresponding to each of the multiple similar days; the flow direction represents the direction from the starting point to the destination of the transport vehicle; obtain the actual vehicle load of each of the at least one flow direction on each similar day; obtain a predicted value of the load of undeparted vehicles for each flow direction on the to-be-predicted day based on an average value of the actual vehicle load of each of the flow directions on the multiple similar days; sum the predicted values ​​of the load of undeparted vehicles for each flow direction on the to-be-predicted day to obtain a predicted value of the load of undeparted vehicles for the to-be-predicted day; the actual vehicle load is obtained from data on historical days, and the undeparted vehicles are vehicles that are planned to be dispatched to a specific site on the to-be-predicted day but have not yet been dispatched; The daily cargo volume forecast value for the day to be predicted is obtained based on the arrived cargo volume, the in-transit cargo volume and the predicted loading volume of the vehicles that have not yet been dispatched on the day to be predicted.

2. The method according to claim 1, characterized in that The step of determining a plurality of similar days corresponding to the day to be predicted from a plurality of historical days includes: Acquire feature vector data corresponding to the plurality of historical days and the day to be predicted; normalize the feature vector data corresponding to each historical day and the feature vector data of the day to be predicted; The multiple similar days are obtained based on the normalized vector distances between the feature vectors of each of the historical days and the day to be predicted.

3. The method according to claim 2, characterized in that The obtaining of the plurality of similar days based on the normalized vector distances between the feature vectors of the historical days and the day to be predicted comprises: Obtaining the Euclidean distance between the vectors of each historical day and the day to be predicted; The Euclidean distances corresponding to the historical days are sorted, and the historical days that meet the sorting threshold are screened and obtained as the multiple similar days.

4. The method according to claim 1, wherein After obtaining the daily cargo volume forecast value for the to-be-forecasted day based on the arrived cargo volume, the in-transit cargo volume, and the predicted loading volume of the vehicles not yet dispatched for the to-be-forecasted day, the method further includes: Obtaining cargo volume deviations corresponding to each of the multiple similar days based on the estimated total cargo volume and the actual total cargo volume of each similar day; The daily cargo volume forecast value of the to-be-forecasted day is adjusted according to the weighted average value of the cargo volume deviations corresponding to the multiple similar days to obtain the final daily cargo volume forecast value of the to-be-forecasted day.

5. The method according to claim 4, characterized in that Before obtaining the cargo volume deviation corresponding to each of the multiple similar days based on the estimated total cargo volume and the actual total cargo volume of each of the multiple similar days, the method further includes: The actual amount of cargo that has arrived, the actual amount of cargo in transit, and the predicted value of the loading capacity of vehicles that have not yet been dispatched on each of the multiple similar days are obtained to obtain the total estimated cargo volume for each of the similar days.

6. The method according to claim 1, characterized in that After obtaining the daily cargo volume forecast value for the to-be-forecasted day based on the arrived cargo volume, the in-transit cargo volume, and the predicted loading volume of the vehicles not yet dispatched for the to-be-forecasted day, the method further includes: According to the actual vehicle loads on the multiple similar days, a predicted value of the load of the undelivered vehicles corresponding to the real-time prediction time on the predicted day is obtained; Obtain the predicted values ​​of the volume of cargo that has arrived, the volume of cargo in transit, and the loading volume of vehicles that have not yet been dispatched at the real-time prediction time, and obtain the daily cargo volume forecast value for the day to be predicted at the real-time prediction time.

7. The method according to claim 6, characterized in that After obtaining the predicted values ​​of the volume of arrived goods, the volume of goods in transit, and the load volume of vehicles not yet dispatched at the real-time prediction time, and obtaining the predicted value of the daily volume of goods on the to-be-predicted day at the real-time prediction time, the method further comprises: Obtaining cargo volume deviations for each of the multiple similar days at multiple data update times; the cargo volume deviations for each data update time are obtained based on the total estimated cargo volume and the total actual cargo volume for each of the similar days at the data update time; Obtaining, from the plurality of data update times of the similar days, a data update time closest to the real-time prediction time of the day to be predicted as a target data update time corresponding to each of the similar days; Obtaining cargo volume deviations at the target data update time corresponding to each of the similar days; According to the weighted average value of the cargo volume deviations at the target data update time corresponding to each of the similar days, the daily cargo volume forecast value at the real-time forecast time is adjusted to obtain the daily cargo volume forecast final value corresponding to the real-time forecast time.

8. A device for predicting cargo volume at a specific location, characterized in that: The device comprises: A similar day determination module is configured to determine, from a plurality of historical days, a plurality of similar days corresponding to the day to be predicted; the plurality of similar days is obtained based on a comparison result of feature vector data of the plurality of historical days and the day to be predicted; the feature vector data includes date feature information of the plurality of historical days and the day to be predicted; A vehicle loading prediction module is configured to obtain at least one flow direction corresponding to each of the plurality of similar days; the flow direction represents the direction from the starting point to the destination of the transport vehicle; obtain the actual vehicle loading capacity of each of the at least one flow direction on each similar day; obtain a predicted value of the loading capacity of undeparted vehicles for each flow direction on the day to be predicted based on the average value of the actual vehicle loading capacity of each flow direction on the plurality of similar days; sum the predicted values ​​of the loading capacity of undeparted vehicles for each flow direction on the day to be predicted to obtain a predicted value of the loading capacity of undeparted vehicles for the day to be predicted; the actual vehicle loading capacity is obtained from data on historical days, and the undeparted vehicles are vehicles that are planned to be dispatched to a specific site on the day to be predicted but have not yet been dispatched; The daily cargo volume forecasting module is used to obtain the daily cargo volume forecast value of the day to be predicted based on the arrived cargo volume, the in-transit cargo volume and the predicted loading volume of the vehicles that have not yet been dispatched on the day to be predicted.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

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

Images