Intelligent dispatching system and method for refined oil

Abstract

The present invention discloses an intelligent dispatching system and method for refined oil, and relates to the field of intelligent dispatching. First, the system obtains oil depot information, oil station information, vehicle information, travel information, and constraint information, then obtains sales forecast information of each oil product at the oil station, then obtains optimization indicators and parameter settings, and finally, according to the oil depot information, the oil station information, the vehicle information, the travel information, the constraint information, the sales forecast information of each oil product at the oil station, the optimization indicators and the parameter settings, uses an iterative greedy algorithm to solve the active distribution dispatching optimization model to obtain the optimal distribution plan. In this way, it can help avoid oversupply or undersupply, and improve distribution efficiency and customer satisfaction.

Description

Technical Field

[0001] The present application relates to the field of intelligent scheduling, and more specifically, to an intelligent scheduling system and method for refined oil products. Background Art

[0002] The dispatch of refined oil is a logistics management plan between oil depots and gas stations. Its purpose is to formulate a reasonable distribution plan based on the inventory of oil depots and the demand of gas stations to ensure the stable supply of gas stations and reduce transportation costs and inventory costs. The traditional dispatch plan of refined oil mainly relies on manual experience to prepare distribution plans. This method has the following problems: First, it is difficult for the manually prepared distribution plan to consider the influence of various factors, such as oil type, oil depot capacity, number of vehicles, transportation routes, traffic conditions, market demand, etc., resulting in unreasonable and unoptimized distribution plans; second, the manually prepared distribution plan takes a long time, the plan preparation efficiency is low, and it is difficult to adapt to changes in market demand, resulting in insufficient flexibility and timeliness of the distribution plan; third, the manually prepared distribution plan lacks data support and analysis, and it is difficult to evaluate the distribution effect and improvement measures, resulting in insufficient scientific and reliable distribution plans.

[0003] Therefore, an optimized intelligent scheduling solution for refined oil products is desired. Summary of the invention

[0004] In order to solve the above technical problems, this application is proposed. The embodiments of this application provide an intelligent dispatching system and method for refined oil products, which can avoid oversupply or undersupply, and improve distribution efficiency and customer satisfaction.

[0005] According to one aspect of the present application, there is provided an intelligent scheduling method for refined oil, which includes:

[0006] Obtain oil depot information, gas station information, vehicle information, itinerary information, and constraint information;

[0007] Obtain sales forecast information for each oil product at a gas station;

[0008] Get optimization indicators and parameter settings;

[0009] And based on the oil depot information, the gas station information, the vehicle information, the trip information, the constraint information, the sales forecast information of each oil product at the gas station, the optimization index and the parameter setting, an iterative greedy algorithm is used to solve the active distribution scheduling optimization model to obtain the optimal distribution plan.

[0010] According to another aspect of the present application, there is provided an intelligent scheduling system for refined oil, comprising:

[0011] The refueling information acquisition module is used to obtain oil depot information, gas station information, vehicle information, itinerary information, and constraint information;

[0012] The sales forecast information acquisition module is used to obtain the sales forecast information of each oil product at the gas station;

[0013] and an indicator and parameter acquisition module for obtaining optimization indicators and parameter settings;

[0014] The distribution plan analysis module is used to solve the active distribution scheduling optimization model using an iterative greedy algorithm to obtain the optimal distribution plan based on the oil depot information, the oil station information, the vehicle information, the trip information, the constraint information, the sales forecast information of each oil product at the oil station, the optimization index and the parameter setting, which includes: obtaining the current optimal solution;

[0015] Select N reconstruction vehicles and trips, and clean up the selected trips;

[0016] Execute a reconstruction sub-process to obtain three reconstruction solutions, and assume that two layers of the reconstruction sub-process are nested and executed on the basis of adopting the three reconstruction solutions to obtain a maximum of 27 solutions for the three reconstruction solutions;

[0017] Verify the legality of up to 27 sets of solutions in each round and calculate the index scores;

[0018] And in response to the indicator score being greater than the indicator score of the current optimal solution, replacing the current optimal solution.

[0019] Compared with the prior art, the intelligent dispatching system and method for refined oil provided by the present application first obtains the oil depot information, gas station information, vehicle information, travel information, and constraint information, then obtains the sales forecast information of each oil product at the gas station, then obtains the optimization index and parameter setting, and finally, according to the oil depot information, the gas station information, the vehicle information, the travel information, the constraint information, the sales forecast information of each oil product at the gas station, the optimization index and the parameter setting, uses the iterative greedy algorithm to solve the active distribution dispatching optimization model to obtain the optimal distribution plan. In this way, it can help avoid oversupply or undersupply and improve distribution efficiency and customer satisfaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. The following drawings are not intentionally scaled to the actual size, and the focus is on illustrating the main purpose of the present application.

[0021] Figure 1 The present invention is a flowchart of an intelligent scheduling method for refined oil according to an embodiment of the present application.

[0022] Figure 2 This is a flowchart of sub-step S120 of the intelligent scheduling method for refined oil according to an embodiment of the present application.

[0023] Figure 3 Schematic diagram of the architecture of sub-step S120 of the intelligent scheduling method for refined oil according to an embodiment of the present application.

[0024] Figure 4 This is a flowchart of sub-step S140 of the intelligent scheduling method for refined oil according to an embodiment of the present application.

[0025] Figure 5 It is a block diagram of an intelligent scheduling system for refined oil according to an embodiment of the present application.

[0026] Figure 6 This is an application scenario diagram of the intelligent scheduling method for refined oil according to an embodiment of the present application.

[0027] Figure 7 It is a schematic diagram of a method for scheduling refined oil according to an embodiment of the present application.

[0028] Figure 8 It is a schematic diagram of a solution process based on an iterative greedy algorithm according to an embodiment of the present application.

[0029] Fig. 9 Schematic diagram of a reconstruction sub-process according to an embodiment of the present application. DETAILED DESCRIPTION

[0030] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work also fall within the scope of protection of the present application.

[0031] As shown in this application and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "include" and "comprise" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0032] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0033] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps may be processed in reverse order or simultaneously as required. Meanwhile, other operations may also be added to these processes, or a certain step or several steps of operations may be removed from these processes.

[0034] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0035] In view of the above technical problems, in the technical solution of this application, an intelligent scheduling method for refined oil is proposed. Figure 1 FIG. 1 is a flow chart of an intelligent dispatching method for refined oil according to an embodiment of the present application. Figure 1 As shown, the intelligent scheduling method for refined oil according to the embodiment of the present application includes the steps of: S110, obtaining oil depot information, gas station information, vehicle information, travel information, and constraint information; S120, obtaining sales forecast information of each oil product at the gas station; S130, obtaining optimization indicators and parameter settings; and, S140, based on the oil depot information, the gas station information, the vehicle information, the travel information, the constraint information, the sales forecast information of each oil product at the gas station, the optimization indicators and the parameter settings, using an iterative greedy algorithm to solve the active distribution scheduling optimization model to obtain the optimal distribution plan.

[0036] The oil depot information includes the working hours and available quantities of each oil product at the oil depot, the gas station information includes the effective tank capacity, pump stop inventory, inventory time and inventory quantity of each oil product at each gas station, the vehicle information includes the parking point, compartment capacity, oil products carried and working hours of each shift of each vehicle, the travel information includes the time and distance required for distribution vehicles to travel from depot to depot, depot to station, station to depot and station to station, and the constraint information includes the restricted time periods for vehicles to each gas station and the restricted time periods for vehicles to each oil depot. The optimization index includes the weights of each optimization index, and the parameter setting includes the business rule parameters.

[0037] Accordingly, considering that in the actual intelligent dispatching process of refined oil, it is crucial to accurately analyze and obtain the sales forecast information of each oil product at the gas station, because the sales forecast information of each oil product at the gas station can provide an estimate of the demand for each oil product in the future. By accurately predicting sales volume, distribution plans can be better planned and arranged, and oil depots can be helped to make reasonable inventory planning to ensure sufficient supply of refined oil when and where needed. This helps to avoid oversupply or undersupply and improve distribution efficiency and customer satisfaction.

[0038] Based on this, for the first oil product among the multiple oil products of the gas station, the technical concept of the present application is to collect the sales value of the first oil product in real time, and introduce data collection and analysis algorithms at the back end to perform time series analysis of the sales value of the first oil product, so as to generate sales forecast information of the first oil product. In this way, it is possible to optimize the distribution plan, improve resource utilization efficiency, improve inventory management, and promote coordination and cooperation among various links in the supply chain, thereby improving the efficiency and quality of finished oil distribution.

[0039] Figure 2 This is a flowchart of sub-step S120 of the intelligent scheduling method for refined oil according to an embodiment of the present application. Figure 3 FIG. 1 is a schematic diagram of the architecture of sub-step S120 of the intelligent scheduling method for refined oil according to an embodiment of the present application. Figure 2 and Figure 3 As shown, according to the intelligent scheduling method for refined oil of the embodiment of the present application, the sales forecast information of each oil product at the gas station is obtained, including: S121, obtaining the sales value of the first oil product for multiple days within a predetermined time period; S122, arranging the sales value of the first oil product for multiple days within the predetermined time period according to the time dimension as a first oil product sales time series input vector; S123, vector segmentation of the first oil product sales time series input vector to obtain a sequence of the first oil product sales local time series input vector; S124, passing the sequence of the first oil product sales local time series input vector through a sales time series feature extractor based on a one-dimensional convolutional layer to obtain the first oil product sales local time series input vector. S125, calculating the sales time series fluctuation measurement coefficient between each two adjacent first oil product sales local time series characteristic vectors in the sequence of the first oil product sales local time series characteristic vectors to obtain a sales time series fluctuation characteristic vector composed of multiple sales time series fluctuation measurement coefficients; S126, embedding the sequence of the first oil product sales local time series characteristic vectors and the sales time series fluctuation characteristic vector into a fusion module through dynamic and static features to obtain the first oil product sales dynamic and static fusion characteristic vector; and, S127, passing the first oil product sales dynamic and static fusion characteristic vector through a decoder to obtain a decoded value, which is the sales forecast information of the first oil product.

[0040] Specifically, in the technical solution of the present application, first, the sales value of the first oil product for multiple days in a predetermined time period is obtained. Then, considering that the sales value of the first oil product has a dynamic change law of time series in the time dimension, therefore, in order to be able to capture the time series change characteristics of the sales value of the first oil product and to predict the sales volume of the first oil product, it is necessary to arrange the sales value of the first oil product for multiple days in the predetermined time period according to the time dimension as the first oil product sales time series input vector, so as to integrate the distribution information of the sales value of the first oil product in time series.

[0041] It should be understood that in the process of actually analyzing the sales volume of the first oil product to make sales volume forecasts, the sales of the first oil product in different time periods may have different characteristics and trends, that is, the sales time series may have time series changes with different trends, such as rising, falling or periodic changes. Therefore, in order to be able to more fully and accurately analyze the time series characteristics of the sales volume value of the first oil product in the time dimension, so as to more accurately predict the sales volume of the first oil product, in the technical solution of the present application, the first oil product sales time series input vector is further vector segmented to obtain a sequence of local time series input vectors of the first oil product sales volume. By segmenting the first oil product sales time series input vector, the sales time series of a long time period can be decomposed into multiple local time series segments, so that each local time series segment is more focused on the time series changes and trends of the first oil product sales volume in a specific time period. In this way, the sales characteristics in each time period can be more accurately extracted and analyzed, including sales peaks, volatility, seasonality, etc.

[0042] Then, the sequence of the local time series input vectors of the first oil product sales volume is subjected to feature mining in a sales time series feature extractor based on a one-dimensional convolutional layer to respectively extract the time series change feature information of the first oil product sales volume in each local time segment within the predetermined time period in the time dimension, thereby obtaining a sequence of local time series feature vectors of the first oil product sales volume.

[0043] Furthermore, considering that the sales value of the first oil product not only has a time series variation characteristic within each local time period, but also the time series characteristics within each local time period have a time series correlation relationship based on the whole, this correlation relationship can better reflect the sales fluctuation of the first oil product. The fluctuation characteristics of the sales time series are of great significance to the intelligent scheduling system of refined oil. It can provide information about the volatility and uncertainty of sales volume, thereby helping the system to better understand the change pattern and trend of sales data. Therefore, in the technical solution of the present application, in order to extract the sales time series fluctuation characteristics of the first oil product, it is necessary to further calculate the sales time series fluctuation measurement coefficient between each two adjacent first oil product sales time series characteristic vectors in the sequence of the first oil product sales time series characteristic vector to obtain a sales time series fluctuation characteristic vector composed of multiple sales time series fluctuation measurement coefficients. This helps to understand the instability and volatility of the sales volume of the first oil product and provide a more accurate basis for the formulation of the distribution plan.

[0044] Accordingly, in step S125, the sales time series fluctuation measurement coefficient between each two adjacent first oil product sales local time series characteristic vectors in the sequence of the first oil product sales local time series characteristic vectors is calculated to obtain a sales time series fluctuation characteristic vector composed of a plurality of sales time series fluctuation measurement coefficients, including: calculating the sales time series fluctuation measurement coefficient between each two adjacent first oil product sales local time series characteristic vectors in the sequence of the first oil product sales local time series characteristic vectors by the following fluctuation formula to obtain the sales time series fluctuation characteristic vector composed of a plurality of sales time series fluctuation measurement coefficients; wherein the fluctuation formula is:

[0045]

[0046] in, and are the characteristic values ​​of each position of each two adjacent first oil product sales local time series characteristic vectors in the sequence of the first oil product sales local time series characteristic vectors, is the scale of every two adjacent first oil product sales local time series feature vectors in the sequence of the first oil product sales local time series feature vectors, is the sales time series fluctuation characteristic vector.

[0047] Next, considering that the sales time series fluctuation feature vector reflects the time series dynamic characteristics of the sales volume of the first oil product, that is, the dynamic information such as the time series volatility and periodicity of the sales volume, and the sequence of the local time series feature vector of the sales volume of the first oil product reflects the time series static characteristics of the sales volume of the first oil product, that is, the local time series change feature information of the sales volume in each local time period. Therefore, in order to comprehensively consider the dynamic and static characteristics of the sales time series to more fully and accurately predict the sales volume of the first oil product, in the technical solution of the present application, the sequence of the local time series feature vector of the sales volume of the first oil product and the sales time series fluctuation feature vector are further embedded in the dynamic and static feature fusion module to obtain the dynamic and static fusion feature vector of the sales volume of the first oil product. It should be understood that the dynamic and static feature embedding fusion module can embed the dynamic and static characteristics of the sales volume of the first oil product, so that the feature information between the two complements each other, so that the feature information of different aspects of the sales time series can be comprehensively considered to improve the feature expression ability and prediction accuracy.

[0048] Correspondingly, in step S126, the sequence of the local time series feature vectors of the sales volume of the first oil product and the sales time series fluctuation feature vector are embedded in a fusion module through dynamic and static features to obtain the first oil product sales volume dynamic and static fusion feature vector, including: passing the sales volume time series fluctuation feature vector through a feature extractor based on a fully convolutional network model to obtain a sales volume fluctuation time series feature vector; linearly processing the sales volume fluctuation time series feature vector to obtain a linearly processed sales volume fluctuation time series feature vector; linearly processing the sequence of the local time series feature vectors of the sales volume of the first oil product to obtain a sequence of the local time series feature vectors of the sales volume of the first oil product after linear processing; fusing the linearly processed sales volume fluctuation time series feature vector and the sequence of the linearly processed local time series feature vectors of the sales volume of the first oil product to obtain a linear initial fusion vector; performing one-dimensional convolution processing on the sequence of the local time series feature vectors of the sales volume of the first oil product to obtain a sequence of the time series neighborhood association feature vectors of the sales volume of the first oil product; and fusing the sequence of the time series neighborhood association feature vectors of the sales volume of the first oil product with the linear initial fusion vector based on a splicing method to obtain the first oil product sales volume dynamic and static fusion feature vector.

[0049] Then, the static-dynamic fusion feature vector of the sales volume of the first oil product is passed through a decoder to obtain a decoded value, which is the sales forecast information of the first oil product. In other words, the embedded fusion features between the static feature information and the dynamic feature information of the sales volume of the first oil product are used to perform decoding regression to generate the sales forecast information of the first oil product. In this way, the distribution plan can be optimized, the resource utilization efficiency can be improved, the inventory management can be improved, and the coordination and cooperation of various links in the supply chain can be promoted, thereby improving the efficiency and quality of the distribution of refined oil products.

[0050] Furthermore, in the technical solution of the present application, the intelligent scheduling method for refined oil also includes a training step: for training the sales time series feature extractor based on the one-dimensional convolutional layer, the dynamic and static feature embedding fusion module and the decoder.

[0051] In one example, the training step includes: obtaining training data, the training data including training sales values ​​of the first oil product for multiple days within a predetermined time period; arranging the training sales values ​​of the first oil product for multiple days within the predetermined time period according to the time dimension as a training first oil product sales time series input vector; performing vector segmentation on the training first oil product sales time series input vector to obtain a sequence of training first oil product sales local time series input vectors; passing the sequence of training first oil product sales local time series input vectors through the sales time series feature extractor based on the one-dimensional convolutional layer to obtain a sequence of training first oil product sales local time series feature vectors; calculating the sales relationship between each two adjacent training first oil product sales local time series feature vectors in the sequence of training first oil product sales local time series feature vectors. The sales time series fluctuation measurement coefficient is used to obtain a training sales time series fluctuation feature vector composed of multiple sales time series fluctuation measurement coefficients; the sequence of the training first oil product sales local time series feature vectors and the training sales time series fluctuation feature vector are passed through the dynamic-static feature embedding fusion module to obtain the training first oil product sales dynamic-static fusion feature vector; the training first oil product sales dynamic-static fusion feature vector is passed through the decoder to obtain a decoding loss function value; and, based on the decoding loss function value and through the direction propagation of gradient descent, the sales time series feature extractor based on the one-dimensional convolutional layer, the dynamic-static feature embedding fusion module and the decoder are trained, wherein, at each weight matrix iteration of the training, the training first oil product sales dynamic-static fusion feature vector is trained and optimized.

[0052] In particular, in the above technical solution, the sequence of the local time series feature vectors of the training first oil product sales volume expresses the time series correlation characteristics of the training sales volume data of the first oil product in the local time domain based on the local time domain of vector segmentation in the global time domain, and the training sales volume time series fluctuation feature vector represents the time series fluctuation characteristics of the training sales volume data of the first oil product based on the local time domain as the measurement unit. Accordingly, after the sequence of the training first oil product sales volume local time series feature vectors and the training sales volume time series fluctuation feature vectors are embedded in the fusion module through the dynamic and static features, the feature representation of the sequence of the training first oil product sales volume local time series feature vectors will be constrained based on the time series fluctuation feature representation of the training sales volume time series fluctuation feature vectors based on the local time domain as the measurement unit. In this way, considering the mismatch of the time series correlation scale of the sequence of the training first oil product sales volume local time series feature vectors and the training sales volume time series fluctuation feature vectors, the overall feature distribution of the training first oil product sales volume dynamic and static fusion feature vectors will have more significant inconsistency and instability, thereby affecting its stability in decoding training through the decoder.

[0053] Based on this, the applicant of the present application performs training optimization on the training first oil product sales volume dynamic and static fusion feature vector in each iteration when decoding the training first oil product sales volume dynamic and static fusion feature vector through a decoder.

[0054] Accordingly, in one example, at each weight matrix iteration of the training, the training first oil product sales dynamic and static fusion feature vector is trained and optimized using the following optimization formula to obtain the optimized training first oil product sales dynamic and static fusion feature vector; wherein the optimization formula is:

[0055]

[0056] in, is the dynamic and static fusion feature vector of the first oil product sales volume in the training, is the dynamic and static fusion feature vector of the first oil product sales volume in the training No. The eigenvalues ​​at the positions, and They are the dynamic and static fusion feature vectors of the first oil product sales volume in the training The 1-norm and 2-norm of is the dynamic and static fusion feature vector of the first oil product sales volume in the training The length of is with The relevant weight hyperparameters, represents an exponential operation of a value, wherein the exponential operation of the value represents calculating the value of a natural exponential function with the value as a power, is the first dynamic and static fusion feature vector of the first oil product sales volume after optimization training. The feature value at each position.

[0057] Here, the first oil product sales dynamic and static fusion feature vector is trained The overall feature distribution of is represented by the structural consistency and stability under the rigid structure of absolute distance and the non-rigid structure of spatial distance, so that the training first oil product sales dynamic and static fusion feature vector The global feature distribution has a certain repeatability for local pattern changes, so that the dynamic and static fusion feature vector of the first oil product sales volume in the training When the classifier is used for classification, the global feature distribution is robust to the scale and rotation changes of the weight matrix of the decoder, which improves the stability of the decoding training. In this way, the sales forecast of oil products can be automatically made based on the time series changes of the sales volume of oil products, thereby optimizing the distribution plan, improving the efficiency of resource utilization, improving inventory management, and promoting the coordination and cooperation of various links in the supply chain, and improving the efficiency and quality of finished oil distribution.

[0058] Further, in step S140, if Figure 4 As shown, according to the oil depot information, the gas station information, the vehicle information, the trip information, the constraint information, the sales forecast information of each oil product at the gas station, the optimization index and the parameter setting, an iterative greedy algorithm is used to solve the active distribution scheduling optimization model to obtain the optimal distribution plan, including: S141, obtaining the current optimal solution; S142, selecting N reconstructed vehicles and trips, and cleaning the selected trips; S143, executing the trip reconstruction sub-process to obtain three trip reconstructed plans, and assuming that on the basis of adopting the three trip reconstructed plans, two layers of the trip reconstruction sub-process are nested and executed to obtain a maximum of 27 sets of plans for the three trip reconstructed plans; S144, respectively verifying the legality of the maximum of 27 sets of plans for the trips, and calculating the index score; and, S145, in response to the index score being greater than the index score of the current optimal solution, replacing the current optimal solution.

[0059] The trip reconstruction sub-process includes: selecting the first empty trip, with the middle time point as the ETA; obtaining a list of gas stations that can be arranged for the first empty trip according to constraint information; calculating the empty capacity of each tank at the ETA time point according to the tank inventory, plan, and sales forecast; allocating oil depots to form multiple plans based on the combination of vehicle compartments, empty tank capacity within the station, and load distribution; and selecting three plans with high full load rate, short delivery mileage, and high delivery urgency from the multiple plans.

[0060] In summary, the intelligent scheduling method for refined oil products based on the embodiment of the present application is explained, which can help avoid oversupply or undersupply and improve distribution efficiency and customer satisfaction.

[0061] Figure 5 FIG. 1 is a block diagram of an intelligent dispatching system 100 for refined oil according to an embodiment of the present application. Figure 5 As shown, the intelligent scheduling system 100 for refined oil according to the embodiment of the present application includes: a refueling information acquisition module 110, which is used to obtain oil depot information, gas station information, vehicle information, travel information, and constraint information; a sales forecast information acquisition module 120, which is used to obtain sales forecast information of each oil product at the gas station; an indicator and parameter acquisition module 130, which is used to obtain optimization indicators and parameter settings; and a distribution plan analysis module 140, which is used to solve the active distribution scheduling optimization model using an iterative greedy algorithm to obtain the optimal distribution plan based on the oil depot information, the gas station information, the vehicle information, the travel information, the constraint information, the sales forecast information of each oil product at the gas station, the optimization indicators and the parameter settings.

[0062] Here, those skilled in the art will appreciate that the specific functions and operations of each module in the above-mentioned intelligent dispatching system 100 for refined oil have been described in the above reference. Figures 1 to 4 The description of the intelligent scheduling method of refined oil has been introduced in detail, and therefore, its repeated description will be omitted.

[0063] As described above, the intelligent scheduling system 100 for refined oil according to the embodiment of the present application can be implemented in various wireless terminals, such as a server with an intelligent scheduling algorithm for refined oil. In one example, the intelligent scheduling system 100 for refined oil according to the embodiment of the present application can be integrated into the wireless terminal as a software module and / or a hardware module. For example, the intelligent scheduling system 100 for refined oil can be a software module in the operating system of the wireless terminal, or can be an application developed for the wireless terminal; of course, the intelligent scheduling system 100 for refined oil can also be one of the many hardware modules of the wireless terminal.

[0064] Alternatively, in another example, the intelligent scheduling system 100 for refined oil and the wireless terminal may also be separate devices, and the intelligent scheduling system 100 for refined oil may be connected to the wireless terminal via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0065] Figure 6 FIG. 1 is an application scenario diagram of the intelligent dispatching method for refined oil according to an embodiment of the present application. Figure 6 As shown, in this application scenario, first, the sales value of the first oil product for multiple days within a predetermined time period is obtained (for example, Figure 6 D), then, the sales value of the first oil product for multiple days within a predetermined time period is input to a server (for example, Figure 6 In S) as shown in , the server can use the intelligent scheduling algorithm of the refined oil to process the sales value of the first oil product for multiple days within a predetermined time period to obtain sales forecast information of the first oil product.

[0066] In one embodiment of the present application, a method for dispatching refined oil is also provided to solve the problem of compiling a distribution plan from an oil depot to a gas station. The method in the present application realizes automatic compilation of a distribution plan through an algorithm, such as Figures 7 to 9 shown.

[0067] Among them, reference Figure 7 First, obtain the oil depot information, including the working hours and the available quantity of each oil product in the oil depot; then, obtain the gas station information, including the effective tank capacity, pump stop inventory, inventory time and inventory quantity of each oil product at each gas station; then, obtain the sales forecast information of each oil product at the gas station, decomposing the daily sales forecast and the sales ratio of the time period into the predicted sales volume per hour; then, obtain the vehicle information, including the parking point, compartment capacity and oil products carried by each vehicle, and the working hours of each shift; then, obtain the constraint information, including the restricted time period for vehicles to each gas station and the restricted time period for vehicles to each oil depot; then, obtain the travel information, including the time and distance required for the distribution vehicles from depot to depot, depot to station, station to depot, and station to station; then, obtain the optimization indicators and parameter settings, including the weights of each optimization indicator and the parameters of the business rules; then, establish the active distribution scheduling optimization model; finally, according to the relevant data obtained, the active distribution scheduling optimization model is solved based on the iterative greedy algorithm to obtain the optimal distribution plan.

[0068] In this process, the solution process based on the iterative greedy algorithm, such as Figure 8 As shown, it includes: first, obtaining the current optimal solution; then, selecting N reconstructed vehicles and trips, and cleaning the selected trips; then, executing the trip reconstruction sub-process to obtain a maximum of 3 reconstructed trips, ABC, and assuming that 2 layers of reconstruction sub-processes are nested and executed on the basis of adopting A, B, and C to obtain 3 reconstructed trips and a maximum of 27 sets of solutions; then, the legality of the 27 sets of solutions is verified and the index scores are calculated; if the index is better than the current optimal solution, the current optimal solution is replaced; if the index is not better than the current optimal solution, the index is discarded.

[0069] In this process, the sub-processes are reconstructed, such as Fig. 9As shown, it includes: first, selecting a certain empty trip, and taking the middle time point as ETA; then, obtaining the list of gas stations that can be arranged for the trip at the ETA time according to the constraint information; then, calculating the empty capacity of each tank at the ETA time point according to the tank inventory, plan, and sales forecast; then, according to the combination of vehicle compartments and tank empty capacity to generate the load distribution within the station, the oil depot is allocated to form X sets of plans; finally, selecting ABC, a total of 3 sets of plans with high full load rate, short delivery mileage, and high delivery urgency from the X sets of plans.

[0070] In addition, it will be appreciated by those skilled in the art that various aspects of the present application may be illustrated and described by a number of patentable categories or situations, including any new and useful process, machine, product or combination of substances, or any new and useful improvements thereto. Accordingly, various aspects of the present application may be performed entirely by hardware, entirely by software (including firmware, resident software, microcode, etc.), or by a combination of hardware and software. The above hardware or software may all be referred to as "data blocks", "modules", "engines", "units", "components" or "systems". In addition, various aspects of the present application may be represented as a computer product located in one or more computer-readable media, which includes computer-readable program code.

[0071] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those of ordinary skill in the art to which this application belongs. It should also be understood that terms such as those defined in common dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an idealized or extremely formal sense, unless explicitly defined as such herein.

[0072] The above is an explanation of the present application and should not be considered as a limitation thereof. Although several exemplary embodiments of the present application are described, it will be readily understood by those skilled in the art that many modifications may be made to the exemplary embodiments without departing from the novel teachings and advantages of the present application. Therefore, all of these modifications are intended to be included within the scope of the present application as defined by the claims. It should be understood that the above is an explanation of the present application and should not be considered to be limited to the specific embodiments disclosed, and modifications to the disclosed embodiments and other embodiments are intended to be included within the scope of the appended claims. The present application is defined by the claims and their equivalents.

Claims

1. An intelligent dispatching method for refined oil, characterized in that: include: Obtain oil depot information, gas station information, vehicle information, itinerary information, and constraint information; Obtain sales forecast information for each oil product at a gas station; Get optimization indicators and parameter settings; And according to the oil depot information, the oil station information, the vehicle information, the trip information, the constraint information, the sales forecast information of each oil product of the oil station, the optimization index and the parameter setting, an iterative greedy algorithm is used to solve the active distribution scheduling optimization model to obtain the optimal distribution plan, which includes: Get the current optimal solution; Select N reconstruction vehicles and trips, and clean up the selected trips; Execute the reconstruction sub-process to obtain three reconstruction solutions, and assume that two layers of the reconstruction sub-process are nested and executed on the basis of adopting the three reconstruction solutions to obtain a maximum of 27 sets of solutions for the three reconstruction solutions; Verify the legality of the maximum 27 trip plans respectively and calculate the index scores; and in response to the indicator score being greater than the indicator score of the current optimal solution, replacing the current optimal solution; The oil depot information includes the working hours and the available quantity of each oil product at the oil depot; the gas station information includes the effective tank capacity, pump stop inventory, inventory time and inventory quantity of each oil product at each gas station; the vehicle information includes the parking point, compartment capacity, oil products carried and the working hours of each shift of each vehicle; the travel information includes the time and distance required for the distribution vehicles to travel from depot to depot, depot to station, station to depot and station to station; the constraint information includes the restricted time period for vehicles to each gas station and the restricted time period for vehicles to each oil depot; The optimization index includes weights of various optimization indexes, and the parameter setting includes business rule parameters; The reconstruction sub-process includes: Select the first empty trip and use the middle time as ETA; Obtaining a list of refueling stations that can be arranged for the first empty trip according to the constraint information; Calculate the empty capacity of each tank at ETA point in time based on tank inventory, plans, and sales forecasts; According to the vehicle compartments, empty tank capacity, station and load distribution, the oil depot is allocated to form multiple solutions; And select three plans with high full load rate, short delivery mileage and high delivery urgency from the multiple plans.

2. The intelligent dispatching method for refined oil according to claim 1, characterized in that: Obtain sales forecast information for each oil product at a gas station, including: Obtain sales value of a first oil product for multiple days within a predetermined time period; Arranging the sales values ​​of the first oil product for multiple days within a predetermined time period according to the time dimension into a first oil product sales time series input vector; Performing vector segmentation on the first oil product sales time series input vector to obtain a sequence of first oil product sales local time series input vectors; Passing the sequence of the first oil product sales local time series input vectors through a sales time series feature extractor based on a one-dimensional convolutional layer to obtain a sequence of first oil product sales local time series feature vectors; Calculating the sales time series fluctuation measurement coefficient between each two adjacent first oil product sales local time series characteristic vectors in the sequence of the first oil product sales local time series characteristic vectors to obtain a sales time series fluctuation characteristic vector composed of a plurality of sales time series fluctuation measurement coefficients; The sequence of the local time series feature vector of the sales volume of the first oil product and the time series fluctuation feature vector of the sales volume are embedded in a dynamic and static feature fusion module to obtain a dynamic and static fusion feature vector of the sales volume of the first oil product; And the dynamic and static fusion feature vector of the first oil product sales volume is passed through a decoder to obtain a decoded value, and the decoded value is the sales forecast information of the first oil product.

3. The intelligent dispatching method for refined oil according to claim 2, characterized in that: Calculating the sales time series fluctuation measurement coefficient between each two adjacent first oil product sales local time series characteristic vectors in the sequence of the first oil product sales local time series characteristic vectors to obtain a sales time series fluctuation characteristic vector composed of a plurality of sales time series fluctuation measurement coefficients, including: The sales time series fluctuation measurement coefficient between each two adjacent first oil product sales local time series characteristic vectors in the sequence of the first oil product sales local time series characteristic vectors is calculated using the following fluctuation formula to obtain the sales time series fluctuation characteristic vector composed of multiple sales time series fluctuation measurement coefficients; Wherein, the fluctuation formula is: ； in, and are the characteristic values ​​of each position of each two adjacent first oil product sales local time series characteristic vectors in the sequence of the first oil product sales local time series characteristic vectors, is the scale of every two adjacent first oil product sales local time series feature vectors in the sequence of the first oil product sales local time series feature vectors, is the sales time series fluctuation characteristic vector.

4. The intelligent dispatching method for refined oil according to claim 3, characterized in that: The sequence of the local time series feature vectors of the first oil product sales volume and the time series fluctuation feature vector of the sales volume are embedded in a dynamic and static feature fusion module to obtain the dynamic and static fusion feature vector of the first oil product sales volume, including: The sales volume time series fluctuation feature vector is passed through a feature extractor based on a fully convolutional network model to obtain a sales volume fluctuation time series feature vector; Performing linear processing on the sales volume fluctuation time series feature vector to obtain a linearly processed sales volume fluctuation time series feature vector; Performing linear processing on the sequence of the local time series feature vectors of the first oil product sales volume to obtain a sequence of the local time series feature vectors of the first oil product sales volume after linear processing; Fusion of the linearly processed sales volume fluctuation time series feature vector and the linearly processed first oil product sales volume local time series feature vector sequence to obtain a linear initial fusion vector; Performing one-dimensional convolution processing on the sequence of local time series feature vectors of the first oil product sales volume to obtain a sequence of neighborhood association feature vectors of the first oil product sales volume time series; And the sequence of the first oil product sales time series neighborhood correlation feature vectors and the linear initial fusion vector are fused based on the splicing method to obtain the first oil product sales dynamic and static fusion feature vector.

5. The intelligent dispatching method for refined oil according to claim 4, characterized in that: The method further includes a training step: for training the sales time series feature extractor based on the one-dimensional convolution layer, the dynamic and static feature embedding fusion module and the decoder; Wherein, the training step includes: Acquire training data, wherein the training data includes training sales values ​​of a first oil product for a plurality of days within a predetermined time period; Arranging the training sales values ​​of the first oil product for multiple days within a predetermined time period according to the time dimension into a time series input vector for training the sales value of the first oil product; Performing vector segmentation on the training first oil product sales time series input vector to obtain a sequence of training first oil product sales local time series input vectors; Passing the sequence of the local time series input vectors of the first oil product sales volume for training through the sales time series feature extractor based on the one-dimensional convolution layer to obtain a sequence of local time series feature vectors of the first oil product sales volume for training; Calculating the sales time series fluctuation measurement coefficient between each two adjacent training first oil product sales local time series characteristic vectors in the sequence of the training first oil product sales local time series characteristic vectors to obtain a training sales time series fluctuation characteristic vector composed of a plurality of sales time series fluctuation measurement coefficients; The sequence of the local time series feature vectors of the training first oil product sales volume and the training sales volume time series fluctuation feature vector are embedded in the dynamic and static feature fusion module to obtain the dynamic and static fusion feature vector of the training first oil product sales volume; Passing the training first oil product sales dynamic and static fusion feature vector through the decoder to obtain a decoding loss function value; And based on the decoding loss function value and through the direction propagation of gradient descent, the sales time series feature extractor based on the one-dimensional convolution layer, the dynamic and static feature embedding fusion module and the decoder are trained, wherein, at each weight matrix iteration of the training, the dynamic and static fusion feature vector of the training first oil product sales volume is trained and optimized.

6. An intelligent dispatching system for refined oil, which adopts the intelligent dispatching method for refined oil as claimed in claim 5, characterized in that: include: The refueling information acquisition module is used to obtain oil depot information, gas station information, vehicle information, itinerary information, and constraint information; The sales forecast information acquisition module is used to obtain the sales forecast information of each oil product at the gas station; Indicator and parameter acquisition module, used to obtain optimization indicators and parameter settings; and a distribution plan analysis module, which is used to solve the active distribution scheduling optimization model using an iterative greedy algorithm to obtain the optimal distribution plan based on the oil depot information, the gas station information, the vehicle information, the trip information, the constraint information, the sales forecast information of each oil product at the gas station, the optimization index and the parameter setting.

Images